Proc. Nat. Acad. Sci. USA
`Vol. 69, No. 10, pp. 2904-2909, October 1972
`
`Biochemical Method for Inserting New Genetic Information into DNA of
`Simian Virus 40: Circular SV40 DNA Molecules Containing Lambda
`Phage Genes and the Galactose Operon of Escherichia coli
`(molecular hybrids/DNA joining/viral transformation/genetic transfer)
`
`DAVID A. JACKSON*, ROBERT H. SYMONSt, AND PAUL BERG
`
`Department of Biochemistry, Stsnfor!l University Medical Center, Stanford, California 94305
`
`Contributed by Paul Berg, July 31, 197S
`
`ABSTRACT We have developed methods for covalently
`joining duplex DNA molecules to one another and have
`used these techniques to construct circular dimers of
`SV40 DNA and to insert a DNA segment containing lambda
`phage genes and the galactose operon of E. coli into SV40
`DNA. The method involves: (a) converting circular
`SV40 DNA to a linear form, (b) adding single-stranded
`homodeoxypolymeric extensions of defined oomposition
`and length to the 3' ends of one of the DNA strands with
`the enzyme terminal deoxynucleotidyl transferase (c)
`adding complementary homodeoxypolymeric extensions
`to the other DNA strand, (d) annealing the two DNA mole(cid:173)
`cules to form a circular duplex structure, and (e) filling
`the gaps and sealing nicks in this structure with E. coli
`DNA polymerase and DNA ligase to form a covalently
`closed-circular DNA molecule.
`
`Our goal is to develop a method by which new, functionally
`defined segments of genetic information can be introduced into
`mammalian cells. It is known that the DNA of the trans(cid:173)
`forming virus SV40 can enter into a stable, heritable, and
`presumably covalent association with the genomes of various
`mammalian cells (1, 2). Since purified SV40 DNA can also
`transform cells (although with reduced efficiency), it seemed
`possible that SV40 DNA molecules, into which a segment of
`functionally defined, nonviral DNA had been covalently
`integrated, could serve as vectors to transport and stabilize
`these nonviral DNA sequences in the cell genome. Ac(cid:173)
`cordingly, we have developed biochemical techniques that are
`generally applicable for joining covalently any two DNA
`molecules.f Using these techniques, we have constructed
`circular dimers of SV40 DNA; moreover, a DNA segment
`containing X phage genes and the galactose operon of Esche(cid:173)
`richia coli has been covalently integrated into the circular
`SV40 DNA molecule. Such hybrid DNA molectJles and others
`like them can be tested for their capacity to transduce foreign
`DNA sequences into mammalian cells, and can be used to
`determine whether these new nonviral genes can be expressed
`in a novel environment.
`
`• Present address: Department of Microbiology, University of
`Michigan Medical Center, Ann Arbor, Mich. 4&104.
`t Present address: Department of Biochemistry, University of
`Adelaide, Adelaide, South Australia, 5001 Australia.
`t Drs. Peter Lobban and A. D. Kaiser of this department have
`performed experiments similar to ours and have obtained similar
`results using bacteriophage P22 DNA (Lobban, P. and Kaiser,
`A. D., in preparation).
`·
`
`MATERIALS AND METHODS
`
`DNA. (a) Covalently closed-circular duplex SV40 DNA
`· [SV 40(I) 1 (labeled with [8H 1dT, 5 X 1 0• cpm/ ~g), free from
`SV 40 linear or oligomeric molecules [but containing 3-5%
`of nicked double-stranded circles-SV 40(II) 1 was purified
`from SV40-infected CV-1 cells (Jackson, D., & Berg, P., in
`preparation). (b) Closed-circular duplex Xdvgal DNA labeled
`with [8H1dT (2.5 X 104 cpm/~g), was isolated from an E. coli
`strain containing this DNA as an autonomously replicating
`plasmid (see ref. 3) by equilibrium sedimentation in CsCl(cid:173)
`ethidium bromide gradients (4) after lysis of the cells with de(cid:173)
`tergent. A more detailed characteri.?;ation of this DNA will
`be published later. Present information ·indicates that the
`Xdvgal (Xdv-120) DNA is a circular dimer containing tandem
`duplications of a sequence of several X phage genes (including
`C1, 0, and P) joined to the entire galactose operon of E. coli
`(Berg, D., Mertz, J., & Jackson, D., in preparation). DNA con(cid:173)
`centrations are given as molecular concentrations.
`Enzymes. The circular SV40 and Xdvgal DNA molecules
`were cleaved with the bacterial restriction endonuclease· RI
`(Yoshimori and Boyer, unpublished; the enzyme was gen(cid:173)
`erously made available to us by these workers). Phage
`X-exonuclease (given to us by Peter Lobban) was prepared
`according to Little et al. (5), calf-thymus deoxynucleotidyl
`terminal transferase (terminal transferase), prepared ac(cid:173)
`cording to Kato et al. (6), was generously sent to us by F. N.
`Hayes; E. coli DNA polymerase I Fraction VII (7) was a gift
`of Douglas Brutlag; and E. coli DNA ligase (8) and exo(cid:173)
`nuclease III (9) were kindly supplied by Paul Modrich.
`Substrates. [a-32P1deoxynucleoside triphosphates (specific
`activities 5-10 Ci/~mol) were synthesized by the method of
`Symons (10). All other reagents were obtained from com(cid:173)
`mercial sources.
`Centrifugations. Alkaline sucrose gradients were formed by
`diffusion from equal volumes of 5, 10, 15, and 20% su!)rose
`solutions with 2 mM EDTA containing, respectively, 0.2,
`0.4, 0.6, and 0.8 M NaOH, and 0.8, 0.6, 0.4, 0.2 M NaCI.
`100-~l samples were run on 3.8-ml gradients in a Beckman
`SW56 Ti rotor in a Beckman L2-65B ultracentrifuge at 4°
`and 55,000 rpm for the indicated times. 2- to 10-drop fractions
`were collected onto 2.5-cm diameter Whatman 3MM discs,
`dried without washing, and counted in PPO-dimethyl
`POPOP-toluene scintillator in a Nuclear Chicago Mark II
`
`2904
`
`BEQ 1006
`Page 1
`
`

`
`Proc. Nat. Acad. Sci. USA 69 (197~)
`
`Insertion of the Galactose Operon into SV40 DNA
`
`2905
`
`scintillation spectrometer. An overlap of 0.4% of 32P into
`the 3H channel was not corrected for.
`CsCl-ethidium bromide equilibrium centrifugation was per(cid:173)
`formed in a Beckman Type 50 rotor at 4° and 37,000 rpm for
`48 hr. SV40 DNA in 10 mM Tris·HCl ~pH 8.1)-1 mM Na
`EDTA-10 mM NaCl was adjusted to 1.566 g/ml of CsCl
`and 350 pg/ml of ethidium bromide. 30-Drop fractions were
`collected and aliquots were precipitated on Whatman GF /C
`filters with cold 2 N HCl; the filters were washed and counted.
`
`Electron Microscopy. DNA was spread for .electron micros(cid:173)
`copy by the aqueous method of Davis et al. (:11) and photo(cid:173)
`graphed in a Phillips EM 300. Projections of the molecules
`were traced on paper and measured with a Keuffel and Esser
`map measurer. Plaque-purified SV40(II) DNA was used as
`an internal length standard.
`
`C01Wersion of SV 40(!) DNA to Unit Length Linear DNA
`[SV40(LR1 )] with R 1 Endonuclease. [3H]SV40(I) DNA (18.7
`nM) in 100 mM Tris·HCl buffer (pH 7.5)-10 mM MgCl2-2
`mM 2-mercaptoethanol was incubated for 30 min at 37°
`with an amount of Rr previously determined to convert 1.5 ·
`times this amount of SV 40(I) to linear molecules [SV 40(LRr)];
`N a EDTA (30 mM) was added to stop the reaction, and the
`DNA was precipitated in 67% ethanol.
`
`Removal of 5'-Terminal Regions from SV40(LRI) with ;\
`Exonuclease. [3H]SV40(LRr) (15 nM) in 67 mM K-glycinate
`(pH 9.5), 4 mM MgCl2, 0.1 mM EDTA was incubated at 0°
`with ;\-exonuclease (20 pg/ml) to yield [3H]SV40(LRrexo)
`DNA. Release of [3H]dTMP was measured by chromato(cid:173)
`graphing aliquots of the reaction on polyethyleneimine thin(cid:173)
`layer sheets (Brinkmann) in 0.6 M NH4HC03 and counting
`the dTMP spot and the origin (undegraded DNA).
`Addition of Homopolymeric Extensions to SV40(LRiexo)
`with Terminal Transferase. (3H]SV40(LRrexo) (50 nM) in 100
`mM K-cacodylate (pH 7.0), 8 mM MgCl2, 2 mM 2-mercapto(cid:173)
`ethanol, 150 pg/ml of bovine serum albumin, [a-32P]dNTP
`(0.2. mM for dATP, 0.4 mM for dTTP) was incubated with
`terminal transferase (3Q-60 pg/ml) at 37°. Addition of
`[ 32P]dNMP residues to SV40 DNA was measured by spotting
`aliquots of the reaction mixture on DEAE-paper discs
`(Whatman DE-81), washing each disc by suction with 50 ml
`(each) of 0.3 M NH4-formate (pH 7.8) and 0.25 M NH4HC03,
`and then with 20 mi of ethanol. To determine the proportion
`of SV40 linear DNA molecules that had acquired at least one
`"functional" (dA)n tail, we measured the amount of SV40
`DNA (3H counts) that could be bound to a Whatman GF/C
`filter (2.4-cm diameter) to which 150 pg of polyuridylic acid
`had been fixed (13). 15-pl Aliquots of the reaction mixture
`were mixed with 5 ml of 0.70 M NaCl-Q.07 M Na citrate
`(pH 7.0)-2% Sarkosyl, and filtered at rooin temperature
`through the poly(U) filters, at a flow rate of 3-5 ml/min.
`Each filter was washed by rapid suction with 50 ml of the same
`buffer at 0°, dried, and counted. Control experiments showed
`that 98-100% of [3H]oligo(dA)l25 bound to the filters under
`these conditions. When the ratio of [32P]dNMP to [3H]DNA
`reached the value equivalent to the desired length of the
`extension, the reaction was stopped with EDTA (30 mM) and
`2% Sarkosyl. The [3H]SV40(LRrexo)-[ 32P]dA or -dT DNA
`was purified by neutral sucrose gradient zone sedimentation
`to remove unincorporated dNTP, as well as any traces of
`SV40(I) orSV40(II) DNA.
`
`Formation of Hydrogen-Bonded Circular DNA Molecules.
`[32P]dA and -dT DNAs were mixed at concentrations of 0.15
`nM each in 0.1 M NaCl-10 mM Tris·HCl (pH 8.1)-1 mM
`EDTA. The mixture was kept at 51° for 30 min, then cooled
`slowly to room temperature.
`
`Formation of Covalently Closed-Circular DNA Molecules.
`After annealing of the DNA, a mixture of the enzymes, sub(cid:173)
`strates, and cofactors needed for closure was added to the
`DNA solution and the mixture was incubated at 20° for 3-5
`hr. The final concentrations in the reaction mixture were:
`20 mM Tris·HCl (pH 8.1), 1 mM EDTA, 6 mM MgCl2,
`50 pg/ml bovine-serum albumin, lOmM NH4Cl, 80 mM NaCl,
`0.052 mM DPN, 0.08 mM (each), dATP, dGTP, dCTP,
`and dTTP, (0.4 pg/ml) E. coli DNA polymerase I, (15 units/
`ml) E. coli ligase, and (0.4 unit/ml) E. coli exonuclease III.
`
`RESULTS
`
`General approach
`Fig. 1 outlines the general approach used to generate circular,
`covalently-closed DNA molecules from two separate DNAs.
`Since, in the present case, the units to be joined are them(cid:173)
`selves circular, the first step requires conversion of the circular
`structures to linear duplexes. This could be achieved by a
`double-strand scission at random locations (see Discussion)
`or, as we describe in this paper, at a unique site with Rr re(cid:173)
`striction endonuclease. Relatively short (5Q-100 nucleotides)
`poly(dA) or poly(dT) extensions are added on the 3'-hydroxyl
`termini of the linear duplexes with terminal transferase; prior
`
`5'PO --...::...-------01-1 3'
`3'HO
`OP5'
`
`D ;>. Exonucleo~
`D Terminal transfer~
`
`dATP or dTTP
`
`o'''IP~O=======~~ OH3'
`3'HO~
`OP5'
`or
`~ OH3'
`~'PO
`3'HO ~~=======co~P5'
`
`dT
`
`dA
`
`dT
`Fm. 1. General protocol for producing covalently closed
`SV40 dimer circles from SV40(I) DNA.
`*The four deoxynucleoside triphosphates and DPN are also
`present for the DNA polymerase and ligase reactions, respec(cid:173)
`tively.
`
`BEQ 1006
`Page 2
`
`

`
`2906
`
`Biochemistry: Jackson et al.
`
`Proc. Nat. Acad. Sci. USA 69 (1972)
`
`either dATP or dTTP resulted in appreciable addition of
`mononucleotidyl units to the DNA. But, for example, after
`addition of 100 residues of dA per end, only a small propor(cid:173)
`tion of the modified SV40 DNA would bind to filter discs
`containing poly(U) (13). This result indicated that initiation
`of
`terminal nucleotidyl addition was
`infrequent with
`SV40(La,1), but that once initiated those termini served as
`preferential primers for extensive homopolymer synthesis.
`Lobban and Kaiser (unpublished) found that P22 phage
`DNA became a better primer for homopolymer synthesis
`after incubation of the DNA with~ exonuclease. This enzyme
`removes, successively, deoxymononucleotides from 5'-phos(cid:173)
`phoryl termini of double-stranded DNA (15), thereby render(cid:173)
`ing the 3'-hydroxyl termini single-stranded. We confirmed
`their finding with SV40(La,x) DNA; after removal of 3Q-50
`
`4.1
`
`2.8
`
`+ +
`
`-8
`3
`25 ~
`I ..
`l5
`
`20
`
`160
`
`140-
`
`r,20
`fro
`8.00
`~
`;F60
`
`40
`
`700
`
`6001
`
`1::
`500-u
`~
`400~
`
`300~
`200
`
`100
`
`Fm. 2. Alkaline sucrose gradient sedimentation of [3JIJSV40-
`(LRrexo)-[31P] (dA)so DNA. 0.16 p.g of DNA was centrifuged for
`6.0hr.
`
`removal of a short sequence (3Q-50 nucleotides) from the
`5'-phosphoryl termini by digestion with~ exonuclease facili(cid:173)
`tates the terminal transferase reaction. Linear duplexes con(cid:173)
`taining
`(dA),. extensions are annealed to the DNA to be
`joined containing (dT),. extensions at relatively low concen(cid:173)
`trations. The circular structure formed contains the two
`DNAs, held together by two hydrogen-bonded homopolymeric
`regions (Fig. 1). Repair of the four gaps is mediated by E.
`coli DNA polymerase with the four deoxynucleosidetriphos(cid:173)
`phates, and covalent closure of the ring structure is effected
`by E. coli DNA ligase; E. coli exonuclease III removes 3'(cid:173)
`phosphoryl residues at any nicks inadvertently introduced
`during the manipulations (nicks with 3'-phosphoryl ends
`cannot be sealed by ligase).
`
`Principal steps in the procedure
`
`Circular SV 40 DNA Can Be Opened to Linear Duplexes by
`Rr Endonuclease. Digestion of SV40(I) DNA with excess R1
`endonuclease yields a product that sediments at 14.5 S in
`neutral sucrose gradients and appears as a linear duplex with
`the same contour length as SV40(II) DNA when examined by
`electron microscopy [ (18); Jackson and Berg, in prepara(cid:173)
`tion; see Table 1]. The point of cleavage is at a unique site
`on the SV40 DNA, and few if any single-strand breaks are
`introduced elsewhere in the molecule (18); moreover, the
`termini at each end are 5'-phosphoryl, 3'-hydroxyl (Mertz,
`J., Davis, R., in preparation). Digestion of plaque-purified
`SV40 DNA under our conditions yields about 87% linear
`molecules, 10% nicked circles, and 3% residual supercoiled
`circles.
`
`Addition of Oligo(dA) or -(dT) Extensions to the 3'-Hydroxyl
`Termini of SV40 (LRr). Terminal transferase has been used
`to generate deoxyhomopolymeric extensions on the 3'(cid:173)
`hydroxyl termini of DNA (7); once the chain is initiated,
`chain propagation is statistical in that each chain grows at
`about the same rate (12). Although the length of the exten(cid:173)
`sions can be controlled by variation of either the time of in(cid:173)
`cubation or the amount of substrate, we have varied the time
`of incubation to minimize spurious nicking of the DNA by
`trace amounts of endonuclease activity in the enzyme prep(cid:173)
`aration; we have so far been unable to remove or selectively
`inhibit these nucleases (Jackson and Berg, in preparation).
`Incubation of SV40(La,1) with terminal transferase and
`
`bottom
`
`5
`
`25
`
`A 6 C 0
`
`20
`115
`10
`Fraction number
`FIG. 3. Alkaline sucrose gradient sedimentation of [1H](cid:173)
`SV40(LRrexo)-[11P](dA)so and -(dT)so DNA incubated 4 hr
`with and without addition of E. coli DNA polymerase I (P),
`·ligase (L ), and exonuclease III (Ill). Conditions are described in
`Methods. 8-Drop fractions were collected. Samples A, C, and D
`'were centrifuged for 60 min, sample B for 90 min. Line A, dA(cid:173)
`ended, plus dT-ended SV40 linears, plus (P+L+III) (31P,
`e; 3JI, O); line B, dT-ended, SV40 omitted, plus (P+L+III)
`( 11P, "'); line C, dA-ended SV40 omitted, plus (P+L+III)
`(11P, •); lineD, dA-ended plus dT-ended SV40 linears, without
`(P+L+III) (11P, .&). 3H profiles are not shown for lines B, C, and
`D, but all show that the SV40 DNA sediments in its normal
`monomeric position. The np and 3H profiles in line A are shifted
`to a faster-sedimenting position with respect to the up profile
`in line D because SV40 strands are covalently linked to one
`another through (dA)so or (dT)so bridges in most of the molecules,
`whether or not covalently closed-circles are formed. Very little
`31P remains associated with the SV40 DNA in lines Band C be(cid:173)
`cause tails that remain single-stranded are degraded to 5'-mono(cid:173)
`nucleotides by the 3'- to 5'-exonuclease activity of E. coli DNA
`polymerase I (7 ).
`The arrows indicate the position in the gradient of different size
`supercoiled marker DNAs; the number is the multiple of SV40
`DNA molecular size (1.0).
`
`BEQ 1006
`Page 3
`
`

`
`Proc. Nat. Acad. Sci. USA 69 (1972)
`
`Insertion of the Galactose Operon into SV40 DNA
`
`2907
`
`nucleotides per 5'-end (see Methods), the number of SV40(Lai)
`molecules that could be bound to poly(U) filters after incuba(cid:173)
`tion with terminal transferase and dATP increased 5- to 6-
`fold. Even after separation of the strands of the SV 40(LRiexo)(cid:173)
`dA, a substantial proportion of the 3H-label in the DNA was
`still bound by the poly(U) filter, indicating that both 3'(cid:173)
`hydroxy termini in the duplex DNA can serve as primers.
`The weight-average length of the homopolymer extensions
`was 50-100 residues per end. Zone sedimentation of [3H](cid:173)
`SV40(LR1exo)-[32P](dA)80 (this particular preparation, which
`is described in Methods, had on the average, 80 dA residues
`per end) in an alkaline sucrose gradient showed that (i) 60--
`70% of the SV40 DNA strands are intact, (ii) the [32P](dA)so
`is covalently attached to the [3H]SV40 DNA, and (iii) the
`distribution of oligo(dA) chain lengths attached to the SV40
`DNA is narrow, indicating that the deviation from the cal(cid:173)
`culated mean length of 80 is small (Fig. 2). SV40(LR1exo),
`having (dT)80 extensions, was prepared with [32P]dTTP and
`gave essentially the same results when analyzed as described
`above.
`
`Hydrogen-Bonded Circular Molecules Are Formed by An(cid:173)
`nealing SV.~O(LRrexo)-(dA)so and SV 40(LRrexo)-(dT)so To(cid:173)
`gether. When SV40(LRiexo)-(dA)so and SV40(LRiexo)-(dT)so
`were annealed together, 3Q-60% of the molecules seen by
`electron microscopy were circular dimers; linear monomers,
`linear dimers, and more complex branched forms were also
`seen. If SV40(LRiexo)-(dA)so or -(dT)so alone was annealed,
`no circles were found. Centrifugation of annealed prepara(cid:173)
`tions in neutral sucrose gradients showed that the bulk of the
`SV40 DNA sedimented faster than modified unit-length
`linears (as would be expected for circular and linear dimers,
`as well as for higher oligomers). Sedimentation in alkaline
`gradients, however, showed only unit-length single strands
`containing the oligonucleotide tails (as seen in Fig. 2).
`
`Covalently Closed-Circular DNA Molecules Are Formed by
`Incubation of Hydrogen-Bonded Complexes with DNA Poly(cid:173)
`merase, Ligase, and Exonuclease III. The hydrogen-bonded
`complexes described above can be sealed by incubation with
`the E. coli enzymes DNA polymerase I, ligase, and exonuclease
`III, plus their substrates and cofactors. Zone sedimentation
`in alkaline sucrose gradients (Fig. 3) shows that 20% of the
`
`14
`
`14
`
`12
`
`• 12
`•
`b10
`-
`10
`~ 8
`fr
`a.6
`Ill
`
`8
`
`6
`
`4
`
`4
`
`I>
`
`-1 1
`
`2
`
`2 A
`
`J
`0
`0
`A ~~2~-4~~6~~8~~10~~12~~14~~~~~~~~~ A
`bottom
`Fraction number
`
`FIG. 4. CsCl-ethidium bromide equilibrium centrifugation
`of the products analyzed in Fig. 4. Line A, dA-ended, plus dT(cid:173)
`ended SV40 linears, plus (P+L+III) (32P, e; 3H, O); line B, the
`same mixture without (P+L+III) (32P, A; aH, t.).
`
`TABLE 1. Relative lengths of SV40 and "Advgal-120
`DNA molecules
`
`DNA species
`
`SV40(II)
`SV40(LRI)t
`(SV40-dA dT).
`>..dvgal-120(1)
`Xdvgal-120(LRI)
`Xdvgal-SV 40
`Xdv-1
`
`Length ± standard
`deviation in
`SV40 units*
`
`Number of
`molecules in
`sample
`
`1.00
`1.00 ± 0.03
`2.06 ± 0.19
`4.09 ± 0.14
`2.00 ± 0.04
`2.95 ± 0.04
`2.78 ± 0.05
`
`224
`108
`23
`65
`163
`76
`13
`
`*The contour length of plaque-purified SV40(Il) DNA is
`defined as 1.00 unit.
`t Data supplied by J. Morrow.
`
`input 32P label derived from the oligo(dA) and -(dT) tails
`sediments with the 3H label present in the SV40 DNA, in the
`position expected of a covalently closed-circular SV 40 dimer
`(7Q-75 S). About the same amount of labeled DNA bands
`in a CsCl-ethidium bromide gradient at a buoyant density
`characteristic of covalently closed-circular DNA (Fig. 4).
`DNA isolated from the heavy band of the CsCl-ethidium
`bromide gradient contains primarily circular molecules, with
`a contour length twice that of SV40(II) DNA (Table 1) when
`viewed by electron microscopy. No covalently closed DNA is
`formed if either one of the linear precursors is omitted from
`the annealing step or if the enzymes are left out of the closure
`reaction. We conclude, therefore, that two unit-length linear
`SV40 molecules have been joined to form a covalently closed(cid:173)
`circular dimer.
`Covalent closure of the hydrogen-bonded SV40 DNA di(cid:173)
`mers is dependent on Mg2+, all four deoxynucleoside triphos(cid:173)
`phates, E. coli DNA polymerase I, and ligase, and is inhibited
`by 98% if exonuclease III is omitted (Lobban and Kaiser
`first observed the need for exonuclease III in the joining of
`P22 molecules; we confirmed their finding with this system).
`Exonuclease III is probably needed to remove 3'-phosphate
`groups from 3'-phosphoryl, 5'-hydroxyl nicks introduced by
`the endonuclease contaminating the terminal transferase
`preparation. 3'-phosphoryl groups are potent inhibitors of
`E. coli DNA polymerase I (14) and termini having 5'-hydroxyl
`groups cannot be sealed by E. coli ligase (8). The 5'-hydroxyl
`group can be removed and replaced by a 5'-phosphoryl group
`by the 5'- to 3'-exonuclease activity of E. coli DNA poly(cid:173)
`merase I (7).
`
`Preparation of the Galactose Operon for Insertion into SV 40
`DNA. The galactose operon of E. coli was obtained from a
`"Advgal DNA; "Advgal is a covalently closed, supercoiled DNA
`molecule four times as long as SV40(II) DNA (Table 1). After
`complete digestion of "Advgal DNA with the R1 endonuclease,
`linear molecules two times the length of SV40(II) DNA are
`virtually the exclusive product (Table 1). This population has
`a unimodal length distribution by electron microscopy and ap(cid:173)
`pears to be homogeneous by ultracentrifugal criteria (Jackson
`and Berg, in preparation). The R1 endonuclease seems, there(cid:173)
`fore, to cut "Advgal circular DNA into two equal length linear
`molecules. Since one Rr endonuclease cleavage per "Adv mono(cid:173)
`meric unit occurs in the closely related "Adv-204 (Jackson and
`Berg, in preparation), it is likely that "Advgal is cleaved at the
`
`BEQ 1006
`Page 4
`
`

`
`2908
`
`Biochemistry: Jackson et al.
`
`Proc. Nat. Acad. Sci. USA 69 (197e)
`
`8
`
`7
`
`!I
`
`..
`
`20
`
`18
`
`16
`
`70 , ..
`
`~
`12 u
`-8
`3
`tO"
`~ ..
`
`8
`
`60
`
`!10
`
`..0
`
`30
`
`tO
`
`6
`
`..
`
`0
`
`~ .................. ~
`
`25
`
`A 6 C
`
`!I
`iO
`20
`1!1
`F"raction numlo4r
`bottom
`FIG. 5. Alkaline sucrose gradient sedimentation of annealed
`(3H]SV40(LRrexo)-[ 81P] (dA)so and [8H]Advgal-1110) (LRrexo)(cid:173)
`[32P] (dT)80 incubated for 3 hr with and without (P+L+III).
`Centrifugation was for 60 min. Line A, dA-ended SV40, plus
`dT-ended Advgal-1110 linea.rs, plus (P+L+III)(IIP, e; 8H, O);
`line B, dT-ended >.dvgal-1110 linears, plus dT-ended SV40 linea.rs,
`plus (P+L+III) (32P, &); line C,dA-ended SV40 linears, plus
`dT-ended Advgal-1110 linears, without (P+L+III) (31P, •).
`The arrows indicate the position in the gradient of supercoiled
`marker DNAs having the indicated multiple of SV40 DNA mo(cid:173)
`lecular size.
`
`same sites and, therefore, that each linear piece contains an
`intact galactose operon.
`The purified Xdvgal (La1) DNA was prepared for joining
`to SV40 DNA by treatment with A-exonuclease, followed by
`terminal transferase and [32P]dTTP, as described for SV40-
`(Lax).
`
`Formation of fJovalenay Closed-Circular DNA Molecules
`Containing both SV40 and Xdvgal DNA. Annealing of [3H](cid:173)
`[1H]Xdvgal(La1exo)-[32P](cid:173)
`SV40(Laiexo)-[12P](dA)so with
`(dT)so, followed by incubation with the enzymes, substrates,
`and cofactors needed for closure, produced a species of DNA
`(in about 15% yield) that sedimented rapidly in alkaline
`sucrose gradients (Fig. 5) and that formed a band in a CsCl(cid:173)
`ethidium. bromide gradient at the position expected for co(cid:173)
`valently closed DNA (Fig. 6). The putative Xdvgal-8V40
`circular DNA sediments just ahead of Adv-1, a supercoiled
`circular DNA marker [2.8 times the length of SV40(II)DNA],
`and behind Xdvgal supercoiled circles [4.1 times SV40(II)DNA]
`in the alkaline sucrose gradient. Electron microscopic measure(cid:173)
`ments of the DNA recovered from the dense band of the CsCl(cid:173)
`ethidium bromide gradient showed a mean contour length
`for the xnajor species of 2.95 ± 0.04 times that of SV40(II)
`DNA (Table 1). Each of these measurements supports the
`conclusion that the newly formed, covalently closed-circular
`DNA contains one SV40 DNA segment and one Xdvgal DNA
`monomeric segment.
`
`Omission of the enzymes from the reaction mixture pre(cid:173)
`vents Xdvgal-8V40 DNA forxnation (Figs. 5 and 6). No co(cid:173)
`valently closed product is detectable (Fig. 5) if Xdvgal and
`SV40 linear molecules with identical, rather than comple(cid:173)
`mentary, tails are annealed and incubated with the enzymes.
`This result demonstrates directly that the formation of co(cid:173)
`valently closed DNA depends on complementarity of the
`homopolymeric tails.
`We conclude from the experiments described above that
`Xdvgal DNA containing the intact galactose operon from E.
`coli, together with some phage X genes, has been covalently
`inserted into an SV 40 genome. These molecules should be
`useful for testing whether these bacterial genes can be intro(cid:173)
`duced into a xnamxnalian cell genome and whether they can
`be expressed there.
`
`DISCUSSION
`The methods described in this report for the covalent join(cid:173)
`ing of two SV40 molecules and for the insertion of a segment
`of DNA containing the galactose operon of E. coli into SV40
`are general and offer an approach for covalently joining any
`two DNA molecules together. With the exception of the for(cid:173)
`tuitous property of the R1 endonuclease, which creates con(cid:173)
`venient linear DNA precursors, none of the techniques used
`depends upon any unique property of SV40 and/or the Advgal
`DNA. By the use of known enzymes and only minor modi(cid:173)
`fications of the methods described here, it should be possible
`to join DNA molecules even if they have the wrong combina(cid:173)
`tion of hydroxyl and phosphoryl groups at their termini. By
`judicious use of generally available enzymes, even DNA
`duplexes with protruding 5'- or 3'-ends can be modified to
`become suitable substrates for the joining reaction.
`One important feature of this method, which is different
`from some other techniques that can be used to join unrelated
`DNA molecules to one another (16, 19), is that here the join(cid:173)
`ing is directed by the homopolymeric tails on the DNA. In
`our protocol, molecule A and molecule B can only be joined
`to each other; all AA and BB intermolecular joinings and all
`A and B intramolecular joinings (circularizations) are pre(cid:173)
`vented. The yield of the desired product is thus increased,
`and subsequent purification problems are greatly reduced.
`
`16
`16
`
`14
`
`12
`
`14
`
`12
`
`4
`"10
`b
`";;a
`§.
`"6
`~ 6
`4
`
`8
`
`2
`
`4
`
`2
`
`8
`
`7
`
`6
`
`5
`
`4
`
`3
`
`2
`
`8
`
`7
`
`6
`
`I>
`5 0
`r:
`4{)
`3
`3~
`~
`
`2
`
`1
`
`FIG. 6. CsCl-ethidium bromide equilibrium centrifugations of
`joined [8H]SV40(LRrexo)-[ 31P](dA)8o and [3H]Advgal-1110(LRr(cid:173)
`exo)-[81P] (dT)so DNA. The samples were those referred to in
`Fig. 5. Line A, dA-ended SV40 linears, plus dT-ended Advgal-1110
`linears, plus (P+L+III) (11P, e; 3H, O); lineB, the same mixture
`without (P+L+III) (81P, &; 3H, .t.).
`
`BEQ 1006
`Page 5
`
`

`
`Proc. Nat. Acad. Sci. USA 69 (1972)
`
`Insertion of the Galactose Operon into SV40 DNA
`
`2909
`
`For some purposes, however, it may be desirable to insert
`Xdvgal or other DNA molecules at other specific, or even ran(cid:173)
`dom, locations in the SV40 genome. Other specific placements
`could be accomplished if other endonucleases could be found
`that cleave the SV40 circular DNA specifically. Since pan(cid:173)
`creatic DNase in the presence of Mn2+ produces randomly
`located, double-strand scissions (17) of SV40 circular DNA
`(Jackson and Berg, in preparation), it should be possible to
`insert a DNA segment at a large number of positions in the
`SV40 genome.
`Although the Xdvgal DNA segment is integrated at the same
`location in each SV40 DNA molecule, it should be emphasized
`that the orientation of the two DNA segments to each other
`is probably not identical. This follows from the fact that each
`of the two strands of a duplex can be joined to either of the two
`W,....W W,....C)§
`strands of the other duplex e.g., C._..C or C._..W
`. What
`(
`possible consequences this fact has on the genetic expression
`of these segments remains to be seen.
`We have no information concerning the biological activities
`of the SV40 dimer or the Xdvgal-SV40 DNAs, but appropri(cid:173)
`ate experiments are in progress. It is clear, however, that the
`location of the Rr break in the SV40 genome will be crucial
`in determining the biological potential of these molecules;
`preliminary evidence suggests that the break occurs in the
`late genes of SV40 (Morrow, Kelly, Berg, and Lewis, in prep(cid:173)
`aration.
`A further feature of these molecules that may bear on their
`usefulness is the (dA·dT)n tracts that join the two DNA seg(cid:173)
`ments. They could be helpful (as a physical or genetic marker)
`or a hindrance (by making the molecule more sensitive to
`degradation) for their potential use as a transducer.
`The Xdvgal-SV40 DNA produced in these experiments is,
`in effect, a trivalent biological reagent. It contains the genetic
`information to code for most of the functions of SV40, all of
`the functions of the E. coli galactose operon, and those func(cid:173)
`tions of the X bacteriophage required for autonomous repli(cid:173)
`cation of circular DNA molecules in E. coli. Each of these
`
`§ The symbols W and C refer to one or the other complementary
`strands of a DNA duplex, and the" connectors" indicate how the
`strands can be joined in the closed-circular duplex.
`
`sets of functions has a wide range of potential uses in studying
`the molecular biology of SV 40 and the mammalian cells with
`which this virus interacts.
`
`We are grateful to Peter Lobban for many helpful discussions.
`D. A. J. was a Basic Science Fellow of the National Cystic
`Fibrosis Research Foundation; R. H. S. was on study leave from
`the Department of Biochemistry, University of Adelaide, Aus(cid:173)
`tralia and was supported in part by a grant from the USPHS.
`This research was supported by Grant GM-13235 from the
`USPHS and Grant VC-23A from the American Cancer Society.
`
`7.
`
`I. Sambrook, J., Westphal, H., Srinivasan, P. R. & Dulbecco,
`R. (1968) Proc. Nat. Acad. Sci. USA 60, 1288-1295.
`2. Dulbecco, R. (1969) Science 166, 962-968.
`3. Matsubara, K., & Kaiser, A. D. (1968) Cold Spring Harbor
`Symp. Quant. Biol. 33, 27-34.
`4. Radloff, R., Bauer, W., Vinograd, J. (1967) Proc. Nat. Acad.
`Sci. USA 57, 1514-1521.
`.5. Little, J. W., Lehman, I. R. & Kaiser, A. D. (1967) J. Biol.
`Chem. 242, 672-678.
`6. Kato, K., Goncalves, J. M., Houts, G. E., & Bollum, F. J.
`(1967) J. Biol. Chem. 242, 2780-2789.
`Jovin, T. M., Englund, P. T. & Kornberg, A. (1969) J. Bicl.
`Chem. 244, 2996-3008.
`8. Olivera, B. M., Hall, Z. W., Anraku, Y., Chien, J. R. &
`Lehman, I. R. (1968) Cold Spring Harbor Symp. Quant.
`Biol. 33, 27-34.
`9. Richardson, C. C., Lehman, I. R. & Kornberg, A. {1964)
`J. Biol Chem. 239, 2.51-258.
`10. Symons, R. H. (1969) Biochim. Biophys. Acta 190, 548-
`.5.50.
`11. Davis, R., Simon, M. & Davidson, N. (1971) in Methods in
`Enzymology, eds. Grossman, L. & Moldave, K. (Academic
`Press, New York); Vol. 21, pp. 413-428.
`12. Chang, L. M. S. & Bollum, F. J. (1971) Biochemistry 10,
`.536-542.
`13. Sheldon, R., Jurale, C. & Kates, J. (1972) Proc. Nat. Acad.
`Sci. USA 69, 417-421.
`14. Richardson, C. C., Schildkraut, C. L. & Kornberg, A.
`(1963) ColdSpring Harbor Symp. Quant. Biol. 28,9-19.
`15. Little, J. W. (1967) J. Biol. Chem. 242,679-686.
`16. Sgaramella, V., van de Sande, J. H. & Khorana, H. G.
`(1970) Proc. Nat. Acad. Sci. USA 67, 1468-14.7.5.
`17. Melgar, E. & Goldthwait, D. A. (1968) J. Biol. Chem. 243,
`44.09-4416.
`18. Morrow, J. F. & Berg, P. (1972) Proc. Nat. Acad. Sci. USA
`69, in press.
`19. Sgaramella, V. & Lobban, P. (1972) Nature, in press.
`
`BEQ 1006
`Page 6