0022-538X/79/11-0483/14$02.00/0
Organization of
Mouse Mammary Tumor
Virus-Specific DNA
Endogenous to BALB/c Mice
J. CRAIGCOHEN,`* JOHNE.MAJORS,2 ANDHAROLDE. VARMUS2
Department of Microbiology and Immunology, Tulane University School of Medicine, New Orleans, Louisiana70112,1 and Department of Microbiology, University of California, San Francisco,
California941432
Received forpublication13 June 1979
We used restriction endonucleases to prepare physical maps of the mouse
mammary tumorvirus(MMTV)-specificDNAendogenous to the BALB/c mouse
strain. The mapping was facilitated by the DNA transfer procedure, using
complementary DNAsspecific for the wholeandforthe 3' terminusof MMTV
RNA todetect fragmentscontaining viral sequences. The strategies used for the
arrangementof fragments into physicalmapsincludedsequential digestionswith
twoorthreeenzymes; preparativeisolationof EcoRIfragments containing viral
sequences;andcomparisons of virus-specific fragments derived from theDNAof
severalmousestrains. MostoftheMMTV-relatedDNAin theBALB/cgenome
is organized into two units (II and III) which strongly resemble proviruses acquireduponhorizontal infectionwithmilk-bornestrainsof MMTV and other
retroviruses. These units contain approximately 6.0 X 106 Mr of apparently
uninterruptedviral sequences, they bear redundant sequences totaling at least
700 to 800 base pairs at their termnini, and the terminal redundancies include
sequencesderived from the 3' end ofMMTVRNA. UnitsIIand III are closely
relatedinthat they share 12of14recognitionsites forendonucleases,butcellular
sequencesflankingunits II and III aredissimilarbythis criterion. Theremainder
of the MMTV-related DNA endogenous to BALB/c mice is found in a single subgenomic unit (unit I) withacomplexity ofca. 2x 106 Mr; thestructure ofthis unit has not been further defined. These results support the hypothesis that
endogenousproviruseshave beenacquired byinfection ofgerminal tissues with
MMTV. Thephysical maps arealso usefulforidentifyingtheMMTV genomes
endogenous to BALB/c mice in studies of the natural history of mammary
tumorigenesis.
DNArelatedtothegenomesof retrovirusesis
endogenoustonormal animals fromseveral
spe-cies (1, 34, 35), but the origin, structure, and function of suchvirus-specific DNA are
uncer-tain. Inparticular, the organization of
endoge-nousviral DNA has notbeen defined because
only
recently have techniques become availableto permit the physical mapping of sequences
occurringinlow numbers ofcopies in the
eucar-yoticgenome.
Afewcopies ofDNArelatedtothe genome of
mouse mammary tumor virus (MMTV) are
foundendogenouslyininbred strains ofmice, in
most wildmice, and in Asian mice (9, 18, 22, 24,
36).Genetically transmitted viruses with
onco-genic activitycanberecovered from some inbred
mousestrains(3, 20,23),indicatingthatatleast
some endogenous viral DNA must encode a
complete viral genome and be arranged as a
template suitable for transcription. In some
strains from which infectious virus hasnotbeen
recovered, most, if not all, of the genetic
infor-mation inoncogenic milk-borne viruseshas been
identified inendogenous MMTV DNA by
mo-lecular hybridization (18, 22), and the
endoge-nous viral DNA is often also transcribed into
RNA (19, 37). In these cases and others,
how-ever, little is known about the
organization
ofendogenous MMTV DNA.
Wehaverecently presented evidencefavoring
the hypothesis thatthe endogenous proviruses
ofMMTV wereacquired by
multiple,
independ-ent, andrelativelyrecentinfections of
germinal
tissues (5). Feral mice vary markedly in the
numbers of copies of MMTV DNA and the
positionsof viral DNA within the hostgenomes,
andinbred strains of miceappeartohave
inher-ited various combinations of viral DNA in a
manner suggesting that independentinsertions
of viral DNA have
segregated
likestablegenetic
markersduringinbreeding.
If endogenous proviruses were, in fact,
ac-483
on November 10, 2019 by guest
http://jvi.asm.org/
484 COHEN, MAJORS, ANI) VARMUS
quired byindependent infections ofgermlines,
theymightbeexpectedtoresemble structurally
theprovirusesfound in cells infected by
horizon-tally transmitted retroviruses. Proviruses
ac-quired byexperimentalinfectioncanbe
accom-modated at many sites in cellular DNA (4, 14,
16, 26, 28, 31), but they have similar features:
they are coextensive with unintegrated viral
DNA (and therefore approximately colinear
withviral RNA); theyareapproximately5 x 106
to 6X106Mrinsize; and they bearlarge terminal
redundancies composed of sequences derived
fromboth the 3' and 5' termini of viral RNA (4,
14, 16, 26, 28).
Inthis report,weused restriction
endonucle-ases to analyze the MMTV DNA endogenousto
onestrainof inbred mice (BALB/c)from which
noinfectious MMTV has been isolated in order
todetermine whether suchDNAisorganizedin
a fashionsimilar to that describedforproviruses
known to be introduced by infection. We
over-came thecomplexity of analyzing multiple units
of viral DNA in asingle cell by isolating EcoRI
fragments containingviral DNA before
second-ary digestions, and we used hybridization
re-agentsspecificforthe3'endof MMTVRNA to
search forterminal redundancies inthe
endog-enous viral DNA. The resultingphysical maps
revealthattheendogenous MMTV DNAinthis
strain is contained within threeunits located at
different positions in the host genome. Two of
these unitscloselyresemble MMTVproviruses
acquired byinfection withrespectto their com-plexity and terminal redundancies; the third unit
containsonlya portion of the MMTV genome.
Thus, the structure of most of theendogenous
MMTV DNA inBALB/cmice conforms tothe
hypothesis that this DNA was introduced by infection.
MATERIALS AND METHODS
DNA extraction. The liverfrom a male BALB/c mouse(Simonson Laboratories, Gilroy,Calif.)was ho-mogenized in DNA extraction buffer (0.02 M Tris-chloride [pH 8.11-0.01 M EDTA-0.1 MNaCl), using a motor-driven, Teflon-glass Dounce homogenizer. Afterdeproteinizationwithself-digested pronase and sodium dodecyl sulfate (SDS) (1 mg/ml and 1.0%, respectively) at 37°C for 12 h, DNA was extracted twice with equalvolumes ofphenol-chloroform (2:1). The preparation was dialyzed extensively against 5 mM Tris-chloride (pH 7.4)-0.1 mM EDTA before analysiswithrestriction endonucleases.
Restriction endonuclease digestion of DNA. EcoRI restriction endonucleases was a generous gift from P. Greene; PstI waspurified from Providencia stuartii by the procedure of Greene et al. (11); and
BglII and HpaI were obtained from New England
Biolabs (Beverly, Mass.). EcoRIdigestions were per-formed in 0.1 M Tris-chloride(pH7.4)-0.05 M
NaCl-0.005 M
MgC92-0.05%
Nonidet P-40. The reaction buffer for PstI included 0.006 M Tris-chloride (pH 7.2), 0.006 MMgC92,
0.05MNaCl,0.006M 2-mercap-toethanol, and100jugofgelatinperml.HpaIdigestion buffer consisted of0.01MTris-chloride(pH 7.5),0.01 M MgCl2, 0.006 M KCl, and 0.001 M dithiothreitol. ReactionswithBgllI wereperformedin 0.006 MKCl, 0.01MTris-chloride(pH 7.4),0.01MMgCl2,and 0.001 Mdithiothreitol. Reactionswere monitored for com-pletenessbyadding either lambdabacteriophageDNA(forEcoRI andBglll) or form Iofplasmid pBR313
(for PstI and HpaI) to a portion of the digestion mixture. A 5- to 10-fold enzyme excess was used in each reaction volume of0.1to 0.5ml, andinallcases reactionswereshown to becomplete by analysis of marker DNAsbygelelectrophoresis.
Gel electrophoresis, DNA transfer, and hy-bridization. After restriction endonuclease digestion, DNAsamples wereelectrophoresed in0.8% agarose (Seakem) gels containing 0.04 M Tris-acetate (pH 8.15),0.02Msodiumacetate, 0.018 MNaCl, and0.002 M EDTA(12).Bromophenol bluewasaddedas track-ing dye, andelectrophoresiswasperformedat 40 to 70
Vforatimesufficientfor thedyetomigrate12 to 18
cm.Ethidium bromide (2.5jig/ml)was usedtostain gels afterelectrophoresis for visualization of marker
DNAs (HindIll-cleavedlambdabacteriophage DNA
labeledwith32Pby nicktranslationwithEscherichia
coliDNApolymeraseI;15).Preparative fractionation
of DNAwasperformed under similar conditions, but thebuffer didnotincludeNaCl.Aspecial electropho-resisapparatus(17)wasusedtoprovide for discontin-uouselectroelutionfrom a0.8%agarosegel2cmthick and6 cmlong. Seventy fractionswere collected and analyzed individually by gelelectrophoresis and the DNA transfer-hybridization procedure of Southern (30).
For transfer of DNA to nitrocellulose sheets, we used the methoddescribed by Southern(30).Briefly, gels were treated with denaturing solution (0.5 N NaOH-1.5MNaCl)for 1 hand thenwithneutralizing buffer (0.5MTris-chloride[pH 7.0]-1.5M NaCl)for 3to 5h. DNA transfer wasaccomplishedbycapillary actionfromthegeltothe filterwith6xSSC (SSC =
0.15MNaCl-0.015M sodium citrate) for24 to 48h,
and thefilterwasdriedat80°Cinvacuofor2h. To detect virus-specific DNAbyhybridization, fil-terswereincubated with32P-labeledMMTV comple-mentary DNA (cDNA) synthesizedbyusingcalf thy-musoligomersasprimers (cDNArep; 16) or oligodeox-ythymidylic acid[oligo(dT)]asprimer(cDNA,3)with either sucrosegradient-purifiedvirion RNA(10 to 20
S)oroligo(dT)-cellulose-selectedMMTV RNA(10 to
20S; 33),respectively. SpecificityofviralcDNA's was evaluatedbyhybridization to filters with restriction endonuclease-digested unintegrated MMTV DNA (29). The annealing mixture included 3x SSC, 50% formamide, 0.05 M HEPES buffer (N-2-hydroxyethyl-piperazine-N'-2-ethanesulfonicacid; pH7.0), 200
jg
of yeast RNA per ml, 50jig of sheared and denatured salmon sperm DNA per ml,1 x 106to2x 106cpmof[32P]cDNA,andDenhardtbuffer (0.02% each ofbovine serum albumin, polyvinylpyrrolidone, and Ficoll; 6). Filters were incubated for12hat41°C with annealing mixlackingcDNAand then with 1 to 2 ml ofannealing J. VIROL.
on November 10, 2019 by guest
http://jvi.asm.org/
mix with cDNAfor 48 to 72 h. After incubation, filters werewashed for 1 h in 2x SSC at roomtemperature; in0.1xSSC-0.1% SDS at 50°C for 90 min; and with multiple rinses at roomtemperature with 0.1x SSC-0.1%SDS and then0.1x SSC only. After air drying, the filters wereexposed to KodakRP-Royal X-Omat film at -70°C in the presence of Dupont Cronex Lightning Plus screens (32). Reannealing filters with different cDNA's after initial analysis required that
filters be placed in Seal-a-meal food container bags
withannealing mixcontaining 1 x 106 to 2 x106cpm/ ml and incubated for 30 min in a 68°C water bath before incubation at 41°C for 48 to72h.
Nomenclature. To simplify identification of
BALB/crestriction endonuclease fragments contain-ingMMTV DNA, we have named them by letter and enzymein order ofmolecular weight, indicating the Mrx 106 in parentheses [e.g., EcoRI-A(10.0) is the largest EcoRI fragment containing MMTV-specific sequencesand hasanMr of10.0x
1061.
RESULTS
Restriction fragments of BALB/c DNA
containing
MMTV-specific sequences. Byannealing virus-specificcDNA tocellularDNA, we estimated that BALB/c mice contain
ap-proximately fivetosixcopiesofviral DNA per
diploid
genome (4, 22).Theextentof annealingofvirus-specificcDNAandRNA tonucleic acids
fromuninfectedcells indicatedthat most, if not
all, of the sequencesin thegenomes of horizon-tally transmitted MMTV were present in the germ lines of inbred strains. Analysis of the
kinetics andextentofannealing, however, does notindicate whether the mass ofvirus-specific
DNA isorganizedinto afewprovirus-likeunits or widely dispersed throughout the mouse
ge-nome in manydisordered,subgenomicunits.
An initial appraisal of this problem can be
obtained by digestionofcellularDNA with
var-iousrestriction endonucleases,followed by iden-tification of fragments containing
MMTV-spe-cific sequences with molecular hybridization. Enzymes chosen for this survey were among
thosepreviously testedfortheir ability tocleave viralDNAsynthesized afterinfection by MMTV
strains transmitted in the milk ofGR or C3H
mice (see Fig. 7A): EcoRI, which recognizes a
single site in the DNA of MMTV(C3H) or
MMTV(GR); PstI, which recognizes five sites;
BgIII, which cleaves at two sites; and HpaI,
whichdoesnotcleave such DNA (29). We made
the provisional assumption, later shown to be
correct, that suchenzymeswould also beuseful
forgenerating a
simple
physical map ofendog-enous MMTVDNA,althoughattheoutset we
did not know whether the maps ofendogenous
DNAwould bear anyrelationship to the maps
ofviral DNAsynthesizedafter horizontal
infec-tion. We also assumed, based upon annealing
experiments noted above,that cDNA's
prepared
from the RNA of milk-borne
MMTV(C3H)
would detectmostorall of the sequencespresent
within the units ofendogenous MMTV DNA.
Again, the results of mapping studies shown
below support the validity of this assumption, thoughminor differenceswould nothave been detected. The cDNA'sused
represented
either the entire genome ofMMTV(C3H)
(cDNA,p,
generated by random priming with oligomers of calf thymus DNA)orthe200 to 300nucleotides
atthe 3'terminus of viral RNA
[cDNA3,
pre-pared by priming
templates
ofpolyadenylated
viral RNA with oligo(dT)]. cDNA3 detects
se-quencesthat are redundantatthe ends of
un-integrated linear and
integrated
MMTV(C3H)
and
MMTV(GR)
DNA (4, 29).Occasionally,
minor bands were found on
single
autoradi-ograms but were not
reproducible
inmultiple
analyses (e.g., bandat3.4inFig.4,lane
C);
thesewere not considered in the
generation
of themaps.
Digestion ofBALB/c DNA with EcoRI
pro-duced five
virus-specific fragments,
all ofwhichannealed with both of the
MMTV-specific
cDNA's (Fig. 1A and 1E; Table 1; 4, 5). This result suggested that the endogenous MMTV
DNA is not widely scattered in small units throughoutthe mouse genome. Itwouldaccord
well
with the estimates of copy number from association kineticsif,
forexample,
twoproviral
elements had a single internal EcoRI site and the third lacked an internal site. In this case,
fourfragments wouldbe
generated
fromthefirsttwo proviruses, and all four
fragments
would anneal withcDNA3,
assuming
theproviruses
bore terminal redundancies that includedse-quencesfromthe 3' end of viral
RNA;
the thirdproviral
element would becontained
within thefifth
fragment.Digestion ofBALB/cliverwithPstI (Fig. 1B and 1F; Table 1) yielded a larger number of virus-specific fragments,aswould have been
pre-dicted from therestriction map of
horizontally
transmitted MMTV DNA. Moreover, some of
these fragments have the same molecular
weights (e.g., 1.1 x
106
and 0.9 x106
Mr)
as
fragments derived from
internal
regions ofMMTV(C3H) and MMTV(GR), suggesting that
homologousregionsmight be present in
endog-enousproviruses. HpaI,ontheotherhand,
gen-eratedfragments
smaller
than would havebeenexpected ifit had failed to cleave provirusesof
standard size (ca. 5.9x
106
Mr; Fig. 1Cand1G),implying either that the endogenous MMTV
DNA contained recognition sites for HpaI or
thatsomeof the DNAwas
organized
insubge-nomicunits.
BglII
alsogeneratedamodestnum-VOL. 32,1979
on November 10, 2019 by guest
http://jvi.asm.org/
486 COHEN, MAJORS, ANI) VARMUS
*
I;
FIG. 1. MMTV-specific fragments obtained from BALBIcliverDNAbydigestionwith various restric-tionendonucleases. Samples (10,ug) of high-molecu-lar-weight DNA extracted from the liverofa male BALB/c mouse were digested with the designated
restriction endonuclease as describedin Materials
andMethods. Thesamplesweresubjectedto electro-phoresisinan0.8% agarosegel,and DNAfragments
weretransferredtonitrocellulosesheets(30), hybrid-izedwith 32P-labeledMMTV-specificcDNArep(lanes A-D)orcDNA3 (lanes E-H), anddetectedby auto-radiography. A HindIII digest of lambda bacterio-phage 32P-labeledDNA(not shown)wasincludedto
obtain theMrx 10' valuesindicatedontheleft.A andE,EcoRIdigests;B andF, PstIdigests; Cand G, HpaI digests;D andH,BglII digests.
ber ofvirus-specific fragments (Fig. 1D and 1H)
without any obvious relationship to the BglII
fragmentsexpected from provirusesacquiredby
infection.
Although definitive deductions would have
beenprematureatthisstageof theanalysis, the
variations inintensity of bands and the
discord-ances between annealing with cDNA3 and
cDNArep could clearly be helpful in ordering
fragments into a restriction map. Forexample,
a large fragment detected readily with cDNA3'
but poorly with cDNArep [e.g., PstI-A(3.4) and
Pstl-D(1.2)] wouldprobably contain mostly host
DNA linked to the end of a provirus; on the
otherhand,asmall fragment thatannealed
ex-tremelyefficiently with either cDNA [e.g.,
PstI-H(0.5)] might be derived from more than one
unit ofendogenousviral DNA.
Preparation of fractions of
EcoRI-cleaved
BALB/c DNA containing
separated
virus-specific
fragments.Cleavage
ofBALB/c
liverDNA withvarious enzymesindicated that
rela-tivelysimple patternsofvirus-specific fragments couldbe obtained, but thepatternsseemedtoo
complicated
topermit
construction ofphysical
maps forindividual unitsofendogenousMMTV
DNA bydirect
sequential
digestions. Wethere-fore elected to separate on a preparative scale
some ofthe EcoRI fragments ofvirus-specific DNA before attemptingto chartmore detailed
mapsby secondary cleavages. The EcoRI
frag-ments were chosen because each fragment
seemed likelytoprovide uswith alargeportion
or
all
ofanendogenous element, basedupon thetentativeinterpretation offeredinthepreceding section. One obvious limitation ofthe choiceof
theEcoRI fragmentswastheinherent
difficulty
ofseparating
EcoRI-C(5.0)
andEcoRI-D(4.7).
Five
milligrams
ofDNAfromBALB/c liverswasdigestedwith
EcoRI
andsubjected topre-parative
electrophoresis
in a thick, short 0.8%agarose gel, using an apparatus equipped for
timed
collection
ofDNAfragmentsexitingfromTABLE 1. Virus-specific fragments from restriction endonucleasedigests ofBALB/cDNA
Frag- M Detectedwith:
Enzyrne mentment Mr 06
cDNAr,p
cDNA3EcoRI A 10.0 + +
B 6.0 + +
C 5.0 + +
D 4.7 + +
E 4.0 + +
PstI A 3.4 - +
B 3.3 +
-C 3.1 +
-D 1.2 - +
E 1.1 +
-F 1.0 + +
G 0.9 + +
H 0.5 + +
I 0.2 - +
HpaI A 4.4 + +
B 4.2 + +
C 1.1 - +
D 0.9 - +
E 0.6 + +
BglII A 4.0 + +
B 3.5 + +
C 2.8 + +
D 2.5 + +
E 1.3 + +
F 1.2 +
-a These v-alueswereobtainedbycomparison ofthe data shown inFig. 1 with HindIII digests of32P-labeled
lambda bacteriophage DNA electrophoresed in par-allellanes.
,J. VIROL.
on November 10, 2019 by guest
http://jvi.asm.org/
[image:4.507.270.456.368.624.2]the end of the gel (see Materials and Methods).
Seventy fractions were then analyzed by
re-peated electrophoresis, DNA transfer, and
mo-lecular hybridization for their content of
virus-specific fragments. Four pools (subsequently
calledI, II,III,and IV) were made from fractions
found to contain the EcoRI fragments
previ-ously identified in digestions of total BALB/c
liverDNA (Fig. 2). Pool I contained the
EcoRI-E(4.0) fragment, apparently uncontaminated
with othervirus-specific fragments; poolII
con-tained theEcoRI-C(5.0) andEcoRI-D(4.7)
frag-ments,contaminated with asmallamountof the
EcoRI-E(4.0) fragment; pool III contained the
EcoRI-B(6.0) fragment,
minimally
contami-nated withsmallerfragments (notvisible in the
test shown in Fig. 2C); and pool IV contained
A
B
C
D
E
.
t0
.5.0
4.7
4.0
FIG 2 Fractionated EcoRI
fragments
ofMMTV-specificDNA
fr-om
BALBIcmouse liver DNA. Fivemilligrams ofBALB/cliver DNAwasdigestedwith EcoRI andsubjectedtopreparative electrophoresis through a 0.8% agarosegel on a
fr-actionating
gelelectrophoresisapparatus (seeMaterials and
Meth-ods). Seventy
fr-actions
were obtained by collectingelectroelutedmaterialat
fr-equent
intervals,and eachfr-action
wasanalyzed by gel electrophoresis, transferto nitocellulose sheets, and hybridization with
MMTVcDNAp.
Fourpoolsweremadefrom
multiplefr-actions
which containedvirus-specificEcoRIfr-ag-ments. A sample of each pool was reanalyzed to evaluatehomogeneity.MrX 10'6foreachviral
fr-ag-ment is indicatedon the rightand wasdeterminedby including32P-labeledHindjjjlaMbdaDNA in a
parallel lane (not shown). A,poolI; B,pool
1I;
C,pool III; D, poolIV; E,
unfr-actionated
EcoRI-di-gestedBALE/cliver DNA.
the EcoRI-A(10.0) fragment. The amount and
purity of these fragmentsweresufficientto
pro-ceed withsecondarydigestions. In some cases,
however, fragments apparently derived from
secondary digestion of contaminating (usually smaller) EcoRI fragments complicated the anal-ysis, butthese could generally be identified as
contaminants without majordifficulty (see, for
example, fragments of2.8 x106 and2.6x106Mr
inFig.3Corthe fragment of1.4 x106MrinFig.
4B).
Secondary cleavages of isolated EcoRI fragments of MMTV DNA.Secondary diges-tions of the DNA in the four pools of EcoRI fragments with PstI, HpaI, andBglII (Fig.3-5;
Table 2) permitted us to make several
deduc-tions importanttothe construction of restriction endonucleasemapsforthe MMTV DNA
endog-enous toBALB/c mice. The following guidelines
were helpful in sorting out the complex data obtained inthese experiments.
First, fragmentsthat arenoveltothe
sequen-tial digestion products (i.e., notpresent among
products of digestion with the second enzyme
alone; Fig. 1andTable 1)mustbepositionedat an end of the EcoRI fragment used for the secondarydigestion.Forexample, PstI digestion of EcoRI-C(5.0) and EcoRI-D(4.7) generated major, novel fragments of2.8 x
106
and 2.6 x106
Mr (Fig. 3B); the appearance of these twofragments indicated that EcoRI sitesmust be
present in PstI-B(3.3) and PstI-C(3.1), as
con-fixned by the absence of these two PstI
frag-ments in all sequential digestions with EcoRI
and PstI (Fig. 3; also see below). Similar
argu-ments apply to the fragments of1.4 x
106,
3.0x
106,
and 1.2x106 Mr derived by HpaI cleavageofDNAfrom pools I, II, and III, respectively, andtothe fragments of0.8x
106
and approxi-mately0.3x 106 Mr obtained byBglII digestion ofDNAfrompoolsI, II, and III.There is a significant possible exception to
this firstrule, in that prior fractionation of the EcoRI
fragments
maypermit
detection offrag-mentsthat have insufficient amountsof
virus-specific sequence to be observed in digests of
unfiactionated
DNA. Thereis,
forexample,
nootherway to
explain
therelatively
faint bands representing fragments of1.5x106
and1.3x106
Mr observed after
HpaI digestion
ofEcoRI-0(5.0)
andEcoRI-D(4.7) (Fig. 4B).Infact, these fragmentsappear tobecomposedlargely
of cel-lularsequences(seebelow), making detection of the minor virus-specific sequences difficult inunfractionatedDNA.
Second, when PstI, HpaI, and
BglII
fragments
ofMMTV DNA(Table 1)arealso foundamong
products of secondarydigestions of EcoRI
on November 10, 2019 by guest
http://jvi.asm.org/
488 COHEN, MAJORS, AND) VARMUS
.a.
*.: ;.'.;:'t :a
::.:::: *: :.:: !.:::.
::
..
':
:...
*: :.:.
+ '".1
..::..
*s *..,'..:. :'.
....:
Pf%.
.:
# At #iiQ
FIG. 3. Products of PstI digestion of separated EcoRIfragmentsof BALB/cmouseliverDNA. Sam-plesof each poolofEcoRIfragment (seeFig. 2)were digested with PstI,electrophoresedon0.8%oagarose gels, and transferredto nitrocellulose sheetsfor hy-bridizationwith either[32P]cDNArep (lanes A-D)or
[32P]cDNA:s (lanes E-H). HindIII-digested
32P-la-beled lambda bacteriophage DNA (notshown) was
includedasmarkertoobtaintheMrX1o-6shownon theleft.AandE,pool I;B andF, pool II;C andG, pool III;D andH, poolIV.
ments(Table2), thesefragmentsarelikelytobe
positionedinternallyin therelevantEcoRI
frag-ment. However, it is possible to generate by
thesedoubledigestionsnewfragments,bounded
by anEcoRI site, whichfortuitously comigrate
with products of digestion with the second
en-zymealone. For example, thePstI-EcoRI
frag-ments of0.5 x 106are likelytobe amixture of
PstI-H(0.5) and theproductsof EcoRIdigestion
ofPstI-B(3.3) andPstI-C(3.1) (Fig. 3Band3F).
Thisinference is substantiatedin alater section
(see Fig. 8).
Third, fragmentslargerthan 0.3 x
106
Mrthatare detected principally or solely with cDNA3
N [e.g., PstI-A(3.4) or PstI-D(1.2)] mustbe
com-N 4 posedmostly ofcellularDNA, since
cDNArep
iscapable of detecting fragments as small as 0.3
X106Mr,whicharelikelytobecomposedmostly
or entirely of viral sequences (see Fig. 5A and
5C).
Fourth, the total molecular weight of
virus-0. 9
n
:f, S jKfa s
.
az ffi
::::
:.'i: ::
.._....':7'....'.
*..je'';.,.:.
U:s3: :; : ..:3t;i.ir!s-
-.i i;
'::*: ....:...'.'i:'?
..;.' '.'^:'...
....?e::
....,.,',
.... i,....eN.:: ::.:.:.:.''S5?. B. :.s.'.2@ ?$'
!g;:;;gj;jE...;,.;..
4
t = asit
Q _ -X.
;e-4,'t.,...:
,,4..:}i::.:8:''
:t.}
'-'.8':'4::;.. ..::.; § + ... ... :. ::.:..
::.:
.: :...
.. ..
.; ..
..
..:
...e.
.::.:
FIG. 4. Products of HpaI digestion ofseparated EcoRIfragmentsfromBALB/cliver DNA.HpaI was used todigestsamples ofpoolsofEcoRIfragments derivedfromBALB/cliver(Fig. 2); the products were analyzed with
cDNAr,,
(lanesA-C) andcDNA3(lanes D-F) as described in thelegend toFig. 1. Numbers and arrows on theleftindicate theMrx10-'of viral fragmentsasdetermined by inclusion of radiolabeled HindIIIfragmentsoflambda DNA (notshown) in a parallel lane. A and D, poolI; B and E, pool II; C andF,pool III.i
I
.J. VIROL.
on November 10, 2019 by guest
http://jvi.asm.org/
[image:6.507.111.451.72.555.2]MMTV DNA 489
cD
NArep
A
B
C
3.5_
2.5 -e
1.2-0.8
--*0.3
-4cDNA3
[image:7.507.55.253.78.478.2]D
E
F
FIG. 5. Products ofBglII digestion ofseparated EcoRIfragments fromBALB/cliver DNA.Samples ofthe variouspools ofEcoRIfragments shown in Fig.2weretreatedwithBglII, subjectedto electro-phoresisin0.8%agarosegels, transferredto nitrocel-lulose sheets, and analyzed by hybridization with eithercDNAp(lanesA-C)orcDNA3(lanesD-F).Mr
X 10-6 indicated onthe left are basedon
bacterio-phage lambda DNA markers included in adjacent
lanes(notshown).AandD, poolI;B andE, poolII; C andF, poolIII. Thefaintbandin lane Catthe position ofca. 3.2 x 106 Mr was not observed in repeatedtestsand hasthereforenotbeen considered furtherin theconstructionofmaps.
specific fragments derived from the secondary
digestions provides an upper limit for the
amountof viralDNA ineachEcoRI fragment.
Forinstance, there canbe no morethan 2.8 x
106Mr of viral DNAin theEcoRI-E(4.0),based
upon the sum of the fragments generated by
secondary digestionwith PstI (Fig.3A). Usingtheseprinciplesand theircorollariesas
guides, it was possibleto draw several
conclu-sions about the viral DNA in the five EcoRI
fragments and,in somecases,toestablish maps
forPstI, HpaI, andBgIII sites within theviral
andcellular DNA in thosefragments.
First, EcoRI-A(10.0) contains less than2.5 x
106
Mr
ofviralsequences, and the EcoRI sitesappeartobelocatedincellular DNA outsidethe
viral DNA (Fig. 3Dand3H). Thus, thisEcoRI fragment harbors a subgenomic unit of viral
DNA which we have termed unit I. Limited recovery ofEcoRI-A(10.0) during fractionation proceduresand the unusualstructure of unit I have compromised further characterization. However, previous data
concerning
thesegre-TABLE 2. Virus-specific fragments from secondary restriction endonuclease digests ofpools of EcoRI
fragments of BALB/c DNA
Pool Detectedwith:
of EcoRI
Enzyme EcoRI frag- M,x 10'
frag- ments cDNA,p cDNA3' ments
PstI I E(4.0)b 1.2 - +
1.1 +
-0.5 + +
II D(4.7) 2.8 +
-C(5.0) 2.6 +
-0.5 + +
0.2 - +
III B(6.0) 3.4 - +
1.1 +
-0.5 + +
IV A(10.0) 1.0 + +
0.9 + +
HpaI I E(4.0) 1.4 +
0.9 - +
0.6 + +
II D(4.7) 3.0 + +
C(5.0) 1.5 +
-1.3 +
-III B(6.0) 3.7 - +
1.2 +
-0.6 + +
BglII I E(4.0) 1.3 + +
1.2 +
-0.3 +
-II D(4.7) 3.5 + +
C(5.0) 0.8 +
-III B(6.0) 2.5 + +
1.2 +
-0.3 +
-aThese
values
were obtained fromHindIII-digested32P-labeledlambdabacteriophageDNA
electrophore-sesin aparallellane.
bMr X 10-6 of EcoRI
virus-specific fragments
indesignatedpool.
on November 10, 2019 by guest
http://jvi.asm.org/
490 COHEN, MAJORS, ANI) VARMUS
gation of endogenous MMTV proviral DNA
in-dicatedthattheEcoRI-A(10.0) fragment
segre-gated as anindependent unit (5).
Second, each of the remaining four EcoRI
fragments contains up to 3 x 106to 4 X 106Mr
ofviral sequences, including sequences specific
tothe 3'end of MMTV RNA.
Third, the products of secondary digestions
indicate that EcoRI-B(6.0) and EcoRI-E(4.0)
areeach derived from one of two units of viral
DNA (units II and III) that also yield
EcoRI-C(5.0) and EcoRI-D(4.7). As one example of evidence for this conclusion, the fragments
gen-erated by EcoRI cleavage of HpaI-A(4.4) and
HpaI-B(4.2)mustbe the 3.0x106Mrfragments
in theHpaI digest ofEcoRI-C(5.0) and
EcoRI-D(4.7) (Fig. 4B) and the 1.4 x 106and 1.2 x 106
Mr fragments in the HpaI digests of
EcoRI-E(4.0) and EcoRI-B(6.0), respectively (Fig. 4A
and4C).Inother words, these data locate EcoRI
sites within HpaI-A(4.4) andHpaI-B(4.2).
Sim-ilar arguments can be made about the location
ofEcoRI sites withinPstI-B(3.3) andPstI-C(3.1)
andwithinBglII-F(1.2).
Fourth, within each of the four EcoRI
frag-mentsfrom thesetwounits, variousHpaI, PstI, andBglIl fragments unaffectedby EcoRI diges-tioncanbearranged (see Table2anddiscussion
ofFig.7,below). Theorderingwasbasedinitially
upontherelationshipof fragment size and rela-tive efficienciesofannealingwith cDNArep and
cDNA3,asdiscussedearlier. Since the
complex-ityof viral sequenceswasconsiderably lessthan
the Mr of each EcoRI fragment (see second
conclusion, above) and since eachEcoRI
cleav-age appeared to affect only one PstI, BglII, or
HpaI fragment (seethirdconclusion,above),we
deduced that the viral sequenceswere
approxi-mately contiguous at one end of each EcoRI
fragment. Theproposalsfor overlappingsetsof
PstI, BglII, and HpaI fragments (see Fig. 7)
werefurther validatedby additional sequential
digestionspresentedinalater section.
Fifth,EcoRI-B(6.0) andEcoRI-E(4.0) appear
to be derived from closely related regions of
MMTVprovirusesasindicated by commonPstI
fragments (1.1 x 106 and 0.5 x 106Mr; Fig. 3A
and3C) andBglIIfragments (0.6x
106
Mr; Fig.5Aand5C).
Usinggenetic segregationof endogenous
viral DNA to facilitate mapping. To
deter-mine which ofthefoursmallerEcoRIfragments
came from the same proviral elements, we
ex-ploited our recent finding that endogenous
MMTV DNA has been segregating like stable
genetic elements during the inbreeding of
con-temporarymousestrains (5). By analysisof
en-donucleasedigests ofDNA from severalstrains
among the progeny ofa cross between a Bagg
albino (forerunner of BALB/c) and a DBA
mouse, wefound thatindependentlysegregating
elements generated pairs of EcoRI fragments
[e.g.,EcoRI-B(6.0) and EcoRI-D(4.7),or
EcoRI-C(5.0) and EcoRI-E(4.0)]; PstI-C(3.1) was also
foundtobe associatedwith the presence of the
former pair of EcoRI fragments. Pertinent
ex-amples of such resultsare illustrated in Fig. 6.
AnEcoRIdigest of liver DNA fromaC3H/HeJ
mouse (lane F) contained EcoRI-C(5.0) and
EcoRI-E(4.0), but lacked EcoRI-B(6.0) and
EcoRI-D(4.7). The PstI digest of C3H/HeJ
DNA did not containPstI-C(3.1), which would
beexpectedtoproduce as strong a band as PstI-B(3.3); however, there was a less readily detected PstIfragment of ca. 3.1 x
106
Mr derived from anendogenousprovirus unique to the C3H/HeJmouse (lane B). These observations indicate
thatEcoRI-B(6.0) and EcoRI-D(4.7)form a sin-gle proviral unit, which we call unit II, and
EcoRI-C(5.0) and EcoRI-E(4.0) form another
proviralunit, called unit III. They also confirm that the PstI-B(3.3) fragment is from unit II and the PstI-C(3.1) fragment is from unit III. The
.0
FIG. 6. Comparison of restriction endonuclease patterns of MMTV-specific DNA in BALB/c and C3H/HeJmouseliver DNAs. A 10-,igsampleofDNA extractedfromthe liverofeitheramaleBALB/cor
a male C3H/HeJ mouse was digested with PstI,
HpaI, or EcoRI and analyzedfor MMTV-specific fragments with
cDNA,
asdescribedinthelegendto Fig. 1. M,x 106 ofsignificant viralfragments are indicated to the leftofeachpanelandwere deter-minedfrom HindIII-digested lambda[32P]DNA in-cluded inparallellanes(not shown). PstIdigests of BALB/c (A)andC3H/HeJ (B) DNAs; HpaI digests ofBALB/c (C) andC3H/HeJ (D) DNAs; EcoRI di-gestsofBALB/c (E)andC3H/HeJ (F)DNAs.eJ. VtIROL,.
OWN.
AL,--11.
)w 40,
on November 10, 2019 by guest
http://jvi.asm.org/
[image:8.507.266.457.338.524.2]additional HpaI site inone of thesetwo
provi-ruses can be mapped by comparison of HpaI
digestionsof DNAs from these two strains (lanes
Cand D). TheHpaI-A(4.4) fragment is common
tobothBALB/c andC3H/HeJ DNAs,
indicat-ing that the HpaI-B(4.2) fragment is derived
from unit III. [The minor fragment of
approxi-mately4.2 x106Mrin C3H/HeJ could represent
either a sequence unique to this strain or a
comigrating fragment from HpaI-B(4.2) in
di-gestsofBALB/cDNA and derivedfromunitI.]
Testing proposed physical maps of en-dogenous MMTV DNA with additional se-quential digestions. With the aid of the ge-neticarguments presented in the previous
sec-tion, it ispossibletoorganizeall ofthe fragments
obtained by secondary PstI, BglII, and HpaI digestionsofEcoRI fragmentsBthrough E (Fig.
3-5)intophysicalmapsofunits II and III (Fig.
7). Theonly arrangement fullyconsonant with
therulesgoverning interpretationof the
double-digestion productsindicates that these units
con-tain approximately 6.0 x
106
Mr of apparentlycontinuousviralsequences, withsequences from
the 3' end of viral RNA positioned near both
ends of both units. In the proposed terminal redundancies, therearethreesites (two PstI and one HpaI) that have the same relationship to
eachotheratboth endsof both units; this finding
supports the evidence for redundancies from
annealings with
cDNA3-,
suggests that theter-minal redundanciesin units IIandIII are very
similar, and shows that thesize of the redundant
sequence is atleast 700 to 800bases. The
ori-entation of themapswithrespect toviral RNA
was basedupon the evidence (discussed below
inrelationtoFig.9) that
cDNA3-
detectedonlythe 100 to 200 nucleotidesadjacentto
polyade-nylic acidatthe3' end of viral RNA. Thus, we
placed to theright the ends of unitsII and III
yielding fragments thatwere composed largely ofcellular DNA and annealed efficiently with
cDNA3.
In these constructions, the two unitsappearclosely related in that they sharemany
restriction sites within the proposed viral
se-quences, but the sites in flanking regions of
cellular DNA are different. Thus, units II and IIIdifferonly withrespect to oneadditional site
forPstI and one for HpaI in unit III, with 12
sitesheld incommon;inflanking cellular DNA,
however, allsites except aBglII site to theleft
of viral DNA are independently positioned in
relation to the viral DNA of units II and III.
(The similarpositions of theBgII sitemay be
coincidental,butfurtherstudyof thehomology
of the cellular sequences flanking these
provi-rusesiswarranted.)
To challenge the proposed maps, we per-A
MMTVIC3H) DNA
F3 R
F3-.I
_w p
P Pp P P
s CELLDNA.-PROVIRAL DNA CELLDNA
I
H1
UNITSU
H? H
I
1
HH H
R || | R
I? H H HH
l l
Al l
I p p0
MrX10-6
H
R
[image:9.507.261.455.70.338.2]9 10 11 12 13
FIG. 7. Physicalmapsofrestrictionendonuclease
siteson unintegrated MMTV(C3H) DNA (panel A;
29)andontwounitsof endogenousMMTVDNA in the BALB/cmousestrain(panel B).Datapresented
inFig. Ithrough6and Tables1 and 2are
summa-rizedaslinearmapsof bothviralandflanking cell sequencesindesignatedunitsIIand IIIof endoge-nousMMTV DNA. Therelativeorientation of
unin-tegratedMMTV DNAto viralRNA isindicatedby the RNA molecule at the top of panelA and the relativepositions of viral and cellularsequences are
indicatedat the topof panelB; the boxes indicate terminally redundant sequences which include
se-quencesfrom the 3'endofviralRNA. The scaleat
the bottomindicatesMrX10-6.Specificcleavage sites
forEcoRI(R),PstI(P), BglII (B),andHpaI (H) are
indicatedfor unintegratedandendogenous (unitsII
andIII)MMTVDNAs.Questionmarks indicatesites notmappedon aspecificunitofMMTV DNA.
formed additional sequential digestions with
various combinations of the fourenzymesused
in thisstudy (Fig. 8 and9). Several features of themapswereconfirmedbythese tests.
First, the datainFig.3suggested,butdidnot
prove,that EcoRIcleavedPstI-B(3.3) and
PstI-C(3.1) from units II and III to generate
frag-mentsof 0.5 x 106Mrfrom theright-handside
of theEcoRI site; these fragmentspresumably
comigrated with PstI-H(0.5), which is derived
from the terminally repeated structure and
hence from all EcoRI fragments from units II
andIII.SincethemapspredictthatPstI-H(0.5),
MMTV(C3H) RNA
(A)n
Iu0 1p p p p
on November 10, 2019 by guest
http://jvi.asm.org/
492 COHEN, MAJORS, AND VARMUS
Second, neither the 0.5 x 106 nor 0.4 x 106
fragments inFig. 8, lanes B andF, annealed with
cDNA3' (datanotshown), indicatingthatHpaI
digestion removed the portion of PstI-H(0.5) capable of reacting with this reagent. This result
supports the proposed orientation of the maps
with respecttoviral RNA bydemonstratingthat
cDNA3' is specific foronly100to200nucleotides
atthe 3' end of viral RNA.
Third, the experiments in Fig. 8 also show, as
predicted by the maps, that PstI-E(1.1) from
unit II cleaved byHpaI toproduce a fragment
of 0.95 x 106Mr (laneB); PstI-E(1.1) fromunit
III was cleaved at an additional HpaI site to
yield a fragmentof 0.6 x 106Mr (lane F). (The newfragmentsof0.1 x 106 to 0.15x106Mr tobe
expected afterdigestion with EcoRI, PstI, and
HpaIwerepresumablytoosmalltobe detected
with ourmethods.)
-9
4
wi+;i''''+
+~
ot
1*
i,
r!:
:---*
04A
....urn
4.
U:
FIG. 8. Virus-specificproductsof digestionof
sep-arated EcoRIfragments from BALB/c liver DNA withtworestrictionenzymes.PoolscontainingEcoRI fragments(seeFig. 2)weredigestedwith either PstI (A, C, and E) orPstI andHpaI (B, D, andF) and analyzed asdescribed inFig. 1 with
[32P]cDNArep-Mrx10-6ofviralfragmentsareindicatedontheleft
and determined by inclusion of HindIII-digested
lambda[32P]DNA. DNA frompool Idigestedwith PstI (A) or PstI andHpaI (B); DNA from pool II digestedwith PstI (C) orPstI andHpaI (D);DNA from pool IIIdigestedwith PstI(E)orPstI andHpaI
(F).
but not the 0.5 x 106 Mr fragments from the
double digestion,should be cleaved with HpaI,
the products of PstI digestion ofEcoRI-E(4.0)
andEcoRI-B(6.0) weresubjected to redigestion
with HpaI (Fig. 8, lanes A, B, E, and F). As
predicted, the relative intensity of the bands in
theregion of0.5 X
106
Mrwasdiminished aftertheHpaIdigestion, andnewfragments of 0.4 x
106
Mrweregenerated (compare lanesA and E [image:10.507.62.256.64.428.2]withlanesBandF).
FIG. 9. Virus-specific products of BALB/c liver DNA digested with two restriction endonucleases. Ten-microgram samples ofliver DNA weredigested sequentially with two restriction enzymes and ana-lyzedwithcDNArrepasdescribed in thelegendtoFig. 1.MrX10-6ofviralfragmentsweredeterminedfrom HindIII-digested 32P-labeledlambda bacteriophage DNA and are indicated to theleft ofeachpanel.The enzymes used were:(A)PstI;(B)PstI andEcoRI; (C) PstI andHpaI; (D)EcoRI andBglII; (E) BglII.
J. VIROL.
2-6
.1,A ., ..,:..
Affiwla-", III&. .,'IRW!,:.:i
4'10'
Z.,V,.:.
:4."';.1... V.,:
t
itij
p
.-y 4.0
It. .,:... S.:.A:..
-t
.,444W
on November 10, 2019 by guest
http://jvi.asm.org/
[image:10.507.264.458.276.567.2]Fourth, themapspredict that the products of
double digestion with PstI and EcoRI should
includetwofragments from the left side of each
provirus, oneof2.8x 106or2.6x 106Mrwhich
isnot susceptibletorecleavage with HpaI and
oneof 0.5 x 106Mr [PstI-H(0.5)]which bearsa
site for HpaI.Sinceweproposethat the ends of
units II and Ill areterminallyredundant, the0.5
X 106 Mr fragments and their HpaI cleavage
products should be identical to those derived
from the right side of units II and III. All of
these predictions are clearly fulfilled, using
EcoRI fragments B and E from the right sides
ofunitsIIand III (Fig. 8, lanes A, B, E, and F)
and EcoRI fragments C and D from the left
sides(Fig. 8, lanes C and D).
Fifth,severalof theresultsillustratedin Fig.
8wereconfmed and extended by redigestion of
PstI fragments inunfractionated BALB/c DNA
witheitherEcoRIorHpaI (Fig.9,lanesA-C).
Thus,EcoRIcleaved onlyPstI-B(3.3)and
PstI-C(3.1) of those PstI fragments detectable with
cDNA,p;
the products of the double digestionwere2.8x 106, 2.6x106, and 0.5 x 106 Mr. HpaI
didnotcleavewithinPstI-B(3.3)orPstI-C(3.1),
butgeneratedfragmentsof 0.95 x 106, 0.6x 106,
and 0.4 x 106Mr bydigestion of PstI-E(1.1) and
PstI-H(0.5).
Sixth, the maps for units II and III predict
that BglII-F(1.2) represents not one but two
fragments which are adjacent in the internal
region of both proviruses. One of the
BglII-F(1.2) fragments from each unit should be
cleaved by EcoRI to produce fragments of 0.8
x
106
and 0.3 x 106 Mr; the otherBgIII-F(1.2)fragmnents should remain intact. Thesepredicted
resultswere observed afterredigestion of BglII
fragments ofunfractionatedBALB/c DNA with
EcoRI (Fig. 9, lanes D and E). The relative
intensity of the band at 1.2 x 106Mr was
de-creased, and the appropriate new bands
ap-peared. EcoRI also cleaved BglII-A(4.0); this
fragmentmust be derived from unit I, but the
products of the secondary digestion with EcoRI
havenotbeenidentified.
DISCUSSION
In this report,weusedrestriction
endonucle-ases to prepare physical maps of the
MMTV-related DNA endogenous to the genome of
BALB/c mice.Themajor conclusionwedrew is
that mostof the MMTV DNA in theBALB/c
genomeisorganized into twounits (II andIII)
which resemble the proviruses acquired after
infection withretroviruses, i.e.,structures with
ca. 6.0x 106Mrofviral DNAcontaining
termi-nally redundant sequences derived from the
end(s) of viral RNA. A small amount of viral
DNAin BALB/c DNA isalsopresent in aunit ofsubgenomic size (unitI).
Thisstudy representsthe firsteffortto
deter-minethestructure ofendogenousviral DNAby
directphysical mapping procedures. The
tech-nical approach required the use ofagarose gel
electrophoresis,
the DNA transfer procedure(30),and molecularhybridization with virus-spe-cific, radiolabeled cDNA's. These methods al-lowed detection of the
small
number ofvirus-specific restriction fragments without purifica-tion of the viral DNAfromtheca.
106-fold
excessof cellular DNA. The mapping procedure
in-cluded conventional
sequential
digestions withvariouscombinations oftwo orthree
endonucle-ases (Fig.8and 9).To
minimize
thecomplexity ofanalyzing multiple units withina singlecel-lulargenome,wealsousedpreparative
fraction-ation (17) of thelarge
EcoRI
fragments bearing MMTV sequencesbeforesomeof thesecondary digestions with other enzymes (Fig. 2-5). The use ofhybridization reagent specific for a few hundred bases at the 3' end of MMTV RNA(cDNA3)was
particularly
helpful, sincewe wereabletoshow that thissequence wasrepeatedat
bothends of unitsIIandIII,asit isinproviruses
acquired
by infection (4, 14, 26). Furthermore,the low complexity of cDNA3
allowed
us todetectcertainfragments that containeda
small
amount ofviral DNA linked to
cellular
DNA.Themostunconventionalaspect of the
mapping
procedure involved the use ofmouse genetics. We have shown elsewhere that endogenous MMTV proviruses have
segregated
like stable genetic elements during inbreeding (5), andwededuced thatsome restriction
fragments
musthave come from the same endogenous unit of viral DNAby showing that the appearance of those fragments was coordinate in digests of DNA from variousstrains (Fig.6;5).
These strategies for physical mapping were
adequate foranunambiguous ordering of
restric-tion fragments in units II and III and their immediately flanking
cell
sequences.Neverthe-less,moredetailed analysis
will
benecessary toexclude the presence of
small
regions of DNAunrelated to MMTV (e.g.,
cellular
intervening sequences) within thestructuresdefined andtodetermine theprecise length and
origins
ofse-quencesrepeatedatthe ends oftheendogenous proviruses. InMMTV DNA introduced by
in-fection, sequences fromboth the 3' and 5' ends
ofviral RNA are
terminally
repeated as ase-quence of ca. 1,200 bases (29; J. E. Majors,
unpublished data). We showed here that
se-quences from the 3' end of MMTV RNA are
presentatbothends of units II and III (asbased upon
annealings
withcDNA3')and that theon November 10, 2019 by guest
http://jvi.asm.org/
494 COHEN, MAJORS, AND VARMUS
minal redundancies are at least 700 to 800 base
pairs in length (as based upon PstI and HpaI
recognition sites repeated at both ends).
How-ever, we have not as yet used cDNA transcribed
from the 5' terminus of viral RNA to analyze
these redundancies, nor have we determined the
precise length of the redundant sequence.
Pre-sumably these and other refinements can now
beapproached afteramplificationofendogenous
proviruses inprocaryotic systems.
Since no endogenous MMTV strains have
been recovered from BALB/c mice, we were
obliged to use cDNA'spreparedfrom the RNA
of milk-borne MMTV. We then assumed that
these cDNA's would detect all the restriction
fragments generated from each unit that
con-tained endogenous DNA related to the genomes ofmilk-borneMMTV. Withinthesensitivity of the analysis, this assumption appears to have been valid, since we did not find anomalous
reductions in thecombined molecularweightsof
virus-specific fragments,usingmultipleenzymes tocleave endogenousDNA. It ispossible,
how-ever, that the unit we have denoted as
subge-nomic(unit I)is aconventionallysizedprovirus,
onlypart of which ishomologoustomilk-borne
strains of MMTV. The size of the other two
units(ca. 6.0x 106 Mr)is notsignificantly
differ-ent fromthe size ofproviruses
acquired by
in-fection withmilk-borne virus (ca. 5.9 x 106 Mr;
29), and we wereabletodetectallthefragments
derivedfrom units II andIII with our
hybridi-zationreagent. These findings support our
pre-viousclaims thatmost, if notall,of thegenomic
sequences of milk-borneviruses havehomologs
inthe DNA of inbred mice(4, 22, 25).However,
the locations ofrestriction sitesfor
HpaI, PstI,
andBglIIwithin theendogenousproviruses
dif-fer appreciably from the location of sites for
those enzymes within the DNA of milk-borne
strainsofMMTV (4, 29),suggestingthat
signif-icantdifferences in base sequences are
distrib-uted along the genomes. It is possible that
re-gionsof greatest difference accountfor the
evi-dence thatendogenousMMTV DNA may lack
sequences presentinmilk-borne viruses (7, 18);
the results presented here, however, argue
against the notion that the genomes of
milk-borne viruses carry genetic elements of
signifi-cantcomplexity totally unrepresentedin
endog-enousproviruses.
Our finding that most endogenous MMTV
DNA is organized into structures resembling
conventional proviruses acquired by infection
offers further support for our contention that
endogenous proviruses of MMTV were acquired
intherelatively recent past,longafterspeciation
bymultiple, independent infections ofgerminal
tissues (5). Our argument waspreviously based
upon a demonstration of the heterogeneity of
wild mice with respect to the number ofcopies
andsites ofintegration ofendogenous MMTV
DNA. Acomplementary analysisofendogenous
MMTV DNA in inbred micesuggested that the
heterogeneityof wild mice was not due to ahigh frequencyofdeletions, transpositions, or
ampli-fication ofendogenousproviral DNA,since such
DNA hassegregatedlike stableelements during
inbreeding. Support for this hypothesis has
come from similar studies of otherendogenous
viruses (2, 8a, 13) and from the demonstration
that the fourendogenousMMTV provirusesin
theA/HeJstrain mouse arelocatedon atleast
three chromosomes (21). The present report
shows that most of the endogenous MMTV
DNA in BALB/c mice has the structure to be
predicted for a provirus acquired by infection,
rather than for sequences that have evolved
directly from cellulargenes. Ofcourse, the pres-ent results, of themselves, donot indicate how
or when theproviruses mighthave been
intro-duced into the germline.
The resultspresented here have an
interme-diatepractical implication in thatthey provide
several markers (inthe form of restrictionsites)
with which to identifythe MMTV genomes
en-dogenoustoBALB/c andtodifferentiate them
from the genomes ofother strains of MMTV,
both horizontally and genetically transmitted.
We have already exploited the differences in
patterns ofrestriction fragments from
endoge-nous MMTV DNA and milk-borne MMTV
DNA in ordertoassay organsof BALB/c mice
forinfectionbyvirusingested via foster nursing with C3H mice(4). V. L.Morriset al. (personal
communication) have used a similar approach
to show that MMTV occasionally recovered
from tumors ofBALB/cmiceisnot endogenous
to those mice and was presumably introduced
by inadvertent infectionwithvirusstrains trans-mitted in the milk of females with high tumor
incidence.
We chose to perform these initial mapping
studies of endogenous MMTV DNA with
BALB/c mice for at least three reasons: (i) the
units of viral DNA appeared relatively low (4,
22);
(ii)
the strain is extremely susceptible tomilk-borne viruses and hence widely used to study virus-induced mammary cancer (3, 20, 23);
and
(iii)
there isvery limited expression ofen-dogenous viral DNA, even in the lactating
mam-maryglands of multiparous animals or in
mam-mary tumors arising in uninfected animals (8, 19, 37). For the last reason, we supposed that
this strain might be the most likely to lack
MMTVDNA organized into conventional
pro-J. VIROL.
on November 10, 2019 by guest
http://jvi.asm.org/
viral structures. In other mouse strains that lack MMTV in the milk, there is evidence for pro-duction of infectious virus late in life (3, 20, 23) or for synthesis of viral RNA similar in size to the MMTV RNA identified in infectedcells (27), thus suggesting the presence of endogenous pro-viruses that resemble viral DNA acquired by infection with MMTV. It would be of obvious interest to define the factors that impair the expression of the two apparently complete pro-viruses endogenous to BALB/c mice (units II andIII).
ACKNOWLEDGMENTS
Wethank J. M. Bishop,S. Hughes, and D. Robertson for useful discussions of these data and the manuscript and Y. W. Kan for use of instrumentation essential to the completion of this work.
Thisstudy was supported by Public Health Service grants CA 12705,CA 19287, and CA 21454 and by American Cancer Society grant VC 70. J.C.C. was supported by a fellowship from the National Institutes of Health (grant CA 04471) and by grant IN-133 from the American Cancer Society.
LITERATURE CITED
1. Aaronson, S.A., and J. R. Stephenson. 1976. Endog-enous type C RNA viruses of mammalian cells. Biochim. Biophys. Acta 458:323-354.
2. Astrin, S. M. 1978. Endogenous viral genes of the White Leghorn chicken: common site of residence and sites associated with specificphenotypes of viral gene expres-sion. Proc. Natl.Acad. Sci. U.S.A. 75:5941-5945. 3. Bentvelzen, P.1974. Host-virus interactions in murine
mammarycarcinogenesis. Biochim. Biophys. Acta 355: 236-259.
4. Cohen,J.C.,P. R.Shank,V.L.Morris,R.Cardiff, andH. E. Varmus. 1979. Integration of the DNA of mousemammarytumorvirus in virus-infected normal andneoplastic tissues of themouse. Cell16:333-345. 5. Cohen, J. C., and H. E.Varmus. 1979. Endogenous mammary tumour virusDNA varies among wild mice and segregates during inbreeding. Nature (London) 278:418-423.
6. Denhardt, D. T. 1966. Amembrane-filtertechnique for the detection ofcomplementary DNA. Biochem. Bio-phys. Res.Commun. 23:641-648.
7.Drohan, W., R. Kettman, D. Colcher, and J. Schlom. 1977. Isolation of the mouse mammary tumor virus sequences nottransmitted as germinal provirus in the C3H and RIII mouse strains. J. Virol.21:986-995. 8. Dudley, J. P., J. M.Rosen, and J. S. Butel. 1978.
Differential expression of poly(A)-adjacent sequences ofmammary tumorvirus RNA in murine mammary cells. Proc. Natl. Acad. Sci. U.S.A. 75:5797-5801. 8a.Frisby, D. P., R. A. Weiss, M. Roussel, and D.Stehelin.
1979.The distribution ofendogenouschicken retrovirus sequences in the DNA ofGalliform birds does not coincidewithavianphylogenetic relationships.Cell 17: 623-634.
9. Gillespie, D., S.Gillespie,R.C.Gallo,J.L.East,and L.Dmochowski. 1973.Geneticoriginof RD 144 and otherRNA tumourvirusesassayed by molecular hy-bridization. Nature(London)New Biol. 244:51-54. 10. Goulian, M.1968.Incorporationofoligonucleotidesinto
DNA. Proc. Natl.Acad. Sci. U.S.A.61:284-291. 11. Greene, P. J., H. L. Heyneker, F. Bolivar, R.L.
Rod-riguez,M.C.Betlach,A.A.Couarrubias,K. Back-man,D. J.Russel,R.Tait,andH. W.Boyer.1978.
A general method for the purification of restriction enzymes.Nucleic Acids Res. 5:2372-2380.
12. Helling, R. B., H. M.Goodman, and H. W. Boyer. 1974. Analysis of endonuclease EcoRI fragments of DNA fromlambdoidbacteriophages and other viruses byagarose gel electrophoresis. J. Virol. 14:1235-1244. 13. Hughes, S. H., F. Payvar, D.Spector, R. T. Schimke, H. L Robinson, J. M.Bishop, and H. E. Varmus. 1979.Heterogeneity of genetic loci in chickens: analysis ofendogenous viral and nonviral genes by cleavage of DNA with restriction endonucleases.Cell 18:347-359. 14. Hughes, S. H., P. R. Shank, D. H. Spector, H.-J. Kung, J. M. Bishop, and H.E.Varmus.1978. Pro-viruses of avian sarcoma virus areterminally redundant, co-extensive withunintegrated linear DNA and inte-grated at many sites. Cell 15:1397-1410.
15. Kelly, R. B., N. R. Cozzarelli, M. P. Deutscher, I. R. Lehman, and A.Kornberg.1970.Enzymatic synthe-sis of deoxyribonucleic acid. XXXII. Replication of duplex deoxyribonucleic acid by polymerase at a single strand breach. J. Biol. Chem. 245:39-45.
16. Keshet, E., and H. M. Temin. 1978.Sites of integration ofreticuloendotheliosis virus DNA inchicken DNA. Proc. Natl. Acad. Sci. U.S.A. 75:3372-3376.
17. Leder, P., S. M. Tilghman, D. C. Tiemeier, K.I. Pol-sky, J. G. Seidman, M. H. Edgell, L W. Enquist, A. Leder, and B. Norman.1978.Thecloning ofmouse globin and surrounding gene sequences in bacterio-phage lambda. ColdSpring HarborSymp. Quant. Biol. 42:915-920.
18. Michalides, R., and J. Schlom. 1975. Relationship in nucleic acid sequencesbetween mousemammarytumor virus variants. Proc. Natl. Acad. Sci. U.S.A. 72:4635-4639.
19.Michalides, R., L. Van Deemter,R.Nusse,G.Ropcke, and L. Boot. 1978. Involvement ofmousemammary tumor virus in spontaneous and hormone-induced mammary tumors in low-mammary-tumor mouse strains. J. Virol. 27:551-559.
20. Moore, D. H.1975.Mammarytumorvirus, p.131-167.In F. B. Becker(ed.), Cancer: acomprehensive treatise, vol. 2. Plenum Press, NewYork.
21. Morris,V.L., C. Kozak, J. C.Cohen,P. R.Shank,P. Jolicoeur, F. Ruddle, and H. E. Varmus.1979. En-dogenousmousemammarytumor virusDNAis distrib-utedamongmultiplemousechromosomes.Virology92: 46-55.
22. Morris, V. L., E. Medeiros, G. M. Ringold, J. M. Bishop, and H. E. Varmus. 1977. Comparison of mousemammarytumorvirus-specific DNA ininbred, wild, and Asian mice, and in tumors andnormal organs frominbred mice. J. Mol. Biol.114:73-91.
23. Nandi,S.,and C. M.McGrath.1973.Mammary neopla-sia in mice.Adv. Cancer Res.17:353-414.
24. Parks, W. P., and E. M.Scolnick.1973.Murine mam-marytumor cellclones with varyingdegreesof virus expression.Virology55:163-173.
25. Ringold, G. M., P. B.Blair,J.M.Bishop,and H. E. Varmus.1976.Nucleotide sequencehomologiesamong mousemammarytumorviruses.Virology70:550-553. 26. Ringold, G. M.,P.R.Shank,H. E.Varmus,J.Ring, andK.R. Yamamoto.1979.Integrationand
transcrip-tion of mousemammarytumor virus DNA in rat hep-atomacells. Proc. Natl. Acad. Sci.U.S.A.76:665-669. 27. Robertson,D.L.,and H. E. Varmus. 1979. Structural analysisof theintracellularRNAs of murinemammary tumor virus.J. Virol.30:576-589.
28. Sabran,J. L.,T. W.Hsu,C.Yeater,A.Kaji,W.S. Mason, andJ. M.Taylor.1979.Analysisofintegrated
avian RNA tumor virus DNA intransformed chicken, duck, and quailfibroblasts. J.Virol.29:170-178. 29. Shank, P. R., J. C. Cohen, H. E. Varmus, K. R.
on November 10, 2019 by guest
http://jvi.asm.org/
496 COHEN, MAJORS, AND VARMUS
Yamamoto, andG. M.Ringold. 1978. Mapping of linear and circular forms ofmouse mammary tumor virusDNAwithrestrictionendonucleases: evidencefor a large specific deletion occurring at high frequency duringcircularization.Proc. Natl.Acad. Sci. U.S.A.75: 2112-2116.
30. Southern, E.M. 1975. Detection ofspecificsequences among DNAfragmentsseparatedbygel electrophore-sis.J.Mol. Biol. 38:503-517.
31. Steffen,D.,and R. A. Weinberg. 1978. The integrated genome of murine leukemia virus. Cell 15:1003-1010. 32. Swanstrom, R., and P.R.Shank. 1978.X-ray intensi-fying screens greatlyenhance the detection by autora-diographyof theradioisotope 82pand 1251.Anal. Bio-chem.86:184-192.
33. Tal, J., H.-J.Kung, H.E.Varmus, and J. M. Bishop. 1977. CharacterizationofDNAcomplementaryto nu-cleotide sequencesadjacenttopoly(A) atthe3'
termi-J. VIROL.
nusoftheavian sarcoma virus genome.Virology 79: 183-197.
34. Todaro,G. J., R. E. Benveniste,R.Callahan,M. M. Leiber,and C. J. Sheer. 1975. Endogenousprimate andfelinetypeCviruses. ColdSpring HarborSymp. Quant.Biol. 39:1159-1168.
35. Todaro, G. J., and R. J. Huebner. 1972. The viral oncogene hypothesis: newevidence. Proc.Natl. Acad. Sci.U.S.A.96:1009-1015.
36. Varmus,H.E.,J. M.Bishop,R.C. Nowinski,and N. H. Sarkar. 1972.Mammary tumorvirusspecific nu-cleotide sequences in mouseDNA. Nature (London) New Biol.238:189-191.
37. Varmus, H. E., N. Quintrell, E. Medeiros, J. M. Bishop, R. C. Nowinski,and N. H. Sarkar. 1973. Transcriptionof mouse mammary tumor virusgenesin tissues from high and low tumor incidence mouse strains. J. Mol. Biol. 79:663-679.