Reprinted from the ProckEDINGS OF THE NATIONAL ACADEMY OF SCIENCES
Vol. 55, No. 6, pp. 1539-1554. June, 1966.

A RATIONALE FOR AN ANALYSIS OF RNA REPLICATION*

By S. SpreEGELMAN AND I. HarRuNA

DEPARTMENT OF MICROBIOLOGY, UNIVERSITY OF ILLINOIS, URBANA

Presented before the Academy, October 13, 1965, by invitation of the
Committee on Arrangements for the Autumn Meeting

The intent is to trace our efforts at understanding the molecular life history of
RNA genomes using the £. coli-RNA phage system.! In the process, an attempt
will be made to provide a background of the thinking that went into the experi-
ments to be described. Because of the usual dominance of serendipity in biological
and biochemical research, it is not often that it is either informative or useful to
record the reasons for a particular set of experiments; noting the outcome is usually
sufficient. However, the present instance may represent an exception.

(1) Problem of Communication between an RNA Virus and Its Host Cell—We
start with the fact that all organisms which use RNA as genetic material are manda-
tory intracellular parasites. They must, therefore, carry out a major part of their
life cycle in cells which use DNA as genetic material and RNA as genetic messages.
On entry, the viral RNA is faced with the problem of inserting itself into the cellu-
lar information flow pattern in order to communicate its own instructions to the
synthesizing machinery. <A possibility one might entertain centers on whether an
RNA virus employs the DNA-to-RNA pathway of information flow. This could
occur either because the DNA of the host already contains a sequence homologous
to the viral RNA (i.e., the “escaped genetic message”? hypothesis), or because such
DNA sequences are generated subsequent to infection by reversal? of the DNA-
dependent RNA synthesizing reaction. It is clear that a decision on the existence
or nonexistence of homology between viral RNA and the host DNA is a necessary
prelude to further experiments designed to delineate the molecular life history of an
RNA genome.

To answer questions of this nature, Doi and Spiegelman* employed the specific
hybridization test combined with the subsequently developed use of RNase to
eliminate ‘noise.’ The sensitivity required had already been achieved in earlier
experiments which identified the DNA complements of sRNA and ribosomal
RNA#*~ Under conditions where complexes between 23S rRNA and DNA were
readily observed, none were detected between the viral RNA and the infected host
DNA.

The negative outcome of the hybridization test implies that the DNA-to-RNA
pathway is not employed, from which it follows that these RNA viruses must have
evolved a mechanism of generating RNA copies from RNA. The existence is then
predicted of an enzymatic mechanism involving an RNA-dependent RNA polym-
erase which we have named® “replicase”? for purposes of brevity and alliterative
usefulness.

It seems highly unlikely that an enzyme of this sort pre-exists in the cell. All
recognized cellular RNA components, including the message fraction,°® the two ribo-
somal components,® * and the translational 4S RNA,’ have been shown to be
complementary to some sequences in the homologous DNA. Furthermore, actino-
mycin D which inhibits" the DNA-dependent RNA polymerase (transcriptase) pre-

1539
1540 BIOCHEMISTRY: SPIEGELMAN AHD HARUNA Proc. N. A. 8.

vents synthesis of RNA in both bacterial!? and animal cells, but does not inhibit the
production of RNA viruses.!3

(2) Viral RNA as a Translatable Message-—The arguments above led us to as-
sume that no mechanism pre-existed in uninfected cells for generating either com-
plements or identical replicas from RNA. It was our feeling from the outset that
the transcriptase reaction would not normally be employed as a step in virus replica-
tion since its ability to employ RNA templates is poor and fragmentary. Conse-
quently, when we began our enzymological investigations, much of our effort was
directed at eliminating transcriptase from our preparations. Complete justification
of this view was ultimately provided [§ 5 and 6] when purified replicases were
obtained and examined. In any event, this line of reasoning did lead to the pre-
diction that the entering RNA must itself serve as a protein program and be
directly translatable in order to communicate with the cell. Direct proof that viral
RNA is directly translatable into protein was achieved in an in vitro system by
Nathans ef al.14 with £2 and confirmed by Ohtaka and Spiegelman! with MS-2
and by Clark et al.!® with STNV.

There is another consequence of this line of reasoning which can be, and was,
subjected to test. Since the new kind of replicase must be synthesized before
replication can begin, it follows that the entering RNA must be conserved while
serving as a protein program. Without conservation, there would be nothing left
to replicate by the time the replicase was completed. Doi and Spiegelman” under-
took to test the validity of this prediction by the use of MS-2 in which the RNA was
doubly labeled with NY and P*. Both isotopes were recovered in the same
strand, leading to the conclusion that the parental strand of an RNA virus is com-
pletely conserved during all the replications and translations required to produce a
full yield of mature virus particles.

(3) Possible a priort Mechanism of RNA Replication Before we attempt to de-
tail the more recent enzymological approaches to the problem of RNA replication,
it is of interest to review briefly the various possibilities that can be entertained
about the nature of RNA replication as derived from arguments of varying plausi-
bility. The kinds of readily imaginable mechanisms can be divided into two
classes: (1) Nature is pleasantly uniform, the nucleic acid universe being com-
pletely describable by ‘‘Watson-Crickery.” Consequently, the replication of
single-stranded RNA will mimic the mechanism employed by its counterpart,
single-stranded DNA. (2) The alternative view considers the possibility that
RNA genomes evolved a different duplicating device. We may now consider
the a priori reasons which can be marshaled in favor of each line of thought.

(a) A variant solution to RNA replication: In the first place, one can plausibly
argue that since RNA genomes are found only in mandatory intracellular para-
sites, they must have arisen after cells evolved. Consequently, the RNA viruses
emerged in a complicated biochemical environment containing highly complex
enzyme molecules fashioned to carry out reactions demanding refined levels of
selective specificity. These were available to the RNA viruses for choice and modi-
fication to suit their particular needs. In contrast, DNA genomes emerged in the
primitive biochemical environment which characterized ‘genesis,’ in which the
problem of replication had to be solved unaided by the subsequently developed
sophisticated protein catalysts. Of necessity, DNA was initially forced to use its
Vou. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1541

hydrogen-bonding capacities and employ the principle of complementarity to du-
plicate. The apparent use of this mechanism today may represent a residue of the
difficulties which DNA. encountered in its early evolution. We emphasize the
qualification ‘“apparent”’ since, despite a widespread belief to the contrary, the fact
is that we do not know how DNA duplicates in detail. No one has, as yet, produced
incontrovertible evidence which compels acceptance of base pairmg on the template
as a step which necessarily precedes the addition of the next complementary resi-
due to the growing chain. One can still entertain a mechanism in which the
enzyme makes the choice via “allosteric instruction” from the template; base
pairing could then occur subsequent to the synthesis of the new diester bond.

Another argument one can offer in favor of a unique solution for RNA replication
stems from the fact that RNA genomes are translatable messages. In view of what
we know about the coding dictionary, the complement of a translatable message is
likely to be nonsense. Complementary transcription of the original strand is
therefore not only unnecessary for information transmission to the cell, but is in fact
useless. Consequently, if complementary copying does occur as an intermediate
step, it would be employed only for replicative purposes. Synthesis of polynucleo-
tide strands is energetically very expensive, and avoiding this step would provide
an obvious advantage to RNA viruses. In any event, because of these and other
considerations, we pointed out”” that RNA viruses could furnish us with a surprising
variation on the Watson-Crick theme.

(b) DNA-like solutions for RNA replication: The most popular and widely
adopted model stems from the studies of the single-stranded* DNA virus ¢X-174,
the relevant properties of which may be briefly summarized. On infection, the
single-stranded DNA is converted into a double-stranded structure which has been
named” the “replicative” form (RF-DNA). It can be shown” by column chroma-
tography and prelabeled virus that this conversion is complete even at elevated
multiplicities of infection.

It is presumed that the RF-DNA then serves as a template for the formation of
single-stranded copies via an asymmetric synthesis, analogous to transcription into
RNA. However, proof that the RF-DNA is a “replicative” form remains to be
provided. Hayashi and Spiegelman?! ?? showed that the single strand found in
the mature virus particle is the nonsense strand. It follows that in order for this
virus to communicate with its host, the complement must be synthesized. As a
consequence, finding a double-stranded structure does not necessarily signify that the
“replicative” form has been identified; its presence is already justified by its require-
ment for transcription. Indced, even if a single-stranded DNA virus were dis-
covered carrying the coding strand, we would still predict the intervention of a
duplex. The cellular transcription mechanism, which DNA viruses must use, is de-
signed to make single-stranded copies of RNA from a double-stranded template so
that, in any event, the duplex would have to be completed.

Despite these reservations, one must grant the attractiveness of assuming that
a single-stranded RNA virus would have the same general problems as a single-
stranded DNA and that therefore RNA might well adopt the same pathway for its
life cycle. Thus, Ochoa e¢ al. proposed ?* that the first step in the replication of an
RNA would be the conversion of the incoming RNA into a double-stranded struc-
ture which could then serve as a “replicative” form for the generation of single-
1542 BIOCHEMISTRY: SPIEGELMAN AND HARUNA Proc. N. A. 8.

stranded copies of the mature viral RNA. According to this view, replication of
RNA viruses introduces no novelties. The same general rules, assumed to function
in the ¢X-174 system, are presumed to apply here also.

(4) The Search for a Double-Stranded RN A.—Not only was the idea of a duplex
replicating form attractive, but what appeared to be supporting evidence quickly
accumulated in the literature on the RNA bacteriophages. The following properties
can be used in a search for evidence of double-stranded RNA: (1) comparative
resistance to RNase, (2) a lower density in a Cs:SO, gradient, and (8) conversion to
RNase sensitivity by heating and fast cooling at low ionic strength. Structures
possessing one or more of these properties were found*+—® in infected cells. We
have elsewhere“ summarized similar findings in our own laboratory.

Everything looks reasonable and consistent with the interpretation involving a
double-stranded intermediate in RNA replication. However, closer examination
reveals certain difficulties in accepting this as compelling evidence that a component
in the replicative pathway of viral RNA has been unveiled. The following disturb-
ing properties of the resistant structure may be noted: (1) Only a very small per-
centage of the injected viral RNA strands is to be found in this structure; and (2)
the resistant material tends to accumulate late in infection, long after many ma-
ture single strands have been made, a feature not easily reconcilable with the resist-
ant structure (RS) being a mandatory initial intermediate. ‘These findings are in
striking contrast to the ¢X-174 situation, where all injected strands are converted
into RF-DNA and where RF-DNA appears in the first stages of infection, long be-
fore appearance of single-stranded viral DNA.

One can argue that since resistant structures are found only in infected cells, they
must have some relevance to replication. While plausible, such arguments do not
have the ring of logical necessity. One can grant that the resistant structure is a
consequence of the infective process without accepiing 11 as an intermediate of the repli-
cative process. It may play some other undetermined role or be a nonfunctional
artifact. One must bear in mind that we are dealing here with infected cells which
are well along the path to death, and it is not outside the realm of possibility that
pathological artifacts might be produced. Under these circumstances, more direct
evidence than mere existence must be provided before the resistant structures are ac-
cepted as demonstrated components of the replicative mechanism.

In any case, our own attempts and those of others to study the process in the in-
fected cell convinced us that it would be difficult to design a truly decisive experi-
ment which could hope to settle the question of the relation of the RNasc-resistant
structures to RNA replication. It seemed necessary to get on with the enzymol-
ogy in the hope that the relevant enzyme system could be purified to the point
where the mechanism of RNA synthesis could be examined in a simple system per-
mitting hard inferences.

(5) The Search for the MS-2 Replicase-—The search for a unique RN A-depend-
ent polymerase is complicated by the presence of a variety of enzymes which can
incorporate ribonucleotides either terminally or subterminally into pre-existent
RNA chains.*! In addition to transcriptase, which can use RNA templates,**: ™
there are other enzymes (e.g., RNA phosphorylase,*? polyadenylate synthetase, **
etc.) which can mediate extensive synthesis of polyribonucleotide chains. It is
obvious that a claim for a new type of RNA polymerase must be accompanied by
VoL. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1543

evidence for RNA dependence and a demonstration that the enzyme possesses some
unique characteristic which differentiates it in one or more of its properties from
previously known enzymes with which it can be confused.

In addition to these enzymological difficulties, we recognized a biological feature
of the situation which influenced, in at least one important detail, the procedure we
chose in the search for replicase. The point at issue may perhaps best be described
in rather naive and admittedly somewhat anthropomorphic terms. Consider an
RNA virus approaching a cell some 10° times its size and into which the virus is
going to inject its only strand of genetic information. Even if the protein-coated
ribosomal RNA molecules are ignored, the cell cytoplasm still contains approxi-
mately 10,000 free RNA molecules of various sorts. If the new replicase were in-
different and replicated any RNA it happened to meet, whaé chance would the single
original strand injected have of multiplying?

Admittedly, there are several ways out of this dilemma. One could, for example,
isolate the new polymerase molecule and the viral RNA in some sequestered corner
where they would be undisturbed by the mass of cellular RNA components. How-
ever, we entertained the unique possibility that the virus is ingenious enough to
design a polymerase which would recognize its own genome and ignore all other
RNA molecules.

At the outset, of course, we did not know which solution had been adopted by the
virus to solve this dilemma, or even if the dilemma was real. However, the possi-
bility that it did exist, and that replicase selectivity might be the chosen solution,
required that its implications not be ignored for, ¢f true, its disregard would guarantee
failure. In particular, this view meant that we could not afford the luxury of
employing any conveniently available RNA in the search for replicase. It de-
manded the use of purified viral RNA in all steps of the purification. Further, one
might perhaps push the selectivity property to its ultimate pessimistic conclusion.
If the cleverness of the replicase extends further, it might well be true that even a
fragment of its own genome would not be recognized and accepted for replication.
This added possibility made it necessary to provide a guarantee that the RNA em-
ployed is not only homologous, but also intact. This in turn introduced the com-
plication that stages of purification preceding the removal of ribonuclease could well
yield ambiguous or indeed false clues even with intact homologous RNA. Thus, one
had to “fly blind” initially and depend on very brief assays to provide the guides
for the direction of the subsequent steps.

Despite all these potential obstacles, most of which were actually realized, the
first success was achieved in 1963. A procedure involving negative protamine frac-
tionation and column chromatography yielded what looked like the relevant enzyme
from FE. colt infected with MS-2. Most important of all, the preparation exhibited
a virtually complete dependence on added RNA, permitting a test of the expecta-
tion of specific template requirement. The response of MS-2 replicase to various
kinds of nucleic acid revealed a striking preference for itsown RNA. No significant
activity was observed with either the host sRNA or ribosomal RNA. The ability
of the replicase to discriminate between one RNA molecule and another does indeed
solve the crucial problem for an RNA virus attempting to direct its own duplication
in an environment replete with other RNA molecules. By producing a polymerase
which ignores the mass of pre-existent cellular RNA, a guarantee is provided that
1544 BIOCHEMISTRY: SPIEGELMAN AHD HARUNA Proc. N. ALS.

replication is focused on the single strand of incoming viral RNA, the ultimate ori-
gin of progeny.

It seems worth noting that sequence recognition by the enzyme can be of value
not only to the virus, but also to the investigator. As already noted, the search for
viral RNA replicases must perforce be carried out in the midst of a variety of highly
active cellular polymerases capable of synthesizing polyribonucleotides. Tf the
enzyme finally isolated possesses the appropriate template requirement, a comfort-
ing assurance is furnished that the effort expended and the information obtained are
indeed relevant to an understanding of RNA replication.

(6) Confirmation of Specific Template Requirements of RNA Replicases——Our
line of reasoning would lead to the expectation that RNA replicases induced by
other RNA viruses would show a similar preference for their homologous templates.
However, this was not a foregone conclusion since it was conceivable that other
viruses might evolve different solutions to the problem of preferential synthesis.
It seemed important to determine whether template selectivity could be observed
in another virus unrelated to MS-2. The QS phage of Watanabe” was chosen
because of its serological®® and other chemical differences. *®

The isolation and purification of the Q8 replicase” essentially followed, with slight
modifications, the procedures worked out earlier® for the MS-2 replicase. The
properties of the Qé replicase on purification to the stage of complete RNA de-
pendence exhibited the same general features as had been observed with MS-2
replicase, including requirements for all four triphosphates, and Mg++. Figure 1
shows the kinetics observed in a reaction mixture containing saturating amounts of
template (1 y of RNA for 40 y protein). Continued synthesis is observed at 35° for
periods exceeding 5 hr, and in 2 hr the amount of RNA synthesized corresponds to
five times the input of template. By variation in the amount of RNA added and
the time permitted for synthesis, virtually any desired level of increase over the
starting material can be achieved. The cessation of synthesis within 5-10 min
reported by others*!—* for presumably similar preparations has been observed by
us only in the early stages of purification prior to the removal of the nucleases.

The abilities of various RNA molecules to stimulate the Qé replicase to synthetic
activity at saturation concentrations of homologous RNA are recorded in Table 1.
The response of the Qf replicase is in accord with that reported for the MS-2

 

Fic. 1.—Kinetics of replicase activity. In addition to
40 ug of enzyme protein, each standard reaction volume
L +eg-RNA—z. of 0.25 ml contained the following in wmoles: Tris HCl
pH 7.4, 21; MgCh, 3.2; CTP, ATP, UTP, and GTP,
0.2 each. The reaction is run for 20 min at 85°C and
terminated in an ice bath by the addition of 0.15 ml of
neutralized saturated pyrophosphate, 0.15 ml of neu-
tralized saturated orthophosphate, and 0.1 ml of 80%
VF , trichloroacetic acid. The precipitate is transferred to a

membrane filter and washed 7 times with 5 ml of cold

fF 1? 10% TCA. The membrane is then dried and counted in a

L liquid scintillation spectrometer. The washing procedure
e yields zero time values of 80 cpm with input counts of
i; 7! 1 xX 108 cpm. The radioactively labeled UTP® was
L NO QB-RNA synthesized as detailed earlier! and was used at a level

fl xe of 1 X 10° cpm/0.2 wzmole (Haruna and Spiegelman“),
30 60 90 120
MINUTES-—~

#g RNA SYNTHESIZED

(CPM INCORPORATED) x 10-4

 

 

 
VoL. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1545

TABLE 1
RESPONSE OF QS RepLicasE TO DIFFERENT TEMPLATES*
Incorporation

Template (cpm)
Qs 4,929
TYMV 146
MS8-2 35
Ribosomal RNA 45
sRNA 15
Bulk RNA from infected cells 146
Satellite virus of tobacco necrosis 61
DNA (10 ug) 36

Conditions of assay are those specified in Fig. 1. RNA-dependent activity was

assayed at 1 wg of RNA for each 40 ug of protein, and DNA-dependent activity at

10 pg of DNA. Control reactions containing no template yielded an average of 30

“Pe Haruna and Spiegelman (ref. 40).
replicase, the preference being clearly for its own template. The only heterologous
RNA showing detectable activity is TY MV, and it supports a synthesis correspond-
ing to 3 per cent of that observed with homologous Q6 RNA. Both of the heterol-
ogous viral RNA’s (M8-2 and STNV) are completely inactive, and, again, so are
the ribosomal and transfer RNA species of the host cell. It is important to note
that, as in the case of MS-2 replicase, the absence of response to DNA shows that
our purification procedure eliminates detectable evidence of transcriptase from our
enzyme preparations.

The following features distinguish the purified replicases described here from the
presumably similar preparations reported*!—* by other laboratories: (1) complete
dependence on added RNA, (2) competence for prolonged (more than 5 hr) syn-
thesis of RNA, (8) ability to synthesize many times the input template, (4) satura-
tion at low levels of RNA, and (5) virtually exclusive requirement for homologous
template under optimal ionic conditions.

It should be evident from the properties listed that the replicases were indeed
approaching a state of purity where it became relevant to examine the nature of the
product in greater detail—a necessary prelude to experiments designed to illumi-
nate mechanism.

The experimental analysis of a replicating reaction centers necessarily on the
nature of the product. If, in particular, the concern is with the synthesis of a
viral nucleic acid, data on base composition and nearest neighbors, while of interest,
are hardly decisive. The ultimate issue is whether or not replicas are in fact
being produced. To answer this question, information on the sequence of the
synthesized RNA is required. Affirmative evidence of similarity between template
and product would provide assurance that the reaction being studied is indeed
relevant to an understanding of the replicative process.

(7) Autocatalytic Synthesis of a Viral RNA.—The ability of a replicase to distin-
guish one RNA sequence from another can be used to provide information pertinent
to the question of similarity of product to the template. Two sorts of readily per-
formed experiments can decide whether the product is recognized by the enzyme as
a template. One approach is to examine the kinetics of RNA synthesis at tem-
plate concentrations which start below those required to saturate the enzyme. If
the product can serve as a template, a period of autocatalytic increase of RNA
should be observed. Exponential kinetics should continue until the product
saturates the enzyme, after which synthesis should become linear.
1546 BIOCHEMISTRY: SPIEGELMAN AHD HARUNA Proc. N. A. 8.

 

 

 

 

 

 

 

3
5 CPM
3
3
= CPM
gis 4 ak
g |% pr LSxlo 1x10
2/3
7 a
2 oa
eye
25XL4
L ixiot Ix toh
r3
l25Xh2 e
+ Sx103 1x102
LL L
f
1 i L J 1 Ix 1o! 1 1 1 1 t
60 (20 180 240 300 60 (20 180 240 300
MINUTES MINUTES

Fie. 2.—Kineties of RNA synthesis. A 2.5-ml! reaction mixture was set. up containing the
components at the concentrations specified in the legend for Fig. 1. The mixture contained 400
ug of enzyme protein and 2 ug of input Q6-RNA so that the starting ratio of template to enzyme
was one fifth of the saturating level. At the indicated times, 0.19-ml aliquots were removed
and assayed for radioactive RNA as detailed in Fig. 1. The ordinates for cpm and yg of RNA
synthesized refer to that found in 0.19-ml samples. The data are plotted against time arith-
metically on the right, and semilogarithmically on the left. The arrows indicate change from
autocatalytic to linear kinetics (Haruna and Spiegelman‘),

A second type of experiment is a direct test of the ability of the synthesized
product to function as initiating template. Here, a synthesis of sufficient extent
is carried out to ensure that the initial input of RNA becomes a quantitatively
minor component of the final reaction mixture. The synthesized RNA can then
be purified and examined for its template functioning capacities, a property readily
examined by means of a saturation curve. If the response of the enzyme to varia-
tion and concentration of product is the same as that observed with the viral RNA,
one would have to conclude that the product generated in the reaction is as ef-
fective a template for the replicase as is RNA from the mature virus particle.

Preliminary experiments established that 40 y of enzyme protein was saturated
by approximately 1 y of Q8B-RNA. An experiment was therefore sct up in which
the ratio of input template to protein was 1/5 of the saturation value. The results
are plotted in Figure 2 arithmetically and semilogarithmically to permit ready
comparison of kinetics. Exponential increase of RNA is evident over a period of
approximately 3 hr. The arrows indicate the time at which the kinetics depart
from exponential and become linear. Extrapolation to the ordinate indicates that
the change to linear synthesis occurs when approximately 1 y of RNA has ac-
cumulated.

The results just described are consistent with the implication that the product
produced in the course of the reaction can serve to stimulate new enzyme molecules
to activity. The enzyme is therefore able to recognize the product as being one
which is homologous to its own genome.

To carry out the more direct test of this conclusion, a 1-ml reaction mixture was
VoL. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1547

 

A= SYNTHESIZED PRODUCT .
O= VIRAL RNA Fic. 3.—Saturation of enzyme by synthe-

sized RNA compared to viral RNA. The
experiment was carried out as detailed in the
legend for Fig. 1. The circles refer to the
values obtained with RNA isolated from virus
particles, and the triangles to the rates ob-
tained with the RNA synthesized. Since, in
the latter case, the template used was labeled
with P*, H*UTP at 1 X 108 cpm per 0.2
umole was used to follow the synthesis. All
preparations and counting of samples were
carried out as described in Fig. 1 (Haruna and
30 Spiegelman‘).

 

 

 

 

2
jig RNA

set up and the synthesis allowed to proceed for 3.5 hr, by which time a more than
60-fold increase of the input material was achieved. The reaction was then ter-
minated and the RNA purified by the phenol method which yielded 55 per cent of
the synthesized product. Examination in a sucrose gradient showed* that much
of the product has the 28S size characteristic of QB RNA.

Figure 3 illustrates the response of the replicase to various input levels of the
product (triangles) compared to the original viral RNA (circles). It is evident that
the RNA synthesized is as effective in serving as a template as the original viral
RNA.

The data just summarized support the assertion that the reaction generates a
polynucleotide of the same molecular weight (1 X 105) as viral RNA and which the
replicase cannot distinguish from its homologous genome. It is clear that the
enzyme is faithfully copying the recognition sequence employed by the replicase to
distinguish one RNA molecule from another.

(8) Synthesis of an Infectious Self-Replicating Viral RNA.—The next question
concerns the extent of the similarity between product and template. Have identi-
cal replicas been in fact produced? The most decisive test would determine
whether the product contains all the information required to program the synthesis
of complete virus particles in a suitable test system. The success we have just
recorded encouraged an attempt at this next phase of the investigation which
would subject the synthesized RNA to this more rigorous challenge. In the ex-
periments to be described, all RNA preparations were first phenol-treated prior to
assay. Further, the phenol-purified synthetic RNA was routincly tested for whole
virus particles by assay on intact cells, and none were found in the experiments
reported.

We now summarize experiments“ in which the kinetics of the appearance of new
RNA and infective units were examined in two different ways. The first shows that
the accumulation of radioactive RNA is accompanied by a proportionate increase in
infective units. The second proves by a serial dilution experiment that the newly
synthesized RNA is infectious.

(a) Comparison of the kinetics of appearance of RNA and infectious units: To
compare the appearance of newly synthesized RNA and the presence of infectious
units in an extensive synthesis, a reaction mixture was set up containing the neces-
sary components at the concentrations required. Aliquots were taken at the times
indicated for the determination of radioactive RNA and purification of the product
for infectivity assay. Figure 4 shows the observed increase in both RNA and
1548 BIOCHEMISTRY; SPIEGELMAN AND HARUNA Proc. N. A. 8.

 

Ixt08 100

   
   
 

INFEGTIVE UNITS

 

ped

 

 

 

 

 

 

 

= xto® +10
ere 1
ee
BF 1
Lf £ J
e |] € 2000+ 150
7 b Jon .
co
oc 4
2 r 4 = /_.—CPM
>
8 P, 7 1 INFECTIVE UNITS —+——, ~
g 4 +
=z O8F 440%
nxiot a1 E x
J go7- a @
L 4 3
N =
L 4 S 2
1 1 z O06 330 2
1 4 o f——s——TOTAL RNA + ws
7 Lost 1000 3
x
| o4f 712.0
1 1 1 L
60 120 180 2a0
MINUTES O3-F 7 4
nd
ou: : x
Fie. 4.—Kinetics of RNA synthesis and 02 E+ 4io
formation of infectious units. An 8-ml 7
reaction mixture was set up containing the ob
components at the concentrations specified
in Fig. 1. Samples were taken as follows: Loa
1 ml at 0 time and 30 min, 0.5 ml at 60 min, 30 60 90 oe

0.3 ml at 90 min, and 0.2 ml at all sub- MINUTES

sequent times. Twenty \ were removed Fria. 5.—Kineties of RNA synthesis and forma-
for assay of incorporated radioactivity. The tion of infectious units. Same conditions as in
RNA was purified from the remainder, Fig. 4, except that the enzyme was purified ina
radioactivity being determined on the final CsCl gradient which decreased the virus particle
product to monitor recovery (Spiegelman count by a factor of 1 x 10* (Spiegelman,
et al.*”). Haruna, and Pace, in preparation).

infectious units. The amount of RNA (0.8 y per ml at 0 time) is well below the
saturation level of the enzyme present. Consequently, the RNA increases auto-
catalytically for about the first 90 min, followed by a synthesis which is linear with
time. We note that the increase in RNA is paralleled by a rise in the number of
infectious units.

Experiments carried out with other enzyme preparations yielded results in
complete accord with those just described. Another example is given in Figure
5 in which the enzyme used was purified pycnographically in a cesium chloride
density gradient which decreases the virus particle content by a factor of 10° with-
out change in the properties of the enzyme. An examination here reveals that
again one has parallel increases in both RNA and infectious units.

(b) Proof that the newly synthesized RNA molecules are infectious: The kind of
experiments just described offer plausible evidence for infectivity of the newly
synthesized radioactive RNA. However, they are not conclusive since they do
not eliminate the possibility that the agreement observed is fortuitous. One could
argue, however implausibly, that the enzyme is “activating” the infectivity of the
input RNA while synthesizing new noninfectious RNA, and that the rather com-
plex combination of exponential and linear kinetics of the two processes coincides
fortuitously.

Direct proof that the newly synthesized RNA is infectious can, in principle, be
Vou. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1549

obtained by experiments which employ N"-H*-labeled initial templates to generate
N4-P®2 product. The two can then, in principle, be separated in equilibrium
density gradients of cesium sulfate. Such experiments have been carried out for
other purposes and will be described elsewhere. However, the steepness of the
cesium sulfate gradient makes it difficult to achieve a separation clean enough to
be completely satisfying.

There exists, however, another approach which bypasses these technical dif-
ficulties by taking advantage of the biology of the situation and of the fact that
we are dealing with a presumed self-propagating entity. Consider a series of tubes
each containing 0.25 ml of the standard reaction mixture, but no added template.
The first tube is seeded with 0.2 y of Q8 RNA and incubated for a period adequate
for the synthesis of several y of radioactive RNA. An aliquot (50 A) is then trans-
ferred to the second tube which is in turn permitted to synthesize about the same
amount of RNA, a portion of which is again transferred to a third tube, and so on.

If each successive synthesis produces RNA which can serve to initiate the next
one, the experiment can be continued indefinitely and, in particular, until the point
is reached at which the initial RNA of tube 1 has been diluted to an insignificant
level. In fact, enough transfers can be made to ensure that the last tube contains
less than one strand of the input primer. Jf, in all the tubes, including the last one,
the number of infectious units corresponds to the amount of radtoactwe RNA found,
convincing evidence ts offered that the newly synthesized RNA is infectious.

A complete account of such a serial transfer experiment may be found in Spiegel-

 

75

  
    

oO
(74 ——~~-INFECTIOUS. UNITS

peg RNA

 

 

MINUTES
300

TRANSFERS

Fig. 6.—RNA synthesis and formation of infectious units in a serial transfer experiment. Six-
teen reaction mixtures of 0.25 ml were set up, each containing 40 y of protein and the other
components specified for the “standard” assay; 0.2 y of template RNA were added to tubes
0 and 1; RNA was extracted from the former immediately, and the latter was allowed to
incubate for 40 min. Then 50 of tube 1 were transferred to tube 3 and so on, each step af-
ter the first involving a 1 to 6 dilution of the input material. Every tube was transferred from
an ice bath to the 35°C water bath a few minutes before use to permit temperature equilibration.
After the transfer from a given tube, 20 A were removed to determine the amount of P*-
RNA synthesized, and the product was purified from the remainder. Control tubes incubated
for 60 min without the addition of the 0.2 7 of RNA showed no detectable RNA synthesis,
nor any increase in the number of infectious units. All recorded numbers are normalized to
0.25 ml. The ordinates represent cumulative increases of infectious units and radioactive RNA
in each transfer. The abscissa records elapsed time and the transfer number. Further details
are to be found in Spiegelman ef al.‘
1550 BIOCHEMISTRY: SPIEGELMAN AND HARUNA Proc. N. A. 8.

man etal.” Aside from controls, 15 transfers were involved, each resulting in a 1
to 6 dilution. By the eighth tube there was less than one infectious unit ascribable
to the initiating RNA, and the 15th tube contained less than one strand of the
initial input. Nevertheless, every tube showed an increment in infectious units
corresponding to the radioactive RNA found.

Figure 6 compares cumulative increments with time in newly synthesized RNA
and infectious units. The agreement between the increments in synthesized RNA
and newly appearing infectious units is excellent at every stage of the serial transfer
and continues to the last tube. Long after the initial RNA has been diluted to
insignificant levels, the RNA from one tube serves to initiate synthesis of biologically
competent RNA in the next. It is clear that every step and component necessary
to complete the replication must be represented in the reaction mixture described.

(9) Prospects for the Resolution of the RNA Replicating Mechanism—We may
conclude this discussion with an assessment of the current status of the RNA
replication problem and an indication of the direction of our present efforts.

Jt must be emphasized that the doubts raised (§ 4) about the ribonuclease-
resistant structures (RS) concern only their function. The structures are real and
their existence must ultimately be explained. Certain quantitative features of the
time, kinetics of appearance, and proportion of input strand involved in “RS” are
difficult to reconcile with a model which insists that they intervene between the
initial template and final product. Further, ribonuclease-resistant structures are
observed with purified replicase whenever it is functioning abnormally (e.g., with
fragments or in the presence of Mn++). 42 On the basis of these and other
difficulties, we maintain that a decision cannot be made at present on whether the RS
are replicative intermediates of unknown structure, nonreplicative intermediates of
unknown function, or simply nonfunctional artifacts.

The unambiguous analysis of a replicating mechanism demands evidence that
the reaction being studied is in fact generating replicas. Ultimately, therefore,
proof must be offered that the polynucleotide product contains the information
necessary for the production of the corresponding virus particle in a suitable test
system. The experiments described demonstrate that this rather rigorous require-
ment has finally been satisfied.

It should now be possible to study RNA replication in a simple system consisting
of purified replicase, template RNA, riboside triphosphates, and magnesium. How-
ever, this is a necessary condition, not one sufficient for success. Possession of an
enzyme of this sort does not, of itself, guarantee that any results observed are
necessarily relevant to the nature of the replicating reaction. Attempts at the
analysis of the replicating mechanism must recognize the implications of the fact
that the enzymes involved are likely to be complicated molecules. High levels of
complexity provide the flexibility which permits the occurrence of abnormalities, a
potentiality which can be accentuated by exposure to either strange environments
or unusual components. Thus, in the absence of primer, the DNA polymerase
eventually initiates the synthesis of an AT-copolymer.® In the presence of Mn++,
the same enzyme will incorporate riboside triphosphates into a mixed polymer.*!
Analogously, the DNA-dependent RNA polymerase synthesizes poly A if supplied
only with ATP, a reaction which is inhibited if the other riboside triphosphates are
added. ** Again, if presented with a single-stranded DNA, the transcriptase
VoL. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1551

synthesizes a DNA-RNA hybrid*4—*" and if the template is RNA, a duplex RNA
results.34: 3

The fact that such variations from the norm can occur makes it difficult to draw
incontestable conclusions from the appearance of a product in a reaction. Thus,
for example, as will be detailed elsewhere,“ replicase makes an RNase-resistant
structure if presented with either fragments of its own genome or intact heterologous
RNA. We recognize that the abnormal has often been fruitfully used in the study
of the normal and that even artifacts can ultimately serve to illuminate the reaction
in which they are generated. However, it is first necessary to identify the normal.
We insist, therefore, that in the test tube even more than in the cell, evidence other than
mere existence must be provided before a component found is accepted as a normal
intermediate of the replicative process.

The study of the normal functioning of the replicases described requires intact
homologous RNA and the avoidance of Mn++. Furthermore, even under optimal
conditions, as we know them, prolonged functioning of these enzymes in the enzy-
mologist’s test tube can create the possibility of accumulating abnormalities.*

Since the enzyme reaction described here does in fact produce RNA strands bio-
logically indistinguishable from the input templates, it should be possible to test
all the implications of any proposed mechanism. If two enzymes are required,
both must be present and it should be possible either to establish their existence or to
prove that one is sufficient. If an intermediate replicating stage intervenes between
the template and the identical copy, then these forms should be demonstrably present
in the reaction mixture. All experiments designed to test these alternatives must
be continually monitored for biologically active product to ensure that the normal
reaction is being followed.

A rather strong negative conclusion can be drawn from the data summarized
concerning the possible role of transcriptase as the “second enzyme” for RNA
replication, a mechanism some find attractive. The complete absence of detectable
transcriptase from our preparation would appear to eliminate it as a participant in
RNA replication.

It seems likely that the most telling data are derivable from experiments in which
the initiation of new chains is synchronized. To begin with, the examination of the
product synthesized, prior to the appearance of mature strands, can be compared
with that formed in more extensive synthesis. The use of different isotopes
on template (e.g., H*) and product (e.g., P*?) permits a sensitive search for inter-
mediate complexes between the two, a prediction of the ¢X-174 model.

We may briefly list potentially informative experiments which use these and
other devices:

(1) There might be a comparison of ribonuclease resistance of product and
template at various stages of synthesis.

(2) A-search could be made for a physical complex between the P*? product and
the H? template in sucrose gradients and in equilibrium density gradients of Cs,8Ox..
In the latter the templates can, in some cases, be additionally labeled with N™ to
give them a unique density position. Here the early (1-5% synthesis) events are
most crucial.

(3) A detailed analysis could be made of the base composition during the
progress of early synthesis. The resulting data are particularly informative in the
1552 BIOCHEMISTRY: SPIEGELMAN AND HARUNA Proc. N. A. S.

case of Q@, since its A/U ratio is 0.75 and that of its complement is, therefore, 1.33.
Consequently, the formation of the complement as an initial step is easily detected.

(4) Along similar lines, a comparison of nearest-neighbor analysis to all four
bases in early and late synthesis should reveal whether a complement or the identical
copy is being made in the early periods.

(5) The degree of complementarity between the product and the original
template at various stages of strand formation could be determined by hybridiza-
tion tests. In this connection, it may be noted that the required annealing ex-
periments are not as simple, either logically or technically, as some recent contribu-
tions would suggest.

(6) The involvement of replicating complexes or complementary strands might
not be detected by any of these experimental devices if they pre-existed in the
enzyme preparations, either free or associated with active enzyme molecules.
Here, however, advantage can be taken of the size (2 X 10° for RF and 1 X 106 for
the complementary RNA) and the density of RNA or RNA-enzyme complexes.
Enzyme can be isolated pycnographically in a density gradient at a density char-
acteristic for nucleic acid-free protein, followed by characterization for size in a
sucrose gradient. If the resulting enzyme is active and still completely satisfied by
viral RNA, pre-existing complements or duplexes can hardly be invoked to explain
their properties.

Virtually all the experiments listed above have been carried out and a few are
in the final stages of completion. The detailed data and conclusions will be re-
corded elsewhere.** Here we may state that, thus far, we have found no evidence to
encourage the idea that a duplex containing the mature strand and its comple-
ment plays a role in replication.

It is important to emphasize that none of this should be taken to mean that our
experiments have eliminated the use of complementarity in RNA replication.
There are readily designed mechanisms which involve complementarity without
requiring the synthesis of an intermediate duplex or the complete complementary
strand. An extreme example may be briefly noted: Consider the possibility that a
representative of each of the four nucleotides is attached to the enzyme. These
could be permanent components or replaceable ones and are used by the enzyme for
complementary reading of the template as a guide, via ‘allosteric instruction,” for
building an identical copy. Other mechanisms involving transient partial com-
plements can also be devised.

It seems likely that many of the uncertainties which still exist about RNA replica-
tion will yield relatively soon to the proper experiments. We are tempted to end
the present discussion with only a slight modification of the conclusion used in an
earlier*® essay on protein synthesis. ‘The crucial experiments have not yet been
executed. However, the systems required for their performance are with us, or
close to hand. The outlook is depressingly bright for the quick resolution of
another interesting problem.”

* This investigation was supported by U.S. Public Health Service research grant no. CA-01094
from the National Cancer Institute and grant no. 2169 from the National Science Foundation.

1 Loeb, T., and N. D. Zinder, these ProcrEpinas, 47, 282 (1961).

2Temin, H. M., these ProckEpiINnes, 52, 323 (1964).

3 Doi, R. H., and 8. Spiegelman, Science, 138, 1270 (1962).
VoL. 55, 1966 BIOCHEMISTRY: SPIEGELMAN AND HARUNA 1553

4 Hall, B. D., and 8. Spiegelman, these ProcrEpINGs, 47, 137 (1961).

5 Yankofsky, S. A., and 8. Spiegelman, these PRocEEDINGS, 48, 1069 (1962).

8 Ibid., p. 1466.

7 Giacomoni, D., and 8. Spiegelman, Science, 138, 1328 (1962).

8 Spiegelman, 8., and R. H. Doi, in Synthesis and Structure of Macromolecules, Cold Spring
Harbor Symposia on Quantitative Biology, vol. 28 (1963), p. 109.

9 Hayashi, M., and S. Spiegelman, these ProcerpinGs, 47, 1564 (1961).

1 Goodman, H. M., and A. Rich, these ProcEEDINGs, 48, 2101 (1962).

Hurwitz, J., J. J. Furth, M. Malamy, and M. Alexander, these ProceEpings, 48, 1222
(1962).

12 Levinthal, C., A. Keynan, and A. Higa, these ProceEpines, 48, 1631 (1962).

13 Reich, E., R. M. Franklin, A. J. Shatkin, and E. L. Tatum, these Procrrpines, 48, 1238
(1962). .

™ Nathans, D., G. Notani, J. H. Schwartz, and N. D. Zinder, these ProcrEpines, 48, 1424
(1962).

156 Ohtaka, Y., and S. Spiegelman, Sczence, 142, 493 (1963).

16 Clark, J. M., Jr., A. Y. Chang, 8. Spiegelman, and M. E. Reichmann, these Procnepings,
54, 1193 (1965).

17 Doi, R. H., and S. Spiegelman, these ProcrEpines, 49, 353 (1963).

18 Sinsheimer, R. L., J. Mol. Biol., 1, 43 (1959).

19 Sinsheimer, R. L., B. Starman, C. Naglar, and 8. Guthrie, J. Mol. Biol., 4, 142 (1962).

20 Hayashi, M., M. N. Hayashi, and 8. Spiegelman, Science, 140, 1313 (1963).

21 Hayashi, M., M. N. Hayashi, and S. Spiegelman, these ProceEpinas, 51, 351 (1964).

22 Thid., 50, 664 (1963).

23 Ochoa, 8., C. Weissmann, P. Borst, R. H. Burdon, and M. A. Billeter, Federation Proc., 23,
1285 (1964).

2% Kelley, R. B., and R. L. Sinsheimer, J. Mol. Biol., 8, 602 (1964).

*% Kaerner, H. C., and H. Hoffman-Berling, Nature, 202, 1012 (1964).

26 Weissmann, C., P. Borst, R. H. Burdon, M. A. Billeter, and 8. Ochoa, these ProcrEpINes,
51, 682 (1964).

27 Krikson, R. L., M. L. Fenwick, and R. M. Franklin, J. Mol. Biol., 10, 519 (1964).

22 Ammann, J., H. Delius, and P. H. Hofschneider, J. Mol. Biol., 10, 557 (1964).

28 Nonayama, M., and Y. Ikeda, J. Mol. Biol., 9, 763 (1964).

30 Fenwick, M. L., R. L. Erikson, and R. M. Franklin, Science, 146, 527 (1964).

31 Smellie, R. M. 8., in Progress in Nucleic Acid Research, ed. J. M. Wavidson and W. E. Cohn
(New York: Academic Press, 1963), vol. 1, p. 27.

32 Grunberg-Manago, M., and 8. Ochoa, J. Am. Chem. Soc., 77, 3165 (1955).

33 August, J. T., P. J. Ortiz, and J. Hurwitz, J. Biol. Chem., 237, 3786 (1962).

34 Nakamoto, T., and S. B. Weiss, these ProcrEpINGs, 48, 880 (1962).

% Krakow, J. 8., and 8. Ochoa, these ProcrEpines, 49, 88 (1963).

36 Haruna, I., K. Nozu, Y. Ohtaka, and S. Spiegelman, these ProceEpines, 50, 905 (1963).

37 Watanabe, I., Nihon Rinsho, 22, 243 (1964).

38 Overby, L. R., G. H. Barlow, R. H. Doi, Monique Jacob, and 8. Spiegelman, J. Bacteriol.,
91, 442 (1966).

33 Tbid., in press.

40 Haruna, I., and 8. Spiegelman, these PRocEEDINGs, 54, 579 (1965).

41 Baltimore, D., and R. M. Franklin, Biochem. Biophys. Res. Commun., 9, 388 (1963).

42 Weissmann, C., L. Simon, and 8. Ochoa, these Procrepines, 49, 407 (1963).

43 August, J. T., S. Cooper, L. Shapiro, and N. D. Zinder, in Synthesis and Structure of Macro-
molecules, Cold Spring Harbor Symposia on Quantitative Biology, vol. 28 (1963), pp. 95-97.

44 August, J. T., L. Shapiro, and L. Eoyang, J. Mol. Biol., 11, 257 (1965).

45 Shapiro, L., and J. T. August, J. Mol. Biol., 11, 272 (1968).

46 Haruna, I., and 8. Spiegelman, Science, 150, 884 (1965).

47 Spiegelman, 8., I. Haruna, I. B. Holland, G. Beaudreau, and D. Mills, these ProcEepINas,
54, 919 (1965).

48 Haruna, I., and 8. Spiegelman, in preparation.
1554 BIOCHEMISTRY: SPIEGELMAN AND HARUNA Proc. N. A. 8.

*8 Spiegelman, 8., in The Chemical Basis of Heredity, ed. W. D. McElroy and Bentley Glass
(Baltimore, Maryland: Johns Hopkins Press, 1957), pp. 232-267.

5° Schachman, H. K., J. Adler, C. M. Radding, I. R. Lehman, and A. Kornberg, J. Biol. Chem.,
235, 3242 (1960).

5! Berg, P., H. Fancher, and M. Chamberlin, in The Synthesis of Mixed Polynucleotides Containing
Frbo and Deoxynucleotides by Purified DNA Polymerase, ed. H. J. Vogel, V. Bryson, and J. O.
Lampen (New York: Academic Press, 1963).

5? Spiegelman, S., “Protein and nucleic acid synthesis in subcellular fraction of bacterial cells,”
in Recent Progress in Microbiology, VII International Cong. for Microbiology (Uppsala: Almquist
and Wiksells Boktryckeri AG.), pp. 81-103.

53 Chamberlin, M., and P. Berg, these ProcrEpiN@s, 48, 81 (1962).

54 Chamberlin, M., and P. Berg, J. Mol. Biol., 8, 297 (1964).

55 Sinsheimer, R. L., and M. Lawrence, J. Mol. Biol., 8, 289 (1964).

56 Bassel, A., M. Hayashi, and S. Spiegelman, these PRoceEDINGs, 52, 796 (1964).

57 Hayashi, M. N., M. Hayashi, and 8. Spiegelman, Biophys. J., 5,231 (1965).