Genomic analyses reveal recurrent mutations in epigenetic modifiers and the JAK-STAT pathway in Sezary syndrome

Abstract

Sézary syndrome (SS) is an aggressive leukaemia of mature T cells with poor prognosis and limited options for targeted therapies. The comprehensive genetic alterations underlying the pathogenesis of SS are unknown. Here we integrate whole-genome sequencing (n=6), whole-exome sequencing (n=66) and array comparative genomic hybridization-based copy-number analysis (n=80) of primary SS samples. We identify previously unknown recurrent loss-of-function aberrations targeting members of the chromatin remodelling/histone modification and trithorax families, including ARID1A in which functional loss from nonsense and frameshift mutations and/or targeted deletions is observed in 40.3% of SS genomes. We also identify recurrent gain-of-function mutations targeting PLCG1 (9%) and JAK1, JAK3, STAT3 and STAT5B (JAK/STAT total ∼11%). Functional studies reveal sensitivity of JAK1-mutated primary SS cells to JAK inhibitor treatment. These results highlight the complex genomic landscape of SS and a role for inhibition of JAK/STAT pathways for the treatment of SS.

ARTICLE

Received 6 Jul 2015 | Accepted 25 Aug 2015 | Published 29 Sep 2015

Mark J. Kiel1,*, Anagh A. Sahasrabuddhe1,*,w, Delphine C.M. Rolland2,*, Thirunavukkarasu Velusamy1,*, Fuzon Chung1, Matthew Schaller1, Nathanael G. Bailey1, Bryan L. Betz1, Roberto N. Miranda3, Pierluigi Porcu4, John C. Byrd4, L. Jeffrey Medeiros3, Steven L. Kunkel1, David W. Bahler5,Megan S. Lim2 & Kojo S.J. Elenitoba-Johnson2,6

Szary syndrome (SS) is an aggressive leukaemia of mature T cells with poor prognosis and limited options for targeted therapies. The comprehensive genetic alterations underlying the pathogenesis of SS are unknown. Here we integrate whole-genome sequencing (n 6),

whole-exome sequencing (n 66) and array comparative genomic hybridization-based copy-

number analysis (n 80) of primary SS samples. We identify previously unknown recurrent

loss-of-function aberrations targeting members of the chromatin remodelling/histone modication and trithorax families, including ARID1A in which functional loss from nonsense and frameshift mutations and/or targeted deletions is observed in 40.3% of SS genomes. We also identify recurrent gain-of-function mutations targeting PLCG1 (9%) and JAK1, JAK3, STAT3 and STAT5B (JAK/STAT total B11%). Functional studies reveal sensitivity of JAK1-mutated primary

SS cells to JAK inhibitor treatment. These results highlight the complex genomic landscape of SS and a role for inhibition of JAK/STAT pathways for the treatment of SS.

1 Department of Pathology, University of Michigan Medical School, Ann Arbor, Michigan 48109, USA. 2 Department of Pathology and Laboratory Medicine, Perelman School of Medicine at University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA. 3 Department of Hematopathology, The University of Texas MD Anderson Cancer Center, Houston, Texas 77030, USA. 4 Division of Hematology, Department of Internal Medicine, The Ohio State University, Columbus, Ohio 43210, USA. 5 Department of Pathology, The University of Utah Health Sciences Center, Salt Lake City, Utah 84112, USA. 6 Center for Personalized Diagnostics, Perelman School of Medicine at University of Pennsylvania., Philadelphia, Pennsylvania 19104, USA. * These authors contributed equally to this work. w Present address: Department of Biotechnology, Pandit Ravishankar Shukla University, Raipur, India. Correspondence and requests for materials should be addressed to M.S.L. (email: mailto:[email protected]

Web End [email protected] ) or to K.S.J.E.-J. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2015 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms9470 OPEN

Genomic analyses reveal recurrent mutations in epigenetic modiers and the JAKSTAT pathway in Szary syndrome

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470

Szary syndrome (SS) is an aggressive mature T-cell leukaemia with a median 5-year survival rate of o20% (refs 1,2). The skin is almost always affected, whereas in

advanced forms of SS lymph nodes and other visceral organs can be involved3. Therapy often involves extracorporeal ultraviolet phototherapy and single-agent cytotoxic chemotherapeutic agents such as methotrexate4. However, despite aggressive therapies, initial response rates are poor and disease recurrence is common5.

To date, efforts to identify genes recurrently targeted by mutation in SS genomes have been largely targeted68, or otherwise limited to a few index samples9,10. The comprehensive genomic landscape of SS has not been explored and opportunities for targeted therapies based on specic genetic mutations have not been fully exploited. To gain insights into the genetic alterations underlying the pathogenesis of SS, we integrated whole-genome sequencing (WGS) and whole-exome sequencing (WES) in combination with high-resolution copy-number variant (CNV) analysis on a large cohort of well-characterized cases of SS. Our studies reveal recurrent mutations targeting epigenetic modiers and JAKSTAT pathway in SS.

ResultsWGS reveals genomic complexity of SS. To obtain a genome-wide view of the molecular genetic alterations underlying SS at a nucleotide resolution level, we performed WGS of highly enriched (490%) pure tumour cells from six cases that fullled established diagnostic criteria including characteristic cytologic, immunophenotypic and karyotypic features3. The data highlight the structural genomic complexity of SS (Fig. 1; comprehensive structural alteration data from WGS can be found in Supplementary Data 1). This analysis revealed a total of 1,010 inter- or intrachromosomal translocations in the six SS genomes (average 16843 translocations per genome). No recurrent

translocations or gene fusions were identied in these six SS cases. However, among 42 potential fusion genes (Supplementary Data2), several noteworthy candidates were identied that may contribute to SS disease pathogenesis in selected cases.

Novel translocations identied in SS included juxtaposition of the N terminal of receptor tyrosine kinase TPR and the C terminal of hepatocyte growth factor receptor MET previously identied to be oncogenic in gastric cancers11; the N terminus of v-myb avian myeloblastosis viral oncogene homologue-like 1 MYBL1 fused with the C terminal of thymocyte high-mobility group box protein of TOX previously implicated in the pathogenesis of mycosis fungoides/SS12; and the N terminus of the Hsp40 homologue DNAJC15 and the C terminus of zinc-nger protein ZMYM2 involved in the translocation t(8;13)(p11;q12), which fuses FGFR1 with ZMYM2 associated with 8p11 myeloproliferative disorders13. Other noteworthy translocations separately targeted the MEIS1 homeobox gene previously implicated in leukemogenesis14,15, the transcription factor IKZF2, and the serinethreonine kinase VRK2 (Supplementary Data 2). WGS also revealed a novel reciprocal translocation event involving chromosomes 3 and 10 leading to interruption of coding elements of CBLB between exons 13 and 14 by insertion of the coding elements of ZEB1 beginning at exon2. This translocation is predicted to result in a fusion gene composed of the N-terminal portion of CBLB (residues 1653 including CBL-PTB, UBA, RING EF-hand and SH2 domains) with the entire protein structure (including all seven zinc-ngers and DNA-binding domains) of ZEB1, a zinc-nger negative transcriptional regulator of interleukin-2-stimulated cytokine signalling in T cells16. Also of note, a translocation event involving elements of the histone-lysine N-methyltransferase EZH2 on chromosome 7 juxtaposed to elements upstream of the forkhead box protein FOXP1 on chromosome 3 was observed in one SS genome. In addition to the large structural variations, WGS analyses revealed recurrent aneuploidies in SS genomes

a

X

Y 1

b

X

Y 1

c

X

Y 1

22

21

20

19

18

17

16

15

14

13

12

11

10

22

21

20

19

18

17

16

15

14

13

12

11

10

22

21

20

19

18

17

16

15

14

13

12

11

10

2

2

2

FOXP1

3

3

3

4

4

4

6

5

6

5

6

5

7

7

8

8

8 7 EZH2

9

9

9

Y

d

X

e

1

22

21

20

19

18

17

16

15

14

13

12

11

10

X

Y 1

Y 1

22

21

20

19

18

17

16

15

14

13

12

11

10

f

22

21

20

19

18

17

16

15

14

13

12

11

10

X

2

2

2

CBLB

3

3

3

4

4

4

6

5

6

5

6

5

7

7

7

8

8

8

9

9

9

ZEB1

Figure 1 | Structural alterations in six Szary syndrome genomes identied by whole-genome sequencing. Circos diagrams for six SS genomes: panels a (case A02), b (case B02), c (case C02), d (case D02), e (case G01) and f (case H01) depict the chromosomes arranged circularly end to end with each chromosomes cytobands marked in the outer ring. The inner ring displays copy-number data inferred from whole-genome sequencing with blue indicating losses and red indicating gains. Within the circle, rearrangements are shown as arcs with intrachromosomal events in red and interchromosomal translocations in blue. The two representative translocations involving EZH2-FOXP1 and ZEB1-CBLB are highlighted in c and d, respectively.

2 NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470 ARTICLE

including trisomy 8 (1/6 cases) and monosomy 10 (2/6); and loss of 17p and/or isochromosome 17 (5/6). Interestingly, losses of chromosome 1p, reported to be recurrently deleted in several aCGH studies1722, were identied in 4/6 genomes with a narrowly restricted region spanning 1p36.21-1p35.3. To date, no candidate gene within this region has been identied and implicated in SS pathogenesis.

CNV and WES analyses reveal recurrent loss-of-function loci. To identify regions of recurrent gain or loss at high resolution and thereby dene genes targeted by numerical aberrations, we performed high-resolution aCGH on a total of 80 SS samples (Supplementary Fig. 1). This analysis revealed a number of recurrent gains (Supplementary Fig. 2) and losses (Supplementary Fig. 3) of chromosomal material (comprehensive CNV data is presented in Supplementary Data 3). These disruptions included previously reported alterations such as recurrent gains of chromosome 8 (B25%) and loss of the short arm of chromosome 17 (B40%), which frequently also involved simultaneous gains of the long arm of chromosome 17 (B25%) of SS cases (Supplementary Fig. 4). A high-frequency of deletions of the short arm of chromosome 19 was observed. The data also dened specic regions of recurrent aneuploidy at chromosome 1p36.11, 3p21.31, 9p21.3, 10p11.22 and 13q14.2 (Supplementary Fig. 3, arrowheads). To precisely identify single-gene candidates within these chromosomal loci that may contribute to SS pathogenesis through gene mutation, we performed WES on 60 additional SS genomes (comprehensive mutational data presented in Supplementary Data 4).

Our WES data demonstrated that genes within regions of recurrent deletions previously demonstrated to be important in SS pathogenesis (TP53 on chromosome 17, CDKN2A on chromosome 9 and PTEN on chromosome 10) also were affected by deleterious mutations. Specically, WES identied loss-of-function mutations in CDKN2A (p.W15X nonsense mutation), a frameshift mutation in PTEN in the C2 functional domain and numerous nonsense, frameshift and splice junction mutations in TP53 (Supplementary Fig. 5). To efciently identify novel genes contributing to the pathogenesis of SS, we selected those genes which showed both a high rate of deletion (410% of 80 SS genomes) as well as genes with at least one known deleterious mutation (either frameshift, nonsense and/or splice donor or acceptor mutation among the 66 SS genomes sequenced). This approach revealed several candidate genes involved in the pathogenesis of SS.

Loss of function of ARID and SMARC family proteins. Examination of the most frequently restricted region of recurrent deletion, the 1p36.11 locus, revealed 18 genes with at least one deleterious mutation (Supplementary Data 5). Of these, ARID1A exhibited the highest frequency of disruption and/or mutation with 27/80 genomes involved by deletions and 7/66 showing novel mutations including a frameshift mutation at residue 1449 within the GR-binding domain where the mutations seemed to co-localize; Figs 2a and 3; Supplementary Fig. 6). Disruption of ARID1A by either deletional events or mutation or both were seen in 25/62 (40.3%) SS genomes on which both WES and aCGH analyses were performed. Loss-of-function mutations in ARID1A have been implicated in the pathogenesis of other hematopoietic and non-haematopoietic malignancies23, but not in SS.

Another member of the ARID family of chromatin remodelling genes, ARID5B, was also implicated as a candidate because deletions of region 10q21.2 were observed in 23/80 genomes and point mutations in 3/66 SS genomes. The latter group included a nonsense mutation p.K239X (Fig. 3; Supplementary Fig. 7.

ARID5B is recurrently mutated in endometrial malignancies24 and deleted in a fraction of paediatric B-cell acute lymphoblastic leukaemia25.

ARID family members play a critical role in the SWI/SNF complex where they exhibit helicase and ATPase activities and regulate transcription of genes by altering the surrounding chromatin structure26. SWI/SNF family members, including the SMARC group, interact with proteins encoded by the ARID family member genes27. Loss of SWI/SNF complex member SNF5 leads to T-cell lymphoma at high penetrance levels28. Interestingly, our studies revealed that SMARCC1 at 3p21.31 was recurrently targeted by deletions (o4 Mb) in 17/80 (21.3%)

of SS genomes (Supplementary Fig. 3). Strikingly, deletions and/ or mutations in either ARID1A, ARID5B or SMARCC1 were identied in 38/62 (61.3%) of SS genomes in which both WES and aCGH were performed. WES revealed mutations in other members of the ARID and SMARC family, namely, ARID1B, ARID4A and ARID2, which also included a case with a nonsense mutation (ARID2 p.Q1462X). Moreover, ARID3A and SMARCA4, recently found to be mutated in small cell carcinoma of the ovary, are both located on the short arm of chromosome 19, which is recurrently deleted in 29/80 and 18/80 SS genomes, respectively. Altogether, deletions and/or mutations of at least one of the 10 ARID or 13 SMARC family genes were identied in 91.9% of SS genomes (Fig. 3).

ZEB1, in addition to being implicated in a fusion event involving CBLB as described above (Fig. 1d; Table 1), was also recurrently targeted by deletions of chromosomal material (29/80 SS genomes) as well as deleterious mutations (7/66 SS genomes) including ve frameshift mutations in four SS genomes (Fig. 2b). In total, 28/62 (45.2%) of SS genomes showed evidence of loss-of-function alterations of ZEB1 strongly implicating this gene as a tumour suppressor involved in the pathogenesis of SS. This concept is supported by the demonstration that TCF8/ZEB1 mutant mice frequently develop a CD4-positive T-cell lymphoma/leukaemia with a median onset of 30 weeks29,30. Further, ZEB1 has been implicated as a candidate tumour suppressor gene in adult T-cell leukaemia/lymphoma30. ZEB1 directly interacts with the trithorax group protein component SMARCA4 (also known as BRG-1) and recruits histone deacetylases HDAC1 and HDAC2 (refs 16,31). Our results also showed that the cyclic AMP-response element-binding protein-binding protein, which is frequently mutated in acute lymphoblastic leukaemia32 and critical for trithorax group protein activity33, was recurrently altered by deletions in 19/80 and mutations in 7/66 SS genomes including a p.Q1108X nonsense mutation and several frameshift mutations (Supplementary Data 4).

MLL and SETDB and KDM6B mutations. The results also implicated members of the trithorax group histone methyltransferases, particularly MLL2, known to be frequently mutated in a variety of human malignancies3436. These genes were deleted in 11 of 80 SS genomes. MLL2 was targeted in 7/66 SS genomes including one with a monoallelic nonsense mutation(p.Q4219X) and another genome with bialleleic nonsense mutations (p.S624X and p.S687X). MLL4 was mutated in 12/66 of SS genomes including one frameshift mutation (p.P622fs; Fig. 3; Table 1). In addition, MLL3 was targeted in 39/66 SS genomes, including four genomes with frameshift or nonsense mutations (Fig. 3). SETD family histone methyltransferases were involved by recurrent alterations including SETD1A deletions in 14/80 SS genomes and mutations in 4/66 genomes including one frameshift (Fig. 3; Supplementary Fig. 8); SETD1B deletions in 12/ 80 and mutations in 11/66 SS genomes including one frameshift and one nonsense mutation (p.E66X; Fig. 3; Supplementary

NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470

d

a 10p11.22

i 0%

5% 10% 15% 20% 25% 30% 35% 40% 45% 50%

ii

iii

b 17p

i 0%

10%

20%

30%

40%

50%

60%

ii

c 10q21.2

i 0%

5% 10% 15% 20% 25% 30% 35% 40% 45%

ii

1p36.11

i 0%

5% 10% 15% 20% 25% 30% 35% 40%

ii

iii

iii

iii

ARID1A

ZEB1

NCOR1

TET2

Figure 2 | Recurrent loss-of-function alterations of chromatin remodelers and DNA modier genes. Evidence of loss of function in Szary syndrome (SS) genomes including (i) proportion of total genome locus affected, (ii) depiction of individual genomes of SS with very narrowly dened deletions specically targeting the highlighted gene and (iii) deleterious mutations including frameshift and missense mutations conrmed to be somatic by Sanger sequencing for (a) ARID1A, (b) ZEB1, (c) NCOR1 and (d) TET2.

Fig. 9); SETDB2 deletions in 18/80 and mutations in 2/66 SS genomes including two nonsense mutations (p.R588X) in separate genomes; and SETD6 deletions in 9/80 and mutations in 2/66 SS genomes including one frameshift mutation (Fig. 3). Altogether, deletions and/or mutations in MLL, MLL2, MLL3, MLL4, SETD1A, SETD1B, SETDB2 and SETD6 were identied in67.7% of SS genomes. Moreover, deletions of the histone-lysine demethylase KDM6B on chromosome 17p were observed in 38/80 SS genome, whereas mutations were seen in 11/66 SS genome including one frameshift mutation p.P471fs.

In addition, we identied genetic alterations affecting other epigenetic modulators including the histone deacetylase NCOR1 on the short arm of chromosome 17 including deletions in 38/80 and mutations in 9/66 including one frameshift mutation(p.Y1997fs) and two nonsense mutations (p.Q600X andp.Q2005X; Figs 2c and 3). NCOR1 is implicated in the pathogenesis of acute lymphoid leukaemia and non-haematopoietic malignancies37,38. Six of 62 SS genomes showed simultaneous mutation and deletion events suggesting biallelic loss of function of NCOR1 (Fig. 3). These results indicate that disruption of homeostatic histone modication may contribute to the pathogenesis of SS.

Further implicating a role for deregulation of epigenetic modiers in SS was our identication of alterations targeting multiple members of the DNA methyltransferase family, including DNMT3A, which has been implicated in the development of acute myeloid leukaemia39. DNMT3A was deleted in 14/80 and mutated in 5/66 SS genomes, including the novel nonsense mutation p.Y394X. In addition, DNMT3B was deleted in 9/80 and mutated in 5/66 SS genomes including the nonsense mutationp.R223X (Fig. 3). Additional genes involved in DNA methylation such as the demethylases, TET1 and TET2, recently implicated in the pathogenesis of myeloid malignancies40,41, also showed evidence of loss-of-function alterations. Strikingly, among the SS genomes showing deletion of any portion of chromosome 10 were eight cases with narrowly focused deletions (o0.5 Mb)

involving the TET1 gene. Moreover, sequencing identied recurrent mutations affecting TET2 in 7/66 SS genomes including ve nonsense mutations and a disrupted splice acceptor mutation (Fig. 2diii). Altogether, deletions and/or mutations in DNMT3A, DNMT3B, TET1 or TET2 were identied in 74.2% of SS genomes (Fig. 3). Taken together, these results support a critical role for deregulated DNA methylation in the pathogenesis of SS.

4 NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470 ARTICLE

Szary samples

Proportion affected

H01

A02

B02

C02

D02

SS1

SS2

SS3

SS5

SS6

SS7

SS8

SS9

SS10

SS11

SS12

SS14

SS15

SS16

SS17

SS19

SS20

SS23

SS25

SS26

SS27

SS28

SS29

SS30

SS31

SS32

SS33

SS34

SS35

SS36

SS37

SS38

SS40

SS42

SS44

SS45

SS47

SS48

SS49

SS50

SS54

SS56

SS59

SS60

SS61

SS62

SS64

SS75

SS76

SS77

%0

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

I I I

o o

o

Chromatin modifiers

and DNA modifiers

Histone modofiers

o o

x

o

o

o

o o o o

o

o

x

o

o o

o

o

o

o

o

o

o

o

o

x

x

x

x

o

x

o

x

o

o

o

o

o

o

o

o

x

o

o

o

x

o

o

o o

o

o

o

xx x x

x x

x

o

o

o

o

o o

x

o o

o

o

oo o

o o o

o

x

x x

o

o oo o oo

o

x x

o

o o

o

x

o

o

x

o

o o

x

I

o

o

o

o

o

o

o

o

o

x

o

o

o

o

x

x

+

JAK-

STAT

ARID1A

ARID18

ARID2

ARID3A

ARID3B

ARID3C

ARID4A

ARID4B

ARID5A

ARID5B TET1

TET2

TET3 DNMT1

DNMT3A

DNMT3B

DNMT3L

NCOR1

NCOR2 MLL

MLL2

MLL3

MLL4

MLL5 SETD1A

SETD1B

SETD2

SETD3

SETD4

SETD5

SETD6

SETD7

SETD8

SETDB1

SETDB2 JAK1

JAK3

STAT3

G01

+

+

+

+

I

Indel mutation

Missense mutation

Frameshift or nonsense mutation

Homozygous or heterozygous deletion

Heterozygous deletion and missense mutation

Heterozygous deletion and deleterious mutation

Gain-of-function mutation +

o

x

o

x

Figure 3 | Epigenetic modiers and JAKSTAT pathway components are targeted by aneuploidy and mutations. Summary of deletions and/or mutations in SS genomes affecting genes involved in epigenetic modications. The loss-of-function of epigenetic modiers including ARID family members, TETs, DNMTs, NCOR1, MLLs and SETdomain containing genes are indicated in blue while gain-of-function mutations in JAK and STAT family members are indicated in red.

Loss-of-function alterations in cell cycle regulators. Our integrated approach highlighted involvement in SS of several novel candidate gene families involved in cell cycle regulation whose members were recurrently altered through both deletions and mutations. Loss of the well-known tumour suppressor gene CDKN1B on 12p13.1 was identied in 12/80 SS genomes, including a case showing a deletion and a single p.W76X nonsense mutation. In addition, the RB1 gene locus at 13q14.2 was deleted in 21/80 SS genomes with a number of narrowly restricted deletions (o0.7 Mb; Supplementary Fig. 5).

Gain-of-function mutations target PLCG1 and CARD11. PLCG1 is an important mediator in the T-cell receptor/CD3-NFAT signalling axis critical for T-cell activation and differentiation42.We identied recurrent gain-of-function mutations in PLCG1 including p.S345F (cases SS26 and SS31) affecting the PLCx catalytic domain and p.S520F (case SS84). In addition, we identied three additional mutations p.R1158H (case SSG01),p.R1163K (case SS9) and p.VYEEDM1161V (case SS86) tightly clustered in the C2 domain, which is a calcium-dependent membrane targeting module involved in signal transduction (Supplementary Fig. 10a).

The multidomain scaffold protein CARMA encoded by the CARD11 gene is a positive mediator for the NF-kB activation involved in antigen receptor signalling43. We identied recurrent missense mutations in CARD11 including p.M360K (case SS12),p.V266A (case SS26), p.D357A (case SS42) and p.S618F (cases SS26 and SS32; Supplementary Fig. 10b).

Overall, 6/66 (9.1%) harboured PLCG1 mutations and 4/66 (6%) harboured CARD11 mutations. These ndings suggest a role for antigen receptor signalling mediators in the pathogenesis of SS.

Mutations in JAKSTAT and Ras pathways. WES revealed somatic gain-of-function mutations affecting JAK1 (2/66, 3%), JAK3 (2/66), STAT3 (2/66) and STAT5B (2/66) altogether present in 11% of cases (Fig. 3; Table 2). Importantly, recurrent mutations of JAK1, JAK3, STAT3 and STAT5B have not been previously reported in SS and were corroborated by Sanger sequencing in all cases (Fig. 4a). The mutations were clustered in the auto-inhibitory pseudokinase domain and included variants detected previously in other malignancies or shown to lead to constitutive activation of JAKSTAT44 signalling including JAK1p.Y654F45,46 and p.L710V. By comparison, the JAK3 mutationsp.S989I and p.Y1023H were located in the kinase domain. This observation is intriguing given the recent observation that some mutations in the kinase domain of JAK3 confer interleukin-3 independence to Ba/F3 cells in vitro47. By WES we identied known gain-of-function mutations p.I498V48 (Table 2) andp.Y640F49 (Table 2) in STAT3, as well as p.N642H and p.Q706L (2/66, 3%) in STAT5B in SS. The JAKSTAT pathway mutations were mutually exclusive in all but one case (SS29) that harboured both STAT3 and JAK1 mutations. Matched constitutional DNA from normal tissue (for example, CD4-negative leukocytes from peripheral blood) conrmed somatic acquisition of these mutations (Fig. 4a; Table 2). Of additional interest, we identied mutations affecting the Ras homologue gene family member RhoA in 4/66 (6%) cases, namely, p.C16R (case SS8),p.N117K (cases SS33 and SS60) and p.K118T (case SS47). We also observed a known oncogenic mutation affecting NRAS(p.Q61K) in the SS-derived cell line Hut-78. Taken together, these mutations implicate a role for mutations affecting the JAKSTAT and RAS family pathways in the pathogenesis of SS.

JAK-mutated primary SS cells are sensitive to JAK inhibitors. To investigate the possible utility of targeting the JAKSTAT

NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470

Table 1 | Recurrent gene mutations in DNA and histone modiers.

Gene ID Consequence Amino-acid change Chr Position Ref Alt Somatic acquisition conrmed*

ARID1A SS34 Missense p.M618V 1 27,059,215 A G Yes

SS18 Missense p.P1163L 1 27,099,072 C T Yes SS57 Missense p.A1710P 1 27,106,168 G C ND SS76 Missense p.R1652Q 1 27,105,995 G A ND SS11 Missense p.R1662Q 1 27,106,025 G A ND SS84 Frameshift p.A1448fs 1 27,101,062 A AT Yes SS84 Missense p.P1885H 1 27,106,694 C A Yes ARID5B SS23 Nonsense p.K239X 10 63,850,666 A T Yes

SS29 Missense p.A375T 10 63,851,074 G A ND SS29 Missense p.V558A 10 63,851,624 T C ND SSA02 Missense p.R505W 10 63,851,464 C T Yes DNMT3A SS5 Missense p.G518D 2 25,463,562 C T Yes

SSG01 Nonsense p.Y395X 2 25,467,123 G T ND DNMT3B SS37 Missense p.R744S 20 31,393,204 G T ND

SS44 Nonsense p.R223X 20 31,379,488 C T ND SS14 Missense p.A288T 20 31,383,238 G A ND SSC02 Missense p.R435C 20 31,386,366 C T ND MLL2 SS16 Missense p.S1762T 12 49,437,685 C G ND

SS33 Missense p.D2769N 12 49,433,066 C T Yes SS38 Nonsense p.Q4219X 12 49,425,833 G A Yes SS49 Missense p.V109A 12 49,448,385 A G ND SS50 Nonsense p.S687X 12 49,445,406 G C Yes SS50 Nonsense p.S624X 12 49,445,595 G C Yes SS77 Missense p.A2925V 12 49,432,365 G A ND SS77 Missense p.L2245V 12 49,434,820 G C ND NCOR1 SS20 Missense p.P1910L 17 15,961,351 G A Yes

SS23 Nonsense p.Q600X 17 16,012,157 G A Yes SS27 Missense p.T649A 17 16,005,030 T C Yes SS31 Missense p.Q26K 17 16,097,808 G T Yes SS47 Nonsense p.Q2005 17 15,960,898 G A Yes SS76 Missense p.E1120D 17 15,983,762 C G Yes SS86 Frameshift p.Y1997fs 17 15,960,917 G GT Yes SSH01 Missense p.K178N 17 16,068,377 C G ND SETD1A SS58 Missense p.P611L 16 30,977,034 C T Yes

SS61 Missense p.R404H 16 30,976,274 G A ND SSH01 Missense p.P595S 16 30,976,985 C T ND SETD1B SS18 Missense p.P842L 12 122,252,646 C T Yes

SS27 Missense p.D920E 12 122,254,983 C G Yes SS31 Missense p.P353L 12 122,247,909 C T ND SSG01 Missense p.P383L 12 122,247,999 C T Yes SETD6 SS16 Frameshift p.M263fs 16 58,552,025 T TA Yes TET1 SS40 Missense p.T87S 10 70,332,355 C G ND

SS52 Missense p.T1421A 10 70,406,747 A G ND SS52 Missense p.D139N 10 70,332,510 G A ND SS1 Missense p.Q683E 10 70,404,533 C G Yes TET2 SS34 Nonsense p.E1492X 4 106,194,012 G T Yes

SS42 Nonsense p.Q640X 4 106,157,017 C T Yes SS50 Missense p.P174H 4 106,155,620 C A Yes SS76 Nonsense p.Q1828X 4 106,197,149 C T Yes SS9 Nonsense p.R550 4 106,156,747 C T Yes SSG01 Missense p.L307R 4 106,156,019 T G Yes ZEB1 SSC02 Frameshift p.E42fs 10 31,750,037 GAAGAAAGTGTTAC G Yes

SS50 Missense p.P106S 10 31,799,636 C T ND SS19 Missense p.Y184C 10 31,803,598 A G Yes SS61 Frameshift p.L349fs 10 31,809,511 T TA Yes SS42 Frameshift p.L528fs 10 31,810,050 TA T Yes SS47 Missense p.T823P 10 31,812,927 A C ND SSA02 Frameshift p.R988fs 10 31,815,779 - GA ND

ND, not determined*NGS and/or Sanger conformation of somatic status in cases where normal tissue was available.

pathway, two primary SS samples harbouring mutations in JAK1 (SS87 p.Y654F and SS29 pL710V) were subjected to JAK inhibition. Exposure to increasing doses of JAK inhibitor I50 led to substantial reduction in cell proliferation of primary SS cells

(Fig. 4b,c). Western blotting studies revealed inhibition of phosphorylation and activation of STAT1, STAT3 and STAT5 signalling in the SS cells harbouring both JAK1 mutations, whereas the mature T-cell lymphoma-derived cell line, HH,

6 NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470 ARTICLE

Table 2 | Recurrent mutations in JAKSTAT and PLCG1.

Gene ID Consequence Amino-acid change Chr Position Ref Alt Somatic acquisition conrmed*

JAK1 SS87 Missense p.Y654F 1 65,312,358 T A Yes

SS29 Missense p.L710V 1 65,310,560 G C Yes JAK3 SS33 Missense p.S989I 19 17,942,049 C A Yes

SS47 Missense p.Y1023H 19 17,941,341 A G Yes STAT3 SS29 Missense p.I498V 17 40,476,837 T C Yes

SS37 Missense p.Y640F 17 40,474,482 T A Yes STAT5B SS55 Missense p.N642H 17 40,359,729 T G Yes

SS86 Missense p.N642H 17 40,359,729 T G Yes PLCG1 SS26 Missense p.S345F 20 39,792,584 C T Yes

SS31 Missense p.S345F 20 39,792,584 C T Yes SS84 Missense p.S520F 20 39,794,139 C T Yes SSG01 Missense p.R1158H 20 39,802,370 G A Yes SS86 Deletion p.VYEEDM1161V 20 39,802,378 GTGTATGAGGAAGACA G Yes SS9 Missense p.E1163K 20 39,802,384 G A Yes

*NGS and/or Sanger conformation of somatic status in cases where normal tissue was available.

a

Y654F

L710V

JAK1

JAK3

A573V

S989I

Y1023H

Pseudokinase B41

Kinase SH2 STAT inhibitory

I498V

Y640F

SS

Confirmed, somatic

Previously identified, somatic

HUT78

STAT3

1

N642H

Q706L

STAT5B

d

HH Mac-1 SS87 SS29

b c

JAK + + + +

inhibitor I

100

70

100

70

55

Relative growth (% of control)

Relative growth (% of control)

STAT1 p-STAT1 (Y701)

STAT3 p-STAT3 (Y705)

STAT5 p-STAT5 (Y694)

-actin

120

120

100

70

100

100

80

80

100

70

**

Mac-1 HH Jurkat SS87 SS29

*** ***

Mac-1 HH Jurkat SS87 SS29

60

60

100

70

40

40

100

70

20

20

0

0 2 3 4 5 Log10 (JAK inhibitor I,nM)

0

48 72

Time (hours)

Figure 4 | Validation of gain-of-function of JAKSTAT pathway in Szary syndrome (SS). (a) Schematic representations of mutations in JAK1, JAK3, STAT3 and STAT5B identied in primary SS samples and the Hut-78 SS-derived cell line. (b) Doseresponse curves for two primary SS leukaemic cells (SS29 and SS87), the negative control cell lines (HH and Jurkat) and the positive control cell line (Mac-1) after 48 h of treatment with JAK inhibitor I. All experiments were performed in triplicate and error bars represent s.ds (***Po0.001, T-test). (c) Timeresponse curve for the two primary Szary syndrome leukaemic cells (SS29 and SS87), the negative control cell lines (HH and Jurkat) and the positive control cell line (Mac-1) treated with 3 mM of

JAK inhibitor I. All experiments were performed in triplicate and error bars represent s.d.s (**Po0.01 and ***Po0.001 T-test). (d) Phosphorylation of STAT1, STAT3 and STAT5 in the two primary SS leukaemic cells (SS29 and SS87), the negative control cell lines (HH and Jurkat) and the positive control cell line (Mac-1) treated with 3 mM of JAK inhibitor I for 4 h.

which does not carry mutations in the JAKSTAT axis, did not show any changes (Fig. 4d; Supplementary Fig. 12). Similarly, the primary SS cells harbouring the JAK1 mutations and the SS-derived cell line Hut-78 carrying the JAK1 p.Y654F and JAK3 p.A573V mutations were sensitive to the JAK1/2-selective inhibitor ruxolitinib. On the other hand, the JAK mutation-negative mycosis fungoides/SS-derived cell line HH and the JAK

mutation-negative T-cell line Jurkat were both unresponsive to ruxolitinib treatment (Supplementary Fig. 11). Taken together, these data indicate that pharmacological antagonism of JAK1STAT signalling inhibits growth of primary SS tumour cells. The effects of these mutations and of pharmacologic targeting of the JAK1STAT pathway are diagrammed in Fig. 4.

NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470

DiscussionThis study provides for the rst time a comprehensive genomic landscape of SS. The integrated genomic analysis of 80 SS genomes we present reveals a complex genomic landscape with numerous structural variations including translocations, loss-of-function deletions and mutations as well as actionable gain-of-function mutations. WGS showed numerous (n 42) fusion

genes including TPR-MET, MYBL1-TOX, DNAJC15-ZMYM2 and EZH2-FOXP1 which, albeit not recurrent, could contribute to the pathogenesis of SS. The WGS analyses also conrms previously observed recurrent aneuploidies, including trisomy 8, monosomy 10 and loss of 17p and/or isochromosome 17 indicating that chromosomal instability is a characteristic of this malignancy and may contribute to the poor clinical behaviour.

The data derived from this integrated genomic analyses of the SS genome show a striking prole of recurrent loss-of-function alterations in the form of deletions and inactivating mutations targeting the epigenetic modication machinery including the ARID/SMARC family chromatin remodelling complex, histone methyltransferases (MLLs and SETD1A/B) and demethylase (KDM6B), DNA methyltransferase family members and TET1/ 2/3 genes. These results highlight the substantial contribution of epigenetic deregulation in the pathogenesis of SS and provide targeted insights to study the mechanisms of sensitivity and resistance to inhibitors of histone deacetylases and DNA methyltransferases in SS patients. Recurrent loss-of-function alterations targeting the well-characterized tumour suppressor genes TP53, CDKN2A, PTEN as well as cell cycle regulators CDKN1B and RB1 were also common in SS.

Dysregulation of T-cell receptor signalling has been implicated in the pathogenesis of T-cell neoplasia51. Our studies reveal that mutational subversion of components of signalling pathways critical for T-cell activation and survival is common in SS. In this regard, recurrent somatic gain-of-function mutations involving PLCG1, JAK1/3 and STAT3/5B are a prominent feature of the SS mutational landscape. Inhibition of activated pathways, particularly those driven by oncogenic tyrosine kinases, has served as a molecularly tractable strategy for the targeted therapy of cancers. In this regard, we demonstrated that pharmacologic antagonism of JAK in primary SS samples harbouring JAK1 mutations led to signicant inhibition of tumour cell proliferation.

In conclusion, the genomic landscape of SS reveals a heterogeneous genetic portrait with frequent disruption of epigenetic modiers, cell cycle regulators and subversion of signalling modules involved in T-cell receptor, chemokine and cytokine signalling pathways critical for T-cell activation, survival and differentiation. Overall, these studies highlight the opportunities afforded by integrated genomic approaches in elucidating the molecular genetic alterations underlying the pathogenesis of SS. We anticipate that the ndings presented here will play a role in rening disease taxonomy and offer opportunities for the denition of personalized biomarkers as well as precision therapeutics for patients with SS.

Methods

Patients and samples. Clinical samples were obtained from the archives of the Department of Pathology at The University of Texas MD Anderson Cancer Center, the University of Michigan and The University of Utah Health Sciences Center. Institutional Review Board approval for access to clinical specimens and information was obtained from the University of Michigan, the University of Texas MD Anderson Cancer Center and the University of Utah. Informed consent was obtained from the patients for use of diagnostic material. All SS cases fullled pathologic criteria for diagnosis of SS according to World Health Organization classication criteria. For a given patient, samples represented either formalinxed, parafn-embedded tissue (FFPE), cryopreserved peripheral blood leukocytes or both. Where applicable, constitutional normal tissue represented either tumour-free FFPE tissue derived from the same patient or otherwise tumour-depleted

peripheral blood leukocytes generated using EasySep column enrichment and B220- and/or Mac-1-positive cell selection (Stem Cell Technologies, Inc.). Relative tumour- depletion of resultant cell suspensions was determined by ow cytometry using an antibody directed against CD4 (BD Pharmingen). DNA was extracted from both FFPE and frozen samples using QIAGEN DNA extraction kits according to manufacturers instructions.

aCGH analysis. CNVs were detected using Nimblegen whole-genome arrays containing 270,000 features (Roche Applied Science) sufcient to detect CNVs of 50 kb or greater. Arrays were prepared according to manufacturers protocol and analysed on NimbleGen MS 200 Scanner followed by data extraction, normalization and processing for segMNT analysis using NimbleScan software according to manufacturers instructions. Downstream data processing and interpretation were performed using custom-designed processing algorithms solely reliant on the data comprising output segMNT data les. Circos plots were generated to display the segMNT data. The aCGH data are available at http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292537

Web End =http://www.ncbi.nlm.nih.gov/ http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292537

Web End =bioproject/PRJNA292537 (index cases) and http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292547

Web End =http://www.ncbi.nlm.nih.gov/biopro http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292547

Web End =ject/PRJNA292547 (discovery cases).

Sequencing. For WGS, 710 mg of high-molecular-weight genomic DNA was extracted from fresh frozen tumour tissue and subjected to WGS by Complete Genomics, Inc. (CGI; Mountain View, CA). CGI performs massively parallel short-read sequencing using a combinatorial probe-anchor ligation chemistry coupled with a patterned nanoarray-based platform of self-assembling DNA nanoballs. (Library generation, read-mapping to the NCBI reference genome (Build 37, RefSeq Accession nos. CM000663-CM00686). Initial read mapping and variant calling were performed using CGAtools v1.3.0 (http://cgatools.sourceforge.net/docs/1.3.0/

Web End =http://cgatools.sourceforge.net/docs/1.3.0/). Additional downstream bioinformatic analyses of WGS data were performed using custom-designed processing routines. WGS yielded a mean of 35113 Gb mapped per sample with 97.497.8% fully called genome fraction and 97.197.7% fully called exome fraction. The median genomic sequencing depth exceeded 60 in all

samples normalized across the entire genome.

For WES, genomic DNA samples were fragmented using a Covaris S2 fragmentation system to a target size of 400 bp. The samples were end-repaired, a-tailed and custom adapters were ligated using the NEBNext DNA Library Prep kit according to the manufacturers recommended protocols. The custom adapters included 6-bp barcodes and were synthesized by Integrated DNA Technologies. After ligation, the samples were size selected to 400 bp on a 2% agarose gel and 1-mm gel slices were retained. Samples were isolated from the gel using the Qiagen QIAquick gel extraction system. Seven microlitres of each ligation product was enriched using the Phusion master mix kit and custom PCR primers with a total of 14 cycles of PCR amplication (TruSeq Universal Adapter 50-AAT-GAT-ACG

GCG-ACC-ACC-GAG-ATC-TAC-ACT-CTT-TCC-CTA-CAC-GAC-GCT-CTTCCG-ATC-T-30 and TruSeq Adapter, Index 1 6 50-GAT-CGG-AAG-AGC-ACA

CGT-CTG-AAC-TCC-AGT-CAC-ATC-ACG-XXX-XXX-TAT-GCC-GTC-TTCTGC-TTG-30). Two PCR amplications were performed for each sample. The PCR products were pooled and puried using AmpureXP beads.

Library QC was performed using the Agilent Tapestation and qPCR. On the basis of qPCR concentrations, 200 ng of each of ve samples were pooled for a total of three pools. Each pool was captured using the Nimblegen SeqCap EZ V3 Exome Enrichment Kit according to the manufacturers recommended protocols. The capture pools were combined and sequenced on the Illumina HiSeq 2,000 platform across four lanes with paired-end 100-bp reads. The average depth of coverage for exome sequencing was 30.411.9 with 495% of coding exons sequenced to a

depth of 30 or more.

Our analysis focused on variations that were not in the Database of single-nucleotide proteins (dbSNP) and/or those that had been associated previously with cancer based on information in the Catalogue of Online Somatic Mutations in Cancer (COSMIC) database. After identifying candidate mutations of interest, 1050 ng of genomic DNA from tumour and matched constitutional tissue or otherwise tumour-depleted, matched normal DNA was subjected to Sanger sequencing. To conrm somatic acquisition, tumour-depleted or tumour-free matched normal DNA was also subjected to WES to identify somatic alterations in SS (CD4-negative cells obtained following immunomagnetic enrichment of CD4-positive SS cells or skin uninvolved by tumour). Three parameters were combined to arrive at threshold criteria for designation of a variant as somatic in the absence of constitutional information. First, all of the variants identied in the present study were ltered to exclude variants in the dbSNP database. To develop a parameter for allele burden based on read counts obtained from next-generation sequencing, we leveraged our NGS database containing NGS data from 4500 genomes and modelled the allele frequency correlation between germline variants and their corresponding read counts. Depending on the read depth, the allele frequencies of germline variants varied from 40 to 60% (variant/wild-type ratio). On the other hand, variants exhibiting r25 or Z75% variant/wild-type ratios consistently and invariably represented somatic alterations. Thus, we employed these thresholds as cutoffs for designation of somatic variants. In addition, Sanger sequencing was performed to validate ndings observed for 75 variants and qualitatively and quantitatively corroborated the results of NGS. Accordingly, in scenarios where constitutional DNA was not available, we assessed the allele burdens using read counts and those with mutant-to-wild-type ratios o25 or

8 NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470 ARTICLE

475% were subjected to Sanger sequencing to corroborate the ratio from peak heights and scored as mutant when these ratios were maintained. Finally, when all parameters (that is, allele burden by NGS conrmed by Sanger and excluded by dbSNP) were fullled, the variants were presumed to be somatic. For all sequencing reactions, PCR amplication was performed using Phusion DNA polymerase (New England Biolabs) followed by conventional Sanger sequencing technology using BigDye version 3.1 chemistry run on an Applied Biosystems 3730 l DNA

Sequencer at the University of Michigan DNA sequencing Core. All sequencing reactions were performed using nested sequencing primers. Sequencing trace analysis was performed using Mutation Surveyor software. All primers were designed using a custom-developed program and purchased from Integrated DNA Technologies lyophilized in 96-well plates.

WES FASTQ sequencing data les were aligned to human NCBI Build 36 reference sequence using BWA 0.6.2 available at http://bio-bwa.sourceforge.net/

Web End =http://bio-bwa.sourceforge.net/ . Merging and deduplication were performed using Picard 1.79 available at http://broadinstitute.github.io/picard/

Web End =http://broadinstitute.github.io/picard/ . The DepthOfCoverage, CountReads, RealignerTargetCreator, IndelRealigner, BaseRecalibrator, PrintReads and UniedGenotyper functions within GenomeAnalysisTK-1.6-9 (available at https://www.broadinstitute.org/gatk/

Web End =https://www.broadinstitute.org/gatk/ ) were used to ascertain coverage and for variant calling. Resultant VCF les were annotated using snpEff v3.1 available at http://snpeff.sourceforge.net/

Web End =http://snpeff.sourceforge.net/ . Variants previously identied as somatic in malignancy were highlighted if present in the COSMIC database (http://cancer.sanger.ac.uk/cosmic

Web End =http://cancer. http://cancer.sanger.ac.uk/cosmic

Web End =sanger.ac.uk/cosmic ). Additional custom in-house bioinformatics pipelines were designed to interpret, collate, organize and display the integrated results obtained from next-generation sequencing and aCGH data open source software tools described above will be published in a subsequent manuscript and are as yet not publically available. Custom-designed data processing algorithms used to process aCGH data for display and interpretation and for automated primer design will likely not be published and will not be made publically available as numerous open source and commercially available tools exist for these analyses. The sequencing data are available at http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292537

Web End =http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292537 (index cases) and http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292547

Web End =http://www.ncbi.nlm.nih.gov/bioproject/PRJNA292547 (discovery cases).

Cell lines. Jurkat, HH and Hut-78 cell lines were obtained from the American Type Culture Collection. Mac-1 was kindly provided by Dr Marshall E. Kardin (Boston University School of Medicine, Massachusetts). All cell lines were grown in RPMI-1644 media supplemented with 10% fetal bovine serum (FBS) except Hut-78, which was grown in IMDM media supplemented with 10% FBS. No myco-plasma contamination was detected at the time of experiments.

Primary cell culture and JAK inhibitor treatment. Primary tumour cells were isolated from peripheral blood of SS patients for routine diagnosis and excess aliquots viably cryopreserved in FBS/dimethylsulphoxide at 140 C. All samples

were obtained with IRB approval from the University of Michigan and consents were obtained. Primary SS cells harbouring JAK1 gain-of-function mutation were thawed and grown overnight in RPMI cell culture medium followed by treatment with JAK inhibitor I (EMD Millipore), a pan inhibitor of JAK kinase activation, at different concentrations as indicated. A measure of 3 mM of JAK inhibitor I was used for time kinetics of cell growth assay on primary SS cells as indicated. Cell growth was determinate at 48 and 72 h in triplicate by WST-1 (Roche Diagnostics) according to the manufacturers instructions. Each assay was performed in triplicate.

Immunouorescence. For immunouorescence studies, cells were plated in chamber slides. Cells were xed in 4% formaldehyde, permeabilized with 0.5% Triton X-100 for 5 min and coimmunostained with anti-ARID1A (clone 3H14, MP Biomedicals, 1/1,000) followed by staining with Alexa Fluor 594-conjugated secondary antibodies (Molecular Probes, Eugene, OR). The slides were mounted with Vectashield mounting medium containing 40,6-diamidino-2-phenylindole (Vector

Laboratories, Burlingame, CA). After staining, cells were photographed ( 40)

using a Nikon Eclipse 80i confocal microscope.

Western blot. For western blot analysis, cells were treated for 4 h with JAK inhibitor I at 3 mM and protein lysates were subjected to western blot analysis. The following primary antibodies were obtained from Cell Signaling Technology: STAT1 (clone 9H2, 1/1,000), p-STAT1 (clone 58D6, 1/1,000), STAT3 (clone 79D7, 1/1,000), p-STAT3 (clone M9C6, 1/1,000), STAT5 (polyclonal, 1/1,000) and p-STAT5 (clone D47E7, 1/1,000). The loading quality was assessed using antibody against b-actin (clone AC-74, 1/10,000) from Sigma-Aldrich.

References

1. Agar, N. S. et al. Survival outcomes and prognostic factors in mycosis fungoides/Sezary syndrome: validation of the revised International Society for Cutaneous Lymphomas/European Organisation for Research and Treatment of Cancer staging proposal. J. Clin. Oncol. 28, 47304739 (2010).

2. Willemze, R. et al. WHO-EORTC classication for cutaneous lymphomas. Blood 105, 37683785 (2005).

3. Ralfkiaer, E., Willemze, R. & Whittaker, S. J. in WHO Classication of Tumours of Haematopoietic and Lymphoid Tissues. (eds Swerdlow, S. H., Campo, E. & Harris, N. L. et al.) 299 (Lyon, 2008).

4. Wong, H. K., Mishra, A., Hake, T. & Porcu, P. Evolving insights in the pathogenesis and therapy of cutaneous T-cell lymphoma (mycosis fungoides and Sezary syndrome). Br. J. Haematol. 155, 150166 (2011).

5. Hughes, C. F. et al. Lack of durable disease control with chemotherapy for mycosis fungoides and Sezary syndrome: a comparative study of systemic therapy. Blood 125, 7181 (2015).

6. Cristofoletti, C. et al. Comprehensive analysis of PTEN status in Sezary syndrome. Blood 122, 35113520 (2013).

7. Kiessling, M. K. et al. High-throughput mutation proling of CTCL samples reveals KRAS and NRAS mutations sensitizing tumors toward inhibition of the RAS/RAF/MEK signaling cascade. Blood 117, 24332440 (2011).

8. Steininger, A. et al. Genomic loss of the putative tumor suppressor gene E2A in human lymphoma. J. Exp. Med. 208, 15851593 (2011).

9. Andersson, E. et al. Novel TBL1XR1, EPHA7 and SLFN12 mutations in a Sezary syndrome patient discovered by whole exome sequencing. Exp. Dermatol. 23, 366368 (2014).

10. Vaque, J. P. et al. PLCG1 mutations in cutaneous T-cell lymphomas. Blood 123,

20342043 (2014).

11. Soman, N. R., Correa, P., Ruiz, B. A. & Wogan, G. N. The TPR-MET oncogenic rearrangement is present and expressed in human gastric carcinoma and precursor lesions. Proc. Natl Acad. Sci. USA 88, 48924896 (1991).

12. Huang, Y., Su, M. W., Jiang, X. & Zhou, Y. Evidence of an oncogenic role of aberrant TOX activation in cutaneous T-cell lymphoma. Blood 125, 14351443 (2015).

13. Xiao, S. et al. FGFR1 is fused with a novel zinc-nger gene, ZNF198, in the t(8;13) leukaemia/lymphoma syndrome. Nat. Genet. 18, 8487 (1998).14. Collins, C. et al. C/EBPalpha is an essential collaborator in Hoxa9/Meis1-mediated leukemogenesis. Proc. Natl Acad. Sci. USA 111, 98999904 (2014).

15. Kroon, E. et al. Hoxa9 transforms primary bone marrow cells through specic collaboration with Meis1a but not Pbx1b. EMBO J. 17, 37143725 (1998).16. Wang, J. et al. The transcription repressor, ZEB1, cooperates with CtBP2 and HDAC1 to suppress IL-2 gene activation in T cells. Int. Immunol. 21, 227235 (2009).

17. Caprini, E. et al. Identication of key regions and genes important in the pathogenesis of sezary syndrome by combining genomic and expression microarrays. Cancer Res. 69, 84388446 (2009).

18. Izykowska, K. & Przybylski, G. K. Genetic alterations in Sezary syndrome. Leuk. Lymphoma 52, 745753 (2011).

19. Lin, W. M. et al. Characterization of the DNA copy-number genome in the blood of cutaneous T-cell lymphoma patients. J. Invest. Dermatol. 132, 188197 (2012).

20. Mao, X. et al. Molecular cytogenetic characterization of Sezary syndrome. Genes Chromosomes Cancer 36, 250260 (2003).

21. Mao, X. & McElwaine, S. Functional copy number changes in Sezary syndrome: toward an integrated molecular cytogenetic map III. Cancer Genet. Cytogenet. 185, 8694 (2008).

22. Scarisbrick, J. J., Woolford, A. J., Russell-Jones, R. & Whittaker, S. J. Allelotyping in mycosis fungoides and Sezary syndrome: common regions of allelic loss identied on 9p, 10q, and 17p. J. Invest. Dermatol. 117, 663670 (2001).

23. Wu, R. C., Wang, T. L. & Shih, Ie, M. The emerging roles of ARID1A in tumor suppression. Cancer Biol. Ther. 15, 655664 (2014).

24. Cancer Genome Atlas Research, N. et al. Integrated genomic characterization of endometrial carcinoma. Nature 497, 6773 (2013).

25. Paulsson, K. et al. Genetic landscape of high hyperdiploid childhood acute lymphoblastic leukemia. Proc. Natl Acad. Sci. USA 107, 2171921724 (2010).

26. Wilson, B. G. & Roberts, C. W. SWI/SNF nucleosome remodellers and cancer. Nat. Rev. Cancer 11, 481492 (2011).

27. Wang, X. et al. Two related ARID family proteins are alternative subunits of human SWI/SNF complexes. Biochem. J. 383, 319325 (2004).

28. Roberts, C. W., Leroux, M. M., Fleming, M. D. & Orkin, S. H. Highly penetrant, rapid tumorigenesis through conditional inversion of the tumor suppressor gene Snf5. Cancer Cell 2, 415425 (2002).

29. Hidaka, T. et al. Down-regulation of TCF8 is involved in the leukemogenesis of adult T-cell leukemia/lymphoma. Blood 112, 383393 (2008).

30. Nakahata, S., Yamazaki, S., Nakauchi, H. & Morishita, K. Downregulation of ZEB1 and overexpression of Smad7 contribute to resistance to TGF-beta1-mediated growth suppression in adult T-cell leukemia/lymphoma. Oncogene 29, 41574169 (2010).

31. Aghdassi, A. et al. Recruitment of histone deacetylases HDAC1 and HDAC2 by the transcriptional repressor ZEB1 downregulates E-cadherin expression in pancreatic cancer. Gut 61, 439448 (2012).

32. Mullighan, C. G. et al. CREBBP mutations in relapsed acute lymphoblastic leukaemia. Nature 471, 235239 (2011).

33. Steffen, P. A. & Ringrose, L. What are memories made of? How Polycomb and Trithorax proteins mediate epigenetic memory. Nat. Rev. Mol. Cell Biol. 15, 340356 (2014).

NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms9470

Acknowledgements

We thank Farah Keyoumarsi for histology and specimen acquisition; Sountharia Rajendran, Nazeena Alvi, Martin Arlt and Thomas Glover for assistance with aCGH analysis; Bo Li, Ayse Ozel, Jun Z. Li, Ashwini Bhasi and James D. Cavalcoli for assistance with bioinformatics analysis of exome-sequencing data; and Constance Eaves, Brendan Tarrier and Robert Lyons for WGS and Sanger sequencing core services. K.E.J was partly supported by R01 CA136905. M.S.L. was partly supported by CA14806. This work was supported in part by the Department of Pathology at the University of Michigan.

34. Huether, R. et al. The landscape of somatic mutations in epigenetic regulators across 1,000 paediatric cancer genomes. Nat. Commun. 5, 3630 (2014).

35. Okosun, J. et al. Integrated genomic analysis identies recurrent mutations and evolution patterns driving the initiation and progression of follicular lymphoma. Nat. Genet. 46, 176181 (2014).

36. Song, Y. et al. Identication of genomic alterations in oesophageal squamous cell cancer. Nature 509, 9195 (2014).

37. Gui, Y. et al. Frequent mutations of chromatin remodeling genes in transitional cell carcinoma of the bladder. Nat. Genet. 43, 875878 (2011).

38. Stephens, P. J. et al. The landscape of cancer genes and mutational processes in breast cancer. Nature 486, 400404 (2012).

39. Ley, T. J. et al. DNMT3A mutations in acute myeloid leukemia. N. Engl. J. Med. 363, 24242433 (2010).

40. Abdel-Wahab, O. et al. Genetic characterization of TET1, TET2, and TET3 alterations in myeloid malignancies. Blood 114, 144147 (2009).

41. Delhommeau, F. et al. Mutation in TET2 in myeloid cancers. N. Engl. J. Med. 360, 22892301 (2009).

42. Huang, W. & August, A. The signaling symphony: T cell receptor tunes cytokine-mediated T cell differentiation. J. Leukoc. Biol. 97, 477485 (2015).

43. Rawlings, D. J., Sommer, K. & Moreno-Garcia, M. E. The CARMA1 signalosome links the signalling machinery of adaptive and innate immunity in lymphocytes. Nat. Rev. Immunol. 6, 799812 (2006).

44. Chen, E., Staudt, L. M. & Green, A. R. Janus kinase deregulation in leukemia and lymphoma. Immunity 36, 529541 (2012).

45. Jeong, E. G. et al. Somatic mutations of JAK1 and JAK3 in acute leukemias and solid cancers. Clin. Cancer Res. 14, 37163721 (2008).

46. Zhang, J. et al. Key pathways are frequently mutated in high-risk childhood acute lymphoblastic leukemia: a report from the Childrens Oncology Group. Blood 118, 30803087 (2011).

47. Degryse, S. et al. JAK3 mutants transform hematopoietic cells through JAK1 activation, causing T-cell acute lymphoblastic leukemia in a mouse model. Blood 124, 30923100 (2014).

48. Bashashati, A. et al. Distinct evolutionary trajectories of primary high-grade serous ovarian cancers revealed through spatial mutational proling. J. Pathol. 231, 2134 (2013).

49. Pilati, C. et al. Somatic mutations activating STAT3 in human inammatory hepatocellular adenomas. J. Exp. Med. 208, 13591366 (2011).

50. Sanda, T. et al. TYK2-STAT1-BCL2 pathway dependence in T-cell acute lymphoblastic leukemia. Cancer Discov. 3, 564577 (2013).

51. Serwold, T. et al. T-cell receptor-driven lymphomagenesis in mice derived from a reprogrammed T cell. Proc. Natl Acad. Sci. USA 107, 1893918943 (2010).

Author contributions

M.J.K., A.A.S., D.C.M.R., T.V., D.W.B., M.S.L. and K.S.J.E.-J. designed the study; M.S.L. and K.S.J.E.-J. supervised and oversaw the project and contributed to manuscript writing; P.P., J.C.B, L.J.M. and S.L.K. contributed to writing of manuscript.; M.J.K. performed WGS and WES bioinformatics analysis with assistance from B.L.B.; M.J.K. and T.V. prepared all samples for Sanger resequencing, WES and aCGH, and performed and analysed Sanger resequencing data and aCGH experiments with assistance from A.A.S.; D.C.M.R. performed primary cell culture experiments; F.C. performed uorescence microscopy; L.J.M., R.N.M., D.B. and N.G.B. provided clinical samples.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Kiel, M. J. et al. Genomic analyses reveal recurrent mutations in epigenetic modiers and the JAKSTAT pathway in Szary syndrome. Nat. Commun. 6:8470 doi: 10.1038/ncomms9470 (2015).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

10 NATURE COMMUNICATIONS | 6:8470 | DOI: 10.1038/ncomms9470 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.