Supplemental Figure S1. A. Venn diagram depicting overlap between anti-correlated genes of

Supplemental Figure S1. A. Venn diagram depicting overlap between anti-correlated genes of 1,000 most differentially expressed genes with NKX2-1 amplification in lung adenocarcinoma cell lines and anti-correlated genes of 1,000 most differentially expressed genes with top 10 th percentile highest NKX2-1 expression in primary lung adenocarcinomas. (hypergeometric p- value of significant overlap = 1.4x10-9 ) B. left, Expression level of the most differentially expressed 38 common genes in 11 NKX2-1-amplified cell lines and 17 cell lines with the lowest NKX2-1 expression are shown in heatmap. Cell lines are ordered from left to right by copy number at NKX2-1 locus. right, Expression level of 38 common genes in primary lung adenocarcinomas with top and bottom 10 th percentile NKX2-1 expression are shown in heatmap. Samples are ordered from left to right by NKX2-1 expression level. Complete list of the genes is available in Supplemental Table 1. C. The gene set of 147 genes overlapped to be positively correlated with NKX2-1 amplification and overexpression (Fig. 1C) was compared to 100 expression cluster signature gene sets from Director s Challenge Consortium s lung adenocarcinoma dataset for significant overlap (hypergeometric p-value is shown in blue bars scaled with log10 on the left Y-axis and overlapping gene numbers per number of genes in each geneset (k/k) in red bars scaled on right Y-axis). 57 clusters with one or more overlapped genes are rank ordered by p-value from left to right. Supplemental Figure S2. A. Scatter plot on NKX2-1 copy and NKX2-1 expression of 42 lung adenocarcinoma cell lines with matched data. X-axis shows copy number of NKX2-1 gene inferred from 250K SNP array and Y-axis shows RAM normalized expression value in log2 order of NKX2-1 from Affymetrix U133 array. 11 cell lines with highest copy number all of which have high expression in red circle in top right and 17 cell lines with low copy number and

low expression in red circle in bottom left were used for analysis in Fig. 1. B. Western blot of the lysates from 5 cell lines, one without NKX2-1 amplification (A549) and 4 cell lines with NKX2-1 amplification (the rest) with anti-foxa1 antibody, anti-nkx2-1 antibody and anti-vinculin antibody. Supplemental Figure S3. A. Average conservation score (PhastCons) is plotted on 3kb around the peak summit of 7,469 NKX2-1-binding regions in NCI-H3122 cells. The score is the highest at the center. B. Views on the SFTPB, CCNB1 and ROR1 gene loci of shifted read counts from ChIP-seq by anti-nkx2-1 antibody as well as read counts from no-chip input control. C. ChIP enrichment (% recovery of input) by anti-nkx2-1 antibody or mock experiment without antibody (no Ab) with NCI-H3122 cells measured by qpcr at two negative control loci (HBG1 and HOXC8 promoters) and 15 arbitrarily selected enriched sites from ChIP-seq data of NCI- H3122 cell line. All enriched sites were verified to be enriched. D. ChIP enrichment (% recovery of input) by three anti-nkx2-1 antibodies (H-190; Santa Cruz, BL3998 (N-term); Bethyl, BL4000 (C-term); Bethyl) or mock experiment without antibody (no Ab) measured by qpcr at one negative control (HBG1 promoter) and 6 enriched sites. All 6 positive sites were verified to be enriched with all three antibodies. The enrichment on these positive loci were abrogated by pre-incubating BL4000 antibody and its corresponding epitope peptides (BL4000+BP). Supplemental Figure S4. A. Histograms showing frequency of statistical p-value of overlapping genomic intervals of 3,358 combined NKX2-1 binding regions in NKX2-1-amplified lung adenocarcinoma cell lines with 5,780 published retinoic acid receptor alpha (RARa) binding regions (top left), 14,505 published estrogen receptor alpha (ERa) binding regions (bottom left),

and 3,648 published GATA3 binding regions in the MCF7 breast carcinoma cell line. RARa and ERa binding regions is much more similar to NKX2-1 binding regions compared to GATA3 binding regions (bottom right). Top right, Venn diagram depicting overlapped genomic regions between common NKX2-1 binding regions (3,358) in lung adenocarcinoma cell lines found in this study, published 4,870 retinoic acid receptor alpha (RARa) binding regions in the MCF7 breast carcinoma cell line and published 14,505 estrogen alpha (ERa) binding regions in the MCF7 cell line. B. 13,291 genes are plotted in rank-order by differential expression SAM score in 47 (top 10 th percentile) primary human lung adenocarcinomas with highest expression of NKX2-1 compared to 47 (bottom 10 th percentile) tumors with the lowest NKX2-1 expression from expression profiles of 470 primary lung adenocarcinoma. Red diamonds indicate positively correlated genes associated with the AP-1 complex or forkhead box transcription factors. Supplemental Figure S5. A. left. Venn diagram depicting 14 overlapped genes (green) that signify high NKX2-1 expressing primary tumors in the expression datasets from Ding et al. paper, (blue) that signify NKX2-1 amplified lung adenocarcinoma cell lines and (red) of that NKX2-1 occupancy is in 3kb upstream or 1kb downstream of TSS in NCI-H3122 cell line. Right. Heatmap showing expression of 14 genes in 11 cell lines with NKX2-1 amplification and 17 low NKX2-1 expressing cell lines (top) and 10 highest NKX2-1 expressing primary tumors and 10 lowest NKX2-1 expressing tumors. B. left/middle. DNA content analysis on NCI-H3122 with control shrna (left) or with shrna against LMO3 (right). Suppression of LMO3 led to increased SubG1 population that indicates apoptotic cells whereas it did not affect the number of total S phase cell population. right. mrna expression of LMO3 after suppression of LMO3 by shrna.

Supplemental Figure S6. A. LMO3 expression is compared across multiple cancer types (datasets from International Genomics Consortium). The expression is highest in thyroid cancers and lung cancers. Tumor type (number of samples): from left to right, Appendix Carcinoma (1), Astrocytoma (2), Benign Fibrous Tumor (1), Bladder Carcinoma (33), Breast Carcinoma (336), Cervical Carcinoma (36), Colon Carcinoma (302), Colorectal Carcinoma (71), Endometrial Carcinoma (177), Ependymoma (1), Esophageal Carcinoma (7), Fallopian Tube Carcinoma (6), Gallbladder Carcinoma (2), Gastric Carcinoma (14), Gastrointestinal Carcinoid Tumor (2), Gastrointestinal Stromal Tumor (7), Glioblastoma Multiforme (1), Head and Neck Carcinoma (11), Liver and Intrahepatic Biliary Tract Carcinoma (13), Lymphoma (21), Melanoma (9), Neuroendocrine Carcinoma (7), Oncocytoma (2), Ovarian Carcinoma (241), Pancreatic Carcinoma (20), Penile Carcinoma (1), Peritoneal Carcinoma (14), Prostate Carcinoma (60), Renal Carcinoma (254), Sarcoma (77), Schwannoma (1), Skin Carcinoma (4), Small Intestinal Carcinoma (4), Testicular Carcinoma (4), Thyroid Gland Carcinoma (33), Ureter Carcinoma (4), Vaginal Carcinoma (1), Vulvar Carcinoma (9) and lung carcinoma (107). B. LMO3 expression is compared between squamous cell carcinoma and adenocarcinoma of the lung (datasets from Bhattacharjee et al. 2001). LMO3 expression is significantly higher in lung adenocarcinoma than squamous cell carcinoma. C. in situ hybridization with murine Lmo3 probe on an adult mouse lung parenchyma (left) and trachea (right). Lmo3 staining is positive in bronchiolar (B) epithelial cells, but negative in alveolar (A), vascular (V) architecture. Lmo3 staining is also negative in proximal airway. D. Heatmap showing most differentially down-regulated genes after knockdown of NKX2-1 via shrna compared to shrna against GFP control in NCI-H2009 cell

lines analyzed on Affymetrix U133 array. Three replicates each for two shrna against NKX2-1 and four replicates for shrna against GFP. Supplemental Figure S7. A, ChIP enrichment (% recovery of input) either by anti-nkx2-1 or anti-foxa1 antibody with the lysates from NCI-H3122 cells on the regions surrounding the TSS of LMO3 gene measured by qpcr. Primers designed to amplify 100-150bp of each region upstream or downstream of TSS were shown from left to right in the order of genomic location. B. 28-species multiple vertebrate and mammalian conservation score and alignment taken UCSC browser at the enriched NKX2-1 binding locus on the LMO3 gene (bottom). Multiple sequence alignment of LMO3 locus (from 823 to 868bp downstream of TSS) containing the enriched binding motif found in this study (Fig. 3; in red circle) in several species (top). C. Firefly luciferase activity was significantly higher with co-transfection of NKX2-1 expression vector (pcdna3.1-nkx2-1) and luciferase reporter construct (pgl4.10) containing the putative endogenous LMO3 promoter region (130 to 1149 downstream of TSS) or SFTPB promoter on 293T cells compared to co-transfection of GFP expression vector (pcdna3.1-gfp), whereas there was no difference with empty control pgl4.10 vector. Renilla luciferase reporter construct under TK promoter was co-transfected in all experiments and Firefly luciferase activity were normalized against Renilla luciferase activity. Supplemental Figure S8. A. Ectopic expression of NKX2-1 does not expression of LMO3 in two cell lines (H2347 and HCC78) that express NKX2-1 at high level but without NKX2-1 amplification. B. Cell growth curve of A549 cells with ectopic NKX2-1 expression or control GFP expression. Forced expression of NKX2-1 slowed cell growth in mucinous lung

adenocarcinoma cell line, A549. C. Segmented copy number on the genomic region (x-axis) that includes NKX2-1 and FOXA1 gene loci (indicated in yellow blocks) from three SNP array datasets; primary NSCLC (top), primary lung adenocarcinomas from TCGA (middle) and lung adenocarcinoma cell lines (bottom). Primary samples or cell lines with significant amplification at the NKX2-1 locus (>3.6 inferred copy number) are shown in order of copy number at the locus from top to bottom. Of these, 67/80 (84%) in primary NSCLC, 25/30 (83%) in TCGA lung adenocarcinomas and 14/15 (93%) in lung adenocarcinoma cell lines are co-amplified at the FOXA1 gene locus. D. top. Ectopic expression of NKX2-1 (in plex-blast vector) only induced ~6-fold expression of LMO3 in GFP (in plex-puro)-expressing A549 cells, whereas it induced ~30-fold expression of LMO3 in FOXA1 (in plex-puro)-expressing A549 cells. bottom. Western blot showing ectopic expression of NKX2-1 and FOXA1 in A549 cells along with a loading control probed with anti-vinculin antibody. E. top. Suppression of FOXA1 in NCI- H3122 cells via three shrnas directed against FOXA1 results in reduced LMO3 mrna expression. bottom. Western blots confirmig effective suppression of FOXA1 protein with three shrnas against FOXA1. F. Significant overlap (47%) was observed between NKX2-1 occupancy and FOXA1 occupancy identified by ChIP-seq with the lysates from NCI-H3122 cells. Supplemental materials and methods Expression analysis.

Cross-tumor-type differential expression analysis was performed on Oncomine to identify differential expression of LMO3 across individual tumor types, and to compare the results between adenocarcinomas and squamous cell carcinomas of the lung. For differential expression analysis on the effects of NKX2-1 knock down in NCI-H2009 cells, mrnas from three replicates of NCI-H2009 with two shrna against NKX2-1 or duplicates of parental NCI-H2009 and with shrna against GFP were extracted and hybridized to Affymetrix Gene ST 1.0 microarray. Differential gene expression comparing shnkx2-1 vs. shgfp was analyzed with use of the limma package in R/Bioconductor. Data are publicly available at GEO (accession no. GSE40270). Signature cluster comparison Hypergeometric p-value and proportion (k/k) of overlap was calculated between 260 genes identified as positively correlated with NKX2-1 amplification and each geneset of 100 cluster signatures in Director s challenge consortium data. DNA content analysis NCI-H3122 cells with shrna against LMO3 or LacZ were harvested after 72 hours of selection with puromycin and fixed with 80% methanol. Fixed cells were stained with 69uM propidium iodine (Molecular Probes) in 0.5mg/ml RNase A containing buffer (3.8mM sodium citrate) for 1 hour at 37C and subject to DNA content analysis with flow cytometry (FACS LSRFortessa, BD Biosciences). DNA contents were analyzed with ModFit LT software (Verity Software).

In situ hybridization Adult mice lungs and trachea were fixed 2 hours in 4% (wt/vol) paraformaldehyde in PBS at 4 C and processed for frozen sections as per established protocols. Mouse Lmo3 cdna was amplified using PCR primers (Table S4) that also contain T7 at the 3 - end. This PCR product was then used to generate specific anti-sense probe using a DIG labeling kit (Roche; Basel, Switzerland). Non-radioactive in situ hybridization was performed using standard protocols. Briefly, 12µm thick frozen sections were washed with RNase-free PBS and acetylated in freshly prepared 0.25% (v/v) acetic anhydride in 0.1 M triethanolamine for 10 min by vigorous shaking at room temperature. Then sections were washed twice in PBS and incubated for 30 min in 0.1% (v/v) Triton X-100 and 0.1 M PBS, ph 7.0. After rinsed in PBS and pre-hybridized for 30 min at room temperature, sections were hybridized with hybridization buffer containing 200ng/mL antisense DIG-labeled riboprobe and incubated at 65 o C for over night (12-16 hours). The slides were incubated in 0.2x saline sodium citrate (SSC) at 65 o C for 2 hours. The slides cooled down and washed once with Tris-buffered saline containing 0.1% Tween 20 (TBST), were then blocked with 20% sheep serum in TBST for 1 hour followed by incubation with anti-dig antibody conjugated to alkaline peroxidase (1:2500 dilution) for 1 hour at room temperature. After two washes in TBST and three washes with alkaline phosphatase activation buffer (100mM NaCl, 100mM Tris-HCl (ph-9.5), 50mM MgCl 2 and 0.1% Tween-20), the slides were stained with BM Purple (Roche cat. #11442074001) at room temperature. Luciferase reporter assay

The NKX2-1-binding locus at LMO3 was PCR-amplified from human genomic DNA, and cloned into the pgl4.10 vectors (Promega) by DNA ligation. All constructs were verified by DNA sequencing. Fifty nanogram of reporter plasmid along with 50 ng of Renilla control vector (prl-tk-renilla) was transfected in quadruplicate into 293T cells in a 96-well plate with the use of FuGENE6. After 24 hours, firefly and Renilla luciferase activities were determined using the Dual-Luciferase Reporter Assay System (Promega) on a SpectroMax5 plate reader. Firefly luciferase activities of individual transfections were normalized against Renilla luciferase activities.