Vol. 174, No. 13

Multiple Mutant of Escherichia coli Synthesizing VirtuallyThymineless DNA during Limited GrowthHIYAM H. EL-HAJJ,t LINGHUA WANG, AND BERNARD WEISS*

Department ofPathology, University of Michigan Medical School, Ann Arbor, Michigan 48109-0602

Received 13 December 1991/Accepted 21 April 1992

The dut gene of Escherichia coli encodes deoxyuridine triphosphatase, an enzyme that prevents theincorporation of dUTP into DNA and that is needed in the de novo biosynthesis of thymidylate. We produceda conditionally lethal dut(Ts) mutation and isolated a phenotypic revertant that had a mutation in an unknowngene tentatively designated dus (for dut suppressor). The dus mutation restored the ability of the dut mutant togrow at 42°C without restoring its enzymatic activity or thymidylate independence. A strain was constructedbearing, in addition to these mutations, ones affecting the following genes and their corresponding products:ung, which produces uracil-DNA N-glycosylase, a repair enzyme that removes uracil from DNA; deoA, whichproduces thymidine (deoxyuridine) phosphorylase, which would degrade exogenous deoxyuridine; and thyA,which produces thymidylate synthase. When grown at 42°C in minimal medium containing deoxyuridine, themultiple mutant displayed a 93 to 96% substitution of uracil for thymine in new DNA. Growth stopped afterthe cellular DNA had increased 1.6- to 1.9-fold and the cell mass had increased 1.7- to 2.7-fold, suggesting a

general failure of macromolecular biosynthesis. DNA hybridization confirmed that the uracil-containing DNAwas chromosomal and that new rounds of initiation must have occurred during its synthesis.

The DNA of almost all organisms contains thymine ratherthan uracil. A plausible reason was provided by Lindahl (22).Uracil can arise in DNA from the mutagenic spontaneoushydrolysis of DNA cytosine, and it is recognized and re-

moved by a ubiquitous repair enzyme, uracil-DNA N-glyco-sylase. Therefore, thymine, rather than uracil, was estab-lished as a normal constituent of DNA. However, phagePBS2 of Bacillus subtilis possesses uracil-containing DNA(11), indicating that at least under some circumstances,thymineless DNA can function normally. To what extenthave other organisms become dependent on thymine-con-taining DNA, and what other properties, if any, are uniqueto such DNA? To answer these questions, we have isolatedmutants of Escherichia coli that incorporate uracil into DNAin place of thymine.

In E. coli, most of the thymidylate needed for DNAsynthesis is manufactured via dUTP with the help of deoxy-uridine triphosphatase (dUTPase) (27, 33). dUTPase cata-lyzes the hydrolysis of dUTP to PPi and dUMP, a substratefor thymidylate synthase. Therefore, mutations in dut, thegene for dUTPase, cause an increased level of dUTP and a

corresponding decrease in dTTP so that large amounts ofuracil are incorporated into DNA in place of thymine. Thisincorporation is only transient (35) because uracil is removedfrom DNA via an excision repair initiated by uracil-DNAN-glycosylase, the product of the ung gene (11). In dut ungdouble mutants, the misincorporated uracil is not effectivelyexcised; in one study, there was a stable replacement of upto 19% of DNA thymine by uracil (37). We wished to obtainhigher levels of replacement, but we were hampered by theleakiness of the existing dut mutations. Therefore, in a

previous study, we generated a tight dut mutation by insert-ing a chloramphenicol resistance gene within a plasmid dutgene (14). However, the insertion was lethal; it could not

* Corresponding author.t Present address: Department of Medicine, Beth Israel Medical

Center, New York, NY 10003.

replace a chromosomal dut+ gene unless there were anotherfunctional copy of dut within the cell. We postulated at firstthat the lethality might be the result of excessive incorpora-tion of uracil into DNA; the uracil-containing chromosomemight function poorly or be irreparably broken during at-tempted excision repair (10, 11). However, we could notrestore viability by supplying large amounts of exogenous

thymidine or by producing combinations of mutations thatwould reduce the formation of dUTP or the removal of uracilfrom DNA. It is possible, therefore, that the lethality of thedut mutation might be unrelated to uracil incorporation intoDNA. To study the mechanism of killing, we have in thisstudy isolated a conditionally lethal dut mutant as well as anew mutation in a gene called dus, which restores theviability of dut mutants without restoring their dUTPaseactivity. We also report some preliminary experiments inwhich a strain of E. coli that contains these mutations incombination with others was able to synthesize virtuallythymineless DNA for almost one generation.

MATERIALS AND METHODS

Microbial strains and plasmids. The bacterial strains used(Table 1) were derivatives of E. coli K-12. The recombinantplasmids pIT15 (34) and pLW2 are derivatives of pBR322(31) that were arbitrarily chosen as hybridization probes forchromosomal DNA. pIT15 bears the soxRS region of E. colion a 2.5-kb insert. Plasmid pLW2 bears the dcd gene (36) ona 1.3-kb HindIII-EcoRV fragment of E. coli DNA subclonedfrom X2E1 (miniset no. 355) from the library of Kohara et al.(20). Phage XBW112 (dut+) is a c+Q+S+ derivative ofXBW111 (32) that was constructed by crossing XBW111 withA wild type. Plasmid pWB30 contains the dut gene on an

881-nucleotide fragment cloned into the EcoRV site ofpBR322. The cloned segment extends from an EcoRV site292 nucleotides from the 3' end of dfp to the XmnI site atnucleotide 44 of the ttk gene (14, 23). It was obtained fromplasmid pWB2, which is similar to pWB1 (32) except that the

4450

JOURNAL OF BACTERIOLOGY, July 1992, p. 4450-44560021-9193/92/134450-07$02.00/0Copyright X) 1992, American Society for Microbiology

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

E. COLI MUTANT SYNTHESIZING THYMINELESS DNA 4451

TABLE 1. Bacterial strains

Description

HfrC (PO-2A), metBi pyrE41 uhp-1 rel-I tonA22 (X)KL16 ung-151::TnlOKL16 dut-1KL16 ung-lKL16 pyrE zia-207::TnlOKL16 recA56 srlC300::TnlOAT2243-11c mutL::TnlOKL16 ttk-l::kanKL16 dut-22(Ts) ttk-l::kanKL16 dut-22(Ts)BW712 dut-22(Ts) ttk-l ::kanKL16 dut-22(Ts) ttk-l::kan ung-JKL16 dut-22(Ts) ttk-1::kan ung-J recAS6 srlC300::TnlOKL16 dus-1 dut-22(Ts) ttk-l::kan ung-1 srlC300::TnJOKL16 deoA22 thr-34::TnlOKL16 deoA22KL16 dut-22(Ts) ttk-l::kan deoA22BW935 ung-1S1::TnJOKL16 dus-I dut-22(Ts) ttk-l::kan ung-JBW938 deoA22 thr-34::TnlOKL16 dus-1 dut-22(Ts) ttk-l::kan ung-J deoA22Same as that for BW940 but ung-151::TnIOKL16 dut-22(Ts) ttk-l::kan ung-151::TnlO deoA22 thyAKL16 dut-21::cat ung-J dus-I srlC300::TnlOmutL::TnlO argE metB gyrA rpoB A(lac-pro) supE? X-F-KL16 dut-21::cat (XpyrE+ dut+ c1857)Hfr Reeves 4 (PO100) deoA22 thr-34::TnlO upp-J udp-1 metBl argF58 reL4lA-Hfr P045 thi-I relAI spoTI

Source or referencea

15B. K. Duncanb9This labc32This labcP1(D6432) x AT2243-11C14P1(BW928) x KL16P1(BW928) x BW322This studyP1(BW741) x BW310P1(BW386) x BW929This studyP1(HH17) x BW35P1(KL16) x BW933P1(BW741) x BW934P1(BD2008) x BW935BW931 -- Srl+d

P1(BW933) x BW938BW939 Thr+d

P1(BD2008) x BW940BW942 -* trimethoprim resistance (26)P1(HH1) x BW9311614142

a Phage P1 transductions are described as follows: Pl(donor) x recipient.' Same as that for BD2007 (12), but cured of X.c Pedigrees available on request.d Selected for spontaneous precise excision of Tn1O.

cloned DNA is in the opposite orientation. Plasmid pES19(dfp+ dut-19::TnlOOO) was previously described (30).

Microbiological methods. Microbiological methods, selec-tive media, and testing of ung and deoA genotypes weredescribed previously (14, 38). For the growth of dut mutants,even rich media were routinely supplemented with thymi-dine at 0.5 mM (14). Minimal media were also supplementedwith thiamine at 1 ,ugIml. Low-phosphate medium was amodification of the 32p labeling medium of Bochner andAmes (4), containing 0.5% Norit-treated vitamin-freeCasamino Acids in place of potassium phosphate.

Production of the dut-22(Ts) mutation. P1 phages weregrown on strain BW719 (ttk-1::kan). The phage lysate wastreated with 0.4 M hydroxylamine (19) for 18 h at 37°C andthen used to transduce strain BW712 to kanamycin resis-tance at 30°C in the presence of thymidine.DNA isolation. E. coli DNA was extracted from cells and

purified as described previously (39). Plasmids were ampli-fied with chloramphenicol and purified by a rapid alkalineextraction method (28). Residual RNA was removed from allDNA preparations by digestion with RNase A, and the DNAwas precipitated with ethanol (28).DNA composition. Purified cellular DNA was digested to

mononucleotides by pancreatic DNase and venom phospho-diesterase (7). The mononucleotides were separated bytwo-dimensional thin-layer chromatography on polyethyl-eneimine cellulose (4). The solvent for the first dimensionwas 1 M acetic acid adjusted to pH 3.5 with NH40H; that forthe second dimension was isobutyric acid-concentratedNH40H-0.5 M potassium phosphate buffer (pH 3.5)-H20(66:1:2:31). Autoradiograms were developed to confirm

completeness of digestion and separation of mononucle-otides from other radiochemical material. Spots produced bycarrier deoxynucleotides (0.2 ,umol each) were visualized at254 nm and scraped into vials for liquid scintillation count-ing.

Filter hybridization of uracil-containing DNA (see Table 4).Strain BW943 was grown with aeration at 37°C in a low-phosphate labeling medium (4) containing 0.25 mM[6-3H]thymidine (25 ,uCi/ml). When the cells reached adensity of 2.5 x 108 ml-', they were washed twice bycentrifugation in unsupplemented medium and resuspendedat 1.25 x 108 ml-l in medium containing 32p; (14 ,Ci/ml) and0.5 mM deoxyuridine for further growth at 42°C. After 3 h,the DNA was isolated from 30 ml of culture, denatured byheating at 100°C for 5 min, and dissolved in 5 ml ofhybridization fluid. Molecular hybridization was performed(28) with 2 ,ug each of the indicated target DNAs bound tonitrocellulose filters.Other methods. dUTPase assays (30) were performed on

sonicates of growing cells (9). Protein was determined by thebicinchoninic acid method (29). Relative cell mass and cellconcentration were determined by turbidity (6). The concen-tration of cellular DNA was measured with diphenylamine(5), and that of purified plasmid DNA was estimated bystaining with ethidium bromide (28).

RESULTS

Isolation of a dut(Ts) mutant. To study the lethal effects ofdut mutations and to produce other mutations that wouldrestore viability, we first had to isolate a dut mutant that we

Strain

AT2243-llcBD2008BW285BW310BW322BW386BW712BW719BW741BW743BW928BW929BW930BW931BW933BW934BW935BW936BW938BW939BW940BW942BW943BW945D6432HH1HH17KL16

VOL. 174, 1992

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

4452 EL-HAJJ ET AL.

could propagate, i.e., a conditionally inviable strain. Thetechnique of Hong and Ames (19) was used to generatemutations within 2 min of the ttk-1::kan allele. ttk is thesecond member of the two-gene dut operon. It encodes a

23-kDa polypeptide of unknown function, and its mutantshave no discernible phenotype (14). The ttk-l::kan insertionmutation (14) consists of a kanamycin resistance determi-nant of Tn9O3, which was ligated into a site at nucleotide 258of the 633-bp open reading frame of ttk (14). A phage P1lysate of a ttk-l: :kan mutant was treated with hydroxylamineand used to transduce another strain to kanamycin resis-tance at 30°C as described in Materials and Methods. Of26,000 transductants that were screened by replica plating,110 had a temperature-sensitive kanamycin resistance.Twenty others were conditional lethal (Ts) mutants, and ofthese, 14 were complemented by XBW112 and therefore hadmutations in a 7-kb region spanning the dut gene. Seven ofthese 14 mutants had <50% of wild-type dUTPase levels at42°C. BW928, the dut mutant with the lowest survival at42°C (<10-3), was chosen for further complementationanalysis with plasmids containing DNA from the dfp-dut-ttkregion. dfp is a vital gene of unknown function, which istranscribed independently of dut and which is its immediateupstream neighbor (30). The ability of strain BW928 to growat 42°C was not restored by plasmid pES19 (dfp+dut-19::TnlOOO) but was restored by pWB30 (dut+). Inplasmid pWB30, dut is the only intact cloned gene; theplasmid contains no more than about 25% of the dfp geneand the initial 44 nucleotides of ttk. The results indicated thatthe conditional lethality was due, at least in part, to a dutmutation.

Conditional lethality does not depend on a ttk mutation.Because ttk is downstream of dut in the two-member dutoperon, a previously studied dut-21::cat insertion mutation(14) should have also affected the expression of ttk. Our new

conditional lethal mutant, BW928 (dut-22[Ts] ttk-1::kan),also had a ttk mutation. However, dut-22 specified temper-ature sensitivity whether or not it was accompanied byttk-l::kan. For example, when BW928 (dut-22 ttk-l::kan)was used to transduce BW322 to pyrE+, some temperature-sensitive, kanamycin-sensitive transductants such as BW743(Table 1) were produced. There is still a remote possibilitythat the truncated Ttk protein specified by the ttk-l::kangene might still be functional and that the temperature-sensitive phenotype of dut-22 mutants is therefore due totwo new mutations, one in dut and the other in the proximalportion of ttk. However, our data indicated that hydroxyl-amine treatment produced mutations for individual genes ata frequency of c0.4%. Therefore, it is extremely unlikelythat the temperature sensitivity of strain BW928 (dut-22[Ts])was due to a double mutation.

Isolation of a dut suppressor (dus) mutant. Previously, we

were unable to reverse the lethality of dut mutations bymutations in known genes affecting the formation or the fateof uracil-containing DNA (14). The temperature-sensitivemutant now enabled us to select directly for new mutationsthat might suppress those in dut. We reasoned that at 420C,death of the mutant might be caused by double-strand breaksresulting from the attempted excision repair of uracil-con-taining DNA. To reduce uracil incorporation, we addedthymidine to all growth media, and to reduce repair, wetransduced the dut-22(Ts) allele into an ung-1 (uracil-DNAglycosylase) mutant. In addition, the strain was made recom-bination deficient by the introduction of a recA mutation. Itspurpose was to reduce pseudoreversion that might resultfrom the spontaneous tandem duplication (1) of a leaky

4

0

0)

6 2 3

Time (h)FIG. 1. Growth curves of mutant strains at 42°C. Cells were

grown with aeration in a thymidine-supplemented tryptone-yeastmedium at 30°C and shifted to 42°C at zero time. The optical densityat 560 nm was measured with a Klett colorimeter (6). The cells werediluted fourfold into fresh prewarmed medium whenever the cellconcentration approached 4 x 108 to 5 x 108 ml-1 (120 to 150 Klettunits), and the optical density at 560 nm was corrected for thedilution. The congenic strains used were BW934 (dut+ dus+ ung+)(@), BW935 [dut(Ts)] (A), BW936 [dut(Ts) ung] (A), and BW940[dut(Ts) ung dus] (0). All strains were deoA22 mutants. dut-22(Ts)mutants also contained ttk-l::kan.

dut(Ts) allele. The resulting strain was BW930 (Table 1). At42°C, temperature-resistant derivatives appeared spontane-ously at a frequency of 10-. One such mutant, BW931, wasarbitrarily chosen for further study. It retained a reduceddUTPase activity and still possessed a dut(Ts) allele thatcould be transferred to other strains via cotransduction withttk-l::kan. This result suggested that the reversion to tem-perature resistance was caused by a mutation outside of dut,which we designate by the tentative gene symbol dus, amnemonic for dut suppressor. Preliminary conjugationalmapping indicated that the dus-1 allele is near the his operon;its location and identity have been subsequently determined(see Discussion).

Properties of the dus mutant. We chose for further studystrain BW931, a temperature-resistant derivative that borethe dus-I and dut-22(Ts) mutations. This strain was found tohave spontaneously reverted to recA +; it was no longer UVsensitive, and it had regained recombinational proficiency sothat it could serve as a transductional recipient (Table 1). Itsreversion to recA+ may have been the result of the mutatorphenotype associated with its dut (18) and ung (11) muta-tions. The dus mutation fully restored the growth rate of adut(Ts) mutant at 42°C (Fig. 1). However, at 42°C, the dusdut(Ts) mutant required thymidine, which was expectedbecause dUTPase is essential for thymidylate biosynthesis(33).

dus-I was also found to suppress a dut insertion mutation.The dut-21: :cat allele was produced by cloning a fragment of

J. BACTERIOL.

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

E. COLI MUTANT SYNTHESIZING THYMINELESS DNA 4453

TABLE 2. dUTPase activities of strains bearing dut anddus mutations

Grwh dUTPaseGrwh(relative sp act)a TempStrain Relevant genotype temp (relative coefficient

(OC) 300C 42°C (Q430b)

KL16 Wild type 30 (1.0) 1.6 1.6BW285 dut-1 30 0.013 0.0088 -cBW741 dut-22(Ts) 30 0.33 0.57 1.7BW931 dut-22(Ts) dus-1 30 0.40 0.65 1.6BW945 dut-21::cat dus-1 30 0.0052 0.0054KL16 Wild type 42 1.8 2.8 1.6BW931 dut-22(Ts) dus-l 42 - 0.20

a Enzymatic activity is reported as a ratio of specific activity of a mutantstrain to that of a wild-type (dut+ dus+) strain grown and assayed at 300C. Arelative specific activity of 1.0 corresponds to 5.54 U of dUTPase per mg ofprotein.

b Q30 is the ratio of dUTPase activity at 42°C to that at 30°C.c, not calculated or not done.

Tn9 that specifies chloramphenicol resistance into the mid-dle of the dut gene. Previously, a dut-21::cat would only betolerated in merodiploids that had a functional second copyof dut (14). Now, we could transduce it into a dus-1 mutantthat was haploid for dut, producing the strain BW945 (Table1). The transduced chloramphenicol resistance was geneti-cally stable, indicating that we had indeed replaced dut-22(Ts) and had not merely inserted dut-21::cat into a tan-demly duplicated dut region. The suppression of thetransduced dut-21: :cat allele confirmed that dus-1 is anextragenic mutation and suggested that it is not a transla-tional suppressor.dUTPase levels. Although the dus-1 mutation suppressed

the lethality of dut-22(Ts) and dut-21::cat mutations, it didnot affect dUTPase levels (Table 2). The properties of themutant dUTPase specified by the dut-22(Ts) allele werecontrary to our expectations (Table 2). First, its residualactivity at 42°C was higher than that of a dut-i mutant that isnot temperature sensitive for growth (18). Therefore, eitherthe relative dUTPase activities of the mutant enzymes invitro do not reflect their activities in vivo or conditionallethality is related to some function of the mutant proteinother than its dUTPase activity. Second, the dut-22 enzymedid not have an altered temperature coefficient. Third, in aseparate experiment, we found that in crude extracts, themutant and wild type had similar Km values at 42°C (6 and 7puM, respectively). One finding, however, indicated theprobable basis for the thermosensitive phenotype. When thedut(Ts) mutant was grown at 30°C, it had 36% of thedUTPase activity of the wild type, but when grown at 42°C,it had only 7% (Table 2). Therefore, the mutant enzyme (orits production) appears to be heat labile in vivo.

Incorporation of uracil into DNA. The availability of viablestrains containing tight dut mutations enabled us to try toreplace all of the thymine in DNA with uracil. We con-structed BW943, a strain with the following relevant geno-type: dut-22(Ts) dus-i ung-Si ::TniO thyA deoA. The strainwas viable at 42°C. The rationale for its construction was asfollows. The dut mutation should lead to the accumulation ofdUTP and to its incorporation into DNA at a high tempera-ture. dus-i, which suppresses the lethality of dut, wouldfavor continued growth of the cells under nonpermissiveconditions. The ung mutation would block the removal ofuracil from the DNA, and the ung-iSi ::TniO insertion mightbe tighter than ung-1. The thyA (thymidylate synthase)

A dCMPQ

dUMPA OdTMP dU

rUMP J

..

X- dGMP

Pj3)fV(Di

C

dTMPiD l

j: 6

FIG. 2. Two-dimensional thin-layer chromatography of 32P-la-beled DNA nucleotides isolated from a dut(Ts) ung dus deoA thyAmutant. Strain BW943 was grown to a density of 2.5 x 108 ml-' at37°C in a thymidine-supplemented tryptone-yeast medium, washedby centrifugation, and transferred to low-phosphate medium at 42°Ccontaining 32Pi at 10 ,Ci ml-' and either 0.5 mM deoxyuridine or 0.5mM thymidine. The cells were harvested at 2 h, by which timegrowth had ceased. DNA was extracted and digested to mononu-cleotides that were separated by chromatography (see Materials andMethods). (A) The positions of markers visualized by UV orautoradiography are shown. Spot X, visible only on autoradiograms(panels B and C), probably represents undigested oligonucleotides.rUMP (dashed outline) was not visible on the autoradiograms(panels B and C). (B) Cells were grown with deoxyuridine. Positionsof UV-visible dTMP and dUMP markers are indicated. (C) Cellswere grown with thymidine.

mutation should prevent the synthesis of dTMP from thedUMP that might arise either through residual dUTPaseactivity or from another pathway (e.g., the hydrolysis ofdUDP). The deoA (thymidine [deoxyuridine] phosphorylase)mutation should enable the more efficient utilization ofexogenous thymidine or deoxyuridine (14).

In the experiment whose results are shown in Fig. 2, themultiple mutant was grown for 2 h at 42°C in a mediumcontaining 32p; and either thymidine or deoxyuridine. Thecells that were fed deoxyuridine appeared to have almostcompletely replaced dTMP by dUMP in newly synthesizedDNA (Fig. 2B); radioactivity measurements indicated a 91%replacement. In contrast, the DNA synthesized in the pres-ence of thymidine (Fig. 2C) contained dUMP amounting tono more than 3% of the dTMP. No radioactivity was seen inthe rUMP (ribouridylate) spots, indicating the absence ofsignificant contamination of the DNA samples by RNA.The experiment whose results are shown in Fig. 2 was

repeated with monitoring of growth and of DNA synthesis.When we attempted to measure rates of DNA synthesis bythe uptake of radioactive thymidine, the results immediatelyafter the medium shift were erratic, suggesting a fluctuationof nucleotide pools. Therefore, we chemically measured theDNA content of the cells. In the experiments whose resultsare shown in Table 3, a growing culture of the multiplemutant was shifted from a complex thymidine-supplementedmedium to a thymidine-free medium containing a high con-centration of deoxyuridine. The culture was then incubatedat 42°C for 2 h, by which time it reached an apparent limit ofgrowth that was less than 25% of that displayed by wild-typecells in a separate experiment. During this time, the cellmass of the culture increased about 1.7- to 2.7-fold and itsDNA content almost doubled. In the new DNA, uracilresidues replaced 93 to 96% of the thymine.

In the experiment whose results are shown in Table 3,

VOL. 174, 1992

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

4454 EL-HAJJ ET AL.

TABLE 3. Synthesis and composition of DNA in a dut(Ts) ungdus deoA thyA mutant grown at 42°C in the presence

of deoxyuridinea

Changes over 2 h

Expt Increase in Increase Uracil content ofcell mass in DNA new DNA [U/(U+T)I

(fold) (fold) (%)

1 1.7 1.8 932 2.7 1.9 96

a Strain BW943 was grown at 37°C to 1 x 108 ml- (experiment 1) or 2.5 x

108 ml-' (experiment 2) in a thymidine-supplemented tryptone-yeast medium,washed, and transferred at time zero to a 32p labeling medium containingdeoxyuridine for growth at 42'C. The experimental conditions were the same

as those described in the legend to Fig. 2.

growth stopped 1.5 h after the shift to deoxyuridine-contain-ing medium, and there was no further increase even after an

additional 18 h of incubation. In a control experiment,cultures were grown at 42°C in minimal medium supple-mented with thymidine rather than deoxyuridine. The cellsgrew continuously to saturation.We do not know at this point what effect a dus+ allele

would have had on the outcome of the experiments whoseresults are shown in Table 3. However, the presence of thedus-1 mutation enabled us to conclude that the cessation ofcell growth in the presence of deoxyuridine at 42°C could notbe attributed to the conditional lethality of dut-22 per se.

Moreover, the dus-1 mutation permitted us to repeat thisexperiment with an analog of the multiple mutant containinga tighter dut mutation, dut-21::cat. The results were similar.Cell growth stopped by 2 h, by which time cell mass hadincreased 2.5-fold. The ratio of uracil to uracil plus thyminein the total cellular DNA was measured by high-performanceliquid chromatography (21) and equaled 46%, a result that iscompatible with the synthesis of almost a full strand ofuracil-containing DNA.The uracil-containing DNA is chromosomal. The experi-

ments described above entailed thymine deprivation, a con-

dition that can induce cryptic prophages (8, 25). Therefore,we used DNA-DNA hybridization to identify the newlysynthesized DNA. The conditions were similar to thosedescribed in Table 3, footnote a, except that the cells were

labeled with [3H]thymidine during preliminary growth inthymidine-enriched medium and with 32Pj during subsequentgrowth with deoxyuridine. During growth in deoxyuridine,the DNA content of the culture increased 1.6-fold. The DNAwas isolated and tested for its ability to hybridize with wholegenomic DNA and with cloned segments of chromosomalDNA. The 3H label served as a convenient measurement ofthe relative concentration of cellular DNA and of the overallhybridization efficiency, whereas the 32P radioactivity was a

specific tracer for the newly synthesized uracil-containingDNA. If the new 32P-labeled DNA were that of an inducedprophage, it should not hybridize to two plasmids carryingwidely separated chromosomal segments. However, thenewly synthesized 32P-DNA and genomic 3H-DNA annealedin similar ratios to whole genomic DNA and to the recom-binant plasmids (Table 4). A vector DNA (pBR322) controldemonstrated the specificity of the hybridization. The resultsindicate that the newly synthesized uracil-containing DNA ischromosomal and is not that of an induced prophage.

TABLE 4. Identification of uracil-containing DNA byfilter hybridizationa

Unlabeled Uracil-DNA Genomic DNA Ratiotarget bound bound R32poDNAb (32p cpm) (3H cpm) ( H)

E. coli 3,578 2,403 1.5pLW2 955 620 1.5pIT15 431 265 1.6pBR322 5 11

a Strain BW943 was grown first at 37°C in a medium containing [3H]thymi-dine and then at 42°C in a medium containing 32Pi and deoxyuridine. The DNAwas extracted and hybridized with the indicated target DNAs (see Materialsand Methods). Values for a blank filter (46 cpm of 32p; 20 cpm of 3H) weresubtracted from the results.

b Plasmids pLW2 and pIT15 are derivatives of pBR322 containing segmentsof E. coli DNA from regions at 45 and 92 min, respectively, of the chromo-somal linkage map (3). (See Materials and Methods.)

DISCUSSION

The vital nature of the dut gene was previously demon-strated via the lethality of a dut insertion mutation (14), butthe mechanism of cell death is unknown. The lethality wasnot reversed by mutations in known genes affecting dUTPformation or the fate of uracil-containing DNA. Therefore,we were led to obtain a conditionally lethal dut mutant and touse it to select for extragenic mutations that would suppressthe lethality. Our dut(Ts) mutant was inviable at 42°C, had alow level of dUTPase at that temperature, and was comple-mented by a dut+ plasmid. A mutation in an unknown gene,dus, suppressed the lethality without restoring dUTPaseactivity. We reasoned that by ultimately identifying the dusgene product and by studying nucleotide pools in dut(Ts) anddus mutants, we might learn why the dut gene is vital. Wealso hoped that by suppressing the lethality of dut withoutrestoring dUTPase, we might maintain cells with high levelsof uracil in their DNA. This study contains our first suchattempt. We were able to synthesize new chromosomalDNA in which over 90% of the thymine was replaced byuracil before there was a shutdown of DNA synthesis andcell growth.Our results suggested that dus-1 is not a translational

suppressor and that the suppressor mutations occur withsuch high frequency (10-5) that they might be null muta-tions. These assumptions have been confirmed recently bythe isolation of additional suppressor mutations by transpo-son insertion and the demonstration that they belong to thesame complementation group that dus-1 does. In work to bepublished separately, Wang and Weiss (36) present evidencethat dus-1 is actually a dcd allele; it is a mutation in thestructural gene for dCTP deaminase, the enzyme that pro-duces about 75% of the dUTP in E. coli (27). This findingsuggested that it is the accumulation of dUTP that causes thelethality associated with dut mutations.

In our dut mutants, dUTPase activity did not seem to becorrelated with viability. At 42°C, the dUTPase of a viabledut-I mutant had 0.6% of the activity of the wild-typeenzyme, whereas the temperature-sensitive dut-22 mutanthad at least 7% residual activity. Superficially, the resultssuggest that lethality is not associated with dUTPase activitybut with some other undiscovered activity of the enzyme.However, because dCTP deaminase mutations suppress dutlethality, it is likely that viability is directly related todUTPase levels. Therefore, our measurements of dUTPaseactivity in crude extracts may not reflect those in the livingcell. Given the important role of the enzyme, it is not

J. BACTERIOL.

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

E. COLI MUTANT SYNTHESIZING THYMINELESS DNA 4455

surprising that a mutant with only 7% residual activity isinviable. What is harder to explain is how a dut-I mutant,which appears to have less than 1% residual activity, isviable. We should like to suggest that our measurements ofdUTPase activity in our dut-I mutant may have been falselylow because of inactivation of the mutant enzyme during thepreparation of cell extracts. A similar effect has been notedwith mutants for other vital enzymes, e.g., valine- andphenylalanine-tRNA ligases; extracts prepared from temper-ature-sensitive conditional lethal mutants had 0.3% residualactivities even at a permissive temperature (13).During the extensive incorporation of uracil into DNA, the

DNA content of the cells nearly doubled (Table 3). There-fore, the cellular DNA should be expected to consist almostentirely of hybrid duplexes; at any point on the chromo-some, one strand should contain uracil and the sister strandshould contain thymine. From the extent of the DNAsynthesis in the deoxyuridine medium, we may concludethat new rounds of DNA initiation must have occurreddespite the unavailability of dTMP. If reinitiation had notoccurred, then we should have seen only a 39% increase inDNA content (24), representing the average amount of DNAthat would be synthesized between the replication forks andthe termini in a logarithmically growing culture. This con-clusion is confirmed by the results shown in Table 4. Thenew DNA hybridized with plasmid pIT15, which bears achromosomal segment only 9 min from the replication origin.Statistically, only a small fraction of the replication forksshould exist proximal to this region. New rounds of initiationmust have occurred in the deoxyuridine medium for thissegment to have been as efficiently labeled with 32P as therest of the chromosome was. Therefore, the replicationapparatus of E. coli supports both the initiation and elonga-tion of uracil-containing daughter strands.

In these experiments, DNA synthesis stopped short ofdoubling. Assuming that both strands were replicated, i.e.,that leading and lagging strand synthesis occurred, replica-tion forks stopped before the point at which uracil-contain-ing DNA would have been made on a uracil-containingtemplate. However, we cannot conclude that such an eventis prohibited. The cessation of cell growth indicates abroader defect in macromolecular biosynthesis, or in energymetabolism, from which an arrest of DNA synthesis mightoccur. If DNA arrest were the sole effect of uracil incorpo-ration, then we should have expected the cell mass toincrease far beyond the point at which DNA synthesisstopped (17). We hoped that the numerous published studitson thymineless death would provide some insight into themechanism of this growth inhibition. Unfortunately, thosestudies concentrated on the synthesis of DNA rather thanthat of other macromolecules, and most of this research waslater found to have been complicated by the induction of aprophage in the most commonly used strain (8). However, inat least one experiment with a nonlysogen (25), a decline ofprotein synthesis was noted 2 h after thymine starvation, aresult which is at least consistent with our finding of arrestedgrowth, although its cause may be different. Therefore, inour experiments, growth arrest might be more directly linkedto thymine starvation rather than to uracil incorporation intoDNA, and there may be a mechanism by which macromo-lecular biosynthesis is regulated by nucleotide pools.Our mutants should enable us now to explore possible

links between dUTP metabolism or thymine starvation andthe synthesis of DNA, RNA, protein, and cell walls. Ofparticular interest is that a possible defect in protein synthe-sis may be the result of the inability of RNA polymerase,

topoisomerases, or transcriptional regulators to recognizeuracil-containing DNA. However, if there are critical thy-mine residues in vital DNA-protein recognition sites, thereprobably cannot be many such sites because E. coli main-tains vigorous growth in the face of at least 10% replacementof thymine by uracil (11). There may even be a selectiveevolutionary pressure against the requirement for thymine atimportant sites because transient incorporation of dUTP,which occurs in wild-type cells, might interfere with thefunction of those sites. Consequently, it might be possible toselect for additional bacterial mutations or to find phages andplasmids that will allow the extensive synthesis of thymine-less DNA in our mutants, thereby further enabling us toexplore evolutionary mechanisms and the interdependenceof pathways of macromolecular biosynthesis.

ACKNOWLEDGMENTS

We gratefully acknowledge the capable technical assistance ofLaura Bliss and Fred Kung.

This work was supported by research grants MV-205R andNP770-S from the American Cancer Society.

REFERENCES1. Anderson, R. P., and J. R. Roth. 1977. Tandem genetic dupli-

cations in phage and bacteria. Annu. Rev. Microbiol. 31:473-505.

2. Bachmann, B. J. 1987. Derivations and genotypes of somemutant derivatives of Escherichia coli K-12, p. 1190-1219. InF. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M.Schaechter, and H. E. Umbarger (ed.), Escherichia coli andSalmonella typhimurium: cellular and molecular biology. Amer-ican Society for Microbiology, Washington, D.C.

3. Bachmann, B. J. 1990. Linkage map of Escherichia coli K-12,edition 8. Microbiol. Rev. 54:130-197.

4. Bochner, B. R., and B. N. Ames. 1982. Complete analysis ofcellular nucleotides by two-dimensional thin-layer chromatog-raphy. J. Biol. Chem. 257:9759-9769.

5. Burton, K. 1956. A study of the conditions and mechanism ofthe diphenylamine reaction for the colorimetric estimation ofdeoxyribonucleic acid. Biochemistry 62:315-323.

6. Chan, E., and B. Weiss. 1987. Endonuclease IV of Escherichiacoli is induced by paraquat. Proc. Natl. Acad. Sci. USA84:3189-3193.

7. Clements, J. E., S. G. Rogers, and B. Weiss. 1978. A DNase forapurinic/apyrimidinic sites associated with exonuclease III ofHemophilus influenzae. J. Biol. Chem. 253:2990-2999.

8. Cohen, S. S. 1971. On the nature of thymineless death. Ann.N.Y. Acad. Sci. 186:292-301.

9. Cunningham, R. P., S. M. Saporito, S. G. Spitzer, and B. Weiss.1986. Endonuclease IV (nfo) mutant of Escherichia coli. J.Bacteriol. 168:1120-1127.

10. Demple, B., and S. Linn. 1982. On the recognition and cleavagemechanism of Escherichia coli endodeoxyribonuclease V, apossible DNA repair enzyme. J. Biol. Chem. 257:2848-2855.

11. Duncan, B. K. 1981. DNA glycosylases, p. 565-586. In P. D.Boyer (ed.), The enzymes, vol. 14. Academic Press, Inc., NewYork.

12. Duncan, B. K. 1985. Isolation of insertion, deletion, and non-sense mutations of the uracil-DNA glycosylase (ung) gene ofEscherichia coli K-12. J. Bacteriol. 164:689-695.

13. Eidlic, L., and F. C. Neidhardt. 1965. Protein and nucleic acidsynthesis in two mutants of Escherichia coli with temperature-sensitive aminoacyl ribonucleic acid synthetases. J. Bacteriol.89:706-711.

14. El-Hajj, H. H., H. Zhang, and B. Weiss. 1988. Lethality of a dut(deoxyuridine triphosphatase) mutation in Escherichia coli. J.Bacteriol. 170:1069-1075.

15. Ferenci, T., H. L. Kornberg, and J. Smith. 1971. Isolation andproperties of a regulatory mutant in the hexose phosphatetransport system of Escherichia coli. FEBS Lett. 13:133-136.

VOL. 174, 1992

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

4456 EL-HAJJ ET AL.

16. Grilley, M., K. M. Welsh, S. S. Su, and P. Modrich. 1989.Isolation and characterization of the Escherichia coli mutL geneproduct. J. Biol. Chem. 264:1000-1004.

17. Hirota, Y., A. Ryter, and F. Jacob. 1968. Thermosensitivemutants of E. coli affected in the processes of DNA synthesisand cellular division. Cold Spring Harbor Symp. Quant. Biol.33:677-693.

18. Hochhauser, S. J., and B. Weiss. 1978. Escherichia coli mutantsdeficient in deoxyuridine triphosphatase. J. Bacteriol. 134:157-166.

19. Hong, J.-S., and B. N. Ames. 1971. Localized mutagenesis ofany specific small region of the bacterial chromosome. Proc.Natl. Acad. Sci. USA 68:3158-3162.

20. Kohara, Y., K. Akiyama, and K. Isono. 1987. The physical mapof the whole E. coli chromosome: application of a new strategyfor rapid analysis and sorting of a large genomic library. Cell50:495-508.

21. Kumagai, M., M. Fujimoto, and A. Kuninaka. 1988. Determi-nation of base composition of DNA by high performance liquidchromatography of its nuclease P1 hydrolysate. Nucleic AcidsRes. Symp. Ser. 19:65-68.

22. Lindahl, T. 1982. DNA repair enzymes. Annu. Rev. Biochem.51:61-87.

23. Lundberg, L. G., H.-O. Thoresson, 0. H. Karlstrom, and P. 0.Nyman. 1983. Nucleotide sequence of the structural gene fordUTPase of Escherichia coli K-12. EMBO J. 2:967-971.

24. Maal0e, O., and P. C. Hanawalt. 1961. Thymine deficiency andthe normal DNA replication cycle. I. J. Mol. Biol. 3:144-155.

25. Medoff, G. 1972. Nucleic acid and protein synthesis duringthymineless death in lysogenic and nonlysogenic thymine auxo-trophs. J. Bacteriol. 109:462-464.

26. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

27. O'Donovan, G. A. 1978. Thymidine metabolism in bacteria, p.219-253. In I. Molineux and M. Kohiyama (ed.), DNA synthe-sis: present and future. Plenum Publishing Corp., New York.

28. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular

cloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory Press, Cold Spring Harbor, New York.

29. Smith, P. K., R. I. Krohn, G. T. Hermanson, A. K. Mallia, F. H.Gartner, M. D. Provenzano, E. K. Fujimoto, N. M. Goeke, B. J.Olson, and D. C. Klenk. 1985. Measurement of protein usingbicinchoninic acid. Anal. Biochem. 150:76-85.

30. Spitzer, E. D., and B. Weiss. 1985. dfp gene of Escherichia coliK-12, a locus affecting DNA synthesis, codes for a flavoprotein.J. Bacteriol. 164:994-1003.

31. Sutcliffe, J. G. 1979. Complete nucleotide sequence of theEscherichia coli plasmid pBR322. Cold Spring Harbor Symp.Quant. Biol. 43:77-90.

32. Taylor, A. F., P. G. Siliciano, and B. Weiss. 1980. Cloning of thedut (deoxyuridine triphosphatase) gene of Eschenchia coli.Gene 9:321-336.

33. Taylor, A. F., and B. Weiss. 1982. Role of exonuclease III in thebase-excision repair of uracil-containing DNA. J. Bacteriol.151:351-357.

34. Tsaneva, I. R., and B. Weiss. 1990. socxR, a locus governing asuperoxide response regulon in Escherichia coli K-12. J. Bac-teriol. 172:4197-4205.

35. Tye, B.-K., P. 0. Nyman, I. R. Lehman, S. Hochhauser, and B.Weiss. 1977. Transient accumulation of Okazaki fragments as aresult of uracil incorporation into nascent DNA. Proc. Natl.Acad. Sci. USA 74:154-157.

36. Wang, L., and B. Weiss. Submitted for publication.37. Warner, H. R., B. K. Duncan, C. Garrett, and J. Neuhard. 1981.

Synthesis and metabolism of uracil-containing deoxyribonucleicacid in Escherichia coli. J. Bacteriol. 145:687-695.

38. White, B. J., S. J. Hochhauser, N. M. Cintr6n, and B. Weiss.1976. Genetic mapping of xthA, the structural gene for exonu-clease III in Eschenchia coli K-12. J. Bacteriol. 126:1082-1088.

39. Wilson, K. 1990. Preparation of genomic DNA from bacteria, p.2.4.1-2.4.5. In F. M. Ausubel, R. Brent, R. E. Kingston, D. D.Moore, J. G. Seidman, J. A. Smith, and K. Struhl (ed.), Currentprotocols in molecular biology, supplement 9. John Wiley &Sons, Inc., New York.

J. BACTERIOL.

on April 30, 2018 by guest

http://jb.asm.org/

Dow

nloaded from