Proc. Nat. Acad. Sci. USAYol. 71, No. 5, pp. 1612-1617, May 1974

Chemical Carcinogens as Frameshift Mutagens: Salmonella DNA SequenceSensitive to Mutagenesis by Polycyclic Carcinogens

(histidinol dehydrogenase/electrophoresis/repeating doublet DNA sequence/amino-acid sequences)

KATSUMI ISONO* AND JOSEPH YOURNOtBiology Department, Brookhaven National Laboratory, Upton, New York 11973

Communicated by Bruce N. Ame8, January 21, 1974

ABSTRACT Other investigators have shown thatseveral polycyclic carcinogens are frameshift mutagens inSalmonella. Mutagenic potency of these compounds isassessed by ability to induce reversion of histidine-requir-ing frameshift mutants to prototrophy. One frameshiftmutation in the histidinol dehydrogenase gene, hisD3052,is unusually sensitive to mutagenesis by certain polycycliccarcinogens. We find that the 3052 mutation is a -1 dele-tion, probably loss of a G C pair from a DNA repeat of-G-G-G-

-C-C-C-. The polycyclic carcinogens tested (e.g., 2-nitroso--G-C-

fluorene) revert 3052 by deleting a -C-G- doublet from the-C-G-C-G-C-G-C-G-

DNA sequence -G-C-G-C-G-C-G-C-, which is close to the3052 site. This rare mispairing-prone sequence represents.the carcinogen-sensitive "hotspot" in 3052. The ICR com-pounds, noncarcinogenic intercalating agents, show abroader specificity of mutagenesis. Reversion with thesemutagens occurs predominantly by two mechanisms: oneidentical to that of the polycyclic carcinogens, and theother by +1 additions in a third DNA tract narrowlyseparated from the 3052 site. The alkylating carcinogenN-methyl-N'-nitro-N-nitrosoguanidine appears to have asimilar dual specificity. DNA base changes in 3052 rever-tants have been correlated with properties of histidinoldehydrogenase in crude extracts. This correlation of DNAbase sequence with electrophoretic and other properties ofthe mutant proteins allows one to analyze easily thespecificity of new mutagens that mutate this strain.

The concept of carcinogenesis as an expression of mutation inDNA has been revived in recent years (1-6). Mutations areof several types: base substitutions, frameshifts, and largerlesions. To type mutations induced by chemical carcinogens,Ames and his coworkers have been developing tester systemswith the bacterium Salmonella (2, 3). The basis of their ap-proach is scoring carcinogen-induced reversion to histidineindependence of histidine-requiring tester strains. Each testerstrain carries a known type of mutation. Reversion requires a

Abbreviations: NF, 2-nitrosofluorene; HC, hycanthone; NQ,nitroquinoline-N-oxide; NG, N-methyl-N'-nitro-N-nitroso-guanidine.* On leave from the Laboratory of Genetics, University of Tokyo,Japan. Present address: Max-Planck-Institut fur MolekulareGenetik, Abt. Wittmann, 1 Berlin 33 (Dahlem), Ihnestrasse63-73, West Germany.t Present address: Dept. of Pathology, University of RochesterMedical Center, Rochester, N.Y. 14620. To whom reprint re-quests should be addressed.

1612

compensating mutation, almost invariably of the same generaltype as the original; i.e., base substitutions are corrected bybase substitutions, frameshifts by frameshifts. In particular,the Ames group has shown that many polycycic compoundswith reactive side groups, interpreted as proximal carcinogens,are frameshift mutagens in Salmonella (3-6).:One tester strain,TA1538, containing the frameshift mutatioi hisDS052, is un-usually sensitive to reversion by many -of these carcinogens;reversion is enhanced many-fold over-the spontaneous fre-quency (3-5). Our interest in examining DNA sequences in3052 and its revertants was stimulated by Ames. We now re-port on the unusual base sequence of a carcinogen-sensitiveDNA tract in 3052.

Frameshift mutations result from additions (+) or dele-tions (-) of DNA base pairs from genes encoding polypeptidechains, so that triplets in the product mRNA are not read inthe correct frame after the mutation (7, 8). Correction of suchmutations may occur by restoration of the original base se-quence, or by introduction of a closely positioned frameshiftof opposite sign. In these double frameshift mutants (+ -)the normal reading frame is restored in mRNA except for thesegment between the two frameshifts. The (+ -) mRNAsegment encodes a segment of altered amino acids in thepolypeptide chain, often disturbing its function. Sequencingthe affected polypeptide segment allows reconstruction ofassociated mRNA and DNA sequences (9, 10).The hisD gene encodes the enzyme histidinol dehydro-

genase (EC 1.1.1.23). To determine nucleic acid base changesin 3052 caused by polycyclic carcinogens and other frameshiftmutagens, we have induced reversion in 3052 by these com-pounds. Revertants carrying double frameshift mutationswere detected by rapid screening of crude extracts from rever-tants for altered histidinol dehydrogenase. The altered en-zyme from each double mutant was then purified and com-pared with the normal enzyme by peptide mapping to detectaltered peptides. These were sequenced to reconstruct al-tered DNA sequences in 3052 and its revertants. We havealso classified the reversions with respect to certain propertiesof histidinol dehydrogenase from crude extracts, and havecorrelated the classes of reversion with specific DNA sequencechanges.We believe that this newly characterized tester system will

be useful for rapid screening of carcinogens as frameshiftmutagens. For the relatively small effort of inducing reversionsin 3052 with carcinogens or other mutagens and examiningcrude extract histidinol dehydrogenase, it should be possibleto assign the reversions to the standard groups with knownbase sequence changes in DNA (see below).

Dow

nloa

ded

by g

uest

on

June

17,

202

1

Mutagenesis by Polycyclic Carcinogens 1613

MATERIALS AND METHODS

A strain of 30S2 carrying the operator constitutive mutationhis01242, TR1693, was kindly constructed for us by J. R.Roth. The 1242 mutation allows production of 12-fold thenormal amounts of histidine enzymes (11). Revertants ofthis 305S strain were induced by spotting a drop of mutagensolution on enriched minimal agar spread plates by the methodof Ames (2). The following carcinogens were used: 2-nitroso-fluorene (NF) synthesized by Bartsch and Miller (3) andfurnished by Ames; hycanthone (HC) (12, 13), furnished byP. E. Hartman; nitroquinoline-N-oxide (NQ), furnished byH. B. Wood; and N-methyl-N'-nitro-N-nitrosoguanidine(NG) from Aldrich Chemicals. The acridine derivative frame-shift mutagens tested, ICR-191 and ICR-364-OH (14), werethe gift of H. J. Creech. All mutagens were used as solutionsat a concentration of 1 mg/ml, except for NG, which wasspotted as crystals.

Histidinol dehydrogenase specific activity and electro-phoretic mobility in polyacrylamide gels were determined bythe method of Yourno, Barr, and Tanemura (15). For se-quence analysis revertant enzymes were purified by the stan-dard method (16), compared with the wild type by peptidemapping, and affected peptides sequenced according toYourno and Heath (17) and Yourno (18).

RESULTS AND DISCUSSION

Frameshift mutation 30S2 was induced with the acridinederivative ICR-364-OH by Oeschger and Hartman, whomapped it in the middle region of the hiD gene. As is charac-teristic of frameshifts, 30S2 is induced to revert by certainframeshift mutagens, such as ICR, but not by pure base-substituting agents such as 2-aminopurine (19). The spon-taneous yield of revertants from 3052 was from 5 to 15 perplate (2 X 108 bacteria) after 72-hr incubation. Reversionwas weakly enhanced by NG and HC; in each case about 10revertant colonies arose per plate around the zone of mutagenkilling. Response to NQ, ICR-191, and ICR-364-OH was quitestrong, averaging about 75 colonies per mutagen spot. NF

T28

TABLE 1. Classification of revertanrs from strain 30Sf

Repre-Number of senta- HDH Relativerevertants tive activity electro-

rever- (% of phoreticClass Total Of each origin tant wild type) mobility

1 1 1/12 (spont.) R12 6 ± 0 1.06 ± 02 2 2/12 (spont.) R2 145 ± 11 1.00 :1 03 17 7/11 (ICR-191) R22

8/11 (ICR-364- R24 9 ± 4 1.00 i00OH)

2/5 (NG) R394 2 (spont.) 18 ± 5 0.90 ± 0.015 31 6/12 (spont.)

6/6 (NF) R144/11 (ICR-191) R192/11 (ICR-364- R25 21 ± 7 0.83 ± 0.01OH

5/5 (HC) R295/5 (NQ) R363/5 (NG) R41

6 2 1/12 (spont.) R5 8 ± 0 0.72 ±' 0.021/11 (ICR-364-OH)

From the left, Column 2: abbreviations used: (spont.), spon-taneous reversions; see Materials and Methods for other abbrevia-tions. Given is the fraction of total revertants of each originassigned to each class. Column 3: representative revertants werethose selected for peptide mapping and amino-acid analyses ofhistidinol dehydrogenase (HDH). Column 4: standard errors aregiven to the nearest integer. Column 5: values were determinedby separate electrophoresis and coelectrophoresis of revertantand wild-type histidinol dehydrogenase. Standard errors to thenearest 1%. Numbers in Columns 4 and 5 were calculated fromseveral determinations.

T9A

was highly mutagenic, producing about 500 colonies per muta-gen spot. A total of 55 revertant strains was isolated as fol-lows: spontaneous, 12 from 2 plates; NG-induced, 5 from 2

*__ Til-GI n-Leu-Ala-G u -Leu-Pro-Arg -AlI a-As p-Thr -AI a -Arg-G In-AI a-Leu -Se r -AoIa-Se r -Arg - ------- proteinCAA CUG GCG GAA CUG CCG CGC GCG GAC ACC CC GG CAG GC* CU ------- mRNA

UUGICR-364-OH -1

A~~~~~hisD03052 CAGCUG GCG GAA CUG CCCCGGGAC ACC GAGGC AGG Ce Uo\ ICR,NG

NF,NO,HC rNG,ICR GCG GAC ACC GCC GGC AGG CU

-CG or -GC R22 A UUG-Alo-Asp-Thr-Ala-GIy-Arg-Pro-Leu-Ser-AI o-Ser-Arg-

+10 nucleotides T9A-T//

CUG C GA CACC G CAG GC CU-Leu-Pro-Arg-Gly-His-Arg-Arg-GIn-AI -Leu-

T2\ T9A T//

CAA CU AC UC. ACU GGC GGA ACU GCC GCG CGC GGA CAC CG'FGG CAG GC CU-R5 UUGAGA A UUG

-Gln-Leu-Thr-Ser-Thr-GIy-GIy-Thr-AIo-Alo-Arg-GIy-His-Arg-Arg-GIn-Ala-Leu-T28 T9A T//

FIG. 1. Reconstruction of altered mRNA nucleotide sequences in 305 and its revertants. Amino-acid residues in histidinol dehy-drogenase are juxtaposed with associated mRNA triplets. Frameshift sites in mRNA are bracketed.

Proc. Nat. Acad. Sci. USA 71 (1974)

Dow

nloa

ded

by g

uest

on

June

17,

202

1

1614 Genetics: Isono and Yourno Proc. Nat. Acad. Sci. USA 71 (1974)

TABLE 2. Amino-acid sequence changes in revertants of strain 3052 (continued on p. 1615)

Part A

Gln Leu Ala Glu Leu Pro -Arg2.07 1.90 1.06 0.98 1.012.03 1.88 1.04 1.04 1.00

0.94 1.07 0.99

Gln - LIu - Ala

1.00 0.98 1.020.95 1.01 1.04

0.19 1.04

T28 S-2 mnole Glu - Leu - Pro - ArgHCI 0.060 0.97 0.98 1.03 1.03Ed. 1 0.080 0.05 1.00 1.03 0.98

2 0.090 - 0.10 1.01 0.993 0.110 - - 0.20 1.00

APM 0.100 0.96 1.03

T28S-3 pmolc (Ala,HCO 0.015 1.00

Glu, Lcu, Pro) Arg

1.07 1.00 1.00 1.00

TlI pnole Gin Ala -1Lcu Scr- Ala - Ser- ArgHCI 0.075 1.06 1.93 1.02 1.83 0.99Ed. 1 0.260 1.07 1.96 0.93 1.78 1.05APM 0.080 1.12 0.50 1.74 1.00

TI S-1 iunole Gln - Ala - L uHCI 0.075 1.04 1.01 0.93Ed. 1 0.065 1.03 1.00 0.97CPA 0.100 1.00

T I S-2 pmnole ser - AlaHCI 0.180 0.92 1.00Ed. 1 0.095 0.13 1.00

T I S-3 wunole Ser - ArgHCI 0.055 1.00 1.00Ed. 1 0.090 0.08 1.00

T9A #mole Ala - Asp - Thr - Ala- ArgHCI 0.070 2.06 1.08 0.96 0.98APM 0.040 1.95 1.10 0.89 1.05Ed. 1 0.045 1.10 1.07 0.84 0.93

2 0.045 0.99

Mutagenesis by Polycyclic Carcinogens 1615

TABLE 2. Amino-acid sequence changes in revertants of strain 3052 (continued)

Part C

T9A* punole Gly - H6s - Arg - ArgHCI 0.010 0.50 0.84 2.00

T9A' pUnole Gly - His - ArgHCI 0.050 0.88 1.20 1.00Ed. 1 0.050 0.11 1.16 1.00

2 0.040 - 0.50 1.00

urnole Gin Leu - Ala - Glu - Leu - Pro - Arg

0.040 1.79 2.08 0.97 0.95 1.00

IsMoIc Gin - Leu - Ala

0.045 1.07 1.00 0.930.075 1.08 1.00 0.920.100 0.11 1.00

Ti 1 pnole Gln - (Ala2, Leu, Scr2) - ArgHCI 0.040 1.07 1.90 1.20 1.83 1.00Ed. 1 0.040 1.08 2.03 1.03 1.84 1.00

T28 S-2HCIEd. 1

23

APM

CA§ pnsoleHCI 0.020APM 0.050Ed. 1 0.075

2 0.0703 0.0704 0.0655 0.0656 0.070

prnole

0.0250.0450.0350.0350.100

Ala

0.931.050.150.120.10

1616 Genetics: Isono and Yourno

Tryptic peptides T9A and TIl were absent from peptidemaps of Class 3 revertants (R22, R24, R39). In their place asingle fused peptide, T9A-T11, was located. In both casesexamined (R22, R24) the amino-acid composition of the"double" peptide was the sum of that of T9A and Til, exceptfor substitution of Gly and Pro residues for Ala and Gln. Thepeptide from the ICR-191-induced revertant, R22, was se-quenced (Table 2A). The double peptide results from altera-tion of the normal -Ala-Arg-Gln- sequence at the T9A-T11border to the trypsin-resistant sequence -Gly-Arg-Pro-. Thechanges found are consistent with the normal electrophoreticmobility of Class 3 enzymes (Table 1). In the case of Class 5revertants (R14, R19, R25, R29, R36, R41) only trypticpeptide T9A was altered. As in R5, this peptide was re-placed by the basic peptides T9A and T9A'. These matchedthe R5 peptides in location and amino-acid composition. Pep-tide T28 was normal in location and composition. Only R41was not analyzed for amino-acid composition. Peptides T9Aand T9A' from the NF-induced revertant R14 were sequencedand were, as expected, identical to the corresponding peptidesof R5 (Table 2C). The reduced electrophoretic mobility ofclass 5 histidinol dehydrogenase (Table 1) is seen to resultfrom a net gain of two positive charges.Chymotryptic peptide maps of Class 5 enzymes revealed

two extra basic peptides, designated CA and CB. These pep-tides from R14 were sequenced and found to represent bridgesbetween the affected tryptic peptides (Fig. 1). CA spans T28and T9A; CB spans T9A and T11. Although correspondingwild-type chymotryptic peptides were not detected by pep-tide mapping, the combined data allow the ordering of trypticpeptides as NHrT28-T9A-T11-COOH.

Histidinol dehydrogenase from the spontaneous Class 2revertant R2 was purified and characterized as above (datanot shown). From amino-acid composition data on purifiedpeptides the sequence of T9A was deduced to be Ala-Asp-Thr-Ala-Gly-Arg; that of Til, Ala-Leu-Ser-Ala-Ser-Arg(changes are italicized). R2 shows normal electrophoreticmobility. This revertant most likely arises from insertion of aG residue in mRNA after the arginine triplet for T9A (Fig.1). Enzyme from spontaneous Class 1 revertant R12 was like-wise examined. In this case peptides T9A and Til were re-placed by a fused peptide. From amino-acid composition ofthe fused peptide the deduced sequence is Ala-Asp-Thr-Gty-Gln-Ala-Leu-Ser-Ala-Ser-Arg (data not shown). The loss of apositive arginine residue explains the fusion of T9A and Ti1and the increased electrophoretic mobility of this enzyme(Table 1). The revertant probably results from deletion of twoadditional DNA base pairs from the 305 site (Fig. 1).The data show that the original 3062 mutation is a -1

deletion affecting the C-terminal portion of T9A (Fig. 1).Most likely this represents deletion of a G* C pair from a mis-pairing-prone DNA repeat of three G C pairs. ICR com-pounds seem to favor this type of repeating sequence asframeshift target sites (17, 18, 22, 23). Other sequences arepossible at the 3062 site, however. While the 3052 frameshiftwas induced with an acridine derivative, ICR-364-OH, theICR compounds tested do not induce reversion to the wildtype DNA sequence in 30562, i.e., restore the deleted G Cpair. As discussed above, neither were spontaneous revertantsof this type detected. We have observed a similar reversionpattern for a -1 frameshift at a different site in the hi8Dgene. This frameshift clearly results from deletion of a G- C

pair from a DNA repeat of three G- C pairs (Ino and Yourno,manuscript in preparation). ICR compounds seem to requireDNA repeats of three or more base pairs for frameshift muta-genesis.The ICR compounds tested here predominantly induce +1

frameshifts in a DNA tract narrowly separated from the 3052site, giving rise to Class 3 revertants (Fig. 1). A Gly-Arg-Prosequence replaces Ala-Arg-Gln in Class 3 revertants. In com-bination with previous observations on ICR mutagenesiscited above, this again suggests that reversion results fromaddition of a G- C pair to a DNA repeat of three G0 C pairs.While we have no composition data bearing on the point, ap-parently the alkylating agent NG is also capable of causing+ 1 reversions at this site. This capacity ofNG is much weakerthan its capacity to cause -1 deletions in similar DNA re-peats. The possibility is slight that Class 3 revertants isolatedas NG-induced are actually from spontaneous background.First, the density of NG-induced revertants on spread plateswas in considerable excess of background. Second, no spon-taneous revertants are included in Class 3.The carcinogen-sensitive site of 3052 proved to be a very

unusual mispairing-proneDNA repeat of -C-G-C-G-C-G-C-G-,close to the 3052 site (Fig. 1). This sequence was variablyreactive to all mutagens tested. In all cases examined the

-G-C-

mutagen-induced event was the deletion of a -C-G-doubletfrom the repeat to yield Class 5 revertants. In addition, sixof the 12 spontaneous revertants were assigned to Class 5 byenzyme properties, although sequence analysis of these wasnot done (Table 1).

Since HC and NG were only weakly mutagenic at this site,the possibility is again raised that some revertants isolated asmutagen-induced are actually spontaneous. The putative -2deletion induced by NG in this instance would indicate, withthe + 1 addition discussed above, newly appreciated, if weak,mutagenic capacities of this alkylating carcinogen. It is likelythat NG causes frameshifts by a less direct mechanism thanthat of intercalating agents. Alkylation damage may causespontaneous mispairing in replication or, by stimulating re-pair, increase the chance of spontaneous mispairing (17, 19).The polycycic carcinogens tested (NF, NQ, HC) reacted

specifically with the repeating doublet sequence. The ICRcompounds, like NG, showed a broader specificity. They re-acted somewhat less avidly at this carcinogen hotspot thanat the proposed DNA repeat of G* C pairs, so characteristi-cally a target of these mutagens. Nitrosofluorene was the mostpotent mutagen at the carcinogen hotspot; reversion responsewas almost an order of magnitude greater than that for in-termediate strength mutagens (NQ, ICR) at this site. Ameset al. have recently reported that NF was the most effectiveframeshift mutagen with 3052 from among 12 fluorene deriva-tives tested. Several other polycyclic carcinogens with anitroso side group are strongly mutagenic for 3052 (4), as areepoxides of some polycycic hydrocarbons (5). The Ames grouphas proposed that reactive metabolites of many polycyciccompounds are carcinogenic by way of causing frameshiftmutations. Requirements for strong activity are a polycycicstructure capable of a stacking interaction with DNA basepairs and a reactive side group to react covalently with DNA(4, 5). The 3052 system should prove useful in examining themutagenic specificity of such chemical carcinogens and othermutagens at the molecular level.

Proc. Nat. Acad. Sci. USA 71 (1974)

Dow

nloa

ded

by g

uest

on

June

17,

202

1

Mutagenesis by Polycyclic Carcinogens 1617

This work was carried out at Brookhaven National Laboratoryunder the auspices of the U.S. Atomic Energy Commission. K.I.was supported by Grant AG-280 to J.Y. from the NationalScience Foundation. We thank Bruce Ames for introducing us tothis problem and for many stimulating discussions. We alsothank William Crockett and Frieda Englberger for excellenttechnical assistance.

1. Miller, J. A. & Miller, E. C. (1971) in Chemical Mutagens:Principles and Methods for their Detection, ed. Hollaender, A.(Plenum Press, New York), Vol. 1, pp. 83-119.

2. Ames, B. N. (1971) in Chemical Mutagens: Principles andMethods for their Detection, ed. Hollaender, A. (Plenum,Press, New York), Vol. 1, pp. 267-282.

3. Ames, B. N., Lee, F. 0. & Durston, W. E. (1973) Proc. Nat.Acad. Sci. USA 70, 782-786.

4. Ames, B. N., Gurney, E. G., Miller, J. A. & Bartsch, H.(1972) Proc. Nat. Acad. Sci. USA 69, 3128-3132.

5. Ames, B. N., Sims, P. & Grover, P. L. (1972) Science 176,47-49.

6. Ames, B. N., Durston, W. E., Yamasaki, E. & Lee, F. D.(1973) Proc. Nat. Acad. Sci USA 70, 2281-2285.

7. Brenner, S., Barnett, L., Crick, F. H. C. & Orgel, A. (1961)J. Mol. Biol. 3, 121-124.

8. Streisinger, G., Okada, Y., Emrich, J., Newton, A., Tsugita,E., Terzughi, E. & Inouye, M. (1966) Cold Spring HarborSymp. Quant. Biol. 31, 77-82.

9. Nirenberg, M., Caskey, R., Marshall, R., Brinacombe, R.,Kellogg, D., Doctor, B., Hatfield, D., Levin, J., Rottman,F., Pestka, S., Wilcox, M. & Anderson, F. (1966) Cold SpringHarbor Symp. Quant. Biol. 31, 11-16.

10. Khorana, H. G., Buchi, H., Gupta, H., Jacob, M., Kossel,H., Morgan, R., Nurang, S., Ohtsuka, E. & Wells, R.(1966) Cold Spring Harbor Symp. Quant. Biol. 31, 29-34.

11. Roth, J., Anton, D. & Hartman, P. (1966) J. Mol. Biol. 22,305-323.

12. Hartman, P. E., Levine, K., Hartman, Z. & Berger, H.(1971) Science 172, 1058-1060.

13. Haese, W. H., Smith, D. L. & Bueding, E. (1973) J. Pharma-col. Exp. Ther. 186, in press.

14. Ames, B. N. & Whitfield, H. J. (1966) Cold Spring HarborSymp. Quant. Biol. 31, 221-225.

15. Yourno, J., Barr, D. & Tanemura, S. (1969) J. Bacteriol.100, 453-459.

16. Yourno, J. & Ino, I. (1968) J. Biol. Chem. 242, 3273-3276.17. Yourno, J. & Heath, S. (1969) J. Bacteriol. 100, 460-468.18. Yourno, J. (1971) J. Mol. Biol. 62, 223-231.19. Oeschger, N. & Hartman, P. (1970) J. Mol. Biol. 101, 490-

502.20. Riddle, D. & Roth, J. (1970) J. Mol. Biol. 54, 134-143.21. Tanemura, S. & Yourno, J. (1969) J. Mol. Biol. 46, 459-

-466.22. Yourno, J. & Kohno, T. (1972) Science 175, 650-652.23. Yourno, J. (1972) Nature New Biol. 239, 219-221.

Proc. Nat. Acad. Sci. USA 71 (1974)

Dow

nloa

ded

by g

uest

on

June

17,

202

1