Supplementary Table 1. Classification of pathogenic BRCA1 mutations in prophylactic mastectomy samples Patient ID Age (yrs) BIC classification * HGVS classification # 1 2 BRCA1 del exon BRCA1g.71598-?_71681+?del (del exon) 2 BRCA1 562 T>A (V188E) BRCA1c.551T>A (p.val188glu) BRCA1 5586 G>A (A182T) BRCA1c.5467G>A (p.ala182thr) 4 27 BRCA1 IVS 6-2 del A BRCA1c.2-2delA 5 44 BRCA1 185_186 del AG (STOP BRCA1c.68_69delAG (p.glu2valfsx17) 9) 6 48 BRCA1 T>G (C61G) BRCA1c.181T>G (p.cys61gly) 7 4 BRCA1 del exon 24 BRCA1g.8296-?_8446+?del (del exon24) 8 9 BRCA1 1996insTAGT BRCA1c.1874_1877dupTAGT 9 BRCA1 5586 G>A (A182T) BRCA1c.5467G>A (p.ala182thr) 29 BRCA1 726 C>T (R1X) BRCA1c.7C>T (p.arg1x) 11 BRCA1 4184_4187 del TCAA (STOP 164) BRCA1c.65_68delTCAA (p.asn155lysfsx) 12 BRCA1 del exons 21_2 BRCA1g.776-?_894+?del (del exons21_2) 1 49 BRCA1 2919 C>T (Q94X) BRCA1c.2C>T (p.gln94x) 14 9 BRCA1 519 G>T (E114X) BRCA1c.G>T (p.glu114x) 15 26 BRCA1 2_21 del AA (STOP 91) BRCA1c.2681_2682delAA (p.lys894thrfsx8) 16 41 BRCA1 Ex 1 Dup n (STOP 14) g4469_5449 dup 6kb BRCA1g.4469_5449 dup6kb (dup exon1) 17 1 BRCA1 74_75 ins GA (STOP 87) BRCA1c.254_255dupGA (p. Leu86AspfsX2) 18 2 BRCA1 2594 del C (STOP 845) BRCA1c.2475delC (p.asp825glufsx21) 19 BRCA1 4744 del CT (STOP 1572) BRCA1c.4625_4626delCT (p.ser1542trp) 1 BRCA1 582_58 ins C (STOP BRCA1c.5266dupC (p. Gln1756ProfsX74) 1829) 21 42 BRCA1 6 del C (STOP 2) BRCA1c.514delC (p.gln172asnfsx62) 22 5 BRCA1 del A (STOP 7) BRCA1c.1961delA (p.lys654serfsx47) 2 42 BRCA1 T>G (C61G) BRCA1c.181T>G (p.cys61gly) 24 BRCA1 _6 del GAAGATACTAG (STOP 116) BRCA1c.481_491delGAAGATACTAG (p.glu1161phefsx) All prophylactic mastectomy samples were from premenopausal women with no personal history of cancer. *BIC: Breast cancer Information Core database (http://research.nhgri.nih.gov/bic/) #HGVS: Human Genome Variation Society (http://www.hgvs.org) Nature Medicine: doi:.8/nm.4118
Supplementary Table 2. and L expression in normal breast tissue Gene Mutation Number p-value L p-value Status of samples H-score H-score BRCA1 BRCA1 mut/+ 78.2.2 24.12.94 Non-BRCA1 6.75 26.45 BRCA2 BRCA2 mut/+ 58 16.67.64 25.98.8 Non-BRCA2 26 11.98 25.89 Normal breast tissue (either tumor adjacent or contralateral breast) from the kconfab cohort 41 was scored for BRCA1 and BRCA2 expression. The H-score incorporates intensity (scale of - ) multiplied by percent cells staining positive for or L (see Methods). P-values (Wilcoxon Rank Sum test) compare BRCA1 mutation carriers with non-brca1 mutant tissue (i.e. women with a positive family history but no BRCA1 germline mutation; includes BRCA2- mutation carriers) and BRCA2 mutation carriers with non-brca2 mutant tissue (i.e. women with a positive family history but no BRCA2 germline mutation; includes BRCA1-mutation carriers). Analysis was not adjusted for menopausal status or menstrual cycle stage. Nature Medicine: doi:.8/nm.4118
Supplementary Table. Incidence scores of and L expression in primary breast tumors Patient Genotype + (%) L + (%) Wild-type (WT) 1/11 (%) 7/12 (12%) BRCA1 mut/+ 61/144 (42%) 1/141 (9%) BRCA2 mut/+ 17/115 (15%) 5/116 (4%) Tumors were scored from tissue microarrays from the kconfab cohort 41. BRCA1- and BRCA2- associated tumors were from known mutation carriers, while WT were from BRCAX cases with a positive family history, where no BRCA1 or BRCA2 germline mutation was identified by germline testing. BRCA1-associated tumors are more frequently + than WT (Fisher s exact test p = 1.8e-14) or BRCA2 (p = 1.4e-6). No association was observed for L staining between different groups. Nature Medicine: doi:.8/nm.4118
Supplementary Table 4. Clinicopathological characteristics of primary human tumors a. Characteristics of the primary tumor cohort Marker Marker BRCA1 BRCA2 WT p-value Status ER Negative 165 (77.1%) 2 (17.9%) 146 (2.1%) 1.9e-49 Positive 49 (22.9%) 147 (82.1%) 486 (76.9%) Total 214 179 62 PR Negative 19 (9.6%) 8 (.%) 25 (9.7%) 6.75e-42 Positive (9.4%) 72 (.%) 79 (.%) Total 21 1 629 HER2 Negative 172 (86.%) 127 (74.7) 54 (85.7%). Positive 28 (14.%) 4 (25.%) 84 (14.%) Total 17 588 Triple-Neg False 71 (4.8%) 15 (84.%) 52 (85.4%) 2.e-4 True 1 (65.2%) 28 (15.7%) 91 (14.6%) Total 4 178 62 EGFR Negative 181 (9.5%) 1 (96.4%) 88 (74.%) 4.e-14 Positive 19 (9.5%) 6 (.6%) 14 (25.7%) Total 166 522 CK5 Negative 59 (.1%) 92 (57.5%) 265 (51.9%) 2.88e-8 Positive 17 (69.9%) 68 (42.5%) 246 (48.1%) Total 196 1 511 Negative 9 (59.2%) 115 (85.8%) 28 (87.1%) 2.6e-11 Positive 62 (.8%) 19 (14.2%) 42 (12.9%) Total 152 14 25 L Negative 121 (9.%) 11 (95.8%) (86.7%).2 Positive 1 (9.7%) 5 (4.2%) 46 (1.%) Total 14 118 46 b. Proportion of + tumors within patient genotypes and tumor subtypes Genotype TNBC HR + HER2 BRCA1.472 (42/89).24 (12/7).64 (4/11) BRCA2.412 (7/17).112 (12/7). (/1) WT.244 (19/78). (19/184).16 (/22) BRCA1 vs WT overall: p =.62e-11 BRCA1 vs WT for TNBC only: p =.2 BRCA1 vs WT for not-tnbc: p =.4 BRCA1 vs WT for HR + : p =.1 BRCA1 vs WT for HER2: p =.186 Tumors were scored from the kconfab cohort 41 (tissue microarrays) and Amgen Tissue Bank (tissue sections). BRCA1- and BRCA2-associated tumors were from known mutation carriers. Wild-type (WT) were cases where no BRCA1 or BRCA2 mutation had been identified on genetic testing. HR, Hormone Receptor. P values were calculated using Fisher s exact test. Nature Medicine: doi:.8/nm.4118
Supplementary Table 5. L levels in normal breast tissue from high-risk women Hormonal Status BRCA1 BRCA2 WT L + L Total L + L Total L + L Total Pre-men 25 (9.1%) 9 (.9%) 64 15 (4.9%) 28 (65.1%) 4 68 (44.2%) 86 (55.8%) 154 Post-men 1 (8.%) 11 (91.7%) 12 (.%) (%) (6.8%) 41 (9.2%) 44 p =.49 p =.46 p = 1.7e-6 L expression was scored in normal breast tissue (either tumor adjacent or contralateral breast) from tissue microarrays in the kconfab cohort 41. High-risk women included patients with pathogenic mutations in BRCA1 or BRCA2, or WT (patients had a positive family history, but no BRCA1 or BRCA2 germline mutation was identified). Menstrual cycle status for premenopausal populations is unknown. The definition of post-menopausal hormonal (Post-men) breast status was based on either: age >55 years, prior oophorectomy, or on tamoxifen at the sample collection day or within days prior to collection day. In all three groups (BRCA1, BRCA2 and WT), the fraction of samples from pre-menopausal (Pre-men) women that was L-positive, was significantly greater than samples obtained from post-menopausal (post-men) women (p <.5, Fisher s exact test). Nature Medicine: doi:.8/nm.4118
Stroma Isotype Stroma Stroma Lin Stroma EpCAM 25 Fwd scatter (x) CD49f Isotype Relative fold change e WT1 WT2.5 Stroma f Isotype Stroma Stroma Stroma Stroma 5 15 25 d Isotype Stroma Fwd scatter (x) L mrna 1. Isotype Percentage of max Percentage of max Stroma BRCA1mut/+ WT Stroma 5 15 2 c Percentage of max Isotype Percentage of max BRCA1mut/+ EpCAM 4 CD1, CD45, CD25-a CD1, CD45, CD25-a b WT Expression relative to GAPDH a CD49f mrna 12 8 6 4 2 L + segments of TDLU L+ duct L M Ba S /M l sa a L TDLU m ro St Supplementary Figure 1. expression in human breast epithelium. (a,b) Additional examples of FACS plots showing delineation of distinct subpopulations isolated from (a) wild-type or (b) histologically normal BRCA1mut/+ breast tissue based on expression of CD1, CD45, CD25-α, EpCAM, CD49f and. Each row of plots correspond to an individual patient. Cells negative for CD1, CD45 and CD25-α define the lineage negative () population. expression (blue) was compared to an isotype-matched control antibody (grey). (c) Representative histograms portraying the lack of expression in the mature luminal () and stroma subsets within wild-type (n = patients) or BRCA1mut/+ (n = 24) breast tissue. (d) Expression analysis of in the, basal/masc-enriched () and luminal progenitor () subpopulations from reduction mammoplasties from three WT patients by qrt-pcr. Data is depicted as mean ± s.e.m. Expression was normalized to GAPDH. (e) Expression analysis of L in the,, and stroma subpopulations from reduction mammoplasties from two patients (WT1 and WT2) by qrtpcr. Expression was normalized to GAPDH and is depicted as fold-change relative to the mature luminal subset. (f) Representative images showing heterogeneous expression of L and in histologically normal human breast tissue by immunohistochemistry. Scale bar = μm. Nature Medicine: doi:.8/nm.4118
a Side scatter (x) With DEAB b 25 15 5 2 4 5 No DEAB 25 15 5 2 4 5 Aldefluor/BAA Relative fold change 2.5 2. 1.5 1..5 Aldefluor/BAA MFI BRCA1 mut/+ WT + c H-score **** * WT BRCA1 BRCA2 Mean = 2. Mean = 1.6 Mean = 5. d L e ALDH1 Supplementary Figure 2. + progenitors exhibit enhanced Aldefluor activity. (a) Representative bodipyaminoacetate (BAA) fluorescence profiles of luminal progenitor cells after incubation with Aldefluor in the presence or absence of the inhibitor diethylaminobenzaldehyde (DEAB). Aldefluor-positive cells are outlined in red (b) Mean fluorescence intensity (MFI) values for the Aldefluor/BAA fluorescence of + luminal progenitor cells isolated from BRCA1 mut/+ (n = 4 patients) and WT (n = 5) breast tissue, relative to the MFI of the corresponding luminal progenitor subset. Data is depicted as mean ± s.e.m ** P <.1, *** P <.1 (c) Elevated expression in BRCA1-mutated breast tumors. expression was determined by immunohistochemistry of breast tumors on kconfab tissue microarrays (TMAs). The overall expression was generated using the H-scoring method (staining intensity multiplied by percent positive cells). Bar, mean score. * P <.5, **** P <.1. (d) L immunostaining in serial sections of a + BRCA1-mutated breast tumor showing distribution within normal breast tissue adjacent to the tumor. Scale bar = μm. (e) Representative images showing overlapping distribution of and ALDH1 immunostaining in sequential sections from the same BRCA1-mutated breast tumor. Scale bar = μm. Statistical significance was determined using two-tailed t-tests. ALDH1, aldehyde dehydrogenase 1. Nature Medicine: doi:.8/nm.4118
a Cell cycle Oocyte meiosis DNA replication Fanconi anemia pathway Fatty acid metabolism Homologous recombination Steroid biosynthesis Biosynthesis of unsaturated fatty acids MicroRNAs in cancer Fatty acid degradation Glycoxylate/Dicarboxylate metabolism Ketone bodies Pyruvate metaolism Butanoate metabolism p5 signalling pathway Tryptophan metabolism Glycosphingolipid biosynthesis Mismatch repair Valine/leucine/isoleucine degradation Fatty acid elongation Pentose/Glucuronate interconversions Arachidonic acid metabolism Lysine degradation Carbon metabolism Fat digestion and absorption c Olive tail moment e 5 UT HU IR UT HU IR + log (P value) 5 15 d Olive tail moment 1 1 b Relative fold change.5. 2.5 2. 1.5 1..5. WT BRCA1 Lum B Basal Normal Basal 4.1 8.6 5.4. 1.5 -.1-1.9 -.6-5.8-9.2-41.1 28.6.8 7. 4.4 2.2.2-1.8-4. -6.6 -.4-1.6 Enrichment 2.7 Enrichment 5.6 Enrichment 2. BRCA1 mut/+ Unsorted Epithelial + + UT HU IR Enrichment 2 Weight P value =.1 P value =.1 Weight Weight Weight Supplementary Figure. BRCA1 mut/+ + luminal progenitor cells have enhanced mitotic activity and correlate with basal-like breast cancers. (a) KEGG analysis following RNA-seq of freshly sorted + and luminal progenitors () revealed significant enrichment of cell cycle, DNA repair and metabolic pathways in + cells (n = 4 patients) (b) Fold change in BRCA1 expression in + cells relative to the corresponding subset for WT (n = patients) and BRCA1 mut/+ (n = ). Data represent mean ± s.e.m. (c) A representative patient sample showing the olive tail moment for individual + or cells in BRCA1 mut/+ tissue. Cells were either untreated (UT), treated with hydroxyurea (HU) or irradiated at Gy (IR), and harvested 4 h post-treatment. Bar, mean score. (d) Comet assays on epithelial subsets of BRCA1 mut/+ cells. The histograms represent the mean olive tail movement ± s.e.m. for at least cells per population. Shown are olive tail movements for unpurified cells (epithelium and stroma), epithelial cells, basal/ MaSC, and and + luminal progenitors. Comet tails are evident in all subsets, but are most pronounced in the + subset (n = patients). (e) Barcode plots depicting association between the + expression signature and basal breast tumors, compared to luminal B and normal-like breast tumors. Genes are ordered from right to left as most upregulated to most downregulated in basal tumors. Vertical red bars designate genes upregulated in + cells compared to cells, whereas blue bars designate downregulated genes. Weight refers to the log-fold change in expression of each gene compared to cells, with a higher fold change represented as a greater bar length. ROAST P values assess overall correlation. Nature Medicine: doi:.8/nm.4118
a Prog Prog + denosumab BRCA1mut/+ #5 BRCA1mut/+ #4 BRCA1mut/+ # BRCA1mut/+ #2 BRCA1mut/+ #1 Vehicle b Post-denosumab Ki67 Pre-denosumab Supplementary Figure 4. L inhibition attenuates progesterone-mediated proliferation in BRCA1mut/+ breast tissue. (a) Representative images from each of the five BRCA1mut/+ patients showing Ki67 immunostaining in freshly isolated organoids that were treated with vehicle (EtOH), progesterone (Prog, nm) or progesterone ( nm) plus the L-inhibitor denosumab ( µg/ml) for 24 h. The enlargement shows each image at x magnification. Scale bars = 5 μm. (b) Representative images showing Ki67 immunostaining of pre- (left) and post-denosumab treatment (right) core biopsies from subject # in the BRCA-D study. Scale bars = 5 µm. Nature Medicine: doi:.8/nm.4118
a b CD29 (HR ) CD49b c (HR+) Vehicle anti-l d Sca1 CD24 59.8% Relative fold change (HR ) 28.5% 8.% 9.5% (HR+) Mat Lum anti-l Sca1 CD24 Vehicle CD29 46.% 9.6% 1..8 *.6.4 CD49b Colony-forming capacity 1.2 Vehicle anti-l **.2. (HR ) 8.5 dp 1 dl MMTV-cre/Brca1fl/fl/p5+/ p5+/ MMTV-Neu Ligand 6 wk virgin (HR+) e Individual female MMTV-cre/Brca1fl/fl/p5+/ (9 weeks) Randomly assigned to treatment group Vehicle or a-l Tumor harvested Nature Medicine: doi:.8/nm.4118 g Tumor-free survival (%) f *** Vehicle a-l 25 5 75 125 15 175 225 25 Days
Supplementary Figure 5. L inhibition attenuates tumor onset in vivo. (a) Representative FACS plots showing expression of CD24, CD29, Sca1 and CD49b in mouse mammary glands pooled from virgin FVB/J mice treated with either anti-l (5 mg/kg) or an isotype-matched control antibody (5 mg/kg) for 7 days. Differential expression of Sca1 and CD49b was used to delineate the mature luminal (Sca1 + CD49b ), hormone receptor-positive luminal progenitor (HR +, Sca1 + CD49b + ) and hormone receptor-negative luminal progenitor (HR, Sca1 CD49b + ) subsets. It has not been possible to sort + and subsets from the mouse mammary gland, owing to a lack of suitable antimouse antibodies for flow cytometry. (b) Representative images showing Giemsa-stained colonies from basal/ MaSC-enriched (basal/ms), HR + and HR cells isolated from vehicle versus anti-l treated mice, plated on fibroblast feeder layers. (c) Bar graph depicting reduced clonogenic capacity in basal/ms and HR subsets isolated from anti-l treated mice compared to control mice. Data represents mean ± s.e.m. (n = 6 mice). *P <.5, **P <.1. (d) Representative images showing immunostaining for mu and mul in mammary glands harvested from virgin, 8.5 day pregnant or 1 day lactating FVB/J mice. Scale bar = 5 μm. (e) Representative images showing enrichment of mu immunostaining in premalignant mammary glands harvested from MMTV-cre/Brca1 fl/fl /p5 +/ mice, compared to p5 +/ and MMTV-Neu mice (n = 5 mice per model). Scale bar = 5 μm (f) Overview of alternative prevention therapy strategy. Individual 9-week old MMTV-cre/Brca1 fl/fl /p5 +/ mice were randomly assigned to receive either an anti-l neutralizing antibody (5 mg/kg) or an isotype-matched control antibody (vehicle, 5 mg/kg). Tumors were harvested once ethical endpoint ( mm ) was reached. (g) Kaplan-Meier survival curves of MMTV-cre/Brca1 fl/fl / p5 +/ mice following treatment with anti-l (n = 9) or vehicle (n = 9). The dotted line depicts median tumor onset. *** P <.1. Statistical significance was determined using two-tailed t-tests. Nature Medicine: doi:.8/nm.4118
a BRCA1-mutant PDX b c Percentage of max Isotype 2 4 5 d Tumor volume (mm ) 7 5 Survival (%) Vehicle OPG-Fc Docetaxel Docetaxel + OPG-Fc 9 Days 5 7 9 1 Days Supplementary Figure 6. L inhibition augments tumor response to docetaxel in a + BRCA1-mutant patient derived xenograft model. (a) Representative FACS plot showing expression in a BRCA1-mutant patient derived xenograft (PDX 1). expression (blue) was compared to an isotype-matched control antibody (grey). Approximately % of tumor cells in this PDX model are +. (b) immunostaining of PDX 1. Scale bar = µm. (c) Tumor growth curve and (d) Kaplan-Meier survival curves for mice bearing tumors that were treated with vehicle, docetaxel ( mg/kg i.p every 21 days), OPG-Fc ( mg/kg, times per week) or both docetaxel and OPG-Fc (n = mice per arm). Tumors were harvested once ethical endpoint ( mm ) was reached. Mice were treated as previously described 55, although estradiol pellets were omitted to minimize the induction of endogenous OPG. The dotted line depicts median tumor onset. Statistical significance was determined using a twotailed t-test. **** P <.1. Nature Medicine: doi:.8/nm.4118