3’-UTR Poly(T/U) tract deletions and altered expression of EWSR1 are a hallmark of mismatch repair-deficient cancers

: The genome-wide accumulation of DNA replication errors known as microsatellite instability (MSI) is the hallmark lesion of DNA mismatch repair (MMR)-deficient cancers. Although testing for MSI is widely used to guide clinical management, the contribution of MSI at distinct genic loci to the phenotype remains largely unexplored. Here, we report that a mononucleotide (T/U)16 tract located in the 3’ untranslated region (3’-UTR) of the Ewing sarcoma breakpoint region 1 (EWSR1) gene is a novel MSI target locus that shows perfect sensitivity and specificity in detecting mismatch repair-deficient cancers in two independent populations. We further found a striking relocalization of the EWSR1 protein from nucleus to cytoplasm in MMR-deficient cancers and that the nonprotein-coding MSI target locus itself has a modulatory effect on EWSR1 gene expression through alternative 3’ end processing of the EWSR1 gene. Our results point to a MSI target gene-specific effect in MMR-deficient cancers. Cancer Res; 74(1); 224-34. ©2013 AACR. Abstract The genome-wide accumulation of DNA replication errors known as microsatellite instability (MSI) is the hallmark lesion of DNA mismatch repair (MMR) deficient cancers. Although testing for MSI is widely used to guide clinical management, the contribution of MSI at distinct genic loci to the phenotype remains largely unexplored. Here we report that a mononucleotide (T/U) 16 tract located in the 3’ untranslated region (3’UTR) of the Ewing sarcoma breakpoint region 1 (EWSR1) gene is a novel MSI target locus that shows perfect sensitivity and specificity in detecting mismatch-repair deficient cancers in two independent populations. We further found a striking re-localization of the EWSR1 protein from nucleus to cytoplasm in MMR-deficient cancers, and that the non-protein coding MSI target locus itself has a modulatory effect on EWSR1 gene expression through alternative 3’ end processing of the EWSR1 gene. Our results point to a MSI target gene-specific effect in MMR-deficient cancers. This study identifies a novel genetic locus that is fully informative and accurate in detecting mismatch-repair deficient cancers, with major implications for routine daily practice in the clinic. be they sporadic or hereditary, with high sensitivity and specificity. Despite its location in a non-coding region of the EWSR1 gene, EWS16T contractions are associated with changes in EWSR1 expression and subcellular localization. Our findings thus directly implicate the RNA-/DNA-binding Ewing sarcoma protein, better known for its fused variants in Ewing sarcoma, in MSI-associated colorectal tumorigenesis. EWS16T were verified using a second set of primers covering the locus (Forward primer: 5’-GCATGCTCAGTATCATTGTGG-3’; Reverse primer: 5’-AGGCCGAGAAGGATGACTCT-3’) for sequencing analysis of selected samples. Sequencing reactions using the Big Dye terminator chemistry (Applied Biosystems, Foster City, CA) were performed according to the manufacturer’s protocol. genetic information. MSH2 ). We further investigated 14 sporadic MMR-deficient CRCs with MLH1 promoter hypermethylation as well as 86 sporadic MMR-proficient CRCs. Assessment of EWS16T tract length by capillary electrophoresis of fluorescently labelled PCR products revealed that all MMR-deficient cancers but none of the MMR-proficient ones displayed novel alleles, i.e. contractions or expansions at the EWS16T tract (Fig. 1). These findings were further confirmed independently in a Finnish cohort of 122 patients. In this patient cohort as well all of the 29 MMR-deficient (12 CRC, 17 gastric cancers) but none of the 93 MMR-proficient (38 CRC and 55 gastric) cancers showed EWS16T tract instability. The majority (72.7%) of somatic alterations consisted of contractions/deletions of 4 or more base pairs in MSI-associated colorectal tumorigenesis.


Abstract
The genome-wide accumulation of DNA replication errors known as microsatellite instability (MSI) is the hallmark lesion of DNA mismatch repair (MMR) deficient cancers. Although testing for MSI is widely used to guide clinical management, the contribution of MSI at distinct genic loci to the phenotype remains largely unexplored. Here we report that a mononucleotide (T/U) 16 tract located in the 3' untranslated region (3'UTR) of the Ewing sarcoma breakpoint region 1 (EWSR1) gene is a novel MSI target locus that shows perfect sensitivity and specificity in detecting mismatch-repair deficient cancers in two independent populations. We further found a striking re-localization of the EWSR1 protein from nucleus to cytoplasm in MMR-deficient cancers, and that the non-protein coding MSI target locus itself has a modulatory effect on EWSR1 gene expression through alternative 3' end processing of the EWSR1 gene. Our results point to a MSI target gene-specific effect in MMR-deficient cancers.

Précis
This study identifies a novel genetic locus that is fully informative and accurate in detecting mismatch-repair deficient cancers, with major implications for routine daily practice in the clinic.

Introduction
With an estimated population incidence of about 1 in 370, Lynch syndrome (LS, formerly known as hereditary non-polyposis colorectal cancer, HNPCC) represents the most common, autosomal dominantly inherited cancer predisposition world-wide(1), caused by germline mutations in DNA mismatch repair (MMR) genes (MLH1, MSH2, MSH6 and PMS2). Mutation carriers are at an increased risk to develop colorectal cancer (CRC) as well as a distinct spectrum of extracolonic cancers (predominantly of the endometrium, ovaries and stomach) at a young age (2).
At the cellular level, biallelic inactivation of the MMR system leads to genome-wide accumulation of DNA replication errors at specific repetitive nucleotide sequences, a condition termed microsatellite instability (MSI). MSI is the hallmark lesion of MMR deficient cancers (3), but is also observed in 10 to 20% of the sporadic colorectal, gastric and endometrial cancers. The tumour's MSI status is increasingly used to guide clinical management (3,4), because genome-wide gene expression data from (sporadic) microsatellite stable (MSS) and unstable (MSI) CRCs demonstrated that tumor development in MMR-deficient cancers follows distinct pathogenetic paths (5).
A large number of distinct genic loci affected by MSI have been described, consisting mainly of mono-and dinucleotide repeats within the 5'UTR (e.g. NR-27), introns (BAT-26) and the 3'UTR (CAT25) of specific genes (6). The functional significance of MSI at these non-coding repeat loci and how they may contribute to the pathogenic process is, however, largely unknown (7). Here we report a novel target gene locus, EWS16T, consisting of a monomorphic polythymine (16T) tract within the 3ƍUTR of the Ewing sarcoma gene (EWSR1).
We assessed this locus in two independent populations and found that EWS16T discriminates MMR-deficient from MMR-proficient cancers, be they sporadic or hereditary, with high sensitivity and specificity. Despite its location in a non-coding region of the EWSR1 gene, EWS16T contractions are associated with changes in EWSR1 expression and subcellular localization. Our findings thus directly implicate the RNA-/DNA-binding Ewing sarcoma protein, better known for its fused variants in Ewing sarcoma, in MSI-associated colorectal tumorigenesis.  (8 Lynch syndrome related CRCs and 5 sporadic CRCs both matched with their tumour free mucosa) was assessed using the TaqMan® Probe-Based Gene Expression Analysis (Applied Biosystems, Foster City, CA), and the EWSR1 probe Hs01580532_g1 (Applied Biosystems, Foster City, CA). The measurements were normalized using the HPRT1 probes Hs01003270_g1 (Applied Biosystems, Foster City, CA) (9,10), and the fold-changes in gene expression were calculated using the standard ǻǻCt method (11). All retrotranscriptase reactions, including no-template controls, were run on an Applied Biosystem 7900HT thermocycler. Each sample was tested in triplicate unless specified otherwise.

Immunohistochemistry
Several cohorts of patients were studied by immunohistochemical analysis of EWS. Briefly, the tissue samples of the following cohorts of patients were analysed: 37 colon adenomas (9 of which Lynch syndrome related), 19 CRCs (10 of which Lynch syndrome related) and a tissue microarray (12). Patient data including complete follow-up were obtained by retrospective analysis of medical records, regional tumour registries and/or treating physicians. Tissue samples were obtained by surgical or endoscopic excision. Tissue sections of 4 μm sections of paraffin embedded tissue were immunostained for primary antibody against EWS (Abcam clone 84389 dilution 1:800). Staining was carried out as previously described (13). Immunoreactivity was scored semi-quantitatively by evaluating the number of positive tumour cells over the total number of tumour cells. Nuclear immunoreactivity scores were assigned using 5% intervals and ranged from 0% to 100%.
Regarding cytoplasmic expression, the staining intensity was scored as described by Allred et al. (14). All samples were examined independently by three different pathologists (S.P., F.T. and L.T.), blinded to clinicopathological and molecular genetic information.

Cell lines
Human cell lines included in this study were purchased from American Type Culture Collection (ATCC, Rockville, MD) and authenticated by ATCC by short tandem repeat (STR) profiling and used for functional studies within 6 months after thaw from liquid nitrogen tank.
All cell lines were maintained at 37°C with 5% CO2. Kanamycin sulphate (all from Invitrogen Basel, Switzerland).
HeLa cells at earlier passages were cultured in DMEM with GlutaMAX (Invitrogen) supplemented with 10% FBS. Antibodies for Western blots against NF90, SFPQ and HuR were obtained from Santacruz

Plasmids and antibodies
Biotech and Antibody against EWSR1 for Western blots and Immunoprecipitation were obtained from Abcam.

In vitro transcription
In vitro transcription for pull down assay using S1 aptamer was performed using T7 RiboMAX™ buffer (100mM NaCl, 50mM Hepes 7.5, 0.5% NP-40 and 10mM MgCl2) with 100ug of in vitro transcribed RNA with S1 aptamer sequence and incubated at 10°C for 40 minutes in a thermomixer with intermittent shaking. The beads were washed twice with one volume of RNA wash buffer prior to incubation with the lysate. HEK293 cell pellet from 15cm 2 dish was lysed in 3ml native lysis buffer (25mM Hepes-KOH pH 7.5, 100mM KCl, 0.5% NP-40, 5mM MgCl2, 0.5mM DTT , protease inhibitor cocktail, 1mM NaF, 1 mM Na4VO4 and 300U of RNasin) for 15 minutes on ice. The lysate was subsequently gently sonicated and centrifuged to remove any cell debris. 200ug of E. coli tRNA was additional added to prevent non-specific binding of proteins to the beads. 1 ml of the lysate was added to the beads coupled to S1 aptamer RNA and also to the beads alone for no RNA control. The mixture was incubated at 4°C on rotation wheel. After 1 hour of incubation, the beads were washed thrice with native lysis buffer. The bound proteins were eluted with 100ul native lysis buffer supplemented with 25mM Biotin for 30-45 minutes at 10°C on a thermomixer with intermittent shaking. 900ul of 100% ethanol was added to the eluate and incubated at -80°C for 2 hours followed by centrifugation to precipitate the eluted proteins. The pellet was air dried and dissolved in 35ul of (SDS loading dye). Prior to loading on the Nuvex gradient gels, the samples were heated at 90°C for 5 minutes. After SDS-PAGE electrophoresis, the gel was stained with colloidal blue and bands of interest were excised and sent for mass spectrophotometry.

Poly(A) site selection assay
HeLa cells were transfected with psiCHECK-2-SPAm constructs for 24 hours. Total RNA was isolated from the HeLa cells using TriReagent separated on a 2 % Agarose gel. The bands were quantified using the ImageJ software (http://rsbweb.nih.gov/ij/).

Luciferase assays
HeLa cells were seeded in a 48 well plate one day prior to transfection. 0.2ug of plasmids

siRNA transfections
Control-siRNAs and siRNAs against hnRNPC, HuR, NF90 and SFPQ and were obtained from Santa Cruz Biotechnology. HeLa cells were reverse transfected with siRNA oligos using RNAiMAX (Invitrogen). After 48 hours the cells were transfected with EWSR1-wt-3'UTR-and EWSR1-6del-3'UTR-psiCHECK-2-SPAm-constructs for another 24 hours. The cells were subsequently harvested and split into two aliquots. One aliquot was used to assess the knockdown efficiency of siRNA using Western blot, while the other was used for RNA isolation and subsequent poly(A) site selection assay.

Statistical Analyses
For statistical analysis, the chi-square test (Ȥ 2 test) and Fisher's exact test for nonparametric variables and Student's t-test for parametric variables were used, with all probabilities reported as 2-tailed, considering a P<0.05 to be statistically significant. Calculations were performed using the software program SPSS 17.0 (IBM Corporation, Somers, NY 10589).

Ethical approval
The study is part of the so-called "Basler Studie über familiaere Tumorkrankheiten", Ref.Nr.
EK258/05 and has been approved by the "Ethikkommission beider Basel". Furthermore, written informed consent was obtained from all Lynch syndrome patients as well as from the sporadic patients.  Fig. S3). Consistently, we found that all MMR-deficient cell lines investigated (LoVo, HCT15, HCT116) carried solely mutated EWS16T alleles (contractions) without evidence for a wild-type tract allele. In contrast, MMR-proficient cell lines (HT29, SW480) as well as 37 microsatellite-stable colon adenomas and 12 MMR-proficient, MSI-low CRCs from Swiss patients had a stable EWS16T locus (Table 1). Thus, the EWS16T tract represents a novel, quasi-monomorphic MSI target locus identifying hereditary and sporadic MMR-deficient cancers with 100% sensitivity (95% CI 97-100) and specificity (95% CI 98-100). the cloned 3'UTR of EWSR1. We then generated variant constructs containing poly(T/U) tracts of variable lengths through deletion mutation. With primers that simultaneously detect both the short and the long 3'UTR isoforms in a multiplexed semi-quantitative PCR, we found that deletions in the ESW16T tract promoted the usage of the distal poly(A) site (Figure 2a-b).
We further investigated the MMR-proficient and MMR-deficient colon cancer cell lines and found that, consistent with our findings in the heterologous system, MMR deficiency is associated with higher expression of the longer EWSR1 isoform (Figure 2c). These results indicate that the EWS16T tract deletions indeed alter poly(A) site selection.

SFPQ but not HuR or NF90 mediates distal poly(A) site usage in EWSR1 pre-mRNAs with EWS16T tract deletions.
To determine the factors involved in EWSR1 poly(A) site selection, we used S1 aptamertagged, in vitro transcribed wildtype 3'UTR and a 3'UTR variants with 6U deletions in the EWS16T region to pull down the proteins that associated with these RNAs ( Supplementary   Fig. S5). In three independent experiments we reproducibly identified a set of A/U-rich element binding proteins that associate with these constructs (Figure 3A and 3B).
Interestingly, we found that nuclear factor 45/90/110 (NF45/90/110), heterogeneous nuclear ribonucleoprotein C (hnRNPC) and human antigen R (HuR) associate with the wildtype but not with the mutant 3'UTR. NF45 and NF90 have been previously shown to be part of a heterodimeric complex, nuclear factor of activated T-cells (NFAT), which is required for T-cell expression of interleukin 2, with NF110 being the larger isoform of NF90. NF90 has been shown to regulate mRNA stability and redistribution of nuclear mRNAs in the cytoplasm(19).
Conversely, the EWS16T mutant preferentially associated with the SFPQ/NONO heterodimer, which is an essential pre-mRNA splicing factor that couples splicing with

Distal poly(A) site usage is associated with decreased EWSR1 expression.
To determine if the choice in polyadenylation site may influence EWSR1 expression, we performed luciferase assays on the constructs that carried the wildtype or various EWS16T deletion variants. The results shown in Supplementary Fig. 6 indicate a significant downregulation (up to 30%) of protein levels associated with EWS16T tract deletions. We To determine the consequences of altered EWSR1 mRNA expression in vivo, we performed immunohistochemical analysis (IHC) of 10 LS-related, MMR-deficient and 9 sporadic, MMRproficient CRCs. Consistent with the data obtained at the mRNA level, the cancers displayed on average an approx. 30% reduction in EWS expression when compared to matched, tumorfree mucosa. Surprisingly however, MMR-deficient and -proficient cancers significantly differed with regard to the subcellular localization of EWS (P<0.001). Tumor-free colon mucosa and adenomas from LS and sporadic CRC patients as well as sporadic adenocarcinomas showed exclusively nuclear expression (Fig. 6). In contrast, LS-related CRCs displayed diffuse cytoplasmic EWS expression (Fig. 6). These results were subsequently confirmed by IHC analysis of a tissue microarray (TMA) containing 64 sporadic and 94 LS-related CRCs: we observed an approx. 30% reduction in EWS expression in both groups, but only the LS-related cancers showed diffuse cytoplasmic staining for EWS (61% vs 3% of the sporadic cancers, p<0.001). Thus, MMR-deficient CRCs, all carrying somatic EWS16T tract alterations, display a distinct subcellular EWS distribution pattern in vivo.
Further studies, ideally performed on EWS16T-mutated CRC cell lines, are needed to assess if this is directly related to EWS16T tract alterations or, rather, an indirect consequence of MSI-associated genetic instability affecting regulators of EWSR1 protein localization such as the methyl-transferases like PRMT1, which is known to regulate the localization of EWS(21) by methylating Glycine/Arginine-rich motifs located in the arginine-glycine-glycine domains of EWS.
on March 12, 2014. © 2013 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Discussion
EWS16T, a polyT/U tract in the 3'UTR of the EWSR1 gene, represents a novel, quasimonomorphic MSI target gene locus that identifies both hereditary and sporadic MMRdeficient colorectal cancers with perfect (100%) diagnostic sensitivity and specificity. Given this high diagnostic accuracy and quasi-monomorphic nature ascertained in more than 300 cancers, EWS16T has the potential to substantially facilitate and improve the accuracy of MSI detection in routine daily practice and prospective studies are now needed to determine whether EWS16T alone could replace the set of multiple diagnostic markers currently employed for MSI testing. The physiological role of EWS is largely unknown but based on its domain structure the protein is thought to be involved in such diverse processes as gene expression, RNA processing / transport and cell signalling. Knockout of EWS in mice results in high postnatal lethality, defects in pre-B cell development, meiotic arrest/germ cell apoptosis, premature cellular senescence, and hypersensitivity to ionizing radiation (IR) (31). These observations suggest roles for EWS in homologous recombination, DNA damage response, and maintenance of genome integrity (32). Indeed, we recently found that EWSR1 binds RNAs that originate at intrinsically unstable genomic loci and its knockdown increases the frequency With respect to tumorigenesis, genetic alterations in EWSR1 were first observed in Ewing sarcoma, the second most common malignant bone tumor in adolescents and young adults after osteosarcoma (16,33,34). In about 85% of cases, an EWSR1-FLI1 fusion protein is observed which has retained the N-terminal transcription activation domain but lost the RNAbinding domains, which are replaced with the DNA binding domain of the fusion partner (35).
The fusion protein is constitutively active and has been shown to alter the transcription of several downstream targets. Ewing Sarcoma is thus largely thought of as a gain of function phenotype. Loss of the EWSR1 function has been largely overlooked, in spite of the fact that the protein has a canonical RNA binding domain and has been shown to regulate several RNA processing events in the nucleus (36,37). As a member of the TET (TLS/FUS, EWS, and TAF15) family of RNA-and DNA-binding proteins it has been involved in transcriptional regulation and RNA processing (38) (39) (40) (41,42). Consistent with this potential dual role, EWS has been shown to regulate cyclin D1 transcripts both transcriptionally and at the level of splicing, with the oncogenic fusion protein EWS-FLI1 promoting the expression of the oncogenic cyclin D1b splice variant in Ewing sarcoma cells (43). More recently, EWS has been shown to regulate alternative splicing (AS) of the p53 repressor MDM2 and to be part of the microprocessor complex that mediates the genesis of microRNAs (32,42)