© 2002 Macmillan Magazines LtdNATURE REVIEWS | GENETICS VOLUME 3 | FEBRUARY 2002 | 87

Lung cancer causes most cancer-related deathsworldwide, and of the different lung tumour types,adenocarcinoma is the most common. Theinvolvement of K-ras in lung cancer was confirmedlast year (see June 2001 Highlights), when TylerJacks’ group showed that the somatic activation ofa constitutively active K-ras allele (K-rasG12D) canalone cause cancer. This study is now followed upby two new K-rasG12D mouse models — one madeby Jacks’ group and the other by Harold Varmus’group — in which conditional gene-activationsystems have been used to switch on this mutant K-ras allele. Importantly, these studies shed muchneeded light on the events required for lungtumour initiation, maintenance and regression,and are a step towards much better mouse modelsof cancer that can be used to develop and test newcancer therapies.

The gene-expression switches used by each teamallowed them to ask different questions about lungtumour biology. Jacks’ team used the Cre/loxPsystem to activate K-rasG12D by targeting theendogenous K-ras locus with a ‘lox–stop–lox’ (LSL)K-ras allele in which loxP sites flank atranscriptional Stop element.When Cre wasintroduced into the lungs of LSL-K-rasG12D micethrough a nasally delivered adenovirus, their lungsbecame covered in precancerous lesions within fourweeks, and there was evidence that Cre-inducedactivation of K-rasG12D was responsible for thishighly penetrant and rapid lung tumorigenesis.

However, such severe tumorigenesis is aproblem — in last year’s study, for example, themice developed so many tumours that many diedbefore the earlier lesions could progress tomalignancy. The Jacks’ group tackled this problemby lowering adenoviral-Cre doses to reduce tumournumbers, allowing the mice to survive and progressto later stages of tumorigenesis. Only then couldthe team solve the long-standing question of whichof several early precancerous lesions give rise toadenocarcinomas — they report that it’s mostprobably a lesion called atypical adenomatoushyperplasia, which seems to originate from oneparticular cell type, the alveolar type II cell. The

authors also identified a new cell type, possibly anew lung stem cell, that might also contribute toadenocarcinoma development.

Fisher et al. used a different trick to turn on theK-rasG12D allele — they created bi-transgenic micethat express both a tetracycline (Tet)-activatableform of K-rasG12D and a reverse Tet transactivatorprotein expressed in alveolar type II cells that canonly activate K-rasG12D in the presence ofdoxycycline. This elegant approach allowed them to look at the events required for the initiation,maintenance and regression of lung tumours.Within one week of receiving doxycycline in theirdrinking water, these mice developed hyperplasticalveolar type II cells; after two months, their lungsbecame laden with large adenomas andadenocarcinomas. Nevertheless, within days ofdoxycycline withdrawal, these tumours regressedand underwent apoptosis.When these mice werecrossed to two mouse strains, each null for a well-known tumour suppressor gene — Trp53 or p16ink4a

— they developed more-aggressive tumours muchmore rapidly. Surprisingly, however, these tumoursstill underwent rapid apoptosis-mediatedregression following doxycycline withdrawal,showing that this regression occurs via a p53-independent apoptotic pathway.

The fact that lung tumours with p16ink4a andTrp53 mutations can be induced to regress is goodnews indeed for those developing anticancertherapeutics, because TP53 mutations areassociated with tumour resistance tochemotherapy. But what is the pathway thatmediates the p53-independent regression of thesetumours? Fisher et al. have already found someclues to this question in their data, and toinvestigate it further, they plan to use microarrayexpression analysis to identify key transcriptionalchanges that occur during tumour induction andregression in these mice.

Jane Alfred

References and linksORIGINAL RESEARCH PAPERS Jackson, E. L. et al. Analysis of lungtumour initiation and progression using conditional expression ofoncogenic k-ras. Genes Dev. 15, 3243–3248 (2002) | Fisher, G. H. et al.Induction and apoptotic regression of lung adenocarinomas by regulationof a K-Ras transgene in the presence and absence of tumour suppressorgenes. Genes Dev. 15, 3249–3262 (2002) WEB SITESTyler Jacks’ lab: http://mit.edu/biology/www/facultyareas/facresearch/jacks.shtmlHarold Varmus’ lab: http://www.ski.edu/lab_homepage.cfm?lab=203

Intricate modelling

C A N C E R G E N E T I C S

The total number of genes in yeastis unlikely to change as a result of thisstudy the number of previouslypredicted genes that turn out to bespurious is likely to be offset by thenumber of new predictions. The trueimpact of this study lies in the factthat it describes a method for re-examination of genome annotationthat is applicable to other genomesand in its ability to predict theexistence of genes that have so fareluded previously known methods.

Magdalena Skipper

References and linksORIGINAL RESEARCH PAPER Kumar, A. et al.An integrated approach for finding overlookedgenes in yeast. Nature Biotechnol. 20, 58–63(2002) FURTHER READING Oliver, S. To-day, we havenaming of parts… Nature Biotechnol. 20, 27–28(2002) WEB SITEMichael Snyder’s lab:http://www.yale.edu/snyder/res.html

together the network of interact-ing proteins that controls thisresponse, so uncovering many newinteractions of probable biologicalsignificance.

Although this approach is clearlyvery powerful, it is not without limi-tations — for example, Gavin et al.could not purify proteins under 15 kDa in size. Both groups alsoreport a significant number of false-positive interactions, while failing todetect some known interactions,perhaps because the tag can inter-fere with a protein’s function or withits physical associations. Althoughthere is still a long way to go beforewe fully understand how a pro-teome’s functional networks respondto the ever changing life of a cell,these two studies provide apanoramic view of protein functionand a wealth of new functional datafor genome annotation.

Jane AlfredReferences and links

ORIGINAL RESEARCH PAPERS Gavin, A.-C.et al. Functional organization of the yeastproteome by systematic analysis of proteincomplexes. Nature 415, 141–147 (2002) | Ho, Y.et al. Systematic analysis of protein complexes inSaccharomyces cerevisiae by massspectrometry. Nature 415, 180–183 (2002)FURTHER READING Kumar, A. & Snyder, M.Protein complexes take the bait. Nature 415,123–124 (2002)WEB SITES These data sets can be found at:http://yeast.cellzome.comhttp://www.mdsp.com/yeast