Constructing andScreening Normalized cDNALibraries MARCELO BENTO SOARES ANDMARIADEFATIMA BONALDO A CDNAlibraryis consldered normalized whenthe frequency of eachCDNAin the libr¿ry is equaly represented reg¿rdless of whetherthe cDNAis derivedfroma rareor a frequently occurring ñRNAspecies. Theffequencies of all clones aretherefore withinthesameor, der of magnltude rángein a typic¿lnormalized library(see,e-9.,Soares et al. 1994; Bonaldoet al. 1996).This chapterprovides detailedprotocols for construction and normaizatlonof oligo(dTfprimed, directionally clonedcDNAlibraries in ph¿gemid vec (Okay¿ma tors. l\4odiflcations in the existingmethodsfor libraryconstruction and Berg 1982;Gublerand Hoffman1983;D'Alessio et al. 1987)andsolutions for the problems thataremostcomr¡only observed in cDNAlibraries arediscussed. Theseproblems are(1) the highfrequency of cloneswithsmallinserts, someof whichcontainnothingexcept the poly(A) (2) the presence tailsof mRNAs, of a longpoly(A) tail at the 3' end of cDNAs, whichnrakesit difficultto obtain3'terminal sequences, and (3) the occurrence of chir¡ericCDNAclones.In addition,a protocol for screenlng cDNAlibrariesby coony hybridization is presented.

49

7

q

50

q q

CDNALibrades andscrceninsNormal¡zed Chaptei2 ¡ Constructing

Primaty UsesandApplications forcDNALibraries

a cDNA librarl requireconrrucling and screenhrg bur ins!cadmay simplyinvolvesearchinglhe pubhonologous!o !he onc lic databases ior sequ€nces

AND EXPRESSION STUDYOF GENESTRUCTURE One of rhe {irst stepstoward cloning a gene oi inlerest typically involves construc¡ing a CDNA library f¡om a tissue or cell type likely to expressit and screening lhe library eirher by hybridizarion (Young and Davis l983a,b; Huynh et al- 1985, Sa¡¡brooket al. 1989)or by PCR (FrohmaneI al. 1988;Saiki et al. 1988;complon 1990;Frohman 1990). AfIer a cDNA clone is isolatedand se quenced lor verificalion tha! it correspondsto the desiredgene, it is ¡hen used as a probe ¡o screena genomic library and lhc¡eby to isolate ils cogn¿¡e gene.It is alsousedfor northernan¡lysisio study r L rp d ¡ l ( r no f e r f r e . ' o n n d r i t e . ( n cr e | . y p e . .r i ' sues,and stagesof developmen¡. The comparativeanalysjs of lhe CDNA and ils gcnomic cione r€veals,al l€asl in corresponding p ¿ - r . r h e \ r r u L t u r eo L " e g e n . . H L ' w e v e l .i " e cDNA clones often representincomplele producls of ¡everse transcripiion, characterization of the enlire transcript frequently involves additional sc¡eeningsof CDNAlibraries to isolate overlapping cDNAs.Alte¡natively, the m¡sing CDNAlragments can be obrainedby RT-PCRapproaches 15' and 3' RACE) with specificoligonucleoddesdesignedon lhe basisol sequenceinformalion derived irom the truncat€dCDNA (Complon 1990;Frohman 1990; seealsoFanningand Gibbs1997).

OF GENESBY PARTIAL RAPIDIDENTIFICATION SEQUENCING OFcDNAs An alternative approach lor gene discovery has emergedin the pas¡yearsthat is expec¡edto lacilirate quite dramalically ihe rvork involved in cloning a gene lAdamr er at. 1991, 1992, l993a,b) Khan e¡ al. 1992; Okubo e¡ a]. 1992; Malsubara and okubo l99l; M¿roba eI al- 1994; Okubo 1994).ln this approach,large numbersoi di¡ec: tionally cloned cDNAs arc randomly chosen from librariesand sequenced i¡om onc or both (l' and ' 5 ) ends.Their nucleotide sequences,and rranslaiions thereof, are comparedwirh those availablein public da¡abasesior idcnriiication of homologies lhat occasionally eiLhe¡ reveal or suggest ¡heir fLrncaion.li is anlicipatedlhat once most genesare identified and lhei¡ cloneEbecome available, lhc p¡ocessoi isolaling a CDNA clone may no long€r

Overview of the Methodsfor Double-stranded Synthesizing cDNA hasbeen madesincerh€ firsl meI:r' Much progress cDNAs was repo¡led(Ehlraliadise: od for cloning "t.

lqTo: RorgFc- ¿-d Md.h

la-o,. l.l he.p -"

1

proiocols, the hairpin-loop struclure narur¿,. iorrneda¡ the l' end of lirsl-s!¡andCDNArs u-.:: ¡o selfprime lhe synthesisof a secoodstrandaj:.: hydroiysis oI thc RNA remplate.The end prodü.i:: DNA ivirh the 5' end :: a hajrpindouble'srranded ' ' , . I R \ A e q J e ' r , ei n r h , f o ' n o f J j n . i . .r'oeo loop ot vdfldblr n/e dnd loc"..on T' rcsul¡ing cDNAs are then trealcd wilh nucleasej r o d r E € .rrh ( ' r n g e - . r . r r r d eDd N A l u . l J r e p : ' : . invariably removes portions ol thc cDNA i:.' r e . p o n d i n rE. , o J r n g o r . ' - r . o n . o d r n ;r . ; ' ¡ rhe mRNA. TJ {ir,,rrvcnr I| F need lor J,; 'r,on . ' n u c l e d \ es l . | " n d e r J l \ l o 8 | ) d e \F l o f ' i€ d l t F r n ¿ te: \p r o L oJ,l I n { h r . h ¿ h o r n o p o l v ' n e : . . is addedto the l' end of !]refirf's¡rand cD\_A=:: ' ' r ' e dl o : : . ' o n p l e | |e n l . r ! u l r t ñ n u . e o r i d e second-strand syn¡hesis. I n r l _ e icr ¿ s 5 rp. " p e r .O k d \ d m rr ' l d B e r ' : : : , l ( \ . f i 1 ( dr . l e l e g r n r- n F . h o dt h d r I n i c d " . , r CDNAbr -,li - . idea of synrhesisof second'strand primed nick translation.In this method, ,l : . I n R NA i r e v . r ' e | | d n \ ( , i b e d i I . o , D N A ' ) . r d i l ( dc l L ' n r nvSc c r ñ r 5 p r j ¡ ¡ e - .' 2 ' " l r n l ' . n r n r r \ l r g d L r l,oi r h ( D \ A : R N A l \ b r r d J : . . : . to ensüretheir circularizadon, and (l) lh. ::::: : s , r , n d i \ . t n r h e . i / e d b v r . : n L ir c o r ' b ' : : RN¿seIl, ¿. roli DNA polymerasc I, ¡nd E ji.. -l . \!::. -ri. ligaseto replacelhe RNA iD lhe hyl.Jrids : R t \ r \ e H r \ ¡ n e n d u r i l , o n u . l eF". p e . i t : . : . : . R N A . r r ¿ . l do f d D N A R N A h y l , ' J l 0 7 t l . T h i \ e n d o r ; b o n J ' l e " rI n ( f i u d u .), - . - . . : smallgapson ¡he RNA frand, rvhichafe.'::::::: { e r u n L c l ( , i lr}: . r . r ' \ I \ by rhe s O N A I o l y m e r d \ eI ( n r c k U - n 1 , , : o D N As r r d n d i , \ y n h e \ i / e d b \r h e o u l , n . ( : . , , Guble¡and Hoiiman (1981) describ.: = , :i: -:: m o d i l i c ¿ . ' o on i t h r O k d ) " r r " r ' r B r

q CI

ct ! l ¡

G ll

a -

I JI ttl I

a

I I rl I

I I I I

I I

a I

of CDNAL¡braries 51 cloningvectoEfor construction

prorocol that combjncsthe classicoligo(dT)-primed íir5t s!¡and synthesiswilh the novel RNaseH-DNA poiymerasei-mediated second'strandsynihesis. Homopolymericrails a¡e added to the ends of both rhe cloning vector (dCTP tail) and the cDNAs,. (dcTP tail) and then ligated.The advantageof this procedure is lbat neither the elaborate Okayama and Berg vector-p¡imer system nor lhe classic by nucleaseSl js used. htri¡pin'loopcleavage A limitation of úe RNaseIl DNA polym€raseimediated method ior second-strandsynthesis is ¡bat ii results in the loss ol sequ€nceinlo¡mation from the 5' end oi thc mRNA template (D'Al€ssio ¿nd Gerard 1988).This can be explainedas follows- Removal of the 5' mor RNA primer by the combination of t. ¿dli RNase H activity and lbe '-Jl' €xonuclcoiyticactivity of -6.col¡DNA poly5 m€raseI mak€s the complcmentary dNTPs wh;ch are located at the 3' end oi a full'length ii$tstrand CDNA, single'sirandcd. Single-stranded DNA, in turn, is hydrolyzed by the I -+5' exonucleolytic activity of E. coli DNA polymeraseI, leaving the end ot the doublc-strandedCDNAproduct blunl or nea¡ly so. Therefore, du¡ing subsequent cloning, sequence information correspoDdingto the 5' most RNA oliSonucleotide that primed se€ond-strandsynrhcsisis ]ost. However, this loss c¿n be minimized by adding RNaseH alter, ;nstead oI wit]n, E. coli DNA polymeraseI in the second srrand synth€sis ¡eaction (D'Alessio and Gerard 1988) sin€e RNase H is not capable ol cleaving hybrid RNA at thc junction belween a ribonrrcleotide and a deoxydbonucleotide (Omer and Faras 1982;Resnicket al. 1984). For furiher discussion and figures illustrating these methods lor synlhesizjng double-stranded cDNA,seeSambrookct at. (1989,Chapter8)

CloningVectorsfor Construction of cDNALibraries Ther€ are two main choices of cloning veclor for construciion ol CDNA liblaries: bacteiophage ¡' v€clors (Murray er al. 1977; Young and Davis 19sla,b;Huynh et al. 1985;Shortet al. 1988tPalazzolo et al. 1990) and plasmid vectors (vieira and Messing1982,1987,t99l; Mead et al l98ó; Dentc and Cortese 1987; Heidecker and Messing 1987; Mead and Kempe¡ 1988). Both cloning systems have advantagesand disadvantáges,and the best

choice !vi1l greatly depend on the specilicapplica'

A VECTOR IN CHOOSING PRINCIPLES GENERAL Cloning cfliciencies (as deiined by the number ol plu or cfu per microgram of bacteriophage;\. or plasmid.vectorDNA, respectively)are no longer a factor in deciding whicb cloning vector to use With the advenl of electroporaiion (Dower et aL. 1988), cloning ellicienciesof recombinant plasmid DNA became comparableto those atiained bY in vlro packaging ol recombinant bacte¡iophageI DNA. Efficiencies of in vitro packaging (using cigapackII Gotd Istratagene2002141)and electroporation (using Electromax E. rolt DHIoB [Life are very similarfor con18290-0151) Technologjes rrol DNA (2 x l0e pfu/lrg ot wild'rype bacreriophagel. DNA and I x l0r0 €Iu/pg of CsClpurified supercoiledpUCl9 plasmid DNA) Thus, a typical ligation containing 50 ng of CDNA (with an ave¡agesizeol I kb) and a twofold mola¡ excessol vector DNA (5 ftg ol bacteriophage¡' DNA or 0.1 pg ol puclg DNA) would be expeciedlo yield approximately three limes more recombinantswith the bacteriophagevector than with the plasmid Bacteriophagel libmries are easier ro screen than plasmid librañes. Mo¡e clones can be screened per plate, replica filters can he more rapidly prepared, and the hybridizarion back ground tends to be lower. Hor4ever,plasmid veclors offer greater versatility than bacteriophage¡" vectorsfor subseqüentmanipulatlon. For examplc, a plasmid DNA library can easily be purified by agarcsegel electrcphoresisto eliminate all of the nonrecombinant vector molecules as lvell as any clones with small inserts that óay have escaped size selection isee pp. 121-l2l). This type oi purilication cannot be as easily accomplishedin bacteriophagc1" iibmries. From a technical sland' point, it is therefore iar easier to obtain 8¡eater - yields of individuaily selectedplasmid DNA than bacteriophagel DNA for subsequentsubcloning and manipulation of the insert I VECTORS BACTERIOPHAGE The most widely used bacteriophage)" vecto¡s for CDNAcloning are ¡,gilo, ¡'8tI l, and lambda ZAP. These bacteriophage), inseÍion ve.lors tl?ically accept inserts of 0-10 kb. Lambda ZA? (Strata-

Ü

52

GDNALibraries Normatized andScreenins 2 r construct¡ns ChaDter

gene) has some advantagesover the other bac¡eriophdgetr cloningvei LoI.bec¿u.eir contd;n'J pld'mid component (pBluescrip!, which has an fl origin oI replicalion and a polylinker wilh multiPle cloning sites) that €an be excisedin vivo after s perir ec¡ion wiü a helper bacteriophagean'l lhen recirculariued,thus eliminating the time involved r'e-s¡er Ld in subcloninS . m b d ¿Z A P i s l h e r c f o r e vecror with which to work. VECTORS PHAGEMID The obse¡vaiionby Dotto er al (1981) thar a plasmid carrying the intergenic region of the Iilamenlous bacteriophageIl can be packaged as single'srrandedDNA into a viral parlicle by a her per bacteriophageled to the construcrionof ve'tors (phagemids)üat combine rhe advantagesof both plasmid and bacreriophagesysiems (zincler and Boeke 1982; Mead ánd Ke¡nper 1986). The p¡ob lem originauy encountered in the use ol these phagemids was the significant reducrion rn tlle amounl ol single-franded DNA próduced as compared io bacreriophage vectors snch ás bacMll. teriophage e) r h ( T h 1 5r e d u ( r i o nw d s d u e r o i n e r f F r e n , b phagemid wiih the replication ol the helper bacteiophage (Enea and Zinder 1982) This problem has now been solved. Wi¡h lhe use of bacteriophage mutants such as bacleiophage MrlKOT, which p¡eferenlially packagesplasmid DNA over bacteriophageDNA (Vieira and Messing 1987), suilicient amounts ol single-strandedplasmid can be ¡outinely obiained irom small-volume cultures il pg ol single-sr¡andedplasmid DNA per millilirer ol culture). A delailed prorocol for con' srruction of tDNA libraies in phagemid vectors is providedon pp. 8ó-123.

for of the Procedure Overview DirectionallY Constructing ClonedcDNALibraries An outline of ihe stepsinvolved in the construc_ tion of direcrionally cloned CDNA libraries in phagemid veclors is depicredin Figure I A r¡o¡e derdrleddenriplion ol rhe processof cloni¡g d bona fide CDNAis presenÍedin Figure 2 and in the protocol on pp. 8ó l2l. Bdeily, a Nr{-(dT)ls oligonucleotide(i-e, an

oligo(dT)t8 flanked by a Nr¡I restrjction srte at lts 5'end) is usedto prime rhe synlhesisoi firsast¡and CDNA wirh RNase H- reverse tra¡rscriptasefuom MMLV (Liie Technologies).(This enzyme is ¡ecommended for ils ability to generate longe¡ reverse lranscriplslhan mos¡ commerciallyavailablelorms of reverse transcriptase,) "one-¡ube" first- and second-strandCDNA synlhesesare perfo¡med essenrially as describedby D'Alessio et a1 (t987) Double-srrandedcDNAs thus genera¡edare lreaied with bacte¡iophageT4 DNA polymeraseto c¡eate blunt ends, size selecredby gel filtration on a Biocel A-50ñ (100-200mesh;Bio'Rad)column,and ljgatedto a largeexcessof adaplermolecules(e g, EcoRl adapters). Adapterligaled cDNAs are then digestedwith No¡I and purified on a sepharoseCL 48 (Pharmacia) column to remove the excess adapter molecules.Size-selecredcDNAs (>350 bp) are úeated wilh bacteriophageT4 polvnucleotide kinase Io phosphorylale the adapler ends (one ol ¡he two oligonucleotidesol the adapter molecule hd. a s'-hydroxyl groJp ro Dfcvcnl concdlemeri_ zation oI adaplersduring liSalion) and ligated dire.Iionally to the NolI- and E oRi-digestedends of a phagemid veclor (pT7Tl-Paq for veclor des' € ption, see p. 86). Bacreriaare rransformedwilh rhe pudlied ligatedma¡erial by electroporalionand F r o p ¿ g " r eudn d e r . e l e , r i o nw : r h d n " p t r o p r i . r e antibiotic.Finally, plasmid DNA is prepared. A recommended option ar üis pornr rs 10 eliminale from lhe libmry all ol the nonrecombinani ("enrp¡y veclor") molecules (seepp. I l l . l 2 l ) ¿ < f o l ' o w \ :A p l " ' m r d D N A p r e F d r ¿ r ¡ o n oi the library can be linearized wiih N¿¡I and the recombinantDNA separatedlrom the emply vecror by agarose gel electrophoresis The recombinanl DNA is then purified using B-agarase(New England Biolabs) and recircülarizedin a large-volume ligadon. The recircülarized DNA is precipiared. bacteria are úansformed by eleclfoporation and L n o e r s e l c c r ' o nw i t h ¿ r ¿ p p l u p r i ¿ l e frop¿gdred antibiotic,and plasmid DNA is prepared

Commonly Problems in cDNALibraries Encountered La4e-scale sequencing of cDNAs has uncovered some of the probtems commonly encountered ir oligo(dT)-primed,directio¡ally cloned CDNAlib|arj e ) ( A d ¿ m \e l a l . l q e l . l o 9 ' l L \ o d r ( 5 l 9 q 4 l ,5 u ( ü as rhe high frequencyof cloneswi¡h a long polytA,

q .l q ;

G G G G G CI CI

G q G G ü -

a a a

I 1 1 a I I I I {

t

in GDNAL¡brades 53 commonlyEncountercd Prcblems

€ rm e r l o ¡ { d T ) t so l i g o n ueco t ¡ d P S y ¡ t h e s ¡ zf ie¡ s ls t r a n dc D N Au s l ¡ g ¡ranscrprase reverse l s y n t h e s i zsee c o n sdt r a n dc D N Ab v n l c k e a n dE c o l ¡ I t r a n sa t o n u s i n gR N a s N D N Ap o l y m e r aIs e t i t hb a c t e r i o P h aTg4eD N A I T r e aw I Poiymer¿se B l u n te n d e dd o u b l es t r a n d ecdD N A l s z es e l e cbt yc o u m nc h r c m a t o g n p l r v S z es ee c t e dd o u be s r a n d € dc D N A( > 35 0 b p ) L gateto excessEcrRladapters Ligaiedc0NA/adapier I D s e s tw t h l o ¡ l i o a l l o wd ¡ r e cot ¡ a l c l o n r n s

t

an'l a'rapter excesslree ¿dapters I Remove c h r c m a tosmPnY b Y c o ! m ¡ d i m e r s { aód¿oo nC IPocpor¿." ho eoo fe4 p o l v n u c l edoei k ¡ a s c b a c t e o r p h a s i 5' e¡d wiih plrosphorvlated Ligate¡cDNA/adapter r N Aa n dc D N A / a d a p t e r L ¡ C a tvee c t oD I l NA c D N A / a d a P tgear t e dt o v e c t oD ( u n a m Pi e d c 0 N Al ¡ b r a r y ) and I lransform6 col/ by e ecfoporatLo¡ qe biotic + p r o p a c aut en d e r e c t i o nw t h ¿ n t P r e p a rPel a s m i0dN A P l a s ñ i dD N AP r e P a roáni ] oigestwith ruot I

. u é P I i 1 u c P ¿ g " r d" r or " - o " m r . o - b r o n rv o ' - o m oo u r ó \ { mt o l e c e us z er e c o ñ b i n a n lReci¡cula, R e cr c u l ¿ ¡ i z eP du , r i l i eP d a s md D N A on and fransiorm E coli bv electrÓporai I p r o p ¿ c a t e s e l é c t l o w n ¡ | r a n tb i o t l c u ' d e r + P r e p a fPei a s m i D d NA t P ¡ a s m ¡DdN A( a ñ p l ¡ f i ecdD N A b r a r vr e a d v _ tnoor r ñ a z a t i o n ) CDNA libraries in Figlre 1 Flow chart of the steps involved in the construction of directionally cloned phagemid veclors. See protocol on pp 8ó-l2l for dctails rail and ih€ occurrence of chimeric clones (ie., CDNA Iragmenls derived frcm differenr mRNA molecules that are join€d by ligation in a single CDNA clone). Ahhough the lormer poscs limitaoi r.q.,..t." Irom ihe l' end i¡"t f"i g."*"1." of a cDN¡, the lalt€r is evcn more serious since it oitcn cannot be idenriljed and may rh€refore resrrlt

ln addition' CDNA in mislcading inlormation clones with small inserts' sometimes conslstrng exclusively oi pieces of mRNArails' prevail in manv l i b r a r i e s( A d a m s e t a l l 9 9 l ' 1 9 9 2 ) T o g c ¡ h e r 't h e s e undesirablc features represent maior impediments to the process of obtaininS thc compl€le nucleotide sequencc ol a transcrrpl'

cDNALibraries Nolmalized andScreening Chapter2 ¡ Construct¡ng

54

3i

g9 H!

tÉ Éá

+38

E

;:

rd

áuE

ti ¡e -

Á

<:;> :s E9

i

t

!fH E" ? t i "; B

4

,?É !r.lE ¡ ñ

¡83

e6 I

,/ ::

i:ü

i"

iE'

Ei{ i3E :9"

;t

l:

2i

z-

;iT

E FC

á

= É_¡

:E

? r'B f,;i

{ --}

: ó2

3:

! l9

;g

Eg

3

ái iqÉ

\-9

e

27

3 \;:,/ !8t

áfÉ :iÉ

<1

2

I

s

¡á Eá

:r

E

E2

,'¡

É!E

+

dÉ /.:

¡

L,:. \ 2 < ._ - , /

5 -

t! iÍ

t! ÉÉ I

!ó

á¡

it !¡

""4

É

a1

a

1i

= 3

CDNAL¡bfaryls Obtained 55 to Enlurelhat a Representative Requiled CloninsEff¡ciencies

presenl in great excessover cDNAs,which may oc' The occurrenc€ol cloneswilh long tails can bc cur, for example, il the amount of CDNA is un_ minimiz€d by taking advantágeof üe fact that the derestimated. It is noleworlhy (as mentioned enzyme reverse traDscriplascis incapable ol dis above) that a signilicantlraction of the synlhesized placing a complementary strand of DNA if its double-strandedCDNA correspondsto small frag_ slnrhesishas b€8un at a siie downslreamfrom ils. mcnts that represent exclusively poly(A) tails of di.cctionot synthesh(Kornbergand Baker r992)' mRNAs. Accordingty,il [32P]dcTPis the radioiso B-v using the No -(dT)ts oligonucleoiide in greal iope that is incorporated duing synthesisto ¿llow exc€ssov€¡ thc mRNA and by controlling the anfor quantitation of the CDNA,all of thesemolecules n€aling conditions very carefully to achieve comwill remain unaccount€d for' Consequ€nily, two plete saturationol the poly(A) tails, only the mosl blunGendedCDNAmoleculescan be joined by ligaex_ proximalpdñer moleculescan be successlully tion and the resulting chimera can then be ligated rended to reversc-transcribethe mRNA. Primers io adaptermolecules.As depicledin Figure 3, two rhái anneal Io rhe poly(A) tracl Iuriher upstream cDNAs can be blunt-end ligated eiúer in a head (respective !o the orientation oi lhe fi¡f-slland io-iail (A) o¡ in a head-to-head (B) orientation. lt DNA) are exrendedfor a very short distanceunüt should be noted, however, that most oi such Lhey encoünter the primer located immediately chimcric CDNA molecules are likelY to be elimi' downstream. The resuhing small Iragmcnts of tail nated at a later step (i.e., digesdon with Notl, can be ellecriv€ly eliminated by a strict slze which precedesthe li¡al ligalion lo the veclor) selectionprocedure The exceptionswould be those casesin which No¡I The occürrenceof cloneswirh small inserts can of the No¡t sitespresbe resolvedby using a reliable systemlor size_sel- fails io cleaveat one or both ent in the chimeric molecule (for delails,seeFigure €clion of cDNAs-Spin .olumns oiten pe¡form rath' 3). In any event, the generationof chimeric molee. poorly for this Purpose cDNAs catl be very effi' cuiesduring ligaiion io ihe adaptercan be sübstan_ ci€ntly size-selectedby gel filtrarion on a Bio-Gel tially minimized if small molec¡rlesare quantitaA-5om (100-2oomesh)column (seepp. l04-107)' tively eliminated by using a reliable systernfor size The occurrenceol chimeric clones is potentially selectionof cDNAs.Furthermore, it is noteworrhy the most seriousol the three problems commonly that both tl?es oi chimeric clones are eliminated encounlered in cDNA tibraries because such during the subsequentpürification slcps invoiving arLirac,sdre of.Fn rndeL€rl¿bleby \equenting lineaization oI a plasmid DNA preparation oi the analysisand consequentlymay ¡esult in misleading library \a,ithNo followed by gel purification, reci¡inlormation. As depictedin Figure l, there áre two cularization of the linearized recombinant moleslepsduring which chime¡ic clones can be generacules,and then t¡ansformation of bacteriaby elecred: The firsr is during the ligation of the adapter troporation and p¡opagation under selecrionwith moleculcsto tbe cDNAs, and the secondis dudng an appropdaieantibiotic. , \ e l : n ¿ rl i C ¿ : o ro l I l - c . D N A . r o l h e ' l o a i n gv € c differen! two have ror. However, since the cDNAs ends Io allow for dircctional cloning and sincevecIor is ahvayspresentjn siight excessover 'DNA in ihat ligation, the lattcr ev€nt is not ljkely to occurln fact, this type of chimeric clone coüld onLv bc generatcd if three CDNA molecules a¡e ligated to one aüother and the resulringchimeric molecule is neededto obtain a ¡epresenihen ligated to lhe vector' The former event, now- - How many clonesare tative CDNA library? The probability that a given ever, can happen if thc adapter molccules are not

to Required CloningEfficiencies EnsureThata RePresentative cDNALibraryls Obtained

the normal processinvolved jn Figure2 P¡ocessof cloning a bona fide CDNAmolecule This figure depicts to lhe cDNAs Note that cl_oninga cDNA molecule ln the lirst s!ep, EcoRIadapte$ are blunt-end-ligated (rhis-prevents avoid multiple con' only o_neof the sÍands of the adapter moiecules is phosphorylated of the restriction lragAdapterligated clÑes are then digesiedwilh N'¡i Note thal one l"ri-.rlrr,l."i. part of the Nod-(dT)18 ; ,mal rrugm"it senerated bv Nol plus EcoRI that consisrs of ;;;'1 adaptermolecule This an E oRI to ligated cDNA oi lirst-irand t*¿ ," prime iire synthesii Thc cDNAs are then size seleciion "iü."".1.*i¿" during elimjnaled be hishlv clonable,vei undesirable,iragmen! mus! oRI ¿nd E Non wilh borh !ccror o'ge\ted oho)DhorvrlFd dnd igdtcdro ¿ phdgemid

Theoret¡cal Considerations ¡n Normaliz¡ns cDNAPopulat¡ons61

:]RNA will be repr€sentedin a CDNA library can ¡e caLculated by the lollowing formula:

r()r)= (t _ tt _¡,) rihere ,7is rhe numberol recombinant clonesandf . ir ihe frequency oi a particular n1RNA. Since a ñpical somatic cell conlains app¡oximately 0.6 pg of mRNA, ir follows rhat there ar€ approximately t00,000 mRNA mol€culesper cell (0.6 pg x 1 molecule/ll x l0 7 pg = 545,454),assumingthat rhe ,ver:ge siz€oi an mRNA is 2 kb (11 x 10-7pg). Accordingly, there is a probability oi 99.9% that rhe rárcsrnRNA 1I copy per cell;/= l/500,000) \vill be ¡ep¡esenledin a library oi 5,000,000 recombinanisirom such a somatic cell (no!e that lhis probability drops to 86.5% lo¡ a library oi 1 . 0 0 0 . 0 0'0e c o m b ' n a n l (Tt .h e r e f o r ei . i s i m p e r ¿ rive that tbe h;ghest cloning efficiencies be áchicved. Transformatjonby clectroporationis the method ihat yields the highest cloning eiliaienciesfor cDNAs ligated to plasmid vectors.T}?ically, a total of 1,000,000-5,000,000 rccombinantclon€scan b€ obtaiued irom 50 ng ol CDNA by elecl¡oporation. High cloning eflicienciescan be attained by electroporation using commercially available com_ pcrent bac¡eria (Electrom¿x [Lite Technotogics]i electroporation efficiency is I x t0l0 cfu/pg ol CsCl purilied supercoiled pUCl9 control DNA). However. existjng prolocols for the preparalion of competent bacleria (Dower er al. 1988) have had some modilications in¡roduced that reproducibly yield cellswirh I x l0r0 to I x l0l0 clu/pgof csclpurilicd supercoiledpucl9 contrcl DNA (see pp. I t8-l t9).

in Considerations Theoretical Normalizing cDNAPopulations The mRNAs of a iypical somatic cell are d¡tibuted into three lrequcncy classes(Bishop er aI 1974; Davidson and Briiten 1979): (I) prevalent, (tI) in' termediaic,ánd (III) rare (complex)-The classesat rhe two exremes (to% and 40-45% of ihe total mRNA species,respectively)include members oc_ curring at vastly difie¡ent relative f¡equencies.On average, the lrost privalent class consistsof ap proximately ten mRNA species,each represented

by 5000 copiesper cell, whereas the classof high complexity comprises 15,000 different species, each represented by only l-15 copies- Even though the rarest mRNA sequencehom any lissue is likcly 1.) be represenredin a cDNA libra¡y of 5,000,000 lo,00o,0oo recombinanis,its idenrificaiion is very difijcult (its frequency of occurrence may be as Iow as 2 x 10-6 on ave¡ageor even l0 7 Ior complex tissuessu€h as the b¡ain). Thus, ior a va¡iety of purposes,ir is advan¡ageousto apply a normalization procedüre and bdng the frequency of each clone in a CDNA library within a narrow rang€. It has been estimated(Soareser al. 1994) üat in a complex organ such as the brain, there are approximately 16 prevalent mRNAs,2lt0 intermediatc mRNAs, and as many as 45,000 different ¡a¡e (complex) mRNAs, which comprise 16Vo, 46"h, and 38% of the total nlRNA mass, respectively. Thus, ó2% ol the clonesin a library correspondto eilher a classI or a classIl transc¡ipt.In contrast, oÍtr'y4.6yó of ihe clones i¡ a normalized library correspond to lranscripts of these two frequency classes([36 + 2l50ll[36 + 2l50 + 45,000]), whereas95.4% of the clonesconespondto classIII transcripts. Therelore, on average, normalization resulrs in a ll.5-lold reduction in the toral frcquency of classI plus classI¡ transcriprsand in a simultaneoüs2.5-iold increasein the iotal frequency ol classIII tanscripts. Two general apprcacheshave been proposedto normalize CDNA libraries (weissman 1987). one approach is based on hybddization to genomic DNA. The frequency ol each hybddized CDNA in the resulting normalized library would be proportional to that of each correspondinggene in the genomic DNA. The feasibility of such an approach has not yet been demonstrat€d. The other approach is based on reassociationkinetics {i.e., the time coürseof the reassociationprocess).lf a population of identical DNA duplexesis dissociatedby heating and then allowed to reassociate,the rate of Lhe rcaction follows second-orde¡kinetics (with rcspect to the concentraiion of single-stranded DNA), that is, dCldt= -kC2 where c is the concentrarion(molesof dNTPs/lite¡) of DNA remaining single stranded at dme t t is time (seconds) spent under reassociationconditions, & is the reassociationÉte constant (liter/

.l

o¿

andScreeringNomal¡zedCDNALibtaries 2 ¡ ConstÍlcting ChaDter

moles'seconds).lntegralion of lhis formula between the limiis Coa! ¡= 0 and C at I = I yjelds the following; c/co=l/(l+kco4 dnd C/co where Coi\ lhe ro!¿l D\A !oncentralro"l expressesthe pe¡centageof DNA ¡emajning srngle silanded at time l. solving lhis equation for hall reassociation(i.e., when C/co = ll2 aÍtd t = tv2) yields the lollowing: v2= ll(l + kcatrl2)

Cotl= llk of thc Lo¡lor h¿lf r(¿r\o'id i.e..t i> thc leciproc¿1 tion. Cñlis the producl of the total DNA concenüation and the time (molesof dNTPs/tite¡x seconds) Accordingly, if CDNA reannealinglollows secondorde¡ kinetics,ra¡er speciesanneallessrapidly, and ¡hus the remaining single-fmnded fuaction of CDNAbecomesp¡ogressivelymore normalizedduring the cours€of the reassociationreacrion (Galau et al. 1977). Specificloss of any speci€soi CDNA, regardlessof its abundance,does not occur at any time (Ko 1990). severdlp¿r"meters¿ffeclIhe rdl€ of re¡ssocia_ tion in a reacdon: . Temperature, The reac¡ion rale incr€asesas the tempe¡ature decreases below Tn (i.e., thc temperatureat which half of the total changein a melting culve has occurr€d),reachinga b¡oad flar maximurn f¡om t5oc to 30oCbelow Th and Lh€n decredringwilh ¿ further decre¿re¡n ¡eactionsare usually temperatü¡e.Reassociation Tm. run at 25oCbelow . Salt co cennation.High salt concentra¡on acceleratesthe reaction. The sÍandard conditions of incubation are 25oc below Tn in 120 mM sodium phospharebufier (l8o mM Na+)- However, if the type of sall is differeni and a correc_ rion is applied to ihe measuredvalue oi Cot the_ cortected vallle is called equivalent Col (ECol), r¡,,hereEcof = cot x 1, co¡ is the col under srándard conditions and 1 is the reassociationrate relativ€ lo 120 mM sodium phosphale bufler. For example, the monovalent cation concentration in th€ ¡eassociationrcaction mixture is o-5 M (ihe ¡eaction is performed in 0.5 M NaCl). Ac-

cordingty,y = 0.t Mlo 12 M = 4.2 r'e.,ihe reac ¡ion is 4.2 times fasler lhan in t20 mM sodium phosphatebufler. DNA wiü higher G + C contenl G + C cofil¿11t. The relalion between rate faster. reassociates G + C is linear lor percentage and constant to 65y. G + a from 3o"A range DNASin the the reactior 1,t) is of constan! Leryth. T:¡rerate of the length of root the square proponional to comreacting the lwo if the DNA rrand length' same oi úe are plementary slrands However, if the reacting singie rrands are representedin sloichiometric quaAtiriesbuI are ol different lengths,lhen k is proportional ro ¡he squareroot of the shorie¡ slrand Increasing the formami'le conOrtanic solvents. centralion decreaseslhe rale constanl of the reaciionby I l% lor each 1% in_ ¡eassociation creasein lormamide (e.g.,the ra!e confant of a rcaction performed in 50% formamide is 045 dmes rhe optimal mlc in an idenlical hybridiza tion solution withoul lormamide). ' viscosílt."fhe reaction ¡ate is inversely propor_ tional to üe microscopic viscosi¡y in high-sah solvenrs; increase ol lhe macroscopicviscosilY (as measuredwiih a viscometer) by a polymer usually has liltle effe.t on the rate (polymersdo nol alf€ci the DNA microenvironmenr). Horvever, certairi polymers (e g., dextran sülfate) al very high concentralions accelera¡e the reas sociationmte as much as 20 times This is prob_ abiy due lo the lac! that lhe volume ex.lusron by ihe added polymer effectively inc¡easesthe DNA concentradonand iherefore the ra!e. . Sequenecomplexit/. f]r,e complexiry (N) ot a populadon of nucleic acid molecules is the lolal length ol different sequencesrepresented.Il is usually expressedin nuctqolides (RNA) or base p " i r s( D N A ) .s i n c el l ' e r ¿ l e - l i m i l i n' l! e p i n r e ¿ ( ' sociation is the nucleation event (i e., lhe processby which a früitful collision ol strand p¿irs 's ¡ecogoc.ur': Lheregion ol .omplemenrdr¡Ly nized and base-pairlormation begins),it follorts that the rate oÍ th€ reaction depe¡ds directly on lhe concenl¡ation of €ach sequencein üe mlxiure. Accordingly, for a given lotal DNA concent¡ation, the greater lh€ complexlty, lhe dower rhe reactio¡, since lhe concenüa¡ion on eách sequenceis lower. In summary, the rale of ¡eassociationof a nucleic acid is inversely proportional to its complexfY.

.l .l

q I

.i .{

d -

d

d

d

CDNAL¡blaries 63 No|malized for constructing overviewof the Methods

of the Methodsfor Overview Normalized Constructing cDNALibraries

ing steps: (1) cloniDg shori cDNAs produced by random priming inro ¡.gtto, (2) ampliiicatjon of cloned cDNAs by PCR; (3) denaturaaon and reassociationto a moderate col (col js concentrationoi the reactingDNA strandsx tim€), (4) separationof Four dill€rent methods have been r€ported thus single strands by chromaiography oo I'laP coli a r r o r L h cL o n ' r r u c l i oor l n o r m d l i / FcdD N Al i b r dumns; (5) amptification by PcR oi flow-througt) i€s on the basis of the reassociationkinelics apet al single-strandedcDNAs from the HAP columni and 1991, Sas¡ki p¡oach(Ko l99O;Paranjalie! ai. (ó) clonirig into ¡.gtl0. Thus, Patanjaii'snormalized 1994; Soareset al. 1994).A bri€l descriptionof consistedol CDNA clones containing both library pro\ belos rdcd edchol rhc'Fpru.edure' is coding and noncoding inlormation. However, the Ko's rnethodinvolvesthe following steps:(l) cDNAshad to b€ relalively shori and homogeneous synthesisof doubie'srrándedcDNAs lrom mRNAs in length to ensure cqual efliciency ol amplificaof (2) shearing by using an No¡l-oligo(dT)primer, rion during PCR. (l) ot ligalion rh€ cDNAsto 200 40o-bp fragmenls; Sasakiet al. (1994) repoÍed a differenl süategy has one cDNAs to a "lone linkcr"'primcr, rvhich generalea normalizcd library involving üe iolto EcoRI blunt end and one overhangand an internal steps: (l) conve¡sion of n]RNA into c¡NA lowing a srngle sire (since th€ ovcrhang is no! sticky, on latex beads;(2) iour cyclesof hvb¡idizaiion bemol€cule oi this linker is attachedto each end of a tween the whole II]RNA population and rhej¡ corcDNA in an ori€ntation specificmanner and is thus responiling cDNAs immobilized on larer beads callcd a "lone linker"); (a) amplification of the (CDNAls presentat a loudold molar excessin each r D N A . b y P . R u . i n g r e l i 8 ¿ r ' dl i n t ¡ r - D ' n e r ' e cycle of hybridization); after each cycle of hybúdiquence; (5) denaiuration and reassociationof the zation, the beads are ¡emoved by centrifügation of the double'stranded cDNAs, and isolation and tbe süpernatanttuactionit likewise hvbridized remaining single-strandedcDNAs by chromatog' using the same amoúnt of new or regene¡ated the oi raphy on HAP columnsi 16) amplificarion cDNA-bead complexes;and (l) ai the end of the in step si;sle-strand€dcDNAsby PcR (asdescrib€d fourth cycte,üe leitover mRNA, which is depleted a); and (7) digestion of the CDNA amPlification of ¡he classI and cl:ss II spcci€s(thoseof the highdilccrion¿l DrodLcr'sith borh N,rl Jnd L.oRldnd est and intermediale abundance),is úsed as temilonrns inro ¿ pld\rrid ve. or' Ac'ord:nglv. alplate for CDNA synthesis and cloning Since apthough both coding and noncoding fragments ar€ p¡oximately 96-98% of the input poly(A)+ RNA is presentduring the reassociationreaclion only ihe removed after lour cycles oi hybddization;'-noncoding fragmentsremain in the normalized subtraction, leaving only 2-4% in the ünhybrid_ library alter the double digestion and djrectional ized supematant fraction to be used for library a norcloning steps.Ko's rationale lor construcling construction, iarge amoünts ol starting poly(A)+ 3'-nonmalizc¡ library consisting primarily ol RNA are required lo obtain a reprcsentative licoding sequenceswas the following: Durjng the brary. Moreover, considerableamounts of stalting r c ¿ . ' o , i d l i o\nt e f ' .e ¿ ( h t ' _ n o n ' o d i n gl e q u c n ' ei s potyin¡* nne are also required lpr synthesis of expected to reann€al io its own complementary CDNAon latex beads. strand only since the l'_noncoding terminal exon The fouÍh procedu¡e developed to normalize is almost always unique to a parlicular t¡anscript libraries (Figure a) (Soareset al 1994) can CDNA In contmst,coding exonsmay be conservedamong as follows: (l) a directionally summadzed be be members of a gene family' some ol which may in the lorm o{ single_stranded CDNA library cloned less represented than others in a given RNA to an oligo(dT)ls primer; is annealed DNA circü]ar source-Thus, during reassociation,ü€ most abün(2) controued primer-extension reactionsare perdant of such coding sequencesmay cross-hybridize fomed to synthesizea short complementarystrand ro a related,but divergent, complementarystrand (200 t 20 bp) on each circular template; (3) the from a lessprevalent family member, which could resülting partialy double-strandedDNA is purified r€sult in the elimination of the rarer family mem_ by chromatographyon HAI columns to separateit ber l¡om the normalizedlibrary' circular DNA; from any remaining single_stranded Paranjaliet at. (t991) generateda normalizedlito and rcassociated is melied DNA (4) HAP-bound brary by a similar procedureinvolving the follow-

I q 64

CDNALibla es Normalized screen¡ng chapter2 r construct¡ngand

I I

a a Pu,li¿d s ngléslfanded0NAtrom

r ¿ s m m L o rE e , D N Ap o r r m s d ¿L d N T P S ddArP ddcrP ddcrP

I t

^oe

boó

J J J

t l f

a t t

I J

I I I I

M e Lp¡ a : y d o u b e r ñ n d ¿ d crclrü DNArd É4eoare

a) (, ^\

oo o

(()

lpur fed ¡orm¿rizedbc¡y) i

DNAby prm{sbmm

I

tr¡dd 5eedioi

!¡th atrlitrioric

Figure4 Normalizafion of directionallyclonedCDNAlibrariesconstructedin phagemidvec¡ors(soareser aL 1994). A preparation oi single-slrandedDNA'ftom a directionally cloned library is annealed lo a. oligo{dT)12ls primer (representedby an arrow), and controlled extensionsare pellormed with the KlenoÑfragment of E. ¿rli DNA polym€ras€I in ihe presenceol dNTPs(dATP,dTTP,dcT", and dGTP)and ddliTP; (ddATP, ddCTP,and ddGTP).The resuliing pa ially double-strandcdcircula¡ DNA is purified on a HAP column to sepa¡atei¡ ftom any remaining single'rranded circular DNA and is then melted and allolved llr reassociareto a relatively low Col (-5-20 seconds-moles/iiter).The lracrion that remains síngle strandei (the normalized lib.ary) is puriiied from the reassocialedmalerial on a HAP column, pa¡tially corvenÉi into double strandsby primer exrension,and amplified using rranslormalion oi bac¡eriaby electropor¿tiD. and p¡opagationünder selectionwith ampicillin. see protocol on pp. 124 148 for details.

a

a I a a a I 1

a a

CDNALibraries 65 fof const¡ucting Notm¿lized overv¡ew ol the Methods

: ..l¿rivelylorv Col;(5) lhe lradion that ¡emains .::qlc-strándcdis purilied from the reassociated -¡lecules by chromatography on HAP columns; :rJ f6) the singleslranded Ilow-through DNA ::¡m lhe HAP column is partiallyconverledinto j¡uble strandsior improvem€niof elect¡oporation ::íiciencies and bacteria are ih€n transformed by :.ectroporation for propagaling a normalized li_ ,-:ar) wiihorrt subcloning sreps.Accordingly, al' participate ::oLrgh only l'-noncoding sequences :lr lh€ reassociationreaclion, the resulting normal-

ized library consistsoi cloneswith large inserts encompassingboih coding and noncoding sequences. This meüod can be applied to any libmry con fructed in a phagemid vector or in any other vector with an fl o¡igin of replicationsuch thar library cDNA rcn bc o,,rired in rhe form o' \inCle stranded circular DNA after superinf€ction of a g¡owing culture with a helper bacteíophage. A derailed protocol and itlrthcr discussion of this methodare providedon pp. 124-148.