Supplementary Information:

T>A/A>T

T>G/A>C

T>C/A>G

C>A/G>T Mutations Type Mutations

C>G/G>C

C>T/G>A

0 2000 4000 6000 8000 10000 12000 14000 Number of Mutations

Supplementary Figure 1. Mutation spectra of single substitutions in melanoma whole exome sequencing. The number of each of the six classes of base substitutions resulting in non-synonymous changes in the whole-exome screen is shown.

1 55N 75N 79N

55T 75T 79T

C51T (F17F) C51T (F17F) C51T (F17F)

81N Ref Ref

81T MB490 MB1110

C51T (F17F) C51T (F17F) C51T (F17F)

Supplementary Figure 2. Detection of BCL2L12 recurrent mutation. Representative examples of the hotspot mutation in BCL2L12. In each case, the top sequence chromatogram was obtained from normal tissue and the lower sequence chromatogram from the indicated tumors. Arrows indicate the location of mutations. The nucleotide and amino acid alterations are indicated below the chromatograms.

2

Supplementary Figure 3. Single nucleotide-resolution evolutionary conservation scores as defined by the GERP algorithm run on an alignment of 35 mammalian species. The nucleotide at position 51 in the coding sequence of BCL2L12 has a GERP score of 2.54 (horizontal blue line), which is above the 2.0 threshold to be considered evolutionarily constrained. This level of constraint is clearer stronger than fourfold degenerate (4D) sites (gray box plot; 23,662 sites total) in the 1 Mb neighborhood and larger than the majority of twofold degenerate (2D) sites (yellow box plot; 20,359 sites total). 4D sites represent nucleotide positions in coding regions where the observed base can change to any other base and the resulting amino acid is not changed; such positions are generally considered to be neutrally evolving. We use nearby 4D and 2D sites as a way to compare the level of evolutionary constraint observed at BCL2L12 position 51.

A

3

B i ii

7

vector BCL2L12 (WT) BCL2L12 (F17F) BCL2L12 ** 6 + + + GFP 5 a-BCL2L12 4

3 *

a-GFP

2 (ratio) Bcl2L12:GFP 1 a-Tubulin 0 Vector WT F17 F Supplementary Figure 4. A qRT-PCR analysis of transiently expressed wild- type BCL2L12 or mutant (F17F) BCL2L12 or empty vector as control in HEK293T cells using BCL2L12- or GAPDH-specific primers. Error bars, sd. (n=3). (* comparison of WT to F17F; student’s t-test). B i. Western blot analysis of co-transfected HEK293T cells with GFP and BCL2L12 (wild-type or F17F mutant). ii. Experimental triplicates were analyzed and graphed using Microsoft Excel. Error bars, sd (n=3). (* p<0.05 comparison of WT or F17F to vector, ** p<0.03 comparison of F17F to WT; student’s t-test)

4

Supplementary Figure 5. Effects of the BCL2L12 (C51T; F17F) recurrent mutation on translation. Autoradiogram of SDS gel electrophoresis of wild-type and mutant BCL2L12 cell-free translation products. Arrows point the position of molecular weight markers and the full-length BCL2L12 product, respectively. In vitro translation was done in the presence of [35S]-methionine in cell-free system prepared from 75T melanoma cells using equal amounts of wild-type and mutant BCL2L12 (in vitro transcribed and capped) mRNAs as described in Materials and Methods section. Note, that there is an unspecific labeled ~48 kDa product present also in no mRNA control.

5 A

51

GLI1

DEAF1

ETS1

STAT1

STAT3

ELK1

ELK4

6 B

400bp STAT3 300bp 200bp 100bp

DEAF1 400bp 300bp 200bp 100bp

400bp 300bp STAT1 200bp 100bp

GAPDH 400bp 300bp 200bp

100bp

79T 81T MB490 75T

C

H3K4me3 ChIP BCL2L12 Sample BRAF STAT1 Chr5 (+) Chr5 (-) Sample H3K4me3 Deaf1 Stat1 Stat3 GFP 79T 10.09 4.7 2.6 0.01 79T 6.093 0.003 0.013 0.022 0.010 75T 18.7 39.9 6.1 0.04 75T 10.236 0.007 0.011 0.018 0.008 Skmel28 3.86 6.33 1 0.01 Skmel28 6.548 0.014 0.013 0.036 0.023

D

7 MALDI-TOF Analysis of ChIP DNA for allelic imbalance

Primer F Primer X Primer R Input DNA Input

BCL2L12 melanoma SNP (C/T alleles)

1) PCR amplification using Primer F and R DNA 2) Single-base extension of Primer X

3) Quantitative MALDI-TOF to determine percentage C or ChIP T allele representation in input and ChIP DNA samples

E

90 79T (C/T) 80 75 T (C/T) 70 Skmel28 (C/C) 60

50

40 % T allele T % 30

20 10 0 gDNA Input H3K4me3 Deaf1 Stat1 Stat3 Neg (GFP)

ChIP epitope Supplementary Figure 6. BCL2L12 C-to-T mutation does not dramatically alter histone modification or factor binding in vivo. A. The UCSC genome browser screen shot is centered on ORF position 51 and includes hg19 coordinates chr19:50,169,116-50,169,146 (31 bps). Transcription factors found to have PWMs with strong sites corresponding to the ORF position 51 are indicated at right (black). None of these PWMs pass the detection threshold with the mutant “T” allele at ORF position 51, which, however, enables the detection of the PWM for GLI1 transcription factor (indicated at left, red). B. STAT3, DEAF1, and STAT1, mRNA expression in melanoma. STAT1, STAT3 and DEAF1 expression was evaluated by reverse transcription of total RNA followed by PCR amplification. Expression analysis of GAPDH was performed as a control. Melanoma tumors analyzed are indicated above. Expected band sizes are 192 bp, 171 bp and 169 bp respectively. C. qPCR analysis of H3K4me3 or Deaf1, Stat1, and Stat3 enrichment in chromatin immunopreciptation (ChIP) assays for two C/T heterozygous mutant (79T, 75T) or C/C homozygous (Skmel28) wild-type melanoma cell culture lines. The left panel indicates enrichment or depletion of H3K4me3 in control regions. Chr5(+) = promoter region of SLC22A5 and serves as another positive control; Chr5(-) = negative control region located near SLC22A5. The right panel indicates enrichment or depletion of the listed factors in the BCL2L12 region containing the C-to-T mutation. Enrichment/depletion is expressed relative to input. D. Schematic of MALDI-TOF (Sequenom) assay design and procedure for assessing allelic representation in input and ChIP DNA samples. The arrow under the right panels indicates the mass of the unextended extension primer (Primer X). The solid

8 and open arrowheads indicate the peak mass for the primer that has incorporated a “C” allele or “T” allele at the SNP position, respectively. Areas under each curve are measured to determine the relative amount of each allele. E. Plots of %T allele for gDNA, input chromatin, or ChIP samples (H3K4me3, Deaf1, Stat1, Stat3, negative control GFP) for three melanoma cell culture samples (79T, 75T, and Skmel28). Genotypes of each sample at the BCL2L12 SNP are indicated in parentheses. Neg=negative

9 A

B BCL2L12 Mutant BCL2L12 Wild Type

TIAL1

GAPDH

75T 79T

MB490 81T

SKMel28

55T

A375 17T 12T 32T

10 C

Supplementary Figure 7. Potential association of BCL2L12 mRNA with mRNA binding . A. Prediction of RNA binding sites using SpliceAid server/database (a database of strictly experimentally assessed target RNA sequences in humans, http://www.introni.it/splicing.html). B. TIAL1 expression was evaluated by reverse transcription of total RNA followed by PCR amplification. mRNA expression analysis of GAPDH was performed as a control. C. Gel shift analysis of potential TIAL1 - BCL2L12 RNA complexes (left panel). Gel shift analysis of RNA- complex formation between TIAL1 and intron 4 enhancer element in Calcitonin/CGRP (positive control).

11 A

6000

5000

levels) 5p

- 4000 671

- 3000

2000

1000

(miR change Fold 0 NC-0 miR-0 miR-50 NC-0 miR-0 miR-50

12T SK-Mel-28

Wild type B

16000

14000

12000 5p levels) 5p -

10000 671 - 8000

6000 4000

2000 (miR change Fold 0 NC-0 miR-0 miR-50 NC-0 miR-0 miR-50

75T 79T Mutant

Supplementary Figure 8. Anti-miR-671-5p sequestration of miR671-5p mimic. A-B. Co-transfection of anti-miR-671-5p suppresses the levels of miR- 671-5p mimic in wild-type A. BCL2L12 melanoma cells (12T and SK-Mel-28) and mutant B. BCL2L12 melanoma cell lines (75T and 79T). Graphs show qRT-PCR

12 of mature miR-671-5p mimic levels in wild-type BCL2L12 (12T and SK-Mel-28) and mutant BCL2L12 (75T and 79T) cell lines in the presence of negative control miR (NC) or hsa-miR-671-5p (miR) plus 0nM or 50nM anti-miR-671-5p. Results are representative of two independent experiments. Error bars, sd.

13

FLAG (F17F) FLAG

(WT) FLAG

- -

BCL2L12 BCL2L12 vector

a-p53

FLAG IP IgG (Heavy chain) a-BCL2L12

a-p53

Lysates a-BCL2L12

a-Tubulin

Supplementary Figure 9. Wild-type and F17F protein made from mutant BCL2L12 transcript interact with p53. Co-immunoprecipitation analysis of BCL2L12 – p53 interaction. HEK293 cells exposed to 50kJ UV with either wild- type BCL2L12-FLAG, protein made from mutant BCL2L12 transcript (F17F) BCL2L12-FLAG constructs and/or empty vector were used for immunoprecipitation with anti-FLAG (M2) beads and subsequent immunoblot analysis with the indicated antibodies; alpha-tubulin was used as a loading control.

14

75T (F17F-BCL2L12) A B 1.2

* *

1.0

sh91 sh92 pLK0.1 0.8 Vector alone Vector + + + BCL2L12 (WT)

a-BCL2L12 0.6

a-Tubulin 0.4

0.2

Fold (transcript changelevels) 0 pLK0.1 sh91 sh92 12T (WT-BCL2L12) SK-Mel-28 (WT-BCL2L12) 1.2 1.2

1.0 * * 1.0 levels)

* *

0.8 0.8 0.6

transcript 0.6 (transcript levels) (transcript

0.4 0.4

change 0.2 0.2 Fold change ( change Fold Fold 0 0 pLK0.1 sh92 sh91 sh92 pLK0.1 sh91

Supplementary Figure 10. Targeted stable depletion of BCL2L12 in melanoma cells. A. shRNA mediated depletion of BCL2L12 was tested using transient transfection and immunoblotting of lysates from HEK293T cells. Lysates from HEK293T transiently transfected with BCL2L12 and either one of two BCL2L12-specific shRNAs or empty vector were immunoblotted using the indicated antibodies to show specificity. B. qRT-PCR analysis of the wild-type BCL2L12 depleted stable pooled cell line (12T) and the mutant (F17F) BCL2L12- depleted stable pooled cell line (75T) using BCL2L12- or GAPDH-specific primers. Error bars, sd.

15 A BCL2L12 (WT)

#2 siRNA #2 #3 siRNA#3 kDa siRNA NC

37 BCL2L12 25

Tubulin

B

1.20

1.00

0.80

0.60

0.40

0.20

levels) (transcript change Fold 0.00 12T SKMel28 75T 79T 55T C025 NC #3 siRNA WT F17F #4 siRNA

Supplementary Figure 11. Depletion of mutant BCL2L12 sensitizes melanoma cells to UV induced cell death. Depletion of mutant BCL2L12 sensitizes melanoma cells to UV induced cell death. A. Transient transfection of siRNA knocksdown exogenously expressed BCL2L12. HEK293T cell lysates co- transfected with BCL2L12 and siRNA were immunoblotted with anti-BCL2L12 and anti-GAPDH as loading control. B. Melanoma cells were depleted of endogenous BCL2L12 using two specific siRNA after transient transfection. Graphs are representative of transient experiments tested for depletion using qRT-PCR analysis in two wild-type BCL2L12 cell lines (12T and SK-Mel-28) or mutant BCL2L12 (75T, 79T, 55T and C025) cell lines using BCL2L12 or GAPDH –specific primers. Error bars, sd.

16

Supplementary Table 1. Splicing potential predictions of BCL2L12

Position of Distance Score with Score with Score splice signal from reference the C51T change (on chr19) mutation allele mutation

54860873 -70 -1.00515 -1.02849 -0.02334 54860912 -31 -2.45644 -2.50231 -0.04587 54860938 -5 -2.21513 -2.47638 -0.26125 54860965 22 -2.71677 -2.85275 -0.13598 54860983 40 -2.29851 -2.36195 -0.06344 54860994 51 -1.86662 -1.95038 -0.08376 54861019 76 -1.59377 -1.52633 0.06744 54861136 193 0.177993 N/A N/A

Supplementary Table 2. Characteristics of melanoma patient samples used in this study

17 Sample Patient Age (years)* Patient Gender Characteristic Primary Tumor Site Metastatic Tumor Site Matched Normal Source 1T† 29 F cutaneous Posterior neck Lung Blood 2T 30 M cutaneous Upper back Pectoral, subcutaneous Blood 3T 18 M cutaneous Scalp Forehead, subcutaneous Blood 4T 33 F cutaneous Arm Supraclavicular, soft tissue Blood 5T† 47 M cutaneous Lower abdomen Iliac Blood 6T 42 M cutaneous Temple Neck, soft tissue Blood 7T† 53 M cutaneous Knee Stomach Blood 8T† 61 M cutaneous Thigh Inguinal Blood 9T† 62 M cutaneous Mid Back Back, subcutaneous Blood 10T 55 M cutaneous Back Axilla Blood 12T† 53 M cutaneous Arm Upper arm, subcutaneous Blood 14T† 58 F cutaneous Foot Small Bowel Blood 15T 39 M cutaneous*** Heel Thigh, subcutaneous Blood 16T 62 M cutaneous Scalp Lung Blood 17T† 33 M cutaneous Unknown Shoulder, subcutaneous Blood 18T† 55 M cutaneous Upper back Clavicle, soft tissue Blood 19T 49 M cutaneous Wrist Scapula, subcutaneous Blood 20T 58 F cutaneous Shoulder Axilla Blood 21T 59 M cutaneous Unknown Omentum Blood 22T† 51 M cutaneous Nipple Chest wall, subcutaneous Blood 23T† 44 M cutaneous Scalp Lung Blood 24T† 49 M cutaneous Upper arm Axilla Blood 26T† 48 F cutaneous Upper thigh Lung Blood 28T 28 F cutaneous Vulva Iliac Blood 29T 51 M cutaneous Thigh Inguinal Blood 30T 53 F cutaneous Upper back Lung Blood 31T 49 F cutaneous Knee Thigh, subcutaneous Blood 32T† 58 M cutaneous Shoulder Omentum Blood 33T 33 M cutaneous Temple Chest wall, subcutaneous, & pleura ** Blood 34T 31 M cutaneous Mid back Shoulder, subcutaneous Blood 35T† 23 F cutaneous*** Heel Thigh, subcutaneous Blood 36T 25 M cutaneous Shoulder Thigh, subcutaneous Blood 37T 38 F cutaneous Neck Omentum Blood 38T 27 M cutaneous Scalp Skull versus Dura Blood 39T 56 M cutaneous Chin Mesentery Blood 41T 45 M cutaneous Scalp Neck, soft tissue Blood 43T† 19 F cutaneous Thigh Popliteal soft tissue Blood 44T 56 M cutaneous Mid back Lung Blood 45T 48 M cutaneous Peri-umbilicus Mediastinum Blood 47T 42 F cutaneous Shoulder Abdomen, subcutaneous Blood 48T 28 M cutaneous Anterior chest wall Back, soft tissue Blood 49T 43 M cutaneous Lower back Thigh, subcutaneous Blood 50T 49 F cutaneous Unknown Inguinal Blood 51T† 50 F cutaneous Medial thigh Adnexa Blood 52T 39 F cutaneous Shoulder Lung Blood 53T 48 F cutaneous Calf Breast Blood 55T† 60 M cutaneous Neck Lung Blood 56T† 52 M cutaneous Thigh Lung Blood 58T 46 F cutaneous Shoulder Hip, subcutaneous Blood

Supplementary Table 2 continued. Characteristics of melanoma patient samples used in this study

18 Sample Patient Age (years)* Patient Gender Characteristic Primary Tumor Site Metastatic Tumor Site Matched Normal Source 59T 64 F cutaneous Back Abdomen, subcutaneous Blood 60T† 46 M cutaneous Abdomen Flank, subcutaneous Blood 62T 58 F cutaneous Toe Thigh, subcutaneous Blood 63T 30 M cutaneous Mandible Small Bowel Blood 64T† 32 F cutaneous Unknown Ovary Blood 67T 29 M cutaneous Scapula Back, subcutaneous Blood 68T 49 M cutaneous Knee Lung Blood 69T 36 M cutaneous Thigh Axilla Blood 71T 67 M cutaneous Anterior shoulder Lung Blood 72T 53 M cutaneous Mid back Liver Blood 73T 45 F cutaneous Scapula Breast Blood 74T 40 F cutaneous Leg Lower extremity, subcutaneous Blood 75T 54 F cutaneous Arm Upper arm, subcutaneous Blood 76T 40 M cutaneous Scalp Neck, soft tissue Blood 77T 39 M cutaneous Posterior shoulder Lung Blood 78T 27 F cutaneous Back Lung Blood 79T 53 M cutaneous Mid back Supraclavicular, soft tissue Blood 80T 36 M cutaneous Calf Popliteal Blood 81T† 60 F cutaneous Arm Upper arm, subcutaneous Blood 82T 48 M cutaneous Scapula Axilla Blood 83T 33 F cutaneous Arm Back, subcutaneous Blood 84T 60 F cutaneous Calf Thigh, subcutaneous Blood 85T 44 M cutaneous Anterior chest wall Chest wall, subcutaneous Blood 86T 42 F cutaneous Forearm Liver Blood 87T 27 M cutaneous Upper arm Small bowel & mesentary ** Blood 88T† 37 F cutaneous Scalp Chest wall, subcutaneous Blood 90T 19 M cutaneous Occipital scalp Neck, soft tissue Blood 91T† 55 F cutaneous Shoulder Subcostal soft tissue Blood 92T 37 F cutaneous Inguinal Femur Blood 93T† 42 F cutaneous Finger Axilla Blood 94T 44 M cutaneous Unknown Adrenal gland Blood 95T 58 F cutaneous Unknown Inguinal Blood 96T† 49 M cutaneous Unknown Inguinal Blood 99T 57 M cutaneous Back Liver Blood 100T 28 M cutaneous Back Chest wall, soft tissue Blood 101T 58 M cutaneous Unknown Omentum Blood 103T 35 F cutaneous Shoulder Axilla Blood 104T 56 M cutaneous Ankle Thigh, subcutaneous Blood 105T 28 M cutaneous Upper back Neck, soft tissue Blood 106T 41 F cutaneous Calf Lung Blood 108T† 25 F cutaneous Heel Thigh, subcutaneous Blood 109T 58 M cutaneous Shoulder Scrotum Blood 110T 51 M cutaneous Unknown Axilla Blood 111T 41 M cutaneous Mid upper back Axilla Blood 112T 46 M cutaneous Lower back Inguinal Blood 113T 38 M cutaneous Posterior shoulder Axilla Blood 114T 22 M cutaneous Unknown Adrenal gland Blood 115T 41 M cutaneous Back Brain Blood 116T 29 M cutaneous Leg Thigh, subcutaneous Blood

Supplementary Table 2 continued. Characteristics of melanoma patient samples used in this study

19 Sample Patient Age (years)* Patient Gender Characteristic Primary Tumor Site Metastatic Tumor Site Matched Normal Source 117T 44 M cutaneous Posterior lower leg Chest wall, subcutaneous Blood 119T 45 M cutaneous Chest Axilla Blood 120T 56 M cutaneous Abdomen Lung Blood 122T 60 F cutaneous Back Lung Blood 123T 51 M cutaneous Unknown Anticubital Blood 124T 44 M cutaneous Back Inguinal Blood 125T 27 M cutaneous Upper back Axilla Blood A11 57 M unknown Unknown primary LN NA B01x 49 M cutaneous R-superior patella LN Blood B06 34 M cutaneous L-upper chest wall LN NA C14 65 M cutaneous*** L-heel LN met NA C16 69 M cutaneous Back LN met NA C19 56 M cutaneous Unknown primary Soft Tissue/Skin NA C22 56 M cutaneous R-arm LN NA C30x 51 M cutaneous R-upper back Soft Tissue/Skin NA C31x 83 F cutaneous*** Left foot LN NA D03x 68 F cutaneous Rectum LN NA D13 70 F cutaneous R-heel LN NA D16 68 M cutaneous*** R-foot LN Blood D22 77 M unknown Unknown primary LN NA D23 69 M cutaneous*** R-foot (plantar surface) LN NA MD_04 61 F cutaneous R-arm LN NA MD_09 27 M cutaneous*** L-foot Soft Tissue NA MD_13 62 F unknown Unknown primary Small Intestine NA MD_14 91 F cutaneous R-5th toe LN NA MD_15 28 F cutaneous R-buttock LN Blood MD_22 58 M unknown Unknown primary Brain Blood MD_35 38 F cutaneous R-calf Liver NA MD_37 53 F ocular R-eye Liver NA MD_40 74 F cutaneous R-arm Liver NA MB532 44 F unknown Unknown Lymph Node Blood MB669 59 M cutaneous Anterior Chest Bowel Small Blood MB930-T 64 M cutaneous L-Shoulder Bone Spine Blood MB1089 56 F cutaneous L-Calf Liver Blood MB1160 61 F cutaneous R-Upper Back Lymph Node Blood MB1245 41 M cutaneous R-Posterior Knee Bowel Small Blood MB1287 83 M cutaneous Forehead Subcutaneous Neck Blood MB1320 57 F cutaneous*** Vulvo-Vag Lymph Node Blood MB706 63 M cutaneous Anterior Chest Subcutaneous Back Blood MB929 69 F cutaneous L-Arm Subcutaneous Arm Blood MB1029 55 M cutaneous R-ankle Primary Skin Blood MB1046-LN 70 M cutaneous*** R-ankle Lymph Node Blood MB1067 47 F cutaneous L-Lower Leg Lymph Node Blood MB1082 61 F cutaneous L-Lateral Ankle Subcutaneous Groin Blood MB1110 61 F unknown Unknown Subcutaneous Arm Blood MB1113 63 F cutaneous*** R-Nasal Cavity Muscle Back Blood MB1157 71 F cutaneous Scalp Primary Skin Blood MB1222 25 M cutaneous L-Upper Back Lymph Node Blood MB87 53 F cutaneous L-Neck Brain Blood

20 Supplementary Table 2 continued. Characteristics of melanoma patient samples used in this study

Sample Patient Age (years)* Patient Gender Characteristic Primary Tumor Site Metastatic Tumor Site Matched Normal Source MB104 65 M cutaneous Posterior Waist Subcutaneous Back Blood MB106 67 M cutaneous Upper Back Bone Spine Blood MB107 47 M cutaneous Unknown Brain Blood MB161 72 F cutaneous*** Vulva Primary Vulvar Vag Blood MB298 53 F cutaneous L-Neck Brain Blood MB327 71 F cutaneous*** R-Vulva Primary Vulvar Vag Blood MB363 69 M cutaneous Posterior Neck Subcutaneous Back Blood MB393 68 M cutaneous R-Ear Brain Blood MB402 52 M unknown Unknown Lymph Node Blood MB404 67 M cutaneous L-Lower Back Muscle Gluteal Blood MB463 52 M cutaneous Rt Buttock Primary Skin Blood MB464 37 F cutaneous Rt Knee Primary Skin Blood MB487 73 M cutaneous L-Upper Arm Lymph Node Blood MB490 65 F cutaneous L-Upper Tragus/Ear Brain Blood MB505 75 F cutaneous R-Posterior Upper Arm Subcutaneous Thigh Blood MB522 52 M unknown Unknown Bowel Small Blood MB1227 41 M unknown Unknown Subcutaneous Trunk Blood MB1335 61 M cutaneous R-Temple Primary Skin Blood MB1337 41 M unknown Unknown Subcutaneous Thigh Blood 130T† 49 F Cutaneous Lower extremity Thigh, subcutaneous Blood 131T 63 M Cutaneous Mid back Thigh & Back, subcutaneous ** Blood 25T 36 M Cutaneous Lower extremity Iliac Blood Colo-829† 43 M Cutaneous Unknown Unknown Blood MB481-Met 47 F Unknown Left Medial Thigh Lymph Node Normal Tissue MB658-LN 56 M Spindle Cell Left Bicep Wide spread metastasis Blood B03 62 F cutaneous R-5th toe LN NA B05 47 F cutaneous R-forearm LN NA B07 77 F cutaneous R-upper arm LN NA B09 55 M cutaneous R-flank Soft Tissue NA B12 72 M cutaneous R-neck LN NA B19 64 M cutaneous R-foot LN/Soft Tissue NA C18 61 F cutaneous L-heel Soft Tissue/Skin NA C29 65 F cutaneous R-arm Soft Tissue/Skin NA MD_16 45 F cutaneous Back Lung NA B08 80 F Acral Lentiginous Rt great toe Inguinal LN NA B13 29 M Melanoma of soft parts Foot Popliteal LN NA B17 53 M Cutaneous Left arm Axillary LN NA C20 41 M Cutaneous Back Inguinal LN NA C23 74 M Cutaneous Abdomen Axillary LN NA C28 50 M Cutaneous Leg Soft tissue NA D14 83 M Mucosal Anus Inguinal LN NA D21 55 F Cutaneous Foot Inguinal LN NA D24 52 F Cutaneous Shoulder Axillary LN NA MD_08 47 M Cutaneous Head Skin NA MD_11 64 M Cutaneous Abdomen Lung NA MD_19 80 M Cutaneous Ear Lung NA MD_28 47 M Cutaneous Back Axillary LN NA MD_34 75 F Mucosal Rectum Lung NA MD_39 54 F Cutaneous Head Liver NA MD_43 52 F Uveal Eye Liver NA

21 Supplementary Table 2 continued. Characteristics of melanoma patient samples used in this study

Sample Patient Age (years)* Patient Gender Characteristic Primary Tumor Site Metastatic Tumor Site Matched Normal Source A02^ 32 F NA R posterior upper arm subcutaneous L axilla LCL A04^ 24 F NA R shin retroperitenial lymph nodes LCL A06^ 39 F NA L posterior upper back (shoulder blade) stomach internal edge, small bowel lymph nodes LCL A11^ 58 M nodular R calf subcutaneous L lower abdominal wall LCL A12^ 40 F Acral base of toe L foot Lower Back, R Axillary, Upper back lymph nodes LCL A13^ 19 M NA L loin L intestine and bowel lymph nodes LCL A15^ 47 F NA R lower back L femoral lymph node LCL C001^ 50 F superficial spreading melanoma R calf R calf nodule LCL C002^ 60 M nodular R upper leg R thigh LCL C004^ 42 F NA NA L axilla lymph nodes LCL C006^ 35 F nodular R lower back R inguinal lymph nodes LCL C011^ 34 F superficial spreading melanoma L upper leg lung and liver LCL C013^ 71 M nodular upper back L and R axilla lymph nodes LCL C016^ 50 F cutaneous post L thigh chest & neck LCL C017^ 57 F cutaneous R mid back R axilla lymph nodes LCL C021^ 38 M superficial spreading melanoma L upper back L axilla lymph nodes LCL C022^ 45 F superficial spreading melanoma R arm R axillary lymph nodes LCL C024^ 63 M NA NA NA LCL C025^ 41 M superficial spreading melanoma L upper back Lymph node LCL C027^ 61 M nodular L upper back L axilla lymph nodes LCL C037^ 27 F NA NA R axilla lymph nodes LCL C044^ 51 F superficial spreading melanoma L lower back L groin LCL C045^ 22 F superficial spreading melanoma R shoulder R axilla lymph nodes LCL C052^ 69 M nodular mid back L axilla lymph nodes LCL C054^ 52 F cutaneous R lower leg R groin LCL C055^ 80 M nodular centre back R axilla lymph nodes and Lgroin LCL C057^ 35 F superficial spreading melanoma L upper back L axilla lymph nodes LCL C058^ 39 M NA NA L axilla lymph nodes LCL C060^ 64 F superficial spreading melanoma R upper leg R groin LCL C062^ 67 M NA NA L and R groin LCL C065^ 39 M superficial spreading melanoma L lower back L axilla lymph nodes LCL C067^ 69 M superficial spreading melanoma upper back R axilla lymph nodes LCL C071^ 33 M superficial spreading melanoma R upper back R axilla lymph nodes LCL C074^ 45 F superficial spreading melanoma mid back axillary lymph nodes LCL C077^ 48 M nodular mid upper back L and R axilla lymph nodes LCL C078^ 60 M superficial spreading melanoma L lower back axillary lymph nodes LCL C081^ 54 M superficial spreading melanoma R upper leg L groin LCL C083^ 33 M superficial spreading melanoma R neck neck LCL C084^ 75 M superficial spreading melanoma R upper back R neck LCL C088^ 55 F superficial spreading melanoma R upper abdomen R axillary lymph nodes LCL C089^ 50 F NA NA L axilla lymph nodes LCL C091^ 53 F superficial spreading melanoma L lower leg groin LCL C092^ 51 M NA NA Neck LCL C094^ 20 F superficial spreading melanoma L upper back L axilla lymph nodes LCL C096^ 45 M superficial spreading melanoma R lower leg groin LCL C097^ 43 F nodular L neck neck LCL C100^ 78 M nodular Mid scalp R neck LCL C106^ 52 M superficial spreading melanoma R upper leg R groin LCL D01^ 54 F occult occult R breast lateral wall LCL D03^ 51 M superficial spreading melanoma L mid back L axilla lymph nodes LCL D04^ 37 F nodular L lower leg pancreas head LCL D05^ 39 M NA L Calf adrenal LCL D08^ 47 F nodular L leg below the knee intransit lesion (shin, ankle) LCL D10^ 64 F acral R foot L scapula LCL D11^ 22 M mucosal R buccal mucosa small bowel lymph node LCL D14^ 24 F nodular L inner thigh small bowel LCL D17^ 31 M NA L calf R leg LCL D18^ 46 F cutaneous NA L chest wall and R axilla LCL D20^ 41 F superficial spreading melanoma L upper outer arm mesenteric lymph node LCL D22^ 49 M superficial spreading melanoma R shoulder R shoulder LCL D23^ 30 M NA NA subcutaneous LCL D24^ 52 M NA NA brain, liver LCL D25^ 40 F NA L ear back LCL D26^ 44 M NA NA mesenteric lymph node LCL D28^ 66 M occult occult R neck lymph nodes LCL D29^ 44 M NA NA subcutaneous nodule LCL D32^ 35 M NA NA R ankle LCL D35^ 43 M occult occult cerebral LCL D36^ 56 M NA L deltoid lung, bowel LCL D40^ 55 M NA NA chest LCL D41^ 44 M superficial spreading melanoma L medial calf L groin lymph node LCL D42^ 74 F NA R knee L scapula LCL D49^ 36 F superficial spreading melanoma R upper arm subcutaneous R forearm LCL D51^ NA NA NA left leg groin lymph node LCL D54^ 51 M NA R forearm R groin illiac lymph nodes LCL D55^ NA NA NA NA NA LCL D64^ 49 M NA NA R calf LCL MM383^ 25 M NA NA R. anterior chest LCL MM540^ 33 M cutaneous R shoulder anterior chest LCL MM548^ 44 M superficial spreading melanoma R thigh R groin lymph node LCL MM576^ 40 F Nodular NA R. medial upper arm LCL MM595^ 50 M NA NA Ant Chest Wall LCL MM608^ 57 M nodular lower back L buttock LCL MM622^ NA M spindle cell melanoma L medial knee NA LCL MM636^ 48 M superficial spreading melanoma R upper posterior arm R Axilla LCL MM647^ 54 F NA NA L thigh LCL MM648^ 46 M nodular R scapula Pulmonary and cerebral LCL * Patient's age when tumor was surgically removed. ** Cell line generated from 2 metastatic melanoma tumors mixed after resection during the same operation. *** Acral lentiginous † samples used in Exome/Genome sequencing. L-left, R-right, LN-lymph node, M-male, F-female, LCL-Lymphoblastoid Cell Line ^Samples from Queensland Institute of Medical Research

22 Supplementary Table 3. Whole-genome sequencing coverage and SNV callability statistics for all normal and tumor genomes

Normal genomes sample quality reads meanCoverage genome callable CDS callable 7 1,839,882,543 32.643 0.947 0.900 8 1,956,417,799 41.757 0.955 0.898 12 1,389,806,896 31.611 0.950 0.877 14 2,417,422,009 61.243 0.951 0.880 26 1,479,982,432 36.174 0.944 0.885 32 1,632,411,627 37.191 0.947 0.851 56 1,858,524,576 40.457 0.948 0.867 81 1,715,080,837 44.423 0.948 0.891 88 1,539,520,660 39.561 0.949 0.904 17 2,778,355,970 47.716 0.956 0.889 mean 1,860,740,535 41.278 0.950 0.884 sum 18,607,405,349

Tumor genomes sample quality reads meanCoverage genome callable CDS callable 7 2,073,932,718 31.846 0.938 0.858 8 2,082,330,768 53.256 0.951 0.867 12 1,151,553,766 30.872 0.935 0.857 14 2,147,567,420 43.592 0.949 0.917 26 1,381,025,264 33.251 0.938 0.881 32 1,855,528,200 32.801 0.941 0.859 56 1,484,627,277 30.050 0.937 0.879 81 2,099,716,396 29.741 0.888 0.866 88 1,187,957,638 32.506 0.940 0.891 17 1,588,246,159 29.640 0.929 0.852 mean 1,705,248,561 34.755 0.935 0.873 sum 17,052,485,606

23 Supplementary Table 4. Primers used for synonymous recurrent mutation confirmation

Gene Primer name Primer Sequence PNLIPRP1 PNLIPRP1 HS synon For GTAAAACGACGGCCAGTtacccattgagttgggcagt PNLIPRP1 HS synon Rev aaacacagttgtgcgtgctc OR4C3 OR4C3 HS synon For TTTCTCCTTGTCTTTATAGGCAATAC OR4C3 HS Synon Rev GTAAAACGACGGCCAGTGAGCCATGCAGCACTCATAA OR8J3 OR8J3 HS Synon For GTAAAACGACGGCCAGTGAGCTCCAGATTCCCCTCTT OR8J3 HS synon Rev CCACCACCATGTAGAGCAGA DNAH9 DNAH9 HS synon For GTAAAACGACGGCCAGTGGATTTCAACATCCCCAAGA DNAH9 HS synon Rev tgggtccccagttaagagtg BCL2L12 BCL2L12 HS synon For GTAAAACGACGGCCAGTTCCGGGTAGCTCTCAAACTC BCL2L12 HS synon Rev CGGTCTCCTCCACTGAACTC BCL2L12 Hs primer set2 GTAAAACGACGGCCAGTGCTGGGGCTTTCTTTTTGAT BCL2L12 Hs primer set2 cagcctgctgtgaggtgtag BCL2L12 Hs primer set3 GTAAAACGACGGCCAGTGTCTTGGAGCTCCGGGTAG BCL2L12 Hs primer set3 tatcagagaagcgggactgg DEMI;PKNP DEMI;PKNP HS synon For GTAAAACGACGGCCAGTagcagttaatggtggggaaa DEMI;PKNP HS synon Rev aagcgtccctctggattgtt TTN TTN HS synon For GTAAAACGACGGCCAGTggcatgcccagagaaaagta TTN HS synon Rev AGACTCAGGTTGGCTGTGCT POTED POTED HS synon For GTAAAACGACGGCCAGTgtgagaaggaagcgaccaag POTED HS synon Rev tctttatgttgcccagtcca GTSE1 GTSE1 HS synon For GTAAAACGACGGCCAGTtcactgagccccttccatac GTSE1 HS synon Rev gttctgcccacCTCACTAGG OR5H6 OR5H6 HS synon For GTAAAACGACGGCCAGTGCAACAATGGCATATGATCG OR5h6 HS synon Rev TAAGAGATGAGCCCCACAGG FILIP1 FILIP1 HS synon For GTAAAACGACGGCCAGTAGCTCTTAGGCCCAGTGTGA FILIP1 HS synon Rev GCCCTTTCTCCTGAGTGATG PPP1R3A PPP1R3A HS synon For GTAAAACGACGGCCAGTAAACAGACTCGGATGCCATT PPP1R3A HS synon Rev TGGCCCTAGAGATTTTTCCA COL14A1 COL14A1 HS synon For GTAAAACGACGGCCAGTtcgatgcattgttttgttttg COL14A1 HS synon Rev atgcacatttcagggtcctc OR2T6 OR2T6 HS synon For GTAAAACGACGGCCAGTCCCCATGTACTTCCTCCTCA OR2T6 HS synon Rev ACTCATGGTAATGGGGGTGA FCRL1 FCRl1 syn HS primer set seq GTAAAACGACGGCCAGTagtcacaaactgggctccac FCRl1 syn HS primer set seq cagcaaaactctcccaggac CPT1A CPT1A syn HS primer set seq GTAAAACGACGGCCAGTtcagacccgctacctgctat CPT1A syn HS primer set seq cccactgggtgaacagtctt

24 Supplementary Table 5. Primers used for plasmid construction and RT-PCR

Gene Name Forward Primer Reverse Primer ctattctagagccaccatgggacggcccgctgggctg ttaagcggccgctcagtccaatggcaagttcaagtcc Cloning wild type BCL2L12 pcDNA3.1 Site-directed mutational primer for ctatgcccttttttgggttttcggccagaggcatgctg cagcatgcctctggccgaaaacccaaaaaagggcatag inserting Phe17Phe change ctattctagagccaccatgggacggcccgctgggctgttcccgcccctatgcccttttttg ttaagcggccgctcacttatcgtcgtcatccttgtaatcgtccaatggcaagttcaagtcc BCL2L12-FLAG tagged cloning primers cgagttcagtggaggagacc gcctaaggaaggcagctagg BCL2L12-real time primer set1 aagacacgctgagggtccta cagggagcagggaagacat BCL2L12-real time primer set2 Hdm2-real time primer set tatataccat gatctacagg ctgtctcactaattgctctc STAT1 reverse transcription primer set ggtacgaacttcagcagcttg gaggtcatgaaaacggatgg STAT3 reverse transcription primer set cttgacacacggtacctgga cttgcaggaagcggctatac DEAF1 reverse transcription primer set gggaggctatgagcgagtg acacggtcaccttctccatc

Online Methods

Tumor tissues All DNA samples used in this study were derived from metastases. Samples used for whole-exome capture and prevalence screen were extracted from cell lines established directly from patients’ tumors as described previously (1) DNA subjected to whole- genome sequencing was extracted from OCT embedded specimens as described previously (1). Tissues used for Validation set 1 were fresh frozen melanoma tumors obtained from the University of Colorado Denver Skin Cancer Biorepository, Division of Medical Oncology. Tissue was collected at University of Colorado Hospital, Anschutz Medical Campus, under Institutional Review Board protocols. DNA was isolated from enriched macrodissected tumor isolates as previously described http://www.riedlab.nci.nih.gov. Tissue processing and storage was previously described by Morente et al (2). Tissues used for Validation set 2 of melanomas, were obtained from Optimum Cutting Temperature (OCT)–embedded frozen clinical specimens from the Melanoma Informatics, Tissue Resource, and Pathology Core (MelCore) at The University of Texas MD Anderson Cancer Center under Institutional Review Board- approved protocols. DNA isolation from the tumor-enriched isolates has been described previously (3). Tissue was further collected and cell lines established at Queensland Institute of Medical Research (41 stage III and 46 stage IV (AJCC) early passage metastatic melanoma cell lines). All cell lines were established as described previously (4-6) with informed patient consent under a protocol approved by the Queensland Institute of Medical Research Human Research Ethics Committee. The clinical information associated with the melanoma tumors used in this study is provided in Table S2.

Melanoma tissue processing for Laser Capture Microdissection (LCM) H&E stained sections of fresh frozen melanoma tissues are prepared for initial histologic assessment. Sections are examined by a pathologist for the presence of tumor, estimation of tumor content, presence of inflammation and necrosis. Tissues with less than 70% tumor and/or significant areas of inflammation and necrosis are subjected to LCM.

Laser Capture Microdissection (LCM) Laser capture microdissection (LCM) was performed in the Pathology Core Facility of MSKCC, New York, NY, using the Veritas Microdissection System (Arcturus). The

25 Veritas system combines ultraviolet laser cutting and laser capture using an infrared laser source. Fresh frozen melanoma tissues sectioned between 8 and 10 μm were transferred to PEN membrane slides (MDS Analytical Technologies) and sections were stained by using a modified protocol described previously (7, 8). Briefly, sections were stained with hematoxylin as follows: slides were immersed in 70% ethanol for 10 min followed by sequential dips in nuclease free water, Mayer's hematoxylin solution for 30 sec, nuclease free water, 75% ethanol, 95% ethanol and finally dehydrated in absolute ethanol by 3 changes of 3 min each. Multiple serial sections (10 - 20) of the tissue are used to maximize cell yields. 5,000 to 10,000 cells were harvested in each LCM cap and material from 5-10 caps was pooled together to maximize yields.

DNA Extraction DNA was extracted using DNeasy® Blood and Tissue kit (Qiagen) following manufacturer’s instructions. DNA was eluted in 35 L of elution buffer. DNA measurements were made using ND-1000 UV-Vis spectrophotometer from NanoDrop technologies.

Whole-genome build statistics We generated ~35.6 billion 100 base paired-end reads that pass the Illumina chastity filter and contain 32 or more Q20 or greater Sanger-scaled quality bases for this study. Reads were aligned to the unmasked hg18 version of the using BWA (9) version 0.5.8c with default parameters. After removing molecular duplicate read pairs (read pairs that map to the same position on the reference sequence are likely an artifact of sample preparation) using samtools (10) version 0.1.11 and considering only reads with a mapping quality of Q30 or greater and bases with Sanger-scaled quality of Q20 or greater, we observe an average base coverage of 41.3x and 34.8x for the normal and tumor genomes, respectively. Using stringent calling criteria, we were able to make confident single nucleotide variant (SNV) calls at, on average, 95.0% and 93.5% of the normal and tumor genomes, respectively (Table S3). When considering only coding regions this translates to an average of 88.4% and 87.3% of normal and tumor positions, respectively.

Whole-genome single nucleotide variants For variant calling, only reads with mapping quality of Q30 or greater and bases with quality of Q20 or greater were considered. We used two related algorithms to make single-position genotype calls in the normal and melanoma genomes. For all genomes, we use a Bayesian genotype caller named Most Probable Genotype (MPG) that has been described previously (11). This genotype caller produces accurate calls in regions that satisfy whole-genome coverage and quality parameters as determined by a separate study (12). Namely, the MPG score must be equal or greater than 10 and the MPG score to base Q20 quality-coverage ratio must be equal to or greater than 0.5.

To identify variant positions, we first developed a new algorithm similar to MPG, called Most Probable Variant (MPV). An important distinction between MPG and MPV is that MPV identifies variant positions relative to a reference genotype, while MPG identifies genotypes without an a priori reference assumption.

26

We validated SSNVs by PCR amplifying the regions in the tumor and normal genomes and then Sanger sequencing the products. Of 192 randomly chosen positions (96 in coding regions, and 96 in non-coding regions), we were able to successfully PCR amplify and sequence 181 in both genomes. Of these, we observed evidence for somatic variants concordant with the whole-genome data at 100% of the positions, suggesting that our method is highly specific.

Exome Capture Exome capture was performed using the SureSelect Human All Exon System (Agilent Technologies, Santa Clara, CA). The manufacturer’s protocol for SureSelect Human All Exon System (Illumina Paired-End Sequencing Library Prep), version 1.0.1 was used, with the following modifications: Bioanalyzer steps were either performed using agarose gel or omitted. In the sample preparation step 9, samples were purified using Ampure XP beads (Agencourt/Beckman Coulter Genomics, Danvers, MA) according to the manufacturer’s protocols. In step 12, samples were purified with the QIAquick MinElute kit (Qiagen Inc., Valencia, CA). One column was used for each sample – the four 250 L post-amplification aliquots were pooled, and passed over the column in several spin steps. Samples were eluted in 12 L buffer EB, and quantitated using the Qubit dsDNA BR Assay kit (Invitrogen Corp, Carlsbad, CA). In the post-hybridization amplification step 2, samples were purified with AMPure XP beads as described above. Samples were then eluted in 30 L buffer EB.

Illumina Sequencing Sequencing was performed on the Illumina GAIIx platform with version 4 chemistry and version 4 flowcells according to the manufacturer’s instructions. 76 base paired-end reads were generated.

Exome read mapping and variant analysis Reads were initially aligned using ELAND (Illumina Inc, San Diego, CA). ELAND alignments were used to place reads in bins of about 100 thousand base pairs. Unmapped reads were placed in the bin of the mate pair if the mate was mapped. Cross_match (Phil Green, http://www.phrap.org) was utilized to align the reads assigned to each bin to the corresponding ~5Mb of genomic sequence. Cross_match alignments were converted to the SamTools bam format, and then genotypes were called using bam2mpg (11), http://research.nhgri.nih.gov/software/bam2mpg/). Bam2mpg was used to implement the Most Probable Genotype (MPG) algorithm, a Bayesian based method to determine the probability of each genotype given the data observed at that position. The quality score represents the difference of the log likelihoods of the most and second most probable genotype. The MPG was divided by the coverage at each position to calculate the MPG/coverage ratio. To eliminate common germline mutations from consideration, alterations observed in dbSNP130 or in the 1000 genomes were removed. We further limited the list to those variants above 5% minor allele frequency. Polymorphisms were further removed by examination of the sequence of the gene in genomic DNA from matched normal tissue. On average a mean depth of 115X or greater was achieved resulting in exome builds with at least

27 89% of the exons covered by high quality genotype calls. To eliminate common germline mutations, we removed any potential somatic mutation that was observed in dbSNP130 or in the 1000 genomes project. To determine which of these alterations were somatic, we compared these data to the matched normal tissue. To discriminate true mutations from the possible sequence alterations identified, we applied criteria that we have previously published (13). These refinements gave us 97.9% coverage, a 2.4% false-negative rate and a sensitivity of 81%. Furthermore, these filters removed ~18% of the alterations. Genotypes were annotated as described previously (14). “Type of Mutation” are: synonymous: changes protein coding region, but not the amino acid; non- synonymous: changes protein coding region, missense variant: changes amino acid; nonsense variant, stop: Introduces a stop codon; DIV-c: in-frame Deletion/Insertion Variant in a coding region and DIV-fs: frameshifting Deletion/Insertion Variant in a coding region. CHASM was used to identify functional mutations (15) (Dataset S1).

Statistical calculation of significance To evaluate whether the frequency of a synonymous mutation is significantly higher than would be expected if the mutation were neutral, we performed a statistical test. We only considered the validation samples to avoid biases. The null hypothesis is that the probability of a mutation at a specific base is the neutral rate of 11.4 mutations/Mb (i.e. p=11.4e-6). We computed a one-sided p-value using the pbinom function in the R statistical software.

Because we evaluated multiple (16) hotspots, this is then corrected for multiple comparisons to arrive at the p-value reported in the text, by using a conservative Bonferroni correction such that the binomial probability is multiplied by the number of hotspots interrogated.

GERP analysis The GERP (Genome Evolutionary Rate Profiling) algorithm, originally described in 2005 (16), identifies evolutionarily constrained positions in multiple genome alignments by quantifying substitution deficits across species. GERP scores are calculated using the reference genome from each species. GERP is a widely used algorithm (16-21) and the scores are available from two big genome browsers: (UCSC) http://genome.ucsc.edu/cgi-bin/hgTrackUi?g=allHg19RS_BW (Ensembl) http://useast.ensembl.org/info/docs/compara/analyses.html#conservation

Previous analyses indicate that disease-causing mutations tend to occur at positions in the genome with high GERP evolutionary constraint scores (22). The designers of the GERP algorithm state that scores above 2 are a strong indication that the position is evolutionarily constrained (see UCSC link above). The position of our BCL2L12 mutation of interest exceeds this threshold.

Splicing prediction analysis To evaluate the strength of donor consensus dinucleotides we used SplicePort (23), which scores sequence features located within 80 bps of the splicing consensus dinucleotide. Specifically, we scanned for the presence of GT dinucleotides within 80

28 bps of the mutation, which we scored with SplicePort independently with the reference and variant alleles (Table S1).

Prediction of protein RNA-binding sites Prediction of RNA binding sites was done using SpliceAid server/database a database of strictly experimentally assessed target RNA sequences in humans, http://www.introni.it/splicing.html (24).

Electrophoretic mobility shift assay (EMSA) Gel shift assays were done following standard protocols. BCL2L12 70 bp long RNA wild-type and mutant fragments harboring the site of mutation were body-labeled with [α-32P]-UTP following in vitro transcription. The RNA fragments were gel-purified and used in electrophoretic mobility shift assays with recombinant TIAL-1 protein available from Novus Biologicals, LLC, (Littleton, CO). A 127 nt long intron 4 enhancer element in Calcitonin/CGRP gene, that has been previously described to bind recombinant TIAL1 (25) was used as a positive control.

Position weight matrix (PWM) analysis Position weight matrix (PWM) information was obtained for vertebrate specific TFs from the TRANSFAC (26) professional (release 11.4) and JASPAR 3 (27) databases. The search for PWM occurrences in the sequences of interest was performed as described in (28). The region containing the C>T variant at BCL2L12 CDS position 51 was searched independently with the reference and variant alleles. Logos of the PWMs were created using Weblogo version 2.8.2 (29) with sequences generated synthetically based on positional frequency information available in the TRANSFAC and JASPAR databases.

PCR, sequencing and mutational analysis of melanoma samples identified to harbor recurrent mutations were confirmed and further screened using two primer sets listed in Table S4 in an additional 153 melanoma samples. Mutational analysis, confirmation and determination of somatic status were carried out as previously described (1, 30).

Chromatin Immunoprecipitation (ChIP) and qPCR enrichment Approximately 2e8 cells from each of three melanoma lines (79T, 75T, skmel28) were prepared for ChIP analysis as previously described (31)with the following modifications. Intact nuclei were isolated and chromatin was sheared on ice to an average size of 200- 1000 bp using a Branson Sonifier 450 (constant duty cycle, 20% output; 9 cycles of 20 second sonication with 1 minute rest between each round of sonication). Chromatin preparations were split into 5 equal portions for ChIP analysis using 15 ug each of anti- H3K4me3 (Abcam, Cambridge, MA) or anti-Deaf1 (Bethyl Laboratories, Dallas, TX), anti-Stat1 (Santa Cruz, Santa Cruz, CA), anti-Stat3 (Santa Cruz), or GFP (Abcam) antibodies.

ChIP enrichment was assessed by qPCR analysis using SYBR Green (Qiagen). Assays were performed in triplicate using 10 ng of input or ChIP DNA. Dissociation curve

29 analysis confirmed a single amplification product in all samples for each amplicon. Enrichment of ChIP DNA compared to input was determined using the delta Ct method. Delta Ct was converted from logarithmic to linear scale using the equation 2e-(deltaCt). Primer sequences for all amplicons are available upon request.

Allelic chromatin immunoprecipitation (ChIP) and expression analyses of melanoma samples Allelic ChIP DNA representation was determined using MALDI/TOF mass spectrometry as previously described(32). For all assays, the BCL2L12 C/T SNP was genotyped using iPlex Gold SBE (Sequenom, San Diego, CA). Input or ChIP (histone H3K4me3, Deaf1, Stat1, Stat3, or non-specific GFP) DNA for each melanoma sample was aliquotted at 10 ng/assay in 384-well format. Eight replicates were tested for each sample SNP assay. Amplification primer sequences for two different amplicons to analyze the melanoma SNP are: ACGTTGGATGCCGCCCCTATGCCCTTTTTT (mel_1F), ACGTTGGATGCGCTCAATTTGCATGTGACG (mel_1R), ACGTTGGATGTTTATCATTCTTTGGGTAACAGAC (mel_2F), ACGTTGGATGGGTCTCCTCCACTGAACTCG (mel_2R). The same extension primer sequence was used to interrogate the SNP position for both amplicons: TGCCCTTTTTTGGGTTT (mel_X). MALDI-TOF analysis for each assay was performed sampling each matrix pad by rastering to 9 independent positions on the pad, accumulating ten laser shots per position. Genotypes were assigned using SpectroCaller software and the peak fitting and area under each allele peak was calculated by SpectroAcquire software (Sequenom, Inc.). Data obtained from each amplicon were comparable, so only data from amplicon one are shown in Fig S6D. To calculate allelic representation of the melanoma SNP, we used the peak areas of the lower (C allele) and higher (T allele) mass alleles to estimate the proportion of the C allele = C/(C+T). Statistical significance of any allelic representation differences between each ChIP DNA sample and its paired input DNA sample were assessed using two-sample two-sided t-test with allowance for unequal variance.

Allelic mRNA/cDNA representation analyses shown in Fig 1A were also determined using iPlex Gold SBE (Sequenom, San Diego, CA) using the mel_1 amplification and the mel_X extension primer sets. For each melanoma sample, 12.5 ng of cDNA or 20 ng of gDNA were aliquotted in 384-well format. Four replicates were tested for each sample SNP assay; the entire experiment was repeated 3 times on 3 different days. Allelic representation was determined as above by measuring the peak areas of the C and T alleles. Allelic cDNA representation was compared to paired gDNA representation for each sample to identify statistically significant differences. Because cDNA variance did not appear to exhibit a normal distribution, the more conservative Wilcoxon rank sum statistical test was used to determine statistical significance. However, the two- sample two-sided t-test with allowance for unequal variance yielded the same general conclusions that 9/12 samples showed statistically significant differences in allelic representation.

30 Construction of wild-type and mutant BCL2L12 expression vectors Human BCL2L12 (NM_138639.1) was cloned by PCR as previously described (1) using clones (# MHS4426-99622734-BCL2L12) purchased from Open Biosystems with primers listed in Table S5. The PCR products were cloned into the mammalian expression vectors pCDF-MCS2-EF1-Puro™ or pCDH-MCS2-EF1-Neo™ (Systems Biosciences, Inc., Mountain View, CA) or pcDNA3.1(-) (Invitrogen) via the XbaI and NotI restriction sites. The F17F BCL2L12 point mutant was made using Phusion PCR for site-directed mutagenesis.

Cell culture and transient expression HEK 293T and HEK293 cells were purchased from ATCC (Manassas, VA) and maintained in complete RPMI-1640 medium supplemented with 10% Fetal Bovine Serum (FBS). HEK 293T cells or HEK293 were transfected with Arrest-IN reagent (Open Biosystems) at a 6:1 ratio with DNA (L:g) using 2-5 g of plasmid DNA.

Immunoprecipitation and Western Blotting BCL2L12 sub-cellular fractionation was performed as previously described (33). Briefly, the BCL2L12 over-expressed HEK293T cells were harvested then washed with ice-cold PBS and lysed in a hypotonic lysis buffer (10 mM Tris pH 7.4, 10 mM NaCl, 3 mM MgCl2, 1 mM EDTA, 1 mM EGTA and cocktail protease inhibitors [Thermo Scientific- 78415]). The cells were re-suspended in 200 µL of lysis buffer, incubated on ice for 10 min, titurated through p2 tip 15-20 times and sonicated for 2-3 times; the total fraction was centrifuged for 15 minutes at 375 xg at 4 oC resulted in a pellet which is the nuclear fraction and supernatant which is the post nuclear fraction (PNF). Post nuclear fractions were loaded on 10% Bris-Tris gel and further analyzed using mouse monoclonal BCL2L12 (Abcam 1:1000 dilution) fragment antibody. The same membrane also processed later with GAPDH antibody.

HEK293 cells transiently transfected with BCL2L12-FLAG (WT, mutant or empty vector) were gently washed 2X in PBS and then lysed using 1.0 ml 1% NP-40 lysis buffer (1% NP-40, 50 mM Tris-HCl pH 7.5, 150 mM NaCl, Complete Protease Inhibitor tablet, EDTA-free (Roche, Indianapolis, IN), 1M sodium orthovanadate, 1 mM sodium fluoride, and 0.1% -mercaptoethanol) per T-75 flask for 20 minutes on ice. Lysed cells were scraped and transferred into a 1.5 mL microcentrifuge tube. Extracts were centrifuged for 10 minutes at 14,000 rpm at 4°C. 800 L of supernatant was immunoprecipitated overnight using 30 L of anti-FLAG (M2) beads (Sigma-Aldrich). The immunoprecipitates were washed and subjected to SDS-PAGE and western blotting as previously described (1). Primary antibodies used to detect BCL2L12 and co- immunoprecipitated proteins were anti-BCL2L12 (#ab57800) (Abcam), anti-p53 (sc- 6243) (Santa Cruz), or anti-Tubulin (#T8203) (Sigma Aldrich).

Preparation of cell-free translation extracts from 75T melanoma cells Melanoma 75T cells cultured in RPMI-1640 medium (Cellgro Mediatech Inc., Manassas, VA) with 10% FBS (Gibco/Life Technologies, Grand Island, NY), 2 mM glutamine and penicillin/streptomycin (HyClone, Logan, UT) and grown to 80% confluency were washed with cold PBS (HyClone, Logan, UT), harvested by scraping in 10-20 mL of

31 cold isotonic buffer containing 35 mM HEPES-KOH pH 7.6, 146 mM NaCl and 11 mM glucose and collected by pelleting (300×g, 7 min, 4°C). Cell pellet was washed twice with cold buffer (above) and resuspended in 1 mL hypotonic solution containing 20 mM HEPES-KOH pH 7.6, 10 mM potassium acetate, 1 mM magnesium acetate, 4 mM DTT, supplemented with complete (EDTA free) protease inhibitor cocktail (Roche, Indianapolis, IN), and incubated on ice for 15 min. Following incubation, cells were transferred to Dounce homogenizer and carefully broken with 25-30 strokes. Extracts were cleared by centrifugation (10,000×g, 10 min, 4°C), aliquoted, frozen in liquid nitrogen and stored at -80°C.

In vitro transcription and translation of BCL2L12 mRNA For in vitro transcription BCL2L12 cDNA was cloned downstream of the T7 promoter in pCR-BluntII-TOPO vector (Open Biosystems, Huntsville, AL). F17F synonymous mutation (TTC→TTT) was introduced by site directed mutagenesis using QuickChange II XL Site-Directed Mutagenesis Kit (Agilent technologies, Santa Clara, CA) using 5´- CTTTTTTGGGTTTTCGGCCAGAGGC-3´ and 5´- GCCTCTGGCCGAAAACCCAAAAAAG -3´ primers. Plasmids carrying wild-type and mutant BCL2L12 alleles were linearized with HindIII (New England Biolabs, Ipswich, MA) and mRNAs were in vitro transcribed (1.5 h 37°C) using Ambion’s mMessage mMachine T7 ultrakit (Ambion/Life Technologies, Grand Island, NY). mRNAs were further purified by lithium chloride precipitation and washed with 70% ethanol. Aqueous mRNA solutions (1 mg/mL) were used. For in vitro translation, 75T melanoma extracts were first treated (in the presence of 1 mM calcium chloride) with 0.01 U/µl micrococcal nuclease (New England Biolabs, Ipswich, MA) for 10 min at 37°C. Nuclease treatment was stopped by EGTA (final concentration of 2 mM), the extracts were transferred on ice for 5 min, spun briefly and the supernatant was further used for in vitro translation. A typical translation reaction (total volume 48 µl) contained 25 µl of nuclease treated cell-free extract, 2 µg capped mRNA, 80 mM potassium acetate, 1 mM magnesium acetate, 0.12 mM GTP (Roche, Indianapolis, IN), 1 mM ATP (Roche, Indianapolis, IN), 40 units RNase inhibitor (New England Biolabs, Ipswich, MA), 0.1 mg/ml creatine phosphokinase (Sigma-Aldrich, St. Louis MO), 20 mM HEPES- KOH pH 7.6, 1 mM amino acids (minus methionine), 0.15 mM spermidine (Sigma-Aldrich, St. Louis MO), 20 µCi [35S]-methionine (MP Biomedicals, Solon, OH) and was done for 3 h at 37°C. Translation products were analyzed on 12.5% SDS-PAGE. After electrophoresis gels were fixed and dried using vacuum gel dyer. Radiolabeled translation products were visualized by autoradiography using a Typhoon 9410 imaging scanner (GE Healthcare Life Sciences, Piscataway, NJ).

Lentiviral shRNA Constructs for stable depletion of BCL2L12 (cat # RHS45330-NM_138639) were obtained from Open Biosystems (Huntsville, AL) and were confirmed to efficiently knockdown BCL2L12 at the protein level. Lentiviral stocks were prepared as previously described (30). Melanoma cell lines (12T and 75T) were infected with shRNA lentiviruses for each condition (vector and two different BCL2L12 specific shRNAs). Selection of stable pooled clones was done in the presence of 3g/mL puromycin

32 containing normal medium for 3-5 days prior to determining knock-down efficiency. Stably infected pooled clones were tested in functional assays.

Quantitative real-time PCR Total RNA was extracted from pooled clones of melanoma cells 12T and 75T stably knocked down for endogenous BCL2L12 as well as from SK-MEL-28 and A375 pooled clones stably expressing BCL2L12 (WT or F17F) or stably transduced with empty vector following the manufacturer’s protocol for the RNeasy Mini Kit (QIAGEN #74101). Total RNA was eluted in 30 μL diethylpyrocarbonate (DEPC)-treated distilled H2O. A total of 1 μg of total RNA was used for single-strand complementary DNA (cDNA) synthesis using a SuperScript III First Strand kit (Invitrogen #18080-051). cDNA was amplified using the olido dT20 primer supplied in the kit. To test for loss of BCL2L12 message, we used 0.4 μL of cDNA in the PCR with either BCL2L12 primers or GAPDH primers (Table S5) mixed with 2× Fast SYBR Green PCR mix at a final volume of 10 μl in triplicate (Applied Biosystems cat # 4355612). qRT-PCR analysis was done using the ABI 7900HT Fast Real-Time PCR system (with a standard program of stage 1: 50 °C for 2 min; stage 2: 95 °C for 10 min; stage 3: 40 cycles of 95 °C for 15 s and 60 °C for 1 min). Results were analyzed using Microsoft Excel and SPSS. siRNA depletion of endogenous BCL2L12 in melanoma cells Specific siRNA was purchased from Dharmacon (Thermo Fisher Scientific) designed using there siRNA design program for human BCL2L12. Four independent siRNA molecules were used to transiently deplete BCL2L12 in malignant melanoma cells. Using DharmaEffect transfection reagent #1 specific for siRNA, melanoma cells were tranfected with 50nM siRNA molecules (#3 and #4) in the presence of OptiMEM-I medium after cells were seeded into 96-well plates at a density of 2000 cells/well 24 prior to transfection. Cells were incubated for 24 hr post-transfection prior to application of any genotoxic stressors.

Cell viability assays Stably depleted pooled clones (12T (WT) or 75T (mutant)) as well as stable pooled clones (SK-Mel-28 and A375) either expressing BCL2L12-FLAG (WT, mutant or empty vector) were seeded into 96-well clear bottom opaque plates at 1,000 cells per well. Cells were incubated 24 hrs prior to exposure to UV light (50kJ) using a UV Stratalinker 2400 (Stratagene). Plates were then incubated for an additional 48 hrs prior to testing for cell viability using Cell-Titer-Glo (cat# G7571). Plates were analyzed on a Thermo Electron Luminoskan reader. Data was then analyzed using Microsoft Excel to generate graphs and statistics. miRNA target site prediction We used the two miRNA target site prediction platforms: PITA (34) and miRanda (35) to search for miRNA target site predictions that overlap the C51T mutated position in BCL2L12. We executed them with default prediction parameters and found that both platforms predict hsa-miR-671-5p to target the wild-type BCL2L12 mRNA overlapping position 51, but not the mutant mRNA.

33 miRNA depletion of endogenous BCL2L12 in melanoma cells A specific miRNA mimetic (hsa-miR-671-5p) was purchased from Sigma Aldrich (HMI0901) which was determined to potentially target human BCL2L12. A negative control scrambled miR (NC) was purchased from Dharmacon (Thermo Fisher Scientific) (CN-001000-01). Using DharmaFECT transfection reagent #1 (T-2001) specific for siRNA or miRNA, melanoma cells were tranfected with 20nM has-miR-671-5p or NC molecules in the presence of OptiMEM-I medium after cells were seeded into 6-well plates at a density of 200,000 cells/well 24 prior to transfection. Cells were incubated for 24 hr post-transfection prior to extraction of miRNA and mRNA and qRT-PCR analysis.

Anti-miR-671-5p rescue assay A specific anti-miRNA mimic (anti-hsa-miR-671-5p) was purchased from Qiagen (MIN0003880) which was determined to inhibit the hsa-miR-671-5p mimic. A negative control scrambled miR (NC) was purchased from Dharmacon (Thermo Fisher Scientific). Using DharmaEffect transfection reagent #1 specific for siRNA or miRNA, melanoma cells were co-tranfected with 20nM hsa-miR-671-5p or NC molecules plus either 0nM or 50nM anti-miR-671-5p in the presence of OptiMEM-I medium after cells were seeded into 6-well plates at a density of 200,000 cells/well 24 prior to transfection. Cells were incubated for 24 hr post-transfection prior to extraction of miRNA and mRNA and qRT-PCR analysis.

Quantitative real-time PCR of miRNA targeted cell lines miRNA and mRNA was extracted from transiently transfected melanoma cells 75T, 79T, 12T and SK-Mel-28 to assess for knock-down of endogenous BCL2L12 following the manufacturer’s protocol for the miRNeasy Mini Kit (QIAGEN #217004). Total was eluted in 30 μL diethylpyrocarbonate (DEPC)-treated distilled H2O. A total of 1 μg of total RNA was used for single-strand complementary DNA (cDNA) synthesis using a miScript II Reverse Transcription kit (QIAGEN #218193). cDNA was amplified the 5X HiFlex buffer to quantitate in parallel the miRNA and mRNA. To test for loss of BCL2L12 message, we used 1 μL of diluted cDNA (10ng/l) in the PCR with either BCL2L12 primers or GAPDH primers or mature miR (Table S5) or precursor-miR (QIAGEN #MP00003479) we followed manufacturer’s protocol and mixed primers and cDNA with QuantiTect SYBR Green PCR master mix at a final volume of 10 μl in triplicate (QIAGEN). qRT- PCR analysis was done using the ABI 7900HT Fast Real-Time PCR system (with a standard program of stage 1: 50 °C for 2 min; stage 2: 95 °C for 10 min; stage 3: 40 cycles of 95 °C for 15 s and 60 °C for 1 min). Results were analyzed using Microsoft Excel and GraphPad Prism v5.0.

Taqman assay – Custom Taqman probes were designed by ABI to detect the wild-type version of BCL2L12, mutant version of BCL2L12 (C51T), and an internal control (TBP). Taqman assays were run using the Universal Taqman PCR master mix (2X), the specific probes, cDNA template from each individual clone and DEPC-dH20 in a 10ml volume in triplicate in a 384-well plate and run on a HT-7900 Fast PCR machine (ABI). Data was analyzed and CTs were normalized against TBP samples and used to generate graphically representations. All experiments were repeated in triplicate (n=3).

34 miRNA rescue experiment – A modified form of the hsa-miR-671-5p was custom made from Sigma Aldrich with a single site changed to represent the synonymous mutation found in melanoma. Melanoma cells 12T (WT) and 79T (C51T/F17F) cells were seeded at ~300,000 cells per well in 6-well plates and incubated overnight prior to transient transfection. Cells were transfected with hsa-miR-671-5p (miR), mod-hsa-miR- 671-5p (mod-miR) or NC in triplicate and allowed to incubate for 24-36 hrs prior to miRNA/RNA purification. Total miRNA/RNA was amplified using the miScript miRNA cDNA kit from Qiagen and levels of BCL2L12 message was detected using SYBR Green master mix (Qiagen) in triplicate. GAPDH was used as an internal control to normal between samples and to generate graphs using Micorsoft Excel. All experiments were repeated at two-three times.

References 1. Palavalli LH, et al. (2009) Analysis of the matrix metalloproteinase family reveals that MMP8 is often mutated in melanoma. Nature genetics 41(5):518-520. 2. Morente MM, et al. (2006) TuBaFrost 2: Standardising tissue collection and quality control procedures for a European virtual frozen tissue bank network. Eur J Cancer 42(16):2684-2691. 3. Davies MA, et al. (2009) Integrated Molecular and Clinical Analysis of AKT Activation in Metastatic Melanoma. Clin Cancer Res 15(24):7538-7546. 4. Pavey S, et al. (2004) Microarray expression profiling in melanoma reveals a BRAF mutation signature. Oncogene 23(23):4060-4067. 5. Castellano M, et al. (1997) CDKN2A/p16 is inactivated in most melanoma cell lines. Cancer research 57(21):4868-4875. 6. Dutton-Regester K AL, Nancarrow DJ, Stark MS, O’Connor L, Lanagan C, Pupo GM, Tembe V, Carter CD, O’Rourke M, Scolyer RA, Mann GJ, Schmidt C, Herington A, Hayward NK (2012) Identification of TFG (TRK Fused Gene) as a putative tumour supressor gene in metastatic melanoma. Genes, and Cancer In Press. 7. Hasel C, Bhanot UK, Heydrich R, Strater J, & Moller P (2005) Parenchymal regression in chronic pancreatitis spares islets reprogrammed for the expression of NFkappaB and IAPs. Lab Invest 85(10):1263-1275. 8. Bhanot U, Kohntop R, Hasel C, & Moller P (2008) Evidence of Notch pathway activation in the ectatic ducts of chronic pancreatitis. The Journal of pathology 214(3):312-319. 9. Li H & Durbin R (2009) Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics (Oxford, England) 25(14):1754-1760. 10. Li H, et al. (2009) The Sequence Alignment/Map format and SAMtools. Bioinformatics (Oxford, England) 25(16):2078-2079. 11. Teer JK, et al. (2010) Systematic comparison of three genomic enrichment methods for massively parallel DNA sequencing. Genome research 20(10):1420-1431. 12. Ajay SS, Parker SC, Abaan HO, Fajardo KV, & Margulies EH (2011) Accurate and comprehensive sequencing of personal genomes. Genome research 21(9):1498-1505. 13. Wei X, et al. (2011) Exome sequencing identifies GRIN2A as frequently mutated in melanoma. Nature genetics 43(5):442-446. 14. Biesecker LG, et al. (2009) The ClinSeq Project: piloting large-scale genome sequencing for research in genomic medicine. Genome research 19(9):1665-1674. 15. Carter H, Samayoa J, Hruban RH, & Karchin R (2010) Prioritization of driver mutations in pancreatic cancer using cancer-specific high-throughput annotation of somatic mutations (CHASM). Cancer biology & therapy 10(6):582-587.

35 16. Cooper GM, et al. (2005) Distribution and intensity of constraint in mammalian genomic sequence. Genome research 15(7):901-913. 17. Thomas DJ, et al. (2007) The ENCODE Project at UC Santa Cruz. Nucleic acids research 35(Database issue):D663-667. 18. Flicek P, et al. (2012) Ensembl 2012. Nucleic acids research 40(Database issue):D84-90. 19. Schmidt D, et al. (2010) Five-vertebrate ChIP-seq reveals the evolutionary dynamics of transcription factor binding. Science (New York, N.Y 328(5981):1036-1040. 20. Consortium EP, et al. (2012) An integrated encyclopedia of DNA elements in the human genome. Nature 489(7414):57-74. 21. MacArthur DG, et al. (2012) A systematic survey of loss-of-function variants in human protein- coding genes. Science (New York, N.Y 335(6070):823-828. 22. Cooper GM, et al. (2010) Single-nucleotide evolutionary constraint scores highlight disease- causing mutations. Nature methods 7(4):250-251. 23. Dogan RI, Getoor L, Wilbur WJ, & Mount SM (2007) SplicePort--an interactive splice-site analysis tool. Nucleic acids research 35(Web Server issue):W285-291. 24. Piva F, Giulietti M, Nocchi L, & Principato G (2009) SpliceAid: a database of experimental RNA target motifs bound by splicing proteins in humans. Bioinformatics (Oxford, England) 25(9):1211-1213. 25. Zhu H, Hasman RA, Young KM, Kedersha NL, & Lou H (2003) U1 snRNP-dependent function of TIAR in the regulation of alternative RNA processing of the human calcitonin/CGRP pre-mRNA. Molecular and cellular biology 23(17):5959-5971. 26. Matys V, et al. (2003) TRANSFAC: transcriptional regulation, from patterns to profiles. Nucleic acids research 31(1):374-378. 27. Sandelin A, Alkema W, Engstrom P, Wasserman WW, & Lenhard B (2004) JASPAR: an open- access database for eukaryotic transcription factor binding profiles. Nucleic acids research 32(Database issue):D91-94. 28. Gotea V, et al. (2010) Homotypic clusters of transcription factor binding sites are a key component of human promoters and enhancers. Genome research 20(5):565-577. 29. Crooks GE, Hon G, Chandonia JM, & Brenner SE (2004) WebLogo: a sequence logo generator. Genome research 14(6):1188-1190. 30. Prickett TD, et al. (2009) Analysis of the tyrosine kinome in melanoma reveals recurrent mutations in ERBB4. Nature genetics 41(10):1127-1132. 31. Scacheri PC, Crawford GE, & Davis S (2006) Statistics for ChIP-chip and DNase hypersensitivity experiments on NimbleGen arrays. Methods Enzymol 411:270-282. 32. Amos-Landgraf JM, et al. (2006) X -inactivation patterns of 1,005 phenotypically unaffected females. American journal of human genetics 79(3):493-499. 33. Fazioli F, et al. (1993) Eps8, a substrate for the epidermal growth factor receptor kinase, enhances EGF-dependent mitogenic signals. The EMBO journal 12(10):3799-3808. 34. Kertesz M, Iovino N, Unnerstall U, Gaul U, & Segal E (2007) The role of site accessibility in microRNA target recognition. Nature genetics 39(10):1278-1284. 35. John B, et al. (2004) Human MicroRNA targets. PLoS Biol 2(11):e363.

36