724 | OCTOBER 2002 | VOLUME 2 www.nature.com/reviews/cancer

H I G H L I G H T S

Medulloblastomas are childhoodtumours of cerebellar origin thatare, in some cases, associated withinappropriate activation of theHedgehog (Hh) signalling pathway.Philip Beachy and colleagues nowreport that inhibitors of this path-way induce regression of establishedtumours in preclinical models of medulloblastoma, revealing a new therapeutic approach for thisaggressive cancer.

Hh encompases a family ofsecreted proteins that are essentialfor developmental patterning events,as well as for the maintenance ofsomatic stem cells and specificationof organ size. Hh binds to the trans-membrane receptor Patched (Ptch),

allowing activation of another transmembrane protein namedSmoothened (Smo). This leads tothe activation of target genes by theCubitus interruptus/Gli (CI/GLI)family of transcription factors.

Ligand-independent activationof this pathway has been shown tooccur in medulloblastoma, causedeither by mutations that renderSmo insensitive to regulation byPtch, or by mutational inactivationof Ptch. The absence of Ptch activ-ity can promote tumour initiation,and subsequent loss of p53 func-tion seems to promote tumour-cell growth in a mouse model ofmedulloblastoma.

Beachy and colleagues thereforedecided to test the ability ofcyclopamine, a plant-derived productthat inhibits Hh signalling at the levelof Smo, to slow medulloblastomagrowth. They found that treatment ofmeduloblastoma-derived cell lineswith cyclopamine inhibited prolifera-

tion and induced differentiation inculture. Furthermore, subcutaneousinjection of cyclopamine into amouse model of medulloblastoma —created by disruption of one Ptchallele and mutation of Trp53 (whichencodes p53 in mice) — reduced thegrowth of established tumours.

Cells from freshly resectedhuman medulloblastomas, whichwere found to have elevated Hh sig-nalling activity, were also suscepti-ble to a derivative of cyclopamine.Treatment with this drug reducedthe viability of medulloblastomacells by almost 50%, and down-regulated the Hh-signalling path-way. Cyclopamine, however, had noeffect on glioblastoma, ependy-moma or fibroblast cells, indicatingthat it is not a general cytotoxin.Nor was the drug observed to havetoxic effects on rodents or othermammals.

So why is Hh-pathway activationa characteristic of medulloblastomas,

Double-stranded RNA molecules that arespecifically designed to block geneexpression — termed short interferenceRNA (siRNAs) — promise to be a usefultool in cancer research. Less than one yearafter the first demonstration of how siRNAscould silence gene expression inmammalian cell culture, Agami andcolleagues, reporting in Cancer Cell, showthat oncogenic KRAS can be specificallyand stably inactivated through the use of aviral RNA interference vector.

The guanine-nucleotide-binding proteinsthat are encoded by RAS are integral tocellular signal-transduction pathways.They regulate proliferation, differentiationand cell survival, and are frequentlymutated in human cancers, particularly inpancreatic (85%) and colon carcinomas(40%). Mutant RAS oncogenes oftencontain point mutations, such as glycine tovaline at codon 12 (KRASV12), and theselead to constitutively active RAS proteins.Wild-type KRAS seems to be required forviability, and antisense strategies to KRAS— which cannot distinguish between the

wild-type and oncogenic forms of the gene— have not been successful.

Brummelkamp et al. used a retroviral vector(pRETROSUPER) carrying the expressioncassette of the plasmid pSUPER — developedrecently by the same group — that directs thesynthesis of siRNAs in mammalian cells.When a human pancreatic cell line (CAPAN-1) was infected with the pRETROSUPERvector that contained sequences that spannedthe mutation region of KRASV12, expression ofKRASV12 was inhibited. This suppression wasspecific — use of a TP53 sequence in thevector did not cause a decrease in KRAS.Infecting the EJ bladder cell line — which hasmutated HRAS but wild-type KRAS — withthe mutant KRAS vector also had no effect onthe expression of KRAS.

In addition, the authors showed that thespecific suppression of KRASV12 leads to lossof tumorigenicity. When KRASV12 expressionwas downregulated in CAPAN-1 cells, thecells could not form tumours in nude mice,whereas mice injected with CAPAN-1 cellsthat had been infected with a control vectorall developed tumours.

So, how can we make use of viral RNAinterference vectors in cancer research? Theyare a powerful tool for investigating thegenetic events that are required for atumorigenic phenotype and for furtherelucidating the events that are needed tomaintain that phenotype. In addition,siRNA-based gene therapy — shown to be apossibility by Brummelkamp et al. in bothcell culture and animal models — should beinvestigated further. Issues that will need tobe addressed include overcoming the poorpenetration of tumours, possibleneutralization of siRNA by the immunesystem and the possibility of escape fromsiRNA-mediated attack.

Ezzie Hutchinson

References and linksORIGINAL RESEARCH PAPER Brummelkamp, T. R. et al.Stable suppression of tumorigenicity by virus-mediated RNAinterference. Cancer Cell 2, 243–247 (2002)FURTHER READING Borkhardt, A. Blocking oncogenes inmalignant cells by RNA interference — new hope for a highlyspecific cancer treatment. Cancer Cell 2, 167–168 (2002) |Tuschl, T. Expanding small RNA interference. NatureBiotechnol. 20, 446–448 (2002)WEB SITEReuven Agami’s lab: http://www.nki.nl/nkidep/agami/

Welcome interference

G E N E T I C T E C H N I Q U E S

Fuzzy logic

S I G N A L L I N G

NATURE REVIEWS | CANCER VOLUME 2 | OCTOBER 2002 | 725

Driving-instructors’ cars are equipped withdual controls, but this might not always be asensible safety feature. Cell-cycle exit andapoptosis are two crucial processes by whichcells limit proliferation, so genes that controlboth of these would be a prime target formutation in tumorigenesis. Now, IswarHariharan and colleagues have identified aDrosophila gene, salvador (sav), that regulatesboth of these processes; the human orthologueis also mutated in cancer cell lines.

The development of the Drosophila eye istightly regulated — cell proliferation occursthroughout the larval stage, differentiationoccurs during the late larval and pupal stages,and excess cells are eliminated by apoptosis.These characteristics make it an ideal system toscreen for mutations that alter proliferation orapoptosis. The authors have identifiedmutations in at least 23 loci that, whenhomozygous, cause an over-representation ofmutant cells compared with wild-type cells,making them good candidates for tumour-suppressor genes. One of these, sav, wascharacterized further.

An increase in cell number could be causedby an increase in proliferation or a decrease inapoptosis, so both of these processes wereexamined in turn. In sav mutants, ectopic BrdUincorporation was observed posterior to themorphogenetic furrow — which moves fromthe posterior to anterior of the eye, causingcells to arrest, after which they synchronouslyenter S phase. This indicates that these cellscontinue to proliferate after wild-type cellshave arrested. Flow-cytometry analysis

confirmed that sav mutants are delayed inexiting the cell cycle.

However, this delay in cell-cycle exit is notsufficient to account for the increase in cell number,so might apoptosis also be inhibited in sav-mutantcells? TUNEL analysis revealed that cell deathseemed to be mostly confined to the wild-typeregions of the eye. Hid and Rpr, which target thecaspase inhibitor DIAP1, were unable to induceapoptosis in sav-mutant cells. DIAP1 protein levelsremained high and the effector caspase Drice wasnot cleaved to generate the active form.

The sav gene was sequenced and contained twoputative WW domains, which are involved inprotein–protein interactions. The warts (wts)gene was also identified in the mutant screen andwas shown to have a similar phenotype to sav inregulating cell-cycle exit and apoptosis. It containsfive PPXY motifs, to which WW domains bind,and a precipitation experiment with GST-taggedSav confirmed that Sav and Wts interact.

So, two genes have been identified that, whenmutated, confer a selective advantage to cells.Might they be mutated in cancer cells? The humanorthologue of sav, WW45, was sequenced in 52tumour-derived cell lines, and WW45 was alteredin three of these. Two renal-cancer cell lines —ACHN and 786-O — had deletions in WW45 thatwould inactivate the protein.

The authors have therefore identified a newpotential tumour suppressor, and have proved, yetagain, that Drosophila can be a useful modelorganism for cancer research.

Emma Greenwood

References and linksORIGINAL RESEARCH PAPER Tapon, N. et al. salvador promotesboth cell cycle exit and apoptosis in Drosophila and is mutated inhuman cancer cell lines. Cell 110, 467–478 (2002) FURTHER READING St. Johnston, D. The art and design of geneticscreens: Drosophila melanogaster. Nature Rev. Genet. 3, 176–188(2002) WEB SITEIswar Hariharan’s lab:http://www.mgh.harvard.edu/depts/CancerCenter/hariharan.html

Dual control

C E L L P R O L I F E R AT I O N

in particular? The authors suggestthat as Hh is a known stem-cell factorand a mitogen in the developing cere-bellum, this pathway might allowcancerous stem cells to continuouslyundergo self-renewal. The highexpression levels of the neuronalstem-cell marker nestin in mouse andhuman tumours supports this theory.If they can be used in humans, Hhantagonists might be developed as auseful therapy for this cancer.

Kristine Novak

References and linksORIGINAL RESEARCH PAPER Berman, D. M.Medulloblastoma growth inhibition by hedgehogpathway blockade. Science 297, 1559–1561(2002) FURTHER READING Ruiz i Altaba, A. et al. Gli and hedgehog in cancer:tumours, embryos and stem cells. Nature Rev.Cancer 2, 361–372 (2002)WEB SITESNature’s Alliance for Cell Signalling (AfCS)update (Hedgehog signalling models: takingthe rough with the Smoothened):http://www.signaling-update.org/featured/200208.htmlPhilip Beachy’s lab:http://www.bs.jhmi.edu/mbg/dept/beachy.html