Proc.Natl.Acad.Sci.USA
Vol.94,pp.8945–8947,August1997
Applied Biological Sciences
Whole genome analysis:Experimental access to all genome quenced gments through larger-scale efficient
oligonucleotide synthesis and PCR
D EVAL A.L ASHKARI*†,J OHN H.M C C USKER‡,AND R ONALD W.D AVIS*§
*Departments of Genetics and Biochemistry,Beckman Center,Stanford University,Stanford,CA94305;and‡Department of Microbiology,3020Duke University Medical Center,Durham,NC27710
Contributed by Ronald W.Davis,May20,1997
ABSTRACT The recent ability to quence whole genomes allows ready access to all genetic material.
The approaches outlined here allow automated analysis of quence for the synthesis of optimal primers in an automated multiplex oligonucleotide synthesizer(AMOS).The efficiency is such that all ORFs for an organism can be amplified by PCR.The resulting amplicons can be ud directly in the construction of DNA arrays or can be cloned for a large variety of functional analys.The tools allow a replacement of single-gene analysis with a highly efficient whole-genome analysis.
The genome quencing projects have generated and will continue to generate enormous amounts of quence data.The genomes of Saccharomyces cerevisiae,Escherichia coli,Hae-mophilus influenzae(1),Mycoplasma genitalium(2),and Meth-anococcus jannaschii(3)have been completely quenced. Other model organisms have had substantial portions of their genomes quenced as well,including the nematode Caeno-rhabditis elegans(4)and the small flowering plant Arabidopsis thaliana(5).This massive and increasing amount of quence information allows the development of novel experimental approaches to identify gene function.
One standard u of genome quence data is to attempt to identify the functions of predicted open reading frames (ORFs)within the genome by comparison to genes of known function.Such a comparative analysis of all ORFs to existing quence data is fast,simple,and requires no experimentation and is therefore a reasonable first step.While finding quence homologies͞motifs is
not a substitute for experimentation, noting the prence of quence homology and͞or quence motifs can be a uful first step in finding interesting genes,in designing experiments and,in some cas,predicting function. However,this type of analysis is frequently uninformative.For example,over one-half of new ORFs visiae have no known function(6).If this is the ca in a well studied organism such as yeast,the problem will be even wor in organisms that are less well studied or less manipulable.A large,experimen-tally determined gene function databa would make homol-ogy͞motif arches much more uful.
Experimental analysis must be performed to thoroughly understand the biological function of a gene product.Scaling up from classical‘‘cottage industry’’one-gene-oriented ap-proaches to whole-genome analysis would be very expensive and laborious.It is clear that novel strategies are necessary to efficiently pursue the next pha of the genome projects—whole-genome experimental analysis to explore gene expres-sion,gene product function,and other genome functions. Model organisms,such visiae,will be extremely important in the development of novel whole-genome analysis techniques and,subquently,in improving our understanding of other more complex and less manipulable organisms. The genome quence can be systematically ud as a tool to understand ORFs,gene product function,and other ge-nome regions.Toward this end,a directed strategy has bee
n developed for exploiting quence information as a means of providing information about biological function(Fig.1).Ef-forts have been directed toward the amplification of each predicted ORF or any other region of the genome ranging from a few ba pairs to veral kiloba pairs.There are many us for the amplicons—they can be cloned into standard vectors or specialized expression vectors,or can be cloned into other specialized vectors such as tho ud for two-hybrid analysis.The amplicons can also be ud directly by,for example,arraying onto glass for expression analysis,for DNA binding assays,or for any direct DNA assay(7).As a pilot study,synthetic primers were made on the96-well automated multiplex oligonucleotide synthesizer(AMOS)instrument(8) (Fig.2).The oligonucleotides were ud to amplify each ORF on yeast chromosome V.The current version of this instrument can synthesize three plates of96oligonucleotides each(25bas)in an8-hr day.The amplification of the entire t of PCR products was then analyzed by gel electrophoresis (Fig.3).Successful amplification of the proper length product on the first attempt was95%.This project demonstrates that one can go directly from quence information to biological analysis in a truly automated,totally directed manner. The amplicons can be incorporated directly in arrays or the amplicons can be cloned.If the amplicons are to be cloned, novel quences can be incorporated at the5Јend of the oligonucleotide to facilitate cloning.One potential problem with cloning PCR products is that the cloned amplicons may contain quence alterations that diminish their utility.One option w
ould be to requence each individual amplicon. However,this is expensive,inefficient,and time consuming.A faster,more cost-effective,and more accurate approach is to apply comparative quencing by denaturing HPLC(9).This method is capable of detecting a single ba change in a2-kb heteroduplex.Longer amplicons can be analyzed by u of appropriate restriction fragments.If any change is detected in a clone,an alternate clone of the same region can be analyzed. Modifying the system to allow high throughput analysis by denaturing HPLC is also relatively simple and straightforward. If amplicons are ud directly on arrays without cloning,it is important to note that,even if single PCR product bands are obrved on gels,the PCR products will be contaminated with various amounts of other quences.This contamination has the potential to affect the results in,for example,expression
The publication costs of this article were defrayed in part by page charge payment.This article must therefore be hereby marked‘‘advertiment’’in accordance with18U.S.C.§1734solely to indicate this fact.
©1997by The National Academy of Sciences0027-8424͞97͞948945-3$2.00͞0 PNAS is available online at http:͞͞
†Prent address:Synteni,Inc.,6519Dumbarton Circle,Fremont,CA 94555.
§To whom reprint requests should be addresd at:Department of Biochemistry,Beckman Center,B400,Stanford University,Stanford, CA94305-5307.e-mail:*********************.edu.
8945
analysis.On the other hand,direct u of the amplicons is much less labor intensive and greatly decreas the occurrence of mistakes in clone identification,a ubiquitous problem associated with large clone t archiving and retrieving.
Any large-scale effort to capture each ORF within a genome must rely on automation if cost is to be minimized while efficiency is maximized.Toward that end,primers targeting ORFs were designed automatically using simple new scripts and existing primer lection software.The script-lected primer quences were directly read by the high-throughput synthesizer and the forward and rever primers were synthe-sized in parate plates in corresponding wells to facilitate automated pipetting and PCR amplifications.Each of the resulting PCR products,generated with minimum labor,con-tains a known,unique ORF.
Large-scale genome analysis projects are dependent on newly emerging technologies to make the studies practical and economically feasible.For example,the cost of the primers,a significant issue in
the past,has been reduced dramatically to make feasible this and other projects that require tens of thousands of oligonucleotides.Other methods of high-throughput analysis are also vital to the success of functional analysis projects,such as microarraying and oligonucleotide chip methods (10–14).
Changes in attitude are also required.One of the major costs of commercial oligonucleotides is extensive quality control such that virtually 100%of the supplied oligonucleotides are successfully synthesized and work for their intended purpo.
Considerable cost reduction can be obtained by simply de-creasing the expected successful synthesis rate to 95–97%.One can then achieve faster and cheaper whole genome coverage by simply adding a single quality control at the end of the experiment and batching the failures for resynthesis.
The directed nature of the amplicon approach is of clear advantage.The quence of each ORF is analyzed automati-cally,and unique specific primers are made to target each ORF.Thus,there is relatively little time or labor involved—for example,no random cloning and subquent screening is required becau each product is known.In the test system,primers for 240ORFs from chromosome
V were systematically synthesized,beginning from the left arm and continuing through to the right arm.At no point was there any manual analysis of quence information to generate the collection.In many ways,now that the quence is known,there is no need for the rearcher to examine it.
The amplicons can be arrayed and expression analysis can be done on all arrayed ORFs with a single hybridization (10).Tho ORFs that display significant differential expression patterns under a given lection are easily identified without the laborious task of arching for and then quencing a clone.Once scaled up,the procedure provides even greater returns on effort,becau a single hybridization will ultimately provide a ‘‘snapshot’’of the expression of all genes in the yeast genome.Thus,the limiting factor in whole genome analysis will not be the analysis process itlf,but will instead be the ability of rearchers to design and carry out experimental lections.Current expression and genetic analysis technologies are geared toward the analysis of single genes and are ill suited to analyze numerous genes under many conditions.Additional difficulties with current technologies include:the effort and expen required to analyze expression and make mutants,the potential duplication of effort if done by different laboratories,and the possibility of conflicting results obtained from differ-ent laboratories.In contrast,whole genome analysis not only is more efficient,it also provides data of much higher quality;all genes are assayed and compared in parallel under
exactly
F I
G .1.Overview of systematic method for isolating individual genes.Sequence information is obtained automatically from quence databas.The data are input into primer lection software specifi-cally designed to target ORFs as designated by databa annotations.The output file containing the primer information is directly read by a high-throughput oligonucleotide synthesizer,which makes the oli-gonucleotides in 96-well plates (AMOS,automated multiplex oligo-nucleotide synthesizer).The forward and rever primers are synthe-sized in the same location on parate plates to facilitate the down-stream handling of primers.The amplicons are generated by PCR in 96-well plates as
well.
F I
G .2.Overall approach for using databa of a genome to direct biological analysis.The synthesis of the 6,000ORFs (orfs)for each gene visiae can be ud in many applications utilizing both cloning and microarraying technology.
8946Applied Biological Sciences:Lashkari et al .Proc.Natl.Acad.Sci.USA 94(1997)
the same conditions.In addition,amplicons have many appli-cations beyond gene expression.For example,one recent approach is to incorporate a unique DNA quence tag,synthesized as part of each gene specific primer,during amplification.The tags or molecular bar codes,when reintro-duced into the organism as a gene deletion or as a gene clone,can be ud much more efficiently than individual mutations or clones becau pools of tagged mutants or transformants can be analyzed in parallel.This parallel analysis is possible becau the tags are readily and quantitatively amplified even in complex mixtures of tags (13).
The ORF genome arrays and oligonucleotide tagged libraries can be ud for many applications.A
ny conventional lection applied to a library that gives discrete or multiple products can u the technologies for a simple direct read-out.The include screens and lections for mutant comple-mentation,overexpression suppression (15,16),cond-site suppressors,synthetic lethality,drug target overexpression (17),two-hybrid screens (18),genome mismatch scanning (19),or recombination mapping.
The genome projects have provided rearchers with a vast amount of information.The data must be ud efficiently and systematically to gain a truly comprehensive understand-ing of gene function and,more broadly,of the entire genome which can then be applied to other organisms.Such global approaches are esntial if we are to gain an understanding of the living cell.This understanding should come from the viewpoint of the integration of complex regulatory networks,the individual roles and interactions of thousands of functional gene products,and the effect of environmental changes on both gene regulatory networks and the roles of all gene products.The time has come to switch from the analysis of a single gene to the analysis of the whole genome.
Support was provided by National Institutes of Health Grants R37H60198and P01H600205.
1.Fleischmann,R.D.,Adams,M.D.,White,O.,Clayton,R.A.,Kirkness,E.F.,et al.(1995)Science 269,496–51
2.
2.Frar,C.M.,Gocayne,J.D.,White,O.,Adams,M.D.,Clayton,R.A.,et al.(1995)Science 270,397–40
3.
3.Bult,C.J.,White,O.,Oln,G.J.,Zhou,L.,Fleischmann,R.D.,et al.(1996)Science 273,1058–1073.
4.
Sulston,J.,Du,Z.,Thomas,K.,Wilson,R.,Hillier,L.,Staden,R.,Halloran,N.,Green,P.,Thierry-Mieg,J.,Qiu,L.,Dear,S.,Coulson,A.,Craxton,M.,Durbin,R.,Berks,M.,Metzstein,M.,Hawkins,T.,Ainscough,R.&Waterston,R.(1992)Nature (London)356,37–41.
5.Newman,T.,de Bruijn,F.J.,Green,P.,Keegstra,K.,Kende,H.,et al.(1994)Plant Physiol.106,1241–1255.
6.Oliver,S.(1996)Nature (London)379,597–600.
7.Lashkari, D. A.(1996)Ph.D.disrtation (Stanford Univ.,Stanford,CA).
8.Lashkari,D.A.,Hunicke-Smith,S.P.,Norgren,R.M.,Davis,R.W.&Brennan,T.(1995)Proc.Natl.Acad.Sci.USA 92,7912–7915.
9.Oefner,P.J.&Underhill,P.A.(1995)Am.J.Hum.Genet.57,A266.
10.Schena,M.,Shalon,D.,Davis,R.W.&Brown,P.O.(1995)Science 270,467–470.
11.Fodor,S.P.,Read,J.L.,Pirrung,M.C.,Stryer,L.,Lu,A.T.&Solas,D.(1991)Science 251,767–773.
12.Chee,M.,Yang,R.,Hubbell,E.,Berno,A.,Huang,X.C.,Stern,D.,Winkler,J.,Lockhart,D.J.,Morris,M.S.&Fodor,S.P.(1996)Science 274,610–614.
13.Shoemaker,D.D.,Lashkari,D.A.,Morris,D.,Mittmann,M.&Davis,R.W.(1996)Nat.Genet.14,450–456.
14.Smith,V.,Chou,K.,Lashkari,D.,Botstein,D.&Brown,P.O.(1996)Science 274,2069–2074.
15.Magdolen,V.,Drubin,D.G.,Mages,G.&Bandlow,W.(1993)FEBS Lett.316,41–47.
16.Ramer,S.W.,Elledge,S.J.&Davis,R.W.(1992)Proc.Natl.Acad.Sci.USA 89,11589–11593.
17.Rine,J.,Hann,W.,Hardeman,E.&Davis,R.W.(1983)Proc.Natl.Acad.Sci.USA 80,6750–6754.
18.Fields,S.&Song,O.(1989)Nature (London)340,245–246.19.
Nelson,S.F.,McCusker,J.H.,Sander,M.A.,Kee,Y.,Modrich,P.&Brown,P.O.(1994)Nat.Genet.4,
11–18.
F I
G .3.Gel image of amplifications.Using the method described in Fig.1,amplicons were generated for ORFs visiae chromosome V.One plate of 96amplification reactions is shown.
Applied Biological Sciences:Lashkari et al .Proc.Natl.Acad.Sci.USA 94(1997)8947
