Post on 21-Jul-2016
description
C L A S S I C
P A P E R
Spliced segments at the 5k terminus ofadenovirus 2 late mRNA{Susan M. Berget, Claire Moore and Phillip A. Sharp
An amazing sequence arrangement at the 5kends of adenovirus 2 messenger RNA{Louise T. Chow, Richard E. Gelinas, Thomas R. Broker andRichard J. Roberts
Reviewed by Tim J. Harrison*Royal Free and University College Medical School, Royal Free Campus, Rowland Hill Street,London NW3 2PF, UK
Accepted 13 July 2000
INTRODUCTIONThe discovery of RNA splicing in adenovirusesgrew out of the stimulating environment createdin the early 1970s by James Watson, Joe Sambrookand others at the Cold Spring Harbor Laboratories(CSHL). In those days, both of the scientists whowere eventually to win the 1993 Nobel Prize forMedicine for the discovery of splicing madesigni®cant contributions to the development ofthe techniques that led to the birth of molecularbiology and stimulated an immediate advance inadenovirus research. Rich Roberts puri®ed andcatalogued a considerable array of restrictionendonucleases [1]; whilst Phil Sharp was involvedin the development of the ethidium bromidestained agarose gel [2] we all take for granted, aswell as techniques to separate the strands ofdenatured restriction fragments of DNA [3].
Adenoviruses were amenable to biochemicalanalysis; they grew to high titre and could bepuri®ed easily. The major structural proteins hadbeen identi®ed by polyacrylamide gel electrophor-esis of disrupted virions and the genome shown tobe a double stranded DNA molecule of around35 000 bp containing inverted terminal repeats. Itwas known that adenovirus DNA was translatedby the host RNA polymerase II [4] and assumed,therefore, that adenoviruses constituted a goodmodel of cellular transcription. Viral RNA wasshown to be polyadenylated [5]. The availability ofisolated restriction endonuclease fragments ofadenovirus DNA as probes enabled the mappingof the regions of the genome expressed early ininfection [6] (DNA replication was inhibited withcytosine arabinoside, preventing late transcrip-tion). The major late genes were mapped by avariety of techniques, including restriction endo-nuclease analysis of recombinants between adeno-virus 2 and adenovirus 5 temperature sensitivemutants with de®ned genetic lesions [7] andhybrid-arrested translation of adenovirus RNAin cell free systems, using de®ned restrictionfragments of adenovirus DNA [8].
Phil Sharp moved to the Massachusetts Instituteof Technology and it was there that Sue Berget,along with Claire Moore, started to map theprecise location of the major capsid protein
{Reproduced from Proc Natl Acad Sci USA 74, 3171--3175 (1977)with kind permission of the publishers.{Reproduced from Cell 12, 1--8 (1977) with kind permission of thepublishers.
*Correspondence to: Dr T. J. Harrison, Royal Free and UniversityCollege Medical School, Royal Free Campus, Rowland Hill Street,London NW3 2PF, UK. E-mail: t.harrison@rfc.ucl.ac.uk
Abbreviations used:CSHL, Cold Spring Harbor Laboratories; hnRNA, heterogeneousnuclear RNA
Reviews in Medical Virology Rev. Med. Virol. 2000; 10: 355±371.
Copyright # 2000 John Wiley & Sons, Ltd.
(hexon) mRNA using R-looping with restrictionfragments of the adenovirus 2 genome. Late ininfection, most RNA on polysomes is viral andhexon mRNA is the most abundant species. R-looping exploits the higher melting temperature ofRNA : DNA than DNA : DNA duplexes, usingelectron microscopy to visualise structures inwhich hybridised RNA has displaced one of theDNA strands from a duplex molecule. Similarstudies were carried out in Roberts' laboratory atCSHL. The results of these experiments were torevolutionise our understanding of eukaryotictranscription.
Spliced segments at the 5k terminus ofadenovirus 2 late mRNA*(adenovirus 2 mRNA processing/5k tails onmRNAs/electron microscopy of mRNA.DNAhybrids)
SUSAN M. BERGET, CLAIRE MOORE,AND PHILLIP A. SHARPCenter for Cancer Research and Department of Biology,Massachusetts Institute of Technology, Cambridge,Massachusetts 02139
Communicated by David Baltimore, May 9, 1977
Reprinted from Proc Natl Acad Sci USA 74, 3171±3175 (1977)
ABSTRACTAn mRNA fraction coding for hexon polypeptide, the major
virion structural protein, was puri®ed by gel electrophoresis
from extracts of adenovirus 2-infected cells late in the lytic
cycle. The mRNA sequences in this fraction were mapped
between 51.7 and 61.3 units on the genome by visualizing
RNA.DNA hybrids in the electron microscope. When hybrids
of hexon mRNA and single-stranded restriction endonuclease
cleavage fragments of viral DNA were visualized in the
electron microscope, branched forms were observed in which
160 nucleotides of RNA from the 5k terminus were not
hydrogen bonded to the single-stranded DNA. DNA sequences
complementary to the RNA sequences in each 5k tail were
found by electron microscopy to be located at 17, 20, and 27
units on the same strand as that coding for the body of the
hexon mRNA. Thus, four segments of viral RNA may be joined
together during the synthesis of mature hexon mRNA. A
model is presented for adenovirus late mRNA synthesis that
involves multiple splicing during maturation of a larger
precursor nuclear RNA.
Most eukaryotic mRNAs bear modi®cations at both termini;
their 3k termini have a tract of poly(A) that ranges in length
from 30 to 200 bases (1±4), while their 5k termini are typically
capped with a methylated guanine joined through a 5k-5k
pyrophosphate linkage to a second nucleotide methylated at its
2k position (5, 6). Both types of modi®cations of eukaryotic
mRNA are known to occur after transcription.
All adenovirus mRNAs are thought to contain poly(A) tracts
at their 3k termini (7) and be capped with a methylated guanine
(8, 9). Speci®c restriction endonuclease cleavage fragments of
adenovirus 2 (Ad2) DNA have permitted the mapping of
regions of the genome expressed as mRNA and viral proteins
during different stages of the lytic cycle (10±12). Little is known
about the molecular mechanisms of viral mRNA synthesis. An
important aspect of late mRNA synthesis is thought to be the
processing and selection of viral mRNAs from the nucleus (13,
14). We have puri®ed a late Ad2 hexon mRNA and found
evidence providing some insight into the mechanism of
synthesis of this mRNA.
MATERIALS AND METHODS
Isolation of Ad2 DNA and RNA
Polyribosomal RNA was prepared from Ad2-infected cells
32 hr after infection as described by Flint and Sharp (14, 15) and
selected by chromatography on poly(U)-Sephadex (16).
R-Loop Mapping
The R-loop hybridization mixture was essentially that of
Thomas et al. (17) and contained 70% (vol/vol) formamide
[Matheson, Coleman, and Bell, 99%, further puri®ed as
described by Duesberg and Vogt (18)]; 0.20 M Tris-HCl,
pH 7.91; 0.50 M NaCl; 0.01 M EDTA; Ad2 DNA at 10 mg/ml;
and puri®ed hexon mRNA at 1±10 mg/ml. This mixture was
incubated at 52.5uC for 2±3 hr and spread on a hypophase of
water with internal length standards of DNA from bacterio-
phage wX174, 5375 bases (19).
Hybridization to Single-Stranded Ad2 DNA
Hybridizations of either polyribosomal poly(A) or puri®ed
hexon mRNA with restriction endonuclease fragments of Ad2
DNA were carried out in reaction mixtures of 80% formamide;
0.40 M NaCl; 0.04 M 2-(N-morpholino)ethanesulfonic acid
(Mes), pH 6.2; 0.01 M EDTA; DNA at 10 mg/ml; and hexon
AbbreviationAd2, adenovirus 2
* We dedicate this work to the memory of Jerome Vinograd, aman who loved science
356 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
mRNA at 1.0±10 mg/ml (20). The sample was incubated at
57±60u for 2±3 hr.
RESULTSAdenovirus late mRNAs begin to appear on polyribosomes
about 13 hr after infection and continue to accumulate in the
cell throughout the lytic cycle (21). Thus, to fractionate the most
abundant late mRNA, polyribosomes were prepared from cells
32 hr after infection with Ad2 and poly(A)-containing mRNA
was selected by chromatography on poly(U)-Sephadex
columns. These mRNAs were then resolved into different
molecular weight fractions by electrophoresis in 2.4±4.0% linear
gradient polyacrylamide gels containing a uniform concentra-
tion of 7 M urea. After staining with ethidium bromide, distinct
¯uorescent bands were present in gels containing mRNA from
virus-infected cells that were not found in gels containing
identically prepared HeLa cell mRNA (Fig. 1A). These virus-
speci®c RNAs were selectively labeled when [32P]phosphate
was added to infected cells 24 hr after infection and the same
mRNA fractions were prepared (Fig. 1B). RNA from the
predominant ethidium bromide-staining band migrating 1.5
times faster than 28S rRNA in Fig. 1A and marked with a large
arrow has been shown to code for the hexon polypeptide by in
vitro translation (S. M. Berget, B. E. Roberts, and P. A. Sharp,
data not shown). Furthermore, this RNA has been mapped by
the R-loop technique (see below) to a region of the genome
known to code for hexon (12) and is complementary to the r
strand of the viral DNA (11). This mRNA species, therefore,
will be referred to in the following sections as hexon mRNA.
R-Loop Mapping of Hexon RNA
The R-loop technique developed by White and Hogness (22)
and Thomas et al. (17) was used to position puri®ed hexon
mRNA on the viral genome. RNA eluted from a gel similar to
that in Fig. 1A was incubated, as described in Materials and
Methods, with either total Ad2 DNA or restriction endonuclease
fragments. Of the 43 total Ad2 DNA molecules scored as
containing R-loops, 41 were observed to have a single region of
hybrid, while two molecules contain a second R-loop,
apparently in the region of the genome coding for the 100K
polypeptide (12).
Fig. 2 shows two examples of R-loops resulting from
hybridizations of hexon mRNA to fragments generated by
the cleavage of Ad2 DNA with the EcoRI restriction endo-
nuclease (see Fig. 3 for fragment location). Hexon RNA spans
the junction between the EcoRI A and B fragments, thus
creating branches at one end of each fragment; one strand of
the branch is single-stranded, the other is double-stranded and
terminated in a ball of single-stranded RNA. Comparison of the
lengths of the two hybrids with an internal standard of double-
stranded wX174 DNA maps the 5k end of the RNA at 51.7t0.5
units (uncertainties are as indicated in (ref. 19)) of the genome
and the 3k end at 61.3t0.5 units, in close agreement with other
estimations based on both R-loop mapping with total cyto-
plasmic RNA (23, 24) and the viral polypeptide mapping of
Lewis et al. (12).
Figure 1. Polyacrylamide gel electrophoresis of Ad2 mRNA.
Polyribosomal poly(A)-containing RNA was prepared 32 hr
after infection as described in Materials and Methods from
either Ad2-infected cells (A) or infected cells to which [32P]phos-
phate was added 24 hr after infection (B). Approximately 25 mg of
this RNA was resolved by electrophoresis on 2.4±4.0% poly-
acrylamide gels containing 7 M urea for 16 hr at 100 V. The gel
was either stained with 0.50 mg/ml of ethidium bromide and
photographed (A) or autoradiographed (B).
Figure 2. R-Loops of puri®ed hexon mRNA and EcoRI cleavage
products of Ad2 DNA. Total Ad2 DNA cleaved by EcoRI and
hexon mRNA were incubated and then spread to visualize R-
loops as described in the text. Two examples of the 60 R-loop
structures photographed and measured are shown; hybrid
structures are observed at the ends of the EcoRI A fragment (A)
and the EcoRI B fragment (B). The junction of these two
fragments maps at 58.5 units on the Ad2 genome. The strand
speci®city of the mRNA sequences from this region (11) and the
EcoRI cleavage map of Ad2 DNA were used to assign the 5k and 3kpolarity of the RNA forming these R-loops. Bars represent 0.1 mm.
Classic Paper 357
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
When hexon mRNA was incubated under conditions to
form R-loops with Ad2 DNA that had been cleaved with the
HindIII restriction endonuclease, R-loops of the type shown in
Fig. 4A, B, and C were observed. The mRNA is totally included
within the HindIII A fragment, forming R-loops terminating
at positions 6.3t0.1% and 50.2t0.8% from the end of the
fragment positioned at 50.1±73.6 map units on the Ad2
genome. Because hexon mRNA is known to be transcribed
in the right-ward direction (11), the map coordinates of hexon
mRNA established from the R-loops to EcoRI fragments
indicate that the 5k end of hexon mRNA is 6.3% from the end
of the HindIII A fragment. Small single-stranded ``tails'' were
visible at both ends of the R-loop; such tails appeared on
Figure 4. Electron micrographs of hybrids of hexon mRNA and fragments of Ad2 DNA. An example of an R-loop hybrid observed after
incubation of hexon mRNA and duplex HindIII A fragment DNA is shown in A and B and is diagrammed schematically in C.
Similarly, two examples of hybrids of hexon mRNA and single-stranded HindIII A fragment are shown in D and E. A schematic of the
hybrid structure shown in E is given in F. The single-stranded RNA at the end of the hybrid region is represented by a wave-like line.
The hybrids of hexon mRNA and single-stranded HindIII A fragment DNA shown in D and C were mounted from an 80% formamide
solution. In A, B, D, and E the positions of the RNA tails at the 5k and 3k ends of the hybrids are denoted by arrows. An example of a
hybrid between single-stranded EcoRI A DNA and hexon RNA is shown in G and diagrammed in H. The hybrid region is indicated by
a heavy line; loops A, B, and C (single-stranded unhybridized DNA) are joined by hybrid regions resulting from annealing of upstream
DNA sequences to the 5k tail of hexon mRNA. Bars on micrographs represent 0.1 mm.
Figure 3. Restriction map of Ad2 DNA. The vertical lines
indicate the positions of cleavage for either the EcoRI or HindIII
restriction endonucleases. The positions of genes coding for the
hexon and 100,000 molecular weight (100K) polypeptides are
from Lewis et al. (12).
358 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
88% of the 5k ends (all polarities given with respect to the
mRNA orientation) and on 75% of the 3k ends of those R-loops
with termini that mapped within two standard deviations of
the mean position of the termini expected for full-length
hybrids. Although the 3k tail might in part be attributed
to the poly(A) tracts of these mRNAs, the tails of the 5kend of the mRNA was not expected and prompted further
investigation.
Hybridization of Hexon RNA to Single-Stranded
DNA
A possible interpretation of single-stranded tail-like structures
at the ends of R-loops would be that branch migration creating
DNA.DNA duplex had occurred, displacing the ends of the
RNA (25). To examine this possibility, hybrids were formed
with hexon mRNA and totally single-stranded HindIII A DNA.
In such hybrids there would be no competing DNA.DNA
renaturation to displace RNA.DNA hybrids. Such RNA.DNA
hybrids were formed by incubating hexon mRNA with
denatured HindIII A DNA at high formamide concentration
(80%) and at 57u (20). After an appropriate incubation the
hybrids were spread and examined under the electron
microscope. As expected, little or no duplex HindIII A
fragment was observed. Fig. 4D, E, and F shows the types of
hybrids that were observed with hexon mRNA and HindIII A
DNA. Double-stranded hybrid segments were terminated with
clearly visible tails at both the 5k and 3k ends of the mRNA
molecules. Of those molecules having full-length hybrid, 90%
had 5k tails and 64% had 3k tails. There may have been a bias in
favor of selecting full-length hybrids with tails for screening
because it was dif®cult to accurately position the ends of a
duplex region that did not terminate in a forked structure.
However, approximately 20% of the total HindIII A strands
displayed a forked structure at the 5k-end position of the hexon
mRNA. This result strongly suggests that the sequences in the
tails are not complementary to the adjacent DNA sequences.
Histograms comparing the lengths of the hybrid regions
observed in the R-loop technique to those produced by
hybridization of hexon mRNA with denatured HindIII A
DNA are shown in Fig. 5A and B. When lengths are calculated
from the hatched area of the histogram (those molecules
containing full-length hexon RNA) the hybrid length is
3330t290 base pairs for R-loop hybrids and 3540t240 base
pairs for hybrids formed with single-stranded HindIII A DNA.
This agrees very well with the length of the RNA itself,
3510t180 bases, as determined by visualization in the electron
microscope following spreading by the urea/formamide
technique (Fig. 5C) (26).
The measured lengths of 5k and 3k tails on the two types of
hybrids are similar; the 5k tails measure 170t40 bases on R-
loops and 160t50 bases on hybrids with single-stranded DNA
(Fig. 5D and E); and the 3k tail measures 150t60 nucleotides on
R-loops and 110t40 nucleotides on hybrids with single-
stranded DNA (Fig. 5F and G). These contour lengths may be
an underestimate because they were calculated assuming that
single-stranded RNA chains were fully extended under the
formamide spreading conditions employed. Those molecules
having hybrids but no tails are also scored on the histograms in
Fig. 5. In both techniques, those molecules having less than a
full-length hybrid region were always missing at least one tail,
suggesting that neither method artifactually generates such
Figure 5. Histograms of contour lengths of various parts of hexon
mRNA and HindIII A DNA hybrids. kb is kilobases, 1000 bases
or base pairs. A is a histogram of the contour length of the
RNA.DNA duplex in R-loops of hexon mRNA and duplex
HindIII A DNA. Similarly, B is histogram of the contour length
of RNA.DNA duplex in hybrids of hexon mRNA and single-
stranded HindIII A DNA. An example of each of these types of
hybrids is shown in A and D, respectively, of Fig. 4. In both A
and B the hatched area represents hybrids found by intact RNA
chains and these molecules were used in calculating average
lengths of RNA.DNA duplex. C is the contour length histogram
of the hexon mRNA as it is eluted from the gel. The molecules
bracketed by the bar were assumed to be intact. D and E are the
contour length histograms of the 5k and 3k single-stranded RNA
tails, respectively, for the R-loop hybrids scored in A. Similar
histograms for the 5k and 3k single-stranded RNA tails of the
hybrids formed with hexon mRNA and single-stranded HindIII
A DNA, respectively, are shown in F and G, respectively. A
histogram of the duplex contour length between the end of the
HindIII A DNA and the beginning of the R-loop hybrid is given
in H. The histogram for the equivalent contour length between
the end of the single-stranded HindIII A DNA and the beginning
of the RNA.DNA hybrid region is shown in I. The hatched and
solid areas scroed on the 0 nucleotide position in D through G
represent molecules falling in the hatched and solid areas,
respectively, of A and B.
Classic Paper 359
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
structures. Fig. 5 also contains two histograms showing the
position of the 5k tail with respect to the 50.1 unit end of the
Ad2 HindIII A fragment (H and I); the 5k tail appears to begin
480t40 base pairs from the end of R-loop molecules and
410t50 bases from an end of the hybrid molecules formed
with the single-stranded HindIII A fragment.
The presence of branched structures suggested that the 5k
end of hexon mRNA may not be complementary to the
adjacent region of DNA. Several alternate explanations
remained to be eliminated by control experiments. The ®rst
involved the possibility that the branched structure was due to
an unusually (A+T)-rich set of DNA sequences at this position
which would be melted at the high formamide±high tempera-
ture spreading conditions employed. To eliminate this possi-
bility, hybridization mixtures of hexon RNA and the puri®ed
denatured HindIII A fragment were diluted 70-fold into either
50% or 40% formamide solution and prepared for electron
microscopy; under these conditions melting of even highly
(A+T)-rich complementary sequences should not be observed.
However, hybrid structures still contained 5k and 3k tails at the
same frequency as those scored at the higher formamide
concentrations (histogram not shown).
Another possibility was that the tails might arise from
palindromic sequences at the ends of the mRNA molecules
which were more stable as RNA.RNA hybrids and thus would
not form hybrids with the complementary DNA. If such
palindromic sequences were present in the Ad2 HindIII A
fragment at this position, they should be visible as hairpins on
single-stranded HindIII A DNA spread under low formamide
conditions. Therefore, single-stranded HindIII A DNA frag-
ment was spread from a 50% formamide solution and
visualized with the electron microscope. A histogram of the
position of all hairpin structures relative to the nearest end of
the single-stranded DNA segment was constructed for 51
molecules. No hairpin structures were observed at the map
position of either end of the hexon mRNA (data not shown).
To ensure that the tails were linked to the hybrid by
ribonuclease-sensitive bonds, hybrids were spread for visuali-
zation after treatment with pancreatic RNase under conditions
where hybrid structures should be resistant to degradation.
After RNase treatment, no 5k or 3k tails were observed, though
the appropriate RNA.DNA hybrid length was seen when
spread from an 80% formamide solution (data not shown).
If the sequences in the 5k tail of hexon mRNA are transcribed
upstream from the same template strand as the other 95% of
the RNA sequences, then a hybrid of this mRNA and single-
stranded EcoRI A DNA should form single-stranded DNA
loops at the 5k terminus of the duplex part of the hybrid. The
single-stranded DNA forming the loop would correspond to
the viral DNA sequences between the two regions of the
template strand that were transcribed and joined during
synthesis of the hexon mRNA. This experiment was performed
and RNA.DNA hybrids of EcoRI A DNA and hexon mRNA
were selected for scoring which, as expected, had a duplex
region on one terminus terminating in a collapsed ball of
single-stranded RNA. The R-loop data described in Fig. 1
predict that 70.9% of the hexon mRNA adjacent to the
5k terminus should form hybrids with single-stranded EcoRI
A DNA; the remaining 29.1% would be collapsed under these
spreading conditions. An example of the RNA.DNA hybrids
observed is shown in Fig. 4G and schematically in Fig. 4H. Data
from 24 such structures were averaged for the following
discussion. At the left end of the structure there is a single-
stranded segment 5770t390 bases (16.8% of the genome) in
length followed by three deletion type loops of single-stranded
DNA originating within 200 bases of the 5k terminus of the
RNA.DNA hybrid segment. The RNA.DNA hybrid has the
expected contour length of 2710t320 bases (7.74%). The order
and contour length of the three loops from the left end of the
genome to the right are: loop A, 1010t130 (2.90%); loop B,
2350t130 (6.70%); and loop C, 8060t830 (23.0%). Loops A and
B are separated by 80t20 bases and loops B and C by 110t10
bases. The simplest interpretation of this structure is that the
5k tail of the hexon mRNA is composed of sequences
transcribed from three different regions of the same strand of
the viral DNA. The map positions of these three regions are
16.8t1.1, 19.8t1.1, and 26.9t1.1. The segment of RNA.DNA
hybrid at the 5k terminus of the hexon mRNA creating loop A is
too short to be distinguished in our electron micrographs. A
comparison of the sum of the lengths of the duplex segments
separating the three loops, 190t30 bases, with the measured
length of the 5k tail on the hexon mRNA, 160t50 bases,
suggests that this region may be quite short. However, this
segment would probably have to be at least 15 bases long to be
stable under the denaturing conditions used for spreading
these samples. Fifty single-stranded EcoRI A DNA molecules
that contained one or more loops were scored from the same
grid; no loop structures were observed that corresponded to
the loops seen in the hexon mRNA/EcoRI A hybrids.
DISCUSSIONThe most abundant viral mRNA found on the polyribosomes of
cells 32 hr after infection with adenovirus 2 maps by the R-loop
technique in the region of genome that codes for the hexon
polypeptide (12). When R-loops between this mRNA and the
HindIII A fragment were examined in the electron microscope,
almost all molecules containing an intact mRNA had single-
stranded RNA tails of 160 nucleotides at their 5k ends. To test
whether this single-stranded 5k-end RNA tail was produced by
branch migration forming homologous DNA.DNA base pairs,
hybrids were formed between the puri®ed mRNA and
denatured HindIII A fragment DNA. A forked structure was
360 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
observed at the 5k end of this mRNA in almost all hybrids
formed by the annealing of an intact mRNA chain to single-
stranded DNA. This forked structure was observed under a
variety of different conditions of mounting for visualization in
the electron microscope and strongly suggests that a segment
of the 5k end of the mRNA is not complementary to the adjacent
viral DNA sequences. The RNA sequences in each 5k tail are
apparently transcribed from the r strand of Ad2 DNA
upstream from those coding for the body of the hexon
mRNA. The structure of hexon mRNA and single-stranded
EcoRI A DNA hybrids (see Fig. 4G and H) suggests that RNA
sequences of unknown length from 16.8t1.1%, of 80t20 bases
from 19.8t1.1%, and 110t10 bases from 26.9t1.1% are joined
in the 5k tail of hexon mRNA.
When total poly(A)-containing polyribosomal RNA was
hybridized to denatured HindIII A under the same conditions,
a second mRNA mapping in the region of the genome coding
for the 100 K polypeptide (12) was observed to have a similar
forked structure at its 5k terminus. Thus, a common short
sequence of RNA might be attached to several late mRNAs.
This is consistent with the observation of R. Gelinas, D. Klessig,
and R. Roberts (personal communication) that a single T1
ribonuclease oligonucleotide containing a capped structure is
found on total viral mRNAs isolated during the late stage of
infection.
The three short segments forming the 5k tail of hexon mRNA
are probably spliced to the body of this mRNA during post-
transcriptional processing. During the late stage of the lytic
cycle the r strand of Ad2 is transcribed into long transcripts
that originate in the left third of the genome and terminate near
the right end (27±30). The region of the genome coding for the
body of the hexon mRNA and the sequences in these three
short RNA segments in the 5k tail of this mRNA are probably
included in this long transcript. Thus, a plausible model for the
synthesis of the mature hexon mRNA would be the intra-
molecular joining of these short segments to the body of the
hexon mRNA during the processing of a nuclear precursor to
generate the mature mRNA. This would result in the
maturation of one mRNA species from each longer precursor
and would explain the large abundance of accumulated viral
RNA sequences in the nucleus of cells during the late stage of
the lytic cycle and the selective transport of certain viral RNA
sequences to the cytoplasm (14). It is interesting to speculate on
how general such a model for the processing of eukaryotic
mRNAs could be. Assuming that eukaryotic mRNA sequences
are adjacent to the 3k terminus of heterogeneous nuclear RNA,
this mechanism would certainly explain the observations by
Perry and Kelley (31) that the 5k-terminal cap 1 structures of
heterogeneous nuclear RNA from mouse cells are conserved
during the processing of these sequences to cytoplasmic
mRNAs, though the lengths of the RNA chains differ by a
factor of 4 between these RNA fractions.
The role of the spliced RNA segment at the 5k end of
adenovirus late mRNA is subject to speculation. This RNA
segment could be involved in the selection of certain viral RNA
sequences for transport to the cell cytoplasm or could be
responsible for the preferential translation of viral mRNA
during the late stage of infection. Because the capped
5k terminus of eukaryotic mRNA is thought to be directly
involved in the initiation of translation of mRNA, an involve-
ment of these sequences in the control of translation would be
expected.
ACKNOWLEDGEMENTSWe would like to thank Arnold J. Berk, Timothy J. Harrison,
Daniel Donoghue, and David Baltimore for comments on the
manuscript, and Ms. Margarita Siafaca for typing the manu-
script. We gratefully acknowledge the suggestion by David
Baltimore that we map the RNA sequences in the 5k tail by
electron microscopy of RNA.DNA hybrids. This work was
supported by an American Cancer Society Grant and career
development support (VC-151A) to P.A.S., a Cancer Center
Core Grant (CA-14051), and a National Institutes of Health
Postdoctoral Fellowship to S.M.B. (CA02391-01).
REFERENCES1. Kates J. (1970) Cold Spring Harbor Symp. Quant. Biol. 35:
743±752.
2. Edmonds M, Vaughan MH Jr, Nakazoto H. (1971) Proc.
Natl. Acad. Sci. USA 68: 1336±1340.
3. Lee SY, Mendecki J, Brawerman G. (1971) Proc. Natl. Acad.
Sci. USA 68: 1331±1335.
4. Darnell JE Jr, Wall R, Tushinski RJ. (1971) Proc. Natl. Acad.
Sci. USA 68: 1321±1325.
5. Furuichi Y, Morgan M, Muthukrishnan S, Shatkin AJ.
(1975) Proc. Natl. Acad. Sci. USA 72: 362±366.
6. Wei CM, Moss B. (1974) Proc. Natl. Acad. Sci. USA 71:
3014±3018.
7. Philipson L, Wall R, Glickman G, Darnell JE. (1971) Proc.
Natl. Acad. Sci. USA 68: 2806±2809.
8. Hashimoto S, Green M. (1976) J. Virol. 20: 425±435.
9. Moss B, Koczot F. (1976) J. Virol. 17: 385±392.
10. Sharp PA, Flint SJ. (1976) Current Topics in Microbiology and
Immunology 74: 137±158.
11. Sharp PA, Gallimore PH, Flint SJ. (1974) Cold Spring Harbor
Symp. Quant. Biol. 34: 457±474.
12. Lewis J, Atkins JF, Anderson C, Baum PR, Gesteland RF.
(1975) Proc. Natl. Acad. Sci. USA 72: 1344±1348.
13. Bachenbeimer S, Darnell JE. (1975) Proc. Natl. Acad. Sci.
USA 72: 4445±4449.
14. Flint SJ, Sharp PA. (1976) J. Mol. Biol. 106: 749±771.
Classic Paper 361
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
15. Flint SJ, Gallimore PH, Sharp PA. (1975) J. Mol. Biol. 96:
47±68.
16. Lindberg U, Persson T, Davis RW. (1972) J. Virol. 10:
909±919.
17. Thomas M, White RL, Davis RW. (1976) Proc. Natl. Acad.
Sci. U S A 73: 2294±2298.
18. Duesberg PH, Vogt PK. (1973) J. Virol. 12: 594±599.
19. Davis RW, Simon M, Davidson N. (1971) in Methods in
Enzymology, eds. Grossman, L. & Moldave K. (Academic
Press, New York), Vol. 21, pp. 413±428.
20. Casey J, Davidson N. (1977) Nucleic Acid Res. 4: 1539±1552.
21. Green M. (1970) Annu. Rev. Biochem. 39: 701±756.
22. White RL, Hogness DS. (1977) Cell 10: 177±192.
23. Westphal H, Meyer J, Maizel J. (1976) Proc. Natl. Acad. Sci.
U S A 73: 2069±2071.
24. Chow LT, Roberts JM, Lewis JB, Broker TM. (1977) Cell, in
press.
25. Lee CS, Davis RW, Davidson N. (1970) J. Mol. Biol. 48: 1±22.
26. Robberson D, Aloni Y, Attardi C, Davidson N. (1971) J. Mol.
Biol. 60: 473±484.
27. Parsons JT, Green M. (1971) Virology 45: 154±162.
28. Wall R, Philipson L, Darnell JE. (1972) Virology 50: 27±34.
29. McGuire PM, Swart C, Hodge LD. (1972) Proc. Natl. Acad.
Sci. U S A 69: 1578±1582.
30. Goldberg S, Weber J, Darnell JE. (1977) Cell 10: 617±622.
31. Perry RP, Kelley DE. (1976) Cell 8: 433±442.
An Amazing Sequence Arrangement atthe 5k Ends of Adenovirus 2 MessengerRNA
LOUISE T. CHOW, RICHARD E.GELINAS, THOMAS R. BROKER, ANDRICHARD J. ROBERTSCold Spring Harbor Laboratory, Cold Spring Harbor,New York 11724
Reprinted from Cell 12, 1±12 (1977)
SUMMARYThe 5k terminal sequences of several adenovirus 2 (Ad2)
mRNAs, isolated late in infection, are complementary to
sequences within the Ad2 genome which are remote from the
DNA from which the main coding sequence of each mRNA is
transcribed. This has been observed by forming RNA displace-
ment loops (R loops) between Ad2 DNA and unfractionated
polysomal RNA from infected cells. The 5k terminal sequences
of mRNAs in R loops, variously located between positions 36
and 92, form complex secondary hybrids with single-stranded
DNA from restriction endonuclease fragments containing
sequences to the left of position 36 on the Ad2 genome. The
structures visualized in the electron microscope show that
short sequences coded at map positions 16.6, 19.6 and 26.6 on
the R strand are joined to form a leader sequence of 150±200
nucleotides at the 5k end of many late mRNAs. A late mRNA
which maps to the left of position 16.6 shows a different pattern
of second site hybridization. It contains sequences from 4.9±6.0
linked directly to those from 9.6±10.9. These ®ndings imply a
new mechanism for the biosynthesis of Ad2 mRNA in
mammalian cells.
INTRODUCTIONIn contrast to the detailed knowledge of the mechanics of
transcription in procaryotic cells (Losick and Chamberlin,
1976), little is known about this process in eucaryotic cells.
Several possible schemes exist: one, analogous to the bacterial
system, requires independent promoters for each mRNA; a
second postulates the production of long primary transcripts in
the nucleus which are subsequently cleaved to yield individual
mRNAs (Darnell, Jelinek and Molloy, 1973); and a third
invokes the use of RNA primers coded at one region on the
genome but acting at some other region(s) and becoming
elongated into mRNAs (Dickson and Robertson, 1976). Experi-
ments to test these hypotheses directly have been hampered by
the complexity of the eucaryotic genome. We have chosen to
study these processes in a simpler system ± lytic infection of
human cells by adenovirus 2 (Ad2).
Ad2 DNA is transcribed by RNA polymerase II (Price and
Penman, 1972; Wallace and Kates, 1972), and its transcription
shows features characteristic of that of the host genome (Lewin,
1975a, 1975b). For example, long polyadenylated transcripts
appear in the nucleus, but only a small percentage of this
nuclear RNA appears as polyadenylated mRNA on cytoplas-
mic polysomes (Philipson et al., 1971). These mRNAs are
``capped'' at their 5k ends (Moss and Koczot, 1976; Sommer
et al., 1976). Gelinas and Roberts (1977) found that most
Ad2 mRNAs isolated at late times during infection contain
the same ``capped'' 11 nucleotide sequence at their 5k ends.
This sequence was sensitive to ribonuclease cleavage in
mRNA : DNA hybrids (Gelinas and Roberts, 1977; Klessig,
1977) and led to the suggestion that this 5k terminal sequence
might not be coded immediately adjacent to the main body of
the mRNA.
Thomas, White and Davis (1976) have shown that individual
RNA molecules can be displayed as RNA displacement loops
(R loops) in the electron microscope, and map coordinates have
been obtained for many Ad2 mRNAs (Meyer et al., 1977; Chow
et al., 1977). In the present studies, we have used mRNAs
visualized in such R loops to examine more closely the
sequences present at the 5k end of late Ad2 mRNAs.
362 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
Figure 1. Hybridization of Rightward-Transcribed Strands (r) of Restriction Fragments to the Common 5k Leader Sequences of Late
Ad2 mRNA (e) or mRNAs in R Loops on Ad2 DNA (a±d). (a) represents Hind III-Br annealed to mRNA for hexon; (b) represents Hind
III-Cr annealed to mRNA for hexon; (c) represents Bal I-Er annealed to mRNA for the 100K protein; (d) represents Xma I-Fr annealed to
mRNA for ®ber; (e) represents Bam HI-Br annealed to free mRNA. Map coordinates covered by each restriction fragment and locations
of the hybridization are given in parentheses. Illustrative tracings are provided. (±) Ad2 DNA; (ÐÐ) restriction fragments; (----) mRNA.
In (c) and (d), most of the RNA ``bridge'' between the R loop and the restriction fragment is due to branch migration of the mRNA. The
remaining portion is due to the unhybridized leader sequence.
Classic Paper 363
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
RESULTSR loops were formed between Ad2 DNA and polysomal RNA
isolated 22 hr after Ad2 infection. The 5k ends of the mRNA
should form single-stranded projections if they are not coded
immediately adjacent to the rest of the mRNA, and so might be
visualized by hybridization to a single-stranded DNA fragment
containing their complement. We therefore prepared a set of
restriction endonuclease fragments of the Ad2 genome,
separated their strands by agarose gel electrophoresis (Hay-
ward, 1972; Sharp, Gallimore and Flint, 1974) and added each
single strand in turn as a third hybridization component after
the preparation of the R loops. Since R loops were formed from
a mixed population of late mRNAs, many different species
were examined simultaneously. By using a restriction endo-
nuclease fragment as the single-stranded probe, complicated
structures which might arise from hybridization of the probe to
the single-stranded DNA segment of the R loop were limited to
one region of the genome. Figure 1a shows the results of such
an experiment using the slow strand of Hind III-B (map
position 17.0±31.5) as the single-stranded DNA probe. The
probe hybridized with the 5k end of hexon mRNA in the R loop
but not with the displaced DNA strand. It adopted a looped
con®guration, indicating that sequences from the 5k end of the
mRNA were complementary to two separate regions within the
probe. The 5k ends of other mRNAs also show identical two-site
hybridization with the slow strand of Hind III-B and are
compiled in Table 1. Length measurements place the contact
points between the Hind III-B single strand and the mRNA
at approximately 900t60 nucleotides (42 measurements)
from the end of the short arm and 1800t120 nucleotides
(42 measurements) from the end of the long arm. The distance
between the two contact points on the DNA (the loop) was
about 2400t90 nucleotides (49 measurements). To orient these
two arms, determine the strandedness and obtain accurate map
positions for the points of hybridization, we used the separated
strands of an overlapping fragment Bam HI-B (map position
0±29.1) in a similar experiment. The results are shown in
Figure 2a, in which the slow strand of Bam HI-B is hybridized
to the 5k ends of both ®ber and hexon mRNAs. In these cases
and in others reported in Table 1, more complicated structures
were observed. Three contact points between the single-
stranded DNA probe and the 5k end of the mRNA are now
evident, and the Bam HI-B fragment is held into two loops.
Length measurements give values of 5800t180 nucleotides
(39 measurements) for the long arm, 950t100 nucleotides (58
measurements) for the short arm, 2400t130 nucleotides
(48 measurements) for the large loop and 1000t100 nucleo-
tides (47 measurements) for the small loop. Comparison with
the hybridization sites on the Hind III-B strand suggests map
positions of 16.6, 19.6 and 29.6 for the three segments of Ad2
DNA which hybridize to the 5k ends of mRNA. Examination of
these structures revealed that the contact point closest to the
main portion of the mRNA was on the long arm of the Hind III-
B fragment and on the short arm of the Bam HI-B fragment.
Thus the 3k end of the leader sequence is at 26.6 and the 5k end
is at 16.6. Because the mRNAs labeled by these probes are
transcribed from the R strand from left to right, and because
nucleic acids form anti-parallel base pairs, we conclude that
these probes are from the R strand. Weingartner et al. (1976)
have also shown that the slow strand of Bam HI-B is the R
Table 1. Map Coordinates of Labeled 5k Termini of Late Ad2 mRNAs
mRNA
Assignment
Previous Map
Coordinatesa
Map Coordinates of
5k Label
(MeantStandard Deviation)
Number R Loop Molecules Labeled with Restriction
Endonuclease Fragments
Bam HI-B Hind III-B Bal I-E Xma I-F Totalb
Core ± 36.6t0.6 2 1 0 2 5
Penton 38.8 38.9t1.0 1 3 3 0 7
Core 45.4 45.0t1.0 1 6 1 1 9
pVI 49.9 49.8t0.5 3 4 2 2 11
Hexon 51.9 52.2t0.8 28 20 4 4 56
100K 67.9 67.9t0.4 3 3 2 3 11
pVIII 74.6 74.1t0.5 5 2 0 2 9
Fiber 86.3 86.4t0.5 29 26 5 2 62
Totals 72 65 17 16 170
aChow et al. (1977).
bNot included in the table are two molecules, each labeled at 66 and 71, that could be alternative 5k ends for 100K and pVIII, respectively.
364 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
strand. All the mRNA species labeled are transcribed from the
R strand (Sharp, Gallimore and Flint, 1974; Pettersson, Tibbetts
and Philipson, 1976).
The 5k terminal leader sequence of an mRNA in an R loop
occasionally formed an intramolecular structure by hybridizing
to its complementary DNA at coordinates 19.6 or 26.6 within
the same DNA molecule. One example involving hexon mRNA
is shown in Figure 2b. Such interaction constrains the inter-
vening DNA, which often assumes a super-coiled con®guration
during spreading for electron microscopy.
To ensure that our interpretation of these structures was
correct, we performed a number of control experiments. In
separate hybridizations, single strands from restriction frag-
ments encompassing the entire Ad2 genome were used as
probes, and a summary of these data is shown in Figure 3. Only
the slow strands of Hind III-B (Figure 1a, two contacts at
19.6 and 26.6), Hind III-C (Figure 1b, one contact at 16.6), Bam
HI-B (Figure 2a, three contacts at 16.6, 19.6 and 26.6), Bal
I-E (Figure 1c, two contacts at 16.6 and 19.6) and Xma I-F
(Figure 1d, one contact at 16.6) showed consistent hybridization
to RNA branches at the 5k ends of R loops. In particular, it
should be noted that the fast strands of these ®ve fragments did
not interact with any of the R loops. When the slow strands of
Bam HI-B or Bal I-E were incubated alone and spread under
identical conditions, no loops of the same size or with the same
coordinates as those formed in the presence of mRNA were
Figure 2. Multiple Site Hybridization of the 5k Leader Sequences of the Hexon and Fiber mRNAs in R Loops. (a) Hybridization with the
R strand (r) of the Bam HI-B restriction fragment. Arrows point to the DNA : RNA hybrids. Arrowheads point to the large and small
loops formed in Bam HI-Br DNA due to the hybridization. An additional 100 nucleotides at the 5k ends of the hexon R loops have been
displaced by branch migration. The spreading force during the preparation of grids has denatured about 200 nucleotides at the 5k end of
the ®ber mRNA/DNA hybrid. (b) The leader of the ®ber mRNA in an R loop was labeled by an added R strand of the Hind III-Br
fragment. The leader on the hexon mRNA was labeled by intramolecular hybridization to complementary DNA at coordinate 19 on the
same molecule. The intervening DNA segment was constrained, and it formed tertiary superhelical twists when solvent conditions
were changed during preparation of the sample for electron microscopy. The hexon RNA formed a convergent R loop with the mRNA
for the E72K protein hybridized to the opposite (L) DNA strand.
Classic Paper 365
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
detected. If polysomal RNA was present during the incubation
of the slow strand, but not the fast strand, of Bam HI-B,
however, loops of the type shown in Figure 1e (identical to
those seen at the 5k ends of many late mRNAs in R loops) were
frequently observed and were associated with collapsed RNA.
The possibility that sequences at map positions 16.6, 19.6 and
26.6 were reiterated on the Ad2 genome was tested by isolating
small fragments of the genome containing these sequences,
labeling them to high speci®c activity in vitro by nick
translation and using them as hybridization probes against
fragments of the Ad2 genome immobilized on nitrocellulose
®lters (Southern, 1975). In each case, as shown in Figure 4, the
fragments rehybridized only to that region of the genome from
which they were derived and failed to hybridize to any other
sequences on the Ad2 genome.
Hybrids between any one component of the leader sequences
of the mRNA in an R loop and single-stranded DNA probes are
stable in 70% formamide, 0.4 M NaCl, 0.1 M HEPES at 30uC,
and yet there is only a hint of a duplex at positions 26.6 and
19.6 when Bam HI-B and Hind III-B are used, or at position 19.6
in the Bal I-E fragment. The duplex regions were measured to
be 50±100 nucleotides at each of these two positions, and we
believe it is improbable that more than a total of 200
nucleotides are involved at all three contact points.
The results described above refer to transcripts located to the
right of position 36. Several other late mRNAs are known to
map to the left of this coordinate. One of these, coding for
polypeptide IVa2 (map position 14.9±11.2), is transcribed from
the L strand, and a second, coding for virion-associated
component IX (map position 9.7±11.0), is transcribed from the
R strand (Chow et al., 1977; U. Pettersson and M. B. Mathews,
manuscript submitted). Both have been visualized in R loops,
but neither showed secondary hybridization with any of the
fragments used in this study. Some of the R loops formed by a
polysomal RNA which contains sequences from coordinates 9.6
(t0.2)±10.9 (t0.2) (24 measurements each), however, have an
unusual structure. Sequences from the 5k end of this RNA form
a second R loop with a noncontiguous region of the Ad2
genome located between coordinates 4.9 (t0.3)±6.0 (t0.2)
(Figures 5a and 5b). As a result, the intervening double-
stranded DNA was held into a third loop, and a short bridge of
displaced RNA between the two R loops is clearly visible.
This structure has frequently been observed in molecules
containing a convergent R loop formed between IVa2 mRNA
and this new RNA (Figure 5c). Because a strand switch at 11.0
can be seen in this structure, as has been observed earlier
(Chow et al., 1977), the new RNA species can be assigned to the
R strand.
DISCUSSIONThe results presented in this paper show that sequences
present at three separated sites (16.6, 19.6, 26.6) on the R
strand of the Ad2 genome are complementary to a
continuous sequence at the 5k end of late Ad2 mRNAs that
are transcribed from the R strand and map to the right of
position 36. Since these sequences are available for hybridi-
Figure 3. Hybridization of Separated Strands of Ad2 DNA Restriction Fragments to the 5k Leader Sequence of Late Polysomal mRNAs.
Strands from restriction fragments spanning the entire Ad2 genome were used in separate experiments to label the R loops. All (+)
slow strands come from the R strand, as discussed in Results. (+) indicates consistent hybridization of the strands to the leader
sequences of the mRNAs in R loops; (x) indicates negative results. Arrowheads point to the locations of hybridization on the R strands
of the fragments. The map coordinates of the restriction fragments are obtained from C. Mulder and R. Greene for Bam HI and Xma I
(unpublished observations) from R. J. Roberts and J. Sambrook for Hind III; from J. R. Arrand and R. J. Roberts for Sal I (unpublished
observations); and from R. E. Gelinas and R. J. Roberts for Bal I (unpublished observations).
366 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
zation when mRNA is displayed in R loops and are not
reiterated elsewhere in the Ad2 genome, we conclude that
they are not coded at a site immediately adjacent to the
main portion of the mRNAs. Biochemical evidence by
Gelinas and Roberts (1977) and Klessig (1977) has been
presented to support this idea. Since it seems improbable to
us that the sequence present at the 5k end of many of these
late Ad2 mRNAs is actually coded by the host genome and
is only complementary to these three Ad2 sequences by
chance, we believe that these sequences are probably
transcribed from positions 16.6, 19.6 and 26.6 on the R
strand of the Ad2 genome, and that their juxtaposition is an
inherent feature of Ad2 mRNA biosynthesis.
Two mRNAs (for polypeptides IVa2 and IX) mapping to the
left of position 30 seem to have a different sequence
arrangement at their 5k ends. Particularly surprising is the
®nding that a polysomal RNA containing sequences from
coordinate 9.6±10.9, the coding region for component IX, has an
additional sequence at its 5k end which is complementary to a
noncontinguous segment from 4.9±6.0. This RNA may be
related to early transcripts for E15K, which map between
5.0±11.0 or between 5.0±6.4 (Chow et al., 1977), and also to the
component IX mRNA, which maps between 9.7±11.0 (Chow
et al., 1977; U. Pettersson and M. B. Mathews, manuscript
submitted). The absence of the tripartite leader and the
occurrence of this new mRNA would account for the
hybridizational and translational data reported for mRNAs
originating from this region of the genome (Lewis, Anderson
and Atkins, 1977).
These observations, together with the results presented in the
accompanying papers on late Ad2 mRNA (Klessig, 1977;
Lewis, Anderson and Atkins, 1977) and on Ad2-SV40 mRNA
(Dunn and Hassell, 1977) are not directly consistent with any
mechanism previously suggested for the biosynthesis of
mRNA in eucaryotic cells. They imply that an alternate
scheme must exist for Ad2 mRNAs, and perhaps for eucaryotic
mRNA in general. One such mechanism is outlined in the
accompanying paper by Klessig (1977). The experiments
described herein provide a convenient method to map
accurately the 5k termini of Ad2 mRNAs, and have con®rmed
many of the previous assignments (Chow et al., 1977) and
established new ones. We have recently learned of similar
experiments by Berget, Moore and Sharp (1977) who used
electron microscopy to examine hybrids between puri®ed
hexon mRNA and single strands of DNA. They observed that
the 5k terminal mRNA sequence appeared as a single-stranded
tail, which was complementary to three noncontiguous regions
of the Ad2 genome with map coordinates essentially identical
to those reported here.
Experimental Procedures
Restriction EndonucleasesBal I (Gelinas et al., 1977) and Xma I (Endow and Roberts, 1977)
were puri®ed as described. Bam HI, Sal I and Xma I were
puri®ed by unpublished procedures of P. A. Myers and R. J.
Roberts. In all cases, DNA was digested at 37uC in 6 mM
Tris±HCl (pH 7.9), 6 mM MgCl2 and 6 mM 2-mercaptoethanol.
Figure 4. Hybridization of Bal I-E DNA (14.7±21.5) and Bal I-D
DNA (21.5±28.5) to Bal I Fragments of Ad2 DNA. Bal I-E and -D
fragments of Ad2 DNA were isolated after two cycles of
puri®cation by agarose gel electrophoresis, labeled with 32P by
nick translation and used as hybridization probes against all Bal
I fragments of Ad2 DNA bound to nitrocellulose membranes.
Slot 1 represents 2.0 mg of Bal I fragments of Ad2 DNA
fractionated on a 1.4% agarose gel and stained with ethidium
bromide. The same amount of DNA was present in slots 2 and 3.
Slot 2 represents 32P-Bal I-D DNA (106 dpm; about 20 mg)
hybridized to Bal I fragments of Ad2 DNA. Slot 3 represents32P-Bal I-D DNA (106 dpm; about 20 mg) hybridized to Bal I
fragments of Ad2 DNA. The minimal length of sequence
homology which can be detected by this method has not been
determined.
Classic Paper 367
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
Isolation of Viral DNA and RNADNA was prepared from Ad2 virions grown on HeLa or
KB cells in suspension cultures as described by Pettersson and
Sambrook (1973) and Pettersson et al. (1973). Fragments of the
Ad2 genome were produced by digestion with the restriction
endonucleases Bal I, Bam H-I, Hind III, Sal I and Xma I. DNA
fragments were fractionated by agarose slab gel electrophoresis
(Sugden et al., 1975) and recovered from the agarose by
chromatography on hydroxylapatite (Lewis et al., 1975), or by
homogenization and diffusion followed by phenol extraction.
Ad2 mRNA was a gift from Dr. J. B. Lewis. Polysomes were
isolated 22 hr after Ad2 infection of KB cells by the method of
Schreier and Staehelin (1973), and the RNA was recovered by
the method of Anderson et al. (1974).
Strand Separation of Endonuclease FragmentsPuri®ed restriction fragments were denatured in 0.25 M NaOH
and subjected to electrophoresis on 1.4% agarose slab gels
Figure 5. R Loops Formed between Ad2 DNA and a Polysomal RNA Containing Sequences from Map Coordinates 4.9±6.0 (X) and
9.6±10.9 (Y). (a) Sequence Y, with the same coordinates as the mRNA for peptide IX, is totally contained in an R loop, whereas sequence
X, possibly coding for the 15K protein, is only partially contained in an R loop with its 5k end displaced as a tail. (b) Y is present in a
collapsed and partially displaced R loop, whereas X is totally contained in an R loop. (c) X is in a partial R loop. Y is in a convergent R
loop with the mRNA (Z), tentatively assigned to peptide IVa2 on the opposite DNA strand. The RNA bridge (indicated by arrowheads)
between X and Y is visible because of some RNA displacement by the intervening DNA segment. D/S and S/D indicate the double-
strand/single-strand junctions in the convergent R loop.
368 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
containing the Tris-phosphate-EDTA buffer described by
Hayward (1972), but at half the stated ionic strength. After
electrophoresis, bands of single strands were located by
staining with ethidium bromide. Single-stranded DNA was
recovered by homogenizing the gel in several volumes of
0.01 M Tris±HCl (pH 7.9), 0.001 M EDTA and allowing the
DNA to diffuse out for several hours. The aqueous supernatant
was extracted ®rst with phenol and then with chloroform. E.
coli rRNA was added as carrier, and the single-stranded DNA
was recovered by ethanol precipitation.
Filter HybridizationsBal I-E (14.7±21.5) and Bal I-D (21.5±28.5) DNAs were labeled in
vitro by nick translation (Kelly et al., 1970) as described by
Maniatis et al. (1975) and were used as probes to challenge Bal I
fragments of Ad2 DNA adsorbed to nitrocellulose membranes
by the method of Southern (1975).
Electron MicroscopyR loops were formed on intact Ad2 DNA at 51.5uC for 14±16 hr
as described previously (Chow et al., 1977). Aliquots were
diluted with an equal volume of the same buffer-formamide
mixture containing puri®ed, separated strands of Ad2 restric-
tion fragments.
The concentration of the single strands was 5±10 mg/ml. The
solution was returned to the water bath and cooled to 42 or
30uC over a period of 3±5 hr. Electron microscope grid
preparation and data processing have been described by
Chow et al. (1977). Single-stranded wX174 (5375 bases) and
double-stranded wX174 RF or Col E1 DNA (6300 base pairs)
were included as internal length standards.
ACKNOWLEDGMENTSWe thank Dr J. B. Lewis for a gift of late polysomal RNA;
J. Bonventre, P. A. Myers and J. Scott for technical assistance;
and M. Moschitta for secretarial help. This work was supported
by a grant from the National Cancer Institute.
Received June 9, 1977; revised July 5, 1977
NOTE ADDED IN PROOFThe mRNA for IVa2 (14.9±11.2, L strand) also has a short
single component leader present at its 5k end. There is only a
short deletion of the RNA sequences between the leader and
the coding sequences, which is visible in Figure 5c as a small
loop.
REFERENCESAnderson CW, Lewis JB, Atkins JF, Gesteland RF. 1974. Proc.
Nat. Acad. Sci. USA 71: 2756±2760.
Berget SM, Moore C, Sharp PA. 1977. Proc. Nat. Acad. Sci. USA
in press.
Chow L, Roberts JM, Lewis JB, Broker TR. 1977. Cell 11:
819±836.
Darnell JE, Jelinek WR, Molloy GR. 1973. Science 181:
1215±1221.
Dickson E, Robertson HD. 1976. Cancer Res. 36: 3387±3393.
Dunn AR, Hassell JA. 1977. Cell 12: 23±36.
Endow SA, Roberts RJ. 1977. J. Mol. Biol. 112: 521±529.
Gelinas RE, Roberts RJ. 1977. Cell 11: 533±544.
Gelinas RE, Myers PA, Weiss GH, Murray K, Roberts RJ. 1977.
J. Mol. Biol., in press.
Hayward GS. 1972. Virology 49: 342±344.
Kelly RB, Cozzarelli NR, Deutscher MP, Lehman IR, Kornberg
A. 1970. J. Biol. Chem. 245: 39±45.
Klessig DF. 1977. Cell 12: 9±21.
Lewin B. 1975a. Cell 4: 11±20.
Lewin B. 1975b. Cell 4: 77±93.
Lewis JB, Anderson CW, Atkins JF. 1977. Cell 12: 37±44.
Lewis JB, Atkins J, Anderson C, Baum P, Gesteland R. 1975.
Proc. Nat. Acad. Sci. USA 72: 1344±1348.
Losick R, Chamberlin M. 1976. RNA Polymerase. Cold Spring
Harbor, New York: Cold Spring Harbor Laboratory Press,
p. 899.
Maniatis T, Jeffrey A, Kleid DG. 1975. Proc. Nat. Acad. Sci. USA
72: 1184±1188.
Meyer J, Neuwald PD, Lai SP, Maizel JV, Jr, Westphal H. 1977.
J. Virol. 21: 1010±1018.
Moss B, Koczot F. 1976. J. Virol. 17: 385±392.
Pettersson U, Sambrook J. 1973. J. Mol. Biol. 73: 125±130.
Pettersson U, Tibbetts C, Philipson L. 1976. J. Mol. Biol. 101:
479±501.
Pettersson U, Mulder C, Delius H, Sharp PA. 1973. Proc. Nat.
Acad. Sci. 70: 200±204.
Philipson L, Wall R, Glickman G, Darnell JE. 1971. Proc. Nat.
Acad. Sci. USA 68: 2806±2809.
Price R, Penman S. 1972. J. Virol. 9: 621±626.
Schreier MH, Staehelin T. 1973. J. Mol. Biol. 73: 329±349.
Sharp PA, Gallimore PH, Flint SJ. 1974. Cold Spring Harbor
Symp. Quant. Biol. 39: 457±474.
Sommer S, Salditt-Georgieff M, Bachenheimer S, Darnell JE,
Furuichi Y, Morgan M, Shatkin AJ. 1976. Nucl. Acids Res. 3:
749±765.
Southern EM. 1975. J. Mol. Biol. 98: 503±517.
Sugden B, DeTroy B, Roberts RJ, Sambrook J. 1975. Anal.
Biochem. 68: 36±46.
Thomas M, White RL, Davis RW. 1976. Proc. Nat. Acad. Sci.
USA 73: 2294±2298.
Wallace RD, Kates J. 1972. J. Virol. 9: 627±635.
Weingartner B, Winnacker E-L, Tolun A, Pettersson U. 1976.
Cell 9: 259±268.
Classic Paper 369
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
COMMENTARYThe surprising result of R-looping hexon mRNAto a de®ned restriction endonuclease fragment ofadenovirus DNA was that sequences at both endsof the mRNA failed to hybridise to the DNA(Berget et al., Figure 2). This was expected at the 3kend because the RNA was known to be poly-adenylated, but the mismatched 5k end wasunprecedented. Although it was theoreticallypossible that secondary structure or sequenceabnormalities in the 5k end of the RNA favouredDNA : DNA duplexes over RNA : DNA duplexes,it seemed more likely that the 5k end was encodedelsewhere in the adenovirus genome or mighteven be of cellular origin. I recall the groupmeeting in Phil Sharp's of®ce when Sue Bergetpresented these electron micrographs; we werestruggling to think of the next experiment whenDavid Baltimore called by on an unrelated matter.He was presented with the problem and immedi-ately suggested hybridising the hexon mRNA to amore extensive, single stranded fragment ofadenovirus DNA (Berget et al., Figure 4, panels Eand H, see also the acknowledgements). Thisexperiment revealed the tripartite structure of theadenovirus late leader sequence.
Meanwhile, at CSHL several lines of evidencewere converging. Gelinas and Roberts [9] showedthat the same 5k terminal structure was shared byat least 12 species of adenovirus 2 late mRNA.Another group at CSHL was translating in vitroadenovirus mRNAs that had been fractionated byhybridisation to restriction fragments of thegenomic DNA and found that most late mRNAspecies exhibited some hybridisation to a commonsecondary site [10]. Chow and colleagues [11], whoalso were R-looping adenovirus mRNAs, noticedbranches at the 5k and 3k ends and speculated thatthe former might be encoded elsewhere. Theysolved the conundrum by hybridising the sepa-rated strands of other restriction fragments to the 5kbranch of the RNA at the end of the R-loop (Chowet al., Figure 1) and also determined that the lateleader sequence was encoded at three separatesites. In addition, they showed that the mRNA forpolypeptide IX, one of the few late transcripts thatis not controlled by the major late promoter, alsocomprised sequences encoded in two separateregions of the genome (Figure 5).
The discovery of RNA splicing in adenovirusinfected cells seemed likely to have wider rami-
®cations for eukaryotic systems generally. RNAtranscribed in the nucleus was known to be ofmuch higher molecular weight than mRNA in thecytoplasm. This so-called heterogeneous nuclearRNA (hnRNA) had been shown to be polyadeny-lated [12] and was thought to be a precursor ofmRNA, perhaps through cleavage of the latterfrom the 3k ends. Soon the globin [13], ovalbumin[14] and immunoglobulin [15] genes were shownto contain introns and RNA splicing was recog-nised as a central process in eukaryotic geneexpression. Although adenoviruses have evolvedto make spectacular use of RNA splicing, manyother families of eukaryotic viruses also encodemRNAs that are spliced.
What is the purpose of RNA splicing? An earlyhypothesis was that it would allow the shuf¯ingof protein domains (exons) during evolution [16].This hypothesis presumes that introns have anancient origin (introns-early theory). However, inat least some instances, introns seem to have beenacquired more recently than the exons whichsurround them [17]. The introns-early versusintrons-late argument continues [18]. Regardlessof evolutionary considerations, we know now thatdifferential RNA splicing is an important mechan-ism in the control of gene expression in eukaryoticcells, as well as in adenoviruses and many othereukaryotic viruses.
REFERENCES1. Roberts RJ. Restriction and modi®cation enzymes
and their recognition sequences. Gene 1978; 4:183±194.
2. Sharp PA, Sugden B, Sambrook J. Detection of tworestriction endonuclease activities in Haemophilusparain¯uenzae using analytical agaroseÐethidiumbromide electrophoresis. Biochemistry 1973; 12:3055±3063.
3. Flint SJ, Sharp PA. Mapping of viral-speci®c RNAin the cytoplasm and nucleus of adeniovirus2-infected human cells. Brookhaven Symp Biol 1974;26: 333.
4. Price R, Penman S. Transcription of the adenovirusgenome by an -amanitine-sensitive ribonucleicacid polymerase in HeLa cells. J Virol 1972; 9:621±626.
5. Philipson L, Wall R, Glickman G, Darnell JE.Addition of polyadenylate sequences to virus-speci®c RNA during adenovirus replication. ProcNatl Acad Sci USA 1971; 68: 2809.
6. Sharp PA, Gallimore PH, Flint SJ. Mapping of
370 Classic Paper
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.
adenovirus 2 RNA sequences in lytically infectedcells and transformed cell lines. Cold Spring HarborSymp Quant Biol 1975; 39 Pt 1: 457±474.
7. Mautner V, Williams J, Sambrook J, Sharp PA,Grodzicker T. The location of the genes coding forhexon and ®ber proteins in adenovirus DNA. Cell1975; 5: 93±99.
8. Paterson BM, Roberts BE, Kuff EL. Structural geneidenti®cation and mapping by DNA-mRNA hybrid-arrested cell-free translation. Proc Natl Acad Sci USA1977; 74: 4370±4374.
9. Gelinas RE, Roberts RJ. One predominant 5k-undecanucleotide in adenovirus 2 late messengerRNAs. Cell 1977; 11: 533±544.
10. Lewis JB, Anderson CW, Atkins JF. Further map-ping of late adenovirus genes by cell-free translationof RNA selected by hybridization to speci®c DNAfragments. Cell 1977; 12: 37±44.
11. Chow LT, Roberts JM, Lewis JB, Broker TR. A mapof cytoplasmic RNA transcripts from lytic adeno-virus type 2, determined by electron microscopy ofRNA : DNA hybrids. Cell 1977; 11: 819±836.
12. Edmonds M, Vaughan MH Jr, Nakazato H.
Polyadenylic acid sequences in the heterogeneousnuclear RNA and rapidly-labeled polyribosomalRNA of HeLa cells: possible evidence for aprecursor relationship. Proc Natl Acad Sci USA1971; 68: 1336±1340.
13. Jeffreys AJ, Flavell RA. The rabbit beta-globin genecontains a large large insert in the coding sequence.Cell 1977; 12: 1097±1108.
14. Breathnach R, Mandel JL, Chambon P. Ovalbumingene is split in chicken DNA. Nature 1977; 270:314±319.
15. Tonegawa S, Maxam AM, Tizard R, Bernard O,Gilbert W. Sequence of a mouse germ-line gene for avariable region of an immunoglobulin light chain.Proc Natl Acad Sci USA 1978; 75: 1485±1489.
16. Gilbert W. Why genes in pieces? Nature 1978; 271:501.
17. Logsdon JM Jr, Stoltzfus A, Doolittle WF. Molecularevolution: recent cases of spliceosomal intron gain?Curr Biol 1998; 8: R560±R563.
18. Logsdon JMJ. The recent origins of spliceosomalintrons revisited. Curr Opin Genet Devel 1998; 8:637±648.
Classic Paper 371
Copyright # 2000 John Wiley & Sons, Ltd. Rev. Med. Virol. 2000; 10: 355±371.