Anal. Chem. XXXX, xxx, 000-000
Droplet-Based Microfluidic Systems for 1 j High-Throughput Single DNA Molecule Isothermal
I Amplification and Analysis
Linas Ma~utis,~Ali Fallah AraghiytOliver J. '~iller,~Jean-Christophe E3aretYtLucas Frenzyt Agnes ~anoshazi,~~~Valerie Talyjt Benjamin J. I\Ailler,* J. Brian Hutchison,* Darren Link,* Andrew D. Griffiths,+')*and Michael Ryckelynckts*
Institut de Science et d'lngenierie Supramoldculaire (ISIS), Universite de Strasbo urg, CNRS UMR 7006, 8 alle'e Gaspard Monge, 67083 Strasbourg Cedex, France, RainDance Technologies, Inc., 44 Hartwell A venue, Lexington, Massachusetts 0242 1, lnstitut de Genetique et de Biologie Moleculaire et Cellulaire (IGBMC), Universite de Strasbourg, CNRS UMR 7104, 1 rue Laurent Fries, 67404 lllkirch Cedex, France
We have developed a method for high-throughputisother- DNA nlolecules can also be col~~parln~entalizedin microdroplets ma1 amplificatioii of single DNA molecules in a droplet- in water-in-oil en~t~lsious,%hichact as microreactors with volulnes based microfluidic system. DNA amplification in droplets down to 1fL.%igital PCR in einulsions (emulsion PCR) is already was analyzed using an intercalating fluorochrome, allow- used to quantily rare mutations using BEAMing7 and to prepare ing fast and accurate "digital" quantification of the tem- the teinplate for two con~n~ercialized"next-generatio11~' DNA plate DNA based on the Poisson distribution of DNA secluencing syste~ns.~ molecules in droplets, The clonal amplified DNA in each. Direct, quantitative screening using emulsion PCR is, however, 2 pL droplet was further analyzed by measuring the comprolnised by the polydispersity of l~ulkemttlsions, Further- enzymatic activity of the encoded proteins after fusion with more, it is difficult to add reagents to droplets after they are a 15 pL droplet containing an in vitro translation system, formed," which limits the range of ass'ays that can be perfor~ned on the atllplifiecl DNA. However, both of these problems can Digital PCR is based on the Poisson distribution o.f DNA potetltially be overcome using rlroplet-based inicrofluidic systems tnolecules in microtiter plate wells (eq 1; in which Y(X = k) is that allow the production oi highly nionodisperse droplets%nd the probabilily to have lz DNA n~oleculesper well, and A is the pailwise droplet electrocoalescence,10014 mean nuinber of DNA n~oleculesper well). Digital PCR has previously been used to quanlily DNA in droplets in niicrofluidic This manuscript, however, eAap describes the digital quantification of DNA using a droplet-based Y (X = lz) = - A?! (1) inicrofluidic system and isothermal "l~yperbrancheclrolling circle
At low /Z (t0.3)) the vast majority of wells contain no illore than a (5) 'l'~rwiik, D. S.; Grifiitl~s,A. D.Nat. Bioteclttrol. 1998, 16, 652-656. single DNA molecule, and fitting the number of PCR competent (6) Gsifiiths, A. I>.; 'hwfilr, I). S, Tretzlls Bloteclt~iol2006, 24, 395-402, wells to eq I allows the DNA concentration to be calculated, Digital (7) Drcssman, I).; Yan, 14.; Traverse, C;.; I 10.1021/ac900403z CCC: $40.75 Q XXXX American Chemical Society Analytical Chemistry, Vol. xxx, No. xx, Month XX, XXXX A Anal. Chem, 2009, 81, 41 69-41 73 uidics: On thelS ope of En ightenmemt Rajendrani Mukhopadhyay Now tliat the hype hzs blown over, will microfluidics live up to its promise of providing marltetable applications? ('Olze coztld argzie that miniaturized che~nicalaftal-ysis systevzs ala just a fashionable craze. However, it is dz3cult to foresee tlze impact a tzew teclznological concept will have, zulzen it is in its early stages of develofimetzt," -Andreas Manz and colleagues (Ch.i?mia 1991, 45, 103- 105). Pessi~nislnand optilnisn~walk hand in hand in the field of n~icrofluidics.Some experts feel that the research endeavor is in danger of falling into a rut. Andreas Manz, curreiltly at the University of Ii'l.eil)urg (Germany), feels a twinge of disappoint- ment with the current state of affairs. He had bought into the hope carried by the first wave of patents, scientific publications, and start-up companies that occurred before 2000. But now he tllinlts, "It's veiy well possible that ~nicrofluidicswill end up in a never-ending cycle where it eventually reaches a steady state, just running t-hrough all ltinds of [academic] applications or all sorts of technical innovatio~lswith no major outlkt." Others take a different attitttde. Micro-fltridics, in and of itself, won't sell. What.wil1sell are the solutioas it: can provide to certain analytical probleil~s.'?IVhen you go to the cornlnercial landscape, technology doesn't tnalte money," states the University of Wash- ington's Daniel Chiu, who helped fo'outld the cotnpany Cellectricon. "Applicatiol~smalte money." The wildly vaiying views of the field are dizzying. Bill: Holger Beclter at the lnicrduidic Chip Shop (Germany) explains that this rift is pelfectly normal, He says that microfluidics is traclting the Gartner Hype Cycle (Med,Device Ikchnol. 2008,19,21-24), inflatecl expectations" crested later that clecade when the hype about what microfluidics cotrld achieve was at ils zenith, According THE HYPE CYCLE to Gartner, during such a peak "tl~eearly publicity produces a Gartner, an information technology research and advisoly comn- number 01 success stories, often accompanied by scores of pany, introduced their Hype Cycle Model in 1995. The model has failures." In the case of microfluidics, the poster child of a five stages: a technology trigger, a peak of inflatecl expectations, successful application was the Agilent Technologies-Caliper Life l a trough of disillusionment, a slope of enlightenment, and finally, Sciences Bioanalyzer. a plateau of produ.ctivity. Over the years, Gartner has applied the Microfluidics experts think that the field's "trough of disil- nioclel to Inany arenas, including the auton~otiveindustiy, Web lusionment" probably happened in the early 2000s, when the 2.0, and higher-education institutions. technology failed to meel inflated expeclations. "Because of The ~nicrofluidics"technology trigger" happened in the early (he bad rap inicrofluidics got at the end of the 1990s, with Ihe 1990s when the early pul~lications11y Manz and his colleagues promises and the laclt of clelively, it was a vely challenging 111ndiug lticlted off the field. Everyone agrees that t.he field's "pealc of environtnent for microfluidics entrepreneurs in the wake of that," 10.1021/ac900638w CCC: $40.75 O 2009 American Chemical Society Analytical Chemistry, Vol. 81, No. 1 I, June 1, 2009 41 69 Published on Web 05/07/2009 The EMBO Journal (2009) 28, 1208-1219 1 2009 European Molecular Biology Organization 1 All Rights Reserved 0261-4189109 www.embojournal,org THE EMBO JOURNAL ' Bacterial cell curvature through mechanical control of cell growth Matthew T Cabeen", Godefroid Charbon". growth depends on cell wall growth. The bacterial peptidogly- Waldemar ~ollmer~,Petra ~orn*, can cell wall is a covalently closed meshwork of rigid glycan Nora ~usrnees~,Douglas B weibe14 strands cross-linlzed by relatively flexible peptide bridges and Christine ~acobs-~agner"*~~~** (Vollmer et al, 2008). It thereby forms a strong but elastic fabric that resists internal turgor pressure and prevents cell '~e~artmentof Molecular, Cellular and Developmental Biology, Yale bursting. Under normal growth conditions, cells are turgid, as University, New Haven, CT, USA, 2~nstitutefor Cell and Molecular Biosciences, Medical School, Newcastle University, Newcastle upon the force exerted by turgor pressure maintains the peptidogly- 5ne, UK, 3~epartmentof Cell and Molecular Biology, Uppsala can in a stretched configuration (IZoch, 1984; Koch et al, 1987; University, Uppsala, Sweden, 4~epartmentof Biochemistry, University Baldwin et al, 1988). To expand, bonds in the peptidoglycan of Wisconsin-Madison, Madison, WI, USA, 'section of Microbial - must be hydrolyzed-to generate-insertion sites for new wall Pathogenesis, Yale Scliool-of Medicine, New Haven, CT, USA and '~oward Hughes Medical Institute, Yale University, New Haven, CT, USA material. The stretching from turgor pressure helps overcome molecular cohesion, facilitating bond hydrolysis and thereby The cytoskeleton is a key regulator of cell morphogenesis. providing a source of energy for cell wall growth (Koch et al, Crescentin, a bacterial intermediate filament-like protein, is 1982; Harold, 2002). To preserve the integrity of the peptido- required for the curved shape of Caulobacter crescentus and glycan fabric, cell wall hydrolases and synthases coordinate localizes to the inner cell curvature. Here, we show that their activities by forming complexes (Romeis and Holtje, 1994; crescentin forms a single filamentous structure that col- von Rechenberg et al, 1996; Holtje, 1998; Schiffer and Holtje, lapses into a helix when detached from the cell membrane, 1999; Vollmer et al, 1999; Bertsche et al, 2006). suggesting that it is normally maintained in a stretched The location and timing of new wall insertion is critical for configuration. Crescentin causes an elongation rate gradient generating nonspherical shapes, as turgor pressure is non- around the circumference of the sidewall, creating a long- directional (Harold, 2007; den Blaauwen et al, 2008). In most itudinal cell length differential and hence curvature. Such rod-shaped bacteria, elongation occurs by insertion of pepti- curvature can be produced by physical force alone when doglycan material along the sidewall of the cell, with little to cells are grown in circular microchambers. Production of no insertion at the poles (de Pedro et al, 1997; Daniel and crescentin in Escherichia coli is sufficient to generate cell Errington, 2003). Cytoskeletal proteins of the actin family curvature. Our data argue for a model in which physical (MreB and its homologs), which form helical structures along strain borne by the crescentin structure anisotropically alters the long axis of the cell beneath the cytoplasmic membrane the kinetics of cell wall insertion to produce curved growth. (Jones et al, 2001; Shih et al, 2003; Figge et al, 2004), are Our study suggests that bacteria mayluse the cytoskeleton thought to spatially restrict insertion of new peptidoglycan to for mechanical. control of growth to alter morphology. the sidewall (Daniel and Errington, 2003; Figge et al, 2004; The EMBO Journal (2009) 28, 1208-1219. doi:10,1038/ Dye et al, 2005; Kruse et al, 2005; Carballido-Lopez et al, emboj.2009,Gl; Published online 12 March 2009 2006; Divalzaruni et al, 2007). Subject Categories: cell & tissue architecture; microbiology Vibrioid (curved rod) or helical ~norphologiesare also com- & pathogens- mon among bacteria, but their morphogenetic mechanisms are I 1208 The EMBO Journal VOL 28 1 NO 9 1 2009 02009 European Molecular Biology Organization ------DinB Upregulation Is the Sole Role of tlie SOS Response in Stress-Ind... littp://www.genetics,~rg~ezp-prodl,hul.harvard.edu/cgi/content~abs~ac... b gA A Publication of The Genetics Society of America lseardh--.------a"" - . - -" ". - "-- .- --- .--. - --- - EE-NETIC,1 Advanced Search Home Journal Information Subscriptions & Services Collections Previous Issues Current Issue Future Issues Institution: Harvard Libraries I Sign In via User NamelPassword Originally published as Genetics Published Arlicles Ahead of Print on March 6, 2009. THIS ARTICLE Genetics, Vol. 182,55-68, May 2009, Copyright O 2009 Fill1 Text doi:10.1534/genetics.l09.100735 Full Text (PDF)-- . Supporting Information DinB Upregulation Is the Sole Role of the SOS Response All Versions of this Article: in Stress-Induced Mutagenesis in Escherichia coli yenetics.109.100735vl 18211I55 most recent Rodrigo S ~alhardo*,Robert DO*,Masiuni yamadat, Errol C. . Alert me when this article is cited ~riedbe~~t,P;J. ~astiilgs*, Takehiko and Susan M. ------oh mi# Alert me if a correction is posted - ~oseilbel*~*~~~1 SERVICES * Department of Molecular and Human Genetics, Baylor College of Medicine. Houston, Texas 77030-3411, 'Department of Pathology, University of Texas Southwestern Medical Center. Dallas. Texas 5390-9072, 'Division of Genetics and Similar articles in this journal Mltagenesie, National Institute of Health Sciences, Tokyo 158-8501, Japan and 'Depamenls of Biochemislty and Molecular Biology and Molecular Virology and Mcrobiology, The Dan L. Duncan Cancer Center, Baylor College of Medicine, Houston, Similar articles in Pi~bMed Texas 77030-3411 Alert me to new issues of the j_qurnal 1 Corresponding author: Department of Molecular and Human Genetics, Baylor College of Medicine, 1 Baylor Plaza, Room Downloacl to citation manager S809, Mail Stop BCM225, Houston, TX 77030-3411. E-mail: [email protected] Stress-induced mutagenesis is a collection of mechanisms observed in bacterial, yeast, and human GOOGLE SCHOLAR cells in which adverse conditions provoke mutagenesis, offen under the control of stress responses. Articles by Galhardo, R. S. Control of mutagenesis by stress responses may accelerate evolution specifically when cells are Articles by Rosenberg, S. M. - maladapted to their environments, i.e., are stressed. It is therefore important to understand how stress responses increase mutagenesis. In the Escherichia coli Lac assay, stress-induced point PUBMED -- - - " mutagenesis requires induction of at least two stress responses: the RpoS-controlled Pubyed Citation generallstarvation stress response and the SOS DNA-damage response, both of which upregulate Articles by Galhardo, R. S. DinB error-prone DNA polymerase, among other genes require'd for Lac mutagenesis. We show that Articles by-Rosenberg, S. M. _ - upregulation of DinB is the only aspect of the SOS response needed for stress-induced mutagenesis. We constructed two dinB(oc) (operator-constitutive) mutants. Both produce SOS-induced levels of DinB constitutively, We find that both dinB(oC) alleles fully suppress the phenotype of constitutively SOS-"off' lexA(1nd-) mutant cells, restoring normal levels of stress-induced mutagenesis. Thus, din6 is the only SOS gene required at induced levels for stress-induced point mutagenesis. Furthermore, although spontaneous SOS induction has been observed to occur in only a small fraction of cells, upregulation of din6 by the din6(oc) alleles in all cells does not promote a further increase in mutagenesis, implying that SOS induction of DinB, although necessary, is insufficient to differentiate cells into a hypermutable condition. Help I Contact Us l~iterriationalAccess Link Online ISSN: 1943-2361 Print ISSN: 0016-6731 Copyright 2009 by the Genetics Society of America phone: 412-268-1812 fax: 412-268-1813 email: [email protected]~~,edu JOURNALOF BACTERIOLOGY,June 2009, p. 3892-3900 Vol. 191, No. 12 I 0021-9193/09/$08.00+O doi:10.112S/JB.00975-08 Copyrigllt O 2009, American Society for Microbiology. All Kiglits Ziescrved. The Dienes Phenomenon: Competition and Territoriality in Swarming Proteus mirabilisV A. E. ~udding,'t*C. J. ~ngharn,~-fW. ~itter,'C. M. ~andenbroucke-~riluls,~and P. M. schneeberger2 VU Me~fiiicnlCelzire, Anzsterclm?~)Tlze Netizerl~~rzcf,~)and Jeroeiz Bosclz Hospital, k Hertogenbosciz, Tlze ~eti~erl~~i~ds~ Received 15 July 2OOSIAccepted 11 February 2009 When two different strains of s~mr~iiingProteus mirabilis encounter one another on an agar plate, swarniing ceases and a visible line of dcinarcation forms, This boundary region is known as the Dicnes line and is associated with the forniation of rounded cells. While the Diencs line appcars to bc the product of distinction between self and nonself, many aspects of its formation and function are unclear, In this work, we studied Dicncs line fonnation ~isirigclinical isolatcs labeled with fluorcscent proteins, Wc show that round cclls in the Dicnes linc originate exclusively from one of the swarlns involved anil that-thcsc round cclls have decreased viability. In this sense one of the swarliis involved is dominant over the other. Close cell proximity is required for Dienes linc formation, anil when strains initiate swarlning in close proximity, the doinillant Dienes typc has a significant conipetitive advantage. When one strain is killed by UV irradiation, a Diciics line does not fonn, Killing of the doniinant strain liniits the induction of round cells. We suggest that both strains are actively involvccl in bouiidary foriliation and that roond cell forination is tlic result of a short-range killing nieclianisni that mediates a coinpetitive advantage, an advaiitage highly specific to the swarming state, Dicncs line forniation has implications for the physiology of swarniing and social recognition in bacteria. The gram-negative bacterium Proteus rtzirwbilis is well ltriowii Dienes line formation. In the 1970s Senior typed strains of P. for its ability to difFerentiate into liyperllagellated, motile, and nzirabilis based on their proticine procluctioi~ and seilsitivity elongated swarlner cells that rapidly spread over a surface. (23, 24). He found that proticine production is not related to When cultured on a nutrient agar plate, a strain of P. nzir(1bilis proticine sensitivity (apart Srom strains being resistant to their typically is able to coloiiizc the whole plate within 24 h. This own proticine) but that there is a good correlation between a pl~enornenonis both of interest in tcrrns of the diffcrentiatioii coinbiiicd proticine production-sciisitivity (PIS) type and and survival strategy 01 the organisin aiid of practical impor- Dienes line formation. Strains with the salnc PIS type do not tance, as contamination,of agar plates by swarming P. mirc~bilis forin a Dienes line, whereas strains with diirerelit P/S types do. is a coninion problem in diagiiostic microbiology. When two The inorc closely related tlie PIS types of two strains, the less different strains of P, n~irabilisswarm on the same agar plate, clearly defined the Dicnes line is, suggesting that relatedness of a visible deniarcation line with lower cell density Eorlns at the strains plays a centsal role. In contrast, a strain of Proteus intessection, and this line is known as a Dienes line (5, 6). A vulgzrris has bccn show4 to be Dienes compatible with a strain Dienes line is seen when both strains are swarming; it is not a of P. 17zirclbilis with tlie same PIS typc (24). Furthermore, property of the snlaller vegetative cells (5). When two identical Dielies incoinpatibility between otherwise identical straiiis call isolates meet, the swarniing edges merge without forination of be triggered by phage lysogeiiy (2). I11 contrast to P/S type, the a Dicnes line, This pl~enomciionhas been used in epideinio- polysaccharide (0)or flagellar (H) scrotype has been shown to logical typing of clinical isolatcs (4, 20, 23, 28) and raises have no relatio~ishipto Dienes line fornlatioii (25, 27). interesting questions concerniiig its inechanisin and biological Research into tlie mechanism governing Dieiles line Sornia- iniportance. Dielies typing in the clinical eiivironinent suggests tion in Proleus lzas revealed soinc i~nportaiitfeatures. The that tlie ~luniberof Dieiies types is large; 81 types were found Dieiies line region contains niaiiy large aiid often rounded cells in oilc study aloiic (25). Incompalibility bctwccil swarming (5). The iiature of these cells reinains controversial, Dienes straii~sinay not be uniclue to 1'. mirnbilis; a co~nparableprocess suggested that r~undcells originated from both strains but that also appcars to occur in I'seudor.zronn,s crerugi~~osa( 19). Like viable rouiid cells always originated fro111 one of the two in tcr- nlaiiy other bacteria, some P. mirnbili,~strailis produce bacte- sectiiig strains. He concluded from this that nonviable round riocins, termed proticiiies (3). It has been shown that Dielies cells should therefore be cells of the otllcr strain (6). In con- line forlnatioii is not directly caused by a proticinc, nor has any trast, Wolstcnhoinc suggested that the round cclls originated other sccrctcd substance or cellular lysate been found to con- mostly from one of the two swarins iiivolved and that only 50% tribute to this plicnotneiion (27). There docs, however, seen1 to of these cclls were viable (32). It has also been noted that bc a circumstantial link between proticine production and round cclls occasionally seem to develop into stable L forins lacking a full cell wall and that they can grow to lorn1 tiny * Corresponding author. Mailing address: VU Medical Centre, De- L-type colonies (5). Additionally, extracellular DNase has been pastmcnt of Medical Microbiology and Infection Preveotiotl, P.O. BOX found at the sito of the Dicllcs line (1). ~l~~ presence of this 7057, 1007 MB Alllslerdam, The Nell~erlfinds. Pllolle: 0031-20- 4440457. Fa:0031-20-4440473, E-mail: cl,budding(@vumc.nl. enzylne has beell intelaPreted to be a of cell lysis, as t A.E.B. and C.J.1, con~ribu~edequally to this work. ~nirl~bilisis ltnown to contain large ainouiits of DNase (22, 26). "Published ahcacl of print on 27 Februa~y2009. Recently, a cluster of six genes, termed ids (identification of ------JOURNALOF BACTER~~OGY,JUG 2009, p. 3437-3444 Vol, 191, No. 11 0021-9193/09/$08.00+O doi:10.1128/JB.00034-09 Copyright O 2009, America11 Society for Microbiology. All Kights Reserved. Gene Expression Profiling and the Use of Genome-Scale In Silico Models of Escherichia coli for Analysis: Providing Context for contentV Nathan E. ~ewis,'Byung-Kwan ~ho,'Eric M. night,^ and Bernhasd 0.~alsson"~ Department of Bioefzgineerii?g, University of Culifor~zia,Sc~z Diem 9500 Gilnznn Drive) Mcijl Code 0412, La Jollu) Cc(lgo,nia 92093-0412) ancl Cenler for Sysle~nsBiology, University oj' Iceland, Valnsr~zyroveglrr16) 101 Reylcjavilc) Icelarrd2 One of the most widely used high-throughput tecliiiologies is nome-scale analysis of cliffcrential gene expressioii patterns the oligonucleotide microarray. From the initial development (21), clustering and classification (17), and gene set enrich- of inicroarrays, liigh expectations were held for their use to aid ment (76) have been successfully deployed. Overall, our in t-mswcrjiig biological questions, clue to their ability to mca- experience shows that informative results can also be ob- sure lnRNA abifndances on a ge%omc sale, 'FTowevcr, accu- tained if data are analyzed in tlie context of abei1oiKe-scale mulating experience is revealing that even when questions of network reconstructioii and with the use of an in silico sample preparation, data processing, and dealing with the in- model. Furthermol+e, greater success in data analysis has herently noisy data (81) are set aside, the large amount of data been achieved on a fine-grain level for more specific hypoth- generated has proven difficult to analyze and interpret (12). It esis generation and assessment than on a coarse-grain ge- is also oftell challenging to narrow down specific novel findings nome-scale level. The lessoils learned from the alialysis of bascd solely on cxprcssion proliling data. this data set can be classified by the granularity of the Here, we present a downloadable compendiuni of gene ex- analysis and the level of use of the in silico model. pression profiles for Esclzericlzi~zcoli and discuss the experience (i) Genome-scale analysis without using a liiodeling frame- from one lab in which expressio~iprofiling data have been work, E. coli grown to mid-log pl~aseon various carbon sources employed in a myriad of studies of E. coli. We will try to and/or with various gcilc knockouts has becn shown to usually address two classes of expression profiling data usage: (i) how evolve a growth phenotype as predicted computationally. How- expression profiling can be analyzed using inore traditional ever, only a modest level of' understanding of the optimal statistical nletllods to provide biological understanding and (ii) growth phcizotype was obtained through the analysis of gene how genome-scale inodels form a context within which expres- expression data alone, sion profiling data content increases in value. (ii) Regulon- and network-level analysis in the inodeling fi*amework. Several studies in which computational models have been tested to predict global properties of E. coli metab- THE DATA olism aiid transcriptional regulation, to suggest pathway usage We present a database of 213 expression profiles produced in strains that clo not grow at the optiii~alrate, and to predict 'in our laboratory aiid used in inany of the case studies reviewed gene expression changes under growth conditioli shifts have hcsc. Thcsc profiles rcprcscnt mcasurcmcnts for about 70 bcci~conducted. Here, expression profiling data have played a con~l~inationsof variations in experin~cntalconclitioils and ge- benelicial role in both validating compulational predictions netic variations in E, coli I<-12 MGZ655. In this set of expcri- and providing necessary ini'orniation to predict novel regula- inents, the following parameters were varied: (i) the carbon tory interactions, source, (ii) the terminal electron acceptor, (iii) the tempera- (iii) Gene-level analysis using in silico models, The most ture, (iv) the number of days the bacteria were grown at mid- inforn~ativeuse of gciic expression profiling data was found log phase, and (v) the genotype (wild-type and gene deletion when simulations predicted growth phenotypes that conilictcd strains were used). Tllcsc data and the corresponcling mini- with experimental obselvatiolls (growthliio growth), suggesting 1num information about a microarray experiment (9) may bc that undiscovered pathways still existed, Microarrays were freely dowi~loacleclat l~ttp://syste~1~sbiology.ucsc1~edu/In~Silicoused in this setting to identify individual genes that may con- -Orgai~is1ns/E_coli/E~coli~expression2. tribute to con~pulatioiiallypredicted pathways. Each of these lcvcls of analysis will be discussed and exam- ples will be provided below, LEVELS OF ANALYSIS Mucll insight into the function of E. coli has been gaincd over tl~epast decade through the statistical analysis of GENOME-LEVEL ANALYSIS WITHOUT USING IN cDNA and oligoiiucleotide arrays. Methods such as the ge- SILICO MODELS Characterizing disparate paths to coinnioli end points in adaptive evolution. Wild-type E. coli I<-12 has been evolved to Vo~~sespondingimthor. Mailing address: Dep~~rtrnc~~tof Biomgi- ilnProvc the growth rate by usillg serial passagillg in mid- to neering, University of California, Sa~iDiego, 9500 Gilmail Drive, Mail Code 0412, La Jolla, CA 92093-0412. Phonc: (858) 534-5668. Fax: late log phase of growth (45, 47). Zntcsestingly, when parallel (858) 822-3120. E-mail: palsson@ucsd,edu. cultures are evolvecJ in this way, their evolutionaty paths are "Published ahead of print on 10 April 2009. not the same, though they usually evolve to have similar growth JOURNALOF BACTERIOI,OGY,May 2009, p. 3 177-3182 Vol. 19.1., No. 9 0021-5,193/09/$08.00+0 doi:10.1128/JB.00011-09 Copyright O 2009, Amcricai~Society for Microbiology. All Kiglits Iiesewcd. The Stationary-Phase Sigma Factor oSIs Responsible for the Resistance of Escherichia coli Stationary-Phase Cells to rnazEF-Mediated Cell ~eath' I Ilana Kolodkin-Gal and Hanna Engelberg-Kulkah I Depnrtntent of Molec~rlarBiolohy, tlze Hebrew Univer,~ity-HudaswhMedical Sclzool, Jelzisulei~z91120, Israel I Received 5 January 2009JAccepted 11 February 2009 Bsclzericltin coli nznzEF is a toxin-antitoxin gene module that mediates cell death during exponential-phase cellular growth through either reactive oxygen species (R0S)-dependent or ROS-independent pathways. Here, we founcl that the stationary-phase signla factor aswas responsible for the resistance to snazEF-nlcdiated ccll death during s taaonaly growth phase. Deleeon of rpoS, the geneencoding crS f~o_mthe bacterial chr_onlosome, - pernlittecl mnzEF-mediated cell death during stationary growth phase. C Toxin-antitoxin systems have been found on the chromo- reactive oxygen species (ROS) (15). ROS have bcciz previously soines of many bacteria (8,10,23,27). One of the best studied inlplicated in prograinined cell death in eultaryotes (21, 29), among these cl~romosomaltoxin-antitoxin systems is Esclze- including in yeast (12, 17), in the liSe span of several organisms riclzici coli MINZEF,which was t11c first to be described as regu- (25,33), in the senescence 01 bacteria (G), and in the mode of latable and responsible for bacterially programmed cell death action of some antibiotics (13-15). It was previously reported (2). E. coli nzcizEF is located dowilstreanl from the re& gene that the stational-y-phasc sigma factor us,eilcodcd by ~poS(16, (18,20), specifying for ppGpp synthase (28). nzazF specif es the 19), positively regulates the Sornlatio~lof catalase and is re- stablc toxin MazF, while rtznzE specifics the labile antitoxin spoiisible for the elevated levels of this enzyme during station- MazE, degraded in vivo by the ATP-dependent ClpPA serine ary growth phase (24, 34, 35). Since calalase detoxilies ROS, proteasc (2). MazF is a sequence-specific endoribonuclease we asked whether resistance of stationary-plzasc cells to that preferentially clqaves single-stranded inRNAs at ACA mazEF-mediated ccll dcath was caused by the elevated Ievcls sequences (36, 37) and thereby inhibits translation (3, 37). of catalase produced at that time. So, we tested whether dc- MazE counteracts the action of MazF. Because MazE is a leting rpoS from E. coli cells would lead to their death through labile protein, prcvcntion of MazF-mediated action requires the 171crzEFsystern during stationary growth phase. Indeed, as the continuous production OS MazE. Thercforc, stressful we have predicted, in ArpoS cells we observed nzazEF-niedi- conditions that prevent tl~cexpression of the cl~romoso~nally ated ccll death even during stationary phase of growth, borne r?~azEFinodule permit the forination of free MazF and Wc used strain MC410relA+ and its Ar71aiEF::lcarz derivative thereby cell death. Tllese stressful coilditions include (i) the (9) and st]-ainMC4100ralA 'ArpoS, which we col~structedby PI use of antibiotics that are general inhibitors of transcription transduction from E, coli strain I<-38ArpoS::tet (kindly pro- and/or translation suclz as rifampin, chloramphei~icol,and vided by Sllosh Alluvia), and its AinctzEF derivative, which we spectinomycii~(31); (ii) extreme amino acid starvation, leading coilstructed by PCR deletion (4). We used plasnlid pQElcatE to the production of ppGpp that inhibits n~azEFtranscription (15), bearing thc catalasc-specifying IcatE gene, which is con- (2,7); and (iii) DNA damage caused by thymine starvation (32) tinuously expressed ill tllc strains describcd here, as well as by DNA-damaging agents like ~nitonlycinC or nali- We grew tllc bacteria in liquid M9 ininiinal ~ncdiu~nwith 1% dixic acid (1 I). Tllc use of these antibiotics and other stressful glucose and a mixture of ainino acids (10 ~-~g/nlleach) (22) and conditions are well known to cause bacterial ccll dcath ('1, 5); then plated then1 on rich LB agar plates, as we have clescribecl wc found that such ccll dcatlz taltes place through the action of previously (I I). the 17lazEF nlodule (31, 32). All the groups of stressful condi- Nalidixic acid, initonlycin C, trimethoprin~,rifampin, serine tions were Sound to trigger nzctzEF-mediated cell death by l~yd~oxarnate,chlorainphe~~icol, spcctinomycin, Trizma base, preveilting the continuous synthesis of MazE and thereby re- sodium dodecyl sulfate, DNasc, and RNasc wcrc obtained ducing its lcvcl (2, 31, 32). We wcrc surprised to firid that from Sigina (St. Louis, MO). Lysozymc was obtained from the nzazEF-mediated ccll death occul.s a t the,exponential stage of United Statcs Biocl~cinicalCorporation (Clcvcland, OH), Am- growth. but does not occur during stationaiy phase (1 1). picillin was obtained from Biochemie GmbH (I NOTES The Conserved Sporulation Protein YneE- Inhibits DNA Replication in Bacillus subtilisVt Lila11 ~ahn-~ee,'Boris ~orbat~uk,'Ole ~kov~aard,~and Richard ~osick'~'. . Department ojMoleculc~r and Cell~rlc1r Biology, Hc~rvardUniversity) Cc/n.zl?riclge,Mc~ssaclztlselts, ' and Depart~?zerzfof Scierzce, Sy,ctenzs arzd Mo(1el.s) Roskil~leUnivgaip, Ro0sIcilde)~elzn~ark~ '. Received 18 February 2009lAccepted 23 March 2009 Cells ofBncillus sribtiRs-triggered to spori~teu~~&rco~~~itionsof rapid growth ug&rgo a nlwked decileas_e in chronlosoine copy number, which was partially relieved by a mutation in the sl)orulation-induced geneyitel?. Cells cnginecrcd to express yneE during growth wcre inlpaired in viability and produced a~lucleatecells. We coilclude that YneE is an inhibitor of DNA replication. The ability of bacteria to adapt to a variety of growtl~con- the illaster regulator for entry illto sporulatioil (2). The dis- ditions demands inecllanisins for coiltrolling DNA replication covery that rapidly growing cells can be triggered to sporulatc and coordiilating chromoson~eduplication with the cell cycle. with high efficiency i~npliesthat B. s~lbtilismust llave an active Bacteria initiate one and only one round of DNA replication at inechanism to shut down DNA replication, given that rapidly the origin of chroinosolnal DNA replication, oriC, for each growing cclls contain multiple replication forks. An appealing cycle of cell division (1). During slow growth, the single copy of hypothesis for how such a mechanisnl might work steins from oriC is duplicated to create two copies of oriC. In contrast, the obseivation that the B. suhtilis origin of replication conlai~ls under conditions of rapid growth, wheil the time requised lo m~iltiplebinding sites for pliospl~orylatedSpoOA (SpoOA-P) duplicate a chrornosoine is longer than the generation time, and that many of these sites ovei-lap with binding sites for the bacteria inherit cl~ron~osomesundergoing active replication replication initiation protein DnaA (5) This suggests that with multiple replication forks. In this case, replication initia- SpoOA-P may impede replication by clirectly blockii~gthe tion occurs synchronously across all origins, leading to multi- action of DnaA. It has not been established, however, whether fork replication with 2" (rz = 1,2,3) copies of oriC present in SpoOA-P binds to the origin region in vivo and, if so, whether each cell (15). Milltifork replication eilsures that each ilewly this binding blocks the initiation of replication. We wondered, divided cell will receive at least one cl~romosome.The transi- therefore, whether SpoOA-P might additionally, or alterna- tion from one growtll condilion to another, such as when tively, block replication indirectly by turning on the expression nutriciits become liinitiiig and cells enter stationaiy phase, of a gene or genes whose product is an inhibitor of replication. requires lnechanislns to adjust the frequency of replication In prcvious work, we identified genes under the direct con- initiation (10). One iiotewortl~ybut poorly understoocl exam- trol of SpoOA, a sinall nu~nbcrof which encode proteins of ple of replicatioil control is exhibited by Bacill~~ssubtilis during unknown function and nlutatlts ol which do ilot exhibit con- the entry into sporulation, Sporulaling cells complete a final spicuous defects in spore formation (7, 13). We used this as a round of replication bebre undergoing a process ol asymmet- starting point to investigate whcther any previously uncharac- ric division that results in a forespore itnd a ~nolhercell (1 1, terized gencs ulldcr SpoOA-P control inhibit DNA replica- 14), Each of these progeny cclls orclillarily inherits a single tion. Here we report the discovery that a SpoOA-P-induced ch~.omosoinc,and inaturc spores, which are derived from the gene, yneE, for which we introduce the name sirA (for sporu- forespore, are known to contain only one chroinosoine (9, 14). latioii inhibitor of replication) cncocles a protein that inhibits Cells nornlally enter the pathway to sporulate in response to DNA nutrieilt limitation. We have previously shown, however, that To deternline if the inductioi~of sporiilation by actively artificial i~iductioiiof syntlicsis of lhc kiiiase KiiiA causes el'- growing cells blocks replication, we turiied on tllc synthesis of ficicnt sporulation even during growth in rich medium (7). IGnA in cclls in the tnidexponcntial phase of growth by placing KinA transfers phosphate groups into a n~ulticoinponentphos- IiirzA under the control of an IPTG (isopropyl-P-D-thiogalac- phorelay that phosphorylates the response regulator SpoOA, topyra11oside)-inducible proinoter and treating tho cells with IPTG. We n~easured chronlosorne conlent by staining for * Corrcspouding author. Mailing add~*ess:Departnlellt of Molecular DNA and analyzing the cclls with a flow cytometer, Either and Cellular Biology, Harvarcl Universily, 16 Divinity Ave., Cam- sa~nplcswcse fixed iin~ncdialclyfix* a "snapsl~ot"oS the total bridge, MA 02138. Phonc: (617) 495-4905. Fax: (617) 496-4642, amount of DNA presei~lat the time that the sample was E-mail: [email protected]. collcctcrl or the cells were allowed to continue growing ill thc t Supplemental illatcrial for this article may be found at http://jb .asm.org/. presence of cl~lorainphenicol,Bccadse clilorarnphenicol blocks ' Puhlishcd ahcod of print on 27 March 2009. protein synthesis, it allows ongoing rounds of replication to Genetic Suppressors and Recovery of Repressed Biocl~ernicalMemory ... http://www.jbc.org.ezp-prod1 .hul.harvard.edu/cgi/content/~l1/284/19... QUICK SEARCH: [advanced] HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS Institution: Harvard Libraries I Slan In via User Name/Password Originally published In Press as doi: 10.'I 074/jbc.X800017200 on Jamary 7,2009 J. Biol. Cl~eem,Vol. 284, Issue 19, 12585-12592, May 8,2009 This Article Reflections 1 FUII Text f PDF~ ) All Versions of this Article: 284/19/12585 most r-ecent Genetic Suppressors and Recovery of X800017200vl k Submit a Letter to Editor ------Repressed-Bioc-hemical-Memory - when this article is cited 1 Alert me when eLetters are tlosted Jon Beclwith Alert me if a correction is posted 1 Citation MaD From the Department of Microbiology and Molecular Genetics, Services Harvard Medical School, Boston, Massachusetts 021 15 ) Email this article to a friend F Similar articles in this iournal Similar articles in PubMed 1 Alert me to new issues of the iournal ) Download to citation manaaer k Reauest Permissions Google Scholar I 1 Articles bv Beckwith, J. I PubMed I PubMed Citation ' 1 Articles bv Beckwith. J. Social Bookmarking I ) INTRODUCTION I am I Many of the Reflections articles in this journal have focused on / = INTROI)IJ(I~'~ON particular biological problems that the authors have devoted their / Y Sunnressors, lac ODCI-on... -r Suppl*essowand l'rotein Export careers to or on particular techniques that have facilitated their /'1 -ol. Misleading Snpprcsson biological studies. Although I began my career as a biochemist, one of the most common threads to my research has been the exploitation of a genetic approach, the analysis of suppressor mutations. Particularly RgPE1'eNCes in the last 20 years, the success of this approach has led me into biological problems that required a remembering or relearning of my early training in biochemistry. This article will describe the voyage from chemistiy and biochemistry to genetics and then to a recovely of remnants of my biochemical memoly. As a graduate student worlting under Lowell Hager at Harvard in the late 1950s, my thesis research was chemical and biochemical, as I performed the chemical synthesis and studied the biosynthesis of the chlorinated mold product caldariomycin (2,2-dichloro-1,3-cyclopentanediol)(L-3). With his postdoctoral fellow Paul Shaw, Lowell had discovered an enzyme, chloroperoxidase, that was capable of Structural and Functional Similarities between a Ribulose- 1,s-bisphos... http://www.jbc.org.ezp-prod1 .hul.harvard.edulcgi/content/abstra~t/28,., Advertisemenl QUICK SEARCH: [advanced] Advertisement $6 ONLINE Author: ~ey-wsrd(s)_:-- ~~_.,-. """* - *. - ." " "3 HOME HELP FEEDBACK SUBSCRIPTIONS ARCHNE SEARCH TABLE OF CONTENTS Year: - ---- vol:-7 -- -:page: L-_I Institution: Harvard Libraries I Si~nIn via User NameIPassword Originally p~~blished111 Press as doi: 10.1074libc.M807095200 on March 11,2009 J. Biol. Chem., Vol. 284, Issue 19, 13256-13264,May 8,2009 This Article Structural and Functional Similarities between I FUII Texf Full Text (PDFI Advertisement a Ribulose-1,5=bisphosphate )- 1 All Versions of this Article: Carboxylase/Oxygenase (RUBisCo)-likee 284/ 19/13256 nrost recertt-- M807095200vl Protein from Bacillus subtilis and b Submit a Letter to Editor b Alert me when this article is cited Photosynthetic R~B~~co*" b Alert me when eLetters are oosted Alert me if a correction is posted Yolltaro snitox', Hiroki AsLida$ , Tomoko Sakiyamat , ) Citation Mao Nicole Tandeau de Marsacs , Antoine ~anchiny, Services Agnieszka sekowskay2, and ~kil~o~okota'~ b Email this article to a friend 1 Similar articles in this iournal $ b Similar articles in PubMed From the Nara Institute of Science and Technology, Graduate School of ) Alert me to new issues of the iournal Biological Sciences, 89 16-5 Takayama, Ikoma, Nara 630-0192, Japan and b Download to citation manaaer $ n b WestPermissions the Departments of Microbiology and Genetics of Bacterial Genomes, Institut Pasteur, 28 Rue du Docteur Roux, 75724 Paris Cedex 15, France Google Scholar ) Articles by Saito, Y. The sequences classified as genes for various ribulose- l,5-bisphosphate ) Articles bv Yokota, A. (RuBP) carboxylase/oxygenase (RuBisC0)-like proteins (RLPs) are PubMed widely distributed ainong bacteria, archaea, and eukaryota. In the PubMed Citation b Articles bv Saito. Y. phylogenic tree constructed with these sequences, RuBisCOs and RLPs b Articles bv Yokota, A. are grouped into four separate clades, forms I-IV, In RuBisCO enzymes Social Bookmarking encoded by form I, 11, and I11 sequences, 19 conserved amino acid IHI $9 2 $9 @ $3 residues are essential for CO2 fixation; however, 1-1 1 of these 19 What's this? residues are substituted with other amino acids in form IV UPS.Atnong form IV RLPs, the only enzymatic activity detected to date is a 2,3-diketo-5-methylthiopentyl l-phosphate (DK-MTP- 1-P) enolase reaction catalyzed by Bacillus subtilis, Microcystis aeruginosa, and Geobacillus kazrstophilus form IV RLPs. RLPs from Rlzodospirillu~~zrubrurn, Rhodopseudonzonaspalustris, Chlorobiurn tepidunz, and Bordetella bronclziseptica were inactive in the enolase reaction. DK-MTP-1-P enolase activity of B. subtilis RLP required M~~+for catalysis and, like RuBisCO, was stimulated by CO2. Four residues that are essential for the enolization reaction of RuBisCO, ~~s~~~,~s~~~~,and were conserved in RLPs and were essential for DK-MTP-1-P enolase catalysis. ~~s~~~,the residue conserved in DI<-MTP-1-P enolases, was also essential for B, subtilis RLP enolase activity. Similarities between the active site structures of RuBisCO and B. subtilis IUP were examined by analyzing the effects of structural analogs of RuBP on DK-MTP- 1-P enolase activity. A transition state analog for the RuBP carboxylation of RuBisCO was a competitive inhibitor in the DIG-MTP-1-P enolase reaction with a Ki value of 103 CLM,RuBP and D-phosphoglycericacid, the substrate and product, respectively, of RuBisCO, were weaker competitive inhibitors. These results suggest that the amino acid residues utilized in the B. subtilis RLP enolase reaction are the same as those utilized in the RuBisCO RuBP enolization reaction. Received for publication, September 12, 2008 , and in revised form, March 10,2009. Inhibitory Mechanism of Escherichia coli RelE-RelB Toxin-Antitoxin, ., http://www .jbc.org.ezp-prod1 .hul.harvard.eddcgi/content/abstract/28.,, QUICK SEARCH: [advanced] j?fkONLINE K-!~!!l!!?K!ls! )L"-.-1 ..""" *... "."."'.'...... - HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS, ...... Institution: Harvard Libraries ( Slqn In via User Name/Password Originally published In Press as doi: l0.1074/jbc.M809656200 on March 18,2009 J. Biol. Chem., Vol. 284, Issue 21, 14628-14636, May 22,2009 This Article Inhibitory Mechanism of Escherichia coli ). FUII Text RelE-ReLB Toxin-Antitoxin Module Involves a ,b 4Full Text PDF Advertisement ) All Versions of this Article: - -- Helix Displacement-Near an mRNA - 284/21/14628 7noStTecent - - M809656200vl Interferase Active site*" k submit a Werto Editor 1 Alert me when this article is cited t f § ). Alert me when eLetters are posted Guang-Yao Li , Yonglong Zlrang , Masayori Inouye , and b Alert me if a correction is posted Mitsulriko Ikura$1 b Citation Mau Services $ From the Division of Signaling Biology, Ontario Cancer Institute, and ). Email this article.... to a friend Department of Medical Biophysics, University of Toronto, Toronto, b Similar artdes n this iournal 5 b Similar articles in PubMed Ontario M5G 1L7, Canada and the Department of Biochemist~y,Robert ) Alert me to new issues of the iournal Wood Johnson Medical School, Piscataway, New Jersey 08854-5635 b Download to citation manaqer I Reauest Permissions In Escherichia coli, RelE toxin participates in growth arrest and cell Google Scholar death by inducing mRNA degradation at the ribosomal A-site under stress 1 Articles bv Li, G,-Y, conditions. The NMR structures of a mutant of E. coli RelE toxin, 1 Articles hy Ikura. M, el^^^^^^^^^, with reduced toxicity and its complex with an inhibitory PubMed peptide from RelB antitoxin, RelBc (L~S~~-L~U~~),have been ). PubMed Citatio~ b Articles bv Li, G.-Y. determined. In the free structure, helix n4 at the C el^^^^^^^^^^ ) Articles bv Ikura, M. terminus adopts a closed conformation contacting with the B-sheet core - Social Bookmarking and adjacent loops. In the el^^^ 1A'R83A-~el~~complex, helix n3 * of la1 j$ q$> .fa $&) @ 0 RelBc displaces a4 of el^^^^^'^^^^ from the binding site on the p-sheet What's this? core. This helix replacement results in neutralization of a conserved positively charged cluster of RelE by acidic residues from n3* of RelB. The released helix n4 becomes unfolded, adopting an open conformation with increased mobility. The displacement of a4 disrupts the geornetly of critical residues, including ~r~~~ and ~~r~~,in a putative active site of RelE toxin. Our structures indicate that RelB counteracts the toxic activity of RelE by displacing a4 helix fiom the catalytically competent position found in the free RelE structure. Received for publication, December 23, 2008 , and in revised form, March 9, 2009. The atolnic coordinates and structure factors (codes 2KC8 and 2KC9) have been deposited in the Protein Data Bank, Research Collaboratoryfor Structural Bioinfornzatics, Rutgers University, New Brunswick, NJ (II~II)://wwM).rcsb. orgo. * This work was supported, in whole or in part, by National Institutes of Health Grant ROlGM081567 (to M. Inouye). This work was also supported by a grant from Canadian Institutes of Health Research (to M. Ikura) and a Canadian Institutes of Health Research operating grant. 6/3/2009 Slipchip (DOI: 10.1039/b908978k) I2..abon a Chip Journals L.ab on a Chip Advance Articles DOl: 10.1039/b908978k Lab Chip, 2009 1 DOI: 10.1039lb908978k 1 Paper Wenbin Du , Liang Li , Kevin P. Nichols and Rus tem y. Is magilov * Department of Chemistry and Institute for Biophysical Dynamics, The University of Chicago, 929 East 57Ih Street, Chicago, Illinois 60637, USA. E-mail: isln a~ilovO,zrchic~zgo,edzt ~ Received 6th May 2009 ,Accepted 12th May 2009 1 First publislzerl on the web 15th May 2009 The SlipChip is a microfluidic device designed to perfonn multiplexed microfl~~idicreactions without pumps or valves. The device has two plates in close contact. The bottom plate coiltains wells preloaded with many reagents; in this paper plates with 48 reagents were used. These wells are covered by tlie top plate that acts as a lid for tlie wells with reagents. The device also has a fluidic path, composed of ducts in the bottom plate and wells in the top plate, which is connected only when the top and bottom plate are aligned in a specific confguration. Sample can be added into the fluidic path, Wu~gboth wells and ducts. Then, the top plate k "slipped7',or moved, relative to the bottom plate so the complementary patterns of wells in both plates overlap, exposing the sample- containing wells of tlie top plate to the reagent- containing wells of the bottom plate, and enabling diffbion and reactions. Between the two plates, a lubricating layer of fluorocarbon was used to hcilitate relative motion ofthe plates. This paper implements this approach on a nanoliter scale using devices fabricated in glass. Stability of preloaded solutions, control of loading, aiid lack of cross-contamination were tested using fluorescent dyes. F~~iictionalityof the device was illustrated via ciystalliition of a model membrane protein. Fabrication ofthis device is simple and does not requise a bonding step. This device requires no pumps or valves and is applicable to resource-poor setting. Overall, this device sl~ouldbe valuable for multiplexed applications that require exposing one sainple to many reagents in small volumes. 0tie may think of the S lipchip as an easy- to-use analogue of a preloaded inulti- well plate, or a preloaded liquid-phase microarray. Introduction Tliis paper describes the Slipchip, a microfluidic device for canyiiig out multiplexed solutioir phase experiments in a simple fonnat. Multiplexed experiments are coilunon, especially in applications that require screenillg. Miniahu'ization and siinplication of these experiments are attractive to miniink both sample volume aiid labor costs and to improve efficiency. Various microfluidic systems have been developed to perform multiplexed experiments on the nanoliter scale.w, Sophisticated approaches have used valve-based systeins~to route the sainple to inany reaction chanibers that can in turn be loaded with many differelit reagents. These systenls requised external equipment for hydraulic control and used gas-permeable materials for dead-end filling, which made storing reagents oii-board dficult.~Compact-disc (CD)-based PAPER www.rsc.org/loc I Lab an a Chip Microwave dielectric heating of drops in microfluidic devices David Issadore," Katherine J. ~um~hr~,?KeithA. Brown," Lori Sandberg,' David A. weitzab and Robert M. ~estervelt*nb Received 12th Decerttbel* 2008, Accepted 2rtd Marlclt 2009 First prcblislieJ as art A(2varice Article or1 tlte web 19th Ma~clt2009 DOI: 10.1039lb822357b We present a technique to locally and rapidly heat water drops in inicrofl~~idicdevices with illicrowave dielectric l~eating.Water absorbs microwave power more efficiently than polymers, glass, and oils due to its perinaileilt illolecular dipole nloineilt that has large dielectric loss at GHz frequencies. The relevant heat capacity of tlle system is a single thermally isolated picolitre-scale drop of water, enabling very fast thermal cycling, We demonstrate microwave dielectric heating in a inicrofluidic device that integrates a flow-focusing drop maker, drop splitters, and inetal electrodes to locally deliver microwave - power from ail-inexpensive, coinn~erciallyavailable 3.0 GHz source and amplifier.-The temperature change of the drops is measured by observing the tenlperature depeildeilt fluorescence inteilsity of cadillium seleilide nanocrystals suspended in the water drops. We denlollstrate characteristic l~eating times as short as 15 ms to steady-state temperature changes as large as 30 "C above tlle base ' temperature of the microfluidic device. Many conlnlon biological and chemical applications require rapid and local control of temperature and can benefit from this new tecl~nique. - Introduction techniques to improve tke localization and response time have been developed, such as those that use noa-contact infrared The miiliaturization of the l~andlingof liquid and biological heating of water in glass microfluidic systemsi4and integrated samples has enabled great advances in fields such as drug micrometre size Peltier Jui~ctionsto transfer heat between two discovery, genetic sequenciilg and synthesis, cell sorting, single chaanels contai~lingfluid at different temperatures.15Fluids have cell gene expression studies, and low-cost portable inedicii~e.'-~At also been cooled oil millisecond time scales with evaporative the forefront of this technology are tlle micro-fabricated pipes, cooling by pumping gasses into the fluid cliani~els.'~ valves, pumps, and mixers of microfluidics that are leading to The focus of our research is to integrate electronics wit11 integrated lab-on-a-clip devices. These integrated microfluidic microfluidics to bring new capabilities to lab-on;a-clip devices are causing a paradigm shift in fluid handling that is systen~s.~~~In this paper we present a technique to locally and ailalogous to what integrated circuit technology did for elec- rapidly heat water ia drop based microfluidic systems with tronics half of a century ago.6 A growing library of eleinents for illicrowave dielectric l~eating.The devices are fabricated using lab-on-a-chip systems have been developed in recent years for soft litl~ographyand are coilnected to iilexpeilsive cominercially tasks such as the nlixiilg of reagents, detecting and couilting of available inicrowave electronics. This work builds on previous cells, sorting cells, genetic analysis, and protein detectio11,'-7 There work in which illicrowaves have been used to heat liquid in is one function, however, that is crucial to inany applicatioils and microfluidic device^'^-^^ by achieving sigi~ificantlyfaster thernlal wllicl~has reillailled a challenge: the local control of tenlperature. response times and a greater ,temperature range. In our device, T11e large surface area to volunle ratios found in micrometre-scale drops of water are thermally isolated from the bulk of tlze channels and the close proxiillity of microfluidic elenleilts make device and this allows exceptioilally fast heating and cooling . temperature coi~trolin such systeills a unique cl~allenge.~*~ tinles z, = 15 ms to be attained and the drops' tenlperature to be Much work has been done in the last decade to iillprove local increased by 30 "C. The coupliilg of microwave electronics with temperature control in microfluidic systems. The most coinilloil microfluidics technology offers an inexpensive and easily inte- technique uses Joule heated inetal wires and thin filills to conduct grated technique to locally and rapidly control temperature. heat into fluid cl~anaels. The thesmal coilductivity of microfluidic devices control the localization of the teillperature chaage and tends to be in the order of cei~timetres.~,"Temporal Theory of dielectric heating control is limited by the heat capacity and thernlal coupling of Dielectric heating describes the pl~enomenonby which a illaterial the microfluidic device to tlle eilvironment and tl~erilialrelaxa- is heated wit11 a time-varying electric field. Induced and intrillsic tion tinles tend to be in the order of secoi~ds.~~~'Alternative dipole lllolllellts all object will aligll tllelllselves a tiine-varyiilg electric field. The energy associated wit11 this "School of E~~girteerir~gcrrzd A11plied Sciences, H(tr11m.d University, aligillllent is V~SCOL~S~Ydissipated as keat into the surrouildiilg Cc1n~6ridge,MA, 02138, USA, E-mail: ~~)estervelt@sens,harvclrrl. eclli solutioa. The power density P absorbed by a dielectric material is hDep(rrt~nerl t of Plzysics, Har~~nrclU~ziversity, C(imbriclge, MA, 02138, given by the frequency o of the applied electric field, the loss USA coflegeof EngiIleerillg ~~~~li~d~~i~~~~~~,(~l~~~~~~i~~ Wyo,niIlg, factor d' of the material, the vacuum perini t tivit y coy and the Lnrmnie, IVY, 82071, USA electric field streagtl~[El with the expressio~~:~~ This journal is O The Royal Society of Chemistry 2009 Lab Chip, 2009,9, 1701-1 706 1.1701 PAPER www,rsc.orylloc I Lab on a Chip Microfluidic valves with integrated structured elastomeric membranes for reversible fluidic entrapment and in situ channel functionalization I Siva A. ~anapalli,"""Daniel ~ijnperle,~~Albert van den ~er~,~'Frieder Mugele"' and Michel H. G. Duits*"' I I I Receivetl23rtl Octobe~~2008, Acceptetl IOtIz Febrimry 2009 I Firest pirblisltetl as nn Atlvnrice Article ort tlte web 31.d MnrAclr2009 DOI: 10.1039/b818712f We report the utility of structured elastomeric inelnbranes (SEMs) as a multifunctional microfluidic tool. These structured membranes are part of a two-layer microfluidic device, analogous to membrane valves, with the novelty that they incoi.porate topographical features on the roof of the fluid cllannel. We deinonstrate that when the topographical features are recessed into the roof of the fluid channel, actuatioil of the ineinbrane leads to effective confinement of fluids down to feintolitres in preformed microfluidic containers. Tlzus, tlle SEMs in this case f~~ilctionas fluidic traps that could be coupled to -- - - - microfl~1idicnetworks for rapid and repeated flushing of solvents.-Alternatively; wllen tlze -- topographical features on the roof protrude into the fluid cllannel, we demonstrate that tlle SEMs can be used to pattern proteins and cells in n~icrocl~ai~nels.Tlzus in tlzis case, the SEMs serve as fluidic stamps for functionalizing inicrochannel surfaces. In addition, we show that the trap or pattern slzape and size call be manipulated simply by varying the topograpky on the elastomeric membrane. Since SEMs, nzembrane valves and puinps use similar fabrication technology, we believe that SEMs call be integrated into microfluidic large-scale circuits for biotecl~nologicalapplications. 1 Introduction surfaces. We also sllow that by incorporating negative relief . features on the membrane, we can effectively confine ferntolitre Soft lithography tools such as inicrocontact printing, micro- volumes in prefornled inicrofl~lidiccontainers of controllable fluidic networks and membrane valves and punlps have emerged shape and size. Ultimately we believe that SEMs are a valuable as powerfill methods for addressing the current needs in addition to tlze toolbox needed for building integrated micro- bi~teclznology.~~~Yet, these tools suffer from some limitations or fluidic circuits capable of high throughput bioanalysis since they need f~~rtlzerimprovement, so that they can be generically can be readily integrated wit11 nzenlbrane valves and pu~nps. applicable to the broad range and complexity associated with biotecl~i~ologicalapplications. For example, nzicrocoiztact printing provides excellent control over material pattern shape 2. Experimental and size on open substrate^.^ However, suclz patterns are not 2.1 Materials easily integrable into nzicrofluidic devices, thereby precluding tlze capability to perform combinatorial assays (e.g, screening several Surface blocking agents bovine serum albumin (BSA, 4 nlg bionzarkers for immunoassays), Alternatively, lnicrofluidic mL-I) and TRITC labeled BSA (1 ing mL-') were procured from networks coupled with lalniilar flows allow patterning on Sigma-Aldricl~.Fibronectin (100 pg inLdl,obtained from a local microchannel surfa~es,~~~but lack colltrol over pattern shape and blood bank) was used as a representative protein to denlollstrate size and therefore are not amenable to fabrication of higll-density patterning of proteins by SEMs. It was detected by using bioarrays, Here we present a soft litlzography-based microfluidic imilzuizofluorescellce staining using fibronectin antibody (0.05 tool called structured elastomeric nlembranes (SEMs) that ing inL-l) and Alexa 488 labeled Anti-rabbit IgG (0,l lng inL-I), overcomes some of these limitations. In particular we incorpo- both of whicll were purchased from Abcam. FITC-pllalloidin rate positive relief features (that are common to inicrocontact and DAPI were used to stain tlze iiztracellular actin network and printing stamps) on the elastomeric defornlable membrane of the nucleus respectively, both of wlliclz were acquired from lnicrofluidic valves6l7 and slzow their application as fluidic Molecular Probes. Plzosphate buffer saline (PBS) solution was stamps. Using this approaclz we generate patterned areas of obtained from the local hospital in Ensclzede, proteins and cells of coiztrollable slzape and size on illicrochai~ilel Human umbilical vein endothelial cells were isolated from umbilical cords by tlle method of Jaffe et a/,,' using a PBS solution containing trypsin (0.05% (wlv) and 0,02% (wlv) EDTA. "Pltysics of Con~l~lex'Fluids, Ui~iversityof Ttlwizte, 7500, AE, Enschede, Tlle obtained endothelial cells were cultured in fibronectin- Tlre Netharlc~izl.ls, E-mail: siva. ~)cri~cipr~lli@ttu,eh; in,h. g. c/~~its@trt114 ut~)erzte.121 coated culture flasks in Endothelial Growth Medi~uin-2(Lonza), "BIOS Lrib on cr C/I@,Urzi1ter'sitj) of T~~leizte,7500, A El E11sc1zc.de,TIE coiltailling 2% fetal bovine serum, ~ultilthey reached confluence. Netl~erIc~r~c/~~ When confluent, cells were detached from the surface using 'MESA+ Institute of Ncr~zoteclzizologj),Uizhersitj) of T~vente,7500, AE, trypsin solution and diluted : 3 in a freslz fibronectin-coated Ensclzecle, Tlze Netherlarzcls 1 "Depclrfinszt of Cltcii?iccrlEizgiizeer'ing, Texns Teclz University, Lubbock, culture flask. Cells were kept in a ll~~midifiedincubator (37 "C, TX,79409, USA 5% COz) up to passage 8, after wlzicll tlzey were discarded. Cells Thisjournal is O The Royal Society of Che~nistry2009 Lab Chip, 2009, 9, 1461-1467 1 1461 I Molecular Microbiology 120U91 72141, 815-829 1 doi:10.111 1/j.1365-295882009,06690.r, - First published online 21 April 2009 Dissection of recognition determinants of Escherichia coli 03*suggests a composite -10 region with an 'extended -10' motif and a core -1 0 element (Gross eta/., 1998). Most transcription in exponentially growing cells is initiated by RNA polymerase holoenzyme carrying the housekeeping o, called 070 in Escherichia coli. Alternative as mediate transcription of regulons acti- vated by specific stress conditions, or by growth or devgl- opmental transitions. The 070 family of proteins has been divided into four groups, based on their phylogenetic rela- tionships and modular structure (Lonetto etal., 1992; Gruber and Gross, 2003). Group 1 os are the essential housekeeping os and contain four domains, of which the Summary first is unique to the Group 1 os. Group 2 os are most 032controls expression of heat shock genes in closely related to Group 1 os, but are not essential. Most Escherichia coli and is widely distributed in Group 3 os contain Domains 2-4, and the Group 4 os proteobacteria. The distinguishing feature of 03*pro- contain only Domains 2 and 4 (Gruber and Gross, .2003). moters is a long -10 region (CCCCATNT) whose Promoters recognized by 070and other Group 1 os tetra-C motif is important for promoter activity. Using have been extensively studied. Promoters are multipar- alanine-scanning mutagenesis of $2 and in vivo and tite, and can have the following potential elements to in vitro assays, we identified promoter recognition mediate recognition: (i) the 'UP elementJ, generally determinants of this motif. The most downstream C located from -61 to -41 relative to the transcription start (-13) is part of the -10 motif; our work confirms and point (Ross et at., 1993; Gourse etal., 2000), (ii) the -35 extends recognition determinants of -13C. Most region, a hexamer (TTGACA) centred about 35 bp importantly, our work suggests that the two upstream upstream of the transcription start point (Gross etal., Cs (-16, -1 5) constitute an 'extended -1 0' recognition 1998), (iii) the 'extended -1 0 motif' (TG) located immedi- motif that is recognized by K130, a residue universally ately upstream of the -1 0 region (Mitchell et at., 2003), (iv) conserved in fb and y-proteobacteria. This residue is the -1 0 region, a hexamer (TATAAT) centred about 10 bp located in the a-helix of oDomain 3 that mediates upstream of the transcription start (Helmann and recognition of the extended -10 promoter motif in deHaseth, 1999) and (v) the discriminator region (GGGa) other os. K130 is not conserved in a- and 6-/Em located downstream of the -10 region of the promoter proteobacteria and we found that 032 from the (Feklistov etal., 2006; Haugen et at., 2006). With the a-proteobacterium Caulobacter crescentus does not exception of the UP element, recognized by the need the extended -10 motif for high promoter C-terminal domain of a, promoter elements are recog- activity. This result supports the idea that K130 medi- nized by the o subunit. The -35 element is recognized by ates extended -10 recognition. $2 is the first Group 3 Region 4.2 in domain 4; the extended -1 0 element by an o shown to use the 'extended -10' recognition motif. u-helix in aDomain 3; the -1 0 by Region 2.4 in oDomain 2 and the discriminator by oRegion 1.2 (Campbell etal., Introduction 2002; Feklistov et a/., 2006; Haugen et a/., 2006). The number of elements in promoters is variable. Bacterial transcription initiation is mediated by RNA poly- Promoters recognized by the housekeeping o in E. coli merase holoenzyme, consisting of a 5-subunit core RNA can be roughly divided into those with an extended -10 polymerase (cc2pp1w) and a dissociable subunit, o, os motif ('extended -10 promotersJ)and those without this contain many of the promoter recognition determinants motif. Extended -1 0 promoters have a lower match to the Accepted 31 March 2009. *For correspondence. E-mail: cgross8 consensus -35 motif than do those lacking the extended cgl.ucsf.edu; Tel. (+I)415 476 4161; Fax (+A) 41 5 514 4080. -1 0 motif (Mitchell etal., 2003). Moreover, at least in vifro, O 2009 The Authors Journal compilation O 2009 Blackwell Publishing Ltd ; I I I I n'dk~re Vol459 128 May 2009 1 doi:10.1038/nature07984 , Irreversibility of mitotic exit is the consequence of systems-level feed back Sandra ~6pez-~vilBes',Orsolya ~apuy~'"BBela ~oviik~& Frank uhlmann1 The eukaryotic cell cycle comprises an ordered series of events, AYC even in the presence of higll Cdk activity owing to mutation of orchestrated by the activity of cyclin-dependent lcinases (Cdks), 11 Cdk phosphorylation sites, This led to efficient degradatioil of the - leading- from chroinoson~ereplication-during S phase to their inajor budding yeastmitotic cyclin Clb2 (Fig. lb), The initotic Polo-- - segregation in n~itosis.The unidirectionality of cell-cycle transi- like lcinase, another APC'~"' target", was also efficiently degmded, tions is fundamental for the successful completion of this cycle. It whereas levels of the S phase cyclin Clb5, a preferential st~bstratefor is thought that irrevocable proteolytic degradation of key cell- ApCCd~2() (ref. 1 I ), re~nai~ledlargely unaffected (S~rpplementaryFig. I ). cycle regulators mdces cell-cycle transitions irreversible, thereby Clb2 destruction was acconlpanied by dephosphorylation of known . enforcing directiondity '". Here we have experimentally examined initotic Cdk substrates, seen by their change in electrophoretic nlobility the contribution of cyclin proteolysis to the irreversibility of (Fig. 1b) . These illcluded three proteins whose dephosphorylatioi~con- mitotic exit, the transition from high n~itoticCdk activity back tril~utesto spindle eloilgation and chrotnosonle segregation, Sli15, to low activity in GI. We show that forced cyclin destri~ctionin Asel and Ask1 (refs 12-14). Their dephosphorylation depended on mitotic budding yeast cells efficiently drives mitotic exit events. the activity of the initotic exit phosphatase Cdcl4 (Stpplernentary However, these remain reversible after termination of cyclin Fig. 2). Mitotic spincues that were present in the metayhase-arrested proteolysis, with recovery of the mitotic state and cyclin levels. cells disassembled as Clb2 levels declined, acco~npaniedby outgrowtll Mitotic exit becomes irreversible only after longer periods of of p rollou~lcedastral microtubules (Fig, I c and Supplen~enta~yFig. 3)) cyclin degradation, owing to activation of a double-negative reminiscent of spindle brealcdown at the end of mitosis. APC~'~"'('"~I)- feedback loop involving the Cdk inhibitor Sicl (refs 4, 5). inediated destruction of the spindle stabilizing factor Asel (Fig. lb), in . Quantitative modelling suggests that feedback is required to addition to its dephospl~oiylation,may contribute to this yhen~type'~. maintain low Cclk activity and to prevent cyclin resynthesis. Our After 50 min, when Clb2 levels became almost undetectable, we findings demonstrate that the unidirectionality of mitotic exit is turned off APC'~" l(lnll) by inactivating a temperature-sensitive APC not the consequence of proteolysis but of systems-level feedback core subunit encoded by the cdcl6-123 allele'" Notably, in response to required to maintain the cell cycle in a new stable state. AI~C~d~'6-123inactivation, C1b2 levels recovered and Sli15, Asel and After coinpletion of chroinosoine segregation during mitosis, the Aslcl reappeared in their mitotic hyperphosphorylated forins. Mitotic activity of the key cell-cycle lunase Cdk is dowilregulated to pron~ote spindles fornled again, suggesting that cells had re-entered a initotic mitotic exit and the return of cells to GI. This involves ubiquitin- state. Spindles appeared n~orpl~ologicallyintact, but were longer after mediated degradation of mitotic cyclins under control of the anaphase cyclin reaccun~ulation(3.9 rt 0.8 11111; mean t: sad.)compared to meta- protnoting colnplex (APC), a mttltisubunit ubiquitin ligase, Mitotic phase spindles at the begirlni~lgof the experime~lt(2,l t: 0.6 in^). cyclins are initially targeted for degradation b the APC in association A probable reason for this lies in con~pronlisedsister chroinatid with its activating subunit Cdc20 (APCcdC2'J.Later, declilliilg Cdk cohesion after some, albeit inefficient, inactivatioil of the anapllase levels and activation of the Cdlc-counteracting phospl~ataseCdcl4 inhibitor securin by APC"~"' (I"' (Supplementary Fig, 4)17. allow a second APC activator, Cdhl, to associate with the APC fluorescence-activated cell sorting (FACS) analysis of DNA content (APC'~'")"" Cyclin proteolysis, a tllerl~lodynan~icallyirreversible confirined that cells n~aintaineda 2C DNA content thi.oughout the reaction, is thought tosbe responsible for the irreversibility of initotic experiment (Fig, Id). This detnollstrates that cyclin destructio~l exit'". However, de novo protein syllthesis can counteract degradation promotes nlitotic exit events, b~stit is not sufficient to re~zderthem and constitutes a similar tl~ennodyl~ai~~icdlyirreversible process, irreversible. Cyclin call reverse initotic exit. Note that driven byK17P hydrolysis. 111 a cellular setting, therefore, proteinlevels reversibility of mitotic exit under these co~~ditionsdid not depend are defined by reversible changes to the rates of two illdividually on tlle persistence of the S pl~asccyclin C1b5 (Supplementary Fig. I), irreversible reactions: protein syntl~esisand degradation. These con- We next addressed what makes lnitotic exit irreversible,if not cycliil siderations have led to the hypothesis that it is not proteolysis itself, destruction. Wheil we repeated the experiment, but contiilued but systems-level fiedbaclc that affects syntl~esisand degradation rates, Cdll(m 11 ) inductio~land inactivated APC~~~~'"-~~~only after 60 min, inaking cell-cycle transitions irreve~*sible~, Clb2 did not i+eaccu~ulateand over half of the cells s~~bsequently To test this hypothesis, we investigated the contribution of cyclin completed cytolciilesis and entered G 1 (Fig, 2a). This suggests that after proteolysis to the irreversibility ofbudding yeast nlitotic exit (Fig. la), longer periods of C1b2 destructioil mitotic exit becomes irreversible, We arrested budding yeast cells in nlitosis with higll levels of initotic Western blotting sl~owedthat around the time when Clb2 loss turned cyclins by depleting Cdc2O under control of the MET3 promoter, 111 irreversible, the Cdk hhibitor Sicl accun~ulated~~~,We therefore asked these cells, we induced Cdlll expressioll from the galactose-inducible whether Sicl accumulation was respo~lsiblefor the irreversibility of GALL promoter (an attenuated version of the GAL1 promoter) for mitotic exit, When we repeated the experiment ~zsinga sicld strain, 30 inin" We expressed a Cdhl variant, Cdhl(mll), that activates the Clb2 reappeared after A~'Jc""'~-'" inactivation, and only a iniilority of 'Chromosome Segregation Laboratory, Cancer Research UK London Research Institute, 44 Lincoln's Inn Fields, London WC2A 3PX, UK. 'oxford Centre for Integrative Systems Biology, Department of Biochemistry, University of Oxford, South Parks Road, Oxford OX1 3QU, UK, 3~udapestUniversity of Technology and Economics, Gell6rt tCr 4,1521 Budapest, Hungary. 592 02009 Macmillan Publishers Limited. All rights reserved L , , I nr;lhxl'e s. ! Vol459 121 May 2009 1 doi:10.1038/nature08012 ! LETTER Non-genetic origins of cell-to-cell variability in TRAIL-induced apoptosis Sabrina L. ~pencer'>~*,Suzanne ~audetl-I*, John G. ~lbeck',John M. ~urkel& Peter K. ~orger' In microorganisms, noise in gene expression gives rise to cell-to- MCPlOA inanlmary epithelial cells in tlle presence aid absence of pro- cell variability in protein concentrations1-'. In n~anzmaliancells, tein synthesis inhibitors. protein levels also vary8-"2nd individual-cells differ widely in their Pai~ofpsister cells expressing a fluorescent sepoTter of MOMP responsive~zessto uniform physiological stimuli1'-'" 111 the case of (mitochondria1 intermeinbrane space reporter protein, IMS-RP'l) apoptosis mediated by TRAIL (tumour necrosis factor (TNP)- born during a 20-30 11 period were identified by time-lapse micro- related apoptosis-inducing ligand) it is coiwnon for some cells in scopy. TRAIL and the protein synthesis inhibitor cyclohexitnide were a clonal population to die while others survive-a striking diver- then added and imaging colltinued for another 8 h. The TRAIL-to- gence in cell fate. Among cells that die, the time between TRAIL MOMP inteival (Td) was calci~latedfor each cell (Fig. la). Alnong the exposure and caspase activation is higlzly variable. Here we image recently divided sisters (<7 11 between division and death), Td was sister cells expressing reporters of caspase activation and mito- highly correlated (R~= 0.93, Fig. lb), whereas Td was ut~correlated chondrial outer men~brmepernleabilization after exposure to (R~= 0.04) for recently divided cells chosen at random. The time TRAIL. We show that naturally occurring differences in the levels since division (Fig. lc) and the position in the dish (data not shown) or states of proteins fegulating receptor-mediated apoptosis are did not correlate with Td, ruling out a role for cell cycle state and cell- the primary causes of cell-to-cell variability in the timing and cell interactions uiider our experiinental conditiois. However, as the probability of death in human cell lines. Protein state is transmitted time since divisioil increased, sister-to-sister correlation in Td from inother to daughter, giving rise to transient heritability in decayed exponentially with a half-life of -11 h, so that sisters lost fate, but protein synthesis proinotes rapid divergence so that sister lnemory of shared ancestry within -50 h or about two cell genera- cells soon become no more similar to each other than pairs of tions (R~5 0.05, the same as random pairs ofcells; Fig. Id, e). Similar cells chosen at random. Our results have iinplications for under- results were obtained with MCFlOA cells (~upplen~entar~Fig. 2). standing 'fractional killing' of tumour cells after exposuse to cheino- High correlation between recently born sisters shows that the vari- therapy, and for vmiability in nzanmalian signal transduction in ability in arises from differences that exist before TMlL exposure, general. and rules out stochastic fluctuatioils in sigilalling reactions: Rapid TRAIL elicits a heterogeneous phenotypic response in both sensitive decorrelation also rules out genetic mutation or coilventional epige- and relatively resistant cell lines: some cells die within 45 min, ot11ei.s netic differences (which typically last 10-10' cell divisioi~s'~). 8-12 h later, and yet others live indefinitely (Supplementaiy Fig. I), However, transient heritability is precisely what we expected for D~uringthe variable delay between TRAIL addition and mitocl~ondrial cell-to-cell differences arising from variations in the concentratiolls outer menlbrane pernleabilization (MOMP), upstream initiator or states of proteins that are partitioned binomially at cell division. caspases are active but downstreain effector caspases are 110t"~'~. Whereas all TRAIL-treated HeLa cells evalttlally died in the presence I'ossible sources of cell-to-cell variability in response to TIWL include of cycloheximide, in its absence a fraction always survived (presumably genetic or epigenetic diEereilces, stochastic fluctuations in biocheinical owing to ind~~ctionofsi~lvival pathways'"). When the fates of sister cells reactions iilvolving low co py nuinber components ('intrinsic i~oise'~)),were comnpared, both lived or both died in most cases (chi-squared test, differences in cell cycle phase, and nati~ralvariation in the concentra- P= 7 X lo-", Supplementary Fig. 3). Variability in Td ac1.0~~the , tions of illlportallt reactants. To distinguish betyen these atld otl~er pop~dationwas large (Fig, Ig and Supplanentary Fi . 4))but recently possibilities, we used live-cell ~nicroscopyto coinpare the tiniing and born sisters were nevertheless correlated in Td (Rf = 0.75, Fig. If), probability of death in sister cells exposed to TRAIL. If phenotypic Again, cell cycle phase was not correlated wit11 fate or the time-to-death variability is caused by genetic or epigenetic differences, sister cells (Fig. lg), Decorrelation it1 TcIanlong sisters was an older of magnitude should behave identically. In contrast, if stocllastic fluctuations in more rapid in the presence of protein syntl~esisthan in its absence i6eactionstriggered by 'l'lUIL predominate; sister cells should be no (-1.5 11 half-life, Fig. 111, iand Supplernaltary Fig. 5), Thus, the length nlore sinlilar to each other than pairs of cells selected at random, Tile of time that Td is heritable is very sensitive to rates of protein synthesis, influence of cell cycle state on apoptosis- - should be readily observable both basal and TRAIL-induced, from time-lapse iinaging of asynchronous cultures. Furtherinore, We then aslted whether the conceiltrations of proteins regulating variability arising from differences in protein levels (or in their activity TRAIL-induced apoptosis are sufl?cientlydifferent from cell to cell to or lnoclification state) sho~~ldproduce a highly distinctive form of account for the variability in Td,Using flow cytometry, we measured inheritance in which newly born sister cells are vc~ysilnilar, because the distributions of five apoptotic regulators for which specific anti- they inherit similar nmnbers of abundant factors fitom their ~nothef"~, bodies are available. All five proteins were log-normally distributed but then diverge as new proteins are made and levels drift1'>'" With this across the population with coefficients of variation ranging between in mind, we exaillined apoptosis in HeLa cells and in non-transfoimed 0.21 and 0.28 for cells of similar size (Fig. 2a), consistent with data on 'center for Cell Decision Processes, Department of Systems Biology, Harvard Medical School, Boston, Massachusetts 02115, USA. 2~omputationaland Systems Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA, ?Present address: Department of Cancer Biology, Dana-Farber Cancer Institute, and Department of Genetics, Harvard Medical School, Boston, Massachusetts 02115, USA. *These authors contributed equally to this work. 428 02009 Macmillan Publishers Limited. All rights reserved j Vol459 114 May 2009 1 doi:10.1038/nature07921 I rerm.r v i LETTERS Snowdrift game dynamics and facultative cheating in yeast b Jeff ore', Hyun ~ouk'& Alexander van ~udenaarden' The origin of cooperation is a central challenge to our understanding expected consequence of diffusion and the known properties of the of evolution1". The fact that microbial interactions c'm be mani- sugar importers (Supplementary Fig. 4). ------ptdated in ways that animal interactions cannot has led to a growing- Giver-that 99%-of glucose created bTa cell'is lost to rieighbauring interest in ~nicrobial~nodels of cooyerati~n~-~~and colnpetiti~n~~*'~. cells, it inay be possible for a 'cheater' strain to take advantage of the For the budding yeast Sacc1zaronzyces cerevisiae to grow on sucrose, cooperato~~sby ;lot secreting iilvertase and instead simply coiisun-iing the disaccharide must first be hydrolysed by the enzyme inver- the glucose created by other cellsL K. ~odal*,K. K. ~sial*& B. ~alalil* Ultrafast real-time optical imaging is m indispensable tool for with a 1-MHz franle rate-t11e worlci's fastest CCD camera-has been studying dynamical events such as shock chemical developed by Shinladzt~Corporation (ref. 15; see also http://www, dyllarnics in living cell^'^^, neural activitf9" laser surgery7-"-and sl~ii1~adzu,com/products/test/hsvc/ol180jtOOO0OO1d6t-att/booe13-02.ydf-)r- microflr~idics'"". However, conventional CCDs (charge-cotrpled Its inlpressive perforlnance stems fronl several technological modifi- devices) and their complenlentary metal-oxide-semicond~~ctor cations aiined at overcoming the loss of sensitivity that results from (CMOS) counterparts are incapable of capturing fast dynamicd the shorter integration time during high-speed imaging, and also at processes with high sensitivity and resolution. This is due in part reducing the tilne it talces to read out tlle data from the 2D pixel arrays, to a technological limitation-it talres time to read out the data These lnodifications incl~lciea larger pixel size for the purpose of from sensor arrays, Also, there is the f~~ndamentalcoinpromise adding pixel-level it1 ssitri nleinoiy (at tlle expense of reduced spatial between sensitivity and frame rate; at high frame rates, fewer resolutioil), and coolillg of the camera to reduce noise (at the expense photons are collected during each frame-a problem that affects of the added con~plexityof refiigeration), The Shiinadzu device also newly all optical imaging systems. Here we report an imaging relies on a hi&-power ill~~nlinatorto ensure adequate signal-to-noise method that overcomes these limitations and offers frame rates ratio and to prevent a drop in sensitivity-a requirement that renders that are at least 1,000 times faster than those of conventional it unsuitable for microscopy, where focusing of the high-power illu- CCDs. Our technique maps a two-dimensional (2D) image into nlination over a sinall field of view can cause damage to a biological a serial time-domain data stream and simultaneously amplifies the , sample. Tllis colnproinise between sensitivity and frame rate is not image in the optical domain. We capture 'an entire 2D image using uniqile to the CCD-it impacts ahnost all imaging and detection a single-pixel photodetector and achieve a net image amplification systems. of 25 dB (a factor of 316). This overcoines the compromise between In scientific applications, high-speed imaging is often achieved sensitivity and hame rate without resorting to cooling and high- using the time-resolved pump-probe tecl~iiique'"~~l'uinp-probe intensity illunlination, As a proof of concept, we perform continu- tecl~niquescan capture the dynamics of fast events, but only if the ous real-time imaging at a frame speed of 163 11s (a fraine rate of event is repetitive, Because they do not operate in real time, they are 6.1 MHz) and a shutter speed of 440 ps, We also demonstrate real- unable to capture non-repetitive rando~nevents that occur only once time imaging of microfluidic flow and phase-explosion effects that or do not occur at regular iiltervals, such as rogue eveiltsL",Detection occur daring laser ablation, of sucll events requires an ilnaging technology with fast, continuous, Optical imaging is a widespread and versatile diag~losticsand real-time capability. illspection tool in use today. Altl~oughmost of the current research Another type of high-speed image sensor is t11e franling streak in imaging is aiined at improving the spatial resolution to below the camera, which has been used for diagllostics in laser filsion, plasma diffraction li~nit'~~'",there are nuillerous applications that denund radiation and con~bustion.This device operates in burst inode only improvement in temporal resolution, Imaging systems will1 high (providing only several franles; see http://lear~~.halnai~~i~t~'~~.com/ temporal resolution are needed to study rapid pllysical pllenomena tutorials/streakcainera/) and requires syncl~ronizationof the camera ranging from shock waves, including ex?sacorporeal shoclc waves wit11 the event to be captured, renderiiig streak cameras also unable to ilsed for surgery', lo diagnostics of laser filsioi12 and file1 iiljection capture unique or ralldoln eveilts. This, along with the high cost of in internal coinbustion engines. High-speed inlaging is beconli~lg the camera, limits its use in practical applications. increasingly iinportant in injc~.oscopybecause on the micrometre Altl~oughCCD imagers will coiltinue to be the most widely used scale even slow-moving p11eiiome11a require l-rigll temporal resolu- imaging modality and pump-probe experiments will ren~aina tion. One example is the spatioteniporal study of biochemical tvaves powerful tool for studying fast repetitive events, a new and comple- in cells and tissues, which requires i~nagingwit11 a ~nicro-to nano- ri~entaryimaging modality that can capture the dyl~arnicsof fast second response time and is iinyortant for the study of cell signallillg single-shot or random events is clearly needed. Serial time-encoded and drug transport'" Another ltcy application is in the field of flow amplified l~licroscopy(STEAM) tcchilology is a new approach to cytoinetryl', where high-speed cameras are needed to provide high- imaging that provides such a capability. The 211 ilnage is encoded throughput cell cl~aracterization, into a serial time-domain waveform that is anlylified and captured, The CCD or CMOS imager is by far the most widcly deployed not by a CCI) camera, but instead by a single-pixel photodiode and optical inlaging techi~ology.It offers a spatial resolution of a few an oscilloscope, The inail1 attributes of the new imager are the in~age n~icrometres,a large number of pixels anc1 relatively low cost. rI'ypical anlplification in the optical domain and the elimination of the imagers used in coIIsun1er electrot~icshave frame sates of 30 Hz, although CCD-when combined, they enable continuous real-time operation high-end versions can operate at rates on the order of 1 kHz by reducing at a fraine rate of 6.1 MHz and a shutter speed of 440 ps. The STEAM the nunlber of pixels that are read out from the arrays. A CC11 imager cainera operates coiltinuously and can capture ultrafast events '~e~artmentof Electrical Engineering, University of California, Los Angeles, California 90(395, USA. "These authors contributed equally to this work. ,1145 02009 Macmillan Publishers Limited. All rights reserved @ SYSTEMS BIOLOGY Using movies to analyse gene circuit dynamics in single cells James C. W. Locke and Michael B. Elowitz -, ------Abstract I Many bacterial systems rely on dynamic genetic circuits to control crucial biological processes. A major goal of systems biology is to understand these behavio~~rsin terms of individual genes and their interactions. However, traditional techniqc~esbased on population averages 'wash out' crucial dynamics that are either clnsynchronized between cells or are driven by fluctuations, or 'noise', in cellular components. Recently, the combination of time-lapse microscopy, quantitative image analysis and fluorescent protein reporters has enabled direct observation of mc~ltiplecellular components over time in individual cells. In conjc~nctionwith mathematical modelling, these techniqc~esare now providing powerful insights into genetic circuit behaviour in diverse microbial systems. As biologists, we inust grapple with, and reconcile, two function that is often fascinating to watch. With movies, very different views of cellular behaviour. On the one the eye often picks out subtle patterns in iildividual liv- hand, we freq~ientlythink of cellular f~inctionsas being ing cells that would be difficult to notice with less direct determined by 'circuits' of interacting genes aid pro- techiliq~ies.Few techniques are more fun. teins. In a loosely analogous way to electronic circuits, How does quantitative movie analysis cotnpare with these cheinical circuits encode genetic progra~nllles alternative techniques for analysing gene circuits? Time- that underlie differentiation, the cell cycle and other lapse microscopy follows a few genes over time in indi- behaviours (FIG. I a). They accurately respond to stimuli vidual living cells. It coillplements approaches such as and generate precise behavioural programmes in indi- microarrays (which provide genome-scale expression vidual cells, On the other hand, there is the 'noisy' view of data averaged over populatioils, but do not allow analysis the cell we get when we actually loolc at cells: they exist of variability) and flow cytoinetry (wllicl~allows high- in squishy, dynamic and heterogeneous populations, the throughpuit accluisition of single-cell fluorescence values, ~norphologies,gene-expression patterns and differenti- but does not allow the same cell to be tracked over time). ated states of which differ fi.on1 one another, even when Movies also colnplenlent new single-cell q~lantitativePCR environtnent and genotype are fixed (FIG. 1 b). How can approaches, which enable analysis of expression of multi- pi.ecisely defined genetic circuits give rise to heteroge- ple genes in individual cells, but, because they require lysis neity and, conversely,how does heterogeneity affect the of the cell, do not permit tracking of expression dynamics1. behaviour of biological circuits? Movies enable researchers to deterinine the 'trajectories' Movies offer a powerfill way to address these qiiestions of gene expression levels in individual living cells. One (FIG. 1 c). By engineeringmicrobial strains to express fluo- potential drawback of inovies is that although many rescent protein reporters for key genes, researchers can genes, and their expression levels, may be inlportant for follow the changing characteristicsof individual cells over a particular process under study, inost studies currently time. Quantitative detection methods, i~nprovedmicro- follow the dynamics of only a few genes at a time owing to scope au~tomationand software, and the range of fluores- the laclc of distinguishable reporters, In the future, multi- cent reporter genes that are now available, in coi~junction spectral techniques may expand the n~lmberof siillultane- Howard Huglies Medical I~atitute,Division of Blology with tnathenx~ticalmodelling, can be combined to analyse ous reporters2.I-Iowever, studies that follow the dynamics and Departnient of Applied gene circuit dynamics. 'liogethel; these techl~iqtiesallow of only two or three genes at a time can still be extremely Physics, Califonlia Institute researchers to characterize epigenetic states, identify new informative. - of T'ech~iology,Pasadena, dy~lamicphenoinena, analyse biocheinical interactions Here we review work in which movies provide new California 9 1 125, USA. within circuits and elucidate the physiological f~lnction insights into the dynamic behaviour of genetic conlpo- Correspondence to M.6. E, esiail: ~ielowitz@callech,edu of genetic circuits, all at the single-cell level. Finally, mov- nents and circuits, Por this Review, we have confined doi: l0.10381nrniicro2056 ies provide an aesthetically compelling view of cellular ourselves to microbial systems and have therefore O 2009 Macmillan Publishers Limited. All rights reserved nature structural & molecular biology Unraveling infectious structures, strain variants and species barriers for the yeast prion [PSI+] Peter M Tessier & Susan Lindquist Prions are proteins that can access mirltiple conformations, readily owing to the ease of genetic ~i~aniyulation,was an iinportant factor at least one of which is P-sheet rich, infectious and self- in winning this battle1'-13. The prions of yeast and other fungi consist $ perpetuating in nature. These infectious proteins show several of conlpletely different proteins whose sequences are unrelated to their $ remarkable biological activities, including the ability to form man~maliancountery art^^*^^""'. Moreovel; fungal p rions are generally v, multiple infectious prion conformations, also Icnown as strains not deleterious and call even be They serve as heritable or variants, encoding unique biological phenotypes, and to elements, producing stable new phenotypes due to a profound change v, establish and overcome prion species (transmission) barriers, in protein conforlnation that is self-templating and transmissible fro111 . In this Perspective, we highlight recent studies of the yeast prion mother to daughter cells3,","' I, Indeed, the recent proposal of a prion-like =a [PSIf], using various biochemical and strrictiiral methods, that mechanism for the perpetuation of synapses and neuronal ~nen~osiesl~, have begun to illuminate the molecular mechanisms by which as well as the discovery of a host of new prions with diverse functiolls in - self-perpetuating prions encipher such biological activities. We yeast (for exanlple, see refs. 15,lG), indicates tliat prions will prove vitally also discuss several aspects of prion conformational change and important in many organisms, . structure that remain either unknown or controversial, and we An important similarity between mammalian and yeast prions propose approaches to accelerate the understanding of these is tliat they forin not just one psion conformation, but a collectioil a enigmatic, infectious conformers. of structurally related yet distinct confor~nations,l 5 9 8 VOLUME 'I6 NUMBER 6 JUNE 2009 NATURE S1'RUCTURAI. & MOLECUI.AR BIOLOGY naturestructural & molecular biology Decapping is preceded by 3' uridylation in a novel pathway of bulk mRNA turnover Olivia S ~issland')~& Chris J or bury' ------Both end structures of eukaryotic mRNAs, namely the 5' cap and 3' poly(A) tail, are necessary for transcript stability, and loss of either is sufficient to stimulate decay. mRNA turnover is classically thought to be initiated by deadenylation, as has been .ri particularly well described in Saccharomyces cerevisiae. Here we describe two additional, parallel decay pathways in the fission yeast Schizosaccharomycespombe. First, in fission yeast mRNA decapping is frequently independent of deadenylation. Second, L, Cidl-dependent uridylation of polyadenylated mRNAs, such as actl, hcnl and urgl, seems to stimulate decapping as part of a novel mRNA turnover pathway. Accordingly, urgl mRNA is stabilized in cidlA cells. Uridylation and deadenylation act a redundantly to stimulate decapping, and our data suggest that uridylation-dependent decapping is mediated by the Lsml-7 complex. 2m m- As human cells contain Cidl orthologs, uridylation may form the basis of a widespread, conserved mechanism of mRNA decay. -L 3 o' -r: In budding yeast, most cytoplasmic mRNA turnover is initiated by uridylation of polyadenylated mRNAs forms the basis for an addi- 6' deadenylationl. These messages are then either decapped and subject tional5'+3' decay pathway, probably conserved in higher euka~yotes, .-o to 5'+ 3' decay2J or degraded by the cytoplasmic exosome4. Although that elicits decapping and seems to be mediated by the Lsml-7 % these decay pathways are conser~ed~~~,metazoans and fission yeast complex. Uridylation and deadenylation have overlapping and distinct contain additional cytoplasmic RNA-processing enzymes; the roles of stimulatory effects on decapping of polyadenylated rnRNAs. a L many of these have not yet been determined, Schizosaccharonzyces 2 ponzbe Cidl is one such enzyme, a cytoplasmic member of a family of RESULTS RNA nucleotidyl tran~ferases~~~. cRACE captures act1 mRNA degradation intermediates Cidl, identified through its involvement in the S-M checkpoint7, is To dissect RNA decay pathways in fission yeast, we used the cRACE now known to be one of a subgroup of this family possessing either technique6. A tail-independent method of capturing 3' and 5' ends, @ poly(U) polymerase (PUP) and/or terminal uridyl transferase this procedure allows distinction between decapped and capped (TUTase) activityg,This subgroup also includes the human enzymes mRNAs (Fig. la). We first examined decapped and capped act1 U6 TUTase, Hs2 and Hs3 (refs. 9-11), although no member of this transcripts fsom exponentially growing wild-type cells. subgroup is present in budding yeast12. We initially wished to determine whether those products isolated Uridylation of mRNAs and noncodillg RNAs has been described in from decapped cRACE analysis were derived from decay intermedi- fission yeast and metazoans. ICtlown substrates include miRNA- ates. To do this, we compared the 5' ends isolated from capped and directed cleavage product^'^ and replication-dependent histone decapped mRNAs (Fig. lb). The 5' ends of products from capped mRNAs, which in metazoans contain a 3' stem-loop structure rather transcripts generally lay 57 nucleotides (nt) or 58 nt upstream from than a poly(A) tail, and decay of which is stimulated by oligouridyl- the start codon (Pig. lb and Supplementary Fig. 1 online). These atioi~'~-'~.In addition, we previously observed terminal uridyl nucleotides presumably represent the major transcriptional start site residues on polyadenylated S, pon~be act1 mRNA during S-phase for actl. In contrast, the 5' ends of decapped products were hetero- arrestg. Although it seemed likely that Cidl-mediated uridylation geneous, always downstream from the inajor transcriptional start was generally involved in inRNA metabolism, the efCect of such a site and distributed significantly differently from the capped species modification on yolyadenylated messages was unclear. (P < 0.0001, two-tailed Mann-Whitney test). Thus, we conclude that Here we have used a circularized rapid amplificatioll of cDNA these products represent mRNAs that have been subject to decayping ends (cRACE) technique to capture mRNA decay intermediates in and subsequent partial 5'+3' decay in vivo. S, ponzbe. Unexpectedly, in contrast to the situation in Sacchnronzyces We next coinpared the 3' ends of adenylated and non-adenylated cerevisine, we find that decapped intermediates often contain sub- transcripts (Fig. lc). Similarly to previous observations17,we detected stantial poly(A) tails, indicative of a novel deadenylation-independent three cleavage and polyadenylation sites in the actl gene: the first is decapping pathway for bulk mRNA in fission yeast. We also show that approximately 1,190 nt downstream from the start codon (and 60 nt 'sir William Dunn School of Pathology, University of Oxford, UK. 2Present address: Whitehead Institute, Cambridge, Massachusetts, USA. Correspondence should be addressed to C.J,N. ([email protected]). Received 29 October 2008; accepted 6 April 2009; published online 10 May 2009; doi:10.1038/nsmb,1601 616 , VOLUME 16 NUMBER 6 JUNE 2009 NATURE STRUCTURAL & MOLECULAR BIOLOGY OPEN ACCESS Freely 8 available online PLoS COMPUTATIONAL BIOLOGY , Ten Simple Rules To Combine Teaching and Research Quentin ~icens',Philip E. ~ourne~" 1 University of Colorado, Boulder, Colorado, United States of America, 2Skaggs School of Pharmacy and Pharmaceutical Science, University of California San Dlego, La Jolla, California, United States of America Thc late Lindley J. Stiles famously made to engage students to brainstorn1 about thc (2) Administer a Web site for your himself an advocate for teaching during risks of GMOs) and your research (e.g., course. Many uilivcrsities and some his professorship at the University of fitlish experiments for this project and start textbooks now offer you the possibility Colorado: "If a better world is your aim, writing before Easter; this week do the of hosting a Web site with course- all must agree: The best should teach" control for my primer binding assay), Make related materials, including automat- (l~ttp://tl~ebestshouldteach.org/).In fact, sure you achieve them. If you don't-this is icaljy gsaded assessments. See, for dispensing high-quality teaching and pro- likely to happen at first-ask yourself how example, the CULearn suite used at fessional education is the priinary goal of legitimate your reason is, Then review and the University of Colorado (http:// - any university [I]. Thus, for most faculty - - adjust the goals accordingly. - - - -- www.colorado.edu/its/culearn/)~-or positio~ls in academia, teaching is a more general automatic grading tools significant requirement of the job. Yet, Rule 3: "Don't Reinvent the presented at http://ctl.stailford.edu/ the higher education programs offered to Wheel" Tomnprof/postings/227.11t1nl, Ph.D. students do not necessarily incorpo- (3) Gather a solid team of motivated rate any form of teaching exposure. We We borrowed the title for this rule from teaching or learning assistants, who offer 10 simple rules that should help you excellent suggestions on How To Pr@areNew will both serve as an intermediary to get prepared for the challeilge of Courses Wzile Ke$ing Your Sa?zi& [2]. Most between you and your students and teaching while keeping some composure. likely, you will not be the first one ever to help you grade. In short, don't be teach a particular topic. So get in touch afraid to ask for help! Rule 1: Strictly Budget Your with the colleagues in your department who Time for Teaching and for have taught the class you are going to " Doing Research teach, or who teach similar topics. You can Rule 4: Don't Try To Explain also use your network and contact former Everything This rule may seem straightfolward, but collcagues or fiiends at other institutions. respecting it actually requires more disci- They will usually be happy to share their Class time should be spent guiding pline and skill than it first appears to. The course material, and along the way you students to create their own explanation key is to set aside time for both teaching and might also glean precious tips fiom their of the material and to develop cognitive research from the beginning, with a well- teaching experience (e.g., a list of do's and abilities that will help them become critical marked separation (e.g., morllings will be don'ts on how to approach a notoriously thillkcrs. In other words, you don't want to devoted to course preparation, afternoons to difficult topic). You will also learn a lot from present all aspects related to a certain topic experiments and manuscript writing). Firm- sitting in one of their classes and watching or to lay out all the explanations for them, ly stick to this agenda, particulasly if this is how they handle their topic and their Thus, an effective way to teach is to get your first time teaching. Failuse to do so students. Here are more examples of students to learn by transformative learn- would eventually affect the quality of your precious time-savers: ing: beyond memorizing and comnprehend- teaching or the progress of your research (or ing basic concepts, they will learn to reflect both). Over time, you will become more (1) Choose a textbook that is accomnpa- on what they learn and how they learn it skillcd at jumping from one commitment to nied by rich online resources such as (see, for example, http://en.wikipedia.org/ annotated figures, prc-made Power- wiki/Transfor~native-lear11ii1g 1 the other, and therefore allowing the and refer- bounduies to fluctuate somewhat. Avoid Point slides, animations, and videos. ences within). Such teaching practices underestimating the time necessary to fulfill Students will thank you for showing require that a significant part of the teaching-related obligations (e.g., office movies, for example, as they often are learning process happens outside the class- hours, test preparation, grading, etc.) by a better option to break down com- room, through reading assigntncnts, home- coitlsulti~lgwith your colleagues. plex mechanisms or sequences of work, writing essays, etc. So make sure you evcnts into distinct steps. budget time to organize these, as specified Rule 2: Set Specific Teaching and Research Goals Citation: Vicens Q, Bourne PE (2009) Ten Simple Rules To Combine Teaching and Research, PLoS Comput Biol 5(4): el000358. doi:10.1371/journal.pcbl.l000358 In ordcr not to have one occupation Published April 24, 2009 overpower the other one-which would transgsess Rulc #I-it is a good idea to Copyright: O 2009 Bourne, Vicens. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, dccide on specific aims for each entell~rise. provided the original author and source are credited, Compile a list of reasonable but spccific Funding: The authors received no specific funding for this article, long-term goals (for the month or the semester) and short-term ones (for the week) Competing Interests: The authors have declared that no competing interests exist. for both your teaching (e.g., finish Chapter * E-mail: [email protected] 3 by Nov. 1; this week propose a discussion Philip E. Bourne is the Editor-in-Chiefof PLoS Computational Biology. !@,: PLoS Computational Biology I www.ploscompbiol.org 1 April 2009 1 Volume 5 1 Issue 4 1 el000358 PLoS COMPUTATIONAL BIOLOGY What Is Stochastic Resonance? Definitions, I Misconceptions, Debates, and Its Relevance to Biology I Mark D. ~c~onnelll*,Derek ~bbott~ 1 Institute for Telecommunications Research, University of South Australia, Mawson Lakes, South Australia, Australia, 2 Centre for Biomedical Engineering and School of Electrical 81 Electronic Engineering, University of Adelaide, Adelaide, South Australia, Australia Stoclzmtic Resona~tce(SR) is the name for a phenomenon that is a Abstract: Stochastic resonance is said to be observed flagship exainple of this idea. It has mostly been studied by when increases in levels of unpredictable fluctuations- , physicists-for Inore than 25 years-but may also be familiar to e.g., random noise-cause an increase in a metric of the 1 some biologists, as well as to those in many other disciplines, quality of signal transmission or detection performance, I Research into SR has had a colorful cvolutiollary journey, and rather than a decrease. This counterintuitive effect relies extgcting importag~pi:i~~ciplesand ~esults&om the litcraty-c can- on system nonlinearities and on some parameter ranges be confusing. ! being "suboptimal". Stochastic resonance has been , ; observed, quantified, and described in a plethora of I In particular, the paradoxical notion of "good noise" is a ! physical and biological systems, including neurons. Being I double-edged sword for SR researchers. To some, working to a topic of widespread multidisciplinary interest, the understand paradoxes and counteri~ltuitiveideas is a significant : definition of stochastic resonance has evolved significant- 1 curiosity-driven challenge. This has drawn many scientists and I ly over the last decade or so, leading to a number of engineers to study SR, leading to many interesting and useful 1 debates, misunderstandings, and controversies. Perhaps theoretical and experimental published results. Others naturally i the most important debate is whether the brain has I focus oidy on the ingrained principle of great utility in engineering, , evolved to utilize random noise in vivo, as part of the where noise needs to be eradicated, and, the Inore it is present, the i ''neural code". Surprisingly, this debate has been for the Inore diminished is performance. This preconceptioil call be a i most part ignored by neuroscientists, despite much sufficient reason for many scientists to ignore or dismiss SR. indirect evidence of a positive role for noise in the brain. The puspose of this essay is to discuss issues that have solnetimes , We explore some of the reasons for this and argue why it clouded the topic. Our first aim is to bring some clarity to the i would be more surprising if the brain did not exploit r debate and to illustrate the pitfalls and controversies for biologists randomness provided by noise-via stochastic resonance unfamiliar with stochastic resonance, or who have held only a I or otherwise-than if it did. We also challenge neurosci- / entists and biologists, both computational and experi- peripheral interest in the area. i The sccond aim is to advocate to readers that \vhen they come ! mental, to embrace a very broad definition of stochastic 1 resonance in terms of signal-processing "noise benefits", 1 across studies of SR, they should focus less on the countcrintuitive and to devise experiments aimed at verifying that random ! idea of "good noise", and instead understand SR in terins of I variability can play a functional role in the brain, nervous 1 "randomness that makes a nonlinearity lcss dctrimental to a system, or other areas of biology. ! signal". Such a change of focus away fi-om "noise" and on to "helpful randomness" may shift thc balance away fiom skepticism of the form "how can tloise be good?" toward thinking "does this Introduction variability have a useful function?" Provoking a discussion on this topic is especially timely, given Noise is an dl-encompassing term that usually describes recent increasing intcrest in the topic of neuronal variability. For undesirable disturbances or fluctuations. In biology, "noise" example, SR is lnentioncd in several recent PLoS articles [I-41, typically refcrs to variability in n~easureddata wl~enidentical while a symposiu~non "Neuronal Variability and Its I?unctional experiments are repeated, or when biosignals cannot bc measurcd Significance" was held in conjunction with the 2008 Society for without background fluctuations distorting the clcsirecl mcasure- Ncuroscieilcc mecting. Furthermore, a recent review on noise in ment. ' Noisc is also thc fundalncntal enemy for colnmunicatioils engineers, whose goal is to ensurc messagcs can be transmitted Citation: McDonnell MD, Abbott D (2009) What Is Stochastic Resonance? Definitions, Misconceptions, Debates, and Its Relevance to Biology. PLoS Comput error-fiee and efficiently from one place to another, at thc fastest Biol 5(5): e1000348, doi:10,137l/journal.pcb1.1000348 possible rate. When ra~mdom nokc in the form of electronic ' Editor: Karl J. Friston, University College London, United Kingdom fluctuations or electromagnetic interference corrupts translnitted messages, this places limits on the rate at which error-fiee Published May 29,2009 , co~nlnunicationcan be achieved. If eveiything else is ideal, then Copyright: O 2009 McDonnell, Abbott. This is an open-access article distrib- noisc is the enemy. uted under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any m,edium, provided But what if not eveiything is ideal? Can an idcal systcm always the original author and source are credited. be implemented in practice? The answer is of course ?to; Funding: MDM is funded by an Australian Research Council Postdoctoral engineering is about clcsigning systems with tracleoffs between Fellowship, DP0770747. The funders had no role in study design, data collection different conflicting objectives, The same could be said of and analysis, decision to publish, or preparation of the manuscript. evolution. Given this, there are circumstances-see below-wllere Competing Interests: The authors have declared that no competing interests ullavoidably present noise or unpredictable fluctuations can be exist. used puyosely, or deliberately introduced to lead to a benefit. * E-mail: [email protected] !@.: PLoS Computational Biology I www.plorcompbiol.org May 2009 1 Volume 5 ( Issue 5 1 el000348 :@.; OPEN a ACCESS Freely available online PLOS one Protein Dynamics in Individual Human Cells: Experiment and Theory Ariel Aharon cohenle, Tomer Kalisky2Y Avi ~a~o'qNaama ~eva- ato or sky', Tamar anon', lrina lssaeva', Ronen Benjamine Kopito3, Natalie ~erzov',Ron ~ilo~,Alex sigals, Uri ~lonl* 1 Department of Molecular Cell Biology, Weizmann Institute of Science, Rehovot, Israel, 2 Department of Bioengineering, Stanford Unlversity and Howard Hughes Medical Institute, Stanford, California, United States of America, 3 Department of Materials and Interfaces, Weizmann lnstitute of Science, Rehovot, Israel, 4Department of Systems Biology, Harvard Medical School, Boston, Massachusetts, United States of Amerlca, 5 Dlvision of Biology, California Institute of Technology, Pasadena, California, United States of America Abstract A current challenge in biology is to understand the dynamics of protein circuits in livlng human cells,,Can one define and test equations for the dynamics and variability of, a protein over time? Here, we address this experimentally and theoretically, by means of accurate time-resolved measurements of endoqenously taqqed proteins In individual human cells. As a model svsw, we choose three stable proteins dis laying cell-ckle-depend% dynamics. We find that protein accumulation with timelylinear for a protein with short mRNA 'lifetime. Both behaviors correspond to a classical model of transcription and translation. A stochastic model, in which genes slowly switch between ON and OFF states, captures measured cell-cell variability. The data suggests, in accordance with the model, that switching to the gene ON state is exponentially distributed and that the cell-cell distribution of protein levels can be approximated by a Gamma distribution throughout the cell cycle, These results suggest that relatively simple models may describe protein dynamics in individual human cells. Citation: Cohen AA, Kallsky T, Mayo A, Geva-Zatorsky N, Danon T, et al. (2009) Protein Dynamics in individual Human Cells: Experiment and Theory. PLoS ONE 4(4): I e4901, dol:10.137l/journal.pone.0004901 I I Editor: Mark lsalan, Center for Genomic Regulation, Spain I I Received December 15, 2008; Accepted January 20, 2009; Published April 17, 2009 I Copyright: O 2009 Cohen et al. This Is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits I unrestricted use, distribution, and reproduction In iny medium, provided the original author and source are credited. I Funding: The authors received funding from the Kahn Family Foundatlon and the Israel Science Foundation. The funders had no role in study design, data I collection and analysis, decision to publish, or preparation of the manuscript. I I Competing Interests: The authors have declared that no competing interests exist. I I * E-mail. [email protected] I 8 These authors contributed equally to this work. I Introduction takes for cells expressing a protein above or below the population nleail to 'relax' towards the average is longer than one cell A goal of systellls biology is to unclerstallcl the clyllalnics of generation, More work is needed to understand the basic protcin lcvels, as proteins are produced and degraded. One aims clyllarnics and differencks of hu~nanprotein lcvels ancl its variation for a mathematical descril~tionof these processes that captures the across the cell cycle in iildiviclual cells. esseiltials and that call be understood intuitively. To address this, we experimentally followed the dyilalnics of Most cument inodels of protein dyilalnics have been established selected human proteins in human cancer cells and test sinlple and tested in micro-organisms. Proteiils in bacterial cells, Tor models to describe these dynamics. The experiments were made example, are well clescribed by production-degradation equations, possible by recent advances in obtaining dynamics of endoge- Tl~cseshow that protein Inass i~measesexl~onentially over time in nously expressed proteins in iildiviclual human cclls at vcry high growing cells, and that proteiil concentration approaches steacly- resdution and accuracy [25], This is based on a library of cell state with exponential dccays [l-51. Such models have been shown cloaes, cach cxl~ressinga clilI'erent fluorescently taggecl protein to be uscful for clcscribing the ineasurcd clynanlics ofprotein circuits expressed from its natural chron~osolnallocation and uilcler its with several intcractions, such as feed-forwarcl loop iletwork 1notifs native regulation. From this library, we chose three p:oteins that [GI and auto-rcgulatory loops [4.,7-91, Similarly, experiment and accumulate tliroughout the cell cyclc, but are clegadecl evely tinle theoly has established an unclerstancli~~gof cell-cell variability of that the cell clividcs. Thcse proteius show little inelnoiy of previous protein levels in bacterial and yeast cclls [lo-1 51. Protein cell cycles, and thus are a good starting point for testing illoclels distributions in micro-organisms liavc been measurccl at steady- based on current cell properties. I11 the present study, we state ancl are well described by Galnma distributions [1 6,171, . calibrated this assay to provide units of protein concentration, Much less is known about the ccll-cell variability of protein Wc find that the proteills are found to follow eitllcl. an dynamics in human cells [18], Direrences in cxpressioll levels al~proxiinatelylinear or a quadratic dependence on time. They between inclivid~~alcells of a clollal population were previously also clilI'er iin their oilset tilne of production ancl in their variability obsel-vcd [19-221 and related to stochastic processes. More across thc cell cycle. M'c fillcl that thcse fcatures call IIC ullderstoocl recently, Raj el a/. [23] have followed lnRNA aild protein to in terms of a single unified model, Approximately linear show bursts of mRNA production, Sigd el al, [24] ~neasui4ecl accum~~lationresults from short mRNA half-life, ancl q~ladratic variability and inelllory in nuclear proteins, Gndiilg that the time it acclun~ilation from long mRNA half-life, as suggcstecl by 1 April 2009 1 Volume 4 1 Issue 4 1 e4901 OPEN 8 ACCESS Freely available online PL"s BIOLOGY - Stochasticity in Protein Levels Drives Colinearity of Gene Order in Metabolic Operons of Escherichia coli I Karoly ~ovacs',Laurence D. ~urst'*,Balazs papp'* 1 Institute of Biochemistry, Biological Research Center, Szeged, Hungary, 2 Department of Biology and Biochemistry, University of Bath, Bath, United Kingdom Abstract In bacterial genomes, gene order is not random. This is most evident when looking at operons, these often encoding enzymes involved in the same metabolic pathway or proteins from the same complex. Is gene order within operons nonrandom, however, and if so why? We examine this issue using metabolic operons as a case study. Using the metabolic network of Escherichia coli, we define the temporal order of reactions. We find a pronounced trend for genes to appear in operons in the same order as they are needed in metabolism (colinearity). This is paradoxical as, at steady state, enzymes abundance should be independent of order within the operon. We consider three extensions of the steady-state model that could potentially account for colinearity: (1) increased productivity associated with+higher expression levels of the most 5' genes, (2) a faster metabolic processing immediately after up-regulation, and (3) metabolic stalling owing to stochastic protein loss, We establish the validity of these hypotheses by employing deterministic and stochastic models of enzyme kinetics. The stochastic stalling hypothesis correctly and uniquely predicts that colinearity is more pronounced both for lowly expressed operons and for genes that are not physically adjacent. The alternative models fail to find any support. These results support the view that stochasticity is a pervasive problem to a cell and that gene order evolution can be driven by the selective consequences of fluctuations in protein levels. Citation: Kovics K, Hurst LD, Papp B (2009) Stochasticity in Protein Levels Drives Colinearity of Gene Order in Metabolic Operons of Escherichia coli. PLoS Biol7(5): el0001 15. doi:l0.1371/journal.pbio.1000115 I I I Academic Editor: Jeffrey G. Lawrence, University of Pittsburgh, United States of America I 1 I Received October 7, 2008: Accepted April 14, 2009; Published May 26, 2009 I Copyright: O 2009 Kovics et al. This is an open-access article distributgd under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited, I Funding: This work was funded by the International Human Frontier Science Program; Hungarian Scientific Research Fund; Bolyai Fellowship of the Hungarian Academy of Sciences; Royal Society Wolfson Research Merit Award; and the Dr. Rollin D. Hotchkiss Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. I I Competing Interests: The authors have declared that no competing interests exist. I I I Abbreviations: SD, standard deviation. I I I * E-mail: L.D.Hurst@bath,ac,uk (LDH); [email protected] (BPI I Introduction patterns. Here, then, wc ask whetller gene order within operons is ~~nderselection and if so kvl~y?I11 particular, we investigate wllether It is well established that the chrolnosornal distribution of genes is witllill operolls of^, colireflect the functional not random, and in mally genolnes, genes that med to .be order ofthe ellcodecl enzymes, i.e., colinearity (Figure 1). coexpressed tcnd to cluster [I+]. Thc coexpression of adjacent genes can be enforced by the action of bidirectional proinotcrs [5], sim~~ltaneo~~sopening and closing of cllrornatin [6,7], trailscrip- tional spill-over [8], or by inclusion within the same operoil [9,10]. Excess of Colinearity in Metabolic Operons Givcn that prol :' :@,: PLoS Biology I www.plosbiology.org 1 May 2009 1 Volume 7 1 Issue 5 I el0001 15 PHYSICAL REVIEW E 79, 03 19 11 (2009) Stochas ticity of gene products from transcriptional pulsing Srividya ~~er- isw was, 11* E ~a~ot,~and C. ~a~a~rakash' '~eportmrr~tof Physics, Ollio State University, WoorlriflAvenrre, Col~nabas,Ohio 43210, USA 'llel,n~-t,aent qf Ncirralogy, Moent Si~~oiSclzoo/ of Medicine, Levy Place, New York, ~e1vYork 10029, USA (Received 28 October 2008; published 23 March 2009) Transcriptional pulsing has bcen observed in both prokaryotes and eukaryotes and plays a crucial role in cell-to-cell variability of protein and mRNA numbers. An important issue is how the time coilstants associated with episodes of transcriptional bursting and lnRNA and protein degradation rates lead to different cellular lnRNA and protein distributio~~s,starting from the transient regime leading to the steady state. We address this by deriving and then i~ivestigati~~gthe exact time-dependent solutio~lof the master equation for a lrilnscrip- tionrrl pulsing ~nodclof mRNA distributions. We fncl a plethora of results. Wc show that, among others, bimodal and long-tailcd (power-law) distributions occur in the steady state as thc rate constailts are varied ovcr biologically significant time scales. Since steady state may not be reached experilllentally we present results for the time evolution of the distributions. Because cellulat*behavior is determined by proteins, we also investigate the- effect-- of the dii'i'erent- -- mRNA dist~ibutionstg the con-esponding protein clist~ibuj,ionsusing numerical - - simulations, DOI: 10,1103/PhysRevE.79.0319 11 PACS numbei*(s): 87.10.Mn I. INTRODUCTION plest model of mRNA and protein number distributioiis is to consider both transcriptio~and translation as Poisson pro- Cell-to-cell variability in rnRNA and protein nulnbers is cesses [7]. Recent experi~nentalstudies of 1nRNA distribu- now recognized as a major aspect of cellular response to tions have showil strong evidence for transcriptional noise stimuli, a variability which is hidden in cell population stud- beyond what can be described by a simple Poisson proccss, ies. The most egregious exainple of the latter is provided in i.e., distributions with variances signilicantly larger than the cases where a graded average response hides the all-or- mean have been observed. In particular, transcriptional puls- nothing behavior of single cells [I-31. Variability of cellular ing, where bursts of transcription alteionatewith quiescent response can have many origins, which are generally classi- periods, has bcen observed in both prokaryotes and eultary- fied as extrinsic and intrinsic noise or fluctuatio~~s[4]. Many otes, Raser and O'Shea [5], who studied intrinsic and extrin- studies, both experimental and theoretical from bacteria to sic noise in Snccliaror.r~~~~cescerevisine, showed that the noise eukaryotes, have been undertaken to characterize intrinsic associated with a particular pror~lotercould be explained in a and extrinsic fluctuations [4-111, Extrinsic fluctuations can transcriptional pulsing model and confirmed it by mutational have nlultiple origins, such as variations from cell to cell in analysis. Transcriptional bursts were recorded in E', coli [12] the number of regulatory n~olecules or signaling cascade by following inRNA production in time and their statistics components. The source of intrinsic fluctuations is the ran- computed. Evidence for a pulsing model of transcription, dom occurrence of reactions that can lead to variability for obtained from fluorescenl microscopy, has also been pre- genetically identical cells in identical fixed environments. sented for the eqression of the discoidin Ia gene of Dicly- When the numnbcr of rnolccules is small, intrinsic noise can ostellunz [13]. Transcriptional bursts have as well been de- play a significa~ltrole in determining the behavior of indi- tected in Chinese l~ainster ovary cells [lo]. In thesc vidual cells, The difficulty in determining rate constants ex- expcri~nentsthe production of' mRNA occurs in a sequence periinentally and the possibility of experinlenls being either of bursts of transcriptional activity separated by quiescent in the steady state or in the transient regime make exact periods. Transcriptional bursting, an intrinsically rand0111 phenomenon, thus becomes an inlportant cl clnent to consider titnc-depe~~dentsolutions to modcls especially useful in the interpreting observed behavior of intrinsic noise. In this pa- when evaluating cell-to-cell variability. One can predict that per we present the exact time-dependent solution to a widely in many cases it will be a significant part of overall noise and most certainly of intrinsic noise. used model for tmnscriptioaal noise and discuss its implica- tions. Our study foc~~seson the consequences of transcriptional bursting in a si~nplifiecl~noclel of transcription that has l3eel.1 thc subject of Inany stuclics and is believcd to encapsulate the 11. EXPEKIMENrl'ALAND THEORETICAL key features of bursting [2,5,10,12,14-161, Related nlodels BACKGROUNDS that include feedback have been studied theoreticafly [17-191. The complex phenomena that can occur in tran- Intrinsic fluctuations arise from either noisy transcription, scription (cbrolllatin remodeling, ellllallceosome formation, noisy translation, or both, the effects of which can be mea- preinitiation assembly, etc,) are lnocleled tllrough positing ill single cell n1RNA and proteill experimellts. The sim- lwo of gellc an inactive stale, where no tran- scription occurs, and an active one, in which transcription ' occurs according to a Poisson process. The production of mRNA is thus pulsatile: tenlporally there are periods of in- 02009 The American Physical Society Studying membrane proteins through the eyes of the genetic code revealed a strong uracil bias in their coding mRNAs Jaime Priluskya and Eitan Bibibll aBioinformatics Unit, Department of Biological Services, and bDepartment of Biological Chemistry, Weizmann Institute of Science, Rehovot 76100, Israel Communicated by H. Ronald Kaback, University of California, Los Angeles, CA, February 27, 2009 (received for review December 30, 2008) Posttranscriptional processes often involve specific signals in mRNAs. offered (10, I 1). Tt wiis soon realized that chemically sinlilar Because mRNAs of integral membrane proteins across evolution amino acicis are often encocled by relatively similar codons (12, are usually translated at distinct locations, we searched for uni- 13) and that vcry hydrophobic amino acids are encoded by versally conserved specific features in this group of mRNAs. Our cadons having iiracil (U) in the second position (14). Our analysis revealed that codons of very hydrophobic amino acids, anczlysis rcvcaled that, in addition to their sccond position, highly represented in integral membrane proteins, are composed codons of very hydrophobic amino acids have aremarkably high of 50% uracils (u). As expected Fom such a strong bias,-che U content in general (Fig. l),Specifically, 50% of the combined calculated U profiles of ~RNASclosely resemble the hydrophobic- numbers of nuclcotidcs in thesc codons arc Us. In contrast, thc ity profiles of their encoded proteins and may designate genes U content in codons of all other groups of amino acids is r22%, encoding integral membrane proteins, even in the absence of and the total U content in all of the 61-aa coding triplets is information on ORFs. We also show that, unexpectedly, the U- 24.6%, suggesting a strong U bills in mRNAs encoding integral richness phenomenon is not merely a consequence of the codon membrane proteins. Next, we investigated whether the proposed composition of very hydrophobic amino acids, because counterin- U bias is an inherent requirement for n1RNAs encoding integral tuitively, the relatively hydrophilic serine and tyrosine, also en- membrane proteins or merely a trivial consequence of the high coded by U-rich codons, are overrepresented in integral membrane U content in codons of very hydrophobic amino acids. As shown proteins. Interestingly, although the U-richness phenomenon is in Fig. 24there are 2 relatively hycirophilic amino acids, serine conserved, there is an evolutionary trend that minimizes usage of and tyrosine, both encoded by U-rich codons (33% and 50% U-rich codons. Taken together, the results suggest that U-richness U, 1~cspcctively).On the basis of the chclnical nature of serine is an evolutionarily ancient feature of mRNAs encoding integral and tyrosine, we predicted that both of them should be more membrane proteins, which might serve as a physiologically rele- abundant in soluble proteins. I11 contrast, analysis of their vant distinctive signature to this group of mRNAs. usage in multipass membrane and cytoplasmic proteins from various organisnls (supporting infc~rmationTables S1 and S2) evolution I hydrophobicity scale I mRNA targeting I U-rich mRNA revealed higher content of serine and tyrosine in the mem- brane protein group (-20% more) (Fig. 2B). These results n addition to protein-coding information, n1RNAs some- strongly suggest that integral membrane protein transcripts I times harbor signals requirecl for posttranscriptional regu- might have been programmed or had evolved to contain high latory puthways, such as processing, translation, degradsrtion, contents of U. and localization (1, 2). For selective targcting, n1RNAs usc various protcin-intcraction detcrlninants (structural, sequence Analysis of the Distribution of U in Membrane Protein mRNAs. specific, or nonspecific) (3), nlostly in untranslated regions, Traditionally, integral membrane proteins are analyzed for altl~oughunique exceptions have been described (e.g., ref, 4). their hydrophobicity profiles (15), using algorithms that help mRNAs of integral mcmbr~neproteins across cvoli~tionare identify their transmembrane helices. We wondered whether usually translated at clistinct locations, and our stilclies in our discovery of the high U content of very hydrophobic Escllel-icl~iacoli have suggested a step through which these codons might be helpful in identifying integral membrane nlIiNAs might be selectively targeted to inenlbrane-boullcl proteins through the analysis of U profiles of genes. Initially, ribosorncs (5-7). Thcreforc, we investigated thc possibility that we compared the hydrophobicity profiles of several integral mRNAs encoding integral membrane proteins have species- membrane proteins with the U profiles of their mRNAs. Fig. 3 independent characteristic features, which might provide an shows several examples in which both curves are strikingly evolutionarily conscrvcd means for their sclcctivc recognition similar. MalF is a complex integral membrane protein with 8 and targeting to the membrane. As a basis for our analysis, we transmembrane helices and a large external hydrophilic domain reasoned that because proknryotes express polycistronic tran- (Fig. 3A, Top) (16). The calculated Kyte-Doolittle-based hydso- scripts, sometimes encoding a mixture of membrane and phobicity profile of MalF (Fig, 3A, Middle) supports the pro- cytosolic proteins, targeting signals might be located inside posed secondary structure model. Remarkably, the calculated U ORFs in addition to untranslated regions. profile of the malF mRNA also supports this model, and the 2 profiles are very similar. Notably, although similar, there are Results and Discussion subtle differences that might indicate the importance of features Analysis of the Genetic Code in mRNAs Encoding Integral Membrane other than the identified relationship between the protein hy- Proteins, A ~~iliqi~eproperty of integral membrane proteins is that they hitve stretches containing very hydrophobic amino acid rcsiducs (-20). Thcrcfore, we iillalyzcd the nuclcotidc compo- Author contributions: E.B.designedresearch; J.P.and E.B. performed research; J.P. and E.B. sition of vcry llydrophobic codons [according to thc Goldman, analyzed data; and E.B. wrote the paper, Engelman, Steitz (GES) scale] (8), in the context of the entire The authors declare no conflict of interest. genetic code. Since the first description of the nearly universal 'To whom correspondence should be addressed. E-mail: e.bibiOweizmann.ac.il. genetic code (for a review see ref. 9), various explunations of its This article contains supporting information online at www.pnas.org/cgl/content/fulll organization and thc assignment of the 64 triplcts havc been 0902029106/DCSuppiemental. 6662-6666 1 PNAS I Apri121,2009 1 vol. 106 I no. 16 www.pnas.org/cgi/doi/10. 1073/pnas50902029106 . Complexity in bacterial cell-cell communication: Quorum signal integration and subpopulation signaling in the Bacillus subtilis phosphorelay llka B. Bisch~fs~~~~l,Joshua A. HugC,Aiwen W. Liua, Denise M. Wolfb, and Adam P. Arkinaebdl Departments of agioengineering and CElectricalEngineering, University of California, 31 1 Hildebrand Hall, MC 5230, Berkeley, CA 94704-3224; and b~hysicalBiosciences Division, Lawrence Berkeley Laboratory, 1 Cyclotron Road, MS Calvin, Berkeley, CA 94720 Edited by John Ross, Stanford University, Stanford, CA, and approved March 20, 2009 (received for review October 30, 2008) A common form of quorum sensing in Gram-positive bacteria is formation unless there is a starvation signal in addition, whereas mediated by peptides that act as phosphatase regulators (Phr) of starvation alone is able to triggcr sporulation, albeit at a reduced receptor aspartyl phosphatases (Raps). In Bacillus subtilis, several efficiency (5,ll). It therefore appears that the Rap-Phr-systems Phr signals are integrated in sporulation phosphorelay signal have evolved toward a mode of yuorunl signaling to which signal transduction. We theoretically demonstrate that the phosphorelay integration is csscntial. can act as a computational machine performing a sensitive division It has recently been shown that even monocultures show a operation of kinase-encoded signals by quorum-modulated Rap large degree of diversity by heterogeneous cell differentiation signals, indicative of cells computing a "food per cell" estimate to ancl populatio~lstructuring (12). Stochastic population splitting decide whether to enter sporulation. We predict expression from has been demonstrated for a variety of stress responses inclucling the rapA-phrA operon to bifurcate as relative environmental sporulation, competence, motility, extracellular matrix produc- signals change in a developing population. We experimentally tion, and cxocnzyrne synthesis (13-16). The coordinated devcl- observe that the rapA-phrA operon is heterogeneously induced in oplnent of specialized differentiated subpopulations iinplies a sporulating microcolonies. Uninduced cells sporulate rather syn- potential need for cell-cell signaling systems that facilitate chronously early on, whereas the RapAIPhrA subpopulation sporu- communication among and across differentiating subpopula- lates less synchronously throughout later stationary phase. More- lions. The Rap-Phr genes comprise a promising family to over, we show that cells sustain PhrA expression during periods of mediate such subpopulation signaling. active growth. Together with the model, these findings suggest Here, we begin to explore these two aspects of Rap-Phrs in that the phosphorelay may normalize environmental signals by the pllosphorelay signaling: signal integration and subpopulation sig- size of the (sub)population actively competing for nutrients (as naling. First, we develop tt co~nputationalmodel to address the signaled by PhrA). Generalizing this concept, the various Phrs could interplay of quorum and e~~vironmentalsignals in pliosphoreli\y facilitate subpopulation communication in dense isogenic commu- signaling and show that the signal transduction architecture nities to control the physiological strategies followed by differen- supports a sensitive ratiometric integration of kinase-encoded tiated subpopulations by interpreting (environmental) signals cnvironmer~talsignals with rcspect to quorum-modulated phos- based on the spatiotemporal community structure. phatase signals. This is i~lclicativeof a cell-densi ty-dependent nornlalization of the environmental stressors such as starvation heterogeneity I Phr I quorum sensing I sporulation I model to clrive clownstream clecision making. Within this theoretical framework, we predict a bifurcation of RapA/PhrA expression as he 11 hon~ologous~vay-phr genes in Blrcill~i~ssubtilis code for rclativc cnvironnlental conditions change. Wc thcn experinlctl- T a family OF proteins in which Rap activity is regulated by tally elucidate the clynamics and fate of the previoirsly observed snlall peptides derivcd from the cognate l~lirgcrlc product (Fig. subpopulation of RiipA/PhrA-expressing cells (17) late into L4) (I). The small Phr precursor peptides are cleaved i~~lcl stationary phasc and suggcst the PhrNRapA may be a subpopu- secreted from the cell, and at least some of them itccilmulate in lation comn~unicationsystem to measure its own population size. the culture supernatant (2). They are subsequently iinported by Although the experilllents do not directly test the theory, the Opp oligopeptide perineases into the cytoplasm where the Phr theory aids in illterpreting thc importance of the cxperimciltal derived pentapeptides (PEPS) intracellularly inhibit the activity observations. Toge ther, our colnpu tational and experimental of their cognatc Raps (3-5). Thus, PEPSs can bc indicators of cell results suggest a model in which phosphorelay-based decision density and have been implied in facilitating cell-cell commu- making ciuring starvation is based on normalizing environmental nication (2, 5-8). A subset of Raps (RapA,B,E,H) act as signals (such as "foocl") with respect to the size of a growing phosphatascs that inhibit signaling through a cc~ltralphosphorc- subpopulation actively competing for nutrients ("Fooci per grow- lay, and their activity is coirnteracted by their cognate PEPSs ing ccll"), Gcncralizillg this concept, we suggest that phosphore- (PhrA,C,E,II), The phosphorelay pli~ys a central role in R. lay-integratecl Rap-Phrs play a central role in determining and siibtilis stress response inductio~lduring stationary phase and nxtintaining cell differentiation development by integrating (en- ultimately controls sporulation induction (Fig. 1B) (9). vironmcntal) signals bascd on the local coinlnu~litystructure, Unlike autoinducer quorum signals, where a~rtoinclucermol- ecules bind to a transcription factor, peptide based quorum Authorcontributions: I.B.B., D.M.W., andA.P.A.designed research;I.B.B.,J.A.H., andA.W.L. signriling indirectly' rcgulatcs transcription by controlling phos- performed research; 1.B.B.contributed new reagentslanalytictools; I.B.B., J.A.H., and A.P.A. phoryl signaling (10). Thc presence of an autoinducer signal is analyzed data; and I.B.B. and A.P.A. wrote the paper. typically sufficient to activate a pathway and induce a cell The authors declare no conflict of interest. response (Fig. 2A Left). 111 co!ltntst, because [he Phr signal iicts This article is a PNAS Direct Submission. on a phosphatasc, it instructs cclls by rcgulating thc phosphoryl 'To whom correspondence may be addressed at: Physical Biosciences Division, Lawrence flux originating from the activation of kinases that are uncler Berkeley Laboratory, 1 Cyclotron Road, MS Calvin, Berkeley, CA 94720. E-mail: control of the environment (Fig. 2A Riglit). For example, it is [email protected] ibbischofsQlbl.gov. well ltnown that sporulation is affected by cell density. However, This article contai~ssupporting information online at www.pnas.org/cgi/contentlfull/ rz high ccll density alone is not sufficient to trigger spore 0810878106lDCSupplemental. www.pnas.org/cgl/doi/10.1073/pnas.0810878106 PNAS I April 21,2009 1 vol. 106 I no. 16 1 6459-6464 Transcriptome transfer produces a predictable cellular phenotype Jai-Yoon Sularl, Chia-wen K. Wua.l, Fanyi Zengaebn1,Jeanine Jochemsa, Miler T. Leecld,Tae Kyung Kima, Tiina Peritza, Peter Buckleyalc,David J. Cappellerie, Margaret Maronskif, Minsun Kimg, Vijay Kumarcte,David Meaneycth, Junhyong Kimc~d~l,and James Eber~ine~~~~~~l~~ Departments of aPharmacology, cPenn Genome Frontiers Institute, d~iology,eMechanical Engineering and Applied Mechanics, fNeuroscience, hsioengineering, and 'Psychiatry, University of Pennsylvania, Philadelphia, PA 19104; bhanghai Institute of Medical Genetics and Institute of Medical Sciences, Shanghai Jiao Tong University School of Medicine, Shanghai 200040, Peoples Republic of China; and 9Department of Physiology, Wonkwang University School of Medicine, lksan 570-749, South Korea Communicated by William T. Greenough, University of Illinois, Urbana, IL, March 5, 2009 (received for review January 23, 2009) Cellular phenotype is the conglomerate of multiple cellular processes differentiated phenotype, as well as the dyilanlic changes of a involving gene and protein expression that result in the elaboration genomc (7). A key question is whcther thcre is n general strategy of a cell's particular morphology and function. It has been thought lo efficiently reprogram a cell to ally target-cell type. that differe-ntiied postmitotic qellshave their genomes hgrd-,red, cDNA microarray analysis has shown thttt phenotypic differences with littleability for phenotypic plasticity. Here we show that transfer at the cellular level are associated with differences in the presence, of the transcriptome from differentiated rat astrocytes into a nondi- :ibsence, ltnd abundance of particular RNAs. As the transcriptome viding differentiated rat neuron resulted in the conversion of the contaiils thc RNAs that encodc the protcins llcccssary to gencrate neuron into a functional astrocyte-like cell in a time-dependent and maintain phenotype, such as kinases, phosphatase, histone manner. This single-cell study permits high resolution of molecular ace tylasw and deace tylases, and other proteins, including transcrip- and functional components that underlie phenotype identity. The tion factors that collaborate in the right proportions to generate the RNA population from astrocytes contains RNAs in the appropriate phenotype, it is reasonable to assume it llas a donlinant role in relative abundances that give rise to regulatory RNAs and translated determining phcnotype. To test this l~ypothcsisusing single-cell proteins that enable astrocyte identity. When transferred into the approaches, we introduced the transcriptome from donor cells into postmitotic neuron, the astrocyte RNA population converts 44% of individuul host cells and qualltitntively assessed conversion froin the the neuronal host cells into the destination astrocyte-like phenotype. host to the donor cell phenotype (12, 13). In support of this observation, q'uantitative measures of cellular morphology, single-cell PCR, single-cell microarray, and single-cell Results functional analyses have been performed. The host-cell phenotypic Modeling of the Phototransfection of Complex RNA Populations. To changes develop over many weeks and are persistent. We call this evaluate whether the astrocyte transcripto~necan directly convert process of RNA-induced phenotype changes, transcriptome-induced host neurons into astrocytcs, we first charactcrized our ccll culti~res phenotype remodeling. to ensilre the purity of the neurons that are to be transfected [supporting information (SI) Fig. Sf]and then established a work- neuron I transcriptome-induced phenotype remodeling I single cell I ing transcriptome-induced phenotype remodeling (TIPeR) proto- Waddington col, which consists of multiple RNA phototransfections over a 10-day period (Fig. 1A). Phototransfection was selected as the means to transfect the astrocyte transcriptome into neurons, as it 11 n~ulticellularorganisn~s, a11 cells contain nearly identical copies can transfect RNA into neurons with high efficiency (12). To Iof the genome but exhibit drastically different phenotypes. Even optimize for the amount of RNA that can be introduced into the a single ncuron has a sct of phcnotypic chrrractcristics that distin- host cells, we modeled the phototransfection of RNA transcripts guish it from other neurons as well as other cell types, such as the that diffuse into the photoinduced pores on the cell membranes nearby astrocytes. Indeed, as Wadclington proposecl in his classical (Fig. S2). This process was modeled with an average transcript size epigenetic Ianclscaye model, genetically predetermined cells can of 1.5 kb + 0,2 kb with the effective radius of the transcript follow any spccific permitted trajcctorics that cveiltually lcad to determined by the Flory approximation (Rtranscript5.5N1I3, where different cellular phenotypes (I). From this point of view, the N is the size of the transcript in bases). Pore sizes were systematically genome serves as a repository of dynamic control information varied to test the relative flux of transcriptome in a typical photo- whose state can be rcpwgran~incdto match thc stable phcnotypic transfection experiment. Simulations with the approximated tran- states. scriptome cargo showed that a single sequence of 16 pulses across Emerging evidence has demonstrated the reversibility and flex- the cell membrane would be sufficient to deliver a large number of ibility of the ccllulnr phcnotype. Gurdon ct al, first showcd that thc transcripts while retaining their relative abundances (Fig. 1B and ability to obtain fcrtile adult nlalc aid fcmalc frogs by injecting endoderm nuclei into enucleatect eggs (2). This result not only forms the founclation of the field in r~~lclearlrailsplantation but also Author contributions: J.-Y.S., C.K.W., F.Z., M.T.L., T.K.K.,T.P., D.J.C., M.K., V.K., D.M., J.K., provides cvidc~lccthat thc cytoplasmic corllponcnts of a diffcrcn- andJ.E.designedresearch;J.-Y.S.,C.K.W.,F.Z.,J.J.,M.T.L.,T.K.K.,T.P.,P.B.,D.J.C.,M.K.,V.K., tiated cell can support nuclear replaogl.aming,Generation of in- J.K., and J.E. performed research; J.-Y.S., M.T.L., D.J.C., M.M., V.K., D.M., J.K., and J.E. contributed new reagentslanalytlc tools; J.-Y.S., C.K.W., F.Z., J.J., M.T.L., T.K.K., T.P., P.B., duced pluripotent stem (iPS) cells by transfection of transcription D.J.C., M.K., V.K., D.M., J.K., and J.E.analyzed data;andJ.-Y.S., C.K.W., F.Z., J.J., M.T.L., T.P., fiictors illto dividing fibroblasts (3), followed by cell selection, P.B., D.J.C., M.M., M.K., V.K., D.M., J.K., and J.E. wrote the paper. reprcqents a new strategy to globally revert ii n~atiirecell into a Conflict of interest: WilliamT. Greenough and J.E. havecollaborated on past research. They different cell type (4-9). Tlle need fc>r redifferentiation of these are not currently collaborating.The remaining authors declare no conflict of interest. embryonic stem cell-like-iPS cells into desired cell types adds a layer Freely available online through the PNAS open access option. of con~plexitythat is difficult to control (10, 11). Nevcrtheless, 1J.-Y. 5, C.K.W., F.Z., J.K., and J.E. contributed equally to this work. studies of nuclear rel>rogramming from genomic and epigenetic 2To whom correspondence should be addressed. E-mail: [email protected]. modification, as seen from somatic-cell nuclear-transfer-clonecl This article contains supporting information online at www.pnas.orglcgi/content/full/ aili~nalsand iPS cells, strongly dcr~lonstratethe flexibility of rr 0902161 106/DCSupplemental. 7624-7629 1 PNAS I May 5,2009 I vol. 106 I no. 18 www.pnas.org/cgi/doi/10.1073/pnas.0902161106 The cost of gene expression underlies a fitness trade-off in yeast Gregory I. Langall, Andrew W. M~rray~~~~~,and David 80tstein~~l~~ alewis-Sigler Institute for Integrative Genomics and the Department of Molecular Biology, Princeton University, Princeton, NJ 08544; and bFAS Center for Systems Biology and the Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138 Contributed by David Botstein, February 12, 2009 (sent for review December 10, 2008) Natural selection optimizes an organism's genotype within the testing their fitncss. Iirrploid, a-mating type (MATa) cclls, arrest context of its environment. Adaptations to one environment can in GI when exposed to the mating pherotnone, alpha-factor decrease fitness in another, revealing evolutionary trade-offs. (aF), and tlli~scannot form colonies on media containing aF, Here, we show that the cost of gene expression underlies a We initiated, from a singlc colony of haploid MAT8 cells, a large trade-off between growth rate and mating efficiency in the yeast llunlber of parallel cultures that were plateci onto either rich Saccharomyces cerevisiae. During asexual growth, mutations that medin or rich media containing af;.On rich media, the vast eliminate the ability to mate provide an =2% per-generation majjority of cells form colonies, but 011 cuF, only the small fraction growth-rate-advantage. Some strains, including-most laboratory of cells that have acquired mutations in pheromone-induced strains, carry an allele of GPAl (an upstream component of the signaling can form colonies. From each culture, we randoinly mating pathway) that increases mating efficiency by -30% per chose a single alpha-factor resistant (uFlZ)or unselected colony round of mating at the cost of an -1 % per-generation growth-rate and measured its relative growth riite by irsing a FACS-based disadvantage. In addition to demonstrating a trade-off between compctitivc growth-rate assay that can dctect growth-rate dif- growth rate and mating efficiency, our results illustrate differences ferences as small as 0.5%). The growth-rate coefficient is a in the selective pressures defining fitness in the laboratory versus measure of the growth-rate advantage ovcr wild type. Fig. 1A the natural environment and show that selection, acting on the shows the growth-rate coefficients (s,) for 27 irnselectecl clones cost of gene expression, can optimize expression levels and pro- and 45 uFKclo~les. AS a control we measured the relative growth mote gene loss. ratcs of 24 sinlilarly selectccl mutants that were resistant to canavanine, a toxic arginine analog. In each case, several clones evolution I GPAl I mating pathway I Saccharomyces cerevisiae have a low growth sate (s, < -I%), suggesting that these strains hrwc become mitochondria1 deficient or hswe acquired a dele- frequent observation in evolutio~lis that traits not main- terious mutation. Exclucli~~gclones wit11 s, < - I(%, the growth- A tained by selection will be lost-this holcls true at the rate coefficients of the u~lselectedclones follow a tight distri- ~norphologicallevel and at the genetic level. Examples of gene bution (Fig. IA, s, = 0.08% 4 0.35%) indistinguishable from the loss include the loss of olfactory receptors in primates (I), the distribution of the canavanine-resistant mutants (Fig. ZA, s, = loss of pig~nentationand vision in Asty(m(u cavefist1 (2), the loss 0.36% + 0.48%, Y > 0.05, Wilcoxon rank sum test); howevcr, tile of the galactose utilization pathway in yeast (3), and the degen- growth-rate coefficients of the uFR inl~tantsshow greater vari- eration of genes involved in c~~rbonutilization during domesti- ation and a positive growth-rate advantage (Fig. 1A, s, = 1.48% cation of Streptococc~lstllernlophilus (4). Such rcgrcssive cvolu- 2 0.85%, 1.' < lW7, Wilcoxon). tion also occurs in laboratory l~opulations;reduction in cdtabolic It appears from these data that at least some sterile nlutants breadth and thermal tolerance is observed during long-tcrn~ have a clear growth-rate advantage over wild type, To determine evolutioll in Escherichia coli (5-9), and sterility frequclltly arises whetlicr all stcrile strains have a similar advantage, and to during long-term asexual propagation of Sacchnl-omn))cesccerevi- cictermine the basis for any growth-rate advantage in the sterile sicle (10). strains, wc used a combination of 4 methods: Phenotypic char- Two mechanisms could account for gene loss durillg evolu- acterization of the spontaneous upRnlutants, growth-rate assays tion. One possibility is that in the absence ol'selection, genes are on targeted gene deletions within the mating pathway, mapping lost because of the neutral accumulatioil of mutations. Alterna- of the mutations in thc most fit stcrilc strains, and expression tively, gene loss events coulcl be clriven by selection. The analysis on uFK strains both with i~ndwithout a growth-rate observation that Inany of these gene-loss events are repealedly t~civantage, observed supports this hypothesis. Repeated loss of all or part of The ycast mating pathway is onc of thc best studicd mitogcn- tlle Rbs operon (whose products catabolize ribose) in E, coli activatecl protein (MAP) kinase cascades (12). At the beginning provides a selective advantage in mininlal glucose media (8), of the pathway is a pheromone receptor (Ste2 in MATa or Ste3 Quantitative analysis of alleles leitding to eye reduction in in MATcY)that binds the cognate mating pheromone. Receptor A,stcijmnm indicates that selection, possibly against the energetic stimulation activates a heterotrimeric G protein (consisting of cost of vision, is responsible for eye degeneration in cavefish Gpal, Stel8, and Ste4), which in turn, activates a MAP kinase populations (11). Thcse studics suggest that haploid ycast that cascade (consisting of the MAP kinase kinase kinase, Stel1, the arc propngatcd for long pcriods without mating partners should MA1' kiniise Icinase, Ste7, the MAP Icinases Fus3 and Kssl, and become sterile, Previous stucties, howevcr, showccl that lincagcs that cvolvcd higher growth rates arid lowcr nlating cfficiclzcics appeared to segregate these traits indepenclently (10). Here, we Author contributions: G.I.L., A.W.M., and D.B. designed research; G.I.L. performed re- set out to directly lest whether selection tlrives yeast to become search; G.I.L. analyzed data; and G.I.L., A.W.M., and D.B. wrote the paper. sterile by determining whether mutations conlerring sterility The authors declare no conflict of interest. provide a selective advantage. Freely available online through the PNAS open access option. 'To whom correspondence may be addressed. E-mail: [email protected], Results amurrayBmcb.harvard.edu, or [email protected]. Sterility Increases Growth Rate by Eliminating unnecessary Gene ~A.w.M.and D.B. contributed equally to this work. Expression, We tested the hypothesis that sterile strains gellerally This article contains information online at www,pnas.org/cgi/content/ml/ have a growth-rate advri~itageby isolating sterile mutants and 0901620106/DCSuppIemental. www.pnas.org/cgl/doi/10.1073/pnas.0901620106 PNAS I April 7, 2009 1 vol. 106 1 no. 14 1 5755-5760 Aleksandra M. Walczakall, Andrew ~ugler~,and Chris H. WigginsC aPrinceton Center for Theoretical Science, Princeton University, Princeton, NJ 08544; and Departmentsof b~hysicsandCApplied Physics and Applied Mathematics, Center for Computational Biology and Bioinformatics, Columbia University, New York, NY 10027 Edited by Peter G,Wolynes, University of California at San Diego, La Jolla, CAI and approved February 18, 2009 (received for review November 25, 2008) The past decade has seen great advances in our understanding of external cues. Cascades play an important role in a diversity of the roleof noise in gene regulation and the physical limitsto signal- cellular processes (23-25), from decision making in development ing in biological networks. Here, we introduce the spectral method (26) to quorum sensing among cells (27) We take a coarse-grained for computation of the joint probability distribution over all species approach, modeling each step of a cascade with a general regula- in a biological network. The spectral method exploits the natural tory function that depends on the copy number of the reactant at eigenfunctions of the master equation of birth-death processes the previous step (Fig. I), Although the method as implemented to solve for the joint distribution of modules within the network, describes arbitrary regulation functions, we optimize the infor- which then inform each other and facilitate calculation of the entire mation tranimission in the We of the most biologically simple joint distribution. We illustrate the method on a ubiquitous case in regulation function: a discontinuous threshold, in which a species nature: linear regulatory cascades. The efficiency of the method is created at a high or low rate depending on the copy count of the makes possible numerical optimization of the input and regulatory species directly upstream. In the next sections, we outline the spec- parameters, revealing design properties of, e.g., the most infor- tral method and present in detail our findings regarding signaling mative cascades. We find, for threshold regulation, that a cascade cascades. of strong regulations converts a unimodal input to a bimodal out- put, that multimodal inputs are no more informative than bimodal Method inputs, and that a chain of up-regulations outperforms a chain We calculate the steady-state joint distribution for L chemical of down-regulations. We anticipate that this numerical approach species in a cascade (Fig. 1).The approach we take involves two may be useful for modeling noise in a variety of small network key observations: the master equation, being linear,* benefits from topologies in biology. solution in terms of its eigenfunctions; and the behavior of a given species should depend only weakly on distant nodes given the gene regulation I noise I signaling I biological network I information theory proximal nodes. The second of these observations can be illdstrated succinctlv by considering a three-gene cascade in which the first gene may ranscriptional regulatory networks are composed of genes and T proteins, which are often present in small numbers in the cell be eliminated by marginalization. For three species obeying s 4 (1, 2), rendering deterministic models poor descriptions of the n % m as in Fig. 1, we have the linear master equation counts of protein molecules observed experimentally (3-9). Prob- abilistic approaches have proven necessary to account fully for the variability of molecule numbers within a homogenous population of cells. A full stochastic description of even a small regulatory network proves quite challenging. Many efforts have been made to refine simulation approaches (10-14), which are mainly based Here, time is rescaled by the second gene's degradation rate, so on the varying step Monte Carlo or "Gillespie" method (15, 16). that each gene's creation rate (g, qs, or qt,) is normalized by its Yet expanding full molecular simulations to larger systems and respective degradation rate; ij and p are the ratios of the first and scanning parameter space is computationally expensive. However, third gene's degradation rate to the second's, respectively. the interaction of many protein and gene types makes analytical To integrate out the first species, we sum over s. We then methods hard to implement. A wide class of approximations to the introduceg,,, the effective regulation of n, by master equation, which describes the evolution of the probabil- ity distribution, focuses on limits of large concentrations or small switches (17-19) Approximations based on timescale separation of the steps of small signaling cascades have been successfully used Here, we have made the Markovian approximation that s is condi- to calculate escape properties (20, 21). Variational techniques tionally independent of m given n. Generally speaking, the prob- have also been used to calculate approximate distributions (22). ability distribution depends on all steps of the cascade. However, In this article we introduce a method for calculating the steady- since there are no loops in the cascades we consider here, we state distributions of chemical reactants (code is available at assume in Eq. 2 that at steady state each species is not affected http://specmark.sourceforge.net). The procedure, which we call by species two or more steps away in the cascade. The validity the spectral method, relies on exploiting the natural basis of a simpler problem from the same class, The full problem is then solved numerically as an expansion in this basis, reducing the mas- Author contributions: A.M.W., A.M., and C.H.W, designed research, performed research, ter equation to a set of linear algebraic equations. We break up the contributed new reagentslanalytic tools, analyzed data, and wrote the paper. problem into two parts: a preprocessing step, which can be solved The authors declare no conflict of Interest. algorithmically; and the parameter-specific step of obtaining the This article is a PNAS Direct Submission. actual probability distributions. The spectral method allows for To whom correspondence should be addressed. E-mail: [email protected]. huge computational gains with respect to simulations. *Although chemical kinetic rates can be nonlinear in the coordinate (e.g., through 9, or We illustrate the spectral method for the case of regulato~y 9" in Eq. I),the master equation itself is a linear equation in the unknown p. cascades: downstream genes responding to concentrations of tran- This article contains supporting information online at www.pnas.orglcgi/content/fulll scription factors produced by upstream genes that are linked to 0811999106/DCSupplemental. www.pnas.org/cgi/doi/ 10.1073/pnas.0811999106 PNAS I April21,2009 1 vol. 106 1 no. 16 1 6529-6534 Evolutionary dynamics in set structured populations Corina E. Tarnitaa, Tibor Antala, Hisashi Ohtsukib, and Martin A. Nowakanl aProgram for Evolutionary Dynamics, Department of Mathematics, Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, MA 02138; and bDepartment of Value and Decision Science, Tokyo Institute of Technology, Tokyo 152-8552, Japan Communicated by Robert May, University of Oxford, Oxford, United Kingdom, April 2, 2009 (received for review February 8, 2009) Evolutionary dynamics are strongly affected by population struc- Let us consider a population of N individuals distributed over ture. The outcome of an evolutionary process in a well-mixed M sets (Fig. 1). Individuals interact with others who belong to the population can be very different from that in a structured popu- same set. If 2 individuals have several sets in common, they lation. We introduce a powerful method to study dynamical interact several times, Interactions lead to payoff from an population structure: evolutionary set theory. The individuals of a evolutionary game. population are distributed over sets. Individuals interact with The payoff of the game is interpreted as fitness (22-26). We others who are in the same set. Any 2 individuals can have several can consider any evolutionary game, but at first we study the sets in common. Some sets can be empty, whereas others have evolution of cooperation, There are 2 strategies: cooperators, C, many members. Interactions occur in terms of an evolutionary and defectors, D. Cooperators pay a cost, c, for the other person game. The payoff of the game is interpreted as fitness. Both the to receive a benefit, b. Defectors pay no cost and provide no strategy and the set memberships change under evolutionary benefit. The resulting payoff matrix represents a simplifie-d- updating. Therefore, the population structure itself is a conse- Prisoner's Dilemma. The crucial parameter is the benefit-to-cost quence of evolutionary dynamics. We construct a general mathe- ratio, blc. In a well-mixed population, where any 2 individuals matical approach for studying any evolutionary game in set struc- interact with equal likelihood, cooperators would be outcom- tured populations. As a particular example, we study the evolution peted by defectors. The key question is whether dynamics on sets of cooperation and derive precise conditions for cooperators to be can induce a population structure that allows evolution of selected over defectors. cooperation. individuals update stochastically in discrete time steps. Payoff cooperation I game I social behavior I stochastic dynamics determines fitness. Successful individuals are more likely to be imitated by others. An imitator picks another individual at uman society is organized into sets. We participate in random, but proportional to payoff, and adopts his strategy and Hactivities or belong to institutions where we meet and set associations. Thus, both the strategy and the set memberships interact with other people. Each person belongs to several sets. are subject to evolutionary updating. Evolutionary set theory is Such sets can be defined, for example, by working for a particular a dynamical graph theory: who interacts with whom changes company, living in a specific location, going to certain restau- during the evolutionary process (Fig. 2). For mat hematical rants, or holding memberships at clubs. There can be sets within convenience we consider evolutionary game dynamics in a sets. For example, the students of the same university have Wright-Fisher process with constant population size (27) A different majors, take different classes, and compete in different frequency-dependent Moran process (28) or a pairwise com- sports. These set memberships determine the structure of human parison process (29), which is more realistic for imitation dy- society: they specify who meets whom, and they define the namics among humans, give very similar results, but some frequency and context of meetings between individuals. aspects of the calculations become more complicated. We take a keen interest in the activities of other people and The inheritance of the set memberships occurs with mutation contemplate whether their success is correlated with belonging rate v: with probability 1 - v, the imitator adopts the parental to particular sets. It is therefore natural to assume that we do not set memberships, but with probability v a random sample of new only imitate the behavior of successful individuals, but also try sets is chosen. Strategies are inherited subject to a mutation rate, to adopt their set memberships. Therefore, the cultural evolu- u. Therefore, we have 2 types of mutation rates: a set mutation tionary dynamics of human society, which are based on imitation rate, v, and a strategy mutation rate, u. In the context of cultural and learning, should include updating of strategic behavior and evolution, our mutation rates can also be seen as "exploration of set memberships. In the same way as successful strategies rates": occasionally, we explore new strategies and new sets. spawn imitators, successful sets attract more members, If we We study the mutation-selection balance of cooperators ver- allow set associations to change, then the structure of the sus defectors in a population of size N distributed over M sets. population itself is not static, but a consequence of evolutionary In the si~pportinginforlnation (SI) Appe~zllix,we show that dynamics. cooperators are more abundant than defectors (for weak selec- There have been many attempts to study the effect of popu- tion and large population size) if blc > (z - h)/(g - h),The term lation structure on evolutionary and ecological dynamics. These z is the average number of sets 2 randomly chosen individuals approaches include spatial models in ecology (1-8), viscous have in common. For g we pick 2 random, distinct, individuals in populations (9), spatial games (10-15), and games on graphs each state; whether they have the same strategy, we add their (16-19). number of sets to the average, otherwise we add 0; g is the We see "evolutionary set theory" as a powerful method to average of this average over the stationary distribution. For study evolutionary dynamics in structured populations in the understanding h we must pick 3 individuals at random: then h is context where the population structure itself is a consequence of the average number of sets the first 2 individuals have in common the evolutionary process. Our primary objective is to provide a model for the cultural evolutionary dynamics of human society, but our framework is applicable to genetic evolution of animal Author contributions: C.E.T., T.A., H.O., and M.A.N. performed research and wrote populations. For animals, sets can denote living at certain the paper. locations or foraging at particular places. Any one individual can The authors declare no conflict of interest. belong to several sets. Offspring might inherit the set member- 'To whom correspondence should be addressed. E-mail: [email protected]. ships of their parents. Our model could also be useful for This article contains supporting information online at www.pn8s.orglcgilcontenVfuill studying dispersal behavior of animals (20, 21). 09030191061DCSupplemental. www.pnas.org/cgi/doi/l 0.1073/pnas.0903019106 PNAS I May 26,2009 I vol. 106 I no. 21 1 8601-8604 REPORTS ized by host cells packed fill1 of fluorescent para- 3. S. Glusliakova, D. Yin, T. Li, 1. Zimmerberg, Ci~rr.Biol. 25. P. 1. Rosenthal, Il~t.1. Parosifol. 34, 1489 (2004). sites unable to egress efficiently, These gl.ossly 15, 1645 (2005). 26. P. Dutt et al., BMC Dev. Biol. 6, 3 (2006). 4. B. L. Salmon, A. Oksman, D. E. Goldberg, Proc. Natl. 27. 1. 5. Arthur, I. S. Elce, C. Hegadorn, K. Williams, swollen cells often detached fi-om tlie n~onolayer Acad, Sci. U.S.A, 98, 271 (2001). P. A. Greer, Mol. Cell. Biol, 20, 4474 (2000). and could be fo~indfloating in the culture ~medi- 5. M. W. Black, G, Arrizabalaga, 1. C. Boothroyd, Mol. Cell, 28. D. 5. Roos, R. G. Donald, N, S. Morrissette, A. L. Moulton, um giving the ap1learaance of a hot air balloon Bial. 20, 9399 (2000). Metl~odsCell Biol. 45, 27 (1994). convention. 6. K. Naganiune et al., Nature 451, 207 (2008). 29, K. A. loiner, D, S, Roos, 1. Cell Biol. 157, 557 (2002). Fibroblast cell lines derived fiom CAPNS 1 7. T. Hadley, M. Aikawa, L. H. Miller, Exp. Parasitol. 55, 306 30. N. Dourdin et al.,]. Biol, Chem. 276, 48382 (2001). (19831, 31. A. Huttenlocher et al., J. Biol. Cheln. 272, 32719 ICO mouse e~nbiyoslack botlth calpain-1 and -2 8. M. E. Wickham, 1. G, Culvenor, A. F. Cowman, I, 6/01, (1997). activity (30), and infection of these cells with 2: Chem, 278, 37658 (2003). 32. B. F. C, Kafsack et al., Science 323, 530 (2009); goriclii produced the same swolle~icell phenotype 9. S, Arastu-l(apur et al., Not. Cl~em.Biol. 4, 203 (2008). published online 18 December 2008 (10.11261 obse~veclin CAPNS 1 knock-down experilllents 10. 5, Yeoh et al., Cell 131, 1072 (2007). science.1165740). 11. D. C. Greenbaum et al., Science 298, 2002 (2002). 33, We thank R, W. Doms, M. Marti, and M. Klemba for (Fig. 3C). Trt~nsgenicexpression of CAPNSl in 12. S. Glushakova, I. Mazar, M, F. Hohmann-Marriott, critical discussions; the Penn Proteolnics Core for niass the I(O mutants restores calpain-1 and -2 activity E. Hania, 1. Zininierberg, Cell, Microbiol. 11, 95 (2009). spectron~etry;and M, A. Lampson for help with imaging, (30) and also coinplernented the I: gondii egsess 13. Materials and methods are available as supporting P. Jalciparum expressing GFP were provided by defect. Parasite tncl~yzoitcswcre readily able to material on Science Online. 0. S. Harb. P,H.D, and D.P,B. are funded by National 14. 1. Russo, A. Oksman, B. Vaupel, D. E. Goldberg, Prac. Research Service Awards, and D.S.R. is an Ellison Medical invade (Fig. 3D) and replicate (Fig, 3E) in WT, Natl. Acad. Sci. U.S.A. 106, 1554 (2009). Foundation Senior Scholar in Global Infectious Disease, CAPNS 1 IWs, and CAPNS 1-co~nplc~ncnted 15. E. M. Pasini et al., Blood 108, 791 (2006). supported by grants from NIH. D.C.G, was supported by fibroblasts. dcinonsbatine" that the bnuact of host 16. D, E. Croall, K. Ersfeld, Gerlome Bial. 8, 218 (2007). tlie Ritter Foundation, the Penn Genome Frontiers cell calpains on 'I: gondii infection is specific to 17. D, E, Goll, V. F, Thompson, H. Li, W. Wei, 1, Cong, Institute, and the Pen11 Institute for Translational- -- - Physiol. 83, 731 (2008, ~edicine-and~herapeutics. !2 cgress, Plaque assays showed a -13-fold reduc- 18. S. C, Murphy et ol., PLoS Med. 3, e528 (2006). tion in plaque size for I: gondii in CAPNSl 19. R. A. Hanna, R. L. Campbell, P. L. Davies, Nature 456, Supporting Online Material mnuitants ve~susparental MEF cells or CAPNS 1- 409 (2008). 20. T. Moldoveanu, K. Geliring, D. R. Green, Natirre 456, 404 www.sciencemag.org/cgilcontent/fulU1171085/DC1 co~n~pleinentedIWs pig. 3, F ruld G). 111 coiitmst to Materials and Methods I! (2008). ,fakcipar+rmz,wl~icli sarely enlel-ge fi0111 calpaili- 21. 5, Gil-Parrado et al., Biol. Cl~an.384, 395 (2003). Figs. 51 to 56 Table S1 depleted e~ytlirocytes(Fig, 2), sorne 7: golzdii 22. A. E. Bianco, F. L. Battye, G. V. Brown, Exp. Parasitol. 62, References parasites did eventually manage to escape from 275 (1986). calpaill-deficient fibroblasts, yielding a small 23. M. Hanspal, V. K. Goel, S. 5. Oh, A. H. Chishti, Mol. 20 January 2009; accepted 10 March 2009 plaque phenotype, Bioche~n.Parasitol. 122, 227 (2002). Published online 2 April 2009; 24. L, Weiss, 1. Iolinson, W. Weidanz, An), 1. Trap. Med. Hyg. 10.1126lscience.1171085 111 sunltllaly, in addition to the many rolcs that 41, 135 (1989). Include this information when citing this paper. parasite-encoded cysteine proteases play in the biology of infection and pathogenesis (25), the apicolllplexans Pln,smo(~i'iumfulc@alwm and Tmo~~li~s~~zcigondii both exploit host cell calpains Human Induced Pluripotent to fi~cilitateescape hmthe inlr.acellular parasito- pl~orousvacuole atldlor host plasma ~nell~brane, The precise lnechailism of calpain-inediated pa1.a- Stem Cells Free of Vector site egress is unknown, but calpains play a role in renlodeling of the cytoskeleton and plasma nlem- and Transgene Sequences brane dul-hig the nligrntioti of manmlalian cells (31), and activated calpain-1 can degsade elytll- Junying YU~~~~~~*Kejin HU,~Kim ~rnuga-~tto,~~~~~ Shulan ~ian,'~~ Ron stewart,'l2 rocyle cytoskelctal proteins in vitro nlid during lgor I. ~lukvin,'~~James A. ~hornsonl~~~~~~* I? ,firlc~~aruniinfection in vivo (fig. S4), The calciurm i+esponsiblefor calpain activalion dul-iiig Reprogramming differentiated human cells to induced pluripotent stem (ips) cells has applications parasite iid?ection nlay be sulpplied tlxough the in basic biology, drug development, and transplantation. Human iPS cell derivation previously action of a parasite-encoded pe~folil~recently required vectors that integrate into the genome, which can create mutations and limit the in~plicatedin I: golzdii cgsess (32). Tlic parasito- utility of the cells in both research and clinical applications, We describe the derivation of pho~.ous vacuole was labeled by the calcium- human iPS cells with the use of nonintegrating episomal vectors. After removal of the episome, iPS specific dye Fluo-4-AM duiing late schizogony, cells completely free of vector and transgene sequences are derived that are similar to human and depletion of internal calcium with the embryonic stem (ES) cells in proliferative and developmental potential. These results demonstrate rne~~~bixt~e-pe~ilieantchelator EGTA-AM blocked that reprogramming human somatic cells does not require genomic integration or the parasile egrcss, wher~~srcliloval of calcium continued presence of exogenous reprogramming factors and removes one obstacle to the h~nthe culture medium did not (fig. S5). We clinical application of human iPS cells. suggest a inodel in whicli a calciulli signal triggered late during parasite infection activates 11e proliferative and develol>mental po- by expressing co~l~binationsof factors such as host cell calpain, which relocalizes to the host tential of both human ciub~yonicstell1 OCT4,SOX2, c-Mjc, I(LF4, NANOG, and LlN28 plasma ~neinb~ailc,cleaving cytoskclctal pro- T(ES) cells ancl human induced pluripotent (2-4, Initial methods used to desive I~umimiPS teins to facilitate parasite egress (fig. SG). stell1 (iPS) cells offers unprecedented access to cclls used viral vectors, in wliich both tlie vector Because parasites that fail to escape from their thc differe~itiatedcells tl~atmake up tlie 11~1man backboilc and transgenes are permanently in- host cells are unable to proliferate, this sug- body (1-3). 111 addition, iPS cells call be deiived tegrated into the genome (2, 3). Such vectoi-s gcsts an intriguing slralegy for anti-parasitic with R s1)eciGc desired genetic background, in- call prochrce insc~tionalmutatio~is that intel-ferc therapeutics. clnding patient-specific iPS cells for disease with the nornial fiinction of iPS cell derivatives, models and for tr.ansplalltaliou tl~erapies,witli- and residual tmnsgene expression can influence References and Notes M, ,,,ishi, K, M, Murray, D, 5, Cell Sci, 121, out the problellis i~ssociatedwith irnii~u~oerejec- differentiatiou into sl?ecific lilleages (2), or even 1559 (2008). tion. Reprogain~~zil~gof both nzouse and h~ullan result in tumorigenesis (5). Vector ii~tegratioll- 2. K. HU et al., MO~.Biol, tell 13, 593 (2002). somatic cells illto iPS cells has bee11 achieved fi-ee mouse iPS cells have been derived fiom www,sciencemag.org SCIENCE VOL 324 8 MAY 2009 REPORTS difference observed between these values may conh-ast is still a subject of debate (32). Our swvey 19. 1. A. Luk'yanchuk, Y, Kopelevich, Pl~ys.Rev, Lett. 93, imply a breal www,sciencemag.org SCIENCE VOL 324 15 MAY 2009 927 REPORTS editing levels, The eillarged set of aoiue~ctitive 13, w. M. Golnlnans et al., RNA 14, 2074 (2008). 30. We thank R. Emeson, M. Po Ball, and F, lsaacs for critical RNA>ditillg targets help unravel l;lles of 14. D. R. ChWrbuck, A. Leroy, M. A. O'Connell, C, A. Selnple, reading of the manuscript; P. Wang and 2. Liu (BioChain Bioiriforrnatics 21, 2590 (2005). Institute) for helping collect humari samples; M. Higuchi RNA editing ill human diseases and behavior. 15, Ohlson, Pedersen, D. M. Ohman, and P. Seeburg for providing ADARZ-'- mouse brain RNA 13, 698 (2007). cDNA; Harvard Biopolymers Facility for help with lllumina References and Notes 16. G. 1. Porreca et ol., Nat. Methods 4, 931 (2007). sequencing; and A. Ahlford, H. Ebling, and 1. B. L. Bass, Annu, Rev. Bioclieni. 71, 817 (2002). 17. A. G. Polson, B. L. Bass, EMBO 1. 13, 5701 (1994). 1. Santosuosso for assistailce with Sanger sequencing, 2. K. Nishikura, Not, Rev. Mol. Cell Biol. 7, 919 (2006). 18. K. Nishikura et al., EMBO]. 10, 3523 (1991). E.Y.1, was supported by the Machiah foundation. Funding 3. M. Higuclii et al., Noture 406, 78 (2000). 19. K. A. Lehmann, B. L. Bass, Biochen~istly39, 12875 (2000). came from National Human Genome Research Institute 4. R. Brusa et a!., Science 270, 1677 (1995). 20. A, Athanasiadis, A. Rich, S, Maas, PLoS Biol. 2, e391(2004). Centers of Excelle~icein Genolnic Science grant to G.M.C. 5. M. Singh et ol., J, Biol, Chem. 282, 22448 (2007). 21, M. Blow, P. A. Futreal, R. Wooster, M. R. Stratton, The lllumina sequencing data are deposited at the 6. M, 1. Palladino, L, P. Keegan, M. A. O'Connell, R. A. Reenan, Genorne Res. 14, 2379 (2004). National Center for Biotechnology Information Short Cell 102, 437 (2000). 22. E. Eisenberg et al., Trends Genet. 21, 77 (2005). Read Archive under accessioll number SRA008181. 7. H. Lomeli et al., Science 266, 1709 (1994), 23. D, D. Kiln et ol., Genome Res. 14, 1719 (2004). Supporting Online Material 8. 5. Maas, Y, Kawahara, K. M. Tamburro, K. Nishikura, 24. E. Y. Levanon et al., Nat, Biotechnol. 22, 1001 (2004). www.sciencemag.org/cgi/co1itent/fuIV32415931/1210/DC1 RNA Biol. 3, 1 (2006). 25. N. Tian, X. Wu, Y. Zhang, Y. ]in, RNA 14, 211 (2008). Materials and Methods 9. 1. Shendure, H. Ji,Nat. Biotechnol. 26, 1135 (2008). 26. C. M. Burns et ol., Nature 387, 303 (1997). Figs. S1 to 510 10. Materials and methods are available as supporting 27. M. Kohler, N. Burnasliev, 8, Saktnann, P. H. Seeburg, Tables S1 to 512 material on Scierlce Online. Neuron 10, 491 (1993). References 11. E. Y. Levanon et al., Nirclek Acids Res, 33, 1162 (2005). 28. B. Sonimer, M. Kohler, R. Sprengel, P. H. Seeburg, 12. B. Hoopengardner, T. Bhalla, C. Staber, R, Reenan, CeN 67, 11 (1990 15 Ianuary 2009; accepted 1April 2009 Scb~ce301, 832 (20031, 29. T. R. Mercer et al., Netrroscientist 14, 434 (2008). 10.11261science,1170995 ol*thologous TR loci ill psoinoters differ in the Unstable Tandem Repeats IIIUII~CSof repelt tulits bctweeil tlie two fidly sequenced strains, S288C and RMl 1 (8)).To in Promoters Confer confi~~nthis, we seclue~lced33 randomly chosen psomoter repeats in seven S. cerevisiae genomes (Fig, JAYfigs. S2 and S3, and data set S3). Transcriptional Evolva bility Twenty-five of the 33 TRs differed in lleyeat units in at least one of the seven sbains. The repeat Marcelo D. ~inces,~'~~~*Matthieu ~egendre,''~* Marina ~aldara,' variation frequency is 40-folcl higher than the Masaki ~agihara,~Kevin 1. ~erstrepenl~~~~t fiequellcy of insel-tionsand deletio~~s(indels) and of point mutations in the sui-sounding non- Relative to most regions of the genome, tandemly repeated DNA sequences display a repetitive seqttencc (P < 10-15) (figs. S2 aid 53). greater propensity to mutate, A search for tandem repeats in the Saccharomyces cerevisiae genome ' To determine wliether psolnoter TR variation revealed that the nucleosome-free region directly upstream of genes (the promoter region) is affects gene exl~ression,we co~lqxiredrepeat var- enriched in repeats. As many as 25% of all gene promoters contain tandem repeat sequences. iablity to expression divergence (ED), which Genes driven by these repeat-containing promoters show significantly higher rates of represents how fast the transcriptional activity of transcriptional divergence. Variations in repeat length result in changes in expression and local each gem evolvcs (9-11). Promoters containing nucleosome positioning, Tandem repeats are variable elements in promoters that may facilitate TRs showed significantly (P 1.75 x l 04) evolutionary tuning of gene expression by affecting local chromatin structure. higher ainounts of ED than did promoters lacking TRs wllen comparing yeast species (S. ce~evisiae, he gcnolnes of inosl organisms are not point mutations (2). u~riablc Tlis arc onen dis- S. ~.)at~~do.~lrs,S, ~nilcntae, and S. kc1 ~~Jrir~vzevii) uniformly proilc to change because they inisscd as iloniiulctioi~al"ju11k" DNA, Howcvel; (Fig. 1, B to D, and fig, S4A) and ,S1 cc www.sciencemag.org SCIENCE VOL 324 29 MA REPORT'S clements silence and activate posttra~isciiptio~~al Synthetic Gene Networks That Count g~leexp~ssion,respective~y,T~ecis-repsessos sequence [cr in Fig, I] is placed between the Ari E. ~riedland," Timothy K. I.U,'~~*Xiao ~ang,' David hi,' tsanscril~tionstart site and the iibosoil~e-binding George James 1. collinslt site (RBS), slnd its complementarity with the RBS causes a stem-loop structure to form upon Synthetic gene networks can be constructed to emulate digital circuits and devices, giving one the transcription, This secondaly structure prevents ability to program and design cells with some of the principles of modern computing, such as binding of the 30s ribosomal s~b~mitto the RBS, counting, A cellular counter would enable complex synthetic programming and a variety of wllicll inhibits translation, A ~11012,transactivat- biotechnology applications, Here, we report two complementary synthetic genetic counters in ing, noncoding RNA (taRNA), driven by the a~b- Escherichia coli that can count up to three induction events: the first, a riboregulated iuose ptunloter PnAD,binds to the cis repressor in transcriptional cascade, and the second, a recombinase-based cascade of tnemory units. These tsans, which selicves RBS repression and allows modular devices permit counting of varied user-defined inputs over a range of frequencies and can tra~lslatioil.With tlis rib~~egulation,each node be expanded to count higher numbers, (is,, gene) in the cascade requires both iiide- pendent kai~scriptionand translation for protein counter- is a ley coinponent in digital In this study, we fist developed a co~u~ter, expression. This cascade is able to count brief circuits and conlputing that retains Inem- termed the iiboregulated tianscriptioaal cascade arabinosc p~~lses[for pulse definition, see (5)] by Aosy of events or ob-jects, rcpresenting (RTC) counter, which is bascd on a tra~lscriplional expressi~lga different proteiu in response to each each ~ignlberof such as a clistillct state.(=oy.n@rs_ casq4ds witll addi tignal _translational_regulation. pulse (Fig. -1 A). _Withcis-repressed T7- RNAP would also be usehl in cells, wliich often must Two such cascades are illustrated in Fig. 1, A w~d - have accurate accounting of tightly controlled C that can count LI~to two and three, respectively '~owardHughes Medical Institute, Departnielit of ~ioniedical processes or bionlolecules to effectively nlailltain (hence, the designatioas RTC two-counter and Engineering, Center for BioDynaniics and Center for Advanced inctabolislll and gsowth. Couilting mecha~isms RTC tlwee-couilter). For tllc RTC two-counter, the Biotechnology, Boston University, Boston, MA 02215, USA. have beell repostedly foulid in telomere length constitutive promoter PL,lelO-Idrives transcription '~arvard-MIT Division of Health Sciences and Technology, regillation (I, 2) and cell aggregation (3), These of T7 RNA polycnerase QW),whose proteiil 77 Massachusetts Avenue, Room E25-519, Cambridge, MA 02139, USA, 3~epartnientof Genetics, Harvard Medical syste~ilbellaviors appear to be the result of a binds the T7 promoter and tsi1nsc1ibes the down- School, Boston, MA 02115, USA. Il~resholdeffect in ~vliichsome critical lnoleculc strea~ngene, in this case, green fluoscscenl *These authors contributed equally to this work. iluinber or dcilsity nust be reached for the ob- protein (GFP). Both genes are additio~lallyregu- ?To whom correspondence should be addressed. E-mail: served phenotypic cha~lge, lated by siboregulators (4), whose cis and balls jcollins@ bu.edu Fig. 1, The RTC two-counter and A C RTC three-counter construct designs and results. (A) The RTC two-counter i ?e*i )Dm, , pm )rb%igg is a transcriptional cascade with two nodes. Shown at the bottom are AmA AilAm A expected expression profiles after PLM&I)=~ Pi7 )mm pkb,&", 1~- ( P* k@m::,Fn bm zero, one, and two arabinose (Ara) pulses, (B) Mean fluorescence of GFP three replicates of RTC two-counter Prqteln @ cell populations over time, measured GFP khn.f ~AAJ by a flow cytometer, Shaded areas mRNA GFP t f represent arabinose pulse duration. Praltrln 448 T3HNAP (C) The RTC three-counter is a tran- Protein @Pv @ scriptional cascade with three nodes, GFP Shown at the bottom are expected mRNA expression profiles after zero, one, two, and three arabinose pulses. (D) Mean fluorescence of three repli- Ti'RNAP @Yf cates of RTC three-counter cell pop- mRNA bwv (stln, ulations over time, measured by a flow cytometer. Shaded areas repre- sent arabinose pulse duration. 0; 0; ' 1.i J 50 100 150 200 250 300 O 0 50 100 150 200 250 300 350 Time (min) Time (min) www.sciencemag.org SCIENCE VOL 324 29 MAY 2009 1199 REPORTS and the size and shape of the tRNA stems, References and Notes ' 17. See supportilig material on Science Online. Finally, a positively charged gsoove extends fiom 1. B. Teng, C. F. Burant, N. 0. Davidson, Scierice 260, 1816 18. C. Prochnow, R. Bransteitter, M. G. Klein, M. F, Goodman, the THUMP domain to thc active sitc of the cn- (19931, X. 5, Chen, Nature 445, 447 (2007). 2. A. P. Gerber, W. Keller, Science 286, 1146 (1999). 19. K. M. Chen et ol., Natirre 452, 116 (20081, zyme. The width of the groove fornled between 3. 5. Maas, A. P. Gerber, A. Rich, Proc. Natl. Acod. Sci, U.S.A. 20. L. G. Holden et al., Nature 456, 121 (2008). the two monomers could provide a snug fit for 96, 8895 (1997). 21. C. 1. McCleverty, M. Hornsby, G. Spraggon, A. Kreusch, the acceptor stein; the length of the groove en- 4. 1. Wolf, A. P. Gerber, W. Keller, EMBO 1. 21, 3841 1. Mol. Biol. 373, 1243 (2007). sures that binding of the 3'-CCA at the THUMP (2002). 22. C. T. Lauhon, W. M. Erwin, G. N. Ton,). Biol. Cheni. 279, 5. H, Grosjean, F. Constantinesco, D. Foiret, N. Benachenhou, 23022 (2004). domains places U8 adjacent to the active site of Nucleic Acids Res, 23, 4312 (1795). 23. R. Marschalek, T. Brechner, E. Amon-Bohm, T. Dingermann, the CDD. 6. S. Steinhauser, S. Beckert, I, Capesius, 0, Malek, V. Knoop, Science 244, 1473 (1989). It is possible that C8 is the result of genetic I, Mol. Evol, 48, 303 (1999). 24. Q. She, K. Brugger, 1. Chen, Res. Microbial. 153, 325 drift. Alternatively, C8 may be belleficial for 7. G. V. Borner, M. Mdrl, A. lanke, 5. Paabo, EMBO J. 15, (20021, 5949 (19961, 25. We thank I, Yuan, P. O'Doiioghue, and R. L. Slierrer for M /c(mdleri at the level of the tRNA genc, thc 8. M. A. Rubio et al., Proc. Natl. Acad, Sci. U.S.A. 104, help and encouragement; K. 0. Stetter for a gift of p~ima~ytRNA t~a~~scripts,or thc mahisation steps 7821 (2007). M, kondleri cells; and tlie staff at the Advanced Plloton that involve folding and lnodification of the tRNA. 7. 1, R. Palmer, T. Baltrus, 1, N, Reeve, C. 1, Daniels, Source (beamline 24-ID), the National Synchrotron Light Because M /cunc/lecr.i groivs at extrcille te~npcr- Biocliitn. Biopl?ys. Acta 1132, 315 (19921, Source (beamlines X25 and X6A1, al~dtlie Center for 10. A. I. Slesarev el al., Proc. Nfltl. Acfld, Sci. U.S.A. 99, 4644 Structural Biology at Yale University. Atomic coorditiates atuses (up to l 10°C), C8 may stabilize the ge- (2002), and structure factors have been deposited in the Protein noine at thc tRNA gene or aid in tertiary iblcling 11. E. Westliof, P. Dumas, D. Moras,). Mol. Biol. 184, 119 Data Bank (code 3G8Q). Supported by NIH grants and modification events of RNA inolccules at (1785), GM22854 (D.S.) atid A1078831 (Y.X.). sucl~extreli~e conditions. Perhaps the 4-thiolation 12,C, N. Jones) C. I- lonest Ws D. Pe F- Agris, Lr 1. Spreniulli, J. Biol. Cl~eni.283, 34445 (2008). - Supporting Online Material - of U8 to s4u8 nlcdiated by lni(lhlbe 13. T. Slerlier, M. lansen, Y. M. Hou, RNA 1. 841 (1995). www.science1nag.org/cgi/content/fulV324/5927/657/DCl regulated by CDAT8, as C8 would 1101 be a ThiI 14. A, A, Smith, D. C, Carlow, R. Wolfenden, 5. A. Short, Materials and Methods si~bstrtitc,A C8 n~ayalso be beneficial at thc levcl Biochemistry 33, 6468 (1994). SOM Text Figs. 51 to 54 of DNA, as a T8C illutation lnigllt prevent the 15. L. Aravihd, E. V. Koonin, Trends Biochern, Sci. 26, 215 (2001). Table 51 i~lteractionof tRNA genes with lnoMle genetic References 16, D. G, Waterman, M, O~iz-Lombardia, M, Fogg, elelllents [e-g., (23, 241 at this otl~elwisefixed E. V. Koonin, A. A. Antson, J. Mol. Biol, 356, 97 22 December 2008; accepted 23 February 2007 and convenient target. (2006). 10.1126kcience.1170123 cliwmosome (I O), but has not yet been applied to A Yeast Hybrid Provides Insight a whole genome. We designed a microarray that enables the into the Evolution of Gene measuring of allclespccific expressioil in a hybrid of Scrcck~~otn~)cescerevisicle and S. pa1~cldoxu,s, two yeast species that diverged -5 mnillioi~years Expression Regulation ago (fig. S I). Hybi*idization of geno~nicDNA fso~uthe two parental species ve~ifiedthe spec- ltay ~irosh,' Sharon ~eikhav,',~Avraham A. LevytZ* Naama ark ail* ificity of the inicroasray (Fig. 1A) and confinled the lack of mdjor variations ill copy il~~~nber.Bio- During evolution, novel phenotypes emerge through changes in gene expression, but the genetic logical repeats were 14gl$y coi~clated[Pearson basis is poorly understood. We compared the allele-specific expression of two yeast species and their co~.sclation(r) -- 0,981, significa~~tlyulore so tllan hybrid, which allowed us to distinguish changes in regulatory sequences of the gene itself (cis) from the con-elation in expression between species (r - changes in upstream regulatory factors (trans), Expression divergence between species was generally 0.85) or betwecn the cossesponding alleles within due to changes in cis. Divergence in trans reflected a differential response to the environment and the hybrid (I.- 0.9) (Fig. lB), explained the tendency of certain genes to diverge rapidly. Hybrid-specific expression, deviating Using the array, we measured the expression from the parental range, occurred through novel cis-trans interactions or, more often, through profiles of the two parei~talspecics and tllcir hy- modified trans regulation associated with environmental sensing. These results provide insights on brid under four diRerent grovirtl~conditions aiid the regulatory changes in cis and trans during the divergence of species and upon hybridization. quantified the relative contrib~~tionof cis and trans effects to the ii~terspeciesdivergence (Fig, IC), As ver the past years, extensive variations ia gence of gene expression is an essential step ill expected, cis (but not t~tns)cKccls were correlated gene expression were ideii ti fied between elucidating its genetic basis, with secluetlce divergence at proomoters and reg- closely related species (1,2). The genetic When coi~lj~ai.ingdifferent strains of tlze same ulatoiy elelnents (fig. S3), whei.eas trans (but not basis of 'inost differences, howevcr, remait~sun- species, cis and trans effects can be approxiillated cis) eflects were enriched with genes whose ex- I / www.sciencemag,org~ SCIENCE VOL 324 1 MAY 2009