Reprinted from the PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES Vol. 63, No. 3, pp. 805-811. July, 1969. FURTHER EXTRACELLULAR DARWINIAN EXPERIMENTS WITH REPLICATING RNA MOLECULES: DIVERSE VARIANTS ISOLATED UNDER DIFFERENT SELECTIVE CONDITIONS BY REUBEN LEVISOHN~ AND S. SPIEGELMAN DEPARTMENT OF MICROBIOLOGY, UNIVERSITY OF ILLINOIS, URBANA Communicated May 9, 1969 Abstract.-Experiments are described which demonstrate that it is possible to isolate in vitro a variety of mutant RNA molecules which exhibit qualitatively distinguishable phenotypes. The results suggest that precellular evolution could have involved selective forces of previously unsuspected diversity and subtlety. Suitable adjustment of the selective conditions leads to the isolation of variants optimally designed to compete successfully with the original viral nucleic acid. One of the properties that can be built into the variants is resistance to the pres- ence of inhibitory analogues of the normal riboside triphosphates. Potentially, such variants could be used as antiviral devices in conjunction with the more usual chemotherapeutic agents. Introduction.-Proof that purified QP-replicase1 catalyzes the synthesis of both normal2, and mutant4 infectious QP-RNA established that the RNA is the in- structive agent in the replicative process. The fact that the RNA niolecule satisfies the operational definition of a self-duplicating entity generated the possibility of performing extracellular Darwinian experiments. The first step in exploiting the inherent potentialities of this system was a serial transfer experiment which resulted5 in the selection of variant V-1. This mutant replicated some 15 times faster than QP-RNA and retained 5.50 of the 3600 residues originally present in the parental molecules. We then showed6 that purified Qp-replicase can be initiated to synthesize copies by a single molecule of template. The resulting clone of descendants pos- sessed evident advantages for sequence studies and in addition made possible the inception of an in vitro genetics of replicating molecules. In performing these ex- periments, a new variant (V-2) was isolated which replicated faster than V-1. Thus, measurable RNA synthesis occurred in a 15-minute reaction when initiated with as little as 0.29 pFpg of V-2. However, more than 300 times as much is re- quired with v-1. This phenotypic difference has been maintained over many transfers. The experiments thus far described were concerned with the isolation of mutants possessing increased growth rates under standard conditions. We here turn our attention to a question of no little theoretical and practical interest and inquire whether other mutant types can be isolated. In effect, we are asking the following question : Can qualitatively distinguishable phenotypes be exhibited by a nucleic acid molecule under conditions in which its information is replicated but never translated? The results to be reported show that numerous differenti- able variants can be isolated, the number depending on the ingenuity expended in designing tl Le appropriate selective conditions. 805 806 GEiVBI'ICS: LEVISOHN AND SPIEGELdfAlV Prtoc. N. A. S. 5 n ; *, *p- `J FIG. l.-CTP concentration cnrve. Re- action mixtures of 0.125 ml containing the indicated concent,rations of CTP and other- 60 80 100 40 2o mp rf)31es CTPK1125ml VOL. 63, 1969 GENETICS: LEVISOHN AND SI'IEGELMAN 807 concentration of 2 mpmoles, the rate of synthesis of V-2 is only 25 per cent' of normal. At 1 mpmole of CTP the rate decreases t,o 5 per cent of normal. With this informatioil available, a search was made for variants which could replicate better than V-2 on limiting levels of CTI'. A serial transfer experiment, at 2 mpmoles of CT1' per reaction was initiated with &@-RKA, culminating aft>er ten transfers with the appearance of V-4. A second series of transfers at 1 mpmole of CTP per reaction was then started with V-4 and after 40 transfers led to the isolation of V-6. .Figure 2 describes the replication of variants 2, 4, and 6 in limit,ing CT1' (1 mpmole per 0.125 ml). The slopes of the semilog plots permit an estimat,ion of the doubling times during logarithmic increase as 1.81 minutes for V-2, 1.41 minutes for V-4, and 1.IG minutes for V-6. Evident'ly both V-4 and V-6 possess a heritable feature which permits them to overcome the disadvantages imposed by the low level of CTP. FIG. 2.-Synthesis of variants on ' a reaction mixtiire with a limiting [ Limiting CTP CTP concentration. A reaction !z5- mixtiire coritaiiiing only 1 Inpmole (instead of 100 nipmoles) CTP, in- cluding 2.2 X lo5 cpni I13--CTP, was incubated at S8"C with 40 pg QP- replicase and 0.001 prg of purified 'i V-2, V-4, or V-6 RNA. The E amolint of trichloroacetic acid-in- soluble radioactive material was determined at the indicated times. In this experiment 1.7 X lo5 cpni is equivaleiit to 1 pg of variaiit RNA. Min. (b) Tlie basis (tf the mufaiit plmotype: The fact that variants 4 arid 6 replicate 28 per cent arid 56 per cent better, reipectively, than V-2 at lorn levels of CTI' might be explained on thc ba45 of smaller size5 or modification of base composi- tion towards a lower cytosine content. Figure 3 compare5 the sixes of V-2 and V-6 on po1yacrgl:~mide gels and shows that there is no sigriificarit difference in chaiii length. 111 passing, it may be noted that similar comparisoiis revealed very little chaiige in length in any of the vari- ants thus far selccted. Herc again, no significant differcrices are detectable. It should he noted that the sensitivity of this test is of the order of 0.5 per cent and chaiigcs iiivolvirig a small number of residues would be difficult to detect. It is evident that the modifications leading to the properties possessed by variants 4 arid 6 do riot involve massive modification in the composition of the molecule. The identification of the changes will require more subtle examina- tions such as oligonucleotide fingerprint patterns, and these are being carried out. The fact that the most obvious pathways for solving the problem of low CTP were not employed leads one to consider more sophisticated devices for achieving the desired end result. It is useful here to recall that the mutant RNA molecules Table 1 compares the have composition of V-2 arid V-G. 808 7EiVETICS: LEVI'ISOHN .AND SI'IEGEIAIAN PROC. N. A. S. 10 mm- FIG. 3.-Gel electrophoresis of V-2 and V-6. Standard reaction mixtures of 0.0125 ml initiated with 0.001 pg of V-2 in the presence of 2.2 X 106 cpm H3-UTP, or 0.001 pg of V-6 in the presence of 2 X IO5 cpm PZ3-UTP were incubated for 18 and 13 min, respectively. The reactions were terminated with 0.01 ml 2.5% sodium dodecyl snlfate and mixed together. The mixture was electrophoresced through polyacrylamide gels and processed as in Materials and Methods. TABLE: 1. Base ratios of variants. Base C A G U v-2 24.8 23.2 26.6 25.4 V-6 24.8 23.5 26.7 2.5.1 The base composition of purified plus strands of variant 2 and variant 6 were determined as de- scribed in Methods. The numbers represent mole per cent. must complex with the replicase. Thus, changes of sequence which would leave such gross features as base composition and size unchanged could, nevertheless, lead to different secondary structures of the mutant molecules. These in turn could have allosteric egects on the replicase, permitting the complex to employ CTP more effectively at suboptimal concentrations. If this were the case, and if there were a common site for the four riboside triphosphates analogous to the DNA polymerase,*O it might be expected that a mutant selected for better replica- tion on low CTP would also exhibit increased capacities to accommodate to low levels of the other riboside triphosphates. Table 2 summarizes data comparing the logarithmic synthesis of variants V-4 V-6 with that of V-2 on limiting levels of each of the four riboside triphosphates. The data show that the two variants selected on low CTP also do much better OII limiting concentrations of the other three substrates. More definitive delinea- tion of the underlying mechanism will require binding studies of substrates with enzyme complexed to mutant and wide-type templates. Tubercidiri is an analogue of adenosine in which the nitrogen atom in position 7 is replaced by a carbon atom. Tubercidin triphosphate (TuTP) inhibits the synthesis of QP-RNA in vitro (Alan Kapular, personal communication). TuTP cannot completely replace ATP in the reaction. It is clear, however, from the following indirect experiment that it is incorporated into the growing chains. A series of reactions (2) Selection of a eariant resistant to an inhibitoyy analogue: VOL. 63, 1969 GENETICS: LEVISOHN AND SPIEGELMAN 809 TABLE 2. Logarithmic synthesis of variant RNA on limiting media. Doubling -Relative Slope--- Limiting time Cornpared Compared nucleotides Mpmoles Variant (min) to v-2 to v-4 None 100 2 0.42 1.00 ... ... 4 0.42 1.00 1.00 ... 6 0.33 1.21 1.21 ATP 4 2 2.41 1.00 ... ... 4 1.54 1.56 1.00 ... 6 1.47 1.64 1.05 CTP 1 2 1.81 1.00 ... ... 4 1.41 1.28 1.00 ... 6 1.16 1.56 1.21 GTP 9 2 3.03 1 .oo ... ... 4 2.25 1.35 1.00 ... 6 2.31 1.31 0.97 ... 4 1.69 1.22 1.00 ... 6 1.54 1.33 1.09 UTP 2 2 2.06 1.00 ... The data are based on the experiment shown in Fig. 2 and similar experiments performed in stand- ard reaction mixture and in reaction mixtures with only 4 mpmoles ATP, 9 mpmoles GTP, or 2 mp- moles UTP. were run at increasing levels of TuTP in the presence of fixed amounts of P-UTP and H3-ATP. The latter permits determination of the U to A ratio in the prod- uct. Figure 4 shows the outcome which indicates that TuTP can replace A but with a less than equal probability. It was of some interest to see whether one could derive a mutant which would show resistance to the presence of this agent. In such experiments, it is desirable to have the ratio of the analogue to ATP as high as possible. To attain this more readily, a variant was isolated on limiting ATP concentration. Variant 6 was chosen to start a series of transfers in a reaction mixture containing 1.5 mpmoles of ATP, and this led to the isolation of V-S. The doubling time of V-5 in the reaction mixture with 1.5 mpmoles of ATP was 2.5 minutes as compared with 5.4 minutes for V-6, the starting variant. The replication rate of V-S on a reaction mixture containing 5 mpmoles of ATP was inhibited fourfold upon the addition of 30 mpmoles of TuTP. A serial transfer was initiated with V-S on the inhibitory medium and led to the isolation of V-9. The doubling time of V-9 in the presence of TuTP was 2.0 minutes as compared with 4.1 minutes for V-S. In the absence of TuTP, both variants are synthesized with a 1.0-minute doubling time. It is clear that V-9 exhibits a specifically increased resistance to the inhibitory effect of TuTP. The resistance mechanism does not involve a more effective exclusion of TuTP as measured by an experiment similar to that described in Figure 3. Thus, at 30 mpmoles of TuTP and 5 mpmoles of ATP, the ratio of U to A in the product was 3.6 for V-8 and 3.5 for V-9, the resistant mutant. Discussion.-Table 3 lists the variants isolated in the experiments described and summarizes the relevant information on their origins and conditions of selec- tion. It will be noted that V-4 is an independent derivative from the parental QP-RNA. Another variant V-3 (not listed) was isolated with limiting CTP starting with V-2 instead of Qp-RNA. V-3 possesses phenotypic properties 81 0 Equ!valent ,' Subst~tution / , / Actual No substitution I I I 0 5 IO Input ratio [LKE' IA T PI Fro. 4.--Substitution of ATP by TriTP. Synthesis of RNA templated by 0.001 pg 1'4 RNA in the presence of 40 pg &@replicase was allowed to take place for 40 min in a reaction niixt,iire coiltailling 5 mpmnles ATP, the indicated amounts of Tu- TP, and I13-ATP and P32-U`l'P. The U to A ratio in plus straiids of variant is 6/5, which we may take as iiiiity. If triljercidiii can replace adenosine with equal probabilit,y, the U to A ratio in lhe product shoiild vary with the ratio of TuTP to ATP in the reactioii mixtine iii the miiiiiier descrit)etl by the upper dashed curve (labeled cquivalcnt suhstitrction). If there is no siibstitiition of adenosine by tuber- cidin, the ratio of U to A shoiild remain normal and illdependent of t.he relat,ive amoiuits of TnTP present (lower dashed ciirve labeled no suhsfitulion). The 1111- broken line indicates the actiittl iiirorporation ratio corrected for an input of 6 X 105 cpm/100 mpmoles PZZ-UTP aiitl 5 x 105 (ym/lOO mpmoles H:<-ATP. TARLI: 3. Conditions used in isolation of variants. RNA used to Variant Selective liniitatioiis >tart selectioii 11-26 None QB v-4 2 mpmolei CTP (20 Y-6 1 (` CTP 1-4 T-8 1.3 ATP v-6 v-9 5 `` .4TP 1-8 +30 `` TuTP ... No. of traiisfers of 1.25 X 104 transfers at dilution Total no. of ... 17 > 10 30 40 11 16 1 5 19 ... .., Variants were selerted on standard reactioii inixtnre, or on a st>andnrd re:tct.ion iiiixt,ure modified to contain one of the four nucleoside triphosphates at the indirated coilcentration. St,arting with the RNA's indicated in column 3, a series of transfers were ~nade with reaction product, diluted 1.25 X IOLfold, as detailed in Mdhods. Snl~sequently, the dilutiori factor between transfers was gradually increased to about 1 X 10". indistinguishable from those of V-4. Thus, one cnn arrive at the V-4 phenotype either from QP-RNA or from V-2. It seems probable that Qp-RXA passes through the V-2 stage before arriving at the V-4 phenotype. The comparatively conscrv:ttive nature of the replicative process is illustrated by the virtual identity of hahe compohitions of V-2 and V-G seen in Table 1. These two mutants are intlependent isolates :ind :ue separated I)y two lengthy and severe selections on limiting CTI'. Xo reflection of this is seen in the base com- positions of the two. It will he of eiiormous irit,erest, to comparc tlic :xtuxl sequence changes among the mutants differing in their rehtedne. lid phenotypic properties. With this informat'ion available, one can begin t'o coiixtriict, the protiable secoridary struc- ture modifications. Only t'lieri will we be in a position t'o begin the attempt t,o understand the molecular hasis of thesc new phenotypes. We pointed out previously5 that extracellular Darwinian selections may mimic one aspect of precellular evolution, Le., when environmental selection operated only 011 the replicating gene arid riot on the gene product. Such experiments pro- vide some insight, into the rules of these early st,:iges of cvolution. It was not a priori obvious what kinds of selective forces could he operative since much de- pended on how many different ways a molecule could bc selected as superior by the environment. The experiments reported here reveal an unexpected wealth of phenot,ypic diff ereiices which :t replicat'irig iiiieleic acid molecule can exhibit. It is t'rue that many of these irivolve interaction bet,ween nucleic acid molecules and a highly evolved protein catalyst,. However, it is possible lo imagine similar t,ypes of interactions with a primitive surface catalyst. Scquciicc changes which would increase slightly the cat,ulyt,ic effectiveiicss could have powerful selective effects in these precellular stages of evolving genetic material. It is apparent from the limited number of ex:tmples described t,hat8 a host of new mutant types possessing predetermined phciiotypes c;tn be isolated by varying other parameters of t.he system. In addit'ioii, one c:tn expand the possibilities by introducing initially neutral agents (e.g., proteins) with which the replicating molecules may interact. Selection c:m then be exerted to favor variants that, can induce these foreign agents to become p:trticip:mts in the replicative process. Finally, we should like to note a practical implication of tlie mutant resistant to the inhibitory analogue TuTI'. We pointed out earlier5 that, t'hese abbreviated variants possess a number of features which makc t,hem potentially powerful tools as chemotherapeutic agents. They combine a very high affiriit,y for the replicase arid a rapid growt'h rate. They compete effectively with the normal viral nucleic acid for the replicase and thus halt the progress of virus production. To these features we can now add a third, iinmely, rcsistarice to a chemothera- peutic agent effective against the origjna.1 virus partJicle. All these features can be built into one variant by the kinds of serial selections described here. This adds anobher dimension to the potential use of these agents as chemotherapeutic devices. * This investigation was siipport,ed by U.S. Public Health Service research grant CA-01094 t Recipient of a 1)amoii Xiinyon Cancer Ilesearch Fellowship. from the National Cancer Institute arid the National Science Fomidation. Harunn, I., and S. Spiegelman, these PROCI.:I.:DINGS, 54, 579 (1965). Spiegelnian, S., I. Hariiiia, I. B. Holland, 0. Beaitdreari, and I). R. Mills, these PROCI:I:I)- Pace, N. R., and S. Spiegelman, these PRO^ Pace, N. R., and S. Spiegelnian, Scicncc, 153, 64 (1966). Mills, 1). R., 12. L. Peterson, and S. Spiegelnian, these PRO Levisohn, It., and S. Spiegelman, these P Bishop, I). 11. I,., J. 11. Clayhrook, and S. Bishop, D. 13. L., 1). It. Mills, and S. Spiegelmmi, Biochonistry, 7, 3744 (1968). INGS, 54, 919 (196.5). DINGS, 55,1608 (1966). DINGS, 58, 217 (1967). DIKGS, 60, 866 (1068). elrnaii, J. Mol. Rid., 26, 373 (1967). * Sanger, F., G. G. Brownlee, arid B. G. Barrel], J. Mol. Hiol., 13, 373 (1965). lo Atkinson, 11. R., J. A. I3itbermnii, 11. B. Kelly, and A. Kornberg, Federation Proc., 28 347 (1969).