Ribonucleoside and deoxyribonucleoside triphosphate pools during 2-aminopurine mutagenesis in T4 mutator-, wild type-, and antimutator-infected Escherichia coli.

Ribonucleoside and deoxyribonucleoside triphosphate pools have been measured in Escherichia coli infected with bacteriophage T4 DNA polymerase mutator, wild type, and antimutator alleles during mutagenesis by the base analogue 2-aminopurine. ATP and GTP pools expand significantly during mutagenesis, while CTP and UTP pools contract slightly. The DNA polymerase (gene 43) alleles and an rII lesion perturb normal dNTP pools more than does the presence of 2-aminopurine. We find no evidence that 2-aminopurine induces mutations indirectly by causing an imbalance in normal dNTP pools. Rather, it seems likely that, by forming base mispairs with thymine and with cytosine, 2-aminopurine is involved directly in causing bidirectional A.T in equilibrium G.C transitions. The ratios for 2-aminopurine deoxyribonucleoside triphosphate/dATP pools are 5-8% for tsL56 mutator and 1-5% for tsL141 antimutator and 43+ alleles. We conclude that the significant differences observed in the frequencies of induced transition mutations in the three alleles can be attributed primarily to the properties of the DNA polymerases with their associated 3'-exonuclease activities in controlling the frequency of 2-aminopurine.cystosine base mispairs.

Ribonucleoside and deoxyribonucleoside triphosphate pools have been measured in Escherichia coli infected with bacteriophage T4 DNA polymerase mutator, wild type, and antimutator alleles during mutagenesis by the base analogue 2-aminopurine. ATP and GTP pools expand significantly during mutagenesis, while CTP and UTP pools contract slightly. The DNA polymerase (gene 43) alleles and an rII lesion perturb normal dNTP pools more than does the presence of 2aminopurine. We find no evidence that 2-aminopurine induces mutations indirectly by causing an imbalance in normal dNTP pools. Rather, it seems likely that, by forming base mispairs with thymine and with cytosine, 2-aminopurine is involved directly in causing bidirectional A-T $ G*C transitions. The ratios for 2-aminopurine deoxyribonucleoside triphosphate/dATP pools are 5-8% for tsL56 mutator and 1-5% for tsL141 antimutator and 43+ alleles. We conclude that the significant differences observed in the frequencies of induced transition mutations in the three alleles can be attributed primarily to the properties of the DNA polymerases with their associated 3'-exonuclease activities in controlling the frequency of Z-aminopurine-cystosine base mispairs.
In this paper, we report measurements of deoxyribonucle- oside triphosphate pools during AmPur mutagenesis in T4 tsL56-rUV199 mutator-, 43+-rUV199-, and tsL141-rUV199 antimutator T4-infected E. coli. These pool size determinations serve three primary purposes with regard to mutagenesis. First, knowledge of relative magnitudes of dAmPurTP/ dATP allows one to assess the potential for dAmPurTP to be inserted into DNA opposite Thy; this step is required for direct involvement of AmPur in causing A. T + G. C transitions. Similarly, a measurement of the dAmPurTP/dGTP ratio relates to AmPur's direct mutagenic potential in the G . C + A -T pathway where dAmPurTP is inserted opposite template Cyt in competition with dGTP. Second, a measurement of dAmPurTP pools for each of the three T4 gene 43 alleles will allow us to determine if the widely different AmPur-induced mutagenic rates observed in mutator, wild type, and antimutator backgrounds (see e.g. Ref. 28) might be attributed to differences in the metabolism of the analogue in the three genetic backgrounds. Alternatively, differences in AmPur mutagenesis in the T4 gene 43 alleles may be caused primarily by differences in the insertion and proofreading properties of the mutator, wild type, and antimutator DNA polymerases. Finally, the pool size measurements should allow us to determine whether AmPur exerts an indirect effect on mutagenesis by perturbing pools of the four common dNTPs.

DISCUSSION
A measurement of the pool size of 2-aminopurine deoxyribonucleoside triphosphate allows one to estimate the mutagenic potential of the base analogue, provided that the mutations occur as a direct result of AmPur incorporation into DNA. The first step in the induction of an A . T + G . C transition by 2-aminopurine presumably requires the replacement of Ade by AmPur opposite a template Thy site. In general, the rates of insertion into DNA for any two competing nucleotide substrates should be in proportion to their relative dNTP pools. A multienzyme deoxyribonucleotide biosynthetic complex encoded by T4 (see e.g. Ref. 31) may act to concentrate dNTPs at the replication fork. Although replication fidelity can be influenced by absolute dNTP concentrations (13,21), it is the ratio of competing dNTP substrates which is expected to play a dominant role in mutagenesis (21).
For the special case of AmPur and Ade deoxyribonucleotides competing for insertion opposite Thy, where I(AmPur)/I(Ade) is the ratio of inserting dAmPurTP and dATP into DNA, [dAmPurTP]/[dATP] is the ratio of the pool concentrations for the two competing substrates, and AG, is a measure of the average free energy difference between A . T and AmPur.T base pairs in DNA. Estimates of AG during DNA replication are in the range 1-1.3 kcal/mol (21,32). Pool concentrations of dAmPurTP appear to be similar, within a factor of 2, in T4 tsL56 mutator, 43+, and tsL141 antimutator backgrounds (Table I). dATP concentrations also appear to be similar in 43+ and tsL141 antimutator backgrounds, but may be about 5-fold lower in tsL56 mutatorinfected cells (Table I). Although we strongly suspect that greater levels of AmPur-induced mutagenesis in L56 mutator backgrounds (see e.g. Ref. 28) are attributable primarily to a reduction in the 3'-exonuclease proofreading activity of L56 DNA polymerase in comparison to the proofreading capabilities for the active 43+ and highly active L141 polymerases (3,21,26), the increased [dAmPurTP]/[dATP] ratio might be responsible for perhaps as much as a 5-fold greater mutation rate in tsL56-infected cells.
The second step in the A T + G . C pathway involves the insertion of HMdCTP in place of dTTP opposite template AmPur. From Table I we note that the pool of HMdCTP is about 1.5-fold larger in tsL141 antimutator-infectedcells compared to 43+ and about 5-fold larger in tsL141 compared to tsL56 mutator. However, these differences would not be expected to have a significant effect on AmPur mutagenesis since the frequency of HMdCTP insertion opposite AmPur should be much more dependent on the HMdCTP/dTTP ratio than on the absolute magnitude of HMdCTP concentration (21). The free energy difference, AG2, for inserting HMdCTP uersus dTTP opposite AmPur is estimated to be about 1.8 kcal/mol (15).
The mutation frequency at each individual site on DNA depends on nearest-neighbor base-stacking partners and surrounding base composition (8)(9)(10)(11)(12)(13)(14)(15)(16)(17). However, one can utilize the dNTP pool size ratios and measurements of base-pairing free energy differences to estimate the average mutation frequency, neglecting site-specific effects. This type of estimate is instructive in the case of mutagenesis by 2-aminopurine as it allows an evaluation to be made of the relative importance of proofreading in controlling the relative mutation frequencies in tsL56 mutator, 43+, and tsL141 antimutator genetic backgrounds.
We will now calculate two extreme cases to characterize  , 32), we obtain an AmPur-induced A . T + G.C mu-tation frequency of for a system without proofreading. A frequency of is similar to the AmPur-induced reversion of rUV199 in the tsL56 mutator background (2 X  Table 111). Perhaps tsL56 mutator represents an "ideal" onestep discrimination enzyme. I n vitro studies on purified L56 polymerase strongly indicate that the mutant enzyme has a diminished 3'-exonuclease activity (3) and exhibits a reduced proofreading capability for AmPur (21,26).
We now consider a second extreme case where proofreading is maximized. We will assume that the same AG discrimination value used in the insertion step is valid for the proofreading step; there is experimental evidence which supports the idea that nucleotide discrimination-free energy differences, governing the frequency of forming AmPur.T versus A . T base pairs, are similar during insertion and excision for purified L56,43+, and L141 DNA polymerases (21). Using the same parameters as before, except that the base pair stabilities AG1 and AG, are each sampled twice (first during insertion and then during excision), we calculate a mutation rate of 6.5 X This value is of the same order as the tsL141 antimutator-induced reversion of rUV199 (1 X Table 111).
Perhaps the L141 antimutator polymerase is behaving similarly to an ideal two-step discrimination enzyme. Reversion of the rUV199 marker to r+ in 43+-infected cells (4 X Table 111) falls between the two extremes of ideal one-and two-step discrimination.
We conclude from the pool size measurements that the metabolism of AmPur appears similar in L56 mutator-, 43+-, and L141 antimutator-infected cells, resulting in a pool of substrate dAmPurTP which is between 1 and 5% of the concentration of dATP pools. However, as we have shown previously (28), AmPur substitution for Ade in T4 L56, 43+, and L141 DNA is at a level which is much less than 1-5%; the actual in vivo dAmPurMP/dAMP incorporation ratios are 1:llOO for L56, 1:1800 for 43+, and 1:8500 for L141. The reduction in dAmPurMP incorporation by more than an order of magnitude below the pool ratio is primarily due to the instability of AmPur .T as compared to A. T base pairs (33, 34). Regarding nucleotide insertion, we would expect this reduction to be roughly similar for the three polymerase alleles. At the excision or proofreading step, a further reduction in the level of dAmPurMP in DNA is expected in all three backgrounds, but a much larger reduction is expected in the case of L141 antimutator.
We find no evidence that the pools of the four common dNTPs undergo significant distortion during AmPur mutagenesis. Thus, rather than causing mutations in an indirect manner by creating an imbalance in the concentration of one nucleotide versus another, it seems likely that AmPur is involved directly in the formation of aberrant base pairs AmPur . T and AmPur . C in causing bidirectional A. T F? G . C transition mutations. lDreOYer, t h e g e n e 4 3 alleles a n d t h e r11 l e s i o n appear t o be more i m p o r t a n t t h a n AmPur m u t a g e n e x s i n p r r u r b i n y

nFcnal DNA P~B E U C B O C pools Icoapare
( R e f . 301 t o T a b l e 1 1 ) .
The dMPurTP/ dATP r a t i o f o r UV199-L56 is s l i g h t l y h i g h e r 1 5 to 8 % ) t h a n t h a t o f UV199-43' 11.7 LO 5 8 ) and UVIY9-LI1I 11.3 to 4.68) [ Table   I ) .
M u t a t i o n rate d a t a a p p e a r i n T a b l e 111.