DispatchR815

orientation at a different spatiallocation, they found that thetraining-induced changes for thesecond orientation transferred tothe first location. Such findings ofbroad location transfer undermine theargument that this learning is due toplasticity in retinotopic visual areas.

The findings of Adab and Vogels [6]that perceptual learning can resultin robust plasticity in V4 providean important counterpoint to theabove-mentioned studies. Thesefindings are significant in that theydemonstrate that perceptual learningcan involve robust plasticity in visualrepresentation areas. Furthermore,that these learning effects manifesteven outside the context of the trainedtasks is inconsistent with the effectsresulting from top-down attentionalmodulation. These results help bringsome balance to the field ofperceptual learning and demonstratethat, while low-level plasticity may notbe ubiquitous to training onperceptual tasks, it can and does occurin certain settings. This brings thedebate in the field back from thequestion of whether perceptuallearning involves low-level plasticity towhen it occurs. Further research will berequired to clarify differences betweenthe studies that have found low-levelplasticity and those that have not. Bynow it is clear that learning a task canresult in a distribution of plasticity that

can include a diverse set of brainregions [20]. What rules determinehow plasticity in a given task isdistributed across brain areas, whysome training procedures yielddifferent distributions of plasticity toothers, and the rules that determinewhether learning occurs in any givenbrain region are important topics offuture research.

References1. Gilbert, C.D., Sigman, M., and Crist, R.E. (2001).

The neural basis of perceptual learning. Neuron31, 681–697.

2. Fahle, M. (2004). Perceptual learning: a case forearly selection. J. Vis. 4, 879–890.

3. Xiao, L.Q., Zhang, J.Y., Wang, R., Klein, S.A.,Levi, D.M., and Yu, C. (2008). Complete transferof perceptual learning across retinal locationsenabled by double training. Curr. Biol. 18,1922–1926.

4. Dosher, B.A., and Lu, Z.L. (1998). Perceptuallearning reflects external noise filtering andinternal noise reduction through channelreweighting. Proc. Natl. Acad. Sci. USA 95,13988–13993.

5. Law,C.T., andGold, J.I. (2008).Neural correlatesof perceptual learning in a sensory-motor, butnot a sensory, cortical area. Nat. Neurosci. 11,505–513.

6. Adab, H.V., and Vogels, R. (2011). Practisingcoarse orientation discrimination improvesorientation signals in macaque cortical area V4.Curr. Biol. 21, 1661–1666.

7. Raiguel, S., Vogels, R., Mysore, S.G., andOrban, G.A. (2006). Learning to see thedifference specifically alters the mostinformative V4 neurons. J. Neurosci. 26,6589–6602.

8. Schoups, A., Vogels, R., Qian, N., and Orban, G.(2001). Practising orientation identificationimproves orientation coding in V1 neurons.Nature 412, 549–553.

9. Fiorentini, A., and Berardi, N. (1980). Perceptuallearning specific for orientation and spatialfrequency. Nature 287, 43–44.

10. Poggio, T., Fahle, M., and Edelman, S. (1992).Fast perceptual learning in visual hyperacuity.Science 256, 1018–1021.

11. Crist, R.E., Li, W., and Gilbert, C.D. (2001).Learning to see: experience and attention inprimary visual cortex. Nat. Neurosci. 4, 519–525.

12. Yang, T., and Maunsell, J.H. (2004). The effectof perceptual learning on neuronal responses inmonkey visual area V4. J. Neurosci. 24,1617–1626.

13. Franko, E., Seitz, A.R., and Vogels, R. (2009).Dissociable neural effects of long-termstimulus-reward pairing in macaque visualcortex. J. Cogn. Neurosci. 22, 1425–1439.

14. Zohary, E., Celebrini, S., Britten, K.H., andNewsome, W.T. (1994). Neuronal plasticitythat underlies improvement in perceptualperformance. Science 263, 1289–1292.

15. Sotiropoulos, G., Seitz, A.R., and Series, P.(2011). Perceptual learning in visualhyperacuity: A reweighting model. Vision Res.51, 585–599.

16. Li, W., Piech, V., and Gilbert, C.D. (2008).Learning to link visual contours. Neuron 57,442–451.

17. Smirnakis, S.M., Brewer, A.A., Schmid, M.C.,Tolias, A.S., Schuz, A., Augath, M., Inhoffen, W.,Wandell, B.A., and Logothetis, N.K. (2005).Lack of long-term cortical reorganizationafter macaque retinal lesions. Nature 435,300–307.

18. Ghose, G.M., Yang, T., and Maunsell, J.H.(2002). Physiological correlates of perceptuallearning in monkey V1 and V2. J. Neurophysiol.87, 1867–1888.

19. Shadlen, M.N., and Newsome, W.T. (2001).Neural basis of a perceptual decision in theparietal cortex (area LIP) of the rhesus monkey.J. Neurophysiol. 86, 1916–1936.

20. Ahissar, M., and Hochstein, S. (2004). Thereverse hierarchy theory of visual perceptuallearning. Trends Cogn. Sci. 8, 457–464.

Department of Psychology, Universityof California, Riverside, 900 UniversityAvenue, Riverside, CA 92521, USA.E-mail: aseitz@ucr.edu

DOI: 10.1016/j.cub.2011.08.042

Morphogen Gradients: Expandand Repress

An expansion–repression mechanism by which morphogen gradients canadjust to size and growth had been postulated as a model. Now, its molecularnature has been uncovered.

Simon Restrepo and Konrad Basler*

The classic 1941 Walt Disney movie‘Dumbo’ tells the story of a youngcircus elephant withdisproportionately large ears. Initiallymocked, Dumbo eventually useshis large ears as wings and becomesa star. However, aside from itsallegorical power, his ordeal mighthave been more dramatic werehe not the hero of an imaginary tale,but one of the heroes of modern

genetics, a Drosophila melanogasterfruitfly. For flies, wing proportionsare best left untouched. In the realworld, the proportions of animalbody plans have been honed for eonsby natural selection and largevariations are rarely observed to bebeneficial. Nature has evolved robustdevelopmental processes thatmaintain well-proportioned bodyplans in the face of environmentalchallenges such as changes innutrition or temperature. For example,

the adult size of Drosophila fliescan be influenced by nutrition up tothe point that starved larvae willmetamorphose into pint-sizedimagos up to fifty percent smallerthan their better fed counterparts [1].Fortunately for such flies, however,their wings scale with their newbody size and Dumbo-like flies arenever observed. Thus, Drosophilaehave developmental mechanismsthat maintain the scale of wing surfaceto body weight best suited to theirphysiology and metabolism. In arecent issue of Current Biology, BenZvi and colleagues [2] describea molecular mechanism that ensuresthat the Drosophila wing disc,from which the adult wings form,scales with tissue size. This scalingmechanism acts through modulationof the activity gradient formed by

C0

C t1

C t2L 2*L

A

DPP PENT

Expansion

Repression

Pent

Dpp

t0 t+1 t+n

L

Pent Pent

B

Current Biology

Figure 1. Scaling morphogen gradients.

(A) The French Flag and scaling. Morphogenspattern organs in a concentration-dependentmanner (upper panels). Low sensitivity tar-gets are expressed close to the source of themorphogen. More sensitive targets have largerexpression domains. In order for the gradientactivity domain to scale it is necessary tobroaden the gradient (increase its length-scale). In this illustration, the wing disc doublesin width and the ‘French flag’ pattern of hypo-thetical target genes remains in scale with therest of the wing disc. The activation thresholdfor all targets remains the same. (B) The expan-sion repression motif. At initial conditions, Dppstarts diffusing from its source and Pentagoneis highly expressed. Pentagone facilitates thediffusion of Dpp, thereby increasing its length-scale (green arrows). As it expands, Dpp startsreaching the expression domain of Pentagone(orange box). Subsequently, the production ofPentagone starts to decrease. Steady-state isreached when enough Dpp has reachedthe Pentagone expression domain to fullyrepress it. The expansion–repression motiflinks the length-scale of the morphogen toorgan size.

Current Biology Vol 21 No 19R816

Dpp in the wing disc and fits anexpansion–repression motif thathad previously been proposed bythe same authors [3] and that couldbe commonly employed throughoutevolution.

Scaling Pattern with SizeAs mentioned above, organismsare able to maintain proportionatebody plans independently of bodysize — a phenomenon called ‘scaling’.The same principle applies to organs;hence, the spatial pattern of the organ,and by extension the underlyingpattern of gene expression domainsin the organ, should scale with its size(Figure 1A). Lately, several studies[2,4,5] have started to shed light on thegenetic and molecular mechanismsthat ensure that gene expressionpatterns scale with size. These studiesused the Drosophila wing imaginaldisc — the wing primordium — asa model to study the activity gradientof the keymorphogenDecapentaplegic(Dpp). Bollenbach et al. [4] confirmedthat in the adult wing, the relativeposition of the wing veins — a goodreadout of Dpp activity — scales withfinal wing size. In a more recent paperfrom the same group, Wartlick et al. [5]showed that the activity gradient ofDpp also scales with wing disc sizeduring development.

Dpp contributes to wing growthand differentiation by establishingthe expression pattern of genesalong the anterior-posterior (A/P) axisof the wing disc. In accordance withthe classic ‘French flag model’ ofmorphogen action [6] — a morphogenwhose concentration decreaseswith distance from its source activatesdifferent target genes at differentconcentrations — Dpp regulatesexpression of target genes ina dose-dependent manner [7]. Fromthis perspective, the notion that theDpp gradient can scale with tissuesize creates interesting challenges.The French flag model states thatcells interpret the physical gradientof a morphogen following athreshold-dependent mechanism:target geneswith high sensitivity for themorphogen have a broad expressiondomain whereas less sensitive targetsare expressed only at highermorphogen concentrations, closer tothe source of the morphogen. Theextent of the different expressiondomains is thus strictly linked to themorphogen concentration at specificpositions along the growth axis.

How can the proportions of geneexpression domains be correctlymaintained in organs of varying sizes?How are these proportions kept ingrowing organs? One could speculatethat the sensitivity of the target genes

could be modified according to organsize in a way that would preservescaled proportions. On the other hand,the gradient itself could be modifiedand pinned to organ size in sucha manner that the boundaries of geneexpression domains remainproportionate and at fixed morphogenconcentration thresholds.Interestingly, for a morphogen gradientof fixed absolute levels this can beachieved by linking the length-scaleof the gradient to the length-axis ofthe tissue (Figure 1A).

Expansion–Repression: Modeland Mechanism for ScalingLast year, Ben Zvi and Barkai [3]proposed a mathematical model thatwould allow morphogen gradients toscale. Their model is based ona feedback motif and ensures that theactivity gradient of amorphogen scalesby pinning its length-scale to the size ofthe tissue. The feedback motif consistsof two main components: an expanderand a repressor. The expander allowsthe morphogen to diffuse better, eitherby diminishing the degradation rateof the morphogen or by facilitating itsdiffusion. The repressor negativelyregulates the production of theexpander. The morphogen playsthe role of the repressor. The roleof the expander could in theory beassigned to any molecule that is bothstable and diffusible.The expansion–repression model

proposes the following course ofevents: initially, the expander isexpressed at high levels at the edgesof the tissue, while the expanderprotein is distributed homogeneouslythroughout the morphogenetic field.As the morphogen starts diffusing, theexpander enhances the morphogen’srange. As the morphogen expands, itsprofile broadens and it starts reachingand restricting the expression domainof the expander. The morphogengradient continues to expand, as thetissue grows, until the morphogenlevels at the most distal portions of thefield are high enough to repress theexpander (Figure 1B). At this point,steady state is reached and thequantity of morphogen at the mostdistal regions becomes pinned tothe amount needed to fully repressthe expander. As a result, theexpansion–repression motif linksthe morphogen profile to the sizeof the tissue and modifies thelength-scale of the gradient

DispatchR817

accordingly. With this simple,two-component feedback motif,Ben Zvi and Barkai [3] provideda parsimonious solution to themorphogen gradient scaling problem.

However insightful and elegant, theexpansion–repression remained still‘only’ a model until Ben-Zvi andcolleagues latest study [2]. In aninteresting act of synthesis, theyanalyzed the molecular properties andexpression domains of putative Dppexpanders. They realized that therecently characterized proteinPentagone [8] met the requirementsthat their model demanded of anexpander: Pentagone is expressedat the edges of the wing disc alongthe A/P axis, is repressed by Dpp andis a stable and diffusible molecule.In addition, Pentagone was shownto modify the Dpp gradient throughinteractions with Dally, a heparansulfate proteoglycan that can affectboth themobility andstabilityofDpp [8].

So, is Pentagone Dpp’s expander?Ben-Zvi and colleagues [2] showthat Pentagone overexpressionleads to a broader Dpp gradientwhile Pentagone loss-of-functionleads to a steeper Dpp gradient.These results suggest that Pentagoneplays the role of the expanderthat ensures that the Dpp gradientis kept in scale during wing discdevelopment (Figure 1B). At themolecular level, Ben-Zviand colleagues [2] propose that theinteraction between Pentagoneand Dally could reduce the affinityof Dpp for its receptor Thickveins.This would decrease Dpp degradationby receptor-mediated endocytosis.However, these propositions aremainly speculative and future workwill show how exactly Pentagone leadsto either a decrease in Dpp degradationor an increase in Dpp mobility.

Expansion–Repression Everywhere?An important feature of theexpansion–repression motif is itssimplicity. Therefore, it seemspotentially very well suited toorchestrating scaling in animals asa whole. Consistent with this, previouswork by the same authors haddemonstrated that the BMP gradient inthe early Xenopus laevis embryo scalesthrough an expansion–repressionmechanism [9]. Dpp and BMP belongto the same protein super-family.However, the molecular details bywhich the system is implemented in the

frog embryo are quite different andinvolve another set of players than inDrosophila. This highlights that, whilethe players might change, theexpansion–repression motif couldremain conserved. Hence, the principleof the expansion–repression feedbackmotif could be commonly employedeven though the details of itsimplementation may vary.

The question of how a morphogengradient could scale with tissue sizehas mainly been treated with the initialpostulate that the magnitude of themorphogen concentration wouldremain fixed while the length-scale ofthe gradient would increase. It is thisisotropic expansion of the morphogengradient that the expansion–repressionmotif explains so well and that, as wementioned previously, has beenobserved in another model organism[9]. However, Wartlick andcollaborators [5] reported also anincrease in the magnitude of the Dppsignalling gradient. They observed thatthe Dpp gradient was not only gettingbroader as the discs grew, but theabsolute levels of morphogen along thefield were also increasing over time andwith size. This raises new questions asto how the concentration thresholds atwhich target genes are activated orrepressed bymorphogens are adjustedto increasing levels of morphogenconcentration. This is a new, differenttype of challenge to the French Flagmodel than the one discussed above.Linking the length-scale of themorphogen gradient to tissue-sizewould not suffice to maintain adequateproportions at fixed morphogenthresholds if the magnitude ofmorphogen concentration indeedincreased over the entire field.

It remains to be seen whetherthe expansion–repression motif hasthepotential to resolve this conundrum,or whether there are other scalingmechanisms yet to be discovered.Drosophila researchers, it seems,still have further, higher peaks toscale, to untangle the logic that ensuresthe development of well-proportionedanimals rather than Dumbo-likespecimens, however cute they may be.

References1. Robertson, F.W. (1963). The ecological genetics

of growth in Drosophila 6. The geneticcorrelation between the duration of the larvalperiod and body size in relation to larval diet.Genet. Res. 4, 74–92.

2. Ben-Zvi, D., Pyrowolakis, G., Barkai, N., andShilo, B.Z. Expansion–repression mechanism forscaling the Dpp activation gradient in Drosophilawing imaginal discs. Curr. Biol. 21, 1391–1396.

3. Ben-Zvi, D., and Barkai, N. (2010). Scaling ofmorphogen gradients by an expansion–repression integral feedback control. Proc. Natl.Acad. Sci. USA 107, 6924–6929.

4. Bollenbach, T., Pantazis, P., Kicheva, A.,Bokel, C., Gonzalez-Gaitan, M., and Julicher, F.(2008). Precision of the Dpp gradient.Development 135, 1137–1146.

5. Wartlick, O., Mumcu, P., Kicheva, A., Bittig, T.,Seum, C., Julicher, F., and Gonzalez-Gaitan, M.(2011). Dynamics of Dpp signaling andproliferation control. Science 331, 1154–1159.

6. Wolpert, L. (1969). Positional information and thespatial pattern of cellular differentiation. J.Theor. Biol. 25, 1–47.

7. Nellen, D., Burke, R., Struhl, G., and Basler, K.(1996). Direct and long-range action of a DPPmorphogen gradient. Cell 85, 357–368.

8. Vuilleumier, R., Springhorn, A., Patterson, L.,Koidl, S., Hammerschmidt, M., Affolter, M., andPyrowolakis, G. (2010). Control of Dppmorphogen signalling by a secreted feedbackregulator. Nat. Cell Biol. 12, 611–617.

9. Ben-Zvi, D., Shilo,B.Z., Fainsod,A., andBarkai,N.(2008). Scaling of the BMP activation gradient inXenopus embryos. Nature 453, 1205–1211.

Institute of Molecular Life Sciences,University of Zurich, Winterthurerstr. 190,CH-8057 Zurich, Switzerland.*E-mail: konrad.basler@imls.uzh.ch

DOI: 10.1016/j.cub.2011.08.041

Plant Development: Light ExposureDirects Meristem Fate

Leaf initiation was previously thought to be self-regulated and not reliant onenvironmental cues. However, a recent study has revealed that light redirectsmeristem fate from maintenance to lateral organ initiation, through theregulation of the plant hormones auxin and cytokinin.

Jayne Griffiths and Karen Halliday*

The shoot apical meristem is located atthe tip of the plant stem and is required

for the production of new leavesthroughout the life of the plant. It isessentially a dome-shaped structurewith undifferentiated cells at the tip,