A B

Supplementary Figure S1

Figure S1 The pattern of LOH amongst distinct tumors from different members of a fNMTC family

is suggestive of a presumable susceptibility haplotype at 2q21 however, the LRP1B germline genetic

variants detected are o f undetermined significance. (A) Heredogram summarizing the clinical

features (age a t diagnosis and histotype) and the pattern of allelic losses in distinct tumors from

different family members. Allelotyping of the 2q21 interval was performed for 14 microsatellite

markers and 1 LRP1B intragenic SNP. A code for each allele was generated on the basis of the size

of the microsatellite fragment. The index case is annotated with an arrow (fl ). The haplotype shared

by the affected individuals is overshadowed in grey. The alleles detected for each marker in the

distinct tumor nodules derived from affected members are represented along side of the respective

case. FTA - Follicular Thyroid Adenoma; PTC - Papillary Thyroid Carcinoma; FV - Follicular

Variant; FTC - Follicular Thyroid Carcinoma; - Retention of Heterozygosity; - Loss of

Heterozygosity; - Non Informative Marker; na - not amplified. (B) Genetic variants detected

upon germline sequencing of the entire LRP1B coding exons and respective intron boundaries in the

index fNMTC case.

Supplementary Figure S2

Figure S2 Methylation analysis of the CpG island reporters after in vitro methylation. In order to

confirm that different patterns of DNA methylation have been generated by M.Sss-I, M.Hpa-II or

M.Msp-I, after in vitro methylation plasmid DNA was c onverted by bisulphite treatment and

subjected to DNA sequencing. Constructs of the CpG island fragment A (-429 to -1) and fragment B

(+1 to +530 , counting from ATG) are represented schematically on the left panel and the

corresponding chromatograms are represented on the right side. Chromatograms illustrate bisulphite-

sequences of the regions delimited by a solid square within the construct. Targets CG, CCGG as well

as the p300 binding site are bordered by doted squares. Arrows indicate sites of methylation.

Supplementary Figure S3

Figure S3 Quantitative Real-Time-PCR quantification of p300 mRNA levels in thyroid cancer

derived cell lines and correlation with LRP1B mRNA expression.

Supplementary Figure S4

Figure S4 A panel of miRNAs predicted to t arget the 3-UTR by computational analysis were

quantified in 10 thyroid cancer cell lines and in normal thyroid tissue and paired tumors of the index

case. Expression values were normalized to the levels of miRNA-U6b determined in the same RNA

samples. A pool of 9 RNAs derived from normal thyroid tissue (NPOOL) was used as the normal

reference.

Supplementary Figure S5

A

B

C

D

Figure S5 mLRP1B expression changes the levels of multiple Cytokines in the extracellular

medium. Conditioned medium was obtained from stable 8505C mLRP1B or 8505C empty vector

control cells by placing 2.5x105 cells in fibronectin-coated 6-well plates for 24 hours in 1ml serum

free media. (A) Map of the antibody arrays used depicting location of antibodies for each of 51

cytokines analyzed. (B) For incubation of a Cytokine antibody arrays media were diluted 1:3 in

sample buffer and 1ml was directly incubated with previously blocked membrane array and stained

with avidin-peroxidase conjugate followed by exposure to X-ray film. (C and D) D ots were

quantified using the Quantity One software and arrays were normalized trough positive controls.

LRP1B relative expression ratio

A

B

Supplementary Figure S6

Figure S6 LRP1B silencing is found across the most prevalent cancer types. (A) LRP1B expression

is considerably higher in normal thyroid relative to control RNA (obtained commercially) derived

from normal tissue of other organs such as lung, stomach, intestine and breast (obtained

commercially). (B) Nevertheless, LRP1B silencing is also a c ommon feature in cell lines derived

from lung, gastric, colorectal, breast, pancreatic and bladder cancer, as is observable from qRT-PCR

quantification of mRNA levels in cancer cell lines relative to n ormal reference from the

corresponding organ.

Cell Line LRP1B Relative Expression*

2q21 Copy Number

DNA Methylation %

miR-548a-5pExpression

8q22.3 Copy Number

Summary of LRP1B Hits

KAT4 -183223 1 94% +2 1 - Copy number loss- DNA Methylation (complete)

ARO -158357 1 94% +2 1 - Copy number loss- DNA Methylation (complete)

8505C -21603 1 0% +6 4 - Copy number loss- miR overexpression

BCPAP -16433 2 0% +15 2 -miR overexpression (high)

K1 -2551 3 0% +35 6 - Copy number gain- miR overexpression (very high)

TPC-1 -1280 3 6% +56 5 - Copy number gain- miR overexpression (very high)

NPA -1241 2 5% +22 2 - miR overexpression (high)

XTC-1 -1117 2 77% +11 4 -DNA Methylation (partial)- miR overexpression (moderate)

C643 -233 1 7% +11 4 -Copy number loss- miR overexpression (moderate)

HTH74 -156 1 0% +5 3 - Copy number loss- miR overexpression (low)

Supplementary Table S3

Table S3 Integration of data on LRP1B inactivation mechanism in cancer cell lines. Cell lines were

sorted by ascending order according to LRP1B expression relative to normal thyroid. The degree of

2q21 copy number, density of DNA methylation, miR-548a-5p overexpression and 8q22.3 copy

number are presented. We can infer biallelic inactivation in KAT4 and ARO cell lines to be the result

of 2q21 copy number loss coupled with high density promoter methylation (>90%) of the remaining

copy. In the absence of promoter methylation, an alternative combination leading to silencing comes

from 2q21 copy number loss accompanied by high miR-548a-5p overexpression (8505C). The

outcome is somewhat similar when there is no 2q21 copy number loss but very high miR-548a-5p

overexpression/8q22.3 gain is observed (BCPAP). Cell lines K1 and TPC-1 also display very high

miR-548a-5p overexpression but have relatively higher levels of LRP1B expression probably

because the miR-548a-5p effect is compensated by an observable low copy number gain at 2q21.

NPA only displays very high miR-548a-5p overexpression. LRP1B under-expression in XTC-1

results from high miR-548a-5p overexpression combined with high DNA methylation, without 2q21

copy number loss. In HTH74 and C643 cells moderate LRP1B under-expression is attributable to

copy number loss alone (HTH74) or together with miR-548a-5p overexpression/8q22.3 gain (C643).