Successive state transitions with I O interface by molecules

Successive State Transitions with I/O Interfaceby MoleculesKen Komiya1,Kensaku Sakamoto,Hidetaka Gouzu,Shigeyuki Yokoyama, Masanori Arita3,Akio Nishikawa2,and Masami Hagiya1Department of Biophysics and Biochemistry,2Department of Information Science,Graduate School of Science,University of Tokyo,7–3–1Hongo Bunkyo-ku,113–0033Tokyo,JAPAN.{komiya,sakamoto,yokoyama}@biochem.s.u-tokyo.ac.jp{nisikawa,hagiya}@is.s.u-tokyo.ac.jp3Electrotechnical Laboratory1–1–4Umezono,Tsukuba-shi,305-8568Ibaraki,JAPAN.arita@etl.go.jpAbstract.This paper reports three experimental achievements in ourcomputation model based on‘whiplash’reactions.Wefirst show thata single-stranded DNA(ssDNA)can serve as an independent machineby using a solid support technique.Second,we show how to append anarbitrary sequence,e.g.a transition state or a PCR primer,to the3’-end of a molecular machine,thus realizing its I/O interface.Finally wedemonstrate the successive state transitions for several steps on solidphase with I/O.1IntroductionAutonomous DNA computing assumes that each molecule works not only as a data carrier but also as a microscopic computing unit.Since we intro-duced Whiplash model,a computation model based on whiplash reactions(Fig-ure1)[4],several advances on our model have been reported in different direc-tions.In the theoretical side,Winfree showed the implementation of GOTO pro-grams and the efficient solution of the directed Hamiltonian path problem[10]. In the experimental side,we showed the more efficient transition steps using an isothermal reaction[9].The isothermal conditions have made the successive state transitions more realistic than do the common thermal schedules for PCR[9].The normal PCR cycle comprises the three steps of denaturation,annealing,and polymerization, each performed at a different temperature.In contrast,the isothermal program intends to perform these three steps simultaneously,thus coupling the dissocia-tion of the‘current state’sequence from the transition table with the polymer-ization of the‘next state’sequence(Figure1).This coupled reaction was found to actually occur at80◦C,with an optional use of64◦C that facilitates a hairpin formation.A.Condon(Ed.):DNA2000,LNCS2054,pp.17–26,2001.c Springer-Verlag Berlin Heidelberg200118K.Komiya et al.Fig.1.Whiplash model is thefirst to implement a program,or state transitions,into ssDNA.(1)The current state A anneals to a transition table forming a hairpin structure.(2)At its3’-end,the next state B is appended by polymerization,which stops at the stopper sequence(white box).(3)Denaturation.(4)The new state B anneals to another position in the transition table.In this paper,we show the successful multiple transition steps on solid phase in Whiplash model.We also report the successful implementation of in-put and output(I/O)interface in the model.The implementation of I/O interface is a significant progress from the al-ready reported experimental results.From the beginning of DNA computing, molecules have been used as a memory device in realistic computation mod-els.Although‘DNA as memory’is the original paradigm shown by Adleman and Lipton,it confined DNA computing to the‘single-instruction,multiple-data (SIMD)’model.In order to realize‘multiple-instruction,multiple-data(MIMD)’model,it is necessary to implement both a program and I/O interface in DNA sequences.Whiplash model is thefirst realistic model to support both with DNA.Independence of each molecular machine is a prerequisite for an autonomous molecular computation.Although each DNA molecule can,in theory,work as one machine in our model,DNA molecules were not immobilized and could freely interact in the experiment of our previous paper[4].In this paper,we employ the surface-based approach to prevent an intermolecular reaction.Thus,Whiplash model made a major progress toward realistic MIMD computation.Reactions on solid phase naturally necessitate I/O interface for a molecu-lar machine.Since DNA molecules are immobilized on solid phase,we cannotSuccessive State Transitions19directlyanalyzethe whiplash reaction by gel electrophoresis,as we did previ-ously[9].In addition,the subsequence at the3’-end of the molecule cannot be used as a PCR primer-binding site,because the3’-end sequence always has a complementary counterpart in the transition table,and the base pairing between them inhibits PCR amplification.The way to avoid this difficulty is to append a ‘readout’sequence to the3’-end in the course of the successive transitions(Fig-ure2).The elongated molecules can be subjected to PCR amplification with the readout sequences as primers,thus reporting the current state of the machine. Note here that,under some experimental conditions,only a minor population will undergo this sequence extension,and the remaining molecules can continue further transition reactions.AA BINPUT OUTPUT20K.Komiya et al.2Experiments 2.1Materials and MethodsDNA.The DNA sequences used here(Table1)were designed using a genetic al-gorithm package,GENESIS.The details of this sequence design will be presented elsewhere [1].Oligonucleotides,including 5’-biotinylated ones,were commercially synthesized by Amersham Pharmacia Biotech (Tokyo,Japan).Hereafter,DNA sequences and oligomers are represented with bold letters.Table 1.DNA sequences output by the genetic algorithmno.sequence 0CCGTCTTCTTCTGCT 1TTCCCTCCCTCTCTT 2CGTCCTCCTCTTGTT 3CCCCTTCTTGTCCTT 4TGCCCCTCTTGTTCT 5CTCCTCTTCCTTGCT 6CTTCTCCCTTCCTCT 7CCTTCCTTCCCTCTT 8TCCCCTTGTGTGTGT 9GAGAGAGAGGCCCCCTATCC 10GAAGAGAAGGGCACCCCTCC 11GGGAAGGGACGCAACACCAC(5’)9121324354Successive State Transitions21 primer10,whose sequence is complementary with that of10,and was then immobilized to streptavidin-coated beads(Dynal)according to the supplier’s protocol.Conversion of Tran7into the single-stranded form was done by alkaline treatment.I/O Interface.An arbitrary sequence can be appended to the3’-end of the state machine by polymerization using an appropriate oligomer as the template. This polymerization was performed under the conditions similar to the whiplash reaction[9];the reaction buffer contains only dATP,dGTP,and dCTP but not dTTP,and the thermal schedule for‘input’reaction was94◦C for30sec,80◦C for30sec,and64◦C for20min,with the addition of r Taq DNA polymerase (Toyobo,Tokyo,Japan)just before the incubation at80◦C.For starting the successive transitions,an initial state was set on Tran7by appending the sequence complementary to a state sequence(from1to8)using an appropriate input oligomer.After this‘input’reaction,the input oligomer was removed by alkaline treatment from Tran7,and then Tran7was subjected to successive whiplash reactions.For readout of the results of compuation,the output oligomers(11–1–0, 11–2–1,···to11–8–7)are to be added after the completion of successive transitions.On the othe hand,for probing the status of the machine(Tran7), the output oligomers were added,into the reaction mixture,in the course of successive transitions.In both cases,the‘readout’sequence(11)is appended to the3’-end of(a subset of)the molecules.Successive Transitions(whiplash reactions).Tran7with the initial state, immobilized on the beads,was added to the buffer provided by the supplier (Toyobo)containing r Taq DNA polymerase(5units),dATP,dGTP,and dCTP (0.2mM each).State transition was performed in a25µl reaction,which contains 5pmol molecules(Tran7).The thermal schedule is as follows:–initial incubations at80◦C for1min–add r Taq DNA polymerase–94◦C for30sec–gradual cooling(in2min)to64◦C–15reaction cycles–64◦C for30sec–shift up to80◦C in1min–80◦C5min2.2ResultsWe developed an isothermal technique to perform successive whiplash reac-tions[9].Since we already achieved the efficient transitions up to2steps,more steps were tried in the present study.The employed DNA molecule(Tran7)is long enough to easily interact intermolecularly.For the purpose of preventing22K.Komiya et al.this intermolecular interaction,each Tran7was immobilized on a solid surface.The initial state 1was appended with the input oligomer (1–0–10)to the 3’-end of the immobilized molecule,which was then allowed to perform the whiplash reactions.The whiplash reaction consists of 15cycles,each of which includes the in-cubation phase at 64◦C,and the trasition phase at 80◦C.Note that the incu-bation phase is only to facilitate hairpin-structure formation.This operation is inherently different from a normal PCR reaction,in which polymerization and denaturation occur at a different temperature.Therefore,our reaction can be called an isothermal-reaction.During the successive transitions,each output oligomer was added to the reaction mixture for the status probing (Figure 4).The oligomer hybridized with the 3’-end of the machine after the completion of the specific transition step.Note here that this output reaction competes with the whiplash reaction for the next transition;the extent of this competition is not yet examined.91870105’3’3’5’Fig.4.The output oligomer 11–1–0appends the readout sequence 11to the 3’-end,and it probes the state 1.In the same way,the output oligomer 11–8–7works for probing the state 8.For converting the state machine into the double-stranded form,dTTP was added to the reaction mixture after the status probing.This conversion is nec-essary to use a restriction enzyme (Bam HI)for cutting offthe part polymerized during successive transitions.The cut-offpart was then amplified by PCR,with primers 10and 11,in order to detect it on a 8%polyacrylamide gel stained with ethidium bromide (Figure 5).The bands with the expected mobilities appeared up to 4succesive steps (lanes 0–4),and were significantly detected for the 5th and 6th steps,with major bands due to the unexpected products.The band for the 7th step was not observed.Successive State Transitions23 The unexpected products in the5th and6th steps were subjected to sequence analysis,and were found to comprise11–6–5–10for the5th step,and11–2–1–0–10for the6th.The reason for the occurrence of these sequences is probably hybridization between the extended part of the machine and PCR primers(or the remains of output oligomers),due to accidental similarities in their sequences.This was a pitfall in our sequence design.In order to show that our technique enables the state machine to start from any state,we appended sequence3–4,forcing the machine to start from the 4th state.We thus succeeded in skipping thefirst4steps,and in performing the following4transitions up to state8(data not shown).M01234567Fig.5.Gel electrophoresis of the transition products.The transiton was successful up to the4th step.Unexpected short major bands appear in the5th and6th steps.3Discussion3.1Benefits from I/O InterfaceWe realized the appending of any transition state or any readout sequence at the3’-end of the state machine.This implementation of I/O interface greatly improved the practicality of Whiplash model.A typical design of readout process for previous DNA computers has been to hybridize a marker DNA to a specific position which is pre-designed in DNA sequences.That is,what can be read is fixed by the initial sequence design.In our approach,on the other hand,reading or outputting an arbitrary state became possible.We list some of its applications.Starting from an Internal State.By appending an input sequence,we can start the state transition from any internal state,and can skip some transitions.24K.Komiya et al.Probing Transition States.By appending a PCR primer-binding site,we can check whether a certain state is transited.Note that appending of a primer-binding site does not alter the execution of other machines.This means that we can observe the transitions,by not stopping the execution of the unprobed machines.Data Transfer between Molecules.With the I/O technique,we could feed the output of one machine as the input of another.We could also preserve the computed data for the execution of different machines.This is a kind of‘DNA to DNA computations’[11],where information on a DNA molecule is transferred to another molecule,according to a logic arbitrarily given[11,12,13]3.2Reaction on Solid PhaseReaction using solid phase techniques has been used in DNA computing,mainly for solving NP-complete problems[6,5].The scalability and the potential for automation are usually referred as the advantages of the solid support,because DNA is used only as a memory device.On the other hand,the importance of solid support for whiplash reactions is guaranteeing the independence of each molecule.In this sense,our idea of this use of the surfaces is hinted by a report on an in vitro selection of RNA enzymes (ribozymes)[2].A ribozyme is a kind of the sophisticated molecular machines that can be artificially produced[3].3.3Sequence DesignSince the melting temperature of DNA molecules depends on their GC content, it is best to uniformly distribute GC in sequences.For this reason,we set the GC content of sequences to be the same.More important is the avoidance of unpexpected polymerization from the3’-end.For this reason,this part should not completely hybridize with other sequences.However,the sequence design satisfying these conditions became extremely difficult,especially when:–we can use only three bases,and–wefix the number of GCs in each sequence.One solution to get over this hardness is to enhance the genetic alphabet by using artificial bases such as isoguanosines(iG)or5-methylisocytidines(iC).We do not intend,however,to fully mix these bases with natural bases.We only need to introduce a single artificial base near each3’-end,because the polymerization of misproducts does not start unless the3’-end completely hybridizes with a ‘false’site.Therefore,a single iC(or iG)is enough to prevent an unexpected extension.Although artificial bases are not used in our experiment of this paper, the prospect of their special use is shown in our previous work[9].Successive State Transitions25 3.4Increasing the Number of Tranistion StepsThe reliability and the efficiency of intramolecular reaction may have been doubted,but we demonstrated the successful multi-step transitions.In this pa-per,we introduced three experimental landmarks in Whiplash model:–state transition on solid phase,–implementation of an I/O interface,–state transition for several steps.In whiplash reactions,the transition from the current state to the next one competes with the‘back annealing’of the current state to the previous position in the transition table.This difficulty was overcome with the isothermal conditions, developed previously[9].Several successive transitions were expected to confer other difficulties.One is that the sequences added at the3’-end during the transitions have comple-mentary counterparts in the transition table.Therefore,the longer becomes the state machine,the more chance it has to take a hairpin structure,which prohibits further transitions in the present design of the transition table.However,the drastic reduction in efficiency after the4th step,as shown above,was probably caused by the mis-annealing of the3’-end subsequence to inappropriate PCR primers or oligomers for output.It should be made clear which type of error is responsible for this result:an error during the state tran-sition,or an error during the PCR amplification for readout.Although either possibility must be separately considered in the design of DNA sequences,these possibilities may at least be attributed to the composition biased toward C and T.(Note that we do not use A in sequences.)They were designed to contain very few G,in order not to hybridize with eath other,but this bias seems to have induced the unexpected similarity between3’-termini of sequences.Careful re-design of sequences with more G is now ongoing.We consider that the transition for10steps will be possible with‘good’sequences.Another difficulty relates to the multiple occurrence of a state in the table, making the machine go along branched paths.This makes the‘back annealing’described above even more serious.This undesired effects of the‘back annealing’was anlyzed in terms of statistical thermodynamics by John A.Rose and Russell J.Deaton(University of Mempshis).Acknowledgment.This work is supported by the Japan Society for the Pro-motion of Science“Research for the Future”Program(JSPS-RFTF96I00101).Thefifth and seventh authors are supported by Grant-in-Aid for Scientific Research on Priority Area“Genome Science”from Ministry of Education,Sci-ence,Sports and Culture,Japan.26K.Komiya et al.References1.Arita,M.,Nishikawa,A.,Hagiya,M.,Komiya,K.,Gouzu,H.,and Sakamoto,K.:Improving Sequence Design for DNA Computing,pp.875–882,Proceedings of GECCO2000,2000.2.Bartel,DP.and Szostak,JW.:Isolation of New Ribozymes from a Large Pool ofRandom Sequences,Nature261,pp.1411–1418,1993.3.Ellington,AD.,Robertson,MP.,James,KD.,and Cox,JJ.:Strategies for DNAComputing,DIMACS Series in Discrete Mathematics and Theoretical Computer Science48,pp.173–184,1999.4.Hagiya,M.,Arita,M.,Kiga,D.,Sakamoto,K.,and Yokoyama,S.:Towards ParallelEvaluation and Learning of Boolean mu-formulas with Molecules,DIMACS Series in Discrete Mathematics and Theoretical Computer Science48,pp.57–72,1999.5.Liu,Q.,Wang,L.,Frutos,AG.,Condon,AE.,Corn RM.,and Smith LM.:DNAComputing on Sufaces,Nature403,pp.175–179,2000.6.Morimoto,N.,Arita,M.,and Suyama,A.:Solid Phase DNA Solution to the Hamil-tonian Path Problem,DIMACS Series in Discrete Mathematics and Theoretical Computer Science48,pp.193–206,1999.7.Ogihara,M.and Ray,A.:Simulating boolean circuits on DNA computers,Proc.1st International Conference of Computational Molecular Biology(ACM Press), pp.326–331,1997.8.Piccirilli,JA.,Krauch,T.,Moroney,SE.,and Benner,SA.:Enzymatic incorpora-tion of a new base pair into DNA and RNA extends the genetic alphabet,Nature 343,pp.33–37,1990.9.Sakamoto,K.,Kiga, D.,Komiya,K.,Gouzu,H.,Yokoyama,S.,Ikeda,S.,Sugiyama,H.,and Hagiya,M.:State Transitions by Molecules,Biosystems52, pp.81–91,1999.10.Winfree,E.:Whiplash PCR for O(1)computing,Proc.4th DIMACS Workshop onDNA Based Computers,pp.175–188,1998.ndweber,LF.,Lipton,RJ.,and Rabin,MO.:DNA2DNA Computations:A po-tential‘Killer App’?,DIMACS Series in Discrete Mathematics and Theoretical Computer Science48,pp.161–172,1999.12.Suyama,A.,Nishida,N.,Kurata,K.,and Omagari,K.:a poster abstract for RE-COMB2000,2000.13.Brenner,S.,Williams,SR.,Vermaas,EH.,Storck,T.,Moon,K.,McCollum,C.,Mao,JI.,Luo,S.,Kirchner,JJ.,Eletr,S.,DuBridge,RB.,Burcham,T.,and Al-brecht,G.:In vitro cloning of complex mixtures of DNA on microbeads:physical separation of differentially expressed cDNAs,Proceedings of National Academy of Science USA97,pp.1665-70,2000.。