innovation: new tools for functional mammalian cancer genetics

PERSPECTIVES

into the genetic alterations that cause phe-notypes such as telomerase activation willallow for the development of new therapeu-tic intervention strategies. One of the waysin which these insights can be obtained isthrough functional genetic approaches.

Functional genetic screens in model organ-isms, such as Drosophila and Caenorhabditiselegans, have provided valuable informationon components of mammalian signal-trans-duction pathways; however, such modelorganisms also have limitations for cancerresearch. First, not all human tumour-suppres-sor genes are conserved in lower eukaryotes.This is perhaps not surprising, as the increasedcomplexity of mammals — both in number ofcells and in number of cell types — mightrequire a far more sophisticated machinery tocontrol faithful DNA replication and cell dif-ferentiation than low-complexity and short-lived model organisms. Because sequenceanalysis of the C. elegans genome has notrevealed the MYC oncogene, and not all of thecomponents of the ARF–MDM2–p53 path-way are present in worms or flies, processesthat are regulated by such genes cannot bestudied in these organisms. Furthermore,control over the number of divisions a cellcan undergo by regulated expression of thetelomerase catalytic subunit is not seen inmice, indicating that even mouse cells are aninappropriate model to study cellular ageingby telomere shortening. Indeed, first insightsinto the regulation of telomerase expressionwere recently obtained in a genetic screen inhuman cells3. An even more complicated task

would be the development of genetic screensto study processes that are important fortumour angiogenesis or metastasis (such ascell adhesion, cell migration or oxygen sens-ing) in the genetic model systems that areavailable at present.

The use of mammalian cells to study thegenetics of cancer development has beenhampered by the lack of effective genetic toolsthat can be applied in these cells. In the pastfew years, our arsenal of genetic tools forfunctional mammalian cancer genetics hasexpanded markedly. So, what are thesegenetic tools and what successes have beenachieved with their use in cancer research?

Gain-of-function genetic screensAn alteration in cellular phenotype can beaccomplished through the introduction ofexogenous genetic material into cells. Suchan altered phenotype can often be readilyselected for in a population of cells. It ismost straightforward to select for a pheno-type that confers a new trait to cells, such asthe ability to form tumours in experimentalanimals, growth in semi-solid media oracquisition of a transformed morphology. Ineach case, it is important that the frequencywith which the recipient cells spontaneouslyacquire the selected phenotype is very low.In fact, engineering the proper cell system tocarry out a genetic screen is often far moretime-consuming and demanding than theperformance of the genetic screen itself. Asthe novel phenotype is introduced in a dom-inant fashion by the foreign DNA, suchgenetic screens are referred to as ‘gain-of-function’ screens. There are various tech-nologies available to introduce foreigngenetic material into cells.

DNA transfection. The discovery of methodsto introduce foreign DNA into mammaliancells, now some 30 years ago, provided thefirst powerful tools to study gene function ona large scale4. By introducing exogenousDNA into mammalian cells — resulting inintegration of the exogenous DNA into the

Knowledge of the function of individualgenes that encode components of cell-signalling pathways is crucial to ourunderstanding of normal growth control andits deregulation in cancer, but we havefunctional information for only ~15% ofhuman genes at present. Several newtechnologies have recently becomeavailable to identify gene function inmammalian cells using high-throughputgenetic screens. These new tools will makeit possible to identify new and innovativeclasses of anticancer drugs, including thosethat show synthetic lethal interactions withcancer-specific mutations.

Cells derived from cancerous lesions oftenresemble their normal counterparts, but dif-fer through the presence of a series of geneticalterations. These alterations are combina-tions of activating mutations in oncogenesand loss-of-function mutations in tumour-suppressor genes (for a review, see REF. 1).Other than these well-studied genetic alter-ations, cancer cells possess a number of phe-notypic alterations that do not have a knowngenetic basis. Tumour-cell immortality, forinstance, is often induced by activation ofthe telomerase catalytic subunit, which dis-ables a counting mechanism that limits thenumber of replications that a normal cellcan undergo2. However, the genetic alter-ations that lead to the activation of thetelomerase gene are largely unknown and itis a complex task to identify the genes thatare responsible for this activation. Insight

NATURE REVIEWS | CANCER VOLUME 3 | OCTOBER 2003 | 781

New tools for functional mammaliancancer genetics

Thijn R. Brummelkamp and René Bernards

I N N O VAT I O N

782 | OCTOBER 2003 | VOLUME 3 www.nature.com/reviews/cancer

P E R S P E C T I V E S

that the phenotype is caused by the cDNA car-ried by the retroviral vector in a second roundof selection. Because the background of false-positive cells in a genetic screen is often causedby genetic or epigenetic alterations of the targetcells, it is mandatory to recover the integratedproviruses from the selected target cells and toperform a second round of selection in a freshpopulation of target cells (the same applies togenetic screens that use other methods tointroduce exogenous DNA into cell popula-tions). When the frequency of cells with theselected phenotype is significantly higher insecond-round selection in the infected cellpopulation compared with a control popula-tion, this indicates that the selected phenotypeis virus-transduced and the selected cells can beused to recover and identify the cDNAs that areresponsible for the acquired phenotype. Toperform such second-round screens, efficienttools are needed to recover the integratedprovirus from the selected cells. Several meth-ods are now routinely used to accomplish this,which greatly facilitate functional cloningstrategies that involve retroviral vectors (BOX 1).

High-complexity retroviral cDNA expres-sion libraries have been used to clone trans-forming oncogenes in NIH-3T3 cells; in oneof the first of these screens, a large numberof genes with transforming activity wascloned, for example, RAF1 and β-catenin16

(for additional examples, see TABLE 1).Recently, other viruses have been adapted

to harbour exogenous DNA for gene-deliverystrategies. For instance, lentiviral vectors havean advantage over Moloney-virus-based vec-tors as they also infect resting cells, whereasMoloney-virus-based vectors only stably integrate into replicating cells20.

Replication-defective adenoviral vectorshave also been used to generate cDNA expres-sion libraries21. Advantages of this vector sys-tem are the very broad range of cell types thatcan be infected by adenoviral vectors (bothdividing and non-dividing cells) and the highstability in solution (resistant to severalrounds of freeze-thawing) of the adenoviralvectors compared with retroviral vectors.

Arrayed-format libraries. Arrayed-formatlibraries are collections of individually-clonedcDNAs deposited in a grid-like fashion inmulti-well plates.A key advantage of the use ofarrayed libraries is that when a desired pheno-type is detected in a screen it is directly knownwhich gene caused the phenotype. Whereasgenetic screens performed with large poly-clonal pools of cDNAs often require repeatedrounds of selection to identify the cDNA thatconfers the selected phenotype, the use ofarrayed-format libraries allows the direct

Viral vectors. DNA transfection into cells alsohas its limitations. First, uptake of foreignDNA varies between cell types and is oftenparticularly inefficient in primary (untrans-formed) cells. The use of such primary cellsfor genetic screens can be useful — forinstance, to identify genes with immortalizingor transforming activity. Second, recovery ofthe DNA fragment that has the selected bio-logical activity from the transfected cell isboth time-consuming and labour-intensive.Both types of problems are bypassed whenretroviral vectors are used as a vehicle tointroduce foreign DNA into cells.

The first retroviral vectors consisted ofsimple replication-defective Moloney murineleukaemia virus derivatives, which can be sta-bly maintained after infection, as they inte-grate into the host genome. High-complexitycDNA expression libraries can be cloned intosuch retroviral vectors and transfected into‘packaging cell lines’13 — which produce theviral gag, pol and env genes that are lackingfrom the replication-defective retroviral vec-tors — to obtain high-titre retroviral super-natants14–16 (FIG. 1). These are then used toinfect target cells, which leads to stable inte-gration of the provirus into the host genomeand expression of the cDNA that is containedby the provirus. These libraries have beenused successfully for the identification ofcDNAs that induce a selectable phenotypicalteration in the target cells14–19. An outline ofsuch a genetic screen is presented in FIG. 1.

When a complex mixture of retrovirus-encoded cDNAs is introduced into a popula-tion of cells and a rare variant with an alteredphenotype is identified, it is important to verify

genome of the recipient cell — it became possible to select for cells that had a stablealteration in phenotype. Probably the firstgain-of-function genetic screen in mam-malian cells that assigned a function to a specific gene was the identification of the frag-ments of adenovirus type 5 DNA that harbourtransforming activity5. About 10 years later,the introduction of sheared human tumourDNA into non-transformed NIH-3T3 cells ledto the identification of the RAS oncogene6,7.

Complex complementary DNA (cDNA)expression libraries (that is, libraries in whichboth abundant and rare messenger RNAs(mRNAs) are represented) can be efficientlyintroduced into mammalian cells using plas-mids that replicate episomally (that is, with-out integrating into the host genome, as theyhave their own origin of replication). Theseplasmids have two marked advantages: thereis no integration into the host genome —often a rather inefficient process — so there isa higher efficiency of cells that stably express acDNA; and the vectors can easily be shuttledback into Escherichia coli to identify thecDNA that encodes the selected biologicalactivity. For cDNA-library expression inhuman cells, Epstein–Barr virus (EBV)-basedcDNA expression shuttle vectors are usedmost often8. These cDNA libraries haveproved to be very useful for cloning genes,especially through complementation ofgenetic defects in cell function. For example,this cloning strategy has allowed the identifi-cation of several Fanconi anaemia genesthrough functional complementation of theintrinsic sensitivity of Fanconi anaemia cellsto mitomycin-C-mediated cell killing9–12.

ATG

AAAAA

Produce virus

Infect targetcells

Phenotypicselection

IdentifycDNAs

Second-roundselection

Infection/selectioncycle

cDNA library inretroviral vector

Figure 1 | Retroviral cDNA library screen. High-complexity retroviral complementary DNA (cDNA)libraries can be obtained through a single cloning step of cDNAs into a retroviral plasmid vector. High-titreviral supernatants are produced through transfection of the cDNA library vectors into retroviral packagingcells13. These viral supernatants are subsequently used to infect the target cells so that all the cDNAs thatare present in the library are expressed in the target-cell population. The infected cells can be used for oneround of phenotypic selection. After one or more cycles of infection/selection, the cells can be used forvirus recovery of the cDNAs that are responsible for the phenotype.


be screened to cover the entire complexity of acDNA library. This is particularly problematicin arrayed-format libraries, as the cDNAs arescreened one-by-one for biological activity,which is a very expensive process.

Another ingenious way to performgenetic screens in an arrayed format is to use‘reverse transfection’. In this approach, thetransfection is ‘reversed’; that is, DNA is notput on top of cells — as is done, for instance,in calcium phosphate transfection — but,rather, the cells are placed on top of spotted-DNA expression vectors. For example, theexpression-vector DNA can be spotted ontoglass slides, akin to the spotting of DNAmicroarrays22. Cells that are grown on theseDNA arrays take up the spotted DNA andcan subsequently be used for phenotypicselection (see BOX 2b).

Loss-of-function genetic screensGenetic screens that aim to identify genefunction through inactivation of a gene (or itscorresponding mRNA) are referred to as loss-of-function screens. Various technologieshave been developed to study the effects ofgene suppression in mammalian cells, includ-ing genetic suppressor elements (GSEs), anti-sense vectors, ribozymes, aptamer librariesand, more recently, RNA interference (RNAi).Below, we discuss these methodologies withemphasis on the recent developments in thefield of RNAi.

Genetic suppressor elements. GSEs are shortand biologically active cDNA fragments thatinterfere with the function of their cognategene. GSEs can either encode peptide frag-ments of proteins that act as dominant-negative inhibitors of the full-length proteinor antisense RNA molecules that interferewith the function of the complementarymRNA. When expressed in retroviral vec-tors, high-complexity GSE libraries can bepowerful tools to identify cancer-relevantgenes, including determinants of cellularsensitivity to anticancer drugs23,24 and regu-lators of transcription25. A limitation of thisapproach is that only a few fragments of anygiven cDNA act as GSEs, which limits theuse of this technology in genome-widescreens. A selection for GSEs with trans-forming activities resulted in the isolation ofING1 as a regulator of cell proliferation26.

Antisense vectors. Expression of antisensecDNA molecules can be an effective way toinhibit expression of a certain gene.Libraries that express antisense RNA mole-cules can be especially useful for studyingtumour-suppressor genes (for a review, see

linkage of genotype to phenotype in a singleround of screening. This can save an enormous amount of time and effort, and,importantly, also allows selection of pheno-types — such as the induction of apoptosis —that are counter-selected when genetic screensare carried out in a polyclonal format. The sta-bility in solution of the adenoviral vectorsgreatly facilitates the generation and handlingof such arrayed-format cDNA libraries21 (see

BOX 2a). An inherent problem that is associ-ated with the generation and use of cDNAlibraries is that some mRNAs are far moreabundant than others, leading to unequalrepresentation of cDNAs in a library. Unlessmeasures are taken to prevent this (for exam-ple, through ‘normalization’; a tediousprocess that is often carried out at theexpense of the average length of cDNAs inthe library), large numbers of cDNAs need to


Box 1 | Recovery of integrated retroviral vectors

Several techniques exist for the efficient recovery of the complementary DNAs (cDNAs) that areharboured by integrated proviruses from the genome of a selected cell population (see figure).First, cDNAs can be recovered from genomic DNA isolated from selected cells by polymerasechain reaction (PCR), using primers located in the retroviral-expression cassette that flanks thecDNA insert. These PCR-amplified cDNAs can then be sequence analysed and re-cloned into aretroviral vector14,16. When the isolated cDNAs are cloned into viral vectors, new viralsupernatants can be obtained by transfection of packaging cells followed by infection of targetcells for second-round selection. PCR isolation, however, is more suitable for the isolation of oneor a few genes from target cells, rather than for the isolation of more complex mixtures of cDNAsfrom the target-cell population (because of selective amplification of certain cDNAs). Toovercome this limitation, shuttle vectors have been developed, which allow simultaneous recoveryof a high number of integrated proviruses from a polyclonal population of target cells. Thesevectors contain sequences — which are located between the viral long terminal repeats (LTRs) —for maintenance as plasmids in Escherichia coli (replicon), so that they are present in theintegrated provirus in the infected cell. Sequences that are recognized by site-specificrecombinases (such as CRE, which recognizes specific sites (LOXP) in DNA) or restrictionenzymes are located in the viral LTRs, so that the proviruses can be excised from the cellulargenome by the recombinase or restriction enzyme, circularized and transformed into bacteria68,69.Plasmid DNA of a mixture of these proviruses can then be isolated and transfected into packagingcells. Finally, integrated proviruses can be mobilized in a natural way by superinfection of thetarget cells with wild-type Moloney virus, which complements the provirus by expressing gag, poland env in the cells70. Replication-defective retroviruses that harbour cDNAs will then bepackaged into retroviral particles, and viral supernatants that contain both replication-deficientcDNA-encoding viruses and wild-type Moloney viruses can be harvested. This mixture of virusescan than be used to infect a fresh batch of target cells for a second round of selection.

Replicon3′ LTR5′ LTR

PCR amplification

cDNA5′ LTR 3′ LTR

Shuttle vector

cDNA

cDNA

5′ LTR 3′ LTR

Moloney mobilization

cDNA

cDNA LTRReplicon

Sequence analysis

Clone in virus Virus production

Second-roundselection

LOXP

LOXP

LOXP



REF. 27), and EBNA (Epstein–Barr-virusnuclear antigen) vectors that express aninducible antisense cDNA library have beenused successfully to clone several antisensecDNA fragments that protect HeLa cellsagainst γ-interferon-induced apoptosis. Thegenes that were affected by these antisensecDNAs were named DAP (for death-associatedproteins) genes28–30. As expected, overexpres-sion of DAP genes is often incompatible withcell survival29.

Ribozymes. Another way to inhibit geneexpression is by using catalytic RNA mole-cules — known as ribozymes — that bindto and cleave certain RNA targets. RNA tar-get molecules are recognized throughsequence complementarity between theribozyme and the target RNA molecule.Ribozymes can be effectively expressed incells using viral vectors, and libraries can becreated through the insertion of random-ized sequences in the ribozyme RNAsequence. These libraries can be used ingenetic screens to identify ribozymes thatinhibit certain mRNA molecules. In thisway, ID4 has been identified as a regulatorof BRCA1 expression31 and PPAN as a genethat inhibits anchorage-independent prolif-eration32. Also, functional differencesbetween the closely related histone acetyltransferases p300 and CBP were identified33.

Aptamer libraries. One method that doesnot depend on cDNA libraries or informa-tion on genome sequence is the use of ran-dom short-peptide (aptamer) expressionlibraries. Aptamer expression libraries canbe used to find peptides with a desired bio-logical activity34. This can be caused by theexpressed peptide activating or inhibiting acellular factor. For stable expression ofpeptides in mammalian cells, libraries areconstructed that express aptamers in thecontext of protease-resistant scaffold struc-tures. These libraries allow the isolation ofpeptides that confer cellular resistance to theanticancer drug taxol, presumably throughupregulation of multidrug transporters35. Alimitation of this approach is that highlycomplex mixtures of aptamers are requiredto identify active aptamers in any pathway(far more complex than, for example, cDNAlibraries). In addition, it can be cumber-some to identify the cellular factor that isaffected by the biologically active peptide. Inmost cases, quite a bit of additional work(for example, a yeast two-hybrid screen) isrequired to identify the cellular protein thatis affected by the interaction with theselected short peptide.

Table 1 | Genes relevant to cancer identified in functional genetic screens

Gene Type of screen Phenotypic selection Reference

HRAS DNA transfection Transformation 6

β-catenin Retroviral cDNA expression library Transformation 16

MIF Retroviral cDNA expression library Inhibits p53-induced 71cell-cycle arrest

Twist/Dermo Retroviral cDNA expression library Inhibits c-MYC-induced 72apoptosis

BCL6 Retroviral cDNA expression library Inhibits p53-dependent 17senescence

FACC EBNA cDNA expression library Complementation of genetic 9defect

NEMO Retroviral cDNA expression library Complementation of genetic 73defect

DAPK EBNA antisense library Resistance against 30γ-interferon-induced apoptosis

CYLD High-throughput RNAi NF-κB-regulated reporter 64

TOSO Retroviral cDNA expression library Inhibits FAS-induced 19apoptosis in T cells

ING1 GSE screen Selection for GSEs that 26promote neoplastic transformation

TFE3 Retroviral cDNA expression library Co-operation with SMAD 70proteins to stimulate TGF-β-induced transcription

BCL6; B-cell non-Hodgkin’s lymphoma 6; CYLD, cylindromatosis gene; DAPK, death-associated-protein kinase; EBNA, Epstein–Barr-virus nuclear antigen; FACC, Fanconi’s anaemia group C; GSE,genetic suppressor element; ING1, inhibitor of growth family, member 1; MIF, migration inhibitory factor;NEMO, NF-κB essential modifier; NF-κB, nuclear factor κB; RNAi, RNA interference; TFE3, transcriptionfactor µE3; TGF-β, transforming growth factor-β.

Box 2 | Arrayed-format libraries

Arrayed-format libraries contain expression vectors that are deposited in an arrayed manner indefined locations. Using this approach, an adenoviral expression library has been generated thatcontains separated viral stocks that are able to induce expression of individual complementaryDNAs (cDNAs; see a)21. These viral stocks can be used to infect target cells and these cells can beused for phenotypic analysis in a high-throughput manner. Another way to screen arrayed-format libraries is the use of reverse transfection combined with microarrays of spotted-cDNAexpression vectors22 (see b). Cells can be transfected (by reverse transfection) on microarrayswith individually spotted cDNAs, resulting in spots of locally transfected cells with a certaincDNA. These cells can then be used for phenotypic screens.

a Adenoviral arrayed-format libraries b Reverse transfection of arrayed cDNA libraries

Virus production

Clone individual cDNAsinto adapter plasmid

Adenoviral stocks

Infect target cells

Functional assay

Cells

Array containing spotted-cDNA expression vectors

Reverse transfection and phenotypic analysis


short RNAs can recruit mammalian RISC totarget mRNAs and induce their specific degra-dation, but do not activate the γ-interferonpathway45. Sequence-assisted annotation of allhuman open reading frames (ORFs), alsoknown as the ORFeome, can function as a guide to construct a collection of siRNAsthat are designed to inhibit the expression ofindividual genes.

Although transfection of human cells withsiRNAs that are produced in vitro is a quickway to silence a gene of interest, there are sev-eral problems associated with the use of thismethod. The first obstacle is the transientnature of the inhibition of gene expression, asthe siRNA molecules are unstable inside thecell. Second, several cell types (for example,primary cells) are difficult to transfect withhigh efficiency. Third, chemically synthesizedsiRNAs are expensive, which becomes anobstacle when genome-wide RNAi screens arecontemplated in mammalian cells.

Several groups have effectively resolvedthese issues through the development of vec-tor systems that mediate stable production ofsiRNA-like molecules in mammalian cells46–50

(FIG. 3). These vectors use RNA-polymerase IIIpromoters to direct the synthesis of shorthairpin RNA (shRNA) molecules, which areprocessed intracellularly into siRNA-likemolecules. The shRNAs contain a perfectlydouble-stranded stem of 19–29 base pairs,which is identical in sequence to the mRNAthat is targeted for suppression, connected bya loop of 6–9 bases, which is efficientlyremoved in vivo. Vector-produced shRNAmolecules are as effective as siRNAs that aregenerated in vitro in inhibiting gene expres-sion, and, in addition, can be used to studyloss-of-function phenotypes that developover longer periods of time. Furthermore,stable integration of these expression cas-settes can be efficiently achieved throughretroviral or lentiviral delivery51–57.

Several observations indicate that vector-mediated RNAi can also be used to study loss-of-function phenotypes in vivo. Humancancer cells that stably express siRNAsdirected against the KRAS oncogene lose theirtransformed phenotype not only in vitro, butalso in vivo after injection into nude mice52.This indicates that gene inhibition by RNAi ismaintained in a living animal. Consistentwith this, in vivo transfection of siRNAs intothe livers of mice can inhibit expression of co-transfected reporter plasmids58. Transgenicembryos have also been generated thatexpress an shRNA directed against a specificmouse gene, and these mice show the samephenotype as mice carrying a null mutationin the same gene59. Together, these studies

RNA interference. The ability to manipulategene expression in the multicellular worm C. elegans — by techniques such as mutage-nesis by transposon tagging, introduction ofDNA deletions by chemical mutagenesis andby antisense RNA — made this organismpopular among geneticists36. The availabilityof these methods, combined with the shortreproductive cycle of C. elegans, has made itpossible to dissect complex genetic pathwaysthat regulate diverse biological processes.The discovery of a cellular response againstdouble-stranded RNA37, however, has pro-vided one of the most powerful tools to turnoff gene expression, and this has revolution-ized loss-of-function genetics, initially in C. elegans and Drosophila and also morerecently in mammalian cells.

This cellular response, known as RNAi, isan ancient defence mechanism to protect cellsfrom foreign invasion by agents such as trans-posons and viruses (for reviews, see REFS

38,39). The double-stranded RNAs that areproduced by integrated transposons or byreplicating viruses are processed into shortdouble-stranded RNAs, named short interfer-ing RNAs (siRNAs), that serve as a signal toactivate the RNA-induced silencing complex(RISC), which degrades homologous mRNAmolecules. The introduction into a worm ofdouble-stranded RNAs that are identical tocellular transcripts will cause degradation ofthe corresponding mRNA molecules, leadingto post-transcriptional gene silencing (FIG. 2).

Genome-wide high-throughput RNAiscreens have been performed with successby feeding worms collections of bacteria-containing plasmids that express long double-stranded RNAs that are homolo-gous to a large number of worm genes.Such large-scale genetic loss-of-functionscreens have allowed the identification ofgenes that affect worm development40,41,regulate fat storage42 or serve to protect thegenome against mutations43. A similarapproach was used in a Drosophila cell lineto isolate Hedgehog pathway components44.An advantage of these screens is that it isimmediately known which gene is inhib-ited. By contrast, mutagenesis screensrequire laborious mapping of the mutationbefore the loss-of-function phenotype canbe linked to a specific gene.

The completion of the sequence of thehuman genome and the identification of theencoded genes opens the possibility to useRNAi for the systematic inactivation of largenumbers of human genes. However, the useof RNAi in mammalian cells is hampered bythe fact that the introduction of long double-stranded RNA molecules (which areeffective in worms and flies to silence geneexpression) leads to nonspecific toxicity inmost somatic cells because the γ-interferonpathway is activated. This problem can besolved by the use of chemically synthesized21–23 base-pair double-stranded siRNAsthat are introduced by transfection. These


mRNA

DNA

siRNA

RISC

Chemically synthesized

Vector expressed

ATG

AAAAA

ATG

ATG

AAAAA

AAAAA

Figure 2 | RNA interference. Short double-stranded RNA molecules — short interfering RNAs(siRNAs) — that are introduced into a cell will be incorporated into the RNA-induced silencing complex(RISC). The antisense strand of the siRNA molecule is then used for the identification of cellularmessenger RNA (mRNA) molecules that contain a perfect complementary sequence. These mRNAmolecules are subsequently cleaved by RISC and become unstable. As a result of this, the proteinencoded by the mRNA will not be produced. The siRNAs can be synthetically synthesized andtransfected into the cells (this will only allow transient inhibition of gene expression) or they can beproduced in the cells by DNA-based vector systems that direct expression of siRNA-like molecules(this can also be used for long-term silencing).



To facilitate the identification of syntheticlethal interactions in mammalian cells withvector-based siRNA libraries, we propose herea technology that we have dubbed ‘siRNA bar-code screens’, which is explained in more detailin BOX 3. Molecular barcodes were first used inyeast genetic screens, in which large numbersof yeast genes were inactivated by insertion of a DNA segment that contained a unique 20-nucleotide sequence (the ‘molecular barcode’) to mark the individual knockout

mutation in TP53, making it a tumour-cell-specific drug target. Inhibition of suchgenes in normal cells (that contain wild-typeTP53) will be much less toxic, providing atherapeutic window for cancer treatment. Theproblem in identifying siRNA vectors thatshow a synthetic lethal interaction with can-cer-specific mutations is that these vectors arespecifically lost from cells, making it hard toidentify interactions when such screens areperformed in a polyclonal format.

indicate that it might be possible to performlarge-scale RNAi screens in mice to identifyloss-of-function phenotypes that cannot bereadily selected in vitro.

One potential complication in usingsiRNAs in genetic screens is that siRNAs arenot always completely target-specific. Some‘off-target’ side effects have been reported60–62,which in most cases can be attributed to par-tial homology of the un-intended target tothe siRNA. Even though these off-targeteffects of RNAi are modest, compared withantisense RNA approaches, for example, ithighlights the need to validate an identifiedphenotype caused by an siRNA with a sec-ond, independent siRNA directed against thesame transcript. In rare cases, shRNA vectorshave been reported to induce an interferonresponse, the molecular mechanism of whichremains unclear63.

The reduced cost of vector-based RNAi alsomakes it possible to carry out genome-widegenetic screens for loss-of-function pheno-types in mammalian cells. High-throughputcloning of short-hairpin-encoding DNAoligonucleotides into retroviral vectors con-taining RNA-polymerase III promoters is nowin progress in several places, including our ownlaboratory. As a proof of concept for thegenome-wide RNAi vector library approach,we have recently constructed a ‘gene family’knockdown library, which targets most ofthe members of the family of de-ubiquitylat-ing enzymes. Using this library, we identifiedthe cylindromatosis tumour-suppressorgene (CYLD) as a key regulator of NF-κBactivity in a high-throughput reporter-basedgenetic screen64.

Novel applications of RNAi vectorsOne of the more attractive applications ofRNAi vector libraries is the identification of‘synthetic lethal’ interactions in mammaliancells — a combination of two non-lethalmutations that, together, result in cell death.Such a genetic association indicates that afunctional interaction between the corre-sponding gene products exists. In simpleorganisms, such as yeast and C. elegans, manysynthetic lethal interactions have been identi-fied; by contrast, there are few examples ofsynthetic lethality in mammalian systems.This is probably because the tools to efficientlyidentify such interactions in mammalian cellshave, so far, been lacking. Identification of syn-thetic lethal interactions might be of keyimportance for the identification of novel andmore powerful classes of cancer drug targets.For instance, inhibition of a gene that shows a synthetic lethal interaction with loss of TP53will only be toxic in cells that contain a

ATG

AAAAA

ATG

AAAAA

ATG

AAAAA

mRNA

Design siRNA sequences for every gene

a

b

e

c

d

Clone DNA oligonucleotidesencoding shRNAs into plasmids

P-III TTTTT19nt 19nt

Viralvector


Viralvector


Viralvector

Produce virus

Transfect siRNA in target cells

ATG

AAAAA

ATG

AAAAA

ATG

AAAAA

5 LTR 3 LTR

Infected target cells

Transfect targetcells

′ ′

Figure 3 | Loss-of-function screens using RNAi. a | To construct a short interfering RNA (siRNA) library,several siRNA sequences (because not every siRNA sequence is active, most groups, including our own,use three or more sequences51) are selected to target and inhibit particular genes. b | These sequences can be used to generate siRNA molecules that can be chemically synthesized anddirectly transfected into target cells in a high-throughput manner, and these cells can then be used forphenotypic selection. c | Alternatively, DNA oligonucleotides can be designed to mediate expression ofshort hairpin RNAs (shRNAs) that mimic siRNAs46–50. These oligonucleotides are then cloned into modified(retroviral, lentiviral or adenoviral) vectors that contain RNA-polymerase-III (P-III) promoter cassettes. d | Retroviral expression of the designed shRNAs results in a population of infected cells in which eachinfected cell expresses one or more different shRNAs designed against one or more mRNA sequences. e | Alternatively, the plasmid vectors can be transfected directly into target cells to drive expression ofshRNA molecules in these cells.


G-protein-coupled receptors (GPCRs),hydrolases and ion channels are, in general,far more likely to be inhibited by small-molecule compounds than, for instance,transcription factors. So, even though we

yeast strains65–67. The molecular barcode servesas a ‘strain identifier’, which makes it possible tofollow the relative abundance of each mutantyeast strain by measuring the abundance of thebarcodes in the population. Quantification ofthe relative abundance of mutant strains undervarious experimental stress conditions wasachieved through polymerase chain reaction(PCR) amplification of the barcodes, followedby hybridization to microarrays containing thebarcode sequences.

By analogy to the yeast barcode approach,we noted that the DNA fragment of eachsiRNA expression vector that encodes theRNA-hairpin transcript has a unique 19-base-pair target-specific sequence. Therefore,introduction of such a vector into mam-malian cells results in the creation of a taggedknockdown cell that carries a permanentgene-specific identifier (known as a knock-down identifier; see BOX 3a). This molecularbarcode can be recovered by PCR amplifica-tion using vector-derived PCR primers thatflank the hairpin-encoding DNA sequence.Fluorescent labelling of the PCR productallows it to be identified by hybridization tomicroarrays that contain the gene-specificknockdown barcode oligonucleotides (seeBOX 3a). The relative abundance of each bar-coded DNA fragment in the cell population isinfluenced by the effect that each knockdownvector has on cellular fitness under the experi-mental conditions and can be quantifiedusing a DNA microarray that contains thebarcode DNA fragments. Comparing theeffects of knockdown vectors on cellular fit-ness in pairs of cells that have defined geneticlesions will make it possible to identify syn-thetic lethal phenotypes in human cells, assiRNA vectors that are specifically lost fromone population can be identified in this way.

Future perspectives/conclusionsMany of the cytostatic drugs that are used atpresent were developed before the molecularbasis of cancer was understood in any detail.As a result, most of these drugs are rather non-specific, acting indiscriminately on all rapidlydividing cells. Therefore, the ‘therapeutic win-dow’ of such drugs (that is, the differentialsensitivity of cancer cells versus normal cells)is limited, which causes many of the undesiredside effects of anticancer therapy. Over the pastdecades, we have gained a detailed knowledgeof the pathways that are deregulated in cancerand many of the components of these cancer-relevant pathways have been identified. Thebenefit to the cancer patient of this quantumleap in knowledge has, with some notableexceptions (such as imatinib (Glivec)), beenlimited. One reason for this is that not all

components of cancer-relevant pathways aresuitable targets for drug development.Pharmaceutical companies have long knownthat certain classes of enzymes are more‘druggable’ than others. For instance,


Box 3 | SiRNA barcode screens

Expression of short hairpin RNA (shRNA)molecules in mammalian cells by stablyintegrated vectors not only creates a gene-specific knockdown phenotype, but alsointroduces a gene-specific fingerprint(molecular barcode) in cells that expressthese siRNAs (see a). The 19-base-pair target-gene-specific insert of the siRNA vector isunique in sequence and, therefore,introduction of such a vector intomammalian cells results in the creation of atagged knockdown cell carrying a permanentgene-specific identifier. This molecularbarcode can be recovered by polymerasechain reaction (PCR) amplification usingvector-derived PCR primers that flank thehairpin-encoding DNA sequence. When alarge collection of siRNA expression vectorsis expressed in a population of cells, PCRamplification of the hairpin inserts willresult in a mixture of barcode sequences thatcorrespond to the mixture of knockdownvectors that is present in the cell population.The relative abundance of each barcodedsiRNA vector in the cell population isinfluenced by the effect that each knockdownvector has on cellular fitness under theexperimental conditions. The abundance ofeach barcode can be quantified by labellingthe PCR product with a fluorescent dye,followed by hybridization to a DNAmicroarray consisting of barcode-complementary DNA fragments (see b) —these oligonucleotides must at least containthe 19 gene-specific nucleotides.

In principle, siRNA barcode screens allowthe detection of genetic interactions betweenlarge sets of genes and almost any biologicalsignal of interest. A particularly useful application of the siRNA barcode screen is to comparetwo cell populations, both of which harbour a collection of siRNA vectors, but only one of whichis exposed to a biological signal of interest (see b) — for example, DNA damage, apoptosis-inducing agents, cytotoxic drugs or inactivation of a tumour-suppressor gene. After PCRamplification, barcoded fragments are labelled with a fluorescent dye (for example, Cy5) andhybridized against barcoded DNA fragments that have been PCR amplified from a controlpopulation of cells that contain the same collection of knockdown vectors but that were notexposed to the biological signal of interest. This control population of barcoded DNA sequencesis then labelled with a different fluorescent dye (for example, Cy3). Simultaneous hybridizationof the Cy5- and Cy3-labelled barcoded DNA fragments allows for the identification of changes inthe relative abundance of knockdown vectors in response to the stimulus applied. When a vectoris lost specifically in the treated cells, but not in the untreated cells, this indicates that theknockdown vector is synthetically lethal with the stimulus used in the assay. The quantitativenature of DNA-array hybridization makes it possible to analyse large numbers of knockdownvectors in parallel assays, which should greatly facilitate the identification of synthetic lethalinteractions in mammalian cell systems.

Culture to allowselection

PCR amplify barcodes

Hybridize to microarray


P-III TTTTT

Gene-specific sequenceMolecular barcode

siRNAvirus

Infect with RNAivector library

Signal of interest

+ –

P-III TTTTT

Label PCR productCy5 Cy3

DNAmicroarray

Array contains barcode oligos

a

b



30. Deiss, L. P., Feinstein, E., Berissi, H., Cohen, O. & Kimchi, A.Identification of a novel serine/threonine kinase and a novel15-kD protein as potential mediators of the γ interferon-induced cell death. Genes Dev. 9, 15–30 (1995).

31. Beger, C. et al. Identification of Id4 as a regulator ofBRCA1 expression by using a ribozyme-library-basedinverse genomics approach. Proc. Natl Acad. Sci. USA98, 130–135 (2001).

32. Welch, P. J. et al. Identification and validation of a geneinvolved in anchorage-independent cell growth controlusing a library of randomized hairpin ribozymes.Genomics 66, 274–283 (2000).

33. Kawasaki, H. et al. Distinct roles of the co-activatorsp300 and CBP in retinoic-acid-induced F9-celldifferentiation. Nature 393, 284–289 (1998).

34. Colas, P. et al. Genetic selection of peptide aptamers thatrecognize and inhibit cyclin-dependent kinase 2. Nature380, 548–550 (1996).

35. Xu, X. et al. Dominant effector genetics in mammaliancells. Nature Genet. 27, 23–29 (2001).

36. Jorgensen, E. M. & Mango, S. E. The art and design ofgenetic screens: Caenorhabditis elegans. Nature Rev.Genet. 3, 356–369 (2002).

37. Fire, A. et al. Potent and specific genetic interference bydouble-stranded RNA in Caenorhabditis elegans. Nature391, 806–811 (1998).

38. Plasterk, R. H. RNA silencing: the genome’s immunesystem. Science 296, 1263–1265 (2002).

39. Hannon, G. J. RNA interference. Nature 418, 244–251 (2002).40. Fraser, A. G. et al. Functional genomic analysis of

C. elegans chromosome I by systematic RNAinterference. Nature 408, 325–330 (2000).

41. Kamath, R. S. et al. Systematic functional analysis of theCaenorhabditis elegans genome using RNAi. Nature 421,231–237 (2003).

42. Ashrafi, K. et al. Genome-wide RNAi analysis ofCaenorhabditis elegans fat regulatory genes. Nature 421,268–272 (2003).

43. Pothof, J. et al. Identification of genes that protect the C. elegans genome against mutations by genome-wideRNAi. Genes Dev. 17, 443–448 (2003).

44. Lum, L. et al. Identification of Hedgehog pathwaycomponents by RNAi in Drosophila cultured cells.Science 299, 2039–2045 (2003).

45. Elbashir, S. M. et al. Duplexes of 21-nucleotide RNAsmediate RNA interference in cultured mammalian cells.Nature 411, 494–498 (2001).

46. Brummelkamp, T. R., Bernards, R. & Agami, R. A systemfor stable expression of short interfering RNAs inmammalian cells. Science 296, 550–553 (2002).

47. Paddison, P. J., Caudy, A. A., Bernstein, E., Hannon, G. J.& Conklin, D. S. Short hairpin RNAs (shRNAs) inducesequence-specific silencing in mammalian cells. GenesDev. 16, 948–958 (2002).

48. Paul, C. P., Good, P. D., Winer, I. & Engelke, D. R.Effective expression of small interfering RNA in humancells. Nature Biotechnol. 20, 505–508 (2002).

49. Miyagishi, M. & Taira, K. U6 promoter-driven siRNAs withfour uridine 3′ overhangs efficiently suppress targetedgene expression in mammalian cells. Nature Biotechnol.20, 497–500 (2002).

50. Lee, N. S. et al. Expression of small interfering RNAstargeted against HIV-1 rev transcripts in human cells.Nature Biotechnol. 20, 500–505 (2002).

51. Paddison, P. J. & Hannon, G. J. RNA interference: thenew somatic cell genetics? Cancer Cell 2, 17–23 (2002).

52. Brummelkamp, T. R., Bernards, R. & Agami, R. Stablesuppression of tumorigenicity by virus-mediated RNAinterference. Cancer Cell 2, 243–247 (2002).

53. Dirac, A. M. & Bernards, R. Reversal of senescence inmouse fibroblasts through lentiviral suppression of p53.J. Biol. Chem. 278, 11731–11734 (2003).

54. Abbas-Terki, T., Blanco-Bose, W., Deglon, N., Pralong, W.& Aebischer, P. Lentiviral-mediated RNA interference.Hum. Gene Ther. 13, 2197–2201 (2002).

55. Rubinson, D. A. et al. A lentivirus-based system tofunctionally silence genes in primary mammalian cells,stem cells and transgenic mice by RNA interference.Nature Genet. 33, 401–406 (2003).

56. Scherr, M., Battmer, K., Ganser, A. & Eder, M. Modulationof gene expression by lentiviral-mediated delivery of smallinterfering RNA. Cell Cycle 2, 251–257 (2003).

57. Qin, X. F., An, D. S., Chen, I. S. & Baltimore, D. InhibitingHIV-1 infection in human T cells by lentiviral-mediateddelivery of small interfering RNA against CCR5. Proc. NatlAcad. Sci. USA 100, 183–188 (2003).

58. McCaffrey, A. P. et al. RNA interference in adult mice.Nature 418, 38–39 (2002).

59. Kunath, T. et al. Transgenic RNA interference in ES cell-derived embryos recapitulates a genetic null phenotype.Nature Biotechnol. 21, 559–561 (2003).

60. Jackson, A. L. et al. Expression profiling reveals off-targetgene regulation by RNAi. Nature Biotechnol. 21,635–637 (2003).

1. Hanahan, D. & Weinberg, R. A. The hallmarks of cancer.Cell 100, 57–70 (2000).

2. Sherr, C. J. & DePinho, R. A. Cellular senescence: mitoticclock or culture shock? Cell 102, 407–410 (2000).

3. Lin, S. Y. & Elledge, S. J. Multiple tumor suppressorpathways negatively regulate telomerase. Cell 113,881–889 (2003).

4. Graham, F. L. & van der Eb, A. J. A new technique for theassay of infectivity of human adenovirus 5 DNA. Virology52, 456–467 (1973).

5. Graham, F. L., van der Eb, A. J. & Heijneker, H. L. Sizeand location of the transforming region in humanadenovirus type 5 DNA. Nature 251, 687–691 (1974).

6. Shih, C. & Weinberg, R. A. Isolation of a transformingsequence from a human bladder carcinoma cell line. Cell29, 161–169 (1982).

7. Goldfarb, M., Shimizu, K., Perucho, M. & Wigler, M.Isolation and preliminary characterization of a humantransforming gene from T24 bladder carcinoma cells.Nature 296, 404–409 (1982).

8. Belt, P. B. et al. Efficient cDNA cloning by directphenotypic correction of a mutant human cell line (HPRT-)using an Epstein–Barr virus-derived cDNA expressionvector. Nucl. Acids Res. 19, 4861–4866 (1991).

9. Strathdee, C. A., Gavish, H., Shannon, W. R. & Buchwald, M.Cloning of cDNAs for Fanconi’s anaemia by functionalcomplementation. Nature 356, 763–767 (1992).

10. Lo Ten Foe, J. R. et al. Expression cloning of a cDNA forthe major Fanconi anaemia gene, FAA. Nature Genet. 14,320–323 (1996).

11. de Winter, J. P. et al. The Fanconi anaemia group G geneFANCG is identical with XRCC9. Nature Genet. 20,281–283 (1998).

12. de Winter, J. P. et al. The Fanconi anaemia gene FANCFencodes a novel protein with homology to ROM. NatureGenet. 24, 15–16 (2000).

13. Pear, W. S., Nolan, G. P., Scott, M. L. & Baltimore, D.Production of high-titer helper-free retroviruses bytransient transfection. Proc. Natl Acad. Sci. USA 90,8392–8396 (1993).

14. Kitamura, T. et al. Efficient screening of retroviral cDNAexpression libraries. Proc. Natl Acad. Sci. USA 92,9146–9150 (1995).

15. Zannettino, A. C., Rayner, J. R., Ashman, L. K., Gonda, T. J.& Simmons, P. J. A powerful new technique for isolatinggenes encoding cell surface antigens using retroviralexpression cloning. J. Immunol. 156, 611–620 (1996).

16. Whitehead, I., Kirk, H. & Kay, R. Expression cloning ofoncogenes by retroviral transfer of cDNA libraries. Mol.Cell Biol. 15, 704–710 (1995).

17. Shvarts, A. et al. A senescence rescue screen identifiesBCL6 as an inhibitor of anti-proliferative p19(ARF)-p53signaling. Genes Dev. 16, 681–686 (2002).

18. Berns, K., Hijmans, E. M., Koh, E., Daley, G. Q. &Bernards, R. A genetic screen to identify genes thatrescue the slow growth phenotype of c-myc nullfibroblasts. Oncogene 19, 3330–3334 (2000).

19. Hitoshi, Y. et al. Toso, a cell surface, specific regulator ofFas-induced apoptosis in T cells. Immunity 8, 461–471(1998).

20. Lewis, P. F. & Emerman, M. Passage through mitosis isrequired for oncoretroviruses but not for the humanimmunodeficiency virus. J. Virol. 68, 510–516 (1994).

21. Michiels, F. et al. Arrayed adenoviral expression librariesfor functional screening. Nature Biotechnol. 20,1154–1157 (2002).

22. Ziauddin, J. & Sabatini, D. M. Microarrays of cellsexpressing defined cDNAs. Nature 411, 107–110(2001).

23. Gudkov, A. V. et al. Cloning mammalian genes byexpression selection of genetic suppressor elements:association of kinesin with drug resistance and cellimmortalization. Proc. Natl Acad. Sci. USA 91,3744–3748 (1994).

24. Levenson, V. V. et al. A combination of geneticsuppressor elements produces resistance to drugsinhibiting DNA replication. Somat. Cell Mol. Genet. 25,9–26 (1999).

25. Novoa, I., Zeng, H., Harding, H. P. & Ron, D. Feedbackinhibition of the unfolded protein response by GADD34-mediated dephosphorylation of eIF2α. J. Cell Biol. 153,1011–1022 (2001).

26. Garkavtsev, I., Kazarov, A., Gudkov, A. & Riabowol, K.Suppression of the novel growth inhibitor p33ING1promotes neoplastic transformation. Nature Genet. 14,415–420 (1996).

27. Kimchi, A. Antisense libraries to isolate tumor suppressorgenes. Methods Mol. Biol. 222, 399–412 (2003).

28. Deiss, L. P. & Kimchi, A. A genetic tool used to identifythioredoxin as a mediator of a growth inhibitory signal.Science 252, 117–120 (1991).

29. Cohen, O. Kimchi, A. DAP-kinase: from functional genecloning to establishment of its role in apoptosis andcancer. Cell Death Differ. 8, 6–15 (2001).

have detailed knowledge of a large number ofcomponents of cancer-relevant pathwayssuch as the RB, p53 and NF-κB pathways,only a few of the components of these path-ways fall into the category of ‘druggable’ tar-gets. As a result, pharmaceutical companiesare still faced with a lack of high-quality drugtargets to specifically inhibit those pathwaysthat are deregulated in cancer.

Functional genetic screens are eminentlysuitable tools to identify novel components ofpathways of interest. Indeed, it is probable thatmany new components of important cancer-relevant pathways can still be identifiedthrough functional genetic approaches.A pow-erful approach in drug-target discovery that isnow within reach is to first select a family of‘druggable’ genes, generate shRNA vectorsagainst all members of the gene family, andthen use such a ‘gene family knockdownlibrary’ to investigate whether any members ofthis gene family act in a pathway that is knownto be deregulated in cancer. The feasibility ofthis approach was recently confirmed by show-ing that one of 50 de-ubiquitylating enzymesacted specifically in the NF-κB pathway64.

In addition, technologies such as thesiRNA barcode screens described here willallow the identification of synthetic lethalinteractions, giving rise to unprecedentednew classes of potent anticancer drug targets.For example, genetic screens could be used inwhich the ability of shRNA vectors to modu-late cellular fitness in the presence or absenceof a second, cancer-specific mutation (forexample, mutant RAS, overexpressed MYC,loss of TP53, PTEN or RB) is compared,allowing the detection of synthetic lethal phe-notypes. A similar approach could also beused to identify siRNA vectors that show agenetic interaction with established pharma-cological agents, such as anticancer drugs.ShRNA vectors that render cells more sensi-tive to a pharmacological agent can be readilyidentified in this way and inhibition of suchtargets should, in theory, act in synergy withthe anticancer drug used in the screen.

We foresee that, together, the advances infunctional genetic screening technology inmammalian cells that are described here willhave a positive impact on the developmentof innovative and new classes of anticancerdrugs that have more tumour-specific effectsand fewer undesired side effects.

Thijn R. Brummelkamp and René Bernards are atthe Division of Molecular Carcinogenesis and

Center for Biomedical Genetics, The NetherlandsCancer Institute, Plesmanlaan 121, 1066 CX

Amsterdam, The Netherlands.Correspondence to R.B.

e-mail: [email protected]: 10.1038/nrc1191


cancer progression could lead to new markersfor early cancer detection and prevention.Recent advances in mitochondrial genomicand proteomic techniques have greatlyimproved our ability to identify these types ofmarkers. Here, we discuss the current status ofmitochondrial proteomics and associatedchallenges in cancer detection.

MtDNA mutations and cancerSome commonly used DNA-based markersof cancer include loss of heterozygosity(LOH), microsatellite instability (MSI), DNAhypermethylation, nuclear and mtDNAmutations, and detection of cancer-associatedviral DNA. Mitochondria contain their ownDNA, which is replicated and transcribedsemi-autonomously. MtDNA is particularlysusceptible to damage by environmental carcinogens because it contains no introns,has no protective histones or non-histoneproteins, and is exposed continuously toendogenous ROS. In fact, the frequency ofmtDNA mutations in cancer cells has beenreported to be tenfold higher than that ofnuclear DNA mutations6,7.

The mitochondrial D loop is a section ofthe mitochondrial genome that is thought tobe involved in replication and that containsshort poly-pyrimidine tracts. The D loopseems to be a hot spot for mutations, althoughmutations have been identified throughoutthe mitochondrial genome1. Many of thesemutations seem to be shared by most or all ofthe mitochondria that are contained in a cancer cell — that is, the mutations are athomoplasmic levels — providing a potentialadvantage for molecular detection. Many cancer-associated mutations have also beenidentified in the 12S and 16S rRNA genes8,9,10.

A number of mutations, deletions andinsertions in the mitochondrial genome havebeen associated with specific cancers.Microsatellite DNA alterations in the non-coding D loop and insertions and deletionshave been observed in the mtDNA genome,particularly in colorectal and gastric cancer.In addition, frameshift mutations have beenidentified in the D loop of pre-neoplastichead and neck cancer cells11. The D310 D-loop variants harbour large C-tract deletions that are likely to interfere with theinitiation of mtDNA replication12 and mito-chondria that undergo rapid replication inchronic hepatitis can acquire and accumulateDNA damage and mutation more readilythan those maintained under resting condi-tions. Base substitutions in the MTND1,MTND4, MTND5 and cytochrome b genes, ortheir noncoding region, can lead to breastcancer13. MtDNA mutations in genes that

61. Semizarov, D. et al. Specificity of short interferingRNA determined through gene expression signatures.Proc. Natl Acad. Sci. USA 100, 6347–6352 (2003).

62. Chi, J. T. et al. Genomewide view of gene silencing bysmall interfering RNAs. Proc. Natl Acad. Sci. USA 100,6343–6346 (2003).

63. Bridge, A. J., Pebernard, S., Ducraux, A., Nicoulaz, A. L.& Iggo, R. Induction of an interferon response by RNAivectors in mammalian cells. Nature Genet. 34, 263–264(2003).

64. Brummelkamp, T. R., Nijman, S. M. B., Dirac, A. M. G. &Bernards, R. Loss of the cylindromatosis tumoursuppressor inhibits apoptosis by activating NF-κB.Nature 424, 797–801 (2003).

65. Shoemaker, D. D., Lashkari, D. A., Morris, D., Mittmann, M.& Davis, R. W. Quantitative phenotypic analysis of yeastdeletion mutants using a highly parallel molecular bar-coding strategy. Nature Genet. 14, 450–456 (1996).

66. Winzeler, E. A. et al. Functional characterization of the S. cerevisiae genome by gene deletion and parallelanalysis. Science 285, 901–906 (1999).

67. Giaever, G. et al. Functional profiling of theSaccharomyces cerevisiae genome. Nature 418,387–391 (2002).

68. Hannon, G. J. et al. MaRX: an approach to genetics inmammalian cells. Science 283, 1129–1130 (1999).

69. Koh, E. Y., Chen, T. & Daley, G. Q. Novel retroviral vectorsto facilitate expression screens in mammalian cells. Nucl.Acids Res. 30, e142 (2002).

70. Hua, X., Liu, X., Ansari, D. O. & Lodish, H. F. Synergisticcooperation of TFE3 and smad proteins in TGF-β-induced transcription of the plasminogen activatorinhibitor-1 gene. Genes Dev. 12, 3084–3095 (1998).

71. Hudson, J. D. et al. A proinflammatory cytokine inhibitsp53 tumor suppressor activity. J. Exp. Med. 190,1375–1382 (1999).

72. Maestro, R. et al. Twist is a potential oncogene thatinhibits apoptosis. Genes Dev. 13, 2207–2217 (1999).

73. Yamaoka, S. et al. Complementation cloning of NEMO, acomponent of the IκB kinase complex essential for NF-κB activation. Cell 93, 1231–1240 (1998).

AcknowledgementsWe thank R. Beijersbergen for helpful discussions and criticalreading of this manuscript and apologize to our colleagues foromission of relevant work due to space constraints. The work ofthe authors was supported by grants from the Dutch CancerSociety, The Netherlands Organization for Scientific Research andthe Center for Biomedical Genetics.

Online links

DATABASESThe following terms in this article are linked online to:LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink/ARF | β-catenin | BRCA1 | CBP | CYLD | env | gag | Hedgehog |ID4 | ING1 | KRAS | MDM2 | MYC | NF-κB | p300 | p53 | pol |PPAN | PTEN | RAF1 | RB


Proteomic analysis of cancer-cellmitochondria

Mukesh Verma, Jacob Kagan, David Sidransky and Sudhir Srivastava

O P I N I O N

Mitochondrial dysfunction and mutations inmitochondrial DNA have been frequentlyreported in cancer cells. Mitochondrial gene-expression signatures of transformed cellshave been identified; however, thephenotypic effects of these geneticalterations remain to be established.Identification of mitochondrial proteins thatare aberrantly expressed in cancer cells hasbeen made possible by the recentdevelopment of mitochondrial functionalproteomics and could identify new markersfor early detection and risk assessment, aswell as targets for therapeutic intervention.

Each human cell contains several hundredcopies of mitochondrial DNA (mtDNA),which encodes 13 Kreb’s cycle and respira-tory-chain subunits, 22 transfer RNAs and 2ribosomal RNAs (rRNAs) — 12S and 16S1.The main function of mitochondria is toproduce cellular energy and, during thisprocess, acetyl-coenyme A — the breakdownproduct of fats and sugars — is passedthrough the Kreb’s cycle to generate elec-trons. These, in turn, are transferred to theproteins of the respiratory chain in the innermitochondrial membrane. This generates a

proton gradient between the mitochondrialmatrix and the intermembrane space, whichis used to generate ATP. In addition to servingas the main intracellular sourceof energy ofthe cell, mitochondria regulate several cellu-lar processes that are linked to apoptosis,which include electron transport and energymetabolism. They are also the storage site fora number of soluble proteins that mediateapoptosis, including cytochrome c, certainprocaspases and apoptosis-inducing factor.Mitochondria can activate apoptosis byreleasing these factors from their intermem-brane space into the cytoplasm, and also byaltering the cellular redox potential2–4. Duringthis process, reactive oxygen species (ROS)are generated that can serve as crucial pro-apoptotic factors. In fact, recent studies haveindicated that mitochondria can undergoself-apoptosis — termed ‘mitoptosis’5.Therefore, mitochondria not only serve as apowerhouse of energy, but also as a supplierof apoptosis-inducing signals (FIG. 1).

We have recently gained much insight intothe connection between mitochondrial dys-function, deregulation of apoptosis andtumorigenesis. Further information aboutmitochondrial proteins that are involved in

innovation: new tools for functional mammalian cancer genetics

Documents