Cancer Genetics 204 (2011) 45e52

Cancer Genetics 204 (2011) 45e52 Exon scanning by reverse transcriptaseepolymerase chain reaction for detection of known and novel EML4eALK fusion variants in nonesmall cell lung cancer Heather R. Sanders a, *, Hai-Rong Li a, Jean-Marie Bruey a,1, Jay A. Scheerle b, Aurelia M. Meloni-Ehrig b, JoAnn C. Kelly b, Constance Novick b, Maher Albitar a,2 a Department of Hematology and Oncology, Quest Diagnostics Nichols Institute, San Juan Capistrano, CA, USA; b Department of Cytogenetics, Quest Diagnostics Nichols Institute, Chantilly, VA, USA Chromosomal inversions within chromosome 2p, resulting in fusions between the echinoderm microtubule-associated protein-like 4 (EML4) and anaplastic lymphoma kinase (ALK ) genes, are a recent focus of treatment options for nonesmall cell lung cancer. Thirteen EML4eALK fusion variants have been identified, affecting eight EML4 exons. We have developed an exon scanning approach using multiplex reverse transcriptaseepolymerase chain reaction (RT-PCR) to amplify known and potential variants involving the first 22 EML4 exons. A total of 55 formalin-fixed, paraffin-embedded lung cancer tumors were screened, of which 5 (9%) were positive for EML4eALK fusions. Four positive cases harbored known fusion variants: variant 3a, 3b, or both in three cases and variant 1 in one case. The fifth positive specimen harbored two novel variants, designated 8a and 8b, involving exon 17 of EML4. Fluorescence in situ hybridization confirmed the presence of EML4eALK fusions in three of the four RT-PCR-positive specimens with sufficient tissue for examination, and also confirmed absence of fusions in all 19 RT-PCR-negative specimens tested. Immunohistochemistry analysis confirmed ALK protein expression in the sample containing the novel 8a and 8b variants. This RT-PCR-based exon scanning approach avoids the limitations of screening only for previously identified EML4eALK fusions and provides a simple molecular assay for fusion detection in a clinical diagnostics setting. Keywords EML4-ALK, anaplastic lymphoma kinase, NSCLC, lung cancer ª 2011 Elsevier Inc. All rights reserved. The introduction of targeted therapies for cancer has provided physicians with a personalized approach to cancer treatment. In nonesmall cell lung cancer (NSCLC), EGFR and KRAS mutations have been the most widely studied in terms of the use of tyrosine kinase inhibitors such as gefitinib and erlotinib. Knowledge of the gene mutations harbored by a tumor provides a significant advantage when treating with targeted therapy. For example, tyrosine kinase inhibitors are much more effective in patients harboring EGFR mutations than in those with KRAS mutations, which are nonresponsive Received July 12, 2010; accepted August 19, 2010. * Corresponding author. E-mail address: heather.r.sanders@questdiagnostics.com 1 Present address: biotheranostics, San Diego, CA. 2 Present address: Health Discovery Corporation, Savannah, GA. to these drugs (1). Clinical EGFR and KRAS mutation testing provides a way to identify patients most likely to respond to such therapies. Recently, inhibitors of anaplastic lymphoma kinase (ALK) have been used successfully in NSCLC patients harboring gene fusions between the ALK and the echinoderm microtubule-associated protein-like 4 (EML4) genes (2,3). These fusions, which result from a paracentric inversion on chromosome 2, inv(2)(p21p23), have been identified in 3e7% of all NSCLC cases (4). To date, 13 variants have been reported, involving eight different EML4 exons (exon 2, 6, 13, 14, 15, 17, 18, and 20) and, invariably, exon 20 of ALK (4e11). The multiplicity of EML4eALK fusion transcript variants already identified, with likely more yet to be characterized, creates a difficult situation for applying diagnostic testing. Detection of ALK rearrangements by fluorescence in 2210-7762/$ - see front matter ª 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.cancergencyto.2010.08.024

46 H.R. Sanders et al. situ hybridization (FISH) is commonly used as a diagnostic tool; however, this method does not identify the gene that ALK is fused with, nor the precise variant. We therefore developed an exon scanning method based on reverse transcriptaseepolymerase chain reaction (RT-PCR), spanning nearly the entire EML4 gene; this method is designed to detect all 13 known EML4eALK variants and also to identify novel variants involving any of the first 22 exons of EML4. Here, we describe the use of this exon scanning approach to detect EML4eALK fusion transcript variants in NSCLC specimens, as well as the characterization and phenotypic expression of two novel EML4eALK fusion transcript variants identified during the study. Materials and methods Tissue specimens De-identified NSCLC formalin-fixed paraffin-embedded (FFPE) tissue blocks from 55 patients were used: 37 adenocarcinoma, 11 squamous cell carcinoma, 5 adenosquamous cell carcinoma, and 2 large cell carcinomas. All samples were residual tissue samples from diagnostic biopsies or resections and were nonenriched by previous molecular testing. Tumor subtype was confirmed by histologic and immunohistochemical (IHC) evaluation. RNA extraction and DNase digestion Tissue blocks were sectioned onto slides for hematoxyline eosin staining; sections for RNA extraction were left unstained. Tumor area was identified by a licensed pathologist, and tissue from this area was scraped for RNA extraction with a HighPure mirna isolation kit (Roche Applied Science, Mannheim, Germany). Extracted nucleic acid samples were treated with DNase I (DNA-free; Applied BiosystemseAmbion, Austin TX) before RNA amplification. Exon scanning RT-PCR and fragment analysis RT-PCR amplification with an RNA UltraSense one-step RT- PCR kit (Invitrogen, Carlsbad CA) was performed using forward primers designed against the first 22 exons of EML4 and reverse FAM-labeled primer to exon 20 of ALK (see Table 1 for primer sequences). The RNA was first reverse transcribed by incubation at 55 C for 30 seconds, followed by a denaturing step at 94 C for 2 minutes. The PCR amplification was performed with 40 cycles of denaturation at 94 C for 15 seconds, annealing at 60 C for 30 seconds, extension at 68 C for 1 minute, and final extension at 68 C for 5 minutes following cycling. The RT-PCR to query alternative splicing in intron 17e18 was performed with the same conditions, except with an annealing temperature of 50 C and a reverse FAM-labeled primer recognizing ALK intron 17e18. The RT-PCR products were diluted 1:10 with moleculargrade H 2 O, denatured in formamide containing ROX GeneScan 350 size marker, and were size-fractionated by capillary electrophoresis in an ABI 3730 genetic analyzer (all from Applied Biosystems, Foster City, CA). Results were analyzed with GeneMapper software (Applied Biosystems). Expected sizes of known fusion transcript variants are presented in Table 2, along with estimated sizes of potential fusions depending on exon involvement. cdna sequencing The RT-PCR products were separated by electrophoresis on a 2% agarose gel. Individual bands were cut out of the gel and DNA was extracted with a MinElute gel extraction kit according to the manufacturer s instructions (Qiagen, Valencia, CA). The forward and reverse primers used in RT-PCR served as forward and reverse primers for sequencing using the ABI Prism BigDye Terminator v3.1 cycle sequencing kit, according to the manufacturer s instructions (Applied Biosystems). FISH Formalin-fixed, paraffin-embedded sections 4e5 mm in size were used for FISH with an ALK dual-color, break-apart probe (Abbott Molecular, Des Plaines, IL); the slides were deparaffinized prior to probe application. The FISH analysis was performed using a Nikon 50i fluorescence microscope (Nikon, Tokyo, Japan). The images were captured using a charge-coupled device camera and the Isis imaging system (MetaSystems, Altlussheim, Germany; Watertown, MA). A total of 200 cells were analyzed on all the normal cases and 100 cells on any abnormal cases. Any tissues with questionable tumor areas were reviewed and marked by a pathologist. On all cases, the entire slide was examined for possible areas where rearrangements may have been missed. The cutoff for rearrangement of the ALK gene was 2%. Immunohistochemistry Unstained slides were deparaffinized and stained with CONFIRM anti-alk1 primary antibody (mouse monoclonal clone ALK01; Ventana Medical Systems, Tucson AZ). All IHC steps were performed using a BenchMark XT system, according to the manufacturer s protocol (Ventana Medical Systems). Results EML4eALK exon scanning for screening of lung cancer tissue For EML4-ALK detection, we designed an exon scanning RT-PCR approach to detect all known fusion transcript variants, as well as variants involving any of the first 22 EML4 exons. EML4eALK fusions were detected by this approach in 5 of the 55 NSCLC FFPE tumor tissue samples (9%). All EML4eALK positives were adenocarcinomas. Four of the positive cases harbored previously described fusion variants: variants 3a or 3b (or both) in three cases and variant 1 in the fourth case. In addition, one case yielded two strong amplification peaks at unexpected sizes (171 bp and 238 bp) in the reaction containing primers for EML4 exons 3, 7, 10, 13, and 17 (Table 2, multiplex mix 3). Repeat analysis conducted

EML4-ALK detection in NSCLC 47 Table 1 RT-PCR primer sequences Primer target Sequence (5 0 to 3 0 ) EML4 exon 1 F EML4 exon 2 F EML4 exon 3 F EML4 exon 4 F EML4 exon 5 F EML4 exon 6 F EML4 exon 7 F EML4 exon 8 F EML4 exon 9 F EML4 exon 10 F EML4 exon 11 F EML4 exon 12 F EML4 exon 13 F EML4 exon 14 F EML4 exon 15 F EML4 exon 16 F EML4 exon 17 F EML4 exon 18 F EML4 exon 19 F EML4 exon 20 F EML4 exon 21 F EML4 exon 22 F ALK exon 20 R EML4 intron 17 R B2M F B2M R Abbreviations: F, forward; R, reverse. CGG TCC GCT GAA TGA AGT AAG ATC ATG TGG CCT CAG TG TGG TGC AAA CAG AAA ACC AA CCC TCT TCA CAA CCT CTC CA ACG ACC ATC ACC AGC TGA AA CTG CAG ACA AGC ATA AAG ATG GTC GGC CAA TTA CCA TGT TC CTT CCG ACC GGG AAA ATA GT ACA TCC TGA CAA AAT TAG GAT TGC CCT CTA CAA CCC CAC GTC AG GCA TAT GCT TAC TGT ATG GGA CTG TTT CAC CCA ACA GAT GCA AA GAC TCA GGT GGA GTC ATG C AAG CTC ATG ATG GCA GTG TG TGT AGC AGA AGG AAA GGC AGA GTC TTG CCA CAC ATC CCT TC CCA GGA CAC TGT GCA GAT TT AGG TGG TTT GTT CTG GAT GC CCT TCC TGG CTG TAG GAT CTC CAG ATA TGG AAG GTG CAC TG ATT CCA AAT GGC TGC AAA CT AGC TGT TGC CGA TGA CTT TT FAM-AGC TTG CTC AGC TTG TAC TC FAM-TTT AAT GAG TTT AAT TTT GGG FAM-TGA CTT TGT CAC AGC CCA AGA TA TGT GCA TAA AGT GTA AGT GTA TAA GCA with these primers in individual reactions revealed that both peaks resulted from amplification with the EML4 exon 17 forward primer, yielding two amplicons of 171 bp and 238 bp (Figure 1A). The ALK rearrangement was also confirmed by FISH using break-apart probes targeting the 2p23 locus. The typical ALK rearrangement FISH pattern consists of overlapping (intact ALK ) and split orange (3 0 region) and green (5 0 region) signals (disrupted ALK locus). Single and multiple copies of the intact ALK fusion together with the abnormal split pattern were observed in the specimen harboring the novel variants, designated 8a and 8b (Figure 1B). In all, 23 of the 55 NSCLC specimens underwent FISH confirmation of RT-PCR exon scanning results: the 1 specimen containing the novel 8a and 8b variants, 3 additional specimens that were EML4eALK fusion-positive by RT-PCR, and 19 specimens that were fusion-negative by RT-PCR (data not shown). Because of insufficient sample, FISH was not performed on the specimen positive for variant 1. All specimens that tested negative by RT-PCR also tested negative by FISH. Three of the four RT-PCR-positive samples tested were also positive by FISH; the fourth sample tested positive for variants 3a and 3b by RT-PCR but negative by FISH, where only polyploidy counts for ALK were seen. Upon repeat RNA extraction and RT-PCR, detection of variant 3a and 3b in this specimen was duplicated. Furthermore, the FISH test was repeated and the initial result from the first analysis was confirmed: no observed rearrangement of the ALK gene. These results suggest that the RT-PCR assay was more sensitive than FISH for this specimen. Nonetheless, to draw general conclusions about assay performance and sensitivities in comparison with other methods, a much larger cohort is needed to include an ample number of positive specimens. Sequence characteristics of two novel EML4eALK fusion transcript variants 8a and 8b Sequencing of the unidentified amplicons described here demonstrated that both contained the complete coding regions of EML4 exon 17 and ALK exon 20; the size difference was attributable to partial intron insertions of varying length. The 171-bp peak (variant 8a) consisted of EML4 exon 17 fused to ALK exon 20 and containing a 30-nucleotide insertion from ALK intron 19e20, E17;ins30A20 (Figure 2A). The 238-bp peak (variant 8b) consisted of EML4 exon 17 with an insertion of 61 non-adjacent nucleotides from EML4 intron 17e18, fused to ALK exon 20 and containing a 34 nucleotide insertion from ALK intron 19e20, E17ins61;ins34A20 (Figure 2B). Thus, the two fusion products likely consist of EML4 exons 1e17 and ALK exons 20e29 with a 30-bp (8a) or 95-bp (8b) insertion (Figure 2C). To determine whether the EML4 intron 17e18 segment may be observed adjacent to exon 17 as a result of alternative Table 2 Expected amplicon size by EML4eALK variant EML4eALK variant EML4 involvement, exon Multiplex reaction no. Amplicon size, bp Published EML4eALK fusion variants 2, 3a, 3b, 4, 7, E20;ins18A20 20, 6, 14, 20 1 183, 112, 145, 163, 187, 201 V5 18 2 167 1, 6, E17;ins68A20 13, 17 3 146, 215, 144 V4, 5a, 5b 15, 2 4 134, 118, 235 Potential EML4 exon involvement a unknown 1, 5, 9, 12, 16, 21 2 152, 199, 137, 181, 186, 170 unknown 3, 7, 10 3 154, 170, 187 unknown 4, 8, 11, 19, 22 4 144, 183, 131, 168, 137 References for published variants are as follows: variant 1 and 2 (11), variant 3a and 3b (10), variant 4, 5a and 5b (9), variant V4 (8), variant V5 (6), variant 6e7 (7), variant E17;ins68A20 and E20;ins18A20 (5). a Amplicon sizes are estimated based on no insertions or deletions.

48 H.R. Sanders et al. Figure 1 Fragment analysis and FISH staining in the specimen positive for the novel EML4eALK fusion variants 8a and 8b. (A) RNA from FFPE tissue was amplified by RT-PCR using unlabeled EML4 exon 17 forward primer and FAM-labeled ALK exon 20 reverse primer. Two peaks were observed, at positions 171 bp and 238 bp, indicating the presence of at least two fusion transcript variants (the actual nucleotide length of variant 8b is 236 bp, slightly shifted from the observed peak position due to mobility shift differences of some primer/pcr products labeled with FAM). (B) FISH using break-apart probes for ALK revealed abnormalities. The normal allele shows cohybridization of 5 0 (green) and 3 0 (red) ALK probes, as demonstrated by overlapping signal (yellow) in the pseudo-colored FISH image (arrowhead). The second allele shows split signal from 5 0 and 3 0 ALK probes (arrows), demonstrating presence of an ALK rearrangement. splicing in normal EML4 transcripts, RT-PCR was performed on three NSCLC cell lines (NCI-H2228, NCI-1299, NCI-H838), two normal lung tissue samples, and the lung cancer tissue sample positive for variants 8a and 8b, using primers specific to exon 17 and the 61-bp segment from intron 17e18. An amplicon of expected size was observed only in the lung cancer specimen harboring the 8a and 8b variants (data not shown). This finding suggests that this specific paracentric inversion results in a new alternative splice site in the premrna transcript. Putative protein characteristics of novel variants 8a and 8b Based on the deduced amino acid sequence, variant 8a yields a 660 amino acid protein (Figure 3A) and variant 8b yields a 1250 amino acid protein (Figure 3B). Fusion transcript variant 8a appears to encode an EML4 truncation with no functional ALK domains, as a result of an early stop codon in the 30-bp insertion (Figure 3C). Variant 8b, however,

EML4-ALK detection in NSCLC 49 Figure 2 Sequencing of transcript fusion sites in novel EML4eALK fusion variants 8a and 8b with electropherograms from reverse sequencing reactions (A,B) and with representation of fusion transcripts (C). (A) The 171-bp peak consisted of a complete EML4 exon 17 and ALK exon 20 separated by 30 bp of adjacent ALK intron 19e20 sequence. (B) The larger product, 238 bp, also contained a complete EML4 exon 17, ALK exon 20, and adjacent ALK intron 19e20 with an added 4 bp (boxed) of that intron. It also contained 60 bp of nonadjacent EML4 intron 17e18 sequence separating the EML4 exon 17 and ALK intron sequence. (C) Fusion transcripts resulted from an EML4 (green) and ALK (red) genetic rearrangement. The precise gene breakpoints were not identified in the present study; dotted borders indicate intronic and exonic sequences that appear to be carried over to the fusion transcripts. contains an in-frame 95-bp insertion, resulting in the presence of a putatively functional ALK protein tyrosine kinase domain (Figure 3C). To determine whether an EML4eALK fusion protein was expressed in the tumor tissue harboring the 8a and 8b variants, IHC analysis of the tissue with ALK1 monoclonal antibodies was performed. This analysis confirmed significant ALK1 staining of the cytoplasm in tumor cells (Figure 4), indicating overexpression of the ALK domain from the fusion protein in the malignant cells. Furthermore, nine additional specimens (1 additional positive specimen and 8 specimens negative by RT-PCR) were confirmed by IHC: both RT-PCRpositive cases were positive by IHC and all 8 RT-PCRnegative cases were negative by IHC (data not shown). Discussion A previous study screening fresh-frozen lung cancer tissue and sputum applied multiplex RT-PCR techniques designed to detect EML4 fusions occurring at any exon that would participate in an in-frame fusion to exon 20 of ALK (9). This approach was designed for high-throughput screening and led to characterization of EML4eALK variant 4 and variants 5a and 5b. The methods from that study are not applicable to fixed tissue containing highly fragmented RNA, however, because they require amplification of fragments much too large to be detected (>400 bp). In the present study, we focused on designing a multiplex RT-PCR assay that can be

50 H.R. Sanders et al. Figure 3 Putative proteins formed as a result of EML4eALK variant 8a and 8b fusions. (A) Amino acid sequence of the putative truncated EML4 protein resulting from EML4eALK fusion transcript variant 8a. EML4 residues are shown in black, intron-derived residues in blue. (B) Amino acid sequence of variant 8b putative EML4eALK fusion protein. EML4 residues are shown in black, intron-derived residues in blue, and ALK residues in red. (C) Putative EML4 truncation containing the N-terminal region of EML4 consisting of HELP and partial WD repeat domains (variant 8a) and fusion of the N-terminal region of EML4 to the C-terminal region of ALK containing the protein tyrosine kinase domain (variant 8b). Inferred chimeric protein fusion sites are indicated with a dotted line. IDS, intron-derived sequence; TM, transmembrane domain.

EML4-ALK detection in NSCLC 51 Figure 4 Immunohistochemistry staining confirms expression of ALK1 protein. Formalin-fixed, paraffin-embedded tissue sections were stained by IHC using ALK1 antibodies (A,B) or nonimmune control antibodies (C,D). Malignant cells stain positive for ALK1. Images were obtained with a 20 objective (ScanScope; Aperio Technologies, Vista, CA) and were digitally magnified an additional 10 (A,C) or 20 (B,D) to achieve 200 and 400 magnification, respectively. All four images are from the case positive for the novel EML4eALK fusion variants 8a and 8b. used as a clinical diagnostic tool in FFPE tumor tissue specimens. The method described here is an exon scanning approach that minimizes amplicon sizes to <250 bp and encompasses the first 22 exons of EML4, where all fusions reported to date have occurred. Screening with this method identified EML4eALK fusions in 5 of the 55 NSCLC specimens examined (9%), including three previously described variants (1, 3a, and 3b) and two novel variants involving exon 17 of EML4 (8a and 8b). Notably, fusions of EML4 exon 17 to ALK exon 20 would require an insertion or deletion to create an in-frame variant. In fact, one fusion transcript variant that we observed (8a) contained an insertion of 30 bp that results in an early stop codon and would not likely have malignant transforming activity on its own. This particular case also expressed variant 8b, with a 95-bp insertion, which is most likely responsible for expression of the ALK domain in this specimen as observed by IHC and the transformation or malignant phenotype. An interesting feature of variant 8b was the presence of a 61-bp sequence of nonadjacent EML4 intron 17e18. This intron sequence is located w1.2kb downstream of exon 17 in the normal EML4 transcript. Based on analysis of normal lung tissue and cells not containing variant 8b, it is clear that this configuration of intron 17e18 results not from normal alternative splicing, but rather from alternative splicing caused by a translocation. The present findings and the increasing number of EML4eALK variants being identified (4e11) highlight the utility of comprehensive testing in ensuring detection of known variants and in identifying novel variants of the EML4eALK fusion. Exon scanning approaches such as that used in the present study may provide an effective solution to the need to identify therapeutic strategies for lung cancer patients in a clinical setting. Acknowledgments The authors thank Julie Broccardo, lead histology technologist at Quest Diagnostics, for technical immunohistochemistry contributions and Jeff Radcliff, senior medical writer at Quest Diagnostics, for editorial contributions. References 1. Pao W, Wang TY, Riely GJ, et al. KRAS mutations and primary resistance of lung adenocarcinomas to gefitinib or erlotinib. PLoS Med 2005;2(1):e17. 2. Solomon B, Varella-Garcia M, Camidge DR. ALK gene rearrangements: a new therapeutic target in a molecularly defined subset of nonesmall cell lung cancer. J Thorac Oncol 2009;4: 1450e1454. 3. Kwak EL, Camidge DR, Clark J, et al. Clinical activity observed in a phase I dose escalation trial of an oral c-met and ALK inhibitor. PF-02341066. J Clin Oncol 2009;27 (15s May 20 Suppl):3509 [Abstract]. 4. Horn L, Pao W. EML4eALK: honing in on a new target in nonesmall cell lung cancer. J Clin Oncol 2009;27:4232e4235. 5. Takahashi T, Sonobe M, Kobayashi M, et al. Clinicopathologic features of non-small-cell lung cancer with EML4eALK fusion gene. Ann Surg Oncol 2010;17:889e897. 6. Wong DW, Leung EL, So KK, et al., University of Hong Kong Lung Cancer Study Group. The EML4eALK fusion gene is involved in various histologic types of lung cancers from

52 H.R. Sanders et al. nonsmokers with wild-type EGFR and KRAS. Cancer 2009; 115:1723e1733. 7. Takeuchi K, Choi YL, Togashi Y, et al. KIF5BeALK, a novel fusion oncokinase identified by an immunohistochemistry-based diagnostic system for ALK-positive lung cancer. Clin Cancer Res 2009;15:3143e3149. 8. Koivunen JP, Mermel C, Zejnullahu K, et al. EML4eALK fusion gene and efficacy of an ALK kinase inhibitor in lung cancer. Clin Cancer Res 2008;14:4275e4283. 9. Takeuchi K, Choi YL, Soda M, et al. Multiplex reverse transcriptionepcr screening for EML4eALK fusion transcripts. Clin Cancer Res 2008;14:6618e6624. 10. Choi YL, Takeuchi K, Soda M, et al. Identification of novel isoforms of the EML4eALK transforming gene in nonesmall cell lung cancer. Cancer Res 2008;68:4971e4976. 11. Soda M, Choi YL, Enomoto M, et al. Identification of the transforming EML4eALK fusion gene in non-small-cell lung cancer. Nature 2007;448:561e566.