Journalof Crystal Growth 122 (1992) 176—180 ou~.*~o. CRYSTALNorth-Holland GROWT H

Microbatch crystallizationunderoil — a new techniqueallowingmany small-volumecrystallizationtrials

Naomi E. Chayen

Blackett Laboratory, Imperial College, London SW7 2BZ, UK

Patrick D. Shaw Stewart

Douglas InstrumentsLtd., 25J, ThamesHouse, 140 BatterseaPark Road,LondonSWJJ4NB, UK

and

David M. BlowBlackett Laboratory, Imperial College, LondonSW72BZ, UK

An approachto rapid protein crystallizationusing very small samplesis described.A computercontrolled microdispenserisusedto makecrystallizationsamplesasmicrobatchdropletsunderoil. Samplesof 1—2 ~l are dispensedready-mixedandwith goodprecision.The samplesare protectedfrom evaporation,contaminationand physical shockby the oil. When favourableconditionsfor crystallization have been foundusing one modeof the system, the size and quantity of crystalsare optimized by a secondprogramwhich generatesa setof conditionsthroughoutthe areaof interest. Crystals of diffraction size and quality have beengrown in 1 pA drops.

1. Introduction (e) automatic execution in a quick and simpleway;

Whentrying to crystallizea protein, the initial (f) flexible changesof operationmodefor differ-aim of the crystallizer is to screenrapidly many ent typesof survey.conditionswith a minimum of labour, and usingas little proteinas possible.

To easethe laboriousandtime consumingtaskof crystallization,severalautomaticsystemshave 2. Crystallizationunder oilbeenintroduced[1—5]— but all previousautoma-tion proceduresare directedtowardvapourdiffu- To reducethe consumptionof material, verysion methods.We are presentinghere an auto- small samplesmust be dispensed.This leads tomated techniquefor batch crystallization which two major problems:(a) evaporation;(b) maccu-fulfils the following requirements: racy of dispensing.(a) Dispensingof very small samples; The problem of evaporationis solved by dis-(b) high levelof accuracy; pensing and incubating the crystallization sam-(c) rapid screeningto determinesolubility prop- pies underoil. Samplesaredispensedasdroplets

erties; into the wells of tissueculture plates(of the kind(d) homing in on conditionsthat producecrystal- known as Terazaki plates, Sterilin, UK) where

lization; they are isolated by a layer of oil (fig. 1). The

0022-0248/92/$05.00© 1992 — ElsevierSciencePublishersB.V. All rights reserved

N.E. Chayen etal. / Microbatch crystallizationunderoil 177

dropletscanbe dispensedunderoil manuallyor torized syringes deliversaccuratevolumesat theautomatically. tip. Delivery from a fine tip under oil cleanly

High accuracyis achievedby usinga microdis- removesthe deliveredvolume as the tip is drawnpensercomprising a bank of Hamilton syringes out from the oil, and eliminates“carry-over”.driven by steppermotors under computercon- Testsshow that 1 ~l dropletscan be dispensedtrol, to dispenseliquid underoil through a fine with a standarddeviationof less than 0.02 pi.multi-bore tip [6]. The highaccuracyof the mo- The microtip is automaticallymoved using a

~.* ~

Fig. 1. 24 dropletsof a coloured solutiondispensedunderoil in a Terazakiplate. Theactualsizeof the plateis 55 X 80 mm.

178 N.E. Chayen etal. / Microbatchcrystallizationunderoil

robot arm (whichmovesvertically) with a moving due to small temperaturechangeswhich causetable(whichmoveshorizontallyin two directions). condensation). Thisdoesnot happenin the mi-Becausethe robot arm does not move horizon- crobatchunderoil.tally, flexing of the tubing is kept to a minimum, (3) Batchmethod. Thesampleis mixed fullywhenwhich improvesthe accuracyof dispensing. it is dispensed.In the microbatch,concentration

We havefound that this methodof crystalliza- gradients and convectionflows are kept to ation hasthe following advantages: minimum, and there is less tendency to form(1) Low protein consumption.Becauseevapora- skins. The volume andcompositionof a droplet istion is reduced,dispensingunderoil enablesthe known exactly, which is useful for theoreticaluseof 0.5—0.7pi of protein,with a final droplet studiesof the crystallization process.This con-volume of 1—2 /Ll. Crystals produced in such trastswith the vapour diffusion methods wheresmall dropletsare of the sameor better quality the physical complicationof the methodsmeansfor diffraction, and of equalor greatersize than that the precise conditionsof crystallizationarethose grown in much larger drops (10 ~l) by notknown.meansof othermethods. (4) Time saving method.Methods for crystalliz-(2) Very stable crystals. In several cases(e.g. ing smallsamplesusuallyinvolve carefulmanipu-reversetranscriptase[7]), crystalsgrewin hanging lation. This technique is mechanically simplerdrops but dissolvedseveral days later (probably than bothvapourdiffusion anddialysis.

IN

.9 ~‘

Fig. 2. Crystalsof carboxypeptidaseG2 in a 2 pA droplet underoil. Crystalsweregrown from 1.6M ammoniumsulphateand8mMzinc sulphate.Theactualsizeof the largest crystalis 700x 100x90~sm.

N.E. Chayenetal. / Microbatchcrystallizationunderoil 179

(5) Protectionof sample.Drops arebuoyedup by and transferredto a depressionglass,containingthe viscous oil, they are therefore easily trans- moreharvestingsolution,usinga micropipette.portable. The layer of oil also protects thedropletsfrom dustandcontamination.(6) Reducedconsumptionof disposables. Linbro 3. Rapid screeningfor crystallization conditionstrays and coverslips are not required;pipettesandchemicalsareused in small quantities. In an alternative configuration,the microdis-(7) Compact.Theuseof the smallTerazaki plates pensercan be usedto carryout largenumbersofsaveson storagespace. crystallization trials usingpre-mixed standard

So far, in all casestried, a crystal form oh- stocksolutionsof precipitating agents andbuffers.tamed by meansof vapour diffusion has been In this case,a 2-bore microtip is usedandthegrownby microbatch with onlyslight modification experimentis carriedout automatically usingtheof the vapourdiffusion conditions.However, on robot arm with two motorized syringes (fig. 3).more than one occasionthis method has given One motorized syringe is used to aspirate typi-crystal forms which could not be obtained by cally 1 pi of stock solution from the wells of ameansof vapourdiffusion, e.g. reverse transcrip- microtitreplate and transfer it into wells of atase[8], carboxypeptidaseG2 (fig. 2) andothers. Terazakiplate in sequence.After eachtransferof

Harvestingthe crystalsdoes not cause prob- solution, the tip is rinsed by aspirating anddis-lems: The dropletsare first enlargedby adding pensing a few microlitres while remainingin aabout 30 ~l of a harvestingsolution (which con- well containingdistilled water. A secondmotor-tainsprecipitantat a slightlyhigherconcentration ized syringe isthen used to dispensetypically 1thanthat in the droplet),crystalsaredetached,if ~tl of proteinsolution into eachof the crystalliza-necessary,by stirringwith a glassfibre, drawnup tion wells of the Terazakiplate.

Fig. 3. A view of theapparatuscomprisingthe liquid handlingsystem,steppermotor drivers.PC, printer.microtip. robotic arm andthe moving table. Two setsof wells are placedon the moving table: a plate with microtitre strips containing stocksolutionsforcrystallizationand a Terazakiplate. The microtip is in the act of dispensinga solution into the Terazakiplate, using the ASP

program.A similarconfigurationis usedwith X-Step,but with only onesetof wells,and a four-boremicrotip.

180 N.E. Chayenetal. / Microbatchcrystallizationunderoil

Both stock solutions and crystallizationdrops using automatic dispensing and a minimumare situated underoil. Thesestock solutionsare amountof protein.The systemdescribedaboveispreparedin approximately20 ml volumes, and routinely used in the following order: When astoredin a freezer.Theymaybe chosento scana new protein is to be crystallized,a standardizedrangeof conditionsin a systematicway, or they initial screenis performedfirst with the use ofmay contain precipitants,additives and buffers the ASP program. In this way, conditions thatchosenat randomusing a randomnumbergener- look favourable for crystallizationare found. Inator, in the style of Carterand Carter [91.Forty the next step, the X-Stepprogram[6] is usedtosolutions can be screenedin one run using ap- fine-tune the conditionsandproduce largecrys-proximately 400 j~gof protein. This run takes tals for X-ray diffraction.approximately1 h but need not be supervised. Severalnewforms of crystalswereobtainedbyThe programis namedASP (AutomatedSample using the systemin this way. Among them arePreparation), and is written in ordinary sen- carboxypeptidaseG2 (fig. 2), reversetranscrip-tences,which makesit easyfor anyoneto use.In tase [81and a protein which regulatesthe ami-the samemanner,up to forty additives can be daseoperonof Pseudomonasa?ruginosa[10].automatically transferredto crystallizationtrialswhich areotherwiseidentical.

Acknowledgements

We thank PeterBaldock for programmingand4. Optimizationof conditions

JonathanRadcliffe for practical assistance.Thiswork has been supported by the Science andTo generatea set of crystallizationconditions

evenlyspacedthroughoutthe areaof interest,the EngineeringResearchCouncil (GR/G32915)andthe UK Medical ResearchCouncil.programX-Step previously describedby Chayen

et al. [6] is used.Two parametersarevariedin asingle run, and a four-bore microtip is used.In Referencessuccessiveruns conditions are fine-tuned, untilthe bestcrystalsareproduced.This programdis- [1]H.A. Kelders,K.H. Kalk, P. Gros and W.G.J. Hol, Pro-

plays a four by six arrayof cells. The conditions tein Eng. (1987)301.

in the top left and bottom right cells can be [2] M.J. Cox and P.C. Weber, J. Crystal Growth 90 (1988)318.

edited, then all other conditions are automati-[3] K.B. Ward, A.M. Perozzo and W.M. Zuk, J. Crystal

cally determinedby the programby linear inter- Growth 90 (1988)325.

polation of concentrationsor pH values. Each [4] D.W. Morris, C.Y. Kim and A. McPherson,BioTech-

cell correspondsto a single microbatchdroplet. niques7 (1989)522.

During execution the droplets from the entire [5]B. Rubin, J. TalafousandD. Larson, J. Crystal Growth110 (1991)156.

arrayareautomaticallydispensedin sequence. [6] N.E. Chayen,P.D. ShawStewart,D.L. MaederandD.M.

Blow, J. App!. Cryst. 23 (1990)297.[7] L.F. Lloyd, personalcommunication.

5. Summary [8] L.F. Lloyd, P. Brick, L. Mei-Zhen, N.E. Chayen andD.M. Blow, J. Mol. Biol. 217 (1991) 19.

[9] C.W. Carter,Jr. and C.W. Carter, J. Biol. Chcm. 254Ourapproachto the crystallizationof proteins (1979) 12219.

startswith the rapid determinationof the ranges [10] S.A. Wilson, N.E. Chayen,A.M. Hemmings,R.E. Drew

of conditionswhereprecipitationcanbe achieved, and LII. Pearl, J. Mol. Biol. 222 (1991)869.