Transcript NF2 - SYSMED

INTRODUCTION
▪ MPM(malignant pleural mesothelioma ) is an aggressive cancer arising from the mesothelial cells of the pleura. About 80% of
mesothelioma cases are linked to asbestos exposure; the remainder may be related to prior chest radiation, genetic predisposition or
spontaneous occurrence. In the United States, ~3,200 new MPM cases and ~3,000 deaths due to MPM occur annually.
▪ MPM tumors are broadly divided into three histological subtypes: epithelioid, biphasic (or mixed) and sarcomatoid. The prognosis for
patients with MPM is dismal, with a median overall survival of 8–36 months depending on stage, but patients with sarcomatoid MPM have
particularly poor outcomes compared to patients with epithelioid histology. Although aggressive surgery is effective in patients with early,
limited MPM with epithelioid histology, most patients present at advanced stages, and current drug regimens are ineffective.
INTRODUCTION
▪ Understanding the genetic alterations that drive MPM is critical for successful development of diagnostics, prognostics and personalized
therapeutic modalities. Because MPM is rare, genomic studies are limited and have typically involved a small number of samples.
▪ Previously, loss-of-function mutations in CDKN2A, NF2 and BAP1 have been reported in MPM. In addition, previous studies have reported
copy gains and copy losses involving multiple regions of the genome. Familial studies have identified germline BAP1 mutations that
predispose carriers to mesothelioma. However, understanding of the mutational landscape of MPM is not yet sufficient to affect
classification or treatment strategies.
▸Here they sequenced transcriptomes and exomes from 216 MPMs.
▸Sample description
Subtype
1
2
3
4
5
Epithelioid
Sarcomatoid
Biphasic-E
Biphasic-S
Unassigned
Total
# of samples
54
29
72
56
6
217*
Targeted exome sequencing
(SPET)
Exome
Tumor
21
18
41
18
1
99
Normal
21
18
41
18
1
99
Paired
21
18
41
18
1
99
Tumor
28
11
27
33
4
103
Normal
28
11
27
33
4
103
Paired
28
11
27
33
4
103
RNA-seq
Tumor
54
29
72
56
0
211
SNP array
Tumor
19
18
39
18
1
95
Normal
19
18
39
18
1
95
WGS
Paired
19
18
39
18
1
95
Tumor
4
7
7
1
1
20
Normal
4
7
7
1
1
20
Paired
4
7
7
1
1
20
1. RESULTS: Expression analysis identifies distinct MPM subtypes
[RNA-seq  EXPRESSION]
▸ Obtained RNA-seq data for 211 MPM samples.
▸ RNA-seq libraries were prepared using TruSeq RNA Sample Preparation kit (Illumina).
▸ paired-end (2 × 75-bp) reads per sample.
▸ All sequencing reads were evaluated for quality using the Bioconductor ShortRead package.
▸ RNA-seq reads were aligned to the human genome version NCBI GRCh37 using GSNAP(Genomic Short-read Nucleotide Alignment Program).
▸ Expression counts per gene were obtained by counting the number of reads aligned concordantly within a pair and uniquely to each gene locus as
defined by NCBI and Ensembl gene annotations and RefSeq mRNA sequences. Differential gene expression analysis performed using edgeR, and DESeq2.
▸Mesotheliomas are histologically classified as epithelioid, biphasic or sarcomatoid and are heterogeneous, with different proportions
of epithelioid and sarcomatoid features. Several studies have analyzed gene expression in MPMs with microarrays; however, none have
established molecular subtypes routinely applied in patient care.
▸ To define molecular subgroups of MPMs, they performed unsupervised consensus clustering of RNA-seq–derived expression data
from 211 of the 216 MPM samples and identified four major clusters: sarcomatoid, epithelioid, biphasic- epithelioid (biphasic-E) and
biphasic-sarcomatoid (biphasic-S) (Fig. 1).
1. RESULTS: Expression analysis identifies distinct MPM subtypes
▸ The sarcomatoid cluster contained all eight samples histologically classified as
sarcomatoid/desmoplastic and 21% (13/62) of histologically biphasic samples. Of
the remaining 49 histologically biphasic samples, 21 were classified as biphasic-E,
27 as biphasic-S and 1 as epithelioid. The biphasic samples that clustered with
sarcomatoid samples contained a high fraction of sarcomatoid cells (Fig. 1a).
▸ Notably, the remaining 62% (88/141) of histologically epithelioid samples,
classified as biphasic-E (n = 51), biphasic-S (n = 29) or sarcomatoid (n = 8), showed
lower overall survival (log-rank test P < 0.0001; Fig. 1b) with a hazard ratio of 2.5
(Cox proportional hazard model P < 0.001; 95% confidence interval 1.6–3.8).
Further, the epithelioid cluster showed greater overall survival than the other
groups (Fig. 1c).
▸ The most significantly upregulated gene in the epithelioid group was CLDN15.
CLDN15 is downregulated in cells undergoing epithelial-to-mesenchymal transition
(EMT). In contrast, LOXL2, known to contribute to EMT, and VIM, which is
upregulated during EMT, were among the most significantly upregulated genes in
sarcomatoid samples. They found that the log2 ratio of CLDN15/VIM (C/V)
expression was significantly different between subtypes (Fig. 1d). A similar trend
for C/V ratio discriminating between subtypes was also observed in a previously
published microarray data set(supplementary Fig. 5).
▪ Epithelioid group: up-regulated gene, CLDN15
▪ Sarcomatoid group: up-regulated genes, LOX2 and VIM
1. RESULTS: Expression analysis identifies distinct MPM subtypes
2. RESULTS: MPM mutational profile
[Exome-seq  MUTATION]
▸ Obtained exome-seq data for 99 MPM samples.
▸ Exome capture was performed using the Agilent SureSelect Human All Exome kit.
▸ To generate paired-end (2 × 75-bp) reads per sample.
▸ All sequencing reads were evaluated for quality using the Bioconductor ShortRead package.
▸ Sequencing reads were mapped to UCSC human genome (GRCh37/hg19) using BWA software set to default parameters.
▸ Somatic variant calling on tumor and its matched normal BAM file was performed using Strelka(accurate somatic small-variant calling from sequenced
tumor-normal sample pairs). Known germline variants represented in dbSNP (Build 131) or 6,515 previously published normal exomes but not represented
in COSMIC (v62) were filtered out for all samples. In addition, germline variants that were present in both the tumor and normal samples were removed.
[TCGA mutation data]
TCGA mutation data used in Figure 2e were retrieved using the GGDSR R package from cBioPortal.
[Mutational signatures]
▸ They analyzed the MPM exome sequence data for the frequency of the possible 96 mutation types. They also included in the analysis TCGA exome data
for 2,437 samples from 8 other cancer types as provided by the SomaticCancerAlterations Bioconductor package as well as data from two small-cell lung
cancer studies. A set of 6 common signatures was detected across the combined data set using non-negative matrix factorization.
2. RESULTS: MPM mutational profile
epithelioid
biphasic-E
biphasic-S
sarcomatoid
LAML, acute myeloid leukemia
THCA, papillary thyroid carcinoma
MESO, mesothelioma
BRCA, breast carcinoma
GBM, glioblastoma
OV, ovarian carcinoma
KIRC, clear cell renal carcinoma
UCEC, endometrial carcinoma
COADREAD, colon and rectal carcinoma
STAD, gastric adenocarcinoma
LUAD, lung adenocarcinoma
BLCA, urothelial bladder carcinoma
SCLC, small-cell lung cancer
LUSC, lung squamous-cell carcinoma
SKCM, cutaneous melanoma
HNSC, head and neck squamous cell carcinoma
▸ They identified a total of 2,529 protein-altering somatic mutations, including 2,069 missense, 190 nonsense, 3 stop-loss, 63 essential splice site and 204
frameshift mutations. A majority of the mutations (85%; 2,144/2,529) were novel, as they were not reported in COSMIC or OncoMD.
▸ They did not observe significant differences in mutation rate between molecular subtypes (P = 0.8; Fig. 2a–d). Notably, with the exception of thyroid
carcinoma and acute myeloid leukemia, MPMs showed a low protein-altering mutation rate compared to other cancers (Fig. 2e).
2. RESULTS: MPM mutational profile
LAML, acute myeloid leukemia
THCA, papillary thyroid carcinoma
MESO, mesothelioma
BRCA, breast carcinoma
GBM, glioblastoma
OV, ovarian carcinoma
KIRC, clear cell renal carcinoma
UCEC, endometrial carcinoma
COADREAD, colon and rectal carcinoma
STAD, gastric adenocarcinoma
LUAD, lung adenocarcinoma
BLCA, urothelial bladder carcinoma
SCLC, small-cell lung cancer
LUSC, lung squamous-cell carcinoma
SKCM, cutaneous melanoma
HNSC, head and neck squamous cell carcinoma
▸ They found S1 and S2 to be the predominant mutational signatures in MPMs (Fig. 2g). The S1 signature, with no predominant transition or transversion, is
probably indicative of a base-agnostic mutagen such as reactive oxygen species (ROS) and is consistent with asbestos exposure known to induce ROS in MPM.
▸ Signature S2 is characterized by C>T transitions at NpCpG trinucleotides and is indicative of the elevated deamination rate of 5-methylcytosine to thymine in
CpG islands. Whereas S4 shows C>T transitions characteristic of repair errors at UV-induced pyrimidine dimer sites observed in melanoma (Fig. 2f), S3 is
characteristic of C>A transversions indicative of cigarette smoking. Notably, the S3 signature associated with cigarette smoking, though predominant in lung
cancer, was not observed in MPMs. Clustering cancers by mutational signatures showed that the mutational processes in MPMs were closer to those observed
in ovarian cancers (Fig. 2h).
3. RESULTS: Significantly mutated mesothelioma genes
▸ To further assess the relevance of the mutated genes, they
identified ten significantly mutated MPM-associated genes (Fig. 3a)
that included BAP1, NF2, TP53, SETD2, DDX3X, ULK2, RYR2, CFAP45,
SETDB1 and DDX51 (Fig. 3a–g). Among the significantly mutated
genes, only BAP1, NF2 and TP53 have been reported in MPM.
Consistent with previous reports, they found tumor suppressors
BAP1 and NF2 to be mutated in 23% (46/202) and 19% (38/202) of
the samples, respectively.
▸ Tumor suppressor TP53 was mutated in 8% (17/202) of MPMs, a
number higher than previously reported (3/53)12 but lower than in
studies involving a smaller sample size (4/20) or (2/22). Notably,
TP53 mutations were absent from the epithelioid subtype. Further,
patients with TP53 mutations showed lower overall survival than
those with wild-type TP53 in this cohort, indicative of the
aggressive nature of TP53-mutant MPMs (Fig. 3h).
4. RESULTS: multiple molecular mechanisms lead to activation and inactivation of genes
▸ In addition to mutations and expression changes, they
assessed 95 MPMs for copy number alterations using 2.5M
Illumina SNP array and/or whole-genome data. They found
regions of recurrent copy loss that include genes such as
BAP1, NF2, CDKN2B, LATS2, LATS1 and TP53 (Fig. 4a),
consistent with previous reports. Copy number loss
correlated with loss of expression in these genes
(Supplementary Fig. 12a).
Supplementary Fig. 12a
4. RESULTS: multiple molecular mechanisms lead to activation and inactivation of genes
▸ Driver gene fusions have been reported in multiple cancers. They analyzed MPM RNA-seq data for presence of gene fusions. Overall, They identified 43 gene
fusions in 22 samples (supplementary Table 10). Although recurrent gene fusions are usually associated with oncogenic activation, they identified many
recurrent fusions involving tumor suppressor genes. They found 13 fusions in NF2, 7 in BAP1, 8 in SETD2, 7 in PBRM1, 2 in PTEN and 6 in other genes (Fig. 4).
RNA Tumor ID
Gene 5'
Gene 3'
Read Evidence
Orthogonal Validation
Chromosome # 5'
Position 5'
Chromosome # 3'
Position 3'
Fusion Protein Length
NF2 predicted fusions
17865241
NF2
THRB
3
No
22
30035201
3
24270492
461
17865267
NF2
IFT140
3
No
22
30000101
16
1561151
106
9259677
NF2
CABP7
5
Validated by sanger
22
30038274
22
30123651
207
9259684
NF2
PIEZO2
13
Validated by sanger
22
30000101
18
10807272
95
9259687
NF2
OSBP2
11
No
22
30061053
22
31090195
1041
9259690
NF2
PI4KA
17
Validated by sanger
22
30000101
22
21193019
2088
9259709
NF2
RHOT1
36
Validated by sanger
22
30038274
17
30530909
396
9259750
NF2
NFATC1
9
No
22
30000101
18
77170403
117
17865239
EWSR1
NF2
14
Validated by sanger
22
29664338
22
30032740
13
17865285
D2HGDH
NF2
3
No
2
242695429
22
30077428
459
9259677
CABP7
NF2
9
Validated by sanger
22
30123794
22
30067815
100
9259683
GSTT1
NF2
25
Validated by sanger
22
24384120
22
30032740
46
9259709
RHOT1
NF2
13
Validated by sanger
17
30529919
22
30050646
890
supplementary Table 10
4. RESULTS: multiple molecular mechanisms lead to activation and inactivation of genes
▸ To identify structurally altered transcripts in MPM, they performed de novo prediction of splice variants from RNA-seq data. They identified 26 candidate
cancer-specific splice alterations in 14 genes frequently mutated in MPM or other cancers (Supplementary Table 11).
Tumor ID
Symbol
Consequence
M680PT
M53PT
M691PT
M35PT
M57PT
M60PT
M632PT
M683PT
606PT
661PT
M91PT
M632PT
M82PT
M680PT
M59PT
651PT
M635PT
M4PT
M46PT
M79PT
M676PT
M3PT
M669PT
M61PT
667PT
M635PT
ABL1
ABL1
BAP1
BAP1
BAP1
BAP1
BAP1
BAP1
CDKN2A
DDX3X
DDX3X
FBXW7
LATS2
LATS2
LATS2
LATS2
MSH2
NCOR1
NF2
NF2
PTPN14
RB1
SETD2
SETD2
SETDB1
STAG2
5' truncation (novel exon)
5' truncation (novel exon)
in-frame deletion
frame-shift insertion
in-frame deletion
frame-shift deletion
in-frame deletion;frame-shift deletion
frame-shift deletion
3' truncation
frame-shift deletion
in-frame deletion
in-frame deletion
frame-shift deletion
frame-shift deletion
in-frame deletion
in-frame deletion;frame-shift deletion
in-frame deletion
5' truncation
in-frame deletion
in-frame deletion
5' truncation
in-frame deletion
in-frame deletion
frame-shift deletion
frame-shift deletion
frame-shift deletion
Supplementary Table 11
Abstract
SGSeq provides a framework for analyzing annotated and previously uncharacterized splice events from
RNA-seq data. Input data must be provided as BAM files containing RNA-seq reads aligned against a
reference genome. Exons and splice junctions are predicted from aligned reads and are assembled into a
genome-wide splice graph. Splice events are identified from the graph and quantified using reads spanning
event boundaries. This vignette provides an introduction to SGSeq, including splice event prediction,
quantification, annotation and visualization.
Package ‘SGSeq’
▸ Analysis based on annotated transcripts
They use the UCSC knownGene table as reference annotation,
which is available as a Bioconductor annotation package
TxDb.Hsapiens.UCSC.hg19.knownGene.
▸ Analysis based on de novo prediction
Instead of relying on existing annotation, SGSeq can predict features from BAM files directly.
4. RESULTS: multiple molecular mechanisms lead to activation and inactivation of genes
Figure 4. (b,c) Mutations, splicing variants and gene fusions that
alter NF2 (b) and SETD2 (c) in MPMs, mapped to domain
architecture. RNA-seq coverage and junction reads supporting
splice alterations in NF2 and SETD2 are shown for patients M79PT
and M669PT, respectively. FERM_N, N-terminal ubiquitin-like
structural domain of the FERM domain; FERM_central, FERM
central domain; FERM_PH, C-terminal PH-like domain; ERM_C,
Ezrin/radixin/moesin, C-terminal. (d) Recurrent splicing changes in
ABL1 observed in patients M680PT, M53PT. The exon-intron
structure of ABL1 and start and stop (*) codons are shown. Top
schematic, ABL1 architecture resulting from the normal isoform;
bottom schematic, alternate isoform encoding the kinase domain
alone. SH2, Src homology 2; SH3, Src homology 3; TyrKc, tyrosineprotein kinase, catalytic domain; FABD, F-actin binding. (e) Quilt
plot of alterations in indicated genes. Each column represents a
sample.
5. RESULTS: Mutations in the splicing factor SF3B1 are associated with specific alterations in mRNA splicing
▸ Alterations in spliceosomal components that affect splicing have
been reported in multiple cancers. Using targeted sequencing, they
identified three tumors with missense mutations in SF3B1, which
encodes a splice factor with a role in branch-point recognition and U2snRNP assembly (Fig. 5a). Differential splicing analysis showed that
these tumors have distinct splicing patterns (Fig. 5b). RNA-seq analysis
in MPMs identified 177 splice events that differ in tumors with SF3B1
mutation, compared to tumors with wild-type SF3B1 (Fig. 5b).
▸ A previous RNA-seq study of cancer cell lines identified an SF3B1
hotspot mutation encoding p.Lys700Glu in NCI-H2595, a mesothelioma
cell line. They assessed splice variant usage in the RNA-seq data and
found that NCI-H2595 showed splice alterations similar to those
observed in SF3B1-mutant tumors (Fig. 5b).
▸ The most common type of splice alteration observed in SF3B1mutant tumors was a change in the 3′ splice site to an upstream (n = 82,
46%) or downstream 3′ splice site (n = 26, 15%; Fig. 5c). Consistent
with previous studies, most 3′ splice site alterations involved increased
usage of an alternative 3′ splice site located 10–30 bp upstream (Fig.
5d).
▸ (Fig. 5e,f) Top, SF3B1-associated splice variants in MAP3K7 (e) or
SMURF2 (f) in transcript isoforms detected in NCI-H2595. Middle,
average per-base read coverage and junction counts after normalizing
for read count; splice variants with increased usage in SF3B1 mutants
are highlighted in red. Bottom, schematic of effect of cryptic splice site
usage on protein architecture. STK, serine/threonine/dual-specificity
protein kinase, catalytic domain; HECT, domain homologous to E6-AP C
terminus; C2, protein kinase C conserved region 2.
5. RESULTS: Mutations in the splicing factor SF3B1 are associated with specific alterations in mRNA splicing
6. RESULTS: Integrated analysis of pathway alteration observed in MPM
▸ MuSiC pathway analysis identified several significantly altered pathways.
Integrated analysis of the genomics data identified Hippo, mTOR, histone
methylation, RNA helicases and p53 signaling pathways (Fig. 6) to be altered
in MPMs.
DISCUSSION
▸ In this study, they performed a comprehensive genomic analysis of transcriptomes and exomes from 216 MPMs. Using RNA-seq, we
identified four distinct molecular MPM subtypes. They found differences in mutational rates and signatures between MPMs and other
cancers. These differences, along with mutations, expression profiles and gene fusions, have the potential to improve MPM diagnosis,
which currently relies on immunohistochemistry and is clinically challenging.
▸ These results substantially expand on previous genomic studies and provide a comprehensive genomics profile of mesothelioma.
Incorporating genomic analysis for the detection of actionable alterations as part of MPM patient care will help in developing rational
individualized therapy.