news & views

128 nature genetics • volume 30 • february 2002

Interestingly, the loci that control the lengthof the first and second spine in the dorsal finmap to distinct linkage groups. This sug-gests that seemingly serially homologoustraits (two spines of one fin), whose expres-sion seems to be correlated within one mor-phological structure, can be controlled to asignificant extent by two different loci ontwo different chromosomes. However, thelength of the second dorsal spine and that ofthe pelvic spine maps to adjacent genomicregions, suggesting that these functionallylinked traits are also under the control (pos-sibly pleiotropic) of the same or at least twoclosely linked loci.

The stickleback linkage map providestantalizing results and is an importantbeginning to build on for future work, butdoes it describe the general genetic nature ofevolutionary change? Not enough compara-tive data are yet available, and as these dataare not easy to come by, no generalities haveyet emerged. We do have an enticinginsight—that both small effects of manygenes and large effects of a small number ofgenes can account for variation in certaincharacteristics that differentiate speciesfrom one another. As Peichel et al.4 suggest,some of the findings of the stickleback studymight be partly explained by the experimen-tal design. By using only one family, one isgoing to overestimate the effect of individual

QTL and underestimate the number of locithat have a significant phenotypic effect.Evolutionary biology is a comparative andhistorical science. Only through additionalstudies can generalities, if they indeed exist,be discovered. Where shall we look? Stickle-backs might be a good model.

Parallel speciationOne of the most remarkable aspects ofsticklebacks is that the distinctive morphslive sympatrically (in the same locality) in atleast six lakes in British Columbia. The dif-ferences between the limnetic and benthicsticklebacks apparently evolved extremelyrapidly, independently and repeatedly—inparallel—from marine ancestors that colo-nized coastal lakes that formed during thelast ice age3. Parallel speciation shows thatevolution repeats itself at the morphologi-cal level, but molecular phylogenetic infor-mation is necessary to be able to recognizesuch cases in the first place. Evolutionarilysimilar morphological solutions to ecologi-cal problems, as well as the occupation ofequivalent ecological niches in differentenvironments, is receiving attention fromevolutionary biologists interested not onlyin sticklebacks but also in other evolution-ary models such as Anolis lizards and cich-lid fishes, where some of the most notableinstances of parallel evolution in several

African lakes have been documented6–9.The exact mechanisms of independent par-allel speciation (whether allopatric or sym-patric) are debated both in general and insticklebacks in particular3,10–12, but each ofthese British Columbian lakes offers anindependent experiment in parallel innova-tion in limnetic and benthic morphologies.If there are genomic rules for diversificationat the phenotypic level to be discovered, it isin these remarkable cases of parallel specia-tion where one might want to look. Are thesame loci used in parallel to bring about thesame adaptations, or does evolution recruitcompletely different genes and/or molecu-lar mechanisms to produce similar pheno-typic responses to equivalent ecologicalchallenges? �1. Bakker, T.C.M., Künzler, R. & Mazzi, D. Nature 401,

234 (1999).2. Reusch, T.B.H., Häberli, M.A., Aeschlimann, P.B. &

Milinski, M. Nature 414, 300–302 (2001).3. Rundle, H.D., Nagel, L., Boughman, J.W. & Schluter,

D. Science 287, 306–308 (2000).4. Peichel, C.L. et al. Nature 414, 901–905 (2001).5. Meyer, A. Biol. J. Linn. Soc. 39, 279–299 (1990).6. Stiassny, M.L.J. & Meyer A. Sci. Am. (February)

64–69 (1999).7. Wilson, A.B. Noack-Kuhnmann, K. & Meyer, A. Proc.

Roy. Soc. Lond. B 267, 2133–2141 (2000).8. Kirkpatrick, M. Nature 408, 299–300 (2000).9. Schluter, D. The Ecology of Adaptive Radiation

(Oxford Univ. Press, Oxford, 2000).10. Johannesson, K. Trends Ecol. Evol. 16, 148–153

(2001).11. Schluter, D., Boughman, J.W. & Rundle, H.D. Trends

Ecol. Evol. 16, 283–284 (2001).12. Johannesson, K. Trends Ecol. Evol. 16, 284 (2001).

Juggling JunJonathan B. Weitzman

Unité des Virus Oncogènes, Institut Pasteur, 25 rue du Dr Roux, 75724 Paris, France. e-mail: jonnyw@pasteur.fr

The Jun and JunB proteins have been proposed to play distinct and antagonizing roles in controlling gene expression and cell pro-liferation. The surprising observation that JunB can functionally replace Jun during mouse development challenges the way wethink about the interplay between members of the Jun family of transcription factors.

Jugglers are full of surprises, keeping allsorts of incongruous objects aloft. On page158 of this issue, Emmanuelle Passegué andcolleagues1 show that juggling genes canlead to surprises as well. Several studies haveindicated that the three mammalian Junproteins have antagonistic effects on genetranscription, cell proliferation, apoptosisand transformation2, allowing distinct bio-logical roles to be assigned to each proteinwith confidence. Passegue et al.1 now showthat Jun family members can functionallysubstitute for each other during mousedevelopment and cellular proliferation.

This result calls for a review of currentassumptions about the behavior of thismultigene family.

Different genes, different functionsWhat made us think that Jun proteins havedistinct functions in the first place? The Junproteins are basic leucine zipper proteinsthat can dimerize with one another, or withmembers of the related Fos and ATF fami-lies, to form the transcription factor com-plex AP-1. In Drosophila, there is one Junprotein and one Fos-like protein—they arecritical to fly development3. As mammals

have three Jun members (Jun, JunB andJunD), it seemed reasonable to assume thatif a duplicated Jun gene has been maintainedsince the divergence of mammals, each ofthe three Jun genes may have a specific role.

Over a decade ago, researchers beganusing transfection experiments to demon-strate that JunB and Jun have different activ-ities4. JunB is a much poorer transactivatorthan Jun. These differences could beaccounted for by small changes in the DNA-binding and dimerization domains5. Jun isan efficient activator of promoters contain-ing a single AP-1–binding site, whereas

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m

news & views

nature genetics • volume 30 • february 2002 129

JunB requires several sites. Jun is also asubstrate for the Jun N-terminal kinases(JNKs), and phosphorylation enhancesits transactivation potential. Further-more, JunB can inhibit Jun-regulatedgene expression by the formation ofinefficient heterodimers.

These differences in transactivationcapacity correlate with seemingly antago-nistic roles in regulating cell proliferationand transformation. Jun is a positive regu-lator of proliferation and drives theexpression of cell cycle regulators such ascyclin D1 (ref. 6). It has recently beenshown that Jun also decreases expressionof the p53 tumor suppressor protein7 andits target gene Cdkn1a (encoding p21).Jun is necessary for transformation andcan cooperate with the Ras oncogene8. Incontrast, JunB (as well as JunD) can nega-tively regulate cell proliferation. Overex-pression of JunB inhibits growth ofcultured fibroblasts and antagonizes Jun-induced transformation. These resultsmay be explained by the recent observa-tions6,9 that JunB reduces Jun-mediatedexpression of the cyclin D1 gene and posi-tively regulates the expression of thecyclin-dependent kinase inhibitorp16INK4a. Thus, many studies in which thedifferent Jun proteins are overexpressedsupport the hypothesis that JunB and Junhave opposing effects on gene expressionand cell proliferation.

Then came the knockouts. Inactivationof mouse Jun or Junb results in quite dif-ferent phenotypes, providing additionalevidence of non-overlapping functions10.Mice lacking Jun die during development(between embryonic day 12.5 and 13.5)and have defects in liver development andheart morphogenesis10,11. Embryos lack-ing JunB die around day 9.5, owing to vas-cular defects in the placenta andextraembryonic tissue12. The extensivecharacterization of cells derived fromthese knockout animals has bolstered thenotion of opposing roles in cell growth,transformation, apoptosis and epidermaldifferentiation7,9,13,14.

No need for Jun?Passegué et al.1 decided to put this modelto the test using an elegant geneticapproach. They replaced Jun with Junbto generate a ‘knock-in’ (ki) allele. Theresultant mice should therefore carry anextra copy of Junb that is expressedwherever Jun would normally be found.It is well known that the levels of Jun areregulated by Jun itself and therefore it is

unlikely that the levels are identical. Sur-prisingly, the Junb knock-in rescuedboth the liver defects and embryoniclethality associated with the absence ofJun. Further analysis revealed that the kiallele restored the normal expression ofseveral AP-1–regulated genes in fibrob-lasts in culture—for example, inductionof cyclin D1 and suppression of p53. Italso rescued the proliferative and apop-totic defects seen in fibroblasts andhepatoblasts lacking Jun.

So, does JunB behave differently fromJun or not, and how can we explain theapparent paradox? Passegué et al.1 con-clude that the regulated expression of Jungenes is more important than differencesin their sequences. The results supportthe notion that JunB is a weaker transac-tivator than Jun, as two copies of the kiallele are required to rescue the embry-onic lethality, whereas mice with onenormal copy of Jun are viable. Many ofthe previous observations are consistentwith a model in which JunB exerts a neg-ative effect by competing with Jun toform less efficient transactivating dimers.Thus, in the presence of Jun, JunBexpression would seem to be inhibitory,whereas in the absence of Jun it can func-tionally replace and activate AP-1 targetgenes. Perhaps it is not just the spatialand temporal regulation of the Jun genesthat is important, but also the relativebalance between Jun and Fos/ATF com-ponents. The existence of three Jun genesprovides mammalian cells with a largerange of potential heterodimeric part-ners and thus the biochemical fodder for

a broad spectrum of activity—and subtledifferences in activity may be sufficientfor suitable response to changes in cellu-lar conditions.

Not entirely redundantIt should be noted that although the kiallele provides an impressive rescue of theliver defects in these embryos, the ki/kimice have malformed cardiac outflowtracts, which is probably responsible fortheir postnatal lethality1. A ubiquitouslyexpressed JunB transgene rescues the car-diac defects, but the mice still die postna-tally1. Thus, it is also possible that even ifJun and JunB are functionally inter-changeable during development, they mayhave distinct functions after birth. TheAP-1 factor has been implicated in theresponse to stress and it may be that dif-ferences in regulation and signal trans-duction are more critical to the health ofadult animals than developing embryos.For example, mice lacking JunD developnormally, but are susceptible to stress-induced conditions15.

Future efforts will lead to a betterunderstanding of when and where the dif-ferent Jun genes are expressed, and howthe overlap of their expression with that ofFos gene expression dictates the AP-1equilibrium. This knowledge, togetherwith insights gained from mice with tis-sue-specific modification of Jun expres-sion, should allow a better understandingof when the Jun proteins do the samething and when they behave differently.When it comes to multigene families weshould interpret gene inactivation andoverexpression experiments with caution.But whether it’s by knocking in or out,these genetic juggling studies will con-tinue to enlighten us about Jun functionin the years to come. �1. Passegué, E., Jochum, W., Behrens, A., Ricci, R. &

Wagner, E.F. Nature Genet. 30, 158–166 (2002).2. Mechta-Grigoriou, F., Gerald D. & Yaniv, M.

Oncogene 20, 2378–2389 (2001).3. Kockel, L., Homsy, J.G. & Bohmann, D. Oncogene

20, 2347–2364 (2001).4. Chiu, R., Angel, P. & Karin, M. Cell 59, 979–986

(1989).5. Deng, T. & Karin, M. Genes Dev. 7, 479–490 (1993).6. Bakiri, L., Lallemand, D., Bossy-Wetzel, E. & Yaniv,

M. EMBO J. 19, 2056–2068 (2000).7. Schreiber, M. et al. Genes Dev. 13, 607–619 (1999).8. Pfarr, C.M. et al. Cell 76, 747–760 (1994).9. Passegué, E. & Wagner, E.F. EMBO J. 19, 2969–2979

(2000).10. Jochum, W., Passegué, E. & Wagner, E.F. Oncogene

20, 2401–2412 (2001).11. Eferl, R. et al. J. Cell Biol. 45, 1049–1061 (1999).12. Schorpp-Kistner, M., Wang, Z-Q., Angel, P. &

Wagner, E.F. EMBO J. 18, 934–948 (1999).13. Szabowski, A. et al. Cell 103, 745–755 (2000).14. Shaulian, E. et al. Cell 103, 897–907 (2000).15. Weitzman, J.B., Fiette, L., Matsuo, K. & Yaniv M.

Mol. Cell 6, 1109–1119 (2000).

Tom

Kiln

er

©20

02 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://g

enet

ics.

nat

ure

.co

m