spring ’15 vol. 28 no. 2 bulletin - hhmi.orgneuroscientists must do as astronomers do ... grow in...

4000 Jones Bridge Road Chevy Chase, Maryland 20815-6789www.hhmi.org

Address Service Requested

Aru

n S

amp

ath

ku

mar

an

d E

llio

t Mey

erow

itz

Plant PaversJigsaw puzzle-shaped cells bulging with chlorophylls (red) cover

the surface of leaves in many flowering plants. The fanciful pattern is more than just a whimsical quirk of nature, however. HHMI Investigator

Elliot Meyerowitz and his team have shown that, as plants grow, the indented regions and lobe-like outgrowths of these so-called pavement cells create internal stresses that reorganize cytoskeletal proteins. These microtubule proteins align along the cell walls (green), a process that in

turn reinforces them against stress – evidence of a subcellular mechanical feedback loop. Learn more about the role of force in the development of

animals, as well as plants, in “Show of Force,” page 12.

bulletin

hh

mi b

ulle

tin

• ho

wa

rd

hu

gh

es m

ed

ica

l inst

itu

te

• ww

w.h

hm

i.or

gv

ol. 2

8 / n

o. 2

Spring ’15 Vol. 28 No. 2

Small Forces,Big Impact

Mechanical stress plays avital role in the development

of both animals and plants

in this issue Neural Theory at Janelia

Marta Zlatic in the FootlightsBacterial Barcodes

12Like the creases in a sheet of origami paper, the stripes that pattern these fruit fly embryos guide forces that will govern their shape. As an embryo becomes longer and thinner, lines of protein are laid down along the growing fly’s head-to-tail axis. Each vibrant band consists of one of three members of the Toll receptor family. The receptors allow the cells to differentiate themselves from their neighbors and perform the directional movements necessary to remodel the growing embryonic tissue. A

dam

Par

é an

d Je

nn

ifer

Zal

len

HHMI Bulletin / Spring 2015

Observations

Vic

to N

gai

With Theory Comes OrderThe gray matter of the human brain may not be as physically vast as the dark matter of outer space. But the challenge of crunching astronomical data pales compared to the complexity of parsing today’s deluge of data about the brain. Making sense of that torrent requires more than banks of supercomputers, says computational neuroscientist and essayist Olaf Sporns. To achieve fundamental insights, he posits, neuroscientists must do as astronomers do and apply a theoretical framework to their immense quantities of empirical data.

Neuroscience, it turns out, is on the brink of its own “big data” revolution. Supplementing more traditional practices of small-scale hypothesis-driven laboratory research, a growing number of large-scale brain data collection and data aggregation ventures are now underway, and prospects are that this trend will only grow in future years. Mapping the roughly one hundred trillion synaptic connections of a human brain would by some estimates generate on the order of a zettabyte of data (a zettabyte corresponds to one million petabytes). For comparison, that’s about equal to the amount of digital information created by all humans worldwide in the year 2010. And recently proposed efforts to map the human brain’s functional activity at the resolution of individual cells and synapses might dwarf even these numbers by orders of magnitude. The coming data deluge is likely to transform neuroscience, from the slow and painstaking accumulation of results gathered in small experimental studies to a discipline more like nuclear physics or astronomy, with giant amounts of data pulled in by specialized facilities or “brain observatories” that are the equivalent of particle colliders and space telescopes.

What to do with all this data? Physics and astronomy can draw on a rich and (mostly) solid foundation of theories and natural laws that can bring order to the maelstrom of empirical data. Theories enable significant data reduction by identifying important variables to track and thus distilling a torrent of primary data pulled in by sophisticated instruments into interpretable form. Theory translates “big data” into “small data.” A remarkable example was the astronomer Edwin Hubble’s discovery of the expansion of the universe in 1929. Integrating

over years of observation, Hubble reported a proportional relationship between redshifts in the spectra of galaxies (interpreted as their recession speeds) and their physical distances. Viewed in the context of cosmological models Albert Einstein and Willem de Sitter formulated earlier, these data strongly supported cosmic expansion. This monumental insight came from a dataset that comprised less than fifty data points, compressible to a fraction of a kilobyte. When it comes to applying theory to “big data,” neuroscience, to put it mildly, has some catching up to do. Sure enough, there are many ways of analyzing brain data that are useful and productive for extracting regularities from neural recordings, filtering signal from noise, deciphering neural codes, identifying coherent neuronal populations, and so forth. But data analysis isn’t theory. At the time of this writing, neuroscience still largely lacks organizing principles or a theoretical framework for converting brain data into fundamental knowledge and understanding.

From The Future of the Brain: Essays by the World’s

Leading Neuroscientists, edited by Gary Marcus

and Jeremy Freeman. © 2015 by Gary Marcus and

Jeremy Freeman. Reprinted with permission

from Princeton University Press.

ContentsSpring ’15 Vol. 28 No. 02

Web-Only Content

Watch a zebrafish’s brain light up as it takes a virtual swim.

Witness tiny pores called stomata develop on a leaf.

See a mouse drink with gusto after a prompt from a light-triggered protein.

Travel inside the body to experience an allergy-induced immune cell reaction.

Observe how mechanical forces can shape cells and tissues.

Departments president’s letter03 AGenerationatRisk centrifuge04 MakingItReal05 Craftsman benchreport06 DiggingDeeperBeyondItch08 Autism’sGeneticRoots10 OpenandShutCase perspectives&opinions30 TheTheoryConnection32 Q&A–Whatbookhave youreadthatinfluenced yourwork? chronicle34 Science Education FromInsighttoAction36 Toolbox BarcodingBacteria38 Lab Book TCellBurnout GotThirst?Here’sWhy SalvagingSNAREs observations WithTheoryComesOrder

Features12 Show of Force Scientistsarelearningmyriad waysthatsmallforcesaddupto abigimpactonthedevelopment oforganisms,fromplants toanimals.

18 All the World’s a Stage MartaZlaticisascapable ofportrayingtheGreekheroine Hecubaassheisofpulling backthecurtainontheneural circuitryoffruitflylarvae. Atbothpursuits,she’swon wideacclaim.

24 Navigating the Torrent Experimentalneuroscientists andtheoristsatJaneliaare joiningforcestomakesense of–andimprove–thedelugeof datacomingoutoflabs.

All the World’s a Stage, page 18

www.hhmi.org/bulletin

Cover illustration:

Martin Nicolausson

This paper is certified by SmartWood for FSC standards, which promote environmentally appropriate, socially beneficial, and economically viable management of the world’s forests.E

liM

eir

Kap

lan

2 Spring 2015 / HHMI Bulletin

Gra

ham

Rou

mie

u,B

ryn

Nel

son

,Car

rie

Arn

old

,Dav

idF

ried

man

Contributors

Telephone (301) 215.8500 • Fax (301) 215.8863 www.hhmi.org© 2015 Howard Hughes Medical InstituteTheopinions,beliefs,andviewpointsexpressedbyauthorsintheHHMI Bulletindonotnecessarilyreflecttheopinions,beliefs,viewpoints,orofficialpoliciesoftheHowardHughesMedicalInstitute.

HHMI TRUSTEESKurt L. Schmoke, Esq., ChairmanPresident / University of BaltimoreJames A. Baker, III, Esq.Partner / Baker Botts. LLPAmbassador Charlene Barshefsky, Esq.Senior International Partner / WilmerHaleJoseph L. Goldstein, MDRegental Professor & Chairman Department of Molecular Genetics University of Texas Southwestern Medical CenterGarnett L. KeithChairman & CEO SeaBridge Investment Advisors, LLCFred R. LummisChairman & CEO Platform Partners, LLCSir Paul Nurse, PhDPresident / The Royal SocietyDame Alison F. Richard, PhDSenior Research Scientist & Professor Emerita Yale UniversityClayton S. Rose, PhDProfessor Harvard Business SchoolAnne M. TatlockRetired Chairman & CEO Fiduciary Trust Company International

HHMI SENIOR EXECUTIVE TEAMRobert Tjian, PhD / PresidentCheryl A. Moore / Executive V.P. & Chief Operating OfficerKathryn S. Brown / Head of CommunicationsSean B. Carroll, PhD / V.P. for Science EducationJennifer L. Farris / Head of Administrative ServicesHeidi E. Henning / V.P. & General CounselMohamoud Jibrell / V.P. for Information TechnologyNitin Kotak / V.P. & Chief Financial OfficerErin O’Shea, PhD / V.P. & Chief Scientific OfficerGerald M. Rubin, PhD / V.P. & Executive Director, Janelia Research CampusKathy A. Wyszynski / V.P. for Human ResourcesLandis Zimmerman / V.P. & Chief Investment Officer

HHMI BULLETIN STAFFMary Beth Gardiner / EditorDana Cook Grossman / Story EditorJim Keeley / Science EditorNicole Kresge / Assistant Editor

ADDITIONAL CONTRIBUTORSElaine Alibrandi, Michelle Cissell, Heather McDonald, Anne Sommer Pentagram Design / DesignAllied Printing / Printing & BindingBryn Nelson(“Autism’s

GeneticRoots,”page8)isaformermicrobiologistwhodecidedhewouldmuchratherwriteaboutmicrobesthanmutatethem.NowanovercaffeinatedsciencewriterbasedinSeattle,hehaspublishedarticlesinThe New York Times,Nature,Discover, Mosaic,andScientific American.

Carrie Arnold(“NavigatingtheTorrent,”page24)isafreelancescienceandhealthwriterlivinginVirginia.WhensheisnotwritingaboutthelivingworldforpublicationslikeNOVA Next,Scientific American,andNational Geographic,youcanprobablyfindherdrinkingcoffee,cycling,knitting,orannoyingthecat.

IllustratorGraham Roumieu(“DiggingDeeperBeyondItch,”page6)isthecreatorofthreefauxautobiographiesofBigfoot.HisworkalsohasappearedinThe Atlantic, The New York Times, Harper’s,ananimatedcommercialforcorndogs,andelsewhere.

Aphotographerandfilmmaker,David Friedman(“TheTheoryConnection,”page30)exploresstoriesofscience,technology,andhumannaturethroughportraitureandstorytelling.Hiscurrentprojectsincludeanongoinglookatthelivesofinventorsfromallwalksoflife.

3HHMI Bulletin / Spring 2015Ja

mes

Keg

ley

President’s Letter

A Generation at Riskearlierthisyear ,IhadthegoodfortunetoattendadayofeventsattheWhiteHousethatcelebratedsciencestudentsandcalledonphilanthropieslikeHHMItocontinuehelpingtodrivescienceforward.ReflectingonsuccessfulSTEMscholars,PresidentObamasaidthis:

“It’snotenoughforustojustliftupyoungpeopleandsay,‘greatjob,waytogo.’You’vealsogottohavelabstogoto,andyou’vegottobeabletosupportyourselfwhileyou’redoingthisamazingresearch.Thatinvolvesusasasocietymakingthekindsofinvestmentsthataregoingtobenecessaryforustocontinuetoinnovateformanyyearstocome.”

Inrecognizingcommitmentstoemergingscientists,thepresidentcalledoutanewprogram,“FacultyScholars,”thatHHMIhascreatedwiththeBill&MelindaGatesFoundationandtheSimonsFoundation.Throughthisprogram,wewilltogetherawardupto70grants,everytwotothreeyears,toearlycareerbasicresearchersandphysicianscientistswhohavethepotentialtomakeuniquecontributionstotheirfields.ThefirstroundoftheFacultyScholarscompetitionlaunchedinMarch,andweplantomakethefirstgrantawardsinthefallof2016.Ourthreeorganizationswillinvestatotalof$148millioninresearchsupportovertheprogram’sfirstfiveyears.

Weplantoidentifyandsupportnotonlyaccomplishedresearcherswhohavealreadydemonstratedtheircapacityforimportantscientificcontributions,butalsopromisingscientistswhohavenotyetreachedtheirpotential.

Today’semergingscientistsfaceincreasinglytoughodds.ThepercentofallNIHresearchgrantfundingawardedtoscientistsundertheageof36hasdroppedfrom5.6percentto1.3percentoverroughlythepastdecade.Theaveragescientistisnow42yearsoldbeforeobtaininghisorherfirstR01NIHresearchgrant.Withthisreality,agrowingnumberofpromisingresearchersareleavingtheU.S.foropportunitiesoverseasor,worse,abandoningsciencealtogether,promptingsometolabelthem“agenerationatrisk.”

Scientistsatthebeginningoftheircareersneedadequatefundingtobeaggressivewiththeirresearchprograms.Inthatearlyphase,generallybetweenyears4and10,start-upfundsfromauniversitybecomeexhausted,justwhenit’stimetostartrampingup.For

thesescientiststopersist,theymusthavehelptoclearthepathway.

HHMIalreadysupportssomeexcellentearlycareerscientistsintheUnitedStatesandabroad.Butthereismuchmoretodotohelplaunchthenextgenerationofscientists.Weneedtoexpandourreach,nurturinggreaternumbersofearlycareerscientistswiththefunding,mentoring,collaboration,andtrainingtheyneedtosuccessfullyestablishcareerstoday.Doingthiswell,atscale,willrequiresignificant,sustainedleadershipandcollaboration–thekindofsupportbestprovidedbybig,strategicorganizationscomingtogethertocreatechange.

IpersonallylookforwardtomeetingthenewestmembersofHHMI’scommunitywhenourfirstcropofFacultyScholarsarrivesatHHMIformeetings.Waitandseewhattheywilldo.

“We need to expand our reach, nurturing greater numbers of early career scientists with the funding, mentoring, collaboration, and training they need to successfully establish careers today.”—roberttjian

Centrifuge4 Spring 2015 / HHMI Bulletin

Making It Realwhenshelectures aboutsexdeterminationtoherLifeSciences1bstudents,evolutionarygeneticistHopiHoekstragetspersonal.Reallypersonal.ShestartsbyscopingouttheundergraduatesinHarvard’sScienceCenterauditoriumandpickingamalestudentwhosayshe’snoteasilyembarrassed.Shestandsbesidehimandaskstheother450-somestudentstowatchhimwhilecountingtofiveinunison.After“...five,”Hoekstraasksforestimatesofhowmanyspermtheyoungmanproducedduringtheprecedingfewseconds.Thenshedeliverstheanswer:about7,500.

“Itrytogetthestudentsexcitedaboutthematerialbeforethelectureevenstarts,”saysHoekstra,anHHMIinvestigatoratHarvardUniversity.Hoekstracallsthis“motivatingthematerial.”

“Startinglectureswithastory–frompersonalstories,tomysteriesofthefield,tohistoricalfacts–helpsgetthestudents’imaginationsgoing.”Withoutthemotivation,Hoekstrasays,“it’sjustabunchoffacts.”

Tosetthestageforherlectureonepigenetics,sheusesafamilystory.Sheprojectsaphotoofhergrandmother,whosurvivedthe1944HongerwinterinNazi-occupiedHolland,andexplainsthathergrandmothersubsistedontulipbulbswhilepregnantwithHoekstra’smother.Wheninutero,Hoekstra’smotherwasformingtheovumfromwhichHoekstraherselfwouldeventuallydevelop.Through

thategg,Hoekstramayhavebeendirectlyaffectedbyhergrandmother’sprivation.Inotherwords,environmentalsignalsmighthavetriggeredepigeneticchangesinhermother’sova–heritablechanges,butonesthatalterhowgenesareexpressedratherthantheDNAitself.

“I’mpartofthislineage,”Hoekstratellstheclass.Thensheshowsasnapshotofherson,three-year-oldHenry.“He’stherealtestcase,”sheexplains–atestcaseforthe indirecteffectsofwhathisgreat-grandmotherendured.Researchhasshownthatepigeneticchangesinpeoplewhosufferedfrommalnutritioncanpredisposetheirdescendantstodiabetesandobesity.

It’sonlythenthatHoekstrasays,“Let’sseehowthisworksatthemolecularlevel.”She’sprettysurethatifshe’dskippedthestory

andsimplyannounced,“Today’slectureisaboutmethylation”–oneformofepigeneticalteration–herstudentswouldhavebeenfarlessengaged.

Hoekstra’sresearchinvolvesevolutionarychangesinwildpopulations,suchasdeermice.Butshethinksalotnotjustaboutevolution,butalsoabouthowtoteachit.Aftereverylecture,shejotscommentsonthebluefolderholdingherlecturenotes–suggestionstoherselfsuchas,“Canmovefasterthroughfirst

part”and“Leavemoretimeforrat-lickingexamples.”InJune2014,HoekstrawasawardedaHarvardCollegeProfessorship,afive-yearappointmentthatrecognizesexcellenceinteaching.

Shebelievesshe’llattractmorestudentstosciencebydemonstratingthatbiologyresearchisverymuchaworkinprogress.Forinstance,lastfallsheinformedherstudentsthat,asrecentlyastwoyearsago,theirtextbookdidn’tevenhaveachapteronepigenetics.“Theythinkofeverythingassolved,butthedoorstothishavejustopened.Theycanthink,‘Wow,there’ssomuchstilltobediscovered!’”—Cathy Shufro

Oli

vier

Ku

gle

r

5HHMI Bulletin / Spring 2015Sa

mC

hiv

ers

Craftsmanwhengreghannon wasagraduatestudentatCaseWesternReserveUniversity,helikedtohangaroundthecivilengineeringdepartment.Friendstherewouldlettheyoungmolecularbiologistusethedepartment’swoodandmetalshops.

Helearnedtorunamillingmachineandoperatealathe.Hebuiltawinerack,bookshelves,andgelboxesforhislab.Thenoneday,inadustybackcornerofawoodstore,HannonfoundwhathebelieveswassomeofthelastCubanmahoganyimportedintotheUnitedStates.“Ispentmorethanamonth’sgradstudentsalaryonit,”recallsHannon,wholookedattherawmahoganyandsawadiningroomtable.“I’dneverbuiltsuchasubstantialpieceoffurniture,butIjustjumpedrightin.”

Hannonhadtheforesighttochooseatrestledesign:thetable’sstructurewassupportedbyfourwedgesthatcouldeasilybeknockedout,allowingittobedisassembledandtransported.WhenheleftCaseWesternin1992andheadedeastforpostdoctoralstudiesatNewYork’sColdSpringHarborLaboratory,Hannonbroughtthetable.

AtColdSpringHarbor,HannonwouldpioneerthestudyofRNAinterference:harnessingforresearchortherapeuticpurposesthenucleicacidsusedbycellstoregulategeneexpressionandproteinsynthesis.Healso

builtbenches,moretables,cabinets,adresser.Headdedaroomtohishouse.

There’ssomeoverlapbetweenhisavocationalandvocationalpursuits.Woodworkandappliedmolecularbiologyarebothexercisesinproblemsolving–handlinginevitableconfrontationswiththeunexpected.

Askilledbuilder“hasathoughtprocessgearedtowardthatandcanthinkaheadanynumberofsteps,”saysHannon.“Thinkingabouthowyoubuildthingsonamolecularlevel,theprocessisthesame.”

Inotherways,though,hefindsbuildingareleasefromthelab.Itrequirestotal,measure-twice-cut-oncefocus.“I’mnotameditationkindofguy,butIimagineit’ssomethinglikethat,”Hannonsaysofpottery-throwing,anotherofhis

manualpursuits.“Theclayknowswhatyourmoodis.”

Bothwoodworkingandpotteryalsooffermoreimmediategratificationthanhisworkusuallyprovides.“Youmakeprogresseveryday,”saysHannon.“That’snotalwaysthecasewithscience,whichispunctuatedbyperiodsofinsightfollowedbythelong,hardslogofmakingthingswork.”

Lastfall,after23yearsatColdSpringHarbor,HannonbegantorelocatehislaboratorytotheUniversityofCambridgeintheUnitedKingdom,withafulltransitionexpectedinJuneofthisyear.Itwasadauntingshift.HewentfrompostdoctofullprofessorandHHMIinvestigatoratalmostliterallythesamebench.

Butasdifficultasitwastomove,Hannonisfindingin

Cambridgeaneededchangeofpace:achancetodomoreappliedresearch,exposuretonewideas,ajoltoutofintellectualcomplacency.

Italsooffersawealthofhomeprojects.Heandhiswifearerenovatinga110-year-oldcountryhouse.They’reputtinginnewfloors,redoingthebathrooms,andbuildingchickenrunsandgoatpensforwhatmaywellbecometheirownhomedairy.

Hannonisalsorebuildingthehouse’smainstaircase.Heapproachesthisnewchallengewithsometrepidation;completingtherestorationcouldtakeyears.Butheknowstheresultswillbeenduring.Inthediningroomofthehousestandshis25-year-oldmahoganytable.—Brandon Keim

Bench Report

Digging Deeper Beyond ItchScientistsidentifyanimmunecellreceptorthatmaybeattherootofsomedrugallergies.xinzhongdonghas beenscratchinganitchforthepast14years.Likeacaseofhivesthatwon’tgoaway,theMrgfamilyofproteinshasbeennaggingatDongeversincehediscoveredthereceptorsonsensoryneuronsin2001.He’sbeenconsumedwithtryingto

figureoutwhattheproteinsdo.Andit’snoeasytask–thereare31differentversionsofthereceptorinmiceand8inhumans.Sofar,theHHMIearlycareerscientisthaslearnedthatsomeofthereceptorssenseitchandotherssensepain.Hislabteam’smostrecentdiscoveryissomewhatofananomaly:they’vefoundanMrgreceptoronimmunecellsthatplaysaroleindrugallergies.

AllergicreactionsoccurwhenimmunoglobulinE(IgE)antibodiesperceiveathreattothebody–oftenanotherwiseharmlesssubstancelikecatdanderorpollen.TheIgEantibodiesbindtothesesubstancesanddockontospecificreceptorsonmastcells.Theseimmunecellsarepackedwithgranulesofhistamineandotherproinflammatorymoleculesthatleadtothehives,irritation,orpainassociatedwithallergies.WhentriggeredbyIgEantibodies,mastcellsreleasetheirpayload,producinganallergicreaction.

Thoughmastcellsaretypicallyactivatedbyanallergy-specificIgEantibody,theycanalsobeactivatedbyawiderangeofotherunrelatedthings,explainsBenjaminMcNeil,apostdocinDong’slabatJohnsHopkinsUniversity.“Peoplewhodisplaysevereallergicreactionstoenvironmentalchemicals,FDA-approveddrugs,andparasites,forexample,oftendon’thaveanyIgEantibodiesagainstthesethings,”hesays.“Theirreactionsaremastcell-mediated,butviaadifferentpathway.”Themajorityofcompoundsthattriggersuchantibody-independentreactionsarecollectivelyknownas“basicsecretagogues,”becauseoftheirpositive,orbasic,chargeandtheirabilitytocausemastcellstosecretetheirgranules.

WhenMcNeiljoinedthelab,in2011,hepickedwhathethoughtwasafairlysimpleproject:figureoutthefunctionofahumanMrgreceptorcalledMRGX2.“Itwassupposedtobeastraightforwardproject,”McNeilsays.“Iwasaneuroscientistandwantedtostudysensorybiology.WethoughtthatMRGX2wasonneurons,too,sothiswasawayformetolearnthetechniquesinthelabwhilecharacterizingthereceptor.”But,asisoften

thecaseinscientificresearch,itwasn’tthatsimple.TheveryfirstthingMcNeildiscoveredwasthatMRGX2isnotexpressedinneurons.

In2006,agroupofJapanesescientistspublishedapapersuggestingthatMRGX2mightbeinvolvedinnon-IgEmastcellresponses.Followinguponthislead,McNeildiscoveredthatampleamountsoftheproteinwereexpressedinhumanandmousemastcells.Infact,hesays,“It’sthemostmastcell-specificgeneintheentiregenome.Theonlywaytoreallydefineamastcellisbythis.”Inotherwords,ifacellexpressestheMRGX2 gene,it’sdefinitelyamastcell.

Thenextstepwasconfirmingthereceptor’sfunction.AfterMcNeildiscoveredthatasimilarMrgprotein–Mrgb2–existsonmousemastcells,hecreatedmicethatdidn’texpressthegenetotesttheprotein’saffinityfordifferentcompounds.Chemicalsthattriggeredthereceptorwouldcauseanallergicreactioninnormalmicebutnotintheknockoutmice.McNeilwentontotesteverybasicsecretagoguehecouldgethishandson.Strikingly,everythinghetriedseemedtoactivatethereceptor.“Afterthat,Irealized,‘Okay,thisisareallybizarrereceptor,’andstartedexpandingthelistofcandidates,”herecalls.

6 Spring2015/HHMIBulletin

Intheend,thelistofcompoundsthatactivatedtheMrgmouseanalogreceptorwashuge.Itincludedinsectvenoms,cancerandHIVmedications,neuromuscular-blockingdrugscommonlyusedforanesthesiaduringsurgery,andseveralmoleculesinaclassofantibioticscalledfluoroquinolones.Allofthecompoundshadtwocharacteristicsincommon–theyweresmallandpositivelycharged.“Theyareotherwisecompletelyunrelated,structurallyspeaking,”McNeilsays.“Theycandifferbyafactoroftenintheirsizeandhavetotallydifferentcompositions.It’sreallyremarkablethatthere’sjustonesinglereceptorforallofthesesubstances.”

Thestudy,publishedinNatureonMarch12,2015,hasopenedthefloodgatesfordozensofnewexperimentsandcollaborationsin

theDonglaboratory.McNeilisscreeningothersuspectedallergy-causingcompoundsforbindingtoMRGX2.DonghasstartedworkingwithscientistsatGlaxoSmithKlinetofindsmallmoleculesthatcanblockMRGX2butstillallowIgE-mediatedactivityinmastcells.He’salsodevelopingadrug-bindingassayforMRGX2thatwillallowpharmaceuticalcompaniestotestwhethertheirnewcompoundselicitallergicreactions.

Andthere’salsomoreworktobedoneonotherMrgfamilymembers.“Westillhavemanyothergenestostudy,”saysDong.“Wedon’tknowexactlywheretheyarelocated,wherethey’reexpressedintissues,andwhattheirfunctionsare.”Clearly,it’sanitchhe’snotreadytostopscratchinganytimesoon.—Nicole Kresge

“It’s really remarkable that there’s just one single receptor for all of these substances.”—benjaminmcneilG

rah

amR

oum

ieu

7HHMIBulletin/Spring2015

Spring2015/HHMIBulletin8

Bench Report

Autism’s Genetic RootsUsing a fresh approach, researchers are making progress in unraveling the genetic complexities of the disorder.autismis notorious forthebroadrangeandvariableintensityofsymptomsthatcanimpairachild’sbehavior,communication,andsocialinteractions.Scientistshavelongknownthatdevelopmentaldisabilityisstronglyinfluencedbygenetics,butpinpointingacauseforautism’sonsethasprovenfrustrating.Evenidenticaltwinscanhavevastlydifferentformsofthecondition.

Apromising“genotype-first”strategypioneeredbyHHMIInvestigatorEvanEichleroftheUniversityofWashingtonisnowgivingthesearchforanswersamuch-neededboost.Insteadofbeginningwithadetailedcharacterizationofanindividual’ssymptomsandthenhuntingfortheresponsiblegeneormutation,EichlersequenceslargeamountsofDNAfrommultiplepatientsandthenseeksoutsharedgeneticvariantsthatmaypointtowardcommonpathstothedisorder.

Theapproachhasyieldedagrowinglistofautismsubtypesinwhichchildrenwiththesamemutationhaveahighlysimilarsuiteofsymptoms.“Ifeellikewe’rekindofwherecancerwasasafield20yearsago,”Eichlersays.Cancer,oncewidelyconsideredasingledisease,hassincebecomeanumbrellaofmorethan200types,eachwithitsowngeneticandenvironmentaldeterminants.“Iliketothinkthatthisiswhatwe’redoinginthecaseofautism:we’rehelpingtounravelthiscomplexity.Butwe’redoingitwithgeneticsfirst,”hesays.

Adecadeago,autismresearchersbeganfindinggeneticcluesinoftensporadic

mutationsknownascopynumbervariations,orCNVs,inwhichbigchunksofchromosomalDNAaremissingorduplicated.EichlerandhiscollaboratorsdiscoveredthatsomeoftheseCNVsoccurredalmostexclusivelyinchildrenwithautismordevelopmentaldelays.

Soonthereafter,aseriesoftechnologicalbreakthroughsbeganallowing“next-generation”DNAsequencingmethodstomorequicklyandcheaplyspelloutthegeneticcodeofautismpatients.Withthenewtools,Eichlerandcolleaguesbegansequencingtheexome–the1.5percentofthegenomethatencodesproteins–ofaffectedindividualsandtheirparents.Thistime,theresearchersfocusedonsinglebase-pairmutationsandsmallinsertionsordeletionstozeroinonspecificgenes.Evenwithonly20familiesinthatstudy,“wewerethrilledwiththecandidategenesthatwewereseeing,”Eichlersays.“Therewasalreadyhandwritingonthewallthatthiswasgoingtobeproductive.”

WhenheandotherresearchersappliedthesamestrategytoamuchlargercollectionofDNAfrompatients,parents,andunaffectedsiblings,theyfoundmoreevidenceforanexcessof“gene-killing”mutationsinautismpatients.Afterpoolingtheirresults,however,thelabteamsfoundthatthevastmajorityofmutationsmappedtoseparategenes.“JustastherearemanyroadstoRome,therearemanywaystogettoanautisticstate,”Eichlerexplains.Despitetheexcitementofseeingastronggeneticsignal,theresearchersrealizedtheywouldneedmanymoreDNAsamplestoprovetheinvolvementofanysinglegene.

Toscaleuptheeffort,Eichlerorganizedalargedata-sharingconsortiumoflabs,andaDecember2012paperinSciencetargeting44candidategenesdemonstratedthegroup’snewfoundpower.Bysequencingthegenesinnearly2,500individualswithautismspectrumdisorders,theconsortiumfounddisruptivemutationsinsixgenesthataccountedforanestimated1percentofallsporadiccases.

Asthesequencingeffortaccelerated,distinctgeneticsubtypesofautismbegantoemerge.Tellingly,individualswiththesamedisruptedgenetendedtosharesimilarphysical,clinical,andbehavioraltraits.BasedonearlyevidenceimplicatingagenecalledCHD8,forexample,Eichlerandcolleaguessequencedthe

geneagainin3,700individualswithautismordevelopmentaldelay.AstheyreportedinaJuly2014Cellstudy,theiranalysisyielded15independentmutations–andthetallyhassincesurpassed20.

Byhavingthechildrenre-evaluatedbypediatricians,psychiatrists,andotherclinicians,theresearchersdiscoveredthatalthoughonlyhalfoftheindividualsareintellectuallydisabled,mosthaveautismandmacrocephaly,oroverlylargeheads.Mostalsohaveexperiencedseveregastrointestinalproblems,andEichlersuspectsthatthesamegeneticsubtypemightcontributetoasleepdysfunctionthatkeepsthechildrenawakeforseveraldaysatatime.Inzebrafish,hislabfoundthatknockingouttheequivalentgeneresultedinfishwithsomewhatlargerheadsandimpaireddigestion.

AmongeightchildrenwithasporadicmutationinaseparategenecalledDYRK1A,bycontrast,Eichler’slabandcollaboratorsinAustraliaandEuropefoundthatallhaveautism,intellectualdisabilities,andunusuallysmallheads,ormicrocephaly.AsdescribedinaFebruary2015Molecular Psychiatrystudy,manyalsohavepointedchinsresultingfromaskulldeformitythatoftenrequiressurgerytoremoveovercrowdedteeth.Infruitflies,researchersdiscoveredthesamegenemutation20yearsagoandnamedit“minibrain”;DYRK1AisalsooneoftwogeneslinkedtocognitivedeficitsinDownsyndrome.

Eichlerandcolleaguesarenowscreeningthousandsoffamiliesformutationsin250candidategenesassociatedwithahalf-dozenproteinnetworkstentativelylinkedtoautism.Whenclinicianseventuallydeveloppotentialinterventionsforthedisorder,hesays,knowingwhichnetworkisimpairedinwhichpatientmayhelpdeterminetherightcourseoftherapy.

Intheinterim,henotes,ageneticdiagnosismayhelpparentsunderstandthebasisofautism,connectwithotherfamiliesforsupport,andrealizethatthey’renottoblamefortheirchild’scondition.“Ithinkitaddsacertainhumanaspecttoallofthisresearch,”Eichlersays.“Afteryearsofbeinginthisbusiness,IfindthisparttobesometimesbetterthantheNature andSciencepapers:thefactthatfortheseindividualgenes,we’rereallymakinganimpactonfamilies’livesinwhatfeelslikeverysubstantialways.”–Bryn Nelson P

atri

ckK

ehoe

“Just as there are many roads to Rome, there are many ways to get to an autistic state.”—evaneichler

Evan Eichler uses a genetics-first approach to unravel autism’s complexities.

Spring2015/HHMIBulletin

Open and Shut CaseResearchers are uncovering details of the molecules that control the development of stomatal pores on a leaf’s surface.it’s nowonder thatDominiqueBergmanntalksaboutherArabidopsiscellsactingliketoddlersandadolescents.Justthewayaparentmarksherkids’heightsonadoorframe,she’sspenthercareerchartingplantcelldevelopment.

“Howdoyougofromanaïvestemcellwithallsortsofpossibilitiestobeing

committedtoafateasadifferentiatedcell?”wondersBergmann,anHHMI-GBMFinvestigatoratStanfordUniversity.“Wedon’thaveanywhereawholerecordofhowanorganisbuiltfromthebeginningtotheend.”

ButBergmannhasnowmadehugestridesintrackingfromstarttofinishthedevelopmentofdiscretemulticellularstructuresonaleafandinplottingthedecisionstheircellsweremakingalongtheway.InanenterprisingnewstudypublishedApril6,2015,in Developmental Cell,herlaboratorytrackedtheshiftinggeneticprogramsofdevelopingleafcellsasstemcellsarise,differentiate,andformacompletestructure.

Theambitiousinvestigationwaspossible,inpart,becausethestructureshestudiesismadeupofjusttwospecializedguardcellsthatformaporeontheleaf’ssurface.Theso-calledstoma,Greekfor“mouth,”allowsplantstobreatheincarbondioxideandexpeloxygen.Stomataformationunfoldssimplyandelegantlyovertwodaysontheleafsurface,whereitcanbewatchedinrealtime.ThatgaveBergmannanedge:in2006and2007,herlabteamidentifiedthethreetranscriptionfactorsthatorchestrate,insequence,aleafcell’sentryintothestomatallineage,itsinitialcommitmenttobecomeaguardcellprecursor,anditsterminaldifferentiationintoaguardcell.

Whenthatsequenceisdisruptedbymutationsinthetranscriptionfactors,theappearanceofstomataonleavesisaltered.SPEECHLESS(SPCH)andMUTEmutantssportno“mouths,”whileFAMAmutants,namedfortheRomangoddessofspreadingrumors,produceanoverabundanceofdysfunctionalmouths.ThesethreemoleculesprovidedBergmannwithahandywayofisolatingcellsateachdevelopmentalstage,sothatherteamcoulddocumenthowchanges

inthecells’geneticprogramsmightdirectdevelopmentaldecisions.

“Ifyouthinkaboutthetransitionoftheadolescentyearsintoadulthood,alotofthingshappenduringthattime,”saysBergmann.“Wewantedtocapturethissortofgrowingupoverdevelopmentaltime.Whatgeneticswitchisflippinginthesecellstosay,‘I’mdonewithbeingastemcell;I’mgoingtocommitmyselftosomething’?”

Theteamsetouttouseknowncell-sortingtechniqueswithfluorescentlytaggedversionsofSPCH,MUTE,andFAMA,amongothermarkers,toidentifyandseparatethedifferentdevelopmentalstages.However,thosetechniqueswereintendedforfree-floatingmammalianimmunecells,notplantcellslockedtogetherbycellwalls.

AdaptingtechniquespioneeredbyHHMI-GBMFInvestigatorPhilipBenfey,Bergmann’sgroupwasabletodissolvethecellwalls,buttheystillhadtoamassenoughofthefleetingintermediatestemcellstogetanaccuratereadoutofthecells’entirearrayofactivelyexpressedgenes.Thesorting,combinedwithgenesequencing,revealedwhichgeneswereactiveateachtransitioninstomatadevelopment.

10

Bench Report To see an animation of 24 hours of stomata development, go to hhmi.org/

bulletin/spring-2015.

11HHMI Bulletin / Spring 2015

theyfound,cellshadactivatedgenesrelatedtochromatinorDNAmodification,whichseemedtosignalcommitment.“Theysettleddown–theyarenotturningonabunchoffunctionalproteinsyet,buttheyarenolongerpluripotent,”Bergmannexplains.“They’velockedintothatdevelopmentalpath.”

Inthefinal,differentiatedguardcellstage,theteamsawsomeofthosefunctionalproteinscomeonline,includingactivegenesthatencodesensorysystemsandionchannelsthatregulatetheopeningandclosingofstomata.

Bergmannnowfeelsshehasanoverarchingviewofthestoma’sdevelopment–fromamessyinfancyfullofpotential,throughanadolescencefocusedonchoosingapath,toanadultphasewithitsfatelockeddown.“Weseethecellgofromachaoticstate,whereitcoulddoanything,tobecomingmoreandmorecontrolled,”shesays.

Thekeystothatprocess,itturnsout,arethetranscriptionfactorsthatherlabidentified:it’sSPCHthatopensthegeneticdoorstovariousnascentpossibilitiesandFAMAthatshutsthosedoorstolockinthecell’smaturefate.—Kendall Powell

Images from a time series in stomata formation. Cells at different developmental stages are marked by nuclei labeled green (SPCH, stem-cell like), pink (MUTE, committed), or blue (FAMA, differentiating). The first two panels show the same group of cells at an early time point; in the third panel, taken a day later, cells are maturing. Scale bar in each is 5 microns.

But,Bergmannadmits,“Itwasamiserableexperimenttodo,”requiringtensofthousandsofseedlingstogetatleast2,000cellsofeachtype.

Onceherteamhadalistoftheactivegenespresentineachdevelopmentalstageofastomatalcell’slife,theycomparedtheirgeneticprofilestothoseofplantcelltypesthathadbeencharacterizedinotherlabs.Theyfoundthattheearlieststemcells,taggedandsortedbySPCH,werehighlyvariableanddidn’tquitematchtheprofilesofanyotherplantcellsoutthere.

“Itwaslikeabunchoftoddlersrunningaround.Therewasgreatpromisethere,butyoucouldn’tpredictanything,”saysBergmann.“Weinterpretedthatasreflectingtheirpluripotency.”(Thisobservationcomplementedanotherrecentfindingbyherlabteam–thatSPCHoscillationsresetsomeleafcells’identitytothestem-cellstate,sothenumberofstomataontheleafcanadjusttoenvironmentalchanges,suchasintemperatureorhumidity.)

Tostudythenextdevelopmentalstage,inwhichcellsbecometheprecursorguardmothercells,theteamtaggedandsortedcellsusingMUTEasamarker.Atthisstage,

“How do you go from a naïve stem cell with all sorts of possibilities to being committed to a fate as a differentiated cell?”—dominiquebergmannB

erg

man

nla

b

byrachelehrenberg

Show of Force Scientists are learning myriad ways that small forces add up to a big impact on the development of organisms, from plants to animals.

illustrationbymartinnicolausson

JK

evin

Wol

f


jenniferzallenhas scrutinizedmillionsofcells,butthedayshewitnessedafruitflycellinatug-of-warstandsout.

Zallen,anHHMIearlycareerscientistatMemorialSloanKetteringCancerCenter,wasexploringtheroleofmechanicalforcesinthedramaticelongationthatafruitflyembryoundergoesduringitsdevelopment.Overameretwohours,roughlyathousandcellsmobilizeenmasseandrearrangethemselvesintoanascentflythat’shalftheembryo’soriginalwidthandtwiceitslength.Previousworkhadimplicatedthemotorproteinmyosininthismassmovement,soZallen’steamlabeledmyosinwithafluorescenttagandriggedavideocameratothemicroscopesotheycouldwatchtheproteininaction.Then,herpostdocRodrigoFernandez-Gonzalezdelicatelypokedthetipofaglassneedleintotheembryoandsuckedinthetiniestbitoffluid,yankingonanearbycell.

Myosinfloodedtothesite,enablingthepinnedcelltocontractandthenescape,essentiallypullingitselfawayfromtheneedle.“Itwasfast,”saysZallen,“toofasttoinvolvechangesingeneexpressioninthenucleus.”Therewasnochangeinthecell’sgeneticmakeupandnochemicalsignalrecruitingmyosintotheneedle’stip.Yetthecell

anditsneighborshadexhibitedanimmediateandcollectiveresponsetothetugofthesuction.Justexperiencingmechanicaltensionappearedtobeenoughtokickmyosinintogear.

Thoseexperiments,publishedinDevelopmental Cellin2009,arepartofagrowingeffortbyscientiststoelucidatetheroleofmechanicalforcesinshapingbiologicaltissuesand,ultimately,entireorganisms.Theresearchisnotonlyyieldingnewinsightsintothestunningaestheticsofanimalandplantmorphology,

butitmayalsoleadtonewtricksforcontrollingplantarchitectureorforhaltingcancer’sspread.

Scientistshavelongappreciatedtheideathatmechanicalforcesareintegraltocreatingshapeandform.Nearlyacenturyago,intheintroductiontohistreatiseOn Growth and Form,ScottishscientistD’ArcyThompsonwrote,“Cellandtissue,shellandbone,leafandflower,aresomanyportionsofmatter,anditisinobediencetothelawsofphysicsthattheirparticleshavebeenmoved,moldedandconformed.”

Theimportanceofthosephysicallawstotheformationofhealthytissuesandorganshasalsolongbeenacknowledged:weight-bearingexerciseiscrucialformaintainingstrongbones,andturgorpressureoncellwallsofplantskeepsleavesandflowersfromwilting.Butduringthemolecularrevolutionofthe1990s,theroleofgenes,signalingmolecules,andproteinsindevelopmenttookcenterstage,partiallyobscuringtheimportantroleofphysicalforces.

Arevivalisnowunderway.Armedwithknowledgegleanedduringthemolecularera,alongwithnewtoolsformanipulatingandimagingcellsandunprecedentedcomputingpower,scientistsarereexaminingtheprofoundimpactofforce.Byzeroinginoncellsastheyaresqueezedandstretchedinrealtime,researcherscanuntanglethedevelopmentalconsequencesofforcegeneration,propagation,anddetection.Effortstoquantifytheseactions,aswellasthecellularplayersinvolved,areleadingtotestablemodelsthatmayeventuallyrevealhowtissuesandfullyfledgedorganismstakeshape.

Push and PullSomestrikingexamplesoftheimportanceofmechanicalforcesarecomingfromstudiesofthedevelopingembryoofthatworkhorseofthelab,thefruitflyDrosophila melanogaster.Unlikeplantcells,whichforthemostpartdon’tmove(thoughthereareexceptions),thecellsofdevelopingfliesandotheranimalsoftentravelastheyrealizetheirultimatefate.

“Duringdevelopment,cellsoftenmoveagreatdistancefromwheretheyareborntowheretheyneedtoendup,”saysZallen.“Thesecellshavetonavigatethroughcomplexmechanicalenvironments,andtheyareconstantlypushingandpullingoneachotherastheymove.”

ThemotorproteinmyosinII,bestknownforitsroleinmusclecontraction,hasemergedastheprimarymediatorofthispushingandpulling.SomeofmyosinII’sjobsincludehelpingcellsdivideandtravelastheycontributetothedevelopmentoftissue-levelstructuressuchasgroovesandtubes.

WhenZallenwasapostdocinthePrincetonUniversitylabofHHMIInvestigatorandNobelLaureateEricWieschausintheearly2000s,shewasinvestigatinghowturningoncertaingenesinaparticularspatialpatterncouldorientcellsastheDrosophilaembryoelongated.Here,myosinIIcameintoplay;duringelongation,theforce-generatingproteinaccumulatedatcellbordersalongthefly’shead-to-tailaxis.Continuingthisworkinherownlab,Zallenfoundthattheaccumulationandcontractionofmyosinwasdrivinga

Jennifer Zallen thinks that cells might use force as a compass, to help them move in the right direction.


coordinated,multicellularmovementthatledtoadramaticchangeintheembryo’sshape.

Zallenbegantowonderifcellsmightdetectandmakeuseoftheforcesgeneratedbytheirneighbors,inspiringthetug-of-warexperimentsandcementingtheroleofmechanicaltensioninregulatingmyosin’sactivity.Herteamshowedthatonceasinglecellstartstoconstrict,thatforcespreads:pullingonneighboringcellscausesthemtoconstrictaswell,producingacontractilecablethatextendsacrosscellularneighborslikealongrubberband.

“We’reveryinterestedinthepossibilitythatcellscouldusetheseforcesasacompasstohelpthemmoveintherightdirection,”saysZallen.“Thereisanincreasingappreciationthatforcescanactassignalsthatinfluencetheshape,fate,andbehaviorofcellstoenablethemtoassembleintotissuesduringdevelopment.”

Zallen’slabgrouphassinceshownthatastheDrosophilaembryoelongates,thecontractionoftheselongcablesofmyosin–andmyosin’spartnerincrime,thecytoskeletalproteinactin–drawscellsintounexpectedintermediatestructures,littleflower-likeconglomeratescalledrosettes.Theserosettesformwhencellsalignintocolumnsandshrinktheirconnectededgestogether,thankstothecontractingmyosincablethatpullsseveralcellsintocontactatasinglepoint.Therosettesthendisassembleinadirectionperpendiculartothewaytheyformed.Theprocesstransformsaclusterofcellsthatstartedoutastallandthinintoonethatisnowshortandwide,promotingelongationoftheflyembryo.

Nowtheresearchersarestartingtopinpointvariouscellularplayersthatguideandrespondtothejostlinginvolvedinembryoelongation.In2012,Zallenandherteamreportedthatrosettesdon’tformproperlyinembryosthataren’tabletomakethesignalingmoleculeAbl,anenzymethatenablescellstoadheretoeachotherundertensionandexecutegroupcellmovements.Herteamalsodiscoveredthecell-surfaceproteinsthatarelaiddowninboldstripesalongtheembryo’shead-to-tailaxisand

thatguidethedirectionofcellmovementstohelpmaketheembryolongerandthinner.Thispatterningdirectsmyosin’scontractilemachinery,orchestratingthemassmovementofelongation,reportedZallenandcolleaguesinNatureinNovember2014.

WieschauscomparesthepatterningthatprecedesforcegenerationinDrosophilatoimaginarydottedlinesonasheetoforigamipapershowingwherecreasesneedtogotofolditintoabird.“Onceyouhaveapattern,itgivesyouawaytolocalizeforces;thenyoucangetform,”hesays.Theideathattissueremodelingmightresultfromtheconcertedactionofassembliesofcellsratherthanfromindividualcellsisaviewofdevelopmentthat’sgainingtraction,headds.

“Webeganbythinkingoftheproblemasawholebunchofbricks.Ifyoucouldunderstandhoweachbrickbehaved,youcouldputthatalltogetherandsayhowitleadstoawholeorganism,”Wieschausexplains.“Butoverthepastcoupleofyears,we’vebecomeawarethatmaybeit’seasier–ormoreusefulormorecorrect–torealizethatchangesinmorphologyarebiggerthansinglecells.”

Mass MovementThere’sgrowingevidencethatnetworksofcontractilemyosincablesareaforcetobereckonedwithduringembryonicelongationinorganismsotherthanthefly.In2012,forexample,GerdWalzoftheUniversityHospitalFreiburginGermanyandHHMIEarlyCareerScientistJohnWallingfordoftheUniversityofTexasatAustinandcolleaguesshowedthatmyosincablesplayacrucialroleintheelongationofkidneytubulesintwomodelvertebrates,thefrogXenopus andmice.Otherresearch,byMasatoshiTakeichiandcolleaguesattheRIKENCenterforDevelopmentalBiologyinJapanandAnnSutherlandandcolleaguesattheUniversityofVirginia,revealedasimilarroleformyosincablesduringthedevelopmentofthechickandmouseneuraltube.

Recentworkalsoimplicatesphysicalforcesatworkinmetastaticcancer.ResearchfromValerieWeaver’slabattheUniversityofCalifornia,SanFrancisco,aswellasin

“During development, cells often move a great distance from where they are born to where they need to end up.”—jenniferzallen

Eric Wieschaus’s lab team studies how force influences the flow of groups of cells.

Kev

inW

olf

Eri

cR

eed

labselsewhere,hasfoundthatthephysicalstiffnessoftheextracellularmatrix,aweboffibersoutsidethecells,playsaprominentroleintheaggressivenessofbreastcancer.Enhancingthemechanicalstiffnessofthismatrixactivatesaproteinthataidsthetumor’sabilitytospread,Weaverandcolleaguesreportedin Cancer Researchlastyear.Thestiffness,whichputsphysicaltensiononepithelialcells,alsodrivesmalignancybydownregulatingthecells’productionofanimportanttumorsuppressorprotein,WeaverandhercolleaguesrecentlyreportedinNature Medicine.

Aviewofcellsassocialconstructs,whichrespondtomechanicalcuesenmasse,mightleadtowaystointerfere

withthosecueswhenthey’reimplicatedindisease,saysZallen.“Itmaybeusefultothinkaboutthemetastasisofcertaincancersasagroupactivity,”shesays.

Wieschaushastakenthisholisticapproachtotheextreme.Hislabhadbeeninvestigatingthedevelopmentoftheventralfurrow,aninwardfoldingofcellsthatcharacterizesthetransitionoftheDrosophilaembryofromasinglesheetofcellsintothreegermlayersthatultimatelydifferentiateintoadultorgansandtissues.Thisstageoccursrightafterthecellmembranesform,whenthefly-to-betransformsfromonegiantcell

withmanynucleitoanembryoof6,000cells.WorkingwithhisPrincetoncolleaguesOlegPolyakov,

KonstantinDoubrovinski,andBingHe,Wieschausdevelopedanapproachthatusestwo-photonimagingandfluorescentbeadstotracktheflowofcellsastheventralfurrowforms.Theresearchersdiscoveredthattheinfoldingresultsfrommechanicalforcesintheformofpulsingcontractionsdrivenbymyosinonthecellsurface.Mathematicalmodelingrevealedthatthecellswereflowingmuchlikeaviscousfluid–behaviorcapturedinthetidyStokesequationsofphysics,whichdescribetheflowofsuchfluidsaspaintandlava.Thisraisedthequestionofhowventralfurrowdevelopmentwouldproceediftheembryoweren’tpartitionedintotheindividualunitswecallcells.

Toinvestigatetheforcesintheabsenceofcells,Wieschaus’steamknockedouttwogenes– slamanddnk –thatdirectthedevelopmentofthecellmembranes.Remarkably,theyfoundthat,duringventralfurrowformation,afruitflyembryowithoutcellmembranesbehavedverysimilarlytoonewithcellmembranes.AstheteamreportedinNature inApril2014,theprocessproceededinamessierandslowerfashionthanitdidinanembryowithcellmembranes,but

theflowpatternswereessentiallythesame,upendingthecell-focusedview.

“Ascellbiologists,webelievethatcellmembranesarereallyimportantforeverything,”Wieschaussays.“Butwefoundthatwecouldeliminateallthepartitioning,havethisbiggooofcytoplasmflowinglikeafluid,andyettheembryogoesonanddoesitsstuff.”

Asimplemodelofmechanicalforcesviamyosinconstrictioncouldaccountforthechangesinshapethatleadtoventralfurrowformation,Wieschaussays.Similarsqueezingatacell’sapicalendoccursduringfoldinginotherplacesandotherembryos,includingduringthedevelopmentoftheDrosophilarespiratorysystemandtheclosureoftheXenopus neuraltube.Suchcommonalitiessuggestthattransmittingforceviaviscousflowmightbeanotherfundamentalmechanismforestablishingform.

Pattern and FormMechanicalforcesmayalsoexplainpatternsobservablewiththenakedeye,includinganenduringmysteryofplantarchitecture.Whilemanyanimalsundergoamassiverearrangementofcellstoformanembryothatthen,inessence,simplyenlarges,plants’growthisoftenindeterminate–thatis,theycanaddnewleaves,branches,andflowersuntildeath.Thisnewgrowthisn’thaphazard;itfollowsaconspicuouslyregularpatternthat’sobservablebylooking,forexample,head-onatthetipofanewshoot.Itturnsoutthatmechanicalforcesgeneratedinthehotbedofplantembryonicgrowth–theshootapicalmeristem,aconcentratedregionofdividingcells–arecrucialindrivingthispredictablegeometricmorphology.

EversincetheancientGreekscholarTheophrastusnotedthisstrikingregularityinthearrangementofleavesaroundaplantstem,botanists,physicists,andmathematicianshavebeentryingtoexplainhowsuchsystematicplacement,calledphyllotaxis,arises.(Moremodern-dayscholarswhohavestudiedthephenomenonincludeD’ArcyThompson,whonotedthecrisscrossingpatternsthatformspiralsinpineconesandsunflowerheads.)ItwasphyllotaxisthatturnedtheattentionofElliotMeyerowitz,anHHMIinvestigatorattheCaliforniaInstituteofTechnology(Caltech),towardtheroleofmechanicalforcesingeneratingform.

Usingliveimagingtechniques,Meyerowitzandhiscolleagueshadbeeninvestigatinghowdifferingconcentrationsoftheplanthormoneauxinrelatedtopatternsofplantgrowth.Ithadlongbeenknownthatauxinwascrucialindeterminingwhereeachnewflowerorleafappearedonaplant;experimentsdatingbacktothe1930sdemonstratedthatdaubingapasteofauxinontothemeristempromptedthegrowthofnewplantorgans.SoMeyerowitz–withateamthatincludedhisthenCaltechcolleaguesBruceShapiroandMarcusHeisler,plusHenrikJönsson,thenofLundUniversityinSweden,andEricMjolsnessoftheUniversityofCalifornia,Irvine–begantrackingtheconcentrationofauxininmeristemcellsofthemodelplantArabidopsis.

PreviousworkinseverallaboratorieshadsuggestedthatthemembraneproteinknownasPIN1wasprobablyanauxinpump,controllingthedirectionofthehormone’sflowoutof

Elliot Meyerowitz investigates the impact of mechanical stress on growth in plants.


To see forces at work in plants and animals, go to hhmi.org/bulletin/spring-2015.

cellsintheplantshoot’smeristem.ButPIN1’slocationisn’tfixedincellmembranes:thepumpingproteincanmovearound,andsomethingwasdirectingittomembraneregionsthatwereadjacenttocellsalreadyhighinauxin.Meyerowitzandhiscolleaguessuspectedthatauxin’sinfluenceonthelocationofthePIN1pumpcouldleadtoaparticularspatialpatternofhighauxinconcentration,andthatthismightaccountfortheregularityofphyllotacticspiralsandwhorls.IfcellssomehowsensedtheirownhighauxinandrecruitedPIN1tothenearestmembraneregionsinadjacentcells,theauxinconcentrationwouldincreaseevenmoreintheoriginalcell.Thiswouldinducegrowthofanewleaforflower.Butasauxinconcentrationsincreasedlocally,neighboringcellswouldbecomedepletedinauxin,leavingaspotwithnoleaforflower.Cellsfartherawayfromthisauxin-depletedsitewould,bycomparison,havemoreauxinandthuswouldrecruitPIN1toattractmoreauxinandagaininduceanewleaforflower.

Theresearcherscreatedamathematicalmodelthatincorporatedthisproposedfeedbackmechanism.Whentheyrancomputersimulationsofthemodel,auxinpeaksemergedatregulardistances,capturingtheregularpatterningofphyllotaxis–afindingpublishedinProceedings of the National Academy of Sciencesin2006.

Butmysteriesremained,includinghowtheplantcellsweresensinglocalauxinconcentrations.Thehormonewasknowntoweakenplantcellwalls,leadingtheteamtowonderwhethermechanicalstressmightsignaltocellsthattheywereadjacenttoneighboringcellshighinauxin.This,inturn,wouldrecruitPIN1totheregionofthecellmembranenearestthehigh-auxinneighbor.

Inaseriesofelegantexperiments,includingonesinwhichthescientistsweakenedorobliteratedparticularplantcellwallswithalaser,mechanicalstressindeedemergedasthemediatorofPIN1’sdynamicbehavior.TheyfoundthatasPIN1directedtheflowofauxinfromanareaoflowtohighauxinconcentration,thecellshighestinauxinexpanded.Inplants,cellwallsareshared.Soitappearedthatmechanicalstressonacommonwallalertedthecellularneighborthatauxinconcentrationwashighnearby,thusbringinginPIN1.

“Basically,alltheobservationswereintheliterature,butnoonehadthoughttoputstressintotheequation,”saysMeyerowitz,whopublishedthefindingswithMarcusHeislerandothercolleaguesinPLOS Biology in2010.

Morerecently,Meyerowitzandhiscollaboratorshaveshownthatmechanicalstressplaysanimportantroleinshapingplantcellsbeyondthemeristem.Itturnsoutthatthe

puzzle-shapedpiecesoftheso-calledpavementcellsonaleaf’ssurfacecreateintracellularstressesthatcausereorganizationofthecytoskeletalproteinscalledmicrotubules.Theseproteinsthenhelpdialuptheproductionofcellulose,which,inturn,reinforcesthecellwallsagainstthestress,theteamreportedineLifeinApril2014.Combiningthemicrotubulefindingswiththeauxin-relatedresearchyieldsasimplemodeloffeedbackdrivenbyphysicalforces:“Mechanicalstresstellscellshowtogrow,andcellgrowthcreatesmechanicalstress–andmorphology,”Meyerowitzandhiscolleagueswroteina2014 Current Biology reviewpaper.

Otherlabsarefindingevidenceofphysicalstressorsaswell.Forexample,AudreyCreffandGwynethIngramoftheLaboratoiredeReproductionetDéveloppementdesPlantesinLyon,France,recentlyshowedthatamechanicallysensitivelayerofcellsintheseedcoatofArabidopsisrespondstostressexertedbytheseed’snutritivetissue,theendosperm,resultinginafine-tuningofseedsize.

TheMeyerowitzlabisnowlookingforcellularstresssensorsresponsibleforthecellwalleffectsonPIN1andforthedifferentsensorsthatmediatetheeffectsofstressonthecytoskeleton.Hebelievesthatagreaterunderstandingoftherelationshipbetweenmechanicalforceandplantgrowthisimportantnotjustforelucidatingaplant’scurrentgrowthpatterns;theresearchmayalsoleadtotechniquesforengineeringsuperiorfoodcrops–forexample,producewithmodifiedleafarrangementsthatmaximizephotosynthesisinaparticulargrowingregion.

“Nowthatwe’rebeginningtounderstandthefeedbackbetweenphysicalstressandgrowth,itmaygiveusanewwaytointervene–oratleastpredictwhatwouldhappenifwechangethings,”Meyerowitzsays.Andfordevelopmentalbiologyingeneral,bringingmechanicsbacktotheforemayhelpresolveoldermysteriesofshapeandform.

“Aspeoplestarttolookatthingsintermsofphysicalsignaling,notjustchemicalsignaling,itmayexplainquiteabit,”heobserves.“Itseemslikewecanmakesomerapidprogressinsolvingoldproblems.”

A view of cells as social constructs, which respond to mechanical cues en masse, might lead to ways to interfere with those cues when they’re implicated in disease.


18

All the World’s a

Stage

Marta Zlatic is as capable of

portraying the Greek heroine

Hecuba as she is of pulling back

the curtain on the neural circuitry

of fruit fly larvae. At both pursuits,

she’s won wide acclaim.

bysarahgoforth

19

photographybyelimeirkaplan


Iit’s hardto imaginetwosettingsmoredissimilarthanthevastrotundaofaGreektheaterandasmall,windowlessmicroscopyroomatHHMI’sJaneliaResearchCampus.ButGroupLeaderMartaZlaticisequallyathomeinbothspaces.Atthismoment,hermicroscope’snarrowbeamisilluminatingaDrosophilafruitflylarvaonitsownstage,thesemitransparentcreaturewrigglinginawaybarelyperceptibletothenakedeye.ToZlatic,thelarva’smovementsrepresentanuancedperformancefine-tunedbymillionsofyearsofevolution–andafoundationforunderstandingtheneuralbasisofbehavior.

Fifteenyearsago,whileshewasanundergraduateattheUniversityofCambridgeinEngland,Zlaticherselfwasinaspotlightonstage–asHecuba,theill-fatedqueenintheEuripidestragedy,The Trojan Women.Astudentoflinguisticsandneuroscience,ZlaticknewancientGreekandhadattendedtheauditionfortheplayoutofcuriosity.Lackingactingexperience,shewasassignedaroleinthechorus.Whentheleadactressquithalfwaythroughrehearsals,however,thedirector,impressedbyZlatic’sfluencywiththelanguageandtext,askedhertostepin.Audiencesandcriticscheeredtheperformance,andZlaticwentontolandastringofrolesknownfortheircomplexity:Electra,Medea,LadyMacbeth,evenOedipusRex.

“Anyonewhohasseenaplayshouldwonder:Howcanthiswonderfulsystem,thebrain,producesomanydifferentemotionsandbehaviors?”Zlaticsaystoday,seatedinhersparselydecoratedofficeupstairsfromthemicroscoperoominJanelia’seastwing.Threeemptychampagnebottles,oneforeachmajorpaperpublishedbyherlabinitsfiveyearsofexistence,sitnexttohercomputer.Thebottles,alongwiththehumblemodelorganismshehaschosenasherlife’swork,makeaquietstatementaboutthepatiencerequiredofscientistswhostudysuchambitiousquestions.Togobig,youmustoftenstartsmall.

Zlaticleansforward.Thehumanbrain,sheexplains,has100billion neurons,roughlyequivalenttothenumberofstarsintheMilkyWay.Fruitflieshaveonly100,000,andafruitflylarvajust10,000.Reduceanervoussystemtothatrelativesimplicity,andsuddenlyitbecomespossibletostudyitcomprehensively–notonlytodescribewhatitlookslike(mapitvisually),butalsotodeterminewhichcellsconnecttooneanother(mapitswiring,or“connectome”)andtofigureouthowthosenetworkslistenandrespondtosignalsfromtheoutsideworld(mapitsbehavior).Zlatic’steam,withaconstellationofcollaborators

atJaneliaandaroundtheworld,isbuildingsomethingakintoGoogleMapsforthelarvalbrain,placingtheselayersofinformationontopofeachotherlikesatelliteimagesoverlaidonroadways.Theyarenowbeginningtooutline,withneuron-by-neuronprecision,theneuralcircuitsthatdeterminehowalarvarespondstoitsenvironmentanddecideswhatactionstotake.

Itisnofrivolousexercise.“Wecanlookatthismuchsimplersystem,anditssmallerrangeofbehaviors,andseethatmanyofthethingswelearnarethesameacrossspecies,”Zlaticexplains.Forexample,Drosophila larvaehavethesamesensorymodalitiesasmostanimals–sound,light,odor,touch,andheat.Furthermore,they’reabletoprocessthisinformation,rememberit,andmakedecisionsbasedonit.Howdoesanervoussystemmakebehavioralchoicesbasedonpreviousexperienceandamultitudeofsensoryinputs?Whatcanunderstandingthissimplesystemtellusaboutthehumanbrainandourcapacityforlanguage,art,andthepursuitofknowledge?

Zlaticwantstoknowtheanswerstothosequestions.Byallaccounts,herteamisalreadytakingstepstowardconstructingamodelforabrainatlas–onegoalofthemultimillion-dollarBRAIN(BrainResearchthroughAdvancingInnovativeNeurotechnologies)InitiativethatPresidentBarackObamaannouncedin2013.

“Wearetryingtounderstandwhat’shappeninginthebrainduringsensingandbehaving,”saysTihanaJovanic,apostdoctoralresearcherinZlatic’slab.“Gainingabetterunderstandingofhowthebrainworkswillhelpuncoverinsightsintohowdifferentneuralprocessesareintegratedsoanorganismcanfunctioninanever-changingenvironment.”

A Renaissance MindZlaticwasborninCroatia,theonlychildofatheoreticalphysicistandaphilosophymajorwhoworkedattheMinistryofCulture.Herparentsweregregariousandecumenicalintheirthinking,andshegrewupimmersedinlivelyconversationaboutartandpolitics,ofteninmixedlanguages.Throughoutheryouth,thefamilywasaslikelytobetravelingabroad–herfather’sresearchcareertookthefamilytoGermany,England,andCaliforniaformonthsatatime–astobeathomeinZagreb.

“IgrewupwiththeideathatIwantedtodobiology,becauseIwasfascinatedbylivingbeings–especiallyanimalsandtheirbehavior,”Zlaticreflects.“Ialsolovedliteratureandart.Ialwayssawthesethingsascomplementary.”Shehadboththetalentandtheopportunitytopickuplanguages,addingEnglish,French,andGermantoCroatian.Byage12,shewasalsostudyingLatinandancientGreekinschool.Atage30,shespokesixlanguages,inadditiontobeingliterateinLatinandancientGreek.

“Ilovedgrammarandtherulesoflanguage,”saysZlatic,“butIwasalsointerestedinbiologybecauseIwantedtounderstandhowthebrainwascapableofcreatingandlearningthesesystems.”Inhighschool,sheheld“thenaïveview”thatshecouldsomedayunderstandindetailhowthebraincreateslanguage,andshesetthatasagoalforherself.ShestudiedlinguisticsandRussianattheUniversityofZagrebinCroatiabeforeearningafullscholarshiptoTrinityCollegeattheUniversityofCambridge,whereshepursuedneuroscience.Everysummer,shewouldreturntoZagrebtocontinueherlanguagestudies.

“Tome,itwaslikeavacation,toreturntoZagrebeverysummer,”saysZlatic.“Itdidn’toccurtomethatIwasdoingsomethingunusual.”

Inherlastyearasanundergraduate,ZlaticwasinspiredbythelecturesofCambridgeneuroscientistMichaelBate,whowasworkingtounderstandhowneuralcircuitsforminDrosophilaembryos.Shefoundirresistibletheideathatonecouldwatchhowanervoussystemcomes

21HHMI Bulletin / Spring 2015C

ard

ona

lab

Shewasnolesstalentedinthelab.WithHuppertandtwootherfriends,shewonfundinginaninnovationcompetitiontoformabiotechstart-upaimedatfillinganeedapparentfromherresearch–fora“virgincollector,”away,thatis,tomoreeasilycollectvirginfemalefruitflies.Thecompanydevelopedasyntheticpheromonethat,intheory,wouldpreventtheinsectsfrommating.Thecompoundsuppressedmatingmostofthetime,butnotallthetime,andthecompanyfailed.Zlatictookthebumpinstride,saysHuppert.“Shewastryingtofillaneed,tosolveaproblem,andsheknewthatwhatshelearnedwouldmakeherbetteratsolvingotherproblems,”hesays.

AboutthetimethatZlaticandHuppertwereestablishingtheircompany,developmentalneuroscientistJimTruman,thenattheUniversityofWashington,visitedtheBatelabonsabbatical.Trumanwasstudyinghowstemcellsdevelopintodifferentkindsofneuronsinthelarvalbrain–workthatwouldlaterprovideafoundationforZlatic’sbrain-behaviormapping.Zlatic,Trumansays,stoodout:“Shewasfullofpointedquestions,tirelessinherresearchandallofheractivities.Irememberthinkingitwasexhaustingjustwatchingher.”

In2006,JulieSimpson,oneofJanelia’sfirstgroupleaders,invitedZlatictojoinherlabaspartofthecampus’svisitingscientistprogram.SimpsonwasusingthegenetictoolsdevelopedbyJaneliaExecutiveDirectorGerryRubinandotherstostudytheneuralandgeneticbasisofadultfruitflybehavior.Zlatic–whobythenwascompletingavisitingpostdoctoralappointmentatColumbiaUniversityonaprestigiousTrinityCollegefellowship–joinedSimpson’sgroupandbeganpiecingtogetherthefirstlarvalbehaviormap.“IcameherewiththeideaoftryingtomapcircuitsandfindthefunctionandbehavioroftheneuronsIusedtostudy,”shesays.“IquicklyrealizedthiswastheplacetobeforthekindsofquestionsIwasinterestedin–therelationshipbetweenstructureandfunctionofthenervoussystem.Atthattime,nooneelseatJaneliawasworkingonthelarva,butJuliejustletmeplaywithlarvaeanyway.”

Ayearlater,sheearnedalabheadpositionatJaneliaandstartedherowngroup.

“Some people play sports, other people study. Marta studies. If she is interested in something, usually she is intensely interested in it.”—albertcardona

intobeing,couldactuallywitnessneuronsfindingtheirpartnercellsandformingconnections,knownassynapses.ShesecuredaPhD-trackpositioninBate’slabtostudytheearlieststepsinthatprocess–herideabeingthatbystudyinghowthenervoussystemdevelops,shewouldgaininsightsintoitsfunctionand,ultimately,intothebiologicalrootsofbehavior.

Zlaticfocusedherattentiononsensoryneuronsinthelarvae,studyinghowtheyformtheextensions,calledaxonsanddendrites,thatallowthemtocommunicatewithotherneurons.Onceinplace,anaxonbranchesatitstipandconnectswithanotherneuron’sdendrite,formingpartofacircuit.Neuronsresponsiblefordifferentsenses,suchasthedetectionofheatortouch,sendaxonstotheregionofthefly’snervoussystemspecifictothatsensation.Zlaticwantedtounderstandwhatwasguidingtheirchoiceoftarget.Weretheaxonsbeingguidedbypositionalcuestoterminateataparticularlocation–aspecific“address”inthenervoussystem–andjustconnectingwithwhatevercelltheyfoundatthatlocation?Orweretheyseekingaparticularpartner,irrespectiveoftheaddress?“Theanswerisposition,becausealteringaxons’abilitytosensepositionalcuesresultedintheirovershootingtheaddresswheretheirpartnersarenormallyfound.Theydidnotappeartocarewheretheirpartnerswere,”shesays.“Butwhatwecouldn’ttellwaswhethertheirpartnerscaredaboutthem,”sheadds,becausenoonehadpiecedtogetheraneuralcircuitfortheDrosophilanervecord.

OutsideBate’slab,sheremainedanavidthespian.Onstage,Zlaticwas“theabsolutestar,”remembersherfriendJulianHuppert,thenafellowPhDstudentatCambridgeandnowamemberoftheBritishParliament.“Shewasjustspellbindinglygood.”

This digital “map” of a fruit fly larva neural circuit highlights a neuron (yellow) that’s able to sense vibrations.


understandatthelevelofsingleneurons;andthecausalrelationshipbetweenneuronsandbehavior.”

Meanwhile,Cardonawasformulatinghisowndreamproject–tolayoutthefruitflylarva“connectome,”amapofneuralconnectivityinthebrain.ThenhegotajobinZurich.Butbythattimetheirrelationshipwasfarfromjustameetingofscientificminds.Theymanagedtonotonlystayintouchfromafar,butalsotogettogetherperiodically,witheachworkingasavisitingscientistattheother’sinstitute,untilCardonawashiredasagroupleaderatJaneliain2011.Theygotmarried–andalsocementedtheirresearchcollaboration.“Somepeopleplaysports,otherpeoplestudy,”saysCardona.“Martastudies.Ifsheisinterestedinsomething,usuallysheisintenselyinterestedinit.”

And,notestheircolleagueJimTruman,“Albert’sapproachtoscienceisverydifferentfromMarta’s,inacomplementaryway.Withouthisabilitiestoreconstructthenervoussystemofthelarva,basically,wewouldbeshootinginthedark.”

A Larva’s LifeSpendmuchtimewithZlaticoranyoneonherteamtoday,andyoumayneverlookatamaggotthesameway.Underthemicroscope,itswrigglesbecomelessrandom,evenpurposeful:Outintheworld,larvaewillsensiblytrytoapproachfoodandpleasantaromas,notesZlatic–butatthesametime,theyhavetoavoidarangeofthreateningsituations.Therearemanywaystoescapefromdanger,dependingonthecontextandtheanimal’spreviousexperience.Inthesamewaythatsomemammalscanwalk,trot,orgallopaway,larvaehavedifferentmeansofescape.Someofthesemeansarefastbutcostlyintermsofenergyexpenditure;othersmaybeslowerbutlesscostly.Often,theyneedtostringtogethersequencesofbehavior.

Zlaticwantedtounderstandhowtheanimal’snervoussystemchoosesamongthesevariousbehaviors.Thisisaproblemfacedbyallnervoussystems,butintheDrosophilalarvaitisparticularlytractable.Herteamusedagenetictoolkit,developedbyRubin,toactivatedifferentsetsofneuronsandwatchwhatthelarvaedidinresponse,likeflippingswitchesinacircuitbreakertotestthewiringbehindthewalls.

Rubin’slabhaddevelopedmorethanathousandmutantflystrainswithagenecalled GAL4 insertedintospecificclustersofneurons.Zlatic’steamcrossedthosestrainswithotherflystrainsthathadbeenengineeredtoproduce,inthepresenceofGAL4,alight-sensitiveproteincalledchannelrhodopsin.ThisgaveZlaticaccesstotheswitchboard.Byshininglightonlarvaefromeachhybridstrain,shecouldselectivelyactivatewhateverneuronshadbeenengineeredtoincludeGAL4.

ThensheandTomokoOhyama,aresearchspecialistinherlab,startedflippingswitches,videotapingthegroupsoflarvaeastheyperformed.Rightaway,theygleanedsomesurprisinginsights,saysOhyama,becausetheysawvariationbetweenindividuallarvaewhenthesameneuronswerestimulated.Thatindicatedtothescientiststhateventhemostdiscretebehaviorsareprobabilisticandnot100-percentpredictable.Still,clearpatternsbegantoemerge.ZlaticandOhyamaenlistedthehelpofateamofmachinelearningspecialistsandstatisticiansintheJohnsHopkinsUniversitylabofCareyPriebe,whodevelopedanalgorithmtoclassifythebehaviorswithgreaterprecisionthanthehumaneyecoulddiscern.Inthespringof2014,thescientistspublishedtheirwork,widelydeemedanewframeworkformappingbehaviortoindividualclassesofneurons,inthejournalScience.Inall,theyhadanalyzedthemovementsofnearly40,000larvae.

Zlaticcallsthisachievement“justthebeginning,”asthepaperdescribesonlywhichneuronsevokeagivenbehaviorandnotthecircuitmechanismsbywhichitoccurs.Thenextstepistorelatethisknowledgetotheconnectomeandaneuralactivitymap.“Martahasbeenasparkplugtokeepthisgoing;again,itisherenthusiasmthatreallymakeslarvalsystemsworkinthiscohesiveway,”saysTruman.

A Marriage of MindsBythattime,JaneliahadbecomeabustlinghubforthestudyoftheDrosophilabrain.AlbertCardona,ayoungSpanishdevelopmentalbiologistandneuroscientistwithawidesmileandenthusiasticmanner,cametothecampusforJanelia’sfirstflybrainconference,organizedbySimpson.

“ThewholefieldwasexcitedabouttheflylinescomingoutofGerryRubin’slab,”heremembers,“andtherewasanenergyaboutwhatthesegenetictoolswouldmakepossible.ButwhatIrememberaboutthatmeetingisonlyMarta.”Twoencountersstandout:Inone,ZlaticwasseatedbyherselfinJanelia’sdarkenedauditoriumbeforeatalk,andshepattedtheseatnexttoher.“Shesaid,‘Sithere,’andIdid,”Cardonalaughs.Intheother,hehadjoinedherinJanelia’spub,Bob’s,todiscussdatacomingoutofherlab.“Irememberthatasweweresittingtheretogether,thisguywhowasabig-shotconferenceattendeewastherealso,talkingwithus.Irememberjustthinking,Go away.”

CardonajoinedSimpson’slabasavisitingscientistonapilotprojectthatallowedhimtotakehundredsofelectronmicroscopeimagesofasegmentoftheflylarvalnervecord.Thiswasproofofconceptthatitwaspossibletoreconstructwhataneuralcircuitlookslikeinaninsectnervoussystematthelevelofsingleneuronsandtheirsynapticconnections.Janelialatercommittedtocompletingthiswiringdiagramforthefulllarva,underCardona’sleadership.

JimTruman,whohadcometoJaneliain2007,wasbythenusinglightmicroscopytoimageRubin’sflystrains,eachofwhichhadbeenengineeredtoallowresearcherstostudysmallsetsofneurons.TheresultofbotheffortscombinedwasthatZlaticcouldnowmanipulateindividualneuronsandstudytheireffectonbehavior.Herambitiousaimwastostudynotonlyhowneuronsaffectedbehavior,butalsohowtheyworktogethertointegratemanykindsofinformation–pain,temperaturechanges,andodors,forexample–andthenselectabehaviorintheanimal’sbestinterest.

“Tounderstandthis,”sheexplains,“weneedtoknowthreethings:circuitstructure;physiologicalpropertiesofneurons,whichwecan

The label “MET” on the fruit fly specimens below refers to methoprene-tolerant protein, which controls metamorphosis in the developing flies.

23HHMI Bulletin / Spring 2015E

liM

eir

Kap

lan

Itisalsoafeatthatfewscientistscanpulloff,giventhatitrequiresmarryingdisciplinesasdiverseasneurobiology,electronmicroscopy,andmachinelearning,saysRubin.“Alotofsmartpeopletalkthemselvesoutofdoingthingsbecausetheycan’tdoitallthemselvesorbeintotalcontrol,”hesays.“ThatisnotMarta’sstyle.Shesays,‘IhavethisbigproblemIwanttosolve,andIknowIcan’tdoitall.Ineedtogetthispiecefromthatperson,andthisotherpiecefromanotherperson,andtogetherwe’llmakeitwork.’That’saskillandatalent–likebeingagoodfieldgeneral.It’sanapproachtoscienceIadmire.”

Inthemeantime,ZlaticandCardonahavealsoembarkedonwhatshecallsthe“ultimateactofcreativity”–parenthood.Theirson,David,nowthree,ispoisedtomatchhismother’slinguisticabilities.

“MartaspeaksCroatiantoDavid,andhespeaksCatalanorEnglishtome.IunderstandabitofCroatian,though,andwhenIspeakittoDavidhecorrectsme,”laughsCardona,anativeofSpain’sCataloniaregion.Thedemandsoftwoactivelaboratoriesplusatoddlerkeepthefamilybusy,buttheymakeitwork;Zlatictakesthe“morningshift,”wakingupwithDavidandmakinghimbreakfast,whileCardonaheadstothecampusatsunrise.Sheworkslatewhilehemakesdinner,andthenthefamilyhasanhourortwotogetherbeforebed.“Itworksforus,”saysCardona,“becauseweknowwhatourprioritiesare.”

AndsoZlatichasgivenupactinginfavorofhergreaterloves–scienceandfamily.“WhatIlovemostaboutscienceisideas,”shesays.“Youlookatthedata,itinspiresideas,thenyoutestthem,thenyoulosethem,thenyoutryoutotherones.Thecreativeaspectofscienceiscomingupwithideasindialoguewiththepast.That’sverybeautifultome.”

“There’ssomethingaboutherthatIthinkistrueofalotofthereallygoodscientistsI’veseeninmycareer,”saysRubin.“They’rehappy.They’rehavingfun.Theysaythattobeasuccessfulscientistyouhavetomaintainyourchildlikecuriosity.Sheexudesthat.”

Marta Zlatic manipulates neurons in fruit fly larvae to study their effect on behavior.

“There’s something about her that I think is true of a lot of the really good scientists I’ve seen in my career. They’re happy. They’re having fun.”—gerryrubin

Navigating the Torrent

Experimentalneuroscientists andtheoristsatJaneliaarejoining

forcestomakesenseof–andimprove– thedelugeofdatacomingoutoflabs.

by carrie arnold

illustrationbycarmensegovia


NJeremy Freeman built an open-source computer platform to help researchers analyze and share massive data sets.

neuroscienceis drowning indata.Sincethe1950s,thenumberofneuronsthatscientistscanrecordsimultaneouslyhasgrownatanexponentialpace,doublingroughlyeverysevenyears.Toutilizethisinformationaboutthebillionsofneuronsthatspitandsputterandmakeus,well,human,researchershavehadtocopewithanexponentialgrowthindata.

“Whatyouwoulddo,backintheday,ismaybelookatacoupleofneuronsinonepartofthebrainduringsimplesensorystimulation–averyfocusedstudy,”saysJeremyFreeman,aneuroscientistandgroupleaderatHHMI’sJaneliaResearchCampus.

Now,thependulumhasswungintheoppositedirection.Neuroscientistscanrecordtheactivityofnearlyalltheneuronsinthebrainsofzebrafishlarvae,andineverincreasingportionsofmouseandDrosophilafruitflybrains.

Asinglesetofexperimentscangenerateterabytesofinformation.Simultaneousincreasesincomputingpowermeanresearcherscanperformmoresophisticatedanalyses,studyingrelationshipsbetweengroupsofneuronsinsteadofanalyzingoneneuronatatime.Tostayafloatinthedelugeofdata,scientistsneedtodevelopanentirelynewwayofthinkingaboutexperiments–andmakingsenseoftheresultingtorrentofinformation.

“It’sreallyabigchangefromthinkingaboutwhatsingleneuronsdoto

thinkingaboutwhatlargepopulationsofneuronsdo.Withasingleneuron,anexperimentalistcouldusehisorherintuitionand,infact,peoplearequitegoodatthat,”saysLarryAbbott,aJaneliaseniorfellowandatheoreticalneuroscientistatColumbiaUniversity.“Whenyouhaveapopulationofneurons,andthey’reallinteracting,it’salmostimpossibletohavethatintuition.Youreallyhavetomakeamodelofitandfigureouthowyouthinkit’sgoingtobehave.”

AbbottandFreeman,togetherwithothertheoreticalandcomputationalneuroscientistsatJanelia,are

workingtobuildlifeboatsandlighthousesforotherscientiststohelpthemnavigatetheswirlingstormsofdata.Buriedinthistsunamiofstatisticsarethepatternsandinsightsthatwillenablethemtocrackthebiggestmysteryinscience:howthehumanbrainworks.

Computation as PartnerToJaneliaGroupLeaderKristinBranson,keepingyourheadabovewaterasdatapoursinrequirescomputerscienceasmuchasitdoesbiology.ShejoinedJaneliain2010,intendingtoimprovethecomputertrackingsoftwareknownasCtraxthatshehadbuiltwhileshewasapostdocattheCaliforniaInstituteofTechnology.ThepremisebehindCtraxwassimple.Atthetime,measuringtheeffectsofneuralactivityonbehaviormeantmeasuringhowafly’sbehaviorchangedafteragroupofneuronswasswitchedonoroff.Tomakesenseofthisbehavior,ascientistwouldhavetodeterminewhattheflywasdoingineachframeofvideo–animpossibletaskwhenasingleexperimentcanyielddaysofvideo.

EnterCtrax.Usingavarietyofcomputeralgorithmsthatallowamachinetoprocessandanalyzeimages,Ctraxenablesresearcherstotrackindividualfliesevenwhentheycongregateinlargegroups.Unlikeotherprogramsavailabletobiologists,Ctraxdoesn’trequireuserstoknowhowtocode.Instead,BransoncreatedaGUI(pronounced“gooey”andstandingfor“graphicaluserinterface”)thatallowsevennon-coderstousetheprogram.

WhenshefirstarrivedatJanelia,BransonhadideasforimprovingCtrax.ButthenumberofotherJaneliascientistsalsoworkingontrackingsoftwaregaveherpause,asdidherrealizationthatthetougherproblemwouldbeanalyzingthefly’sbehavior,notjusttrackingit.SoBransondevelopedJAABA,theJaneliaAutomaticAnimalBehaviorAnnotator,whichisfreelyavailabletoallresearchersthroughBranson’swebsite.Researcherscan“teach”JAABAtherelevantbehaviorstorecognizeandrecord,whichallowsthemtobegintofigureoutwhathappenstotheanimalwhenneuralactivityisaltered.

“Theideaisyouwanttoautomaticallybeabletosayforeveryframeandforeveryfly,Isthisflydoingthiscertainbehaviorornot?”Bransonexplains.“Isthisflychasinganotherfly?Isitwalking?Isitturning?”

BransonhasbegunusingCtraxandJAABAtoscreenalloftheneuronsinthefruitflybrain,tolinkspecificgroupsofcellstobehaviors.Tobegin,shetookadvantageofthe10,000linesoffruitflies,createdbyJaneliaExecutiveDirectorGerryRubinandhislabgroup,thatexpresstheproteinGAL4,whichscientistsusetoaffectgeneexpressionindifferentcells.ScientistscanselectthefliesthatexpressGAL4inspecificsetsofneuronsandthenusethefluorescentmarkerGFPtovisualizetheseneuronswithamicroscope.Bransoncrossed2,000differentlinesofGAL4flieswithfliescontainingTrpA,atemperature-sensinggene,whichallowshertoactivatetheseneuronsbyraisingthefly’sbodytemperatureafewdegrees.Placingthesefliesinanenclosedarena,BransontracksthemwithCtraxandrecordstheirbehaviorwith M

attS

tale

y


Kristin Branson uses computation and machine learning to understand behavior.

JAABA.Usedtogether,thesystemsgiveherawaytolinktheactivityofspecificneuronsinflieswithmeasurablebehaviordifferences,suchaswalkingspeed.

Machinelearningsystemshavebenefitsbeyondjustmanagingtidalwavesofdata;Bransonhasalsofoundthatthesesystemscanrecognizeeffectsthatmightgounnoticedbyhumans.

“Itmightbeasubtlebehavioralchangethat’shappening,butyou’reobservingitlotsandlotsoftimes.Soifyouanalyzeabigenoughdataset,it’sgoingtocomeout,”shesays.“Maybeit’snotahugephenotypethatyou’reseeing–it’snotsomethingthatahumanwouldtypicallynotice–butifyoucompoundthiswithhowmanytimesyou’veseenit,it’ssomethingthatcan’tbeignored.”

Massive ComputationCtraxandJAABAmaybeabletohelpneuroscientistsidentifyandcatalogrelevantbehaviors,butresearchersalsohaveothertypesofdatatoanalyze.Theproblem,Freemanpointsout,isthatsomeofthesedatasetsarefartoobigforasinglecomputertohandle.Soresearchersareturningtoclustercomputing,usingmultiplecomputersworkingtogethertomakesenseoftheirfindings.

Historically,Freemansays,scientistshaveworkedtosolvethesetypesofproblemsonalab-by-labbasis.“Individuallabsbuildtheirownlittlelocalsolutionstosolveproblemsthataretotallycustomizedtotheirlabsandaremeanttoworkonasinglemachine,”hesays.“We’reat

apointwhereanindividuallabisstartingtoreachalimitofwhatitcandoinareasonableamountoftime.”

Thisprocessofcontinuallyreinventingthewheeldidn’tseemproductive,nordiditallowscientiststoeasilysharetheirdatawitheachother.SoFreeman,incollaborationwithothersatJaneliaandelsewhere,builtanopen-sourcecomputerplatformthatwouldallowresearcherstoanalyzeandsharemassivedatasets.HeutilizedtheApacheSparkplatform–anopen-sourceframeworkabletoprocesslargedatasetsoncomputerclusters–tocreateThunder,alibrarydesignedspecificallytoanalyzeneuraldata.DetailedinNature MethodsinSeptember2014,thelibraryisfreelyavailableforresearcherstouseandcontributeto.FreemanestimatesthatThunderisnowbeingusedbysome10to20labsaroundtheworld.He’susingittolookatlargedatasetstogainamoreholisticunderstandingofneuralfunction.

Inoneofhiscollaborations,hehasbeenworkingwithfellowJaneliaGroupLeaderMishaAhrenstoimagetheentirebrainofthezebrafishusingtwo-photonmicroscopy,whichcanrecordfluorescingcellsatadepthofuptoonemillimeter.FreemanandAhrensdisplayavisualpatterninfrontofthefish.Theywanttoknowwhichneuronsfirewhenthefishseesthepattern,whichonesareactiveasittriestoswim,andwhethertheanimal’smovementisfeedingbackintotheneuralactivity.Freemanhopestoanswerthesequestionsbothattheleveloftheentirebrainandonthelevelofindividualneurons.Justanhouroftheserecordings,however,cangeneratemorethanaterabyteofdata(theroughequivalentof16millionbooks),makingtheuseofThunderoranothertypeofclustercomputingsoftwareanecessity.

TheanalysesperformedbyThunder,however,areonlyasgoodastheexperimentsthatprovidethedata.ToFreeman,theoryandcomputationaren’tjustthingsyoudoafteryougetyourresults–theyneedtobeintegratedintoeveryaspectofthescientificprocess.

“Ifpeoplehandedmedata,andIwentawayforsixmonthsandanalyzedit,itwouldbeboring,”hesays.“That’snotreallythewaytoprogress.Youneedtobeconstantlyinteractingandfindingcoolstuffinthedatatogether.”

Modeling NetworksTheclosemarriageofcomputation,theory,andexperimentisn’tuniquetobiology–it’slongbeenafeatureofphysics.Perhapsthatexplainswhysomanytheoreticalneuroscientists,includingmanyofthoseatJanelia,gottheirstartnotinbiologybutinphysics.ShaulDruckmann,forexample,wasallsettostartagraduateprograminhigh-energyphysicsbutchangedhisfocustoneuroscienceafter

“If you could give me the structure of the circuit, it would serve as crucial inspiration for hypotheses regarding what the circuit is trying to do.”

—shauldruckmann

Mat

tSta

ley

28

Ph

oto:

Mat

tSta

ley;

Sci

ence

imag

e:F

reem

anla

b

hearingalecturebyatheoreticalphysicistwhoworkedonneuralnetworks.AftercompletinghisPhDincomputationalneuroscienceatJerusalem’sHebrewUniversity,whereheconcentratedondetailedmodelsofsingleneurons,DruckmannmovedtoJaneliaforapostdocpositionin2013andwassubsequentlyhiredasagroupleader.Sincecomingtotheresearchcampus,hehasfocusedondevelopingmodelsofworkingmemory,theprocessthatholdsinformationforbriefperiodsoftime.

Researchershadpreviouslybelievedthatifyouaskedmicetoremembersomethingforashortperiodoftime,theirneuralactivitywouldberelativelyconstant,sincealltheyweredoingwasremembering.Butwhenscientistsactuallyrecordedwhattheneuronsweredoing,theyfound

thatneuralactivityinthemicewasalloverthemap,constantlychanging.Druckmannwantedtobuildamodeltoshowwhysuchactivitywasalwaysshifting.Duringhispostdoc,Druckmannwasabletoshowthatasimplemodel,wherebyneuronsexchangeinformationbetweenthemselvesinaspecificmanner,showshowshiftingactivityisperfectlyconsistentwithrepresentinginformationthatisn’tchangingovertime.

“Theprobleminthebrainisthattypicallyyouhavenetworksofneuronsthatare

extremelystronglyconnectedtoeachother,soAtriggersB,BtriggersC,andthenCgoesbackandtriggersA.ButitalsotriggersD,E,andF,andeachoneofthesealsotriggersX,Y,andZ,andtheyallfeedbackintoeachother,”saysDruckmann.“And,it’sallnonlinear,whichmeansitisconsiderablymoredifficulttointuitcausesfromeffects,andallofthesemodelsbecomehardertothinkthrough.

Onceonedoesthemath,andunderstandshowthiscancomeabout,itisstraightforwardtoexplaintheintuitionbehindit,butwithoutgoingthroughthecalculations,itwouldbehardtoreachthatintuition.”

Atthesametime,anotherJaneliagroupleader,KarelSvoboda,wastestingshort-termmemoryinmice,whichgaveDruckmannachancetoseeifhisideaswouldmatchexperimentaldata.Thememorytaskpresentedtothemicewasrelativelystraightforward.First,theresearcherstrainedamousetorespondtoasensorycue–apoleitcouldlocatewithitswhiskers.Dependingonthelocationofthepole,themousewouldrespondbymovingrightorleftafterhearingabeep.Importantly,thesoundwasnotplayedimmediatelyafterthemousefoundthepole–theanimalhadtowaitsecondsbeforethebeepsounded.Thismeantthatthemousehadtorememberboththelocationofthe

poleandthedirectioninwhichitneededtomovewhileitwaited.Svoboda’steamalsousedoptogenetics–atechniquethatallowsscientiststocontrolneuralactivitywithlight–tostrategicallyturnoffdifferentgroupsofneuronsinthemousebraincortex.Theyfoundthatactivityinanareaofthebraincalledtheanteriorlateralmotorcortexwascrucialtotheanimalbeingabletoperformthememorytask.Whenscientistsswitchedoffthatarea,performancedramaticallydeclined.

Nowthathehasthisinformation,DruckmannisrevisinghismodelstomoreaccuratelyreflectSvoboda’sdata.Byunderstandingthesefluctuationsinneuralactivityontimescalesofjustafewhundredmilliseconds,Druckmannhopestounderstandwhatcomputationsaregoingonduringthistimeperiodandwhatkindofmechanismsmayberesponsibleforit.Akeypieceofinformationthatishardtogetatisthestructureofthebrain’sneuralnetworks,whichcantellscientistsalotabouthowthedifferentpartsofthebrainfunction.“Ifyoucouldgivemethestructureofthecircuit,itwouldserveascrucialinspirationforhypothesesregardingwhatthecircuitistryingtodo,”Druckmannsays.

Unifying PrinciplesCrackingjustoneneuralnetworkcangivescientistsinsightintomanyothernetworks.TheoreticalneuroscientistSandroRomani,alsoagroupleaderatJanelia,spenttheearlyyearsofhiscareermodelinghowprimatesperformworkingmemorytasks.DuringhispostdocattheWeizmannInstituteofScienceinIsrael,heworkedtounderstandhowhumansrememberlonglistsofwords.Later,atColumbiaUniversity,hewentontostudyneuralcircuitdynamicsinthehippocampus,aseahorse-shapedregiondeepinthebrain

This visual representation of neurons from the larval zebrafish brain shows calcium responses in the cells – an indicator of activity – measured using light sheet microscopy. Each circle marks a neuron’s position, with colors based on a functional categorization. Curves connecting the circles reflect a measure of neuronal coupling.

Shaul Druckmann focuses on developing models of working memory, the brain’s ability to hold information short-term.



thatcontrolsnavigationandlong-termmemory.Specifically,Romaniwaslookingatrecordingsfromplacecellsinthehippocampusoftherat,eachofwhichmapsaspecificplaceintheanimal’sworld.Askarattorunafamiliarmaze,andresearcherscantrackwhereitislocatedbywatchingwhichplacecellsfireinitsbrain.

Whenscientistslookedatthisfiringpatternmoreclosely,however,theyfoundthatthesignalswereevenmoreintriguingthanthey’dexpected.Inparticular,theynoticedaregularlyrepeatingsequentialactivationoftheplacecellsthatmovedmorequicklythantheratdid,indicatingthattheanimal’smovementwasn’tdrivingthisactivity.Instead,thecells’activitieswerelikelypredictingtherat’sfuturemovements.

Justasaratexploresnearbyplaces–whicharesimilarbutnotthesame–inasequence,itemsinmemorytendtoberecalledaccordingtohowsimilartheyare.Forexample,grabbingaboxofcakemixatthegrocerystoreremindsyoutopickupfrostingaswell,insteadoftriggeringyourmemorythatyoualsoneedsteak.Scientistsstudythiskindofmemorybyaskingpeopletorememberalonglistofwordsinwhat’sknownasthefreerecalltask.Asthelistbecomeslonger,peoplecanrecallmorewordsbutthefractionofthetotalgetssmaller.

Romaniandcolleaguesrealizedthatthenervoussystemmayusesimilardynamicsforbothsequencesofplacesandsequencesofwords,andtheydevelopedaneuralnetworkmodeltoshowthatthiscanbedonewitharealisticneuralarchitecture.

AtJanelia,RomanihasworkedwithGroupLeaderEvaPastalkovaonamodifiedversionofthemodeltoaccountforsomeofherexperimentalresultsonagroupofhippocampalneuronsshehasdubbed“episode”cells.Theseneuronsareactiveinasequencewhenaratstopsitsmotionthroughtheenvironmentwhileitisengagedinamemorytask.

“Onecontributionthattheoreticalneurosciencecanbringtothetableistostepbackalittlefromthephenomenaandtrytofindunifyingmechanismsandprinciplesbehindthem,”Romanisays.

Theoreticalmodels,Abbottpointsout,areanothersetoftoolsforpeopledesigningandcarryingoutexperiments–andshouldbetreatedassuch.“Inthepast,yougeneratedawholebunchofdataandyoujustthrewitatatheoristandsaid,‘Well,whatdoesitmean?’Butit’sbettertohavepeoplewithdifferentexpertiseintheprocess,”heexplains,“thesameasyoumighthaveanexpertmicroscopistintheprocessandsay,‘Howdowedesignthisexperimentsowecanget

thebestuseoutofthetools?’Theoryisoneofthetoolsindesigningandgettingthemostoutofanexperiment.”

It’saprocessAbbotthimselfuseswhiletryingtounderstandhowananimalrespondstostimuli.Forexample,theolfactorysysteminaflyhelpsittorecognizedifferentodorsinitsworld,everythingfromnutritiousfoodtothepresenceofpotentialmates.Butthefly’sbraindoesn’tstopatjustrecognizingodors;italsohelpsdrivethefly’sbehaviorinresponsetothoseodors.Aflymightmovetowardhealthyfoodbutawayfromfoodthatislessnutritious–responsesthatAbbott,workingwiththelabgroupsofGerryRubinatJaneliaandHHMIInvestigatorRichardAxelatColumbiaUniversity,hasbeguntoidentifyintheolfactorycircuit.

“Youseethistransformationfromarepresentationthat’sdominatedbytheoutsideworld–manyodorscomeinandactivatecells–tosomethingthat’sinternaltotheanimal,”Abbottsays.“Itdependsontheirexperience,whetherthey’rehungryornot,thosekindsofthings.”

Thetasknow,hesays,istoworkwithotherscientiststodesignexperimentstoidentifythoseexperiencesandinternalfactorsthatdrivechoices.AllthetheoreticalandcomputationalneuroscientistsatJaneliaagreethattheirworkisiterative.

“WhatIhopewillhappenatJaneliaisthattherewillreallybethiscloseshoulder-to-shoulderinteractionwithexperimentallabs:thedevelopmentofmodelsbasedonexperimentalobservations,whichinturngeneratespredictionstobetestedwithnewexperiments,”Romanisays.

Thispartnershipamongtheory,computation,andexperimentinneurosciencegivesresearchersanenhancedabilitytodeviseeffectivewaystopeerintothemysteriousdepthsofthebrain.Asbetterexperimentsyieldaclearerunderstandingofneuralprocesses,thesescientistsarehelpingresearcherstonavigatetheturbulentseaofneuralactivityandrowtowardthetruth.

“Theory is one of the tools in designing and getting the most out of an experiment.”—larryabbott

Sandro Romani studies the ability of the brain’s hippocampal cells to store sequences of places and words.

To see more of the data being analyzed by Janelia’s theorists, go to hhmi.org/

bulletin/spring-2015.

Mat

tSta

ley

Perspectives & Opinions


The Theory ConnectionNeuroscientist Jeremy Freeman believes scientific theory is all about making connections: between people, between labs, between biological systems. And he considers Janelia Research Campus, where he’s a group leader, to be an ideal environment for fostering those relationships.

wealllearned thescientificprocess.Formulateahypothesis,collectexactlytherightdatatotestthehypothesis,andthenevaluateit.Butit’sneverthatstraightforward.Moreoften,wecollectnot-really-the-rightdataandendupswimminginobservations,manyofthemunexpected.Somearecurious,somearetrivial,somearegroundbreaking.Howdowesortthroughthem?Howdoweplanthenextexperiment?Howdoweputitalltogether?

Tome,theoryisacollectionofstrategies,tools,andconceptsformakingsenseofexperimentaldata.It’sthehalfwaypointbetweendataandideas.Bydesign,IrarelyenteracollaborationorexperimentwithapreconceivednotionofwhatIhopeorexpecttofind.Iletmythinkingbedrivenbythedata,andallowmyselftobesurprised.Butdataanalysisquicklysuggestshypotheses

andmoretargetedexperiments.Idomybesttositinthemiddleofthatloop.

Thisapproachbecomesallthemorepressingaswestudyincreasinglycomplexsystems.Themagnitudeofdatabeingcollectedandtheexperimentalcapabilities,atJaneliaandelsewhere,areoutpacingourabilitytoanalyzethedata,letalonedevelopprinciplesforwhattheymeanandwhatexperimentstodonext.Wehaveunprecedentedaccesstotheinnerworkingsofthenervoussystemandcandoexperimentsthatafewyearsagowouldhaveseemedimpossible.Tokeepup,weneedtomodernizeouranalysis,makingitmoreefficient,morecollaborative.Anddoitfastenoughtoadjustourexperimentsinrealtime.

Onegreatexampleismylab’scollaborationwithNickSofroniew,inKarelSvoboda’slab.Nickhasdesignedanincrediblevirtualrealitysystemthatsimulatesatactileenvironmentthroughwhicharealmousecannavigate,runningonaballwhilewallsmovebackandforthstimulatingitswhiskers.Wecanmonitorneuralresponsesoverincreasinglylargeportionsoftheanimal’sbrainwhile

itperformsthisbehavior.Thecomplexityisdaunting–notonlyofthedatathemselves,butthesheernumberofpossiblebehavioralparadigmsandexperimentalmanipulations.Onewaytostartparingdownthatlististoanalyzethedataonline.Thisletsusmakeexperimentaldecisionsormanipulateneurons

basedonthemeasureddata,allwhilethemouseisstillontheball.Someofwhatwe’redevelopingisspecifictothissystem,butmuchofitisgeneral–boththetechniquesandtheideas–andwe’reactivelycollaboratingwithotherstodevelopsimilarapproachesinothersystems.

Establishingacomputationalcollaborationstartswithmakingapersonalconnection.Weneedtoopenandmaintainconversations,betweentheoristsandexperimentalgroups,andamongdifferentgroupswhomightfindit

usefultobeexposedtoeachother’sideas.Idon’tthinkit’sasurprisethatmostofthegreatdiscoveriesinsciencehavetwoormorenamesattachedtothem,andthatwillonlycontinueaswetacklelargerandmorecomplexproblems–together.

Janeliaisaphenomenalplacetodocomputationandtheorybecausetherearealmostnobarrierstobuildingtheserelationships.Everyoneissoopentolookingattheexperimentaldatawithothers.Janelianswanttocollaborate.

Thephysicalspaceinwhichweworkmattersalot,especiallywhenitcomestoproximity.Janeliaisrelativelysmall,soit’seasytorunintopeople.Onabig,sprawlingcampusitwouldbehardertogetcollaborationsofftheground.AtJanelia,ifIjustsitalldaybythecoffeebar,I’llhaveachancetotalkwithallofmycollaborators–andprobablyendupwithnewones.

Theorists’activitiesatJaneliaarediverse,butonethingwehaveincommonisthatweworkacrossdifferentsystems.TheoristsatJaneliahaveauniqueopportunitytositattheintersectionbetweenlabs,revealingprinciplesthatspananimalsandmodalities,andsharingideasandtechniques.Partofmyownworkisbuildingplatformsforanalysisofneuraldatathatmakeiteasiertocollaborateandcoordinate.Asthateffortgrowsbeyondourcampusandintothebroaderneurosciencecommunity,Janeliaitselfhasbeenamodelforwhatopennessandcollaborationcan,andshould,looklike.

Ioftenhearthatneuroscienceneedsmoretheorists.Maybeit’snotthatweneedmore;it’sjustthatweneedtodissolvethebarrierstomakingconnections.Allscientistsshouldworkclosertotheintersections.We’dallbethinkingalittlemoretheoretically,alittlemorecomputationally.Andthat’swherethebigideaswillcomefrom.–Interview by Nicole Kresge D

avid

Fri

edm

an

Jeremy Freeman—Jeremy Freeman is a group leader at the Janelia Research Campus.

31HHMI Bulletin / Spring 2015Jeremy Freeman applies computational approaches to data analysis and experimental design.

32

Perspectives & Opinions

Spring 2015 / HHMI Bulletin

Erich JarvisHHMI InvestigatorDuke UniversityIhadmylabreadGeorgeGopen’sExpectations: Teaching Writing from the Reader’s Perspective,abookaboutwritingskillsfocusedonwhatscientistsandotherprofessionalsexpecttoreadinarticles,grantproposals,andbooks.Gopen,aDukeUniversityprofessorwhostudieswriting,claimsthatourprosecanbemoresuccessfulifweconsiderthecommonreader’sperspectiveandexpectations.AfterIstartedusingGopen’ssuggestions,Ifoundthatmygrantsandpapersweremoreunderstandable.Itdidn’tmeanthatmyreadersagreedwithme,butatleasttheygraspedwhatIwastryingtoexplain.

Nicole KingHHMI InvestigatorUniversity of California, BerkeleyWhenIwasingraduateschoolatHarvard,paleontologistAndyKnollhandedmeacopyofThe Evolution of IndividualitybyLeoBuss,whichopenedupanentirelynewwayofthinkingaboutevolutionanddevelopmentforme.Thebookprovidesabeautifullywrittenandcompellingframeworkforthinkingaboutthemeaningof“individuality”–atopicthatiscentraltomylab’sworkonmulticellularityandtheinteractionsbetweeneukaryotesandenvironmentalbacteria.Ithasbecomerequiredreadingforallnewmembersofmylab.

Oliver HobertHHMI InvestigatorColumbia UniversityAdisparagingremarkoftenheardfromreviewersandeditorsisthatapieceofworkis“toodescriptive.”Thisstatementneverceasestoperplexme.JavierDeFelipe’swonderfullyillustratedbookCajal’s Butterflies of the Soul: Science and Art–itspoeticnameisderivedfromCajal’sdescriptionofpyramidalneuronsintheneocortex–servesasabrilliantreminderoftheimportanceofdescriptivescience.Thebookconveysthebeautyofsuchworkandencouragesmetokeeppursuingseveralhighlydescriptivelinesofresearch.

Q&A What book have you read that influenced your work, and what effect did it have?Bookshavetheabilitytotakeusplaceswe’veneverbeen,toteachusconceptswedidn’tknow,andtointroduceustopeoplewe’venevermet.Often,longafterwe’veclosedabook,itscontentswillstaywithus.Here,fourHHMIinvestigatorsrecallsomeoftheirmemorablereads.–Edited by Nicole Kresge K

ing

:Ali

son

Yin

,Hob

ert:

Kev

inW

olf,

Jarv

is:J

onat

han

Fre

din

,Lee

:Kev

inW

olf

Jeannie LeeHHMI InvestigatorMassachusetts General HospitalInGeorgeOrwell’sAnimal Farm,IcametounderstandthatoneofSnowball’ssevencommandments–thetransfigurationof“Fourlegsgood,twolegsbad”to“Fourlegsgood,twolegsbetter”–wasasigntovalidateanimalmodels.Iamonceagainembracinghumancellularmodels.

ChronicleJa

net

Iwas

a,U

niv

ersi

tyo

fU

tah

34 scienceeducation FromInsighttoAction36 toolbox BarcodingBacteria38 labbook TCellBurnout GotThirst?Here’sWhy SalvagingSNAREs

Mostscientistshavelongbelievedthateachaminoacidinaproteinisspecifiedbyathree-lettercodefoundinmessengerRNA(mRNA).ButateamledbyHHMIInvestigatorJonathanWeissmanhasfoundanexceptiontotherule,illustratedinthisstructure.Withhelpfromacell’stransferRNA(darkblueandteal)andpartoftheribosome(whitetangles),aproteincalledRqc2(yellow)canassembleaprotein(green)intheabsenceofanmRNAblueprint.Butthere’sonecaveat.Rqc2recognizesonlyalanineandthreonine(abbreviated“A”and“T”),soitcreateswhatWeissmanreferstoas“CATtails.”Read more about the research in “A CAT Tale” on page 38.

Spring 2015 / HHMI Bulletin

Chronicle / Science Education


From Insight to ActionPULSE, a national network of science education leaders, is helping transform college-level science teaching and student learning.s o m e t i m e s t h e b e s t waytosolvebig,nationalproblemsistostartbytalkingtoyourneighborjustdowntheroad.TakeJennyMcFarland,forexample,abiologyfacultymemberatEdmondsCommunityCollegeinLynnwood,Washington.Manyofhersciencestudentscontinuetheirstudiesatnearbyfour-yearuniversitiesandcolleges–butseldomdofacultymembersfromthevariousschoolsinteractwitheachother.Yetifprofessorsdon’tcoordinatetheirteach-ingofkeyconcepts,likethestructureofcellsorthegrowthoforganisms,thatcanhurtstudentswhotransfer.

Recently,McFarlandjoinedfacultyteamsfrom14schoolsintheNorthwestforathree-dayworkshoptothinkabouthowtheinstitutionscouldlearnfromoneanotherforthebenefitoftheirstudents.“Wewantedtolookatourregionasasystem,”saysMcFarland.“Howcouldweworktogether?Whatresourcescouldwedrawfromeachother?”Forsomeinstitutions–eventhosejust

afewmilesapart–itwasthefirsttimetheirfacultymembershadtalkedtooneanother.

TheworkshopwaspartofaregionalinitiativeconductedbythePartnershipforUndergraduateLifeSciencesEducation(PULSE),agroupformedafterthe2011publicationofareporttitled Vision and Change in Undergraduate Biology Education: A Call to Action.ProducedbytheAmericanAssociationfortheAdvancementofScience,withsupportfromtheNationalScienceFoundation(NSF),thereportofferedanumberofrecommendationstomakehigher-educationscienceteachingmoreeffective–includinggivingstudentsmoreresearchexperiences,urgingfacultymemberstomovebeyondthetraditionallectureformat,andcatalyzingdepartmental-levelchange.Tohelpturntheseideasintoconcreteactions,officialsatHHMI,theNSF,andtheNationalInstituteofGeneralMedicalSciencesworkedtogethertofoundPULSE.

“Wewantedtomakesurethatthisreportdidn’tjustgatherdustoneverybody’sshelves,”saysHHMI’sCynthiaBauerle,assistantdirectorofundergraduateandgraduateprograms(andacoauthorofVision and Change).

Theeffortgotunderwayin2012withtheappointmentof40PULSEFellows;oneofthemwasMcFarland,andallweremid-careerorseniorlifesciencefacultymembersatschoolsrangingfromcommunitycollegestolargeresearchuniversities.Sincethen,PULSEhasmadesignificantprogressinidentifying,measuring,andsupportingschools’effortstoimplementVisionandChangeprinciplesforexceptionallifescienceseducation.

Oneofthegroup’sfirstprojectswasdevelopingrobustregionalnetworks.Thesenetworksfocusonarea-specificinterestsandopportunities.Forexample,intheSoutheast,teamsofcampusleadersfromcommunitycolleges,smallprivatecolleges,researchinstitutions,andhistoricallyblackcollegesmetlastsummertodiscussstrategiesthatcouldhelpunderrepresentedminoritiespersevereandsucceedinthesciences.Upuntilthatmeeting,interactionbetweentheregion’shistoricallyblackcollegesandpredominantlywhiteinstitutionshadbeenrelativelyrare,saysBauerle.

Asecondkeyinitiativeisaprogramtorecognizeinstitutionsthathavedone

“We wanted to make sure that this report didn’t just gather dust on everybody’s shelves.”—cynthiabauerle

exceptionalworkimplementingVisionandChangerecommendations.Thisprocessdigsdeepintoschools’curriculaandapproaches:Areconceptssuchasevolutiontaughtexplicitlyandimplicitlythroughoutthecurriculum?Aremorethanhalfofsciencestudentsabletotakeadvantageofmentoredresearchopportunities?Areclassroomssmallanddesignedforrealinteractionbetweenstudentsandfacultymembers,asopposedtocavernous,theater-stylelecturehalls?

Groupmembersrecentlycompletedasuccessfulpilotstudywitheightschoolstoseeifthecriteriathey’ddevelopedaccuratelymeasuredwhetherschoolswereachievingVisionandChangegoals.Theresults,whichwillbesharedonthePULSEwebsitelaterthisspring,werepromising,saysKathyMiller,aPULSEFellowandchairofthebiologydepartmentatWashingtonUniversityinSt.Louis.Thegroupwilllikelyassessmoreschoolsinthenearfuture,sheadds.

Therecognitionprogramwillbothcommendtop-performingschoolsandhelpallschoolsunderstandsmartnextsteps,saysMcFarland.“Ifpeopleareassessingtheirdepartmentswiththeserubrics,andtheywanttoseewaysthattheycanimprove,theyhavearoadmap,”shesays.Peopletryingtomakeacaseforinvestmentinteaching,shecontinues,“aren’tjustsaying,‘Ihaveanideathat I wanttotryintheclassroom.’They’remakingthecasewithrubricsthathavebeenstandardized,vetted,andvalidated.”

AthirdmajorinitiativeisthecreationofVisionandChangeAmbassadors–teamsof

HHMI Bulletin / Spring 2015 35C

ath

erin

eL

epag

eHHMIBulletin/Spring2015

facultytrainedtoguideconversationsaboutreformatotherinstitutions.“Theambassadorsmeetfacultymembersontheirownturfandtrytokick-startdepartmentalchange,”explainsMcFarland,“whetherthat’stogetthingsstarted,helpthemchangedirections,orguidetheminsomeway.”

AsthesevariousinitiativeswithinPULSEtakeshape,agrowingonlinecommunity–nownumberingsome1,500members,atpulsecommunity.org–isactivelysharingitssuccesses,challenges,andteachingpractices.It’sprovingthatastrongvisionstillallowsfor

varied,effectiveteachingapproaches,saysJimCollins,anecologistandevolutionarybiologistatArizonaStateUniversity.“I’mseeingalotofdiversityinthewaythatindividualsaremovingawayfromthe‘sageonthestage’model,”hesays.“Theymightuseclicker[technology]togatherinformation,theymighttryflippingtheclassroomwithvideo-basedlecturesthatstudentswatchontheirowntime,theymayworkwithsmallergroups.There’ssomuchrichness,andthat’sgreat.”

AsthemembersofPULSElookahead,theyhopetocontinuetofosterclose

connectionsamonginstitutions,toscaleuptherecognitionprogram,andtoprovideincreasedsupporttoinstitutionsandindividualswhowanttoimplementVisionandChangeprinciplesintheirclassroomsanddepartments,saysChuckSullivan,aprogramdirectorfortheNSF’sDivisionofBiologicalInfrastructure.“PULSEisexpandingitscircleofinfluence,liketheconcentricringsyouseeafterastoneisthrowninapond,”hesays.“Morepeoplearegettinginvolvedandreformingtheircourses.Andthat’swhatwehopedwouldhappen.”– Erin Peterson

36

Chronicle / Toolbox

Barcoding BacteriaDNA barcodes and a half-century-old equation help scientists track infection in the gut.f o r t h e b a c t e r i a thatcausecholera,thejourneythroughahost’sgutisnopicnic.Thefirststopisthestomach,filledwithgastricjuicessodeadlythatthevastmajorityofVibrio cholerae bacteriawon’tmakeitoutalive.Survivorsburrowintothewallofthesmallintestine,hopingtoavoidthetideofdigestedfoodrushingby.Theluckyoneswill

reproduce;thelessfortunatewillbeattackedbytheimmunesystemor,insomecases,killedbyantibiotics.

Theseebbsandflowsinpopulationsizehavelongfascinatedbiologists,whocanusetheinformationbothtogarnercluesaboutthegeneticfitnessofanorganismandtopinpointthebesttimestooverpowerapathogen.

“Fromatheoreticalperspective,it’smucheasiertofightaverysmallpopulationthanabigpopulation,”explainsPiaAbelzurWiesch,apostdoctoralfellowwhoworkswithTedCohen,aninfectiousdiseaseexpertatBrighamandWomen’sHospitalinBoston.“Ifyouhaveanantibioticorvaccineactingwhenthepopulationisespeciallyvulnerableandsmall,youhaveamuchhigherchanceofsuccessbecauseit’seasiertoeliminateafewbacteriathanalotofthem.”

Buthowcanresearcherstrackapopulationofmicroscopicbacteriainsideananimal’sgut?ZurWieschandherhusband,SörenAbel,alsoaninfectiousdiseasepostdoctoralresearcher,wereponderingthisquestionatdinneronenight,shortlyafterhejoinedthelabofHHMIInvestigatorMatthewWaldor,alsoatBrighamandWomen’sHospital.Thetwopostdocsdecidedtocombinehermathematicalexpertisewithhismicrobialknowledgetoanswerthequestion.Theirsolutioninvolvedcoupling500DNAbarcodeswitha50-year-oldmathematicalformulatoproduceatechniquethatcantracktheupsanddownsofprettymuchanycellpopulation–pathogenorotherwise.

ThetechniqueiscalledSTAMP,for“sequencetag-basedanalysisofmicrobialpopulations.”It’satwo-partprocedurethatboilsdowntotackingakindofbarcodeontobacteriaandthentrackingthebarcodefrequencytocalculatethesizeofafoundingpopulation.

ThebarcodesareshortstretchesofDNAthathelpthescientistsdistinguishonebacterialcellfromanother.“Thelabelsarejustlikelastnames,”sayszurWiesch.“Theydon’treallychangeanythingabouttheindividuals.Ifyouthinkaboutafewpeoplefoundingavillage,forexample,andyoujustfollowtheirlastnames,youcaninferhowmanypeoplethereweretobeginwith.”

Todeducethesizeofthefoundingpopulationfromthebarcodes,zurWieschmadeuseofamathematicalequationestablishednearlyahalf-centuryago.In1971,geneticistsCostasKrimbasandSpyrosTsakasshowedthatitwaspossibletousemarkerstoinferthesizeofapopulationattwodifferenttimepoints.KrimbasandTsakasusedtheirequationtofollowtheeffectsofaninsecticideonthepopulationdynamicsoffruitfliesthatfedontreatedolivetrees.

Inasimilarfashion,Abel,zurWiesch,andtheircolleaguesusedSTAMPtotracethegrowthanddeclineof V. choleraeinfectioninrabbits.Abeladdedabout500differentbarcodestothegenomesofabatchof V. cholerae.Heinfectedrabbitswiththelabeledbacteriaandthencollectedsamplesfromtheanimals’gutsatvarioustimepoints.NewdeepDNAsequencingtechnologyallowedhimtodeterminehowmanyofeachtypeoftaggedbacteriawereinthesamples.


37Ju

stin

Ren

teri

aHHMIBulletin/Spring2015

“STAMP is applicable, in principle, not only to all pathogens, but also to commensal organisms and even to eukaryotic cells.” —matthewwaldor

Withthosenumbersinhand,theresearchersusedKrimbasandTsakas’sequationtotracethepopulationdynamicsofthetaggedV. choleraebacteria.Theycomparedtherelativeabundanceofthetagsintheinjectedbacteriatothenumbersinthedifferentintestinalsamples.Asamplewithalargechangeintagfrequenciesmostlikelyexperiencedabottleneck–someeventthatdrasticallyreducedthepopulationsizeand,intheprocess,alteredthedistributionoftags.Conversely,asamplewithasmallerchangeintagfrequenciesprobablywentthroughamorebenignpathwaythathadaminimaleffectonthebacteria,resultinginapopulationverysimilartotheonethathadinitiallyinfectedtherabbits.

Thecalculationsyieldedsomesurprisinginformationaboutthebacteria’sjourneythroughthegut.“Frankly,whenweappliedSTAMPtoVibrio choleraeinrabbits,wethoughtitwouldbekindofaboringexperiment,”saysWaldor.“Wethoughtthepatternofbacterialmigrationfromthestomachtotheanuswouldbeaforegoneconclusion.Butthat’snothowitturnedout.”Instead,theteamdiscoveredthat,afteracertainpointintheinfection,

thebacteriachangeddirection,goinguptheintestinaltractinsteadofheadingdown.

“Wecouldn’thavefiguredthatoutwithanyothermethod,becausewewouldn’thaveknowntheidentityofthebugs,”saysWaldor.Thoughthescientistsaren’tsurewhytheV. cholerae reversedcourse,they’reinvestigatingafewideas.

Theresearchers’technique,publishedintheMarch2015issueofNature Methods,isn’tlimitedtobacteria,accordingtoWaldor.“STAMPisapplicable,inprinciple,notonlytoallpathogens–bacteria,viruses,andparasites–butalsotocommensalorganismsandeventoeukaryoticcells,”hesays.“Forexample,youcouldusethisincancermetastasisstudies,todeterminehowmanycancercellsmetastasizetoanewsite.”

Theimportanceofthemethod,saysWaldor,isthatitenablesscientists,inaretrospectivefashion,tofigureouthowmanycellsorbacteriawerethefoundersofaparticularpopulationandtoinferwherebottlenecksoccur.Thisisgoodnewsforhumans,butbadnewsforV. choleraeandotherpathogens,nowthatscientistscanpinpointwhenthey’remostvulnerable.–Nicole Kresge

38

Chronicle / Lab Book

I N B R I E F

SEEING REDAs a fish swims, nerve cells fire in its brain, sending signals racing along a neural network ending in muscles that make its fins flap and its tail swish. By using a molecule called CaMPARI to permanently mark neurons as they fire, scientists can now watch as signals light up such neural networks in live animals.

CaMPARI came out of a collaborative project spearheaded by Eric Schreiter, a senior scientist in Group Leader Loren Looger’s lab at the Janelia Research Campus. The team started with a protein called Eos, which emits a green glow until it’s exposed to violet light. The light changes the molecule’s structure, causing it to glow red. By combining Eos with the calcium-sensitive protein calmodulin, the researchers were able to couple Eos’s color

change to the burst of calcium that accompanies

neuronal signaling. The resulting molecule indelibly tags firing neurons with a red glow in the presence of violet light.

“Ideally, we would flip the [violet] light switch on while an animal is doing a behavior that we care about, then flip the switch off as soon as the animal stops the behavior,” Schreiter explains. “So we’re capturing a snapshot of neural activity that occurs only while the animal is doing that behavior.”

The scientists published their results February 13, 2015, in Science. Although they are still tinkering with CaMPARI to make it more sensitive and reliable, they’ve already made it available to scientists on Addgene and the Bloomington Drosophila Stock Center. Janelia Group Leader Misha Ahrens is also distributing CaMPARI-expressing zebrafish.

IMMUNE RALLY CRYLike a Good Samaritan, a cell that’s been attacked by a virus warns

neighboring cells to shore up their defenses. These alerts are sent via a family of proteins called interferons, which are produced when surveillance proteins in the infected cell detect a pathogen. Although several different surveillance proteins scout for signs of pathogens, new research shows how the proteins all activate a single molecule called IRF3 to turn on interferon production.

There are three known pathways that trigger type 1 interferon production. In each case, the individual pathway’s unique surveillance protein uses its own adaptor protein to relay the message that an invader is present and interferon is needed. Siqi Liu, a graduate student in HHMI Investigator Zhijian “James” Chen’s lab at the University of Texas Southwestern Medical Center, noticed that all three adaptor

proteins have similar stretches of five amino acids that became tagged with phosphate groups. As the team reported March 13,

2015, in Science, the addition of a phosphate molecule to that stretch of amino acids causes the adaptors to activate IRF3.

Now that they know how the interferon pathways converge, Chen and his team are examining them in more detail. Eventually, they hope to develop small molecules that treat immune disorders by interfering with the pathways.

A CAT TALEA central tenet of biology is that each amino acid in a protein is specified by a three-letter code found in messenger RNA (mRNA). That may be true most of the time, but HHMI Investigator Jonathan Weissman at the University of California, San Francisco (UCSF), along with Onn Brandman at Stanford University and UCSF P

hot

o:N

atio

nal

Inst

itu

teo

fAll

erg

yan

dIn

fect

iou

sD

isea

ses;

Illu

stra

tion

:Lei

fP

arso

ns

Winter2015/HHMIBulletin

T Cell BurnoutSome immune cells function better after taking a break.everylivingbeing needstorest.EvenourimmunecellsenteradysfunctionalstatecalledTcellexhaustionifthey’reoverworked.Infact,thisfatigueandtheensuingdowntimeareimportantpartsoftheimmuneprocess.

KillerTcellsgettheirnamefromtheirfunction.Theytargetsickenedcellsinvolvedinchronicinfections,killingthembefore

thevirusesinsidecanreplicateandspread.Butchronicinfectionscanbelengthy,andafterafewweeksofsustainedaction,cellularexhaustionoftensetsin.TheTcellsbecomelesseffectiveatslayingtheirtargetsandbeginproducingproteinsthatpreventthemfromrecognizinginfectedcells.

Oneoftheseproteinsiscalledprogrammedcelldeathprotein1,orPD-1.SusanKaech,anHHMIearlycareerscientistatYaleUniversity,discoveredafeedbackloopbywhichPD-1activationleadstoanincreaseinaproteincalledFOXO1.FOXO1,inturn,producesfactorsthatpromoteTcellexhaustion,includingmorePD-1.KaechsuspectedthateliminatingFOXO1mightcurtailTcells’exhaustionandmakethembetterkillers.

Whatshefoundwastheexactopposite.Whenherteamcreatedmicelackingthe FOXO1gene,therodents’TcellsdidproducelessPD-1comparedtoanimalswiththegene.Butthecellsweren’tbetteratcontrollingtheviralinfection.Instead,withouttherestaffordedthemintheirexhaustedstate,thecellsdied,andviralreplicationincreased.

The immune system’s T lymphocytes, like this one, need downtime to do their job properly.

KaechconcludedthatTcellsneedthisrespitetofunctionproperly.“Theyhavetoturndowntheirresponseorelsethey’llgetover-activated,andwethinkthatcausesthecellstodeteriorateanddie,”shesays.“We’restartingtoappreciatethatexhaustionisanimportantprocessthatishelpingtomaintainthispreciouspoolofTcells.”

Thefindings,publishedNovember20,2014,in Immunity,couldleadtodrugsthatmodulateFOXO1andhelpreinvigoratetheimmuneresponseofpatientsbeingtreatedforchronicviralinfectionsorevencancer.“Theremaybeapointwhereyoucanstopthecellsfromfullyenteringtheexhaustedstate–whereyousuppressPD-1,butnotsomuchthatthecellsdie,”Kaechsays.– Nicole Kresge

39

colleague Adam Frost, has discovered an exception to the rule.

Generally, protein synthesis requires three components: an mRNA template, a ribosome to read the template and assemble the protein, and transfer RNA (tRNA) molecules to supply the necessary amino acids. Weissman and his colleagues discovered that sometimes, when protein synthesis stalls, elongation can still continue without the mRNA template. This type of synthesis relies on only two components: a part of the ribosome called the 60s subunit and a protein named Rqc2 that recruits tRNA molecules. With these components, cells will continue to randomly add amino acids to the stalled peptide, with one caveat. Rqc2 recognizes only alanine and threonine (abbreviated “A” and “T”), so it creates what Weissman refers to as “CAT tails.”

“This is a mode of protein synthesis that we’ve missed for 50 years. It’s totally unprecedented,” says Weissman. He and his colleagues

published their findings January 2, 2015, in Science. They don’t yet know the function of CAT tails, but they have some ideas they’re pursuing. For example, the extra amino acids may trigger a stress response in the cell, or they could tag an incomplete peptide for destruction.

HOX PARADOX NO MOREDuring development in animals, the Hox gene family directs the formation of segment-specific anatomy along the head-to-tail axis. The genes are responsible for putting the head, thorax, abdomen, and other body parts in the right places. Surprisingly, all Hox proteins bind with high affinity to very similar stretches of DNA, which begs the question: how can Hox proteins find and activate their target genes if all binding sequences look essentially the same?

Janelia Group Leader David Stern recently revealed the answer to this long-standing paradox.

Stern’s team, in collaboration with Richard Mann’s lab at Columbia University Medical

Center, figured out that Hox proteins activate genes via weak interactions at

previously unrecognized DNA binding sites. The scientists discovered this by showing that a Hox protein

called Ubx controls expression of a gene called shavenbaby by binding weakly to two enhancer regions on DNA near the gene. These low-affinity binding sites conferred shavenbaby’s specificity for Ubx over other Hox proteins.

The findings, published January 15, 2015, in Cell, explain why scientists have been unable to predict where Hox proteins bind to DNA simply by looking at its sequence. “They’re not regulating genes by binding to the sites that everybody thought they were – they’re binding to these

sites that don’t look like good Hox binding sites,” explains Stern.

A TARGET FOR DYSKINESIASome human diseases occur episodically in people who appear otherwise healthy. One of these disorders is paroxysmal nonkine-sigenic dyskinesia (PNKD). The neurological disease’s symp-toms – involuntary movements in the limbs, torso, and face – are triggered by coffee, alcohol, and stress. In 2004, HHMI Investigator Louis J. Ptáçek at the University of California, San Francisco, discov-ered the gene that causes PNKD. Now, just over a decade later, he’s figured out what the gene does.

To deduce the function of the PNKD gene, Ptáçek’s team raised antibodies against the PNKD protein. This allowed them to isolate PNKD and its binding partners from mouse brain tissue. From these isolates, they discovered that PNKD interacts with two other proteins, called RIM1 and RIM2, that are involved P

hot

o:Z

uk

erla

b;I

llu

stra

tion

:Lei

fP

arso

ns

HHMIBulletin/Spring2015

Got Thirst? Here’s WhyScientists have pinpointed the neurons in the brain that control thirst.humanscantypically surviveonlythreetofivedayswithoutwater.Butwe’veallsaid“I’mdyingofthirst”wellbeforewe’vebeenwaterlessthatlong–albeitwithlittleunderstandingofthemechanismbehindtheurge.Butnowthebrainsignalsthatgovernthirstarenolongeramystery.AteamledbyHHMIInvestigatorCharlesZukerhasfiguredoutthatthemotivationtodrinkwateriscontrolledbytwosetsofneurons:onethatprovokesthestimulustosipandonethatquenchesit.

Researchershavelongsuspectedthatthesignalsdrivinganimalstodrinkoriginateinthebrain’ssubfornicalorgan,orSFO.Locatedoutsidetheblood-brainbarrier,whereithastheopportunitytodirectlysensetheelectrolytebalanceinbodyfluids,theSFOshowsincreasedactivityindehydratedanimals.YukiOka,apostdoctoralfellowinZuker’sColumbiaUniversitylab,tookacloserlookattheSFOinmiceandidentifiedtwotypesofnervecells:excitatoryCAMKII-expressingneuronsandinhibitoryVGAT-expressingneurons.Usingoptogenetics,Okaaddedalight-sensitiveproteintothecellsintheanimals’SFOs,allowinghimtoselectivelyactivatethemwithbluelight.WhenheflippedtheswitchandactivatedtheCAMKIIneurons,therodentsdrankwithgusto.

“Youhaveawater-satiatedanimalthatishappilywanderingaround,withzerointerestindrinking,”saysZuker.“ActivatethisgroupofSFOneurons,andthemousejustbeelinestothewaterspout.Aslongasthelightison,thatmousekeepsondrinking.”Okashowedthattheanimalsbecamesuchaviddrinkersthattheyconsumedasmuchas8percentoftheirbodyweightinwater–theequivalentof1.5gallonsforhumans.WhenOkausedthesametechniquetostimulateVGATneurons,thirstyanimalsimmediatelystoppeddrinkingandreducedtheirwater

intakebyabout80percent.TheresearcherspublishedtheirfindingsonlineJanuary26,2015,inNature.

AccordingtoZuker,theopposingneuronslikelyworktogethertoensureanimalstakeinenoughwatertomaintainfluidhomeostasis,includingbloodpressure,electrolytebalance,andcellvolume.Itremainstobeseenwhetherthesamecircuitcontrolsthirstinhumans;ifso,thefindingscouldonedayhelppeoplewithanimpairedsenseofthirst.– Nicole Kresge

The mouse brain contains CAMKII neurons (red) that trigger thirst and VGAT neurons (green) that quench thirst.

Watch a mouse begin to drink thirstily when prompted by a blue light.

Go to hhmi.org/bulletin/spring-2015.

40

Chronicle / Lab Book

in neurotransmitter release. As the scientists reported March 10, 2015, in the Proceedings of the National Academy of Sciences, PNKD normally suppresses neurotransmitter release, but the mutant form found in the disease does so less effectively. The result is excessive synaptic transmission, which leads to the involuntary movements experienced by people with the disease.

According to Ptáçek, RIM1 and RIM2 may be important targets for developing better medications for PNKD and other dyskinesias, such as those seen in Parkinson’s disease.

THE LONG LOST RECEPTORIn our cells, there is a widely expressed secreted factor called pigment epithelium-derived factor (PEDF), which affects a bewildering array of bodily processes. The cellular receptors that bind PEDF are at the root of many useful activities, such

as preventing the formation of new blood vessels, protecting cells in the retina and brain from damage, and stopping cancer cells from growing. Despite considerable effort by scientists in academia and industry, the identity of the receptors had, until recently, remained a mystery. This long-standing puzzle was so intractable that it took HHMI Early Career Scientist Hui Sun and his team at the University of California, Los Angeles, seven years to solve.

“We were initially intrigued by this problem because of PEDF’s broad medical value in treating major diseases,” says Sun. “But the hunt for the receptor turned out to be an extremely challenging adventure.” His team methodically searched for PEDF’s binding partner by looking at various native tissues and cells, as

well as in databases of molecules with unknown functions. Finally, they found two proteins that fit the bill – PLXDC1 and PLXDC2.

As Sun’s team reported

December 23, 2014, in eLife, these two proteins confer cell-surface binding to PEDF, transduce PEDF signals into distinct types of cells that respond to PEDF, and function in a cell type-specific manner. Because of the critical role of these cell-surface receptors in PEDF signaling, they have high potential as therapeutic targets for cancer and other diseases.

UNIVERSAL PROTEIN SYNTHESISThe process of gene expression is, at its core, universal. But different organisms use different methods to carry out the “DNA makes RNA and RNA makes protein” recipe. For example, the signals used to recruit ribosomes and start protein synthesis in bacteria are not the same signals used by eukaryotes. Yet much of the structure and function of the ribosome, the molecular machine that translates RNA into proteins, is conserved in both types of organisms. This made HHMI

Early Career Scientist Jeffrey Kieft at the University of Colorado Denver wonder if a universal start signal existed that could be recognized by both eukaryotic and prokaryotic ribosomes.

A structured region of an RNA molecule dubbed an internal ribosome entry site (IRES), which is used by some viruses to initiate translation in eukaryotes, fit the bill. Taking a close look at the molecule, Kieft’s team discovered that it could also initiate protein synthesis in prokaryotes. This particular IRES, it turns out, binds both bacterial and eukaryotic ribosomes in a similar structure-dependent manner, but the bacterial interaction appears transient and weaker.

The findings, published March 5, 2015, in Nature, suggest there might be other naturally occurring structured RNAs that can initiate bacterial protein synthesis. One of Kieft’s next steps will be to determine if other such structures do indeed exist, and, if so, what they mean in terms of evolution and gene regulation.


I N B R I E F

Scie

nce

imag

e:R

epri

nte

db

yp

erm

issi

onfr

omM

acm

illa

nP

ub

lish

ers

Ltd

:Nat

ure5

18,6

1-67

,cop

yrig

ht2

015;

Il

lust

rati

on:L

eifP

arso

ns

Salvaging SNAREsScientists catch a protein-recycling machine in the act.cellsarehuge proponentsofrecycling.Theyreuseeverythingfromtinyelectronstomassivemolecularcomplexes.Despitethispenchantforsalvage,scientistshaveneverbeenabletoseeoneformofcellularrecyclinginaction–untilnow.AgroupledbyHHMIInvestigatorAxelBrungerrecentlycapturedaproteinintheactofpullingapartaspentclusterofmembranefusionmolecules.

“Membranefusionisaprocessakintothemergingofsoapbubblesintolargerones,”saysBrunger,astructuralbiologistatStanfordUniversity.“Butthat’swheretheanalogyends,sincebiologicalmembranesdonotmergeallthateasily.”Tofacilitatetheprocess,

cellsusemembrane-boundSNAREproteins.Duringfusion,SNAREslocatedonopposingmembranesziptogetherintoastablecomplex,linkingthemembranes.Whenthetwomembraneshavebecomeone,aproteincalledNSFrecyclestheSNAREcomponents.

Brunger’steamusedatechniquecalledsingle-particleelectroncryomicroscopytofreezeNSFmoleculesboundtoSNAREcomplexesatseveralstagesandthencaptureimagesofthem.Theresultingsnapshotsprovidenear-atomicorbetterdetailoftheSNAREdisassemblyprocess.Likeaseriesofmoviestills,theimagesshowNSFlatchingontoaSNAREcomplex,thenNSFboundtoATP–themoleculethatpowersthesalvageoperation.YetanotherimagecapturedNSFafterithadfinishedworking,boundtoanenergy-depletedformofATP,calledADP.

TheSNAREcomplexresemblesaropewithaleft-handedtwist;theteam’simagesrevealedthatNSFusesadapterproteinscalledSNAPstograspthe“rope”inmultipleplaces.TheSNAPswraparoundtheSNAREcomplexwitharight-handedtwist,suggestingthatthedisassemblyoccursviaasimpleunwindingmotionthatfreesthezippedSNAREproteins.

Theresults,publishedFebruary5,2015,inNature,raiseotherquestionstheteamiseagertopursue.“ThereisalottobedoneinordertounderstandthemotionsandconformationalchangesneededtodisassembletheSNAREcomplex,”saysBrunger.“Ourelectronmicroscopestructuresnowenableustodesignfollow-upbiophysicalexperimentstoanswerthesequestionsataverydeeplevel.”– Nicole Kresge

A cell’s NSF complex uses SNAP proteins (orange) to grasp and unwind membrane-embedded SNARE proteins (red and blue helices).

12Like the creases in a sheet of origami paper, the stripes that pattern these fruit fly embryos guide forces that will govern their shape. As an embryo becomes longer and thinner, lines of protein are laid down along the growing fly’s head-to-tail axis. Each vibrant band consists of one of three members of the Toll receptor family. The receptors allow the cells to differentiate themselves from their neighbors and perform the directional movements necessary to remodel the growing embryonic tissue. A

dam

Par

é an

d Je

nn

ifer

Zal

len

HHMI Bulletin / Spring 2015

Observations

Vic

to N

gai

With Theory Comes OrderThe gray matter of the human brain may not be as physically vast as the dark matter of outer space. But the challenge of crunching astronomical data pales compared to the complexity of parsing today’s deluge of data about the brain. Making sense of that torrent requires more than banks of supercomputers, says computational neuroscientist and essayist Olaf Sporns. To achieve fundamental insights, he posits, neuroscientists must do as astronomers do and apply a theoretical framework to their immense quantities of empirical data.

Neuroscience, it turns out, is on the brink of its own “big data” revolution. Supplementing more traditional practices of small-scale hypothesis-driven laboratory research, a growing number of large-scale brain data collection and data aggregation ventures are now underway, and prospects are that this trend will only grow in future years. Mapping the roughly one hundred trillion synaptic connections of a human brain would by some estimates generate on the order of a zettabyte of data (a zettabyte corresponds to one million petabytes). For comparison, that’s about equal to the amount of digital information created by all humans worldwide in the year 2010. And recently proposed efforts to map the human brain’s functional activity at the resolution of individual cells and synapses might dwarf even these numbers by orders of magnitude. The coming data deluge is likely to transform neuroscience, from the slow and painstaking accumulation of results gathered in small experimental studies to a discipline more like nuclear physics or astronomy, with giant amounts of data pulled in by specialized facilities or “brain observatories” that are the equivalent of particle colliders and space telescopes.

What to do with all this data? Physics and astronomy can draw on a rich and (mostly) solid foundation of theories and natural laws that can bring order to the maelstrom of empirical data. Theories enable significant data reduction by identifying important variables to track and thus distilling a torrent of primary data pulled in by sophisticated instruments into interpretable form. Theory translates “big data” into “small data.” A remarkable example was the astronomer Edwin Hubble’s discovery of the expansion of the universe in 1929. Integrating

over years of observation, Hubble reported a proportional relationship between redshifts in the spectra of galaxies (interpreted as their recession speeds) and their physical distances. Viewed in the context of cosmological models Albert Einstein and Willem de Sitter formulated earlier, these data strongly supported cosmic expansion. This monumental insight came from a dataset that comprised less than fifty data points, compressible to a fraction of a kilobyte. When it comes to applying theory to “big data,” neuroscience, to put it mildly, has some catching up to do. Sure enough, there are many ways of analyzing brain data that are useful and productive for extracting regularities from neural recordings, filtering signal from noise, deciphering neural codes, identifying coherent neuronal populations, and so forth. But data analysis isn’t theory. At the time of this writing, neuroscience still largely lacks organizing principles or a theoretical framework for converting brain data into fundamental knowledge and understanding.

From The Future of the Brain: Essays by the World’s

Leading Neuroscientists, edited by Gary Marcus

and Jeremy Freeman. © 2015 by Gary Marcus and

Jeremy Freeman. Reprinted with permission

from Princeton University Press.

4000 Jones Bridge Road Chevy Chase, Maryland 20815-6789www.hhmi.org

Address Service Requested

Aru

n S

amp

ath

ku

mar

an

d E

llio

t Mey

erow

itz

Plant PaversJigsaw puzzle-shaped cells bulging with chlorophylls (red) cover

the surface of leaves in many flowering plants. The fanciful pattern is more than just a whimsical quirk of nature, however. HHMI Investigator

Elliot Meyerowitz and his team have shown that, as plants grow, the indented regions and lobe-like outgrowths of these so-called pavement cells create internal stresses that reorganize cytoskeletal proteins. These microtubule proteins align along the cell walls (green), a process that in

turn reinforces them against stress – evidence of a subcellular mechanical feedback loop. Learn more about the role of force in the development of

animals, as well as plants, in “Show of Force,” page 12.

bulletinh

hm

i bu

llet

in • h

ow

ar

d h

ug

he

s me

dic

al in

stit

ut

e • w

ww

.hh

mi.o

rg

vo

l. 28

/ no

. 2

Spring ’15 Vol. 28 No. 2

Small Forces,Big Impact

Mechanical stress plays avital role in the development

of both animals and plants

in this issue Neural Theory at Janelia

Marta Zlatic in the FootlightsBacterial Barcodes

spring ’15 vol. 28 no. 2 bulletin - hhmi.orgneuroscientists must do as astronomers do ... grow in...

Documents