BioFeature

Multichannel PCR and Serial TransferMachine as a Future Tool in EvolutionaryBiotechnology

A. Schoberl, N.G. Walten,U. Tangenl, G. Strunk, T.Ederhof, J. Dapprich andM. EigenMax-Planck-Institut fiirBiophysikalische Chemie,Gottingen; and 'Institut fiirMolekulare B iotechnologie,Jena. Germanv

ABSTRACT

As an improved strategy for producing

functional macromolecules by in vitro evo-lutionary optimization, we propose anautomated machine that can process up to960 samples in parallel. It consists of a960-well PCR machine with special sealedplastic reaction vessels and appropriatehandling devices. We show that the heat-sealing technique does not significantly af-fect the activity of Taq DNA Polymerase orthat of the temperature-sensitive Q0 RNApolymerase but avoids cross-contamina-tion and evaporation of the samples. In-itial experiments demonstrate the suitabil-ity of the apparatus to uniformly processthe samples and to perform the thermocy-cling. Serial transfer of reaction productsinto fresh reaction solution was used to in-itiate further rounds of amplification as atypical experimental setup for an evolu-tionary biotechnology application.

652 BioTechniques

INTRODUCTION

Optimizing macromolecules in ac-cordance with the rules of naturalevolution is a new strategy that has re-cently been introduced into modernbiotechnology. This evolutionary bio-technology is based on applying theselection principle to populations ofreplicating molecules, viruses, micro-organisms or cell populations that aresubjected to externally controlled se-lection pressures (2,6,'7). In this way,biological macromolecules evolve to-wards final products, which are opti-mized for a specific application, e.g., li-gand binding (9,13,23) or catalyticfunction (1-3).

The classical experiments of "evolu-tion in the test tube" by Spiegelmanand coworkers (16,21) were based onthe amplification of specific short se-quences of RNAs using the enzyme QBreplicase. Recently other enzymaticamplification techniques have been in-troduced into molecular in vitro evolu-tion: polymerase chain reaction (PCR)(e.g., References 1-3,23), 3SR (self-sustained sequence amplification) (5)and SDA (strand displacement amplifi-cation) (24).

The statistical analysis of evolution-ary phenomena requires screening oflarge numbers of molecules in isolatedcompartments (7). Different speciesmust be studied under identical reac-tion conditions, and, vice versa, theevolution of one species must be ob-

served under different environmentalconstraints. Data obtained from theseanalyses can assist in the evaluation ofnew optimization strategies (20). Toapply these statistical methods on evo-lutionary experiments with biologicalmacromolecules utilizing amplificationreactions like PCR (12,19), a machinefor homogeneous and fast temperaturecycling of many samples in parallel isneeded.

Many automated systems have beendeveloped that accomplish accuratetemperature cycling in appropriate re-action vessels (e.9., References 11,18,25). Some systems employ a roboticarm and a rack to move the sampletubes between water or oil baths. Thesesystems generally ensure a uniformtemperature distribution in the reactionvessels that is required for homogene-ous reaction conditions. More commonare programmable devices that producedifferent temperatures by heating andcooling a single metal block. However,these PCR machines do not always pro-vide a uniform temperature distributionacross the block, especially whencycling rapidly. Some commerciallyavailable temperature cyclers circum-vent this problem by using a stirredwater bath or oven at the expense offast cycling (22).

Here we present an automated tem-perature cycler for evolutionary experi-ments with up to 960 parallel PCRsamples of 10 - 40 pL in volume, util-izing three separate thermostating

Vol. 18, No.4 (1995)

aluminum blocks. A vacuum systemensures close contact between theseblocks and the moved sample carier.To deal with such a large number ofsamples and to avoid cross-contamina-tion, it was necessary to develop newtechniques of sample handling andprocessing. We therefore constructedspecial sealable and disposable plasticreaction vessels. All components of theapparatus were tested with RNA andDNA amplification reactions and a

PCR serial transfer system modelingthe technical requirements necessaryfbr evolutionary experiments. Bothcritical requirements-the unifbrmityof temperature for all samples andrapid heating-are met, and the overallprocessing time is short.

MATERIALS AND METHODS

The main components of the PCRmachine (Figure 1) are (l) plastic reac-tion vessels, each of which is a plasticsheet with 96 reaction vessels in a mi-croplate fomat (see Figure 24' andSealing of the Samples section below);(ii) a heat sealer to seal the PCR sam-ples in the reaction vessels (see Figure28 and Sealing of Samples); (lll) a

liquid delivery system including twoindependent pipetting robotic arms tofill and empty the reaction vessels con-taining the samples; (lv) a transporta-tion unit for moving the sample carrier(holding l0 plastic sheets with 96 reac-tion vessels each, i.e., a total of 960samples) between the diff'erent pipet-ting and temperature stations; (v) threethermostats to perform f-ast temperatureadjustments required for PCR and a

freeze-storage device to store the reac-tion solutions for further analysis; and,(vl) a process control unit with severalcomputers based on the VME (VersaModule Eurocard) data bus (a specialcomputer system.).

Sealing of the Samples

Sealed plastic sheets with reactionvessels were developed to ensure f'ast

processing under identical conditionsl'or each sample and to minimizeevaporation and cross-contamination(Figure 2A). The reaction vessels are

made from films extruded from Macro-lon@ (Kunststoffe Arthur Kriiger, Ham-burg, Germany); these proved to bechemically neutral in the test series.The bottom sheets (107 x 71 mm) withan initial thickness of 210 lrm are

placed on an aluminum mask contain-ing the wells and then subjected to hotair (270"-300'C) from above. Thereaction Vessels are sealed after thewells har e been filled with biochemicalreaction solution. Sealing is achievedby welding a 1OO-,um-thick Macroloncover foil to the sheet containing the re-action vessels b1" a heat sealer (Figure2B). The finished and sealed reactionvessels have a volume of .10 uL and awall thickness oi J0 uni.

As an alternative sealin_s method,two commercially available ultrasonicwave generators (Model 900 M [20

Figure 2. The plastic reaction vessels and theheat sealer. A) 96-reactron uells are containedon one plastic sheet. Left: corer sheet. middle:bottom sheet with wells. right: sealed plasticsheet. B) Photo of a sealing derice lor a plasticsheet with 96 rvells. A stepping motor lll drivesthe block with the heating cartridges f2l downonto a Tetlon block f.ll carry ing the plastic sheets

that are fired by an aluminum rnask [3]. Afterreaching the surf'ace of the plastic sheets, the car-tridges-at a temperature of 273'C-are slowlylowered a total of one millimeter in five steps

thereby sealing the tuo sheets vacuum-tightwithin 10 s. The insertion of the sheets into theheat sealer is manually done on a separate bench.A patent application for this heat sealer has been

tlled (Reference 8).

Figure 1. Scheme of the multichannel PCR machine. The transportation unit Il moves the samplecan'ier l2l between the diff'erent pipetting [3] and temperature stations [4,5]. The liquid delivery system

l3l includes two independent pipetting robot arms to fill and empty the plastic sheets containing the sam-ples. The three thermostating facilities f5l required fbr the PCR and the freeze-storage facility [4] are

shown. The aluminum cover of the central thermostat is removed. The sample carrier is machined with960 wells in the shape of the reaction vessel cavities to accommodate up to i0 plastic sheets with 96wells each.

Vol. 18. No. 4 (1995)

BioFeature

KHzl and Model 941 M [40 KHz];Branson, Hannover, Germany) weretested. By placing a sonotrode directlyon top of the two sheets, the reactionvessels are sealed within several milli-seconds.

Liquid Delivery System

A pipetting robot RSP 5052 withtwo arms (Zinsser Analytic; Tecan Ro-botics, Frankfurt, Germany) automat-ically delivers the samples to the reac-tion vessels using aerosol-resistant tips(ART) (Biozym, Hameln, Germany) onan arm with an 8-channel adaptor. fill-ing eight vessels in parallel. io etnptythe reaction vessels, the second arm ofthe robot automatically punctures thecover film and then extracts the reac-tion solutions. Sterile cannulas withluer manus norn are used in order tominimize cross-contamination.

Tiansportation Unit and SampleCarrier

To guarantee fast temperature transi-

tions together with convenient samplehandling, we used an x, y, z - manipula-tor (CNC machine 82Y4; Fcihrenbach,Lcifflingen, Germany). It moves thesample carrier made of aluminum (444x 260 x 5 mm) among the three differ-ent temperature stations required forthe PCR. The sample carier is ma-chined with 960 wells in the shape ofthe reaction vessel cavities to accom-modate up to 10 plastic sheets with 96wells each.

Thermostating and Freeze StorageFacilities

All thermostats are constructed fromlarge aluminum blocks (.490 x 210 x290 mm) to provide a high heat capac-ity. This ensures that their temperatureremains constant even when the cooleror warmer sample carrier is placed onit. Furthermore, this guarantees tem-perature homogeneity across the alumi-num block.

The desired temperatures are main-tained by heating elements placed in-side the blocks and by cooling with

water. The temperature of the samplecarrier is measured by a Pt 100 thermo-couple probe (type M-FK, measure-ment based on the temperature depend-ency of a resistance; Heraeus, Hanau,Germany), directly affixed to thesample carrier with silicon glue. Ho-mogeneity of static temperature acrossthe sample carrier was experimentallydetermined to be 0.5"C. To measuretemperature inside a water-filled wellof a heat-sealed reaction vessel, a Ni-CrNi sensor (type GTF 300; measure-ment based on the Seebeck effect;Greisinger, Regenstauf, Germany) wasinserted through the cover sheet, andthe sheet was sealed with hot-melt glue(Henkel, Diisseldorf, Germany).

After moving the samples to the ap-propriate thermostats, the sample car-rier is pressed onto the chosen stationby a vacuum system integrated intoeach block. The high thermal conduc-tivity of aluminum and the small thick-ness (5 mm) of the sample carrier ac-count for a high heat transfer rate. Toachieve a homogeneous and fast tem-perature adjustment for all 960 sam-ples, an additional preheated aluminumcover is lowered onto the samples auto-matically after the sample carrier is po-sitioned on the appropriate station.

BIOCHEMICAL TESTS

QB System

To test the compatibility of the seal-ing procedures with the Bp system, thereaction vessels were filled with 25-pLaliquots of reaction solution containing50 mM Tris-HCl pH 7.5, l07o (vol/vol)glycerol, l0 mM dithiorheirol,0.5 mMeach of ATP, CTP, GTP and UTP, 1.0pM ethidium bromide, 100 mM NaCl,1 ttM 0F replicase and I nM of theself-replicating RNA MNV 11 (4). Thereaction vessels were sealed as de-scribed above and immediately chilledon ice. After removing the sampleswith a sterile microsyringe, the con-tents of every five adjacent reactionwells were pooled to provide enoughsolution for fluorescence measure-ments in ultra-micro cuvettes with 100-ptL volumes. RNA amplification by Qppolymerase was monitored by measur-ing the fluorescence increase ofethidium bromide during amplificationat 30"C in a thermostated fluorometer

Figure 3. Fluorescence monitoring of RNA amplification by QB replicase after sealing the enzyme-containing reaction mixture and removing it from the reaction vessel. Sealing techniques and sub-sequent reaction conditions are described in Materials and Methods. Relative fluorescence signals areplotted vs. the reaction time ol RNA replication. Curve l: Ref'erence (unsealed reaction mixture);curve 2: a liquid QB reaction mixture was heat sealed; curve 3: a frozen Qp reaction mrxture washeat-sealed; curve 4: the reaction mixture was liquid, but the block for the heat-sealing device wasthermostated to -10'C, curve -5: a liquid reaction mixture was sealed with ultrasound; and curve 6:a frozen reaction mixture was sealed with ultrasound.

654 BioTechniques

o

I

va(IJ(Jc(u.Joo(-

(l)

t-

s00 1000 1s00

time (s )

2000 2500

Vol. 18, No.4 (1995)

o=oEa

e

100 150 2000 50 100 150 200 ?50 300 350

time (sl250 300 350 100

time

Figure 4. pcR cycling by the machine. A) Temperature profile of pCR cycles measured in the sample carrier. Time constant (half-time period): 7 s B) Tem-

perature profile of pcR cycles measured in the solution inside a plastic reaciion vessel placetl onto the corner of the sample carrier. Time constant (half-time pe-

LS5B (Perkin-Elmer, Ueberlingen, above. All reaction mixtures were ana-

Germany). Excitation wavelength was lyzed on a nondenaturing llTo poly-

l, = 514-nm, emission wavelength was acrylamide gel, which was photo-

l, = 600 nm. RNAreplication was initi- graphed after ethidium bromide

ated after 5-min preincubation by add- staining.

ing I pL of a I M MgCl2 stock solu-

tion. A control reaction was pedormed pCn Amplificationusing the same solution except that itwas"not sealed in plastic. DNA template (Sequence: s'-GGG-

Theseexperimenrswererepeatedby ATTTTTCCTTTTTCTCCCTCGCG-sealing a siock solution of 4.2 pM TAAGCTAGCTACGCGAGGTGAC-MNVII-RNA in water and amplify- CCCCCCCCGGGGGTTCTCCCCT-ing the RNA with fresh QB replicase ATAGTGAGTCGTAITA-3') and oli-

aciording to rhe protocol deicribed gonucleotide primers (primer I: 5'-

riod): 8.5 s.

Table 1. Time Needed for the Different Pipetting Steps

TAATAC GACTCACTI{TAGGGGAG-AACC-3'; primer II: 5'-GGGAITTTTCCTTTTTCTCCCT-3) were

synthesized on a Gene Assembler@

Plus DNA Synthesizer (Pharmacia Bio-tech, Freiburg, Germany) and purifiedby denaturing polyacrylamide gel elec-

trophoresis (PAGE). To test the differ-ent sealing techniques, 25-pL aliquots

of PCR mixture containing 0'5 PMDNA template, 0.5 PM of each PCRprimer, 10 mM Tris-HCl PH 8.3' 50

mM KCl, 1.5 mM MgCl2, 0.0l%o gela-

tin,200;.rM each of dATP, dCTP, dGTP

Process

No. of different 1

solutions

Technique ofdispensing

Time

Filling of oneplastic sheet

(96 wells),8-channel

arm

2 min

Fillingof 10 sheets(960 wells),8-channel

arm

1

20 min

Serialdilution

(5 times),8-channel

arm'l

6:30 min

Filling o{ oneplastic sheet

(96 wells),8-channel

arm

z

3:35 min

Emptying8 wells,

1 -channelarm

8

2:30 min

multi-dispensing, multi-dispensing, different positions, different positions, different positions'

lxStips toxbtips sxStips 2x8tips singlecannulas

The pCR process time for 34 temperature cycles of the described system is 1 .33 h. The whole process time of the serial trans-

fer pcR experiment (3 subsequent pcRs, 25 cycles each) including all automatic steps and sealing is 3'5 h. Insertion of the

olastic sheet into the heat sealer and subsequent sealing is done manually on a separate bench and takes 2 min'

Vol. 18. No.4 (1995)

BioFeature

I

rNs<h***-9:19:39:SgRnSE

r,g,'."s. Non-u""**ingPAGE analysisof lea.:tiol products of PCRsamdes P::':*::"d ?{i::i::::i::Tii:1,t::#J:t:?Tti:J;::ffiT:#l

lj1".li,i;j):,1-ir'jili[iililtff;:, ffiilil"ia,,i^.,.pr", (lanes r-22frrom an areas oriample canier. Both gels contain a sample processed in a com-

mercial TRlO-Thermocycler in lane R and 2 pL DNA marker i lnoenrmger l'tannheim, Mannheim''Germany) in lane M. Molecular weights are given in bp'

1241048980645'7

5l

t2110489ri06451

II

t

and dTTP and 0.04 tJllt'LTaq DNA Po-

lymerase (Stratagene, Heidelberg, Ger-

many) were distnbuted on the plastic

sheets and subjected to the weldingprocedures described above' Twentymicroliters were removed from each re-

action chamber with a microsyringeand transferred to commercral 0.5-pLplastic tubes. An additional 20 pL ofPCR mixture were not sealed and

served as a control solution. Ten mt-

croliters of mineral oil were added to

each PCR sample before subjecting

them to 34 cycles of amplitication. The

following temperatures and reaction

times were employed by a commerical

TRlO-ThermocyclerrM (Biometra, Got-

tingen, Germany): 30 s at 93oC, 30 s at

55"C, 30 s (7 min during the last cycle)

at 72'C. Finally. thel r.tere \tored at

10'C until they were analyzed by non-

denaturing 1l% PAGE.Similar expenments were Per-

formed by sealing a stock solution of 1

pM of DNA temPlate in water and

subsequently subjecting it to a PCR

with fresh primers and laq DNAPollrncrasc. according to tlte protocol

described above.T0 test the uniform Processing oi

parallel PCR samples by the robot de-

scribed here. 96- x 30-ptl- volumes otthe above-described PCR mixture. but

with 90 template molecuies Per PL

656 tsrdlecirniques

(0.15 fM) and 0.02 Pg/PL of salmon

sperm DNA (Sigma-Aldrich' Deisen-

hofen, Germany), were heat-sealed in

the reaction vessels. The plastic sheet

was cut by hand into Pieces wirh 2 x 2reaction chambers that were distributeduniformly over the whole area of the

sample carrier. After a total of 45 tem-perature cycles as described above, 10

pL of each samPle were removed foranalysis by nondenaturing 107o PAGE.

Finally, to test the suitability of the

multichannel PCR apparatus for evolu-

tionary experiments, a serial transfer

experiment was performed employingall the components of the machine.

Forty-eight reaction chambers in fourrows (every other row: rows B, D, Fand H in the microPlate format) of a

plastic sheet were filled by the piperting robot with the above-described

PCR mixture but with 1'4 PM of DNAtemplate and 0.02 Pg/;tL of salmon

soerm DNA. The 48 remaining cham-

bers in the rows in between (rows A, C,

E and G) were likewise filled withreaction mixture excluding template.

After heat-sealing the reaction vessels,

the sheet was subjected to 25 tempera-

ture cycles on the sample carrier as de-

scribecl above. Subsequently' the pipet-

ting robot removed 3.6 PL from the

first reaction chamber of each row, di-lutecl it in water 1:86000 times by five

serial dilutions in a microplate and then

mixed 15 pL of the final dilution withl5 uL of twice-concentrated PCR mix-ture excluding template, in each of the

l2 chambers of the appropriate row in a

fresh plastic sheet (so that the fillingpattem with lin rows B, D, F and H]and without [in rows A, C' E and G]

temolate was maintained)' Again, 25

temperature cycles were performed af-

ter heat-sealing the reaction vessels,

and the whole procedure of diluting the

products and initiating a further roundof PCR amplification by serial transfer

to the appropriate row of a new plastic

sheet was repeated. Volumes of l0 pLwere removed from each individual re-

action chamber of the three plastic

sheets and analyzed by nondenaturing

107o PAGE.

RESULTS

The biological compatibility of dif-ferent sealing lechniques was tcsted

with RNA-dependent RNA polymerase

of QB phage that replicates a number ofnonphysiological short RNA templates

14; and with Taq DNA PolYmerase thal

amolifles DNA molecuies in a sub-

seouent PCR ( 12,19). Neither the ther-

mostable ltlq DNA Polymerase nor the

nucleic acids were significantly af:

Vol. 18, No. 4 (1995)

fected by the heat or ultrasound-sealingprocedures (results not shown). How-ever, the heterotetrameric and quite un-stable 0p replicase (14,15) showed agreat loss of activity. This was evi-denced by the lower increase in ethid-ium bromide fluorescence in a sub-sequent amplification reaction (Figure3) and confirmed by gel electrophore-sis. This inactivation is attributable todenaturation of QB replicase during thesealing procedure with heat (and notcooling the sample) or with ultrasound.We found, that the most reliable sealingtechnique to minimize the loss of activ-ity of sensitive biological samples wassealing by heat on a Teflon@ block withcooled or frozen samples (curve 3, Fig-ure 3). Based on this technique, a paral-lel sealing device was constructed thatcan handle up to 96 samples on a singleplastic sheet (Figure 2 and Materialsand Methods).

Multiple thermostating blocks wereused to obtain maximal cycling ratesfor a large number of samples. Insteadof employing one block that must beheated and cooled repeatedly, a sepa-rate block exists for each desired tem-perature. Rapid changes among thesetemperatures can be achieved by trans-fering the samples among the differentstations and pressing them onto theblocks with a vacuum svstem. The use

of separate blocks also ensures spatialand temporal homogeneity of the dif-ferent temperatures, whereas devicesthat heat and cool the samples in onemetal block often show substantial dif-ferences in the temperature profiles ofdifferent samples when operating inrapid cycles (22).

Typically obtained PCR temperatureprofiles are given in Figure 4. Figure4A shows a temperature profile mea-sured in the sample carrier; Figure 48shows a profile measured in the plasticreaction vessels. The temperature jumpbegins when the sample carrier isplaced on the appropriate block and thevacuum system is switched on. Thetime required for a temperature adjustment of an aqueous solution in the re-action vessels from 55' toJ2"C is 30 s,

from72" to 93'C is 30 s and decreasingthe temperature from 93" to 55oC takes45 s. In a dynamic measurement at dif-ferent locations all over the sample car-rier, we obtained an average time con-stant (half-time period) of 14.9 s

(standard deviation of 5.3 s) in solutionin comparison to 7 s (standard devia-tion of 1 s) measured in the sample car-rier. A further reduction of the timeconstant (8.5 s in solution, standard de-viation of 1.3 s) can be achieved by us-ing heat-conducting grease between thereaction vessels and the samole carrier.

This proved not to be necessary for thedescribed PCR system.

The time constant for temperaturechanges is about 9 times larger whenthe vacuum system is not used. An arbi-trary temperature profile between thetwo extremes can be simulated by peri-odically switching the vacuum systemon and off to achieve the desired heattransfer rates. This flexibility is impor-tant because it is known that DNA am-plification is critically dependent ontemperature-time profiles (25).

Different profiles were tested fortheir efficiency and accuracy of ampli-fication across the sample area. As atest system, we used a PCR with lowtemplate (0.15 fl\4) and high nontem-plate (0.02 ttC/FL) (salmon sperm)DNA as described in Materials andMethods. Employing the vacuum sys-tem, uniform products were found in anexperiment with 96 separate samples ofeach 30-uL volume distributed uni-formly over the whole sample ca.rrier(Figure 5). An experiment with 960ten-pl samples in 10 plastic sheetsprocessed in parallel, which were ana-lyzed by PAGE of each 12 pooled sam-ples, showed equivalent results.

Finally, a serial transfer experimentwas used to test all components of thedevice in an evolutionary type of pro-cess. A defi4ed pattern of reaction

| :+\lf1

I ii489itt )

64575l

t{i-ltt9ti{ )

\1 -')l

Figure 6. Non-denaturing PAGE analysis of twice serially transferred PCR products. Two representative gels of the products of the last of three subsequentPCRs are shown as described in Materrals and Methods. Both gels contain 2 ;LL of DNA marker V in lane M. Molecular weights are given in bp. The lefi gelshowstheproductsofwellsCltoCl2andDltoDl2l therightgelshowstheproductsofwellsEltoEl2andFltoFl2afterrwoserialtransfers.RowsCandE of the microplate-fbrmatted plastic sheet show no products, while rows D and F do show products, indicating that the initial filling pattern is maintainedthroughout the serial transfer experiment wlthout cross-contamination.

Vol. 18. No. 4 (1995)

BioFeature

chambers was filled by the pipetting ro-bot with PCR mixture including tem-plate and neighboring reaction cham-

bers with PCR mixture excludingtemplate (see Materials and Methods).After two serial transfers including the

dilution of reaction products and main-

taining the filling pattern in the plastic

sheet, we found that all samples were

correctly processed without cross-con-tamination (Figure 6). Table I shows

the times necessary for the differentprocessing steps.

DISCUSSION

Our experiments with PCR mixturesin 96 and up to 960 sample wells show

that our device is suitable for parallelprocessing of this biological reactionunder reproducible and homogeneousconditions. In contrast to other ap-

proaches (10,17), it is optimized forevolutionary types of experiments.This was shown by applying the serialtransfer method (21,24) to a PCR.

Similar experiments simulating condi-tions for Darwinian evolution (5,9,24)are now possible with up to 960 sam-

ples that can be processed in parallel.We also showed that solutions

containing QB RNA polymerase can be

sealed without loss of activity so thatthis isothermal amplification systemcan be employed in our machine as

well. As demonstrated previouslY(4,16), RNA molecules replicated by

QB polymerase can be used as a modelsystem for studying evolution at themolecular level. Moreover, all compo-nents of our system (the plastic materi-als, the setup with three heating blocksand a sample carrier that is accessible

from above) are suitable for performingfluorescence measurements directly inthe reaction chambers. Experiments are

in progress where both PCR and repli-cation by the Q$ enzyme are monitoredusing fluorescent dyes to follow theevolution of the nucleic acid moleculesdirectly invitro.

ACKNOWLEDGMENTS

We are very grateful to W. Simmand his technical group as well as R.

Weise and G. Goldmann for their help

in constructing the device. We wish tothank W. Gotz of the Technical Univer-sity of Miinchen for providing a tem-perature simulation program. We thankS. Volker for technical assistance in the

experiments. This work was supportedfinancially by the Bundesministeriumfiir Forschung and Technologie, Ger-many (F<irderungsnummer 03102484).N.G.W. was partly supported by a doc-toral grant from the Stiftung Sti-pendien-Fonds des Verbandes der

Chemischen Industrie e.V.

REFERENCES

1.Bartel, D.P. and J.W. Szostak. 1993. Isolation

of new ribozymes from a large pool of random

seouences. Science 261 : 141 1-1418.

2.Beaudry, A.A. and G.F. Joyce' 1992. Directed

evolution of an RNA enzyme. Science

257:635-641.3.Berzal-Herranz, A., S. Joseph and J.M.

Burke. 1992. 1n vltro selection of active hair-pin ribozymes by sequential RNA-catalyzedcleavage and ligation reactions. Genes Dev.

6:129-134.4.Biebricher, C.K. 1988. Replication and evolu-

tion of short-chained RNA species replicatedby Qp replicase. Cold Spring Harbor Symp.

Quant. Biol. 5 2 :299 -306.

5.Breaker, R.R. and G.F. Joyce. i994. Emer-gence of a replicating species from an in vitroRNAevolution reaction. Proc. Natl. Acad. Sci.

USA 91:6093-6097.

6.Eigen, M. and W.C. Gardiner, Jr. 1984. Evo-lutionary molecular engineering based on

RNA replication. Pure Appl. Chem. 56:967-978.

7.Eigen, M. 1986. The physics of macromolecu-1ar evolution. Chem. Scripta 268:13-26.

8.Eigen, M. 1990. German patent application P

4022'7 92.8, P 40427 93.6 and P 40221 9 4 -4.

g.Ellington, A.D. and J.W. Szostak. 1992. Se-

lection in r.,ltro of single-stranded DNA mole-

cules that fold into specific ligand-bindingstructures. Nature J55: 850-852.

10.Garner, H.R., B. Armstrong and D.M' Lin-inger. 1993. High-throughput PCR. BioTech-niques 14:1 12-1 15.

1 l.Giordano, P,C., R. Fodde, M. Losekoot and

L.F. Bernini. 1989. Design of aprogrammableautomatic apparatus for the DNA polymerase

chain reaction. Technique I :16-20.12.Innis. M'A', D'H' Gelfand, J'J' Sninsky and

T.J. White. 1990. PCR Protocols: A Guide to

Methods and Applications. Academic Press,

San Diego.l3.Jenison, R.D., S.C. Gill, A. Pardi and B.

Polisky. 1994. High resolution moiecular dis-

crimination by RNA. Science 26J:1425-1429.14.Kamen, R. 1970. Characterization of the

subunits of Qp replicase. Nature 228:52'7 -533'

15.Kondo, M., R. Galleraniand C. Weissmann.1 970. Subunit structure ofQp replicase. Nature

228:525-521.16.Krarner, F,R., D.R' Mills, P.E. Cole, T.

Nishihara and S. Spiegelman. 1974. Evolu-lion in vitro'. Sequence and phenotype of a

mutant RNA resistant to ethidium bromide.

J. Mol. Biol. 89:719-136.17.Meier-Ewert. S., E. Meier, A. Akmadi' J.

Curtis and II. Lehrach. 1993. An automated

approach to generating expressed sequence

catalogues. Nature J61 :375-376.l8.Ponder, S., T. Pappas, W.J. Wells' E. Kam,

A. Mckay, B.S. Keenan and C.S. Watson.1990. An inexpensive automated method toperform the polymerase chain reaction. Tech-

nioue 2:2O2-205.lg.Saiki, R.K., D.H. Gelfand, S. Stoffel, S.J.

Scharf, R. Higuchi, G.T. Horn, K.B' Mullisand H.A. Erlich. 1988. Primer-directed enzy-

matic amolification of DNA with thermostableDNA polymerase. Science 239:487 -494

20.Schober, A., M. Thiirk and M. Eigen. 1993'

Optimization by hierarchical mutant produc-

tion. Biol. Cybernetics 69 :493 -501.

2l.Spiegelman, S., I. Haruna, I.B. Holland' G.Beaudreau and D. Mills. 1965. The synthesis

of a self-propagating and infectious nucleic

acid with a purified enzyme. Proc Natl. Acad.

Sci. USA54:919-927.Z2.Taylor, G.R. 1991. Polymerase chain reac-

tion: Basic principles and automation, p. 1.1n

M.J. McPherson, P. Quirke and G.R. Taylor(Eds.), PCR A Practical Approach. OxfordUniversity Press, Oxford.

23.Tuerk, C. and L. Gold. 1990. Slstematic

evolution of ligands by exponential enrich-

ment: RNA ligands to bacteriophage T4 DNApolymerase. Science 249:505-5 I 0.

24.Walter, N.G. and G' Strunk. 1994. Strand

displacement amplification as in vitro model

for rolling-circle replication: Deletion forma-

tion and evolution during serial transfer. Proc.

Natl. Acad. Sci. USA 9/:7937-7941.25.Wittwer, C.T. and D.J. Garling. 1991. Rapid

cycle DNA amplification: Time and tempera-

ture optimization. BioTechniques I 0:7 6-83.

Received 9 September 1993; accepted16 December 1994.

Address corresPondence to:

Andreas Schober

D e partment of B ioc hemical K ine tic s

Am Fassberg

D - 37077 Giittingen, G ermanylnternet: aschobe@ gwdgv 1.dnet. gwdg.de and

aschobe@ imb-ienrt.de

658 BioTechniquesVol. 18, No. 4 (1995)