Supplementary figure legends

Supplementary figure legends
Supplementary Figure 1 – DNA sequence of DICER1 exon 23 from blood and tumor specimens using discriminating primers across germline mutant sequence (c.4407_4410delTTCT). Tumor DNA shows c.5125G>A (p.D1709N) in wild-type allele (germline mutant-negative allele).
Supplementary Figure 2 – Expression levels of miRNAs with at least three-fold difference in median expression between PPB and normal lung tissues from DICER1 germline carriers. Hierarchical clustering suggests a depletion of 5’-derived miRNAs relative to 3’-derived miRNAs in PPB tumors with DICER somatic missense mutations in addition to germline loss-of-function variants.
Supplementary Figure 3 – qRT-PCR from DICER1 mutant tumors demonstrating decreased expression of 5p-microRNA in PPB (type II and III) patient tissues. Five members of let-7 family including miR-98, plus miR-34a and c shown.
Supplementary Figure 4 – Distribution of sequence coverage across each miRNA, normalized for the length of each miRNA primary transcript with at least one sequencing read aligned. Columns correspond to different subsets of expressed miRNAs. All miRNAs: All miRNAs with a 5p, 3p, or 5p and 3p arms annotated in miRbase build 19. Two arms: Only miRNAs with 5p and 3p arms annotated. One arm: miRNAs with a single arm annotated in the first 50% (5p) or latter 50% (3p) of the primary transcript. Each row corresponds to a different sample subject to miRNA sequencing. Normal: Reads from normal fibroblasts mapped predominantly to mature 5p and 3p miRNA transcripts resulting in a characteristic “valley” of lower coverage of cleaved and degraded hairpin sequence. PPB_14 tumor: Coverage profiles from primary PPB tissue have increased coverage of hairpin sequences. “One arm” 5p miRNAs have increased coverage beyond the boundary of annotated 5p miRNA region whereas profiles from one-arm 3p miRNAs are more similar to normal tissue. PPB_14 cell line: In the presumably clonal tumor cell line, hairpin sequences are retained to a greater degree than the heterogeneous tumor tissue specimen. This suggests the small hairpin “valley” seen in PPB_14 tumor may be due to infiltrating normal cells. As with the tumor sample, the retention appears to be specific to 5p-derived miRNAs as one-arm 3p miRNAs do not have significant coverage beyond the annotated boundary of mature transcripts. miRNA arm pos: Line segments depicting the location of each annotated mature transcript mapped to the percent location along each miRNA primary transcript. Transcripts with both 5p and 3p arms annotated are marked in black and red. Transcripts with only a single annotated mature transcript are marked in blue. Y-axis is a count miRNAs in each miRNA subset.
Supplementary Figure 5 – Photomicrographs of PPB tumors stained with anti-p53 antibodies shows diffuse positivity, subclonal positivity or negative staining. Results compared with mutation analysis performed on DNA extracted from a 5mm core from the same FFPE tumor block. Anti-p53 rabbit monoclonal with hematoxylin counterstain; original magnification x 200 A, C and D; x 400 B. A) PPB_5 tumor shows strong nuclear positivity in the majority of tumor cells. Exome sequence analysis identified an exon 5 c.481G>A; p.A161T missense mutation in TP53 predicted to be deleterious by SIFT; B) PPB_16 tumor is negative for p53. Ion Torrent sequence analysis identified an exon 5 c.513-514 insA frameshift mutation in TP53; C) PPB_11 tumor was also negative for p53. This tumor had an exon 6 c. 642_643delTA frameshift mutation and deletion of the wild-type allele identified by exome sequencing; D) PPB_17 tumor shows diffuse anaplasia with focal areas of cells with nuclear p53 staining approximating 20% of the tumor. Ion Torrent sequencing showed an exon 5 c.481G>A; p.A161T missense mutation supported by 7% of reads. We interpret the immunohistochemistry result as a mixed population of cells with 20% mutant TP53/deletion TP53 and 80% cells with biallelic TP53 deletion with the mutant TP53 population not represented in the core taken for genetic analysis (region of tumor selected by H&E only). Note the cytologic features of the p53 positive nuclei and p53 negative nuclei are identical showing significant anaplasia.
Supplementary Figure 6 – Fraction of read counts with start and end points confined to functional units of 1,595 miRNA primary transcripts. 5p and 3p regions as annotated by miRbase build 19. Hairpin loop regions were defined as the genome segment between 5p and 3p regions. 5p+loop+3p denotes reads that include 5p, loop, and 3p sequence. The “Unannotated” category denotes miRNAs without an explicit 5p or 3p designation. On each page, the three rows correspond to the three samples presented in Figure 4: a normal fibroblast cell line, a PPB primary tumor, and a PPB cell line derived from the PPB primary tumor tissue. miRNAs are ordered by the number of reads in the normal sample.

Supplementary table descriptions
Supplementary Table 1 – Matrix of clinical features, copy number summary, and mutation calls by case.
Supplementary Table 2 – List of annotated somatic mutation calls.
Supplementary Table 3 – Somatic mutation significance table (MutSig v1.5).
Supplementary Table 4 – Summary of somatic DICER1 mutations in discovery and extension cohorts
Supplementary Table 5 – Summary of samples analyzed by quantative RT-PCR to support Supplemental Figure 3
Supplementary Table 6 – Summary of somatic TP53 mutations and immunohistochemistry patterns in 27 tumors
Supplementary Table 7 – Broad somatic copy number variant significance table (GISTIC2).
Supplementary Table 8 – Focal somatic copy number variant significance table (GISTIC2).
Supplementary Table 9 - Counts of reads with start and end points confined to functional units of 1,595 miRNA primary transcripts across a normal fibroblast cell line, a PPB primary tumor, and a PPB cell line derived from the PPB primary tumor tissue. These are the raw data that underlie panels B-D of Figure 4 and Supplementary Figure 6.