Ribonucleic acid chain elongation by Escherichia coli ribonucleic acid polymerase. I. Isolation of ternary complexes and the kinetics of elongation.

Abstract The chain elongation portion of enzymatic RNA synthesis has been studied by employing a ternary complex containing RNA polymerase, a DNA template, and product RNA. Ternary complexes were isolated by passage through a gel exclusion column and were shown to carry out RNA chain elongation under conditions that preclude RNA chain initiation or termination. The properties of the elongation reaction have been studied by employing ternary complexes formed with synthetic DNA templates. A simple ping-pong kinetic model was derived and was shown to fit the data obtained with alternating copolymer templates. The model is also able to account for the inhibition observed at high concentrations of nucleoside triphosphates. A general rate equation can be devised for RNA chain elongation with DNA templates of complex sequence if one assumes that the Michaelis constants for the substrates are independent of the DNA sequence. This assumption appears to hold for the various synthetic DNA templates that we have examined but has not been adequately tested for more complex templates.


J. CHAMBERLIN
From the Departments of Genetics and Biochemistry, University of California, Berkeley, California 9&???0 SUMMARY The chain elongation portion of enzymatic RNA synthesis has been studied by employing a ternary complex containing RNA polymerase, a DNA template, and product RNA. Ternary complexes were isolated by passage through a gel exclusion column and were shown to carry out RNA chain elongation under conditions that preclude RNA chain initiation or termination.
The properties of the elongation reaction have been studied by employing ternary complexes formed with synthetic DNA templates.
A simple ping-pong kinetic model was derived and was shown to fit the data obtained with alternating copolymer templates. The model is also able to account for the inhibition observed at high concentrations of nucleoside triphosphates.
A general rate equation can be devised for RNA chain elongation with DNA templates of complex sequence if one assumes that the Michaelis constants for the substrates are independent of the DNA sequence.
This assumption appears to hold for the various synthetic DNA templates that we have examined but has not been adequately tested for more complex templates.
In order to minimize contributions due to initiation, chain elongation has been studied by employing complex trapped on filters (6), assaying in the presence of elevated salt (5, 7, S), or using oligonucleotide primers (9, 10). Some kinetic analysis has also been done assuming that incorporation in a normal assay system reflects only the chain elongation phase of the reaction (11). However, recent studies make it clear that extensive RNA chain termination and re-initiation can occur in the complete reaction system (12)(13)(14) ; hence, study of the RNA chain elongation phase requires that these reactions be positively excluded.
In addition, it is desirable to avoid the use of elevated salt concentrations to block RNA chain initiation, since such salt concentrations alter the rate of RNA chain elongation (15,16) and also enhance RNA chain termination (12)(13)(14). In this report we describe the isolation of ternary complexes of Escherichia coli RNA polymerase with several synthetic DNA templates and their complementary RNA products. These complexes actively carry out the chain elongation phase of RNA synthesis at a constant rate, even in the presence of the drug rifampicin, which selectively blocks RNA chain initiation. Hence the properties of the RNA chain elongation reaction can be studied with these complexes in the absence of complexities due to initiation or termination.
A simple steady state model for enzymatic RNA chain elongation is presented.
Synthesis of RNA by DNA-dependent RNA polymerase in vitro proceeds in a series of steps. These normally include DNA template binding, RNA chain initiation, RNA chain elongation, and RNA chain termination and enzyme release (1). To understand the mechanism of RNA synthesis fully it is necessary to know the properties of each of these partial reactions.
However, study of the partial reactions is not normally possible in the in u&o reaction system, where all of the steps occur together in a complex temporal sequence. Thus, to study the partial reactions, assay procedures must be devised which allow the characterization of each step separately from the others. nucleotide, and 1 mM UTP. This mixture was incubated for 1 min at 37" and then ATP was added to give a final concentration of 1 mM. The reaction was incubated for 1 min and svnthesis was stotmed bv the addition of 0.2 ml of 0.1 mM EDTA (PH 8.0). The mixture &as chilled and was loaded onto a Bio-Gel P-100 column (2 X 60 cm) equilibrated with a solution containing Studies of the chain elongation phase of RNA synthesis have shown that after RNA Chain initiation a highly stable ternary complex containing RNA polymerase, template DNA, and the nascent RNA chain is formed.
Chain elongation by this complex * This research was supported by Research Grant GM12010 from the National Institute of General Medical Sciences. Reprints of this article will not be available. 0.02 M Tris-HCl (pH 8.0), 1 rnM MgCL, 0.1 mM dithiothreitol, 0.1 mM EDTA. and 5% glvcerol: the complex was then eluted with the same buffe;.
The-&&lex &as located by ultraviolet absorption, and peak fractions w&e pooled and asiayed for elongation activitv.
Procedures for the isolation of comalexes with different temilates were similar, except that the necessary nucleoside triphosphate substitutions were made. Complexes were stored at 4'; the dAT complex was found to lose elongation activity slowly, with one-half of the activity being lost in approximately 1 week.
Assay-Ternary complexes were assayed by employing a procedure similar to that used by Berg el al. (17), with the following modifications.
Assay reaction mixtures (0.25 ml) contained 40 mM Tris-HCl @H 8.0), 10 mM MgCls, 10 mM 2-mercaptoethanol, 50 mM KCl, and 10 pg of rifamycin per ml. After the reaction was terminated with ice-cold perchloric acid, 0.1 ml of yeast RNA (2 mg per ml) was added to e&h assay tube to facilitate precipitation of the labeled Dolvnucleotide.
One unit of elongation is defined as that amount of"complex which incorporates one pmole of total nucleotide into polynucleotide in 10 min, all substrates at a concentration of 200 PM.
Error Analysis-Kinetic constants were calculated from the experimental data by the statistical procedures outlined by Cleland (18). This analysis also allows one to calculate variances in the fitted constants; these error estimates were in the range of 1 to 370 of the constants.
Thus the experimental data obtained in any single experiment appear to fit quite closely the theoretical expressions used. The experimental deviation among KS values obtained in 12 independent experiments with poly[d (A-T)] complexes showed a variance of from 10 to 12Yo (Table I) One can test the purity of the nucleoside triphosphate solutions by using an elongation complex with an alternating copolymer template and adding one substrate together with another nucleoside triphosphate which is not a substrate.
Any incorporation reflects contamination with the missing substrate, and the concentration of the contaminant can be calculated from Eauation 11 below. With this method it was found that the GTP &paration that we have used was free of ATP but contaminated kith approximately 0.01% UTP. The CTP preparation employed was essentiallv free of both ATP and UTP. The trace contamination of the GTP was below that needed to affect any of the kinetic parameters qualitatively.
Materials-Escherichi coli polymerase was purified from E. (20). The products of the reaction were characterized by nearest neighbor analysis of RNA transcribed from the template (22); the sequence homogeneity of all DNA templates used was judged to be greater than 95y0 by this method.
[or-81P]ATP was synthesized according to the method of Symons (23) and was purified by a procedure devised by James Dahlberg.2 Rifamycin (rifampin, Calbiochem grade B) solutions were dissolved in aqueous buffer and stored at 4' in the dark. Unlabeled nucleoside triphosphates were purchased from Sigma Chemical Co., tritiated UTP from New England Nuclear Corp., oligo[d(A-G)*d(T-C)] from Collaborative Research Inc., and deoxynucleoside triphosphates from Calbiochem.

Chain Elongation by PoZy[d(A-T)]-containing Complex
The properties of ternary complexes obtained after gel exclusion chromatography were studied in greatest detail by employing the complex isolated with poly[d (A-T)] copolymer as template. Fractions eluted from the gel exclusion column which contained the poly[d (A-T)] copolymer template as monitored by 1 W. Mangel, personal communication.
absorbance at 260 nm were active in incorporating [u-~~P]AMP into poly[r (A-U)] copolymer in a standard chain elongation assay. The kinetics of poly[r (A-U)] synthesis under these conditions (Fig. 1) shows a constant rate of incorporation for at least 10 min, followed by a slow reduction in rate during the next 30 min. All rate measurements given in this paper were made in the initial, linear range of the elongation assay and the linearity of the initial rate was checked at several triphosphate concentrations whenever the assay conditions were varied.
As expected, the rate of poly-[r(A-U)] synthesis is directly proportional to the amount of ternary complex added over the assay range from 0.5 to 200 elongation units.
If one could determine the number of RNA polymerase molecules in active ternary complexes, one would be able to determine directly the rate of poly[r (A-U)] chain elongation and the concentration of active ternary complexes in a given solution. This has not yet been possible by any direct means that we have tried. An estimate of this number can be obtained by measuring the initial rate of poly[r (A-U)] synthesis in a normal RNA polymerase assay. The maximum specific activity reported for E. coli RNA polymerase holoenzyme is 24,000 units per mg (13, 17), which gives an initial rate of 34 0, assuming that all enzyme molecules are active. At least 80% of the enzyme molecules are active in such a preparation, as measured by their ability to initiate T7 RNA chains (13,24). Thus, unless there is an appreciable alteration of the rate of RNA chain elongation as the length of the poly[r (A-U)] chain is increased, a reasonable estimate of V,,, for poly[r (A-U)] chain growth is probably 35 to 45 s-1. From this estimate one can approximate the concentration of active ternary complexes in the gel exclusion fractions and calculate the fractional recovery of active enzyme in ternary complexes as compared with the amount initially employed in the reaction (Vmax in this case is calculated from the value of u observed in a standard elongation assay employing the equations determined below). These recoveries range from 5 to 7% of the active enzyme molecules initially added to the reaction in which ternary complex is formed. This low value may reflect an inactivation of enzyme in ternary complexes during isolation of the complex.
Alternatively, an appreciable fraction of enzyme molecules which initiate a poly[r- (A-U)] chain may terminate the chain and fail to re-initiate during formation of the ternary complex.
The drug rifampicin efficiently blocks the initiation of new RNA chains (25) and has been routinely included in all chain elongation assays to preclude any contributions due to RNA chain initiation.
There was no alteration in the rate of poly- ternary complex, except that Tris-HCl buffer at the desired pH was used. The concentration of UTP in the reaction was 466 PM, that of ATP was 6 bad.
[r(A-U)] synthesis nor in the kinetic parameters obtained for poly[(A-II)] synthesis (I',,,, KA, or Ku) when rifampicin was omitted from the reaction mixture or when the rifampicin concentration was increased to 500 kg per ml. These results indicate, first, that under standard elongation assay conditions there is essentially no poly[r(A-U)] chain termination and re-initiation by enzyme in active ternary complexes during the initial phase of chain growth.
Second, it appears that there is no pool of active enzyme (in binary complexes, for example) which can initiate under these conditions. It also appears that rifampicin does not alter the kinetic properties of the enzyme in the ternary complex. This is expected, since Eilen and Krakow (26) have shown that rifampicin does not detectably bind to RNA polymerase in the ternary complex.
The effect of pH on the chain elongation assay is shown in Fig. 2. The rate of chain elongation shows a broad pH optimum, with a maximum at 8.0. This is similar to the relationship obtained for the complete reaction (27-30).
At pH 7.4 the rate of chain elongation is 65% of the optimal rate. Kinetic studies employing the procedures described below indicate that this decrease is due to an increase in the apparent Ks value for ATP, whereas the V,,, for chain elongation is unchanged.
A dependency of Ks on pH (actually v/Ks) might be due to either an ionization of substrate or of a residue in the protein which alters or prevents substrate binding (31). The former possibility is excluded by the fact that the magnesium-nucleoside triphosphate substrate complexes have no pK in this range.
The dependence of the rate of a chain elongation on divalent cations (Fig. 3) shows that for manganese ion there is a sharp maximum of activity at 2 InM.
The rate of elongation in the presence of magnesium ion showed a much broader optimum between 10 and 20 mM. The response of the rate of chain elongation to the divalent cation concentration is similar to that observed for the over-all reaction (27)(28)(29).
The maximal velocity is the same for both of these ions at their optimal concentrations, although the K, for Mg*+ATP is higher than for M&ATP.
A divalent metal ion is required for the elongation reaction.
A trace of activity observed when no ion was added to the reaction mixture could be accounted for by the residual Mg2+ contained in the buffer used to elute the complex from the column. This activity was abolished by the addition of EDTA to the reaction mixtures.
The rate of poly[r(A-U)] chain elongation is markedly enhanced at elevated concentrations of monovalent cations such as NaCl, KCl, or NH&l (Fig. 4). The rate of elongation is maxi- [Mn++] (mM) [ b++] (mM) ternary complex, except that the indicated divalent metal was added at the desired concentration.
The concentration of UTP in the reaction was 400 PM and that of ATP was 6 JLM. Each assay contained 10 nmoles of MgClz which came from the Bio-Gel P-100 buffer used in isolation of the ternary complex. ternary complex, except that NH&l at the indicated concentrations was substituted for KC1 and the reaction times were shortened to 1 min. The concentration of UTP in the reaction was 8 JLM ma1 at 0.2 to 0.3 M NH&l.
The rate of chain elongation is diminished at 1 M NH&l; however, this reduction may not reflect a true decrease in the rate of elongation.
At 1 M NH&l there is rapid termination of poly[r(A-U)] chains growing in the ternary complex,3 and the rate measured in a 60-s assay may not be an accurate reflection of the true initial rate. Synthesis of RNA in the over-all reaction is completely eliminated by NH&l concentrations of more than 0.3 M in the reaction mixture, probably due to the inhibition of DNA template binding by the salt (32, 33).

Steady State Kinetic Model for Chain Elongation by Ternary Complexes
A simple model for RNA chain elongation with a poly[d(A-T)] template postulates two states of the ternary complex.
One state, which we will call Cr, contains an RNA chain with a 3'terminal uridine residue and is a potential acceptor for an AMP residue.
The complementary complex (CA) contains a terminal adenosine residue.
The simplest sequence of steps for chain elongation then involves: Here Steps 1 and 3 involve a reversible binding of substrate, whereas Steps 2 and 4 involve formation of the phosphodiester bond and translocation of the enzyme. A steady state rate equation can be derived for the above model (34) and is where v is the initial velocity of the reaction at concentrations of MgATP and MgUTP of A and U, respectively; V is the maximal velocity obtained at saturating substrate concentration; and KA and Ku are the Michaelis constants for each substrate.
This equation falls into the general class of kinetic mechanisms which Cleland has described as ping-pong reactions (34).
It is clear that the steps in Equations 1 to 4 are probably complex; for example, conversion of CT (MgATP) t,o CA is likely to involve, first, formation of the phosphodiester bond, then perhaps release of pyrophosphate, and finally isomerization or translocation to align the active site with the next template base. However, for a ping-pong reaction mechanism the presence of these substeps in the reaction sequence does not alter the over-all kinetic equation (34).
To t.est the degree to which poIy[r(A-U)] chain growth fits the above kinetic model, measurements were made of the rate of chain elongation at a variety of nucleotide concentrations.
Initially, rate measurements were made by varying the concentration of one of the two substrates while holding the concentration of the other constant.
The data are represented graphically in Lineweaver-Burk reciprocal plots (Fig. 5) and have a form which is characteristic of a ping-pong kinetic mechanism.
Similar results have been obtained with all of the templates tested.
To determine KS values for MgATP and MgUTP in the elongation reaction, the intercepts on the ordinate of each of the lines in Fig. 5 were graphed in a secondary plot (Fig. 6) as a function of the reciprocal of the concentration of the nucleotide which previously had been held constant.4 As expected from Equation 5 (34), this treatment gives two linear relationships from which the constants are obtained.
It is also predicted from Equation 5 that KS values for ATP and UTP can be obtained directly, by varying one substrate while the other is present in great excess. However, this procedure cannot be used without additional information because high concentrations of a competing nucleoside triphosphate competitively inhibit chain elongation (see below).
By applying the steady state kinetic treatment derived above, we have determined several KS values for the alternating co- template several analogues of UTP have also been tested. Although the number of different templates and substrates for which kinetic parameters have been determined is not large, several conclusions appear from the data, which is tabulated in Table I.
First, within experimental error, the values of KA and Ku do not vary for the different templates (Note that the value of KA measured with the alternating copolymer templates has to be multiplied by 2 in order to be compared with the value obtained from poly[d (A).d(T)J. This is discussed further under "Dis-  cussion"). This result may have significance in terms of the mechanisms of substrate recognition by the enzyme and is discussed later.
As other workers have shown, the apparent KS values for chain elongation are substantially lower than those measured for the complete DNA-dependent reaction (9, 35), and this generalization holds for the true KS values which we report here. The very high apparent KS values obtained in the complete reaction may be related to the requirement for high ATP or GTP concentrations in the chain initiation phase of the reaction (5,9).
The values of KS differ significantly for the different nucleoside triphosphates.
It is interesting in this regard that UTP, the unique nucleotide in RNA, has the highest KS value. These differences do not have any obvious control function; the con-  centrations of nucleoside triphosphates within the bacterial cell are 10 to 1000 times higher than the Ks values (36,37).
The KS values for several nucleotide analogues were also measured in order to determine how structural pertubations might affect the reactivity.
All of these Ks values are much higher than those found with the natural substrates, and the maximal velocity of the reaction is also affected in some instances. The alteration of the Vm,, makes quantitative interpretation of the analogue results difficult (see below).

Inhibition of Chair Elongation by Nucleoside Triphosphates and Derivatives
Inhibition by Substrate Nucleoside Triphosphates-The derivation of Rate Equation 5 assumed that only the nucleoside triphosphate which is to be incorporated into an RNA chain would bind to the complex (Reactions 1 and 3). Since misincorporation by RNA polymerase is very low (38), one might expect that either (a) a complex CT is unable to bind UTP at all or (b) CT binds UTP but the complex is unable to carry out phosphodiester bond formation.
In the latter case, if we raise the UTP concentration high enough one might expect to see inhibition of poly[r(A-U)] chain elongation by UTP even though it is a substrate. As shown in Fig. 7, such inhibition is observed.5 It is a simple matter to extend the model for elongation to account for inhibition of this sort. For example, if one is concerned with elongation by a dAT complex, then one can consider that the complex called CT, defined as the complex which must insert ATP as the next nucleotide, can instead bind MgUTP to form a complex CT-MgUTP which can react no further: CT + MgUTP e CT(MgUTP) Following the method of Cleland (40), one can show that the addition of Reaction 6 to Reactions 1 to 4 adds a single term to the rate equation yielding V/V = 1 + KU/U + (KA/A)U + U/K,U,) where Kz is an inhibition constant for the inhibition of MgATP incorporation by MgUTP. The constant KL in Equation 7 is 6 The inhibition is not due to binding of the divalent cations by the excess nucleoside triphosphates (39), since the Mg2+ concentration is always higher than the highest triphosphate concentration. Furthermore, doubling the Mga+ concentration does not change the Ki values. The numerical value for KU, which is about 15 PM, is much less than K$, which is of the order of magnitude of 1 mM. This means that at MgUTP concentrations with which one must be concerned with substrate inhibition the ratio Ku: U is very much less than 1 and can be neglected.
In this case Equation 7 can be simplified to V/V = 1 + (KAIA) (1 + UlK?A) (8) Thus at elevated MgUTP concentrations, MgUTP can be neglected as a substrate, and one can proceed to determine K$ as though MgUTP were simply a competitive inhibitor. Experimentally, if one chooses a UTP concentration and determines v while varying the ATP concentration, one obtains a linear double reciprocal plot as before (Fig. 7). At a second UTP concentration, a new linear relationship is obtained in which the slope is changed but not the l/v intercept.
This behavior is characteristic of competitive inhibition in which the substrate and inhibitor compete for the same form of the enzyme (40,41). By an appropriate replot of the data shown in Fig. 7 one can determine numerical values for the Kis. A summary of the results is given in Table II. Inhibition by Triphosphates That Are Not Substrates-A substance which is not a substrate for the elongation reaction can also react with the complex in a competitive manner.
In this case it can potentially interact with both forms of the complex: CAfI$cA-I (9) CT-f-I*CCT-I (10) and introduces two new factors into the rate equation where KfA and K& are not necessarily equal. One can measure these constants by selecting the proper experimental conditions. For example, to measure KfA, one can work at substrate concentrations in which the UTP concentration is much greater than  Ki values for various templates are shown in Table II. We find that within experimental error KL = Kfu and Kz = Kit,.

This indicates that MgCTP and MgGTP
can interact equally well with both CA1 and CT. There are considerable differences in both the size of the bases and the positions of the reactive groups between the substrates MgATP and MgUTP, used by these complexes, and the inhibitors MgCTP and MgGTP.
The results suggest that this binding site is not appreciably altered as the enzyme interacts with different bases in the DNA template. This interpretation is reinforced by the observation that the inhibitory constant for binding of tripolyphosphate is essentially the same as those for the nucleoside triphosphates themselves. Thus the nucleoside triphosphate binding site employed in chain elongation seems to be primarily directed toward the tripolyphosphate portion of the substrate. DISCUSSION Our experiments show that ternary complexes containing RNA polymerase, template DNA, and a growing RNA chain can be isolated free from substrates under conditions such that a portion of the enzyme molecules is still able to elongate RNA chains actively.
The initial rate of this elongation reaction is constant for a reasonable length of time, and this result, together with the insensitivity of the reaction to the presence of the drug rifampicin, assures that the repeated initiation of RNA chains or chain termination events do not occur to a significant extent in this system under standard conditions. Consequently these ternary complexes provide an attractive approach to the study of the kinetic and biochemical properties of the chain elongation phase of enzymatic transcription.
The rate of RNA chain elongation carried out by these ternary complexes follows simple steady state rate equations for the several synthetic DNA templates that we have tested. For DNA templates of alternating nucleotide sequence, a so-called "ping-pang" kinetic model is obtained in which the enzyme oscillates alternately between two states, each of which is specific for a different substrate ( By employing the approach we have discussed above, this model generates the rate equation  The equivalent homopolymer Ke values were calculated as described under "Discussion." If no information was available as to the base frequencies of the RNA synthesized, this was approximated by using base frequencies of the DNA template (47). Abbreviations used in the template column are: CT, calf thymus; ML, Micrococcus lysodeiklicus; ST, Salmonella typhimwium.

The parameters
Ke and fs are defined in the text. .5 .5 .285 .14 .5 .5 .5 .22 -ion of the complex and pyrophosphate release must not depend on the substrate incorporated.
One can extend the kinetic treatment to include any template by assuming that the KS values do not depend on the DNA sequence. In this case the equivalent homopolymer value, Kssh, becomes an intrinsic Ke value for the substrate, and one can write a general rate equation (12) where the sum is over all nucleotides contained in the RNA product.
Both Bremer (11) and Hyman and Davidson (9) devised a kinetic treatment of RNA chain elongation based on these assumptions.
The equation they derive is identical with Equation 12.
If Ke does depend on a nucleotide sequence, the kinetics of chain elongation will still fit an equation of the form V/v = 1 + cs Kc/S, but the values of Ks will depend on the template DNA used in the measurements.
Our extremely limited data indicate that K&J, does not vary appreciably for the templates poly[d (A-T)   The values of KS do vary appreciably when base analogues such as FUTP, BUTP, and deoxyuridine triphosphate replace UTP; variation of apparent KS values in the complete RNA polymerase reaction had been reported previously for these analogues (20). The interpretation of these variations is made difficult by the lack of information in our experiments as to the true value of V, that is the true turnover number of the RNA polymerase molecules engaged in chain elongation.
However, one can tentatively say that the variations do not follow a simple pattern of' dependence of Ke on base-stacking or basepairing interactions.
Instead, all of the analogues have substantially higher Ks values than does UTP and this suggests that interactions between the protein and the substrate rather than interactions between substrate and the DNA template are most critical in determining KS. Again, this fits our tentative conclusion that KS does not show a substantial dependence on nucleotide sequence for these DNA templates.
The elongation of RNA chains by ternary complexes is inhibited at elevated concentrations of nucleoside triphosphates. This effect has been reported previously with the complete RNA polymerase reaction as an ability of noncomplementary nucleoside triphosphates to inhibit synthesis of polynucleotides when templates of restricted base composition were employed (43). This "low efficiency inhibition" was attributed to a low general affinity of RNA polymerase in the ternary complex for nucleoside triphosphates (43) and is consistent with a model of RNA polymerase which postulates a single common site on the enzyme at which nucleoside triphosphates are bound for RNA chain elongation (38,43).
Our results confirm the earlier reports of low efficiency inhibition and show that this inhibition reflects a general ability of nucleoside triphosphates and related compounds to inhibit RNA chain elongation competitively. Alternative possibilities to explain this inhibition, such as complexing of metal ions in the reaction at elevated nucleoside triphosphate concentrat.ions (39), can be ruled out. Since the values of Ki obt,ained for different nucleoside triphosphates reflect the affinity of the enzyme in the ternary complex for that nucleoside triphosphate, a comparison of Ki for the different forms of the ternary complex allows a rough probe of t.he structure of the complex. One can ask, does the ternary complex C, have the same affinity for GTP as does CT? The results indicate that, within the limits of accuracy, all forms of ternary complex have equal affinity for noncomplementary nucleoside triphosphates.
This result has two important negative consequences; it makes less likely models of RNA polymerase in the ternary complex which postulate (a) four separate sites, one for each nucleoside triphosphate, or (b) a single site which is altered appreciably in its conformation by the DNA template base that it is to read. Instead, the data fit well with a model proposed some time ago (38,43) in which selection of the correct nucleoside triphosphate proceeds through a random, weak binding of all possible substrates at a common site on the enzyme followed by a conformational change to form a much stronger complex when the correct triphosphat,e enters that site. Studies of t.he specificity of incorporation of base analogues into RNA by RNA polymerase make it very likely that transition into this second "active" complex is determined primarily by the ability of the base of the incoming triphosphate to fit into a site on the enzyme that also contains the template base to be read (44).
It would be of interest to probe the conformation of this complex with complementary nucleoside triphosphate analogues which are unable to form phosphodiester bonds (45,46); the isolated ternary complexes that we have described and the availability of a steady state kinetic model for elongation by these complexes provide an attractive system for such a study.
A large number of studies of the kinetics of RNA synthesis by RNA polymerase in vitro and in viva have been carried out since the initial reports of the enzyme. A comparison of the equivalent homopolymer Ks values that we have determined with those calculated from the dat.a of other workers is difficult or impossible in most instances because of the complexities introduced by competing reactions such as RNA chain initiation and RNA chain termination and by the competitive inhibition of RNA chain elongation at elevated triphosphate concentrations.
In addition, values obtained by others for Ke for a given nucleoside triphosphate normally include contributions from Ks values for the other nucleoside triphosphates: KS ~~~~~~~~~~~ = 1 + c (K2/CI) where I/ is the sum over all nonvaried substrates. Unless the concentrations of the other substrates were fixed at sufficiently high concentrations to make K,/C, negligible, these Ks values cannot be compared with our values in any simple way.
Within these limits, the values of Ks obtained by other workers usually fall within the range of values that we report here (Table  III) for simple synthetic templates.
Slapikoff and Berg (20) determined Ku values for the over-all and these are found to agree with the values that we report. The "eouivalent homopolymer values" for phage and bacterial DNAs may be somewhat higher than the values obtained for synthetic templates. However, a definite conclusion regarding this possibility must await careful measurements in which initiation is bypassed and the unvaried substrate is judiciously chosen, as discussed above. Several groups have shown that the identity of a base analogue can profoundly influence the nearest neighbor frequencies with which it is incorporated into an RNA chain when synthesis occurs in the presence of both analogue and the naturally occurring nucleoside triphosphate.
For example, when a mixture of UTP and%TP are used for RNA synthesis, UTP preferentially follows A or U, whereas 9TP preferentially follows GTP. If these differences reflect differences in incorporation in the elongation phase of the reaction, then, since there appears to be no alteration of KS with different nearest neighbors, how can these nearest neighbor preferences be expressed? The simplest answer would be that the maximal rate of incorporation, V, can vary appreciably with the nucleotide sequence. Again, it would be extremely valuable to have a method for measuring this parameter in the current system. The notion that the rate of RNA chain elongation is sensitive to the nucleotide sequence of the region being transcribed has been suggested by others (1, 44,46), and a number of models can be imagined in which this phenomenon could play an important role in the process of RNA synthesis on DNA templates and its regulation.
Acknowledgments-We would like to thank Dr. W. W. Cleland for his many helpful comments concerning the manuscript. We are also grateful to Dr. W. Mange1 for many discussions and suggestions during the course of these experiments.