Review Article Advances and Computational Tools towards Predictable Design …downloads.hindawi.com/journals/cmmm/2014/369681.pdf · 2019. 7. 31. · toggle switch [ ]andanoscillator(therepressilator

Review ArticleAdvances and Computational Tools towards Predictable Designin Biological Engineering

Lorenzo Pasotti12 and Susanna Zucca12

1 Laboratory of Bioinformatics Mathematical Modelling and Synthetic Biology Department of ElectricalComputer and Biomedical Engineering University of Pavia 27100 Pavia Italy

2 Centre for Tissue Engineering University of Pavia 27100 Pavia Italy

Correspondence should be addressed to Lorenzo Pasotti lorenzopasottiunipvit

Received 3 April 2014 Accepted 9 June 2014 Published 3 August 2014

Academic Editor Huiru Zheng

Copyright copy 2014 L Pasotti and S Zucca This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited

The design process of complex systems in all the fields of engineering requires a set of quantitatively characterized components and amethod to predict the output of systems composed by such elementsThis strategy relies on themodularity of the used componentsor the prediction of their context-dependent behaviour when parts functioning depends on the specific context Mathematicalmodels usually support the whole process by guiding the selection of parts and by predicting the output of interconnected systemsSuch bottom-up design process cannot be trivially adopted for biological systems engineering since parts function is hard to predictwhen components are reused in different contexts This issue and the intrinsic complexity of living systems limit the capabilityof synthetic biologists to predict the quantitative behaviour of biological systems The high potential of synthetic biology stronglydepends on the capability of mastering this issueThis review discusses the predictability issues of basic biological parts (promotersribosome binding sites coding sequences transcriptional terminators and plasmids) when used to engineer simple and complexgene expression systems in Escherichia coli A comparison between bottom-up and trial-and-error approaches is performed for allthe discussed elements and mathematical models supporting the prediction of parts behaviour are illustrated

1 Background

In order to handle complexity in the design of customizedsystems engineers usually rely on a bottom-up approachcomponents are quantitatively characterized and the outputof an interconnected system is predicted from the knowledgeof individual parts function [1] This process is applied in allthe fields of engineering and is useful to hide the complexityof the individual components functioning thus using themas input-output modules [2]

This strategy is successful only in a modular frameworkwhere parts behaviour does not change upon intercon-nections and in general when the same parts are reusedin a different context [3 4] Even if this property doesnot persist the bottom-up approach is still feasible whenengineers are able to predict how parts behaviour varies asa function of environmental changes or interconnections [5]In electronics examples of the latter situation are resistors

they are characterized by an electrical resistance which doesnot change upon connection in different circuits Howeverit is well established that resistance changes as a functionof temperature and for this reason datasheets of electriccomponents report the temperature-resistance characteristicin order to make the output of complex circuits predictablewhen used in different environments Another example is acircuit with a nonzero impedance it can exhibit a differentinput-output behaviour when interconnected to differentloads However it is still possible to predict the output ofsuch interconnected systems since mathematical models ofelectrical circuits are able to describe voltage and currentthroughout the network

Mathematical models are widely used in many areasof engineering to support the early design steps of a sys-tem guide the debugging process measure nonobservableparameters and finally predict the quantitative behaviourof systems composed by precharacterized parts Likewise

Hindawi Publishing CorporationComputational and Mathematical Methods in MedicineVolume 2014 Article ID 369681 16 pageshttpdxdoiorg1011552014369681

2 Computational and Mathematical Methods in Medicine

models also play an important role in a biological systemsframework in fact they are often used to study com-plex metabolic interactions like those occurring in diseaseconditions to understand the underlying processes andorpredict the effect of drugs [25] Some mathematical modelsof biologicalphysiological systems have also been approvedby the US Food and Drug Administration (FDA) for usein simulated clinical trials thus enabling researchers forexample to support or even skip expensive in vivo trials [26]

Synthetic biology aims to realize novel complex biologicalfunctions with the same principles on which engineeringdisciplines lay their foundationsmodularity abstraction andpredictability [2 27 28] As a result synthetic biologists sofar have mainly focused on the definition of biological partsand on their abstraction and standardization in order todeal with well-defined components with specific function[29] This process has brought to the creation of biologicalparts repositories including DNA parts that can be sharedby the scientific community like the MIT Registry of Stan-dard Biological Parts [30ndash32] to standardized and easy-to-automate DNA assembly strategies [33ndash35] and to standardmeasurementmethodologies to share characterization resultsof parts like promoters [36 37] Researchers have alsofocused on the realization of engineering-inspired functionsto learn the complexity that could be reached in a biologicalcontext Towards this goal researchers built up devices thatimplement logic gates and functions [19 38ndash41] memories[42] oscillators [43ndash45] other waveform generators [46 47]signal processing devices [48ndash50] and the likeMany of themrelied on mathematical models to support the early designsteps and to capture the behaviour of the designed circuit Forexample two of the synthetic biologymilestones are a genetictoggle switch [42] and an oscillator (the repressilator) [43]both implemented in Escherichia coli via genetic networks ofproperly connected transcriptional regulators A semiquan-titative investigation of the features required for a correctcircuit behaviour was performed via mathematical modelsby using dimensionless equations or reasonable parametervalues Thanks to the model analysis the authors could learnuseful guidelines for correct design of circuits exhibiting thedesired functioning for example fast degradation rates ofrepressor proteins encoded in the oscillatory network [43]

The realization of complex functions has brought tosome biological systems of high impact An engineeredpathway was implemented in recombinant yeast to producethe antimalarial drug precursor artemisinin [51] a biosensor-encoding genetic device was implemented in microbes todetect arsenic in drinking water and to provide a colourchange of its growth medium as visual output [52 53]microbes were recently engineered to produce bioethanolfrom algal biomass [54] or advanced fuels from differentsubstrates [55]

However despite many examples of complex engi-neering-inspired function implementation and also of indus-trially relevant solutions to global health environmental andenergy problems a rigorous bottom-up design process is notcurrently adopted because the predictability boundaries stillhave to be clearly defined [3 56 57] The high potentialof synthetic biology strongly depends on the achievement

of such task [58] Trial-and-error approaches represent analternative if synthetic biologists cannot design a systemfrom the bottom-up they can rely on random approacheswhere for example circuit components are mutated andthe best candidate implementing the function of interest isselected [38 59 60] Depending on the reliability of pre-dictions and of mathematical models this process could becompletely random or partially guided In general trial-and-error approaches are time- and resource-consuming and arecharacterized by a low efficiency However recent advancesin the construction of biological systems for example DNAandor strain production via automated proceduresmay pro-vide a good alternative to the rational bottom-up approachespecially when accurate automated and possibly low-costscreeningmethods are available to rapidly evaluate the outputof the constructed circuits [60]

This review discusses the predictability issues of basicbiological parts (promoters ribosome binding sitesmdashRBSscoding sequences transcriptional terminators and plasmids)when used to design the desired biological function in theform of a simple or complex gene expression system Eventhough synthetic biological systems may be implementedin several organisms (or even in vitro [61]) and may havedisparate architectures and regulatory mechanisms [62 63]the reviewwill focus on predictability of parts in vivo in the Ecoli bacterium according to the biological information flowdescribed in the central dogma of molecular biology [64]protein-coding DNA sequences (herein called genes) aretranscribed into mRNA molecules which are converted intoproteins by ribosomes and finally DNA sequences can bereplicated in living cells to propagate the encoded function tothe progenyThus in the considered framework the possiblebasic architectures are shown in Box 1 promoters can triggerthe expression of a single gene (monocistronic architecture)or a set of genes (polycistronic or operon architecture) eachgene is transcribed with its RBS upstream and finally termi-nators stop transcription Ribosomes complete the process bytranslation of mRNA molecules into the proteins of interestfrom the start codon (generally AUG) to the stop codon(generally UAA) Complex genetic circuits can be realizedwith a set of such gene expression units implementing theinteractions of interest and giving the desired product asoutput Genetic circuits can be placed on a plasmid vector orotherwise they can be integrated into a target position of thebacterial chromosome

Even if other classes of parts can be used to constructcomplex genetic systems and other elements can also affectcircuit behaviour we will focus only on the abovemen-tioned genetic parts and architectures given a specificstrain and environment Other important contexts like thehost (the reciprocal variation of parts behaviour and hostmetabolism when a circuit is incorporated) environmental(the reciprocal variation of parts behaviour and environ-mental parameters) ecological (changes of synthetic circuitand surrounding community parameters as well as strainsfitness) and evolutionary (changes of DNA composition)contexts are reviewed elsewhere [57] Other reviews are

Computational and Mathematical Methods in Medicine 3

Genomic DNA

Replication origin

(c)

(a) (b)

(d)The simplest expression system(panel A) includes a promoter (curved arrow)a ribosome binding site (RBS oval) a gene (straight arrow) and a transcriptional terminator(hexagon) A more complex architecture has an operon structure (panel B) where multiple geneswith their own RBSs are expressed by a single promoter a two-gene operon is shownIndividual or composite gene expression systems can be placed in a plasmid vector (panel C)which is maintained by the host strain at a specific copy number per cell depending on theparticular replication origin or in the bacterial chromosome (panel D) Dashed line indicates thegenomic DNA of the hostPromoters control the transcription of the downstream DNA Their sequence determines theinformation on transcriptional strength and regulation by other factors Different promoterclasses are present in E coli according to the specific sigma factor that determines theirregulation Other factors such as activatorrepressor proteins which are widely used in syntheticcircuits can provide an additional degree of controlRBSs are small sequences that are placed upstream of a gene They are transcribed into mRNAbut are not translated since protein synthesis starts with the AUG codon of the coding sequenceRBS sequence determines the binding affinity between ribosome and mRNA thus affecting thetranslation initiation rate In particular mRNA binding to ribosomes occurs at theShine-Dalgarno region of an RBS with the rRNA of the 30S ribosomal subunit of E coliTranscriptional terminators of the rho-independent class exploit hairpin structures to stop RNApolymerase activity during transcription This class of terminators is commonly used insynthetic circuits as opposed to rho-dependent terminators that mediate transcription stop viaan ATP-dependent process which has not been completely understoodGenes encode the desired proteins which can actuate disparate specific regulatory ormetabolic functions Gene sequence can affect the mRNA folding which may result in differenttranslation initiation rate and different decay rate of the transcript Moreover codon usage biasdetermines preferences of specific codons over others for a given organism For these reasons twoidentical proteins encoded by two different genes (designed exploiting genetic code redundancy)can have extremely diverse synthesis rates

Box 1 Genetic parts and architecture of a gene expression system

complementary to the presentwork describing software toolsfor partspathway identification [65] and cellular behaviourmodelling at different scales [65ndash67]

Each of the biological parts and architectures describedin Box 1 will be considered We will discuss to which extenttheir function can be predictable and then a comparisonbetween bottom-up and trial-and-error approaches will becarried out For each part and architecture the contributionof mathematical models supporting the prediction of circuitbehaviour will be highlighted Even though many computer-aided design (CAD) tools are available for synthetic circuits[68] only mathematical analysis tools (also including toolsfrom the field of systems biology) and predictive models ofparts function will be considered while no software toolfor database accessdevelopment or for the assembly processsupport [65] will be taken into account In particular theconsidered tools can be ordinary differential equation (ODE)

models (or derived steady-state equation models) based onempirical or mechanistic functions or predictivemodels ableto infer parts behaviour given their sequence andor theirDNA context

2 Research Studies and Tools to SupportBottom-Up Design

The kit of parts architectures and contexts available tosynthetic biologists will be discussed Then interconnectionissues will be considered A summary of the selectedmethodsand tools available for partsdevices quantitative prediction isreported in Table 1

21 Promoters Promoters are intrinsically context-depend-ent parts since it is known that their upstream and down-stream elements may affect transcriptional activity [69ndash73] The research studies on the predictability of promoters


Table 1 Selected computational methods and tools that support the bottom-up design in biological engineering

Part architecture orcontext Description Reference

Promoters

Strength prediction tool for sigmaE promoters using aposition weight matrix-based core promoter model andthe length and frequency of A- and T-tracts of UPelements

[6]

Strength prediction tool for sigma70 promoters usingpartial least squares regression

[7]

Promoter-RBS pairs Strength prediction tool for sigma70 promoter-RBSpairs using an artificial neural network [8]

RBSs

RBS Calculator a web-based tool for RBS strengthprediction and forward engineering frequentlyupdated and able to design RBS libraries

[9]

RBS Designer a stand-alone tool for RBS strengthprediction and forward engineering it considerslong-range interactions within RNA and it can predictthe translation efficiency of mRNAs that maypotentially fold into more than one structure

[10]

UTR Designer a web-based tool for RBS strengthprediction and forward engineering able to design RBSlibraries and with the codon editing option to changeRNA secondary structures

[11]

Genes

GeMS web-based tool for gene design using a codonoptimization strategy based on codon randomizationvia frequency tables

[12]

Optimizer web-based tool for gene design using threepossible codon optimization strategies ldquoone aminoacid-one codonrdquo randomization (called ldquoguidedrandomrdquo) and a hybrid method (called ldquocustomized oneamino acid-one codonrdquo)

[13]

Synthetic Gene Designer web-based tool for genedesign with expanded range of codon optimizationmethods full (ldquoone amino acid-one codonrdquo) selective(rare codon replacement) and probabilistic(randomization-based) optimization

[14]

Gene Designer stand-alone tool for gene design using acodon randomization method based on frequencytables and with the possibility to filter out secondarystructures and Shine-Dalgarno internal motifs

[15]

Terminators

Termination efficiency prediction tool based on a linearregression model using a set of sequence-specificfeatures identified via stepwise regression

[16]

Termination efficiency prediction tool based on abiophysical model using a set of free energiespreviously identified as important features

[17]

Interconnected networks

A range of empirical or mechanistic ODE orsteady-state models can be used to predict complexsystems behaviour from the knowledge of individualpartsdevices parameters

[5 18ndash21]

ArchitectureProtein expression prediction for the first gene of anoperon given the downstream mRNA length via alinear regression model

[22]

Context

Mechanistic ODE models where the DNA copy numberis explicitly represented [23]

Protein expression prediction tool based on linearregression model given the chromosomal position ofthe gene and its orientation

[24]


have focused on their context-dependent variability and onactivity prediction given their nucleotide sequence Context-dependent variability studies aim to evaluate whether pro-moters show the same activity in different contexts forexample when promoters have different sequences upstreamwhen expressing different genesmRNAs or when otherindependent gene expression cassettes are present in thesame circuit Generally the activity of a set of promoterscan be indirectly measured via reporter proteins providedthat the downstream sequences are the same (ie identicalRBSs reporter genes terminators and similar transcriptionstart sitesmdashTSSs) so that mRNA primary and secondarystructures do not significantly vary among the promotermeasurement systems [37] Using the same architecture theactivity can be evaluated via qPCR by directly measuringthe mRNA level [74] Davis et al [73] quantified a set ofconstitutive promoters and found that activity was affectedup to 4-foldwhen a specific upstream (UP) sequence is placedbefore promoters even though in some cases activity was notaffected Other studies showed that the upstream sequence-dependent activity change could be as high as 300-foldand the consensus sequences that can affect such differenttranscriptional activity were identified [72 75 76] this effectwas observed when using the rrnB P1 promoter but activitychange was also observed for the lac promoter On the otherhand specific ldquoantirdquo sequences downstream of promoterscan limit the RNA polymerase escape process thus affectingpromoter activity [77] such elements were found to decreasesigma70 and sigma32 promoter activity up to 10-fold [69]Davis et al also tested the effects of different sequences flank-ing promoters downstream including an ldquoantirdquo sequence ordifferent reporter genes (GFP dsRed and Gemini) with thesame RBS yielding an activity change up to 2-fold [73] Asimilar fold change was observed in analogous experimentswhereMartin et al [78] testedGFP lacZ-alpha andGemini asreporter genes In their work however the 2-fold differencepersists for the strongest promoter which might be affectedby an excess of the lacZ-alpha fragment compared to theomega fragment needed for complementation A study of ourgroup [20] yielded a lower estimate of activity change for a setof 5 widely used promoters expressing the green fluorescentprotein (GFP) with the BBa B0032 RBS or the red fluorescentprotein (RFP) with the BBa B0032 or BBa B0034 RBS onlyone of the tested promoters showed a significant activitychange among the three conditions with a coefficient of vari-ation (CV) of 22 The abovementioned studies expressedpromoter activities in RPUs in order to provide comparablemeasurements among the different reporters used Recentadvances in DNA synthesis assembly and high-throughputcharacterization techniques enabled the quantification ofvery large libraries of single gene expression cassettes com-posed by different promoters RBSs and target genes bymeasuring the fluorescence of reporter gene as well asmRNAlevel via qPCR or next generation sequencing In particularKosuri et al performed the so far largest scale experimentalstudy where 114 promoters and 111 RBSs were combinedupstreamof aGFP gene [60] Promoterswere found to triggerconsistent RNA levels of the downstream transcript amongthe different RBS-gene combinations By using an ANOVA

model for data interpretation it was found that promotersequence accounted for about 92 of total variability ofmRNA level demonstrating that promoters are the mainfactors affecting mRNA level even though they expresseddifferent mRNAs RBSs accounted for 4 of total variabilitywhich could be due to transcription rate modulation by thesequence downstream of promoter or to other phenomenanot involving transcription such as RBS-dependent mRNAdegradation or sequestration (see discussion in Section 22)

The majority of flanking sequence-dependent studieson promoters are relative to downstream sequences whileupstream sequences are less frequently studied Even thoughhighly stimulatory or inhibitory effects may be obtained viaUP or ldquoantirdquo sequences promoters were found to changetheir activity within a reasonably low fold-change when notflanked by such difficult elements

Although such data gave a significant contributiontowards the understanding of promoter reusability geneexpression systems composed by independent expressioncassettes are not similarly well studied and could yield unpre-dictable effects Hajimorad et al [79] studied the mRNAlevels produced by different gene expression cassettes totest the superposition of the effects in synthetic biologicalsystems at different copy number levels they found condi-tions where even three cassettes could provide predictablelevels of mRNA while in other configurations cassettescould not be considered as modular systems Similarly ourgroup [20] used two cassette-systems expressing GFP andRFP under the control of a set of promoters detectingfluorescence as output Cassette position was also studiedContext-dependent variability was higher than for individualcassette expressing different reporters (maximum CV of 33versus 22) A part of this variability could be explainedby a different upstream sequence that is promoters couldbe flanked by the transcriptional terminator of the upstreamcassette or by the plasmid sequence upstream of the cloningsite

Activity prediction studies given the nucleotide sequenceof promoters have not yet produced accurate tools for thewidely used sigma70 promoters Promoter strength can beaffected bymany sequence features which are not completelyunderstood yet including the minus35minus10 sequences the spacerbetween them and the above discussed flanking sequencesRecent efforts towards prediction include the works ofRhodius et al [6 80] who developed position weight matrix-based models to predict the activity of sigmaE promotersas a function of their sequence as well as their flankingsequences (UP elements) with good predictive performance(119903 = 086 after cross-validation) [6] However the samemethods are not likely to work for sigma70 promoters due totheir complex structure [80] De Mey et al used partial leastsquares (PLS) regression to classify promoter strength as afunction of nucleotide sequence [7] this approach accuratelypredicted the activity of 6 out of 7 promoters used as a testset Meng et al developed an artificial neural network (ANN)to predict the strength of regulatory elements composed bya promoter and an RBS [8] this approach brought to theaccurate prediction of an initial test set of 10 promoter-RBSpairs (119903 = 098) and good performance was also obtained on


a second set of 16 newly constructed pairsThedescribed toolsprovided promising results but additional work is needed toindependently validate such methods on other datasets andto fully understand promoter sequence features

In summary reproducible context-dependent variabilitystudies should be performed to fully understand the factorsaffecting promoter activity in individual expression cassettesand in multiple cassette systems Large libraries of parts arenow affordable and for this reason the analysis of suchfactors will be facilitated as well as activity prediction givenpromoter sequence Standard [37] and multifaceted [74]characterization approaches have been proposed to providerobust measurements that can be shared and reproduced inmany laboratories

22 RBSs RBSs are strongly context-dependent elementssince their surrounding sequences can affect ribosome bind-ing and as a result the translation initiation rate per tran-script In particular even a few nucleotide changes in theRBS or in the surrounding sequences can dramatically affecttranslation [10] and the use of different genes downstreamof an RBS can provide completely different translationalefficiencies [81] Given the sequence of a gene and its 51015840UTRbiophysical models have been used to predict the translationinitiation rate by modelling local and global folding as wellas the interaction between RBS and 16S ribosomal RNAComputational tools such as the RBS Designer (stand-aloneapplication [10]) the RBSCalculator (web-based application[9]) and the UTR Designer (web-based application [11])are available to perform such tasks They take into accountthe 51015840UTR sequence as well as the first portion of codingsequence to predict the translation initiation rate level TheRBS Calculator and UTR Designer use similar biophysicalthermodynamics-basedmodels while the RBSDesigner usesa steady-state kinetic model of stepwise-occurring reactions[82 83] These tools showed similar and reasonably goodpredictive performance (1199032 gt 08) and can also be used toforward-engineer novel RBSs with a desired strength [83]They differentiate for the use of different external tools forenergy computation [83] and for some specific peculiaritiesfor example RBS Calculator provides indication of confi-dence and it is frequently updated [84] RBSDesigner consid-ers long-range interactions within RNA and can predict thetranslation efficiency ofmRNAs thatmay potentially fold intomore than one structure while UTR Designer enables codonediting to minimize secondary structures [83] Other effortstowards RBS prediction include an artificial neural networkalready cited above to evaluate the strength of promoter-RBSpairs [8]

The RBS Calculator is one of the most commonly usedtools in the synthetic biology community it was used in basicresearch studies to tune the response of a synthetic AND gate[9] to generate a set of RBSs of graded strengths to evaluatethe transcriptiontranslation processes [85] and to test DNAassembly platforms [33 35] as well as in applied researchto optimize biosynthetic pathways [86 87] Although it wasproved to be useful to guide the choice of proper RBSsequences given a downstream gene its accuracy is limited

and additional tools should be developed to improve thepredictability of RBSs [57 81]

RBSs could also affect the mRNA decay rate by causingdifferent secondary structures [16] In addition Kosuri etal also observed a mutual interaction between transcriptionand translation in fact translation efficiency can affectmRNA levels probably because the most translated mRNAmolecules are protected from degradation compared to theleast translated mRNAs [60]

In summary as in the case of promoters large datasetshave been useful to show the contributions of differ-ent context-dependent factors Due to the strong context-dependent nature of RBSs experimental studies mainlyfocused on flanking sequences while the evaluation of RBSmodularity in complex circuits still needs to be studied

23 Genes Given a target protein its coding sequence canaffect both transcription and translation processes [15 88]As described above mRNA secondary structures couldaffect mRNA degradation and limit RBS accessibility toribosomes and in addition AT-rich sequences can causepremature transcriptional termination [89] Codon usage hasbeen reported to affect the translation process [90] In thisframework most of the efforts towards the prediction of thecontribution of gene sequence to transcriptiontranslationprocesses have focused on the development of gene optimiza-tion algorithms To define them several sequences need tobe constructed to cover a sufficient number of hypothesesalthough the cost of synthetic genes is greatly decreasinggene synthesis still brings to expensive studies [88] For thisreason the process of sequence optimization is not fullyunderstood and no consensus rules have been found for geneoptimization Some research studies identified strong sec-ondary structures as the primary limiting factors in proteinsynthesis [91] while other studies did not find a correlationbetween predicted secondary structure and expression level[92] On the other hand in some studies expression levelhas been found to correlate with the codon adaptation index(CAI) [93 94] often used to express the codon bias of agene towards common codons [95] while in other studiesthis correlation was null [88 91] The codon randomizationmethod where codons are extracted from codon usagefrequency tables was found to be superior to the ldquoone aminoacid-one codonrdquo strategy where the CAI is maximized [1592] Finally codon context that is the influence of codonpair usage was found to affect protein expression althoughno ready-to-use software tool is available to carry out anoptimization procedure based on such feature [90]

All the features described above might be gene andvariant dependent [88] and for this reason several studiesshould be conducted to identify the correct features ofgene sequence affecting transcription translation and otherprocesses In particular the simultaneous measurement ofmRNA and protein level can provide exhaustive data todecouple the effects of gene sequence changes on cellularprocesses In a large-scale study performed by Goodman etal a library of gt14000 expression systems was constructedto test the contribution of the N-terminal codons on gene


expression [96] they measured DNA RNA and proteinlevels and confirmed that mRNA secondary structure is acrucial factor which can tune gene expression up to sim14-fold

The research efforts carried out so far have brought todifferent gene optimization tools currently used by syntheticbiologists and gene synthesis companies to optimize proteinexpression according to codon usage frequency tables globalGC content minimization of hairpin structures within thegene andor of secondary structures in the N-terminalcodons [97 98]The free software tools proposed in literatureinclude for instance GeMS (web-based application [12])Optimizer (web-based application [13]) Synthetic GeneDesigner (web-based application [14]) and Gene Designer(stand-alone application [15]) All the tools mainly differ-entiate for their available options for designing genes (egavoid unwanted restriction sites and inverted repeats designframework of oligonucleotides for gene synthesis) and fortheir codon optimization strategy (eg ldquoone amino acid-onecodonrdquo method probabilistic methods or hybrid solutionsbased on codon frequency tables from different sources) totake into account codon usage and constraints Becausemanyavailable tools are proprietary of gene synthesis companiesan accurate comparison of the implemented methodologiesis not feasible and in addition their performances still needto be experimentally evaluated on different gene sets

In summary although prediction tools have been pro-posed no widely accepted algorithm is available to predictthe effects of gene sequence on transcription translation ormRNA degradation

24 Terminators Rho-independent terminators are hereinconsidered Although very efficient terminators are available(eg the popular BBa B0015 double terminator from theMITRegistry of Standard Biological Parts) the repeated use ofa small set of elements in a genetic circuit may result inpoor evolutionary stability [99 100] For this reason reliablemethods to design new terminators with predictable strengthand methods to predict the efficiency of already existingterminators given their sequence are required

Terminator efficiency can be characterized via an operon-structuredmeasurement system where a promoter drives theexpression of two different reporter geneswith the terminatorsequence to bemeasured that is assembled between these twogenes The two reporter proteins are quantified and termi-nation efficiency is computed from their values consideringthe operon without the terminator of interest as a control[16 17 101]

Like promoters and RBSs also terminator efficiency hasbeen found to be dependent on the surrounding contextIn particular Cambray et al [16] tested different minimalterminators including only the hairpin and U-tail sequencesand compared their termination efficiency to the respectivefull-length terminators Efficiencies significantly changedbetween the two contexts for almost all the 11 tested termi-nators demonstrating that sequences flanking the essentialterminator parts are crucial The authors also used a multiplelinear regression model to build up a predictive tool fortranscriptional termination given the terminator sequence

using a set of features identified via stepwise regression butthe resulting predictor gave poor performance on the 54terminators used (119903 = 061 after cross-validation) Only byexcluding the low efficiency terminators low predicted fold-ing frequency terminators and extended terminators classesthe Pearson correlation coefficient 119903 increased to 085 aftercross-validation Through a complementary approach Chenet al [17] experimentally characterized a large set of termi-nators (582) and analyzed how sequence features contributeto their strength The dominant features were used to buildup a biophysical model that aimed to capture terminationstrength (Ts) as a function of the U-tract hairpin loop stembase and A-tract-free energies The model was used to fit vialinear regression the experimentally determined Ts yieldinga squared 119903 value of 04 which results in low predictiveperformances Although not currently available to users thetools developed in the above publications [16 17] can beimplemented through the provided regression coefficientsweb-based nucleic acid folding tools and specific indexescomputed from terminator sequences These two recentstudies relied on experimental measurements performed viathe abovementioned operon structure with reporter genesHowever Cambray et al constructed measurement plasmidswith RNAse sites flanking the terminator to be measured inorder to avoid terminator-dependent mRNA folding whichmight affect the translation efficiencies of the two reportergenes The authors tested RFP-GFP and GFP-RFP operonswith terminators flanked by RNAse III RNAse E or non-functional RNAse III sitesThe configuration giving the lowercoefficient of variance for the upstream gene level was theRFP-GFP operon with RNAse III sites which was used for allthe characterization experiments of their paper ConverselyChen et al used a GFP-RFP operon without RNAse sitessince they found that in their configuration RNAse Esites presence affected the downstream gene expression Inlight of these findings a standard measurement method forterminators still needs to be defined in order to enable reliablequantifications and to avoid potential mechanisms that maycomplicate the measurement of terminator efficiency forexample promoters that might arise at the interface of theterminator to be measured and the downstream gene of theoperon [17]

In summary sequence features affecting terminatorsbehaviour have been recently evaluated on large datasets butpredictive models with good performances are not availableyet demonstrating that different models and additionalknowledge on transcriptional termination are needed as wellas a standardized setup for experimental measurements

25 Interconnected Networks and Retroactivity In the phi-losophy of bottom-up composition of biological systemsarbitrarily complex networks are considered as black-boxmodules that can be interconnected Their characterizationcan provide the essential elements to describe their steady-state and dynamic behaviour In a modular framework suchknowledge enables the prediction of composite networksfunctioning To quantitatively test themodularity boundariesof biological systems recent studies have focused on the


characterization of systems subparts and on the predictionof the behaviour of composite systems obtained upon theirinterconnection Wang et al [18] tested different regulatedpromoters (inducible by arabinose AHL and IPTG) as theinputs of ANDNAND gates whose output was visualizedvia GFP at two different temperatures After a fitting processinvolving one specific configuration (ie one of the citedinput modules) the fluorescence output of the other config-urations was predicted from the individual characterizationof input devices and ANDNAND gates Experimental dataand predictions exhibited a Pearson correlation coefficient of086 to 098 even though some specific input combinationsyielded highly different values Moon et al [19] constructedand characterized a set of AND gates Then they used themto engineer composite two layered logic functions a 3-inputsystem including 3 input devices connected to two ANDgates and a 4-input system including 4 input devices and 3AND gates The latter represented one of the largest geneticprograms built up so far with a total of 11 regulatory proteins21 kbp-length on three plasmids The basic AND gates wereindividually characterized as before and the output of thecomplex 3- and 4-input systems was predicted and comparedwith experimental data The 3-input system yielded a lowerdeviation between prediction and data compared to the 4-input system Our group also faced prediction problems withsimple interconnected networks composed by an input device(inducible promoters or constitutive promoters of differentstrengths) assembled with a TetR-based NOT gate whichprovides GFP as output [20] The individual input deviceswere characterized via RFP measurements and the steady-state transfer function output of the NOT gate driven by eachof the input systems was quantified These data were fittedwith a Hill function they had similar maximum activity andHill coefficients while the switch point varied about 44which was considered as an estimate of interconnection errorwith these elements

The mentioned studies evaluated interconnection-dependent variability in considerably complex systems butthey did not characterize the causes of such deviationsOne ofthe best characterized and formalized interconnection errorsis retroactivity a phenomenon that extends the electronicengineering notion of impedance or loading to biologicalsystems [5] The functioning of a given system can changewhen a downstream or upstream system is connected forexample because of unwanted sequestration of transcriptionfactors by the connected modules In this case the individualsystems cannot be considered to be modular howevergiven the knowledge of the parts to be combined suchunwanted interactions can be modelled thus having aninterconnected system with predictable behaviour Jayanthiet al [21] experimentally tested a model system includingan ATc-inducible LacI production module connected toa lac-repressible promoter with GFP downstream Thiscomposite system was placed in a medium-copy plasmid andtested individually or in presence of a downstream ldquoclientrdquoincluding lac operator sites in a high-copy plasmid thusproviding additional binding sites for LacI The presence ofthe client significantly affected the induction and deinductiondynamics This phenomenon was captured by a mechanistic

model describing the LacI-occupied DNA sites upstream ofGFP and in the client binding as a function of ATc induction

26 Circuit Architecture Most of the research studiesdescribed above are based on single gene cassettes Thepolycistronic operon structure could be preferred whenexpressing genes carrying out similar functions that can becontrolled by the same promoter Although predictable RBStuning in operons has been reported [87] the prediction ofprotein levels encoded by genes in operons is not trivial andcannot be simply inferred by the protein levels of individualgene cassettes In particular the specific operon structurecan affectmRNAdegradation rate and ribosome accessibilityLim et al developed and experimentally tested a mathemat-ical model of transcription and translation coupling whichpredicts the protein level encoded by the first gene as afunction of the operon length [22]They found and predictedprotein level variations up to 2- to 3-fold In a complementaryframework Levin-Karp et al studied the translational cou-pling of an operon that is the mutual relationships betweenthe translation efficiencies of neighbouring genes [102] Theyindividuated a gt10-fold change for the protein level encodedby the second gene as a function of the translation rate of thefirst gene However the findings of Lim et al and Levin-Karpet al were not valid for all combinations of genes and the samephenomena were not observed in different studies [61 102]

Themeasurement ofmRNA levels of a transcribed operonhas been useful to decouple the effects of RNA stabilityand translation rate change [102] In summary other math-ematical analyses are needed to develop predictive tools thatcan guide biological engineers in the composition of operonstructures with quantitatively predictable function whichcan be inferred by the knowledge of promoter RBSs genesequence genes position operon length and other possiblefeatures [22]

27 Genetic Context The context in which a gene expressioncassette or a complex circuit is placed can affect its quantita-tive behaviour Genetic contexts include plasmids replicatingat different copy numbers per cell or the bacterial chro-mosome Given a single gene expression cassette plasmidsequence can affect promoter or terminator activity bymeansof the sequences flanking the cloning site as described abovefor these two part classes Moreover intuitively DNA copynumber determines different levels of all the species (mRNAand protein) but such levels could be unpredictable sincecells may exhibit metabolic overloading when copy numberis increased thus showing nonlinear changes This effectis commonly observed in expression cassettes at high copynumber [20 79 103] and needs to be characterized when thecassette copy number is to be tuned Furthermore plasmidcopy number can be intrinsically noisy [104 105] and can alsochange when multiple plasmids are incorporated in the samecell [106] To test the latter case Lee et al [106] showed thatlow copy plasmids with the heat-sensitive pSC101 replicationorigin maintain their copy number (about 5 copies per cell)in single plasmid systems and in 3-plasmid systems whileplasmids with the medium or high copy replication origins


(p15A and ColE1 resp) showed copy number increase whenused in the 3-plasmid system compared to the single plasmidsystem

Mathematical models of gene regulatory networks oftenuse empirical Hill functions to describe activation or repres-sion of cellular species butDNAcopynumber is not explicitlypresent in the equations [23 103] For this reason even byassuming a linear change of cellular species as a functionof DNA copy number mechanistic mathematical modelsshould be defined to easily study the copy number effectsAlthough such models are also widely used to describebiochemical reactions they are more difficult to study andidentify than empirical models thus requiring additionalwork to fully characterize the system of interest Mileyko etal used such class of models to study the copy number effectson different gene network motifs [23]

The integration of the desired expression cassette in thebacterial chromosome determines the maintenance of itsDNA in a single copy replicated with the genome Howeverthe quantitative behaviour of parts in the genomic context canbe difficult to predict For example the real copy number ofthe desired DNA could change when integrated in differentgenomic positions because the sequences near the bacterialreplication origin are expected to be replicated earlier thanthe other sequences [24 107] and thus the specific DNAsegment is actually present in the cell at a slightly highercopy number on averageThe complexity of genomic contextis not limited to this effect and other not fully understoodphenomena could limit the prediction of an integratedcassette For example transcriptional read-through fromflanking genomic cassettes could affect the expression of thesynthetic cassette

3 Trial-and-Error Approaches

Thedesign of a desired biological function can be achieved byrandomly changing its DNA-encoded elements In particularpromoters RBSs architectures and contexts are varied viadisparate experimental methods and the resulting circuit isscreened The success of all these methods relies on partsgeneration and screening efficiency which should allowan easy and high-throughput construction and recognitionof the desired phenotype [60] Here only representativestudies are illustrated which randomly optimize promotersRBSs genes architectures and context towards a targetcircuitpathway functioning

Promoters upstream of one or more target genes israndomly changed by directly synthesizing new promotersequences or by assembling the genes under the controlof a collection of promoters In the first case degenerateprimers can be used to insert a new random promotersequence upstream of a gene [108] In the second casepromoters from existing collections of parts [55] or randomfragments [109 110] can be used in the same manner andthe resulting constructs are screened In this latter casethe characterization of promoters (or the quantification ofthe transcriptional activity of random fragments) is notrequired because only the circuit outcome is considered to

optimize the process These two methods can be combinedby producing libraries of synthetic random promoters whenrequired with the desired design constraints (eg the desiredoperator sites) [74 111] that are screened by reporter genes toyield a collection of parts with diverse and graded activitythen elements can be randomly assembled to tune thedesired circuitpathway [74 111] Such procedure could bepartially rational inducible promoters can be used to probethe optimal activity of a target gene and only a subset of thecandidate newly generated promoters having a constitutiveactivity similar to the optimal one can be tested [20 112 113]

By following a similar procedure RBSs can be randomlychanged and selected Anderson et al [38] and Kelly [101]repaired a nonfunctional AND gate and a logic inverterrespectively by random mutagenesis of the RBS upstreamof a regulatory gene The two gates were nonfunctionalbecause their activity range in input did not match theactivity range provided by the upstream promoter used in thefinal interconnected circuit The RBS sequence mutagenesisand screening process produced circuits with the expectedbehaviour The use of existing collections of RBSs can also beexploited instead of creating new ones [42 114] The randommutagenesis of promoters and RBSs can be performed viadifferent widely used molecular biology methods includingerror-prone PCR or DNA amplification with degenerateprimers High-throughput techniques have been recentlyproposed to simultaneously mutate the sequence of severalelements also in the genome via automated procedures Themultiplex automated genome engineering (MAGE) approachwas used coupled with a microfluidic automatic system andwith degenerate single-strandedDNAs to enable the lycopenepathway optimization through RBSmutagenesis for 24 targetgenes in plasmid or genome [115]

Genes have been randomly mutated mainly to obtaindifferent functional protein variants with improved perfor-mance [59] Since this approach causes amino acid variationinstead of synonymous codon replacement the resultingprotein is different Such approaches are beyond the focus ofthis review Codon change studies without affecting proteinsequence are not widely used and they are limited to theexperimental works carried out to find gene optimizationrules as described in Section 23 of this review Similarly ter-minators are not commonly targeted for random mutations

When dealingwith polycistronic designs the architectureof gene expression cassettes can be randomly varied bychanging the position of the genes in an operon or by flankinggenes with libraries of tunable intergenic regions (TIGRs)[116] Since the target protein level produced by genes inoperons is not currently predictable the first intuitivemethod relies on random change of gene position This inseveral studies yielded highly diverse protein levels amongthe shuffled constructs For example bicistronic operonsincluding the 1a-hydroxylase adrenodoxin and NADPH-adrenodoxin reductase genes (called ADX and ADR) usedas redox partners to characterize the 25-hydroxyvitamin D31a-hydroxylase gene were switched (yielding ADX-ADR andADR-ADX constructs) and both ADR and ADX expressionlevels varied up to 5-fold [117] On the other hand theuse of TIGRs relies on the assembly of various control


elements (mRNA secondary structures RNAse cleavage sitesRBS sequestering sequences etc) within operon genes Thisrandom approach has proved to enable a gt100-fold range ofenzyme levels and a 7-fold improvement of productivity for asynthetic mevalonate pathway [116]

The genetic context can also be randomly optimizedPlasmid copy number change is an intuitive method totune the output of circuits and pathway Kittleson et al[118] constructed different-allele (DIAL) strains that had thesame genetic background except for an expression cassetteproviding different protein levels of a trans-acting replicationfactor (Pi or RepA) plasmids with the R6K and ColE2replication origins can be maintained at disparate copynumber per cell levels due to the regulation by Pi and RepArespectively The resulting strains were successfully usedto optimize a violacein biosynthetic pathway Consideringgenetic context at genomic level different methods wereused to optimize integration position and copy numberof synthetic DNA-encoded production pathways via ran-dom approaches Santos et al developed a recombinase-assisted genome engineering (RAGE) approach where loxsites recognized by the Cre recombinase are exploited tointegrate very large synthetic DNA fragments into the desiredgenomic position thus enabling the trial-and-error searchamong several predefined candidate loci [119] They usedit to optimize a 34Kb heterologous pathway for alginatemetabolism On the other hand the random insertion of thedesired DNA parts is often carried out through transposableelements By randomly optimizing promoter activity andgenomic position at the same time Yomano et al optimizedthe expression of an ethanol production pathway [120] Inparticular they integrated a promoter-less 3-cistron ethanolproduction cassette in random positions of the strain ofinterest via a mini-Tn5 cassette (transpososome) relying onthe random placement of the cassette under the control ofpromoters with optimal strength in the optimal genomicposition

Chromosomally integrated circuits or pathways can bealso optimized by randomly changing their copy numberMethods to carry out this task rely on genomic integrationof the DNA of interest together with an antibiotic resistancecassette subsequently recombinant strains are evolved inpresence of increasing antibiotic concentration to promotethe tandem duplication of the recombinant DNA cassetteuntil a target efficiency is reached This method has providedrecombinant strains containing more than 25 copies of theDNA-encoded ethanol production pathway to be optimized[121 122] A further refinement of the methods was carriedout by Tyo et al where the chemically inducible chromoso-mal evolution (CIChE) was described [123] It is analogousto the previously described procedure but when the desiredefficiency is reached the recA gene (promoting homologousrecombination) is knocked out CIChE was applied to poly-3-hydroxybutyrate (PHB) and lycopene production yieldingsignificant pathway improvement (4-fold and 60 resp)This method produced approximately 40 consecutive copiesof the DNA-encoded pathway and 10-fold improvement ongenetic stability [123]

4 Interventions on Circuit Structure toImprove Predictability

Although individual parts networks architectures and con-texts have the abovementioned predictability issues severalefforts have been undertaken to modify some of theseelements to decrease their context-dependent variability andimprove their predictability

Davis et al designed a set of insulated promoters thatextend from minus105 to +55 from the transcription start site [73]These elements had amore predictable activity than noninsu-lated promoters when tested in different contexts Mutalik etal proposed a bicistronic design (BCD) of gene expressioncassettes to effectively predict the translation initiation rateof a downstream gene [81] This design includes a smallopen reading frame (ORF) with its own RBS assembleddownstream of the promoter of interest The stop codon ofthis ORF is fused to the start codon of the gene of interest(thus having TAATG) which is assembled downstreamThe RBS of the gene of interest is included in the smallORF upstream With this design inhibitory RNA structuresaround the gene of interest start codon or RBS are eliminatedby the intrinsic helicase activity of ribosomes arriving at thestop codon of the upstream ORF By forward-engineering anexpression cassette via BCD users should obtain the expectedrelative expression within 2-fold of the target value with 93probability which represents a great improvement over state-of-the-art predictive tools for RBSs [9 81]

Qi et al proposed the use of bacterial clustered regularlyinterspaced short palindromic repeat (CRISPR) pathwayelements to engineer specific posttranscriptional cleavage ofmultigene operons to yield predictable expression of the indi-vidual genes also when placed in different positions [124]Via a complementary approach Lou et al used ribozymesassembled downstream of a promoter to improve the pre-dictability of gene expression [125] ribozymes cleave themRNA eliminating their 51015840 end and also act as transcriptioninsulators

Del Vecchio et al [5 126] proposed a system able to over-come retroactivity issues upon interconnection of biologicalsystems thus implementing a buffer (or insulator) deviceIt strongly relies on engineering-inspired insulators such asnoninverting operational amplifiers The biological imple-mentation of this mechanism includes phosphorylation-dephosphorylation reactions which act with fast timescalesbut it needs to be experimentally validated

5 Conclusions

This review has described several aspects of the designof genetic circuits with predictable function Bottom-upapproaches have been recently investigated to mimic thetraditional design processes in engineering areas In thiscontext research studies have been carried out to evaluatethe predictability boundaries of biological systems composedby precharacterized parts providing the expected intercon-nection error estimated from the study of model systemsand highlighting situations where circuits cannot behave


(i) What is still needed to fully understand the context-dependent variability ofbiological parts

(ii) How can we develop accurate computational tools for the bottom-up biologicalengineering

(iii) How can we support the construction of novel synthetic biology-basedsolutions

(iv) How can we understand model and predict cell-to-cell variability in thefunctioning of biological systems

(v) If the fully predictable engineering of biological systems is not possible how canwe improve the efficiency of trial-and-error approaches

Box 2 Outstanding questions

as intended Mathematical models support the bottom-updesign steps from the early feasibility study of complexfunctions to the quantitative prediction of circuit behaviourfrom the knowledge of basic parts function and finally to thedebugging step

To exploit the full potential of synthetic biology via anengineering-inspired bottom-up design of circuits severalchallenges need to be facedThemain crucial issues identifiedin the context of this work are delineated in Box 2 in the formof outstanding questions and they are herein discussed

Predictable biological engineering requires deepeningour knowledge on context dependency and reusability of bio-logical parts by discovering the features that play importantroles in parts function predictability Technology advancesin the DNA synthesis field can support the testing of largenumber of hypotheses by providing huge libraries of con-structs at affordable price In fact although large-scale studieshave been reported to support the investigation of differentaspects of parts predictability [60 81 96] the cost and scale ofDNA synthesis are still a major bottleneck for basic researchsince many studies require a very large number of constructvariants as in the case of codon usage dependency in pro-tein expression [88] The development of high-throughputmethods for parts measurement plays a complementary rolebecause multifaceted characterization of parts performanceneeds to be carried out In particular to fully characterizethe activity of parts the simultaneous quantification of DNARNA and proteins is required to accurately decouple effectsdue to circuit copy number transcription and translationto improve the knowledge of all the atomic steps involvedin parts function In addition ad hoc experimental designsdata analysis tools and mathematical models can supportthe above procedures for example models can be of helpin the estimation of nonobservable parameters useful tocharacterize parts function [36]

Empirical mathematical models of gene regulatory net-works are currently used to summarize the function ofparts and predict the quantitative behaviour of higher-orderdevices Although they are widely used in some cases mech-anistic models could bemore appropriate tools such as in thestudy of DNA copy number variations or retroactivity effectsOther tools enable the prediction of parts activity from theknowledge of their nucleotide sequence Although promisingresults have been obtained particularly in the case of RBSs

that are already optimized via these computational methodsthese tools need to be significantly improved The data andknowledge gained in the above ldquodiscoveryrdquo step are to beexploited in the development of predictive computationaltools with greater accuracy than the current ones In thiscontext novel tools can be based on the acquired biologicalknowledge which will be used to define essential rulesfor parts function prediction or can be data-based wheremachine learning methods are used to learn the relationshipsof interest for parts prediction Context-dependent activitychange of individual parts and mathematical models ofinterconnected networks should ultimately be integrated tocontribute unique tools for interconnected circuit designfrom parts sequence

In addition to existing parts prediction an ambitiousgoal of synthetic biology is the construction of unnaturalparts with finely tuned customized function To this aim thecomputational design tools need to be expanded to supportthe forward engineering of new components according tospecific design rules learned from data examples or fromthe acquired biological knowledge Again the currentlyavailable RBS design tools already enable the design ofRBSs with desired strength given the downstream genesequence although their performance needs to be signifi-cantly improved [57] Specifically the RBS Calculator com-putes novel RBS sequences with about 47 chance to showthe target strength within 2-fold [81]

Even though most of our current biological knowledgeis based on population-averaged data and central tendencyvalues cell-to-cell variability is a crucial issue and can bring tounpredictable system behaviour Although the main aspectsof this point are described elsewhere [127] and are beyondthe scope of this review we want to highlight that biologicalnoise can be detrimental for circuits function even whencentral tendency values are predictable For this reasonthe full characterization of biological components shouldalso take into account cell-to-cell variability which needsto be propagated throughout an interconnected network ofwell-characterized modules to obtain reliable quantitativepredictions of network output

In this review trial-and-error approaches involving therandom-based optimization of partscircuit function havealso been briefly illustrated These approaches rely on


affordable parts construction methods and efficient high-throughput-compatible screening methods to select the bestcombination of genetic parts while these approaches cannotbe efficiently applied when this condition does not persistThe technology advances mentioned above could greatlysupport the generation of large libraries to be screenedvia appropriate high-throughput measurement techniqueseven without significant improvements in biological discov-eries about context-dependent variability However whilethe learning of predictability boundaries is expected tocontribute definitive predictive tools to handle the com-plexity of biological systems trial-and-error approaches donot ensure the success of synthetic biology In fact largenumbers of candidate constructs can be built up but high-throughput measurement methods are not always availablefor the quantitative evaluation of circuit activity and theimpact of pure trial-and-error approaches remains limitedto specific projects For this reason bottom-up approachesurgently need to be refined to exploit the full potential ofsynthetic biology A mixture of prediction tools even withnonoptimal accuracy and trial-and-error approaches couldrapidly boost the efficiency of biological engineering byproviding a smaller search space than fully random-basedapproaches

Finally intense interventions on genetic circuits havebeen reported which can provide considerable improve-ments to the predictability of promoters RBSs architectureand retroactivity issues in different contexts Since suchimprovements are highly promising these modificationsshould be used in different studies to demonstrate theirbenefits on large scale and they should be considered in allthe previously mentioned issues

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

References

[1] H M Sauro ldquoModularity definedrdquo Molecular Systems Biologyvol 4 p 166 2008

[2] D Endy ldquoFoundations for engineering biologyrdquo Nature vol438 no 7067 pp 449ndash453 2005

[3] R Kwok ldquoFive hard truths for synthetic biologyrdquo Nature vol463 no 7279 pp 288ndash290 2010

[4] M Muers ldquoSynthetic biology quality and quantityrdquo NatureReviews Genetics vol 14 no 5 article 303 2013

[5] D Del Vecchio A J Ninfa and E D Sontag ldquoModularcell biology retroactivity and insulationrdquo Molecular SystemsBiology vol 4 article 161 2008

[6] V A Rhodius V K Mutalik and C A Gross ldquoPredicting thestrength of UP-elements and full-length E coli 120590119864 promotersrdquoNucleic Acids Research vol 40 no 7 pp 2907ndash2924 2012

[7] M De Mey J Maertens G J Lequeux W K Soetaert andE J Vandamme ldquoConstruction and model-based analysis of apromoter library for E coli an indispensable tool for metabolicengineeringrdquo BMC Biotechnology vol 7 article 34 2007

[8] H Meng J Wang Z Xiong F Xu G Zhao and Y WangldquoQuantitative design of regulatory elements based on high-precision strength prediction using artificial neural networkrdquoPLoS ONE vol 8 no 4 Article ID e60288 2013

[9] H M Salis E A Mirsky and C A Voigt ldquoAutomated design ofsynthetic ribosome binding sites to control protein expressionrdquoNature Biotechnology vol 27 no 10 pp 946ndash950 2009

[10] D Na and D Lee ldquoRBSDesigner software for designingsynthetic ribosome binding sites that yields a desired level ofprotein expressionrdquo Bioinformatics vol 26 no 20 pp 2633ndash2634 2010

[11] S W Seo J-S Yang I Kim B E Min S Kim and GY Jung ldquoPredictive design of mRNA translation initiationregion to control prokaryotic translation efficiencyrdquo MetabolicEngineering vol 15 no 1 pp 67ndash74 2013

[12] S Jayaraj R Reid andDV Santi ldquoGeMS an advanced softwarepackage for designing synthetic genesrdquo Nucleic Acids Researchvol 33 no 9 pp 3011ndash3016 2005

[13] P Puigbo E Guzman A Romeu and S Garcia-Vallve ldquoOPTI-MIZER a web server for optimizing the codon usage of DNAsequencesrdquo Nucleic Acids Research vol 35 no 2 pp W126ndashW131 2007

[14] G Wu N Bashir-Bello and S J Freeland ldquoThe SyntheticGene Designer a flexible web platform to explore sequencemanipulation for heterologous expressionrdquo Protein Expressionand Purification vol 47 no 2 pp 441ndash445 2006

[15] A Villalobos J E Ness C Gustafsson J Minshull and SGovindarajan ldquoGene Designer a synthetic biology tool forconstructuring artificial DNA segmentsrdquo BMC Bioinformaticsvol 7 article 285 2006

[16] G Cambray J C Guimaraes V KMutalik et al ldquoMeasurementand modeling of intrinsic transcription terminatorsrdquo NucleicAcids Research vol 41 no 9 pp 5139ndash5148 2013

[17] Y J Chen P Liu A A K Nielsen et al ldquoCharacterization of 582natural and synthetic terminators and quantification of theirdesign constraintsrdquo Nature Methods vol 10 no 7 pp 659ndash6642013

[18] B Wang R I Kitney N Joly and M Buck ldquoEngineeringmodular and orthogonal genetic logic gates for robust digital-like synthetic biologyrdquo Nature Communications vol 2 no 1article 508 2011

[19] T S Moon C Lou A Tamsir B C Stanton and C A VoigtldquoGenetic programs constructed from layered logic gates insingle cellsrdquo Nature vol 491 no 7423 pp 249ndash253 2012

[20] L Pasotti N Politi S Zucca M G Cusella De Angelis and PMagni ldquoBottom-up engineering of biological systems throughstandard bricks a modularity study on basic parts and devicesrdquoPLoS ONE vol 7 no 7 Article ID e39407 2012

[21] S Jayanthi K S Nilgiriwala and D Del Vecchio ldquoRetroactivitycontrols the temporal dynamics of gene transcriptionrdquo ACSSynthetic Biology vol 2 no 8 pp 431ndash441 2013

[22] H N Lim Y Lee and R Hussein ldquoFundamental relationshipbetween operon organization and gene expressionrdquo Proceedingsof the National Academy of Sciences of the United States ofAmerica vol 108 no 26 pp 10626ndash10631 2011

[23] Y Mileyko R I Joh and J S Weitz ldquoSmall-scale copynumber variation and large-scale changes in gene expressionrdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 105 no 43 pp 16659ndash16664 2008

[24] D H S Block R Hussein L W Liang and H N LimldquoRegulatory consequences of gene translocation in bacteriardquoNucleic Acids Research vol 40 no 18 pp 8979ndash8992 2012


[25] M Simeoni G De Nicolao P Magni M Rocchetti andI Poggesi ldquoModeling of human tumor xenografts and doserationale in oncologyrdquo Drug Discovery Today Technologies vol10 no 3 pp e365ndashe372 2013

[26] B P Kovatchev M Breton C Dalla Man and C Cobelli ldquoInsilico preclinical trials a proof of concept in closed-loop controlof type 1 diabetesrdquo Journal of Diabetes Science and Technologyvol 3 no 1 pp 44ndash55 2009

[27] E Andrianantoandro S Basu D K Karig and R WeissldquoSynthetic biology new engineering rules for an emergingdisciplinerdquoMolecular Systems Biology vol 2 2006

[28] D E Cameron C J Bashor and J J Collins ldquoA brief historyof synthetic biologyrdquo Nature Reviews Microbiology vol 12 pp381ndash390 2014

[29] G M Church M B Elowitz C D Smolke C A Voigt andR Weiss ldquoRealizing the potential of synthetic biologyrdquo NatureReviews Molecular Cell Biology vol 15 pp 289ndash294 2014

[30] MIT Registry of Standard Biological Parts httppartsreg-istryorg

[31] R P Shetty D Endy and T F Knight ldquoEngineering BioBrickvectors from BioBrick partsrdquo Journal of Biological Engineeringvol 2 article 5 2008

[32] J C Anderson J E Dueber M Leguia G C Wu A P Arkinand J D Keasling ldquoBglBricks a flexible standard for biologicalpart assemblyrdquo Journal of Biological Engineering vol 4 article 12010

[33] J E Norville R Derda S Gupta et al ldquoIntroduction ofcustomized inserts for streamlined assembly and optimizationof BioBrick synthetic genetic circuitsrdquo Journal of BiologicalEngineering vol 4 article no 17 2010

[34] M A Speer and T L Richard ldquoAmplified insert assembly anoptimized approach to standard assembly of BioBrick geneticcircuitsrdquo Journal of Biological Engineering vol 5 article 17 2011

[35] M Leguia J A N Brophy D Densmore A Asante and JC Anderson ldquo2ab assembly a methodology for automatablehigh-throughput assembly of standard biological partsrdquo Journalof Biological Engineering vol 7 no 1 article 2 2013

[36] B Canton A Labno and D Endy ldquoRefinement and stan-dardization of synthetic biological parts and devicesrdquo NatureBiotechnology vol 26 no 7 pp 787ndash793 2008

[37] J R Kelly A J Rubin J H Davis et al ldquoMeasuring the activityof BioBrick promoters using an in vivo reference standardrdquoJournal of Biological Engineering vol 3 article 4 2009

[38] J C Anderson C A Voigt and A P Arkin ldquoEnvironmentalsignal integration by a modular and gaterdquo Molecular SystemsBiology vol 3 p 133 2007

[39] A Tamsir J J Tabor and C A Voigt ldquoRobust multicellularcomputing using genetically encoded NOR gates and chemicalldquowiresrdquordquo Nature vol 469 no 7329 pp 212ndash215 2010

[40] L Pasotti M Quattrocelli D Galli M G Cusella de Angelisand PMagni ldquoMultiplexing and demultiplexing logic functionsfor computing signal processing tasks in synthetic biologyrdquoBiotechnology Journal vol 6 no 7 pp 784ndash795 2011

[41] B Wang and M Buck ldquoCustomizing cell signaling usingengineered genetic logic circuitsrdquo Trends in Microbiology vol20 no 8 pp 376ndash384 2012

[42] T S Gardner C R Cantor and J J Collins ldquoConstruction ofa genetic toggle switch in Escherichia colirdquo Nature vol 403 no6767 pp 339ndash342 2000

[43] M B Elowitz and S Leibier ldquoA synthetic oscillatory network oftranscriptional regulatorsrdquo Nature vol 403 no 6767 pp 335ndash338 2000

[44] J Stricker S Cookson M R Bennett W H Mather L STsimring and JHasty ldquoA fast robust and tunable synthetic geneoscillatorrdquo Nature vol 456 no 7221 pp 516ndash519 2008

[45] T Danino O Mondragon-Palomino L Tsimring and J HastyldquoA synchronized quorumof genetic clocksrdquoNature vol 463 no7279 pp 326ndash330 2010

[46] S Basu RMehreja SThibergeM T Chen andRWeiss ldquoSpa-tiotemporal control of gene expression with pulse-generatingnetworksrdquoProceedings of theNational Academy of Sciences of theUnited States of America vol 101 no 17 pp 6355ndash6360 2004

[47] S Basu Y Gerchman C H Collins F H Arnold and RWeissldquoA synthetic multicellular system for programmed patternformationrdquo Nature vol 434 no 7037 pp 1130ndash1134 2005

[48] J J Tabor H Salis Z B Simpson et al ldquoA synthetic genetic edgedetection programrdquo Cell vol 137 no 7 pp 1272ndash1281 2009

[49] A E Friedland T K Lu X Wang D Shi G Church and J JCollins ldquoSynthetic gene networks that countrdquo Science vol 324no 5931 pp 1199ndash1202 2009

[50] R Daniel J R Rubens R Sarpeshkar and T K Lu ldquoSyntheticanalog computation in living cellsrdquo Nature vol 497 no 7451pp 619ndash623 2013

[51] C J Paddon P J Westfall D J Pitera et al ldquoHigh-level semi-synthetic production of the potent antimalarial artemisininrdquoNature vol 496 no 7446 pp 528ndash532 2013

[52] K de Mora N Joshi B L Balint F B Ward A Elfick andC E French ldquoA pH-based biosensor for detection of arsenicin drinking waterrdquo Analytical and Bioanalytical Chemistry vol400 no 4 pp 1031ndash1039 2011

[53] C E French K de Mora N Joshi A Elfick J Haseloff andJ Ajioka ldquoSynthetic biology and the art of biosensor designrdquoin Institute of Medicine (US) Forum on Microbial ThreatsThe Science and Applications of Synthetic and Systems BiologyWorkshop Summary National Academies Press WashingtonDC USA 2011

[54] A J Wargacki E Leonard M N Win et al ldquoAn engineeredmicrobial platform for direct biofuel production from brownmacroalgaerdquo Science vol 335 no 6066 pp 308ndash313 2012

[55] F Zhang J M Carothers and J D Keasling ldquoDesign of adynamic sensor-regulator system for production of chemicalsand fuels derived from fatty acidsrdquo Nature Biotechnology vol30 no 4 pp 354ndash359 2012

[56] L Serrano ldquoSynthetic biology promises and challengesrdquoMolec-ular Systems Biology vol 3 article 158 2007

[57] A P Arkin ldquoA wise consistency engineering biology for con-formity reliability predictabilityrdquo Current Opinion in ChemicalBiology vol 17 no 6 pp 893ndash901 2013

[58] T K Lu A S Khalil and J J Collins ldquoNext-generationsynthetic gene networksrdquo Nature Biotechnology vol 27 no 12pp 1139ndash1150 2009

[59] Y Yokobayashi RWeiss and F H Arnold ldquoDirected evolutionof a genetic circuitrdquo Proceedings of the National Academy ofSciences of the United States of America vol 99 no 26 pp16587ndash16591 2002

[60] S Kosuri D B Goodman G Cambray et al ldquoComposability ofregulatory sequences controlling transcription and translationin Escherichia colirdquo Proceedings of the National Academy ofSciences of the United States of America vol 110 no 34 pp14024ndash14029 2013

[61] F Chizzolini M Forlin D Cecchi and S S Mansy ldquoGeneposition more strongly influences cell-free protein expressionfrom operons than T7 transcriptional promoter strengthrdquo ACSSynthetic Biology 2013


[62] F Ceroni S Furini E Giordano and S Cavalcanti ldquoRationaldesign of modular circuits for gene transcription a test of thebottom-up approachrdquo Journal of Biological Engineering vol 4article 14 2010

[63] D Na S M Yoo H Chung H Park J H Park and S Y LeeldquoMetabolic engineering of Escherichia coli using synthetic smallregulatory RNAsrdquo Nature Biotechnology vol 31 no 2 pp 170ndash174 2013

[64] F Crick ldquoCentral dogma of molecular biologyrdquoNature vol 227no 5258 pp 561ndash563 1970

[65] M H Medema R Van Raaphorst E Takano and R BreitlingldquoComputational tools for the synthetic design of biochemicalpathwaysrdquo Nature Reviews Microbiology vol 10 no 3 pp 191ndash202 2012

[66] J Ang E Harris B J Hussey R Kil and D R McMillenldquoTuning response curves for synthetic biologyrdquo ACS SyntheticBiology vol 2 no 10 pp 547ndash567 2013

[67] N Crook and H S Alper ldquoModel-based design of syntheticbiological systemsrdquo Chemical Engineering Science vol 103 pp2ndash11 2013

[68] Y Cai M L Wilson and J Peccoud ldquoGenoCAD for iGEMa grammatical approach to the design of standard-compliantconstructsrdquoNucleic Acids Research vol 38 no 8 pp 2637ndash26442010

[69] W Kammerer U Deuschle R Gentz and H Bujard ldquoFunc-tional dissection of Escherichia coli promoters information inthe transcribed region is involved in late steps of the overallprocessrdquoThe EMBO Journal vol 5 no 11 pp 2995ndash3000 1986

[70] S Leirmo and R L Gourse ldquoFactor-independent activationof Escherichia coli rRNA transcription I Kinetic analysis ofthe roles of the upstream activator region and supercoilingon transcription of the rrnB P1 promoter in vitrordquo Journal ofMolecular Biology vol 220 no 3 pp 555ndash568 1991

[71] T Caramori and A Galizzi ldquoThe UP element of the promoterfor the flagellin gene hag stimulates transcription from bothSigD- and SigA-dependent promoters in Bacillus subtilisrdquoMolecular and General Genetics vol 258 no 4 pp 385ndash3881998

[72] S T Estrem T Gaal W Ross and R L Gourse ldquoIdentificationof an UP element consensus sequence for bacterial promotersrdquoProceedings of the National Academy of Sciences of the UnitedStates of America vol 95 no 17 pp 9761ndash9766 1998

[73] J H Davis A J Rubin and R T Sauer ldquoDesign constructionand characterization of a set of insulated bacterial promotersrdquoNucleic Acids Research vol 39 no 3 pp 1131ndash1141 2011

[74] H Alper C Fischer E Nevoigt and G Stephanopoulos ldquoTun-ing genetic control through promoter engineeringrdquo Proceedingsof the National Academy of Sciences of the United States ofAmerica vol 102 no 36 pp 12678ndash12683 2005

[75] S T Estrem W Ross T Gaal et al ldquoBacterial promoterarchitecture Subsite structure of UP elements and interactionswith the carboxy-terminal domain of the RNA polymerase 120572subunitrdquo Genes and Development vol 13 no 16 pp 2134ndash21471999

[76] W Ross A Ernst and R L Gourse ldquoFine structure of E coliRNA polymerase-promoter interactions 120572 subunit binding tothe UP element minor grooverdquo Genes and Development vol 15no 5 pp 491ndash506 2001

[77] C L Chan and C A Gross ldquoThe anti-initial transcribedsequence a portable sequence that impedes promoter escaperequires 12059070 for functionrdquo The Journal of Biological Chemistryvol 276 no 41 pp 38201ndash38209 2001

[78] L Martin A Che and D Endy ldquoGemini a bifunctionalenzymatic and fluorescent reporter of gene expressionrdquo PLoSONE vol 4 no 11 Article ID e7569 2009

[79] M Hajimorad P R Gray and J D Keasling ldquoA frameworkand model system to investigate linear system behavior inEscherichia colirdquo Journal of Biological Engineering vol 5 article3 2011

[80] V A Rhodius and V K Mutalik ldquoPredicting strength andfunction for promoters of the Escherichia coli alternative sigmafactor sigmaErdquo Proceedings of the National Academy of Sciencesof the United States of America vol 107 no 7 pp 2854ndash28592010

[81] V K Mutalik J C Guimaraes G Cambray et al ldquoPreciseand reliable gene expression via standard transcription andtranslation initiation elementsrdquo Nature Methods vol 10 no 4pp 354ndash360 2013

[82] DNa S Lee andD Lee ldquoMathematicalmodeling of translationinitiation for the estimation of its efficiency to computationallydesign mRNA sequences with desired expression levels inprokaryotesrdquo BMC Systems Biology vol 4 article 71 2010

[83] B Reeve T Hargest C Gilbert and T Ellis ldquoPredictingtranslation initiation rates for designing synthetic biologyrdquoFrontiers in Bioengineering and Biotechnology vol 2 article 12014

[84] A Espah Borujeni A S Channarasappa and H M SalisldquoTranslation rate is controlled by coupled trade-offs betweensite accessibility selective RNA unfolding and sliding atupstream standby sitesrdquo Nucleic Acids Research vol 42 no 4pp 2646ndash2659 2014

[85] G Pothoulakis F Ceroni B Reeve and T Ellis ldquoThe SpinachRNA aptamer as a characterization tool for synthetic biologyrdquoACS Synthetic Biology vol 3 182 no 3 p 187 2014

[86] C Bi P Su J Muller et al ldquoDevelopment of a broad-host syn-thetic biology toolbox for ralstonia eutropha and its applicationto engineering hydrocarbon biofuel productionrdquoMicrobial CellFactories vol 12 article 107 2013

[87] F F Nowroozi E E K Baidoo S Ermakov et al ldquoMetabolicpathway optimization using ribosome binding site variantsand combinatorial gene assemblyrdquo Applied Microbiology andBiotechnology vol 98 no 4 pp 1567ndash1581 2014

[88] MWelch S Govindarajan J E Ness et al ldquoDesign parametersto control synthetic gene expression in Escherichia colirdquo PLoSONE vol 4 no 9 Article ID e7002 2009

[89] C Gustafsson J Minshull S Govindarajan J Ness A Villalo-bos and M Welch ldquoEngineering genes for predictable proteinexpressionrdquo Protein Expression and Purification vol 83 no 1pp 37ndash46 2012

[90] B K S Chung and D Y Lee ldquoComputational codon optimiza-tion of synthetic gene for protein expressionrdquo BMC SystemsBiology vol 6 p 134 2012

[91] G Kudla A W Murray D Tollervey and J B PlotkinldquoCoding-sequence determinants of expression in Escherichiacolirdquo Science vol 324 no 5924 pp 255ndash258 2009

[92] H GMenzella ldquoComparison of two codon optimization strate-gies to enhance recombinant protein production in EscherichiacolirdquoMicrobial Cell Factories vol 10 article 15 2011

[93] C Gustafsson S Govindarajan and J Minshull ldquoCodon biasand heterologous protein expressionrdquo Trends in Biotechnologyvol 22 no 7 pp 346ndash353 2004

[94] M Graf T Schoedl and R Wagner ldquoRationales of genedesign and de novo gene constructionrdquo in Systems Biology and


Synthetic Biology P Fu and S Panke Eds pp 411ndash438 JohnWileyamp Sons Hoboken NJ USA 2009

[95] P M Sharp and W H Li ldquoThe codon adaptation indexmdashameasure of directional synonymous codon usage bias and itspotential applicationsrdquoNucleic Acids Research vol 15 no 3 pp1281ndash1295 1987

[96] D B Goodman G M Church and S Kosuri ldquoCauses andeffects of N-terminal codon bias in bacterial genesrdquo Science vol342 no 6157 pp 475ndash479 2013

[97] C Elena P Ravasi M E Castelli S Peiru and H G MenzellaldquoExpression of codon optimized genes in microbial systemscurrent industrial applications and perspectivesrdquo Frontiers inMicrobiology vol 5 article 21 2014

[98] G L Rosano and E A Ceccarelli ldquoRecombinant proteinexpression in Escherichia coli advances and challengesrdquo Fron-tiers in Microbiology vol 5 p 172 2014

[99] S C Sleight B A Bartley J A Lieviant and H M SauroldquoDesigning and engineering evolutionary robust genetic cir-cuitsrdquo Journal of Biological Engineering vol 4 p 12 2010

[100] L Pasotti S Zucca M Lupotto M G Cusella De Angelisand P Magni ldquoCharacterization of a synthetic bacterial self-destruction device for programmed cell death and for recom-binant proteins releaserdquo Journal of Biological Engineering vol 5article 8 2011

[101] J R Kelly Tools and reference standards supporting the engineer-ing and evolution of synthetic biological systems [PhD thesis]Massachusetts Institute of Technology 2008

[102] A Levin-Karp U Barenholz T Bareia et al ldquoQuantifyingtranslational coupling in E coli synthetic operons using RBSmodulation and fluorescent reportersrdquo ACS Synthetic Biologyvol 2 no 6 pp 327ndash336 2013

[103] S Zucca L Pasotti GMazziniMG Cusella DeAngelis and PMagni ldquoCharacterization of an inducible promoter in differentDNA copy number conditionsrdquoBMCBioinformatics vol 13 no4 article S11 2012

[104] N J Guido X Wang D Adalsteinsson et al ldquoA bottom-upapproach to gene regulationrdquoNature vol 439 no 7078 pp 856ndash860 2006

[105] Y Dublanche K Michalodimitrakis N Kummerer MFoglierini and L Serrano ldquoNoise in transcription negativefeedback loops simulation and experimental analysisrdquoMolecular Systems Biology vol 2 article 41 2006

[106] T S Lee R A Krupa F Zhang et al ldquoBglBrick vectors anddatasheets a synthetic biology platform for gene expressionrdquoJournal of Biological Engineering vol 5 article 12 2011

[107] S Zucca L Pasotti N Politi M G Cusella De Angelis and PMagni ldquoA standard vector for the chromosomal integration andcharacterization of BioBrick parts in Escherichia colirdquo Journal ofBiological Engineering vol 7 no 1 article 12 2013

[108] C SolemandP R Jensen ldquoModulation of gene expressionmadeeasyrdquoApplied and EnvironmentalMicrobiology vol 68 no 5 pp2397ndash2403 2002

[109] L O Ingram and T Conway ldquoExpression of different levelsof ethanologenic enzymes from Zymomonas mobilis in recom-binant strains of Escherichia colirdquo Applied and EnvironmentalMicrobiology vol 54 no 2 pp 397ndash404 1988

[110] A Martinez S W York L P Yomano et al ldquoBiosyntheticburden and plasmid burden limit expression of chromosomallyintegrated heterologous genes (pdc adhB) in Escherichia colirdquoBiotechnology Progress vol 15 no 5 pp 891ndash897 1999

[111] T Ellis X Wang and J J Collins ldquoDiversity-based model-guided construction of synthetic gene networks with predictedfunctionsrdquo Nature Biotechnology vol 27 no 5 pp 465ndash4712009

[112] K Temme D Zhao and C A Voigt ldquoRefactoring the nitrogenfixation gene cluster from Klebsiella oxytocardquo Proceedings of theNational Academy of Sciences of the United States of Americavol 109 no 18 pp 7085ndash7090 2012

[113] N Politi L Pasotti S Zucca et al ldquoHalf-life measurements ofchemical inducers for recombinant gene expressionrdquo Journal ofBiological Engineering vol 8 article 5 2014

[114] R Weiss Cellular computation and communications using engi-neered genetic regulatory networks [PhD thesis] MassachusettsInstitute of Technology 2001

[115] H H Wang F J Isaacs P A Carr et al ldquoProgramming cellsby multiplex genome engineering and accelerated evolutionrdquoNature vol 460 no 7257 pp 894ndash898 2009

[116] B F Pfleger D J Pitera C D Smolke and J D KeaslingldquoCombinatorial engineering of intergenic regions in operonstunes expression of multiple genesrdquo Nature Biotechnology vol24 no 8 pp 1027ndash1032 2006

[117] N Sawada T Sakaki S Kitanaka K Takeyama S Kato and KInouye ldquoEnzymatic properties of human 25-hydroxyvitaminD3 1120572-hydroxylase Coexpression with adrenodoxin andNADPH-adrenodoxin reductase in Escherichia colirdquo EuropeanJournal of Biochemistry vol 265 no 3 pp 950ndash956 1999

[118] J T Kittleson S Cheung and J C Anderson ldquoRapid optimiza-tion of gene dosage in E coli using DIAL strainsrdquo Journal ofBiological Engineering vol 5 article 10 2011

[119] C N Santos D D Regitsky and Y Yoshikuni ldquoImplementationof stable and complex biological systems through recombinase-assisted genome engineeringrdquo Nature Communications vol 4article 2503 2013

[120] L P Yomano S W York S Zhou K T Shanmugam andL O Ingram ldquoRe-engineering Escherichia coli for ethanolproductionrdquoBiotechnology Letters vol 30 no 12 pp 2097ndash21032008

[121] K Ohta D S Beall J P Mejia K T Shanmugam andL O Ingram ldquoGenetic improvement of Escherichia coli forethanol production chromosomal integration of Zymomonasmobilis genes encoding pyruvate decarboxylase and alcoholdehydrogenase IIrdquo Applied and Environmental Microbiologyvol 57 no 4 pp 893ndash900 1991

[122] P C Turner L P Yomano L R Jarboe et al ldquoOptical mappingand sequencing of the Escherichia coli KO11 genome revealextensive chromosomal rearrangements and multiple tandemcopies of the Zymomonas mobilis pdc and adhB genesrdquo Journalof Industrial Microbiology amp Biotechnology vol 39 no 4 pp629ndash639 2012

[123] K E J Tyo P K Ajikumar and G Stephanopoulos ldquoStabilizedgene duplication enables long-term selection-free heterologouspathway expressionrdquo Nature Biotechnology vol 27 no 8 pp760ndash765 2009

[124] L Qi R E Haurwitz W Shao J A Doudna and A P ArkinldquoRNA processing enables predictable programming of geneexpressionrdquoNature Biotechnology vol 30 no 10 pp 1002ndash10062012

[125] C Lou B Stanton Y Chen B Munsky and C A VoigtldquoRibozyme-based insulator parts buffer synthetic circuits fromgenetic contextrdquo Nature Biotechnology vol 30 no 11 pp 1137ndash1142 2012


[126] D Del Vecchio ldquoA control theoretic framework for modularanalysis and design of biomolecular networksrdquo Annual Reviewsin Control vol 37 pp 333ndash345 2013

[127] B Li and L You ldquoPredictive power of cell-to-cell variabilityrdquoQuantitative Biology vol 1 no 2 pp 131ndash139 2013

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


MEDIATORSINFLAMMATION

of


Behavioural Neurology

EndocrinologyInternational Journal of



Disease Markers


BioMed Research International

OncologyJournal of



Oxidative Medicine and Cellular Longevity


PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of



Computational and Mathematical Methods in Medicine

OphthalmologyJournal of


Diabetes ResearchJournal of



Research and TreatmentAIDS


Gastroenterology Research and Practice


Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom


models also play an important role in a biological systemsframework in fact they are often used to study com-plex metabolic interactions like those occurring in diseaseconditions to understand the underlying processes andorpredict the effect of drugs [25] Some mathematical modelsof biologicalphysiological systems have also been approvedby the US Food and Drug Administration (FDA) for usein simulated clinical trials thus enabling researchers forexample to support or even skip expensive in vivo trials [26]

Synthetic biology aims to realize novel complex biologicalfunctions with the same principles on which engineeringdisciplines lay their foundationsmodularity abstraction andpredictability [2 27 28] As a result synthetic biologists sofar have mainly focused on the definition of biological partsand on their abstraction and standardization in order todeal with well-defined components with specific function[29] This process has brought to the creation of biologicalparts repositories including DNA parts that can be sharedby the scientific community like the MIT Registry of Stan-dard Biological Parts [30ndash32] to standardized and easy-to-automate DNA assembly strategies [33ndash35] and to standardmeasurementmethodologies to share characterization resultsof parts like promoters [36 37] Researchers have alsofocused on the realization of engineering-inspired functionsto learn the complexity that could be reached in a biologicalcontext Towards this goal researchers built up devices thatimplement logic gates and functions [19 38ndash41] memories[42] oscillators [43ndash45] other waveform generators [46 47]signal processing devices [48ndash50] and the likeMany of themrelied on mathematical models to support the early designsteps and to capture the behaviour of the designed circuit Forexample two of the synthetic biologymilestones are a genetictoggle switch [42] and an oscillator (the repressilator) [43]both implemented in Escherichia coli via genetic networks ofproperly connected transcriptional regulators A semiquan-titative investigation of the features required for a correctcircuit behaviour was performed via mathematical modelsby using dimensionless equations or reasonable parametervalues Thanks to the model analysis the authors could learnuseful guidelines for correct design of circuits exhibiting thedesired functioning for example fast degradation rates ofrepressor proteins encoded in the oscillatory network [43]

The realization of complex functions has brought tosome biological systems of high impact An engineeredpathway was implemented in recombinant yeast to producethe antimalarial drug precursor artemisinin [51] a biosensor-encoding genetic device was implemented in microbes todetect arsenic in drinking water and to provide a colourchange of its growth medium as visual output [52 53]microbes were recently engineered to produce bioethanolfrom algal biomass [54] or advanced fuels from differentsubstrates [55]

However despite many examples of complex engi-neering-inspired function implementation and also of indus-trially relevant solutions to global health environmental andenergy problems a rigorous bottom-up design process is notcurrently adopted because the predictability boundaries stillhave to be clearly defined [3 56 57] The high potentialof synthetic biology strongly depends on the achievement

of such task [58] Trial-and-error approaches represent analternative if synthetic biologists cannot design a systemfrom the bottom-up they can rely on random approacheswhere for example circuit components are mutated andthe best candidate implementing the function of interest isselected [38 59 60] Depending on the reliability of pre-dictions and of mathematical models this process could becompletely random or partially guided In general trial-and-error approaches are time- and resource-consuming and arecharacterized by a low efficiency However recent advancesin the construction of biological systems for example DNAandor strain production via automated proceduresmay pro-vide a good alternative to the rational bottom-up approachespecially when accurate automated and possibly low-costscreeningmethods are available to rapidly evaluate the outputof the constructed circuits [60]

This review discusses the predictability issues of basicbiological parts (promoters ribosome binding sitesmdashRBSscoding sequences transcriptional terminators and plasmids)when used to design the desired biological function in theform of a simple or complex gene expression system Eventhough synthetic biological systems may be implementedin several organisms (or even in vitro [61]) and may havedisparate architectures and regulatory mechanisms [62 63]the reviewwill focus on predictability of parts in vivo in the Ecoli bacterium according to the biological information flowdescribed in the central dogma of molecular biology [64]protein-coding DNA sequences (herein called genes) aretranscribed into mRNA molecules which are converted intoproteins by ribosomes and finally DNA sequences can bereplicated in living cells to propagate the encoded function tothe progenyThus in the considered framework the possiblebasic architectures are shown in Box 1 promoters can triggerthe expression of a single gene (monocistronic architecture)or a set of genes (polycistronic or operon architecture) eachgene is transcribed with its RBS upstream and finally termi-nators stop transcription Ribosomes complete the process bytranslation of mRNA molecules into the proteins of interestfrom the start codon (generally AUG) to the stop codon(generally UAA) Complex genetic circuits can be realizedwith a set of such gene expression units implementing theinteractions of interest and giving the desired product asoutput Genetic circuits can be placed on a plasmid vector orotherwise they can be integrated into a target position of thebacterial chromosome

Even if other classes of parts can be used to constructcomplex genetic systems and other elements can also affectcircuit behaviour we will focus only on the abovemen-tioned genetic parts and architectures given a specificstrain and environment Other important contexts like thehost (the reciprocal variation of parts behaviour and hostmetabolism when a circuit is incorporated) environmental(the reciprocal variation of parts behaviour and environ-mental parameters) ecological (changes of synthetic circuitand surrounding community parameters as well as strainsfitness) and evolutionary (changes of DNA composition)contexts are reviewed elsewhere [57] Other reviews are


Genomic DNA

Replication origin

(c)

(a) (b)












Promoters


[6]


[7]


RBSs


[9]


[10]


[11]

Genes


[12]


[13]


[14]


[15]

Terminators


[16]


[17]



[5 18ndash21]


[22]

Context



[24]






















































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

















Genomic DNA

Replication origin

(c)

(a) (b)












Promoters


[6]


[7]


RBSs


[9]


[10]


[11]

Genes


[12]


[13]


[14]


[15]

Terminators


[16]


[17]



[5 18ndash21]


[22]

Context



[24]






















































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of



















Promoters


[6]


[7]


RBSs


[9]


[10]


[11]

Genes


[12]


[13]


[14]


[15]

Terminators


[16]


[17]



[5 18ndash21]


[22]

Context



[24]






















































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of





































































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of































































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of





















































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of











































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of




































5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

























5 Conclusions






















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of




































References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of





















References










































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

































































































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of



























































































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of

























































of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of
























of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of





















of






Disease Markers



OncologyJournal of





PPAR Research



Journal of

ObesityJournal of
















Documents

Review Article Advances and Computational Tools towards Predictable Design …downloads.hindawi.com/journals/cmmm/2014/369681.pdf · 2019. 7. 31. · toggle switch [ ]andanoscillator(therepressilator