Kenichi Yoshida et al.,2012——结合癌症研究文献无重点介绍了几种测

Deep Sequencing in Cancer ResearchKenichi Yoshida,Masashi Sanada and Seishi Ogawa *Cancer Genomics Project,Graduate School of Medicine,The University of Tokyo,Tokyo,Japan *For reprints and all correspondence:Seishi Ogawa.E-mail:sogawa-tky@umin.ac.jpReceived August 26,2012;accepted November 4,2012Cancer is caused by alterations in the cellular genome including single-nucleotide variations,small insertions and deletions (indels),copy number changes and other structural variations and,as such,their detection in a comprehensive manner is of critical importance for fully understanding cancer pathogenesis,improvement of diagnosis as well as the development of novel therapeutics.In this point of views,the recent development of massively parallel (or ‘next-generation’)sequencing technologies has provided an unprecedented possibility to ac-complish this need by enabling single-nucleotide resolution analysis of the entire genome of cancer cells as well as more targeted analysis of coding sequencing or transcriptomes.Through international co-operations,a wide variety of cancer cell types have now been analyzed using these technologies to help unmask their pathogenesis.In this review,we briefly overview the recent advances in cancer research obtained through the massive effort of sequencing cancer genomes.Key words:massively parallel sequencing –cancer genome –somatic mutationINTRODUCTIONCancer is caused by accumulation of genomic abnormalities (1).Therefore,comprehensive detection of various types of genetic changes,such as point mutations,copy number alterations and structural variations,are critically important for the understanding cancer biology,the improvement of diagnosis as well as the development of targeted therapies.Cancer genomes are highly complex,which are composed of as many as 3billion bases,including 24000–25000of protein-coding genes.The initial clue to accommodate this complexity of cancer genomes was obtained by the construc-tion of the first draft sequence of the human genome in 2001through the international collaborative effort under the human genome project (2).Since then,the sequence has been used as the reference human genome,based on which researches have been analyzing somatic mutations in cancers.However,the real breakthrough to facilitate the de-tection of cancer-specific gene mutations/alterations has been brought about by the recent development of massively parallel sequencing technologies;the 13years requiredto read an initial human genome has been dramatically reduced to 2weeks with typical next-generation sequen-cing.This revolution of sequencing technologies envisages a new era of cancer genetics and biology.In this review,we overview the second-generation sequencing technologies and their application to cancer research.ANALYSIS OF CANCER GENOMES USING SECOND-GENERATION SEQUENCING TECHNOLOGIESA typical second-generation (also known as ‘next-generation’)sequencer goes through billions of fragmented DNAs (reads)simultaneously in a highly parallel way (mas-sively parallel sequencing).Inevitably high sequencing errors in reading numerous bases are compensated by reading multiple (.30–40times)DNA fragments for an ac-curate determination of somatic mutations and allelic status of polymorphisms in diploid sequences.For example,typic-ally 90Gb of sequence or 30-fold coverage are obtained#The Author 2012.Published by Oxford University Press.All rights reserved.For Permissions,please email:journals.permissions@doi:10.1093/jjco/hys206Advance Access Publication 5December2012at University of Melbourne Library on September 25, 2014/Downloaded fromto call99%of SNP alleles(3).The need of multiple reads is also realized by the fact that the mutations under interest may exist only in low allele frequencies in many situations; primary cancer samples are more or less‘contaminated’with normal tissues and there also exist multiple tumor subclones with different genetic alterations within a tumor,comprising intra-tumor heterogeneity(4–6).It may be necessary to deeply sequence to capture such low-frequency alleles de-pending on the apparent frequencies of the relevant alleles. To further compound matters,pathogenesis of cancer is heterogeneous in terms of the spectrum of gene mutations even within the same histology type.Thus,sequencing mul-tiple tumor samples may be necessary to understand the full spectrum of causative mutations depending on the extent of the heterogeneity,except for rare cases where the tumor is highly homogeneous and therefore only a few samples were sufficient to capture the relevant mutations.In fact,several ongoing projects are analyzing large numbers of cancer spe-cimens;in the Cancer Genome Atlas(TCGA)(7)promoted by NIH/NCI in the USA and the International Cancer Genome Consortium(ICGC)participated in by institutes from multiple nations(8),more than500specimens for each of the major cancer types will be subjected to integrated genome,epigenome,transcriptome studies(9–14). WHOLE-GENOME SEQUENCINGIn2008,thefirst sequencing of an entire cancer genome was reported,providing a comprehensive registry of gene mutations in a case with acute myeloid leukemia(AML) with a normal karyotype.In whole-genome sequencing,fragmented genomic DNA is directly subjected to massively parallel sequencing after minimum processing with several cycles of PCR amplification.Although simple in manipula-tion,this approach provides least biased,full-range informa-tion about genetic alterations at a single-nucleotide level (Fig.1).Major caveats include the cost and time required for the analysis,compared with sequencing with lower complex-ity.It also consumes much more computer resources. Because of its uniform coverage across all the genomic regions except for GC-rich sequences,whole-genome se-quencing captured mutations that occur in both coding and non-coding regions,including promoters,enhancer,introns and non-coding RNAs as well as intergenic sequences.Not only detecting single-nucleotide substitutions,whole-genome sequencing can also identify a wide variety of structural ab-normalities,such as chromosomal translocations,deletions and insertions,as well as copy number alterations,in a com-prehensive manner.Especially,currently no other methods are available for detecting chromosomal translocations sys-tematically,which would be underscored by the fact that a number of fusion genes caused by chromosomal transloca-tions play a critical role in the cancer pathogenesis and are also exploited for contriving novel diagnostics and therapeu-tics(15,16).Moreover,whole-genome sequencing can reveal a more complex aspect of cancer genomes,in which tens to hundreds of genomic rearrangements involved localized genomic regions(‘chromothripsis’),which were thought to be acquired in an apparently one-off cellular crisis and reported to occur in2–3%of all cancers and 25%of bone cancers(Fig.2)(17).Finally,whole-genome sequencing is capable of detecting non-human sequencing derived from microbial genomebyFigure1.Various types of genetic alterations that can be detected by whole-genome sequencing.Sequenced fragments(reads)are depicted as bars andpaired-end reads connected with a dotted line are mapped to different genomic positions in the case of translocation.Copy number alterations can be detectedby change in the number of sequence reads(depth)compared with a normal control.at University of Melbourne Library on September 25, 2014/Downloaded fromanalyzing sequences that are not mapped to human reference genome.Sung et al.performed whole-genome sequencing of 76HBV-positive hepatocellular carcinomas and detected HBV integration in 76.4%of tumor samples.They also revealed that HBV integrations in TERT ,MLL4and CCNE1genes accounted for 40%of cases with HBV integrations (18).As the cost of sequencing continues to reduce,whole-genome sequencing is becoming a standard method of cancer genome sequencing feasible for many researchers.TARGETED/WHOLE-EXOME SEQUENCINGConsidering the current cost and throughput of sequencing,alternative to whole-genome sequencing would be focusing on particular sequences under interest,for example,entire coding sequences that account for ,1.3%of the total genomes,which are now widely used for cancer genome se-quencing (targeted sequencing).In whole-exome sequences,entire protein-coding sequences encompassing from 50to 62Mb are captured using liquid–solid phase hybridization to chemically synthesized nucleotides or ‘baits’(Fig.3),which were subjected to massively parallel sequencing.By limiting the targets to approximately one-hundredth of the entire genome,much higher sequence coverage ( 100times)can be achieved with considerably less sequence data compared with whole-genome sequencing,which far outperforms capillary sequencing-based approach employed in exome sequencing of glioblastoma (19).Potential drawbacks in whole-exome sequencing are add-itional costs and procedures required for target capture andcompromised sensitivity due to less uniform coverage and inability to detect fusion genes and other structural varia-tions.Nevertheless,whole-exome sequencing provides a rea-sonable alternative to whole-genome sequencing,in terms of a high throughput of analyzing multiple samples at higher coverage and lower costs and the higher sensitivity of detect-ing mutations with lower allele frequencies,which may exist only in minor subclones or be associated with lower tumor burden,although it is challenging to detect somatic muta-tions in cases where germline control sample is contami-nated with tumor tissues.In fact,whole-exome sequencing has been successfully applied to identify a number of novel targets of recurrent mutations (10,20–24).TRANSCRIPTOME SEQUENCINGOne of the major shortcomings in whole-exome sequencing is its inability to detect gene fusions.The deficit could be complemented in part by transcriptome sequencing (RNA sequencing,RNA-seq)at an acceptable cost (Fig.4).In fact,transcriptome sequencing has been shown to provide a powerful method to detect biologically relevant fusion genes in that the detected fusion genes are actually expressed in tumor cells.The exact sequences of the expressed fusion transcripts can be directly determined.Steidl et al.performed RNA-seq of Hodgkin lymphoma cell lines and identified gene fusion involving MHC class II transactivator CIITA (25),which were also found in primary mediastinal B-cell lymphoma (38%)and Hodgkin lymphoma samples (15%).Several algorithms are available for detecting fusiongenesFigure 2.Schematic overview of chromothripsis and a circos plot of a chromothripsis-positive cancer.(A)Schematic overview of chromotripsis.(B)Chromothripsis detected in a case of clear cell renal carcinoma by whole-genome sequencing.Chromosomes (outside of the circular plot)and chromosom-al rearrangements are shown as arcs connecting the two relevant genomic regions in the middle.Chromothripsis rearrangements are indicated by green arrows.112Cancer research and deep sequencingat University of Melbourne Library on September 25, 2014/Downloaded frombased on transcriptome sequencing,many of which are no-torious for high false-positive rates even for recurrently detected candidates,which may necessitate intensive valid-ation tasks using RT–PCRs.RNA-seq can be used to detect somatic mutations and some important discovery of recurrent mutations in cancer has been made by RNA-seq,such as FOXL2mutations in ovarian granulosa cell tumors (26).Although biologically relevant mutations tend to be enriched expressed sequences,coverage of sequencing could be highly biased depending on expression levels of different targets,leading to loss of power for detecting some mutations.This is in contrast to whole-genome/exome sequencing,with which relatively un-biased detection is expected.Difficulty in obtaining appro-priate normal control tissues is another significant problem with transcriptome-based detection of mutations.RNA-seq also allows analysis of gene expression profiles in highersensitivity.Novel transcripts or alternative splicing forms could be detected also,although computational analysis could be a challenging task.However,the higher representa-tion of 30sequences could prevents unbiased estimation of relative expression levels,especially estimation of large tran-scripts,in which significant underrepresentation of the 50sequences occurs.DEEP SEQUENCING OF SELECTED TARGETSIntratumoral heterogeneity seems to be a common feature of cancers,providing a major source of tumor recurrence or re-sistance to anti-cancer therapies.It could be best evaluated by deep sequencing in combination with whole-genome/exome sequencing,where accurate allele frequencies of a set of mutations detected by genome/exome sequencingareFigure 3.Targeted sequencing using in-solution captureapproach.Figure 4.FUS -DDIT3fusion gene of liposarcoma detected by transcriptome sequencing.Jpn J Clin Oncol 2013;43(2)113at University of Melbourne Library on September 25, 2014/Downloaded fromdetermined by reading at higher depth(typically.1000 or10000).Target DNA was amplified by individual PCRs or captured by high-throughput hybridization to the target baits.As low as1%of minor alleles could be sensitively detected by target deep sequencing depending on experimen-tal conditions and sequencing platforms,although back-ground errors inherent to sequencing chemistries would prevent further detection of even lower allele frequencies. Saha et al.performed deep sequencing(median20000 times)of somatic mutations in triple-negative breast cancers detected by whole-exome sequencing,in which existence of subclones with different mutant allele frequencies were clearly demonstrated in most of cases(5).Walter et al.also showed the presence of subpopulation in secondary AML cases derived from myelodysplastic syndromes by whole-genome sequencing followed by target deep resequeing (average640times)of304–872candidates of somatic muta-tions using solid-phase target capture(6).Another application of target deep sequencing is mutation screening involving a large number of genes and tumor samples.Actually,it becomes more and more unrealistic in detecting mutations of a number of large genes among hun-dreds of specimens using Sanger sequencing and plausible application of high-throughput sequencing.Deep sequencing also enabled a sensitive detection of minor mutant alleles with low frequencies together with an accurate determination of allele frequencies at the same time(10,20,27). SUMMARY AND FUTURE DIRECTIONS Considering the ever-decreasing cost for sequencing,major mutational targets will be identified for common cancer types within a few years.However,many issues should be remained to be unanswered.For many newly identified mutations,their biological/clinical significance as well as their molecular mechanics in cancer development is unknown.There is an urgent need to achieving the clinical application of these technologies for diagnosis and treatment stratification,eventually leading to cure for cancer,although special attention should be paid before clinical use.For example,validation of the candidate mutations using alterna-tive methods such as the conventional Sanger sequencing is indispensable at the current moment considering a relatively high error rate of second-generation sequencing.While understanding of the intratumoral heterogeneity is one of the critical issues in carcinogenesis,especially with regard to re-currence or resistance to anti-cancer therapies,higher reso-lution/throughput sequencing technologies,ideally of single-cell genomes,will be required to approach this issue. Promising news is the recent announcement of the coming newer generation sequencing platforms,including Ion tor-rent’s and Nanopore technologies(28,29).Roles of epigenet-ic abnormalities together with their interactions with gene mutations need to be clarified in further details,for which another technological breakthrough should be warranted.FundingThis work was supported by Grant-in-Aids from the Ministry of Health,Labor and Welfare of Japan and from the Ministry of Education,Culture,Sports,Science and Technology,and the Japan Society for the Promotion of Science(JSPS)through the‘Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST Program),’initiated by the Council for Science and Technology Policy(CSTP)and also by‘project for develop-ment of innovative research on cancer therapies(p-direct)’. Conflict of interest statementNone declared.References1.Stratton MR,Campbell PJ,Futreal PA.The cancer genome.Nature2009;458:719–24.nder ES,Linton LM,Birren B,et al.Initial sequencing and analysisof the human genome.Nature2001;409:860–921.3.Li H,Ruan J,Durbin R.Mapping short DNA sequencing reads andcalling variants using mapping quality scores.Genome Res2008;18:1851–8.4.Gerlinger M,Rowan AJ,Horswell S,et al.Intratumor heterogeneity andbranched evolution revealed by multiregion sequencing.N Engl J Med 2012;366:883–92.5.Shah SP,Roth A,Goya R,et al.The clonal and mutational evolutionspectrum of primary triple-negative breast cancers.Nature2012;486:395–9.6.Walter MJ,Shen D,Ding L,et al.Clonal architecture of secondaryacute myeloid leukemia.N Engl J Med2012;366:1090–8.7./.8./.9.Integrated genomic analyses of ovarian carcinoma.Nature2011;474:609–15.10.Papaemmanuil E,Cazzola M,Boultwood J,et al.Somatic SF3B1mutation in myelodysplasia with ring sideroblasts.N Engl J Med 2011;365:1384–95.11.Wang L,Lawrence MS,Wan Y,et al.SF3B1and other novel cancergenes in chronic lymphocytic leukemia.N Engl J Med2011;365:2497–506.prehensive molecular characterization of human colon and rectalcancer.Nature2012;487:330–7.13.Fujimoto A,Totoki Y,Abe T,et al.Whole-genome sequencing of livercancers identifies etiological influences on mutation patterns and recurrent mutations in chromatin regulators.Nat Genet2012;44:760–4.14.Jones DT,Jager N,Kool M,et al.Dissecting the genomic complexityunderlying medulloblastoma.Nature2012;488:100–5.15.Tomlins SA,Rhodes DR,Perner S,et al.Recurrent fusion of TMPRSS2and ETS transcription factor genes in prostate cancer.Science 2005;310:644–8.16.Soda M,Choi YL,Enomoto M,et al.Identification of the transformingEML4-ALK fusion gene in non-small-cell lung cancer.Nature2007;448:561–6.17.Stephens PJ,Greenman CD,Fu B,et al.Massive genomicrearrangement acquired in a single catastrophic event during cancer development.Cell2011;144:27–40.18.Sung WK,Zheng H,Li S,et al.Genome-wide survey of recurrent HBVintegration in hepatocellular carcinoma.Nat Genet2012;44:765–9. 19.Parsons DW,Jones S,Zhang X,et al.An integrated genomic analysisof human glioblastoma multiforme.Science2008;321:1807–12.20.Yoshida K,Sanada M,Shiraishi Y,et al.Frequent pathway mutations ofsplicing machinery in myelodysplasia.Nature2011;478:64–9.21.Tiacci E,Trifonov V,Schiavoni G,et al.BRAF mutations in hairy-cellleukemia.N Engl J Med2011;364:2305–15.114Cancer research and deep sequencingat University of Melbourne Library on September 25, 2014/Downloaded from22.Pasqualucci L,Dominguez-Sola D,Chiarenza A,et al.Inactivatingmutations of acetyltransferase genes in B-cell lymphoma.Nature 2011;471:189–95.23.Agrawal N,Frederick MJ,Pickering CR,et al.Exome sequencing ofhead and neck squamous cell carcinoma reveals inactivating mutations in NOTCH1.Science2011;333:1154–7.24.Stransky N,Egloff AM,Tward AD,et al.The mutational landscape ofhead and neck squamous cell carcinoma.Science2011;333:1157–60. 25.Steidl C,Shah SP,Woolcock BW,et al.MHC class II transactivatorCIITA is a recurrent gene fusion partner in lymphoid cancers.Nature 2011;471:377–81.26.Shah SP,Kobel M,Senz J,et al.Mutation of FOXL2in granulosa-celltumors of the ovary.N Engl J Med2009;360:2719–29.27.Puente XS,Pinyol M,Quesada V,et al.Whole-genome sequencingidentifies recurrent mutations in chronic lymphocytic leukaemia.Nature2011;475:101–5.28.Rothberg JM,Hinz W,Rearick TM,et al.An integrated semiconductordevice enabling non-optical genome sequencing.Nature2011;475:348–52.29.Branton D,Deamer DW,Marziali A,et al.The potentialand challenges of nanopore sequencing.Nat Biotechnol2008;26:1146–53.Jpn J Clin Oncol2013;43(2)115at University of Melbourne Library on September 25, 2014/Downloaded from。