Mutagenicity of radiations at low doses.

The mutagenic effectiveness of low-LET radiations was studied in E. coli Sd-4, with emphasis on low doses. At the lowest dose tested, 0.75 Gy, the mutation frequency per unit dose was found to be highest. It then dropped to an approximately linear curve in the range 5–50 Gy and became again higher at still higher doses. The system is suitable for studies of the mechanism of deviation from linearity at low doses, of possible importance to risk estimation.

As a result of large scale application of atomic energy, there is public concern about the effects of ionizing radiations. The interest is mainly centered on low doses, of the order of I mGy* or less.
At intermediatehigh radiation doses, experimentally induced effects with a genetic mechanism (mutation, cancer) often exhibit a linear dose response that may be extrapolated through the spontaneous frequency at dose zero. I t has, therefore, become a practice in risk estimates to assume the dose-response curve to be linear all the way down to very low doses: [ R = response; C = spontaneous frequency; CY = slope; D = dose] I t has also become a practice, on the basis of studies of specific locus mutations in mice (Rus-SELL et al. 1959). to consider the risk to be approximately three times lower at dose rates appreciably lower than those usually obtained with standard equipment, of the order of I Gy min- ' (cf. UNSCEAR 1977).
Both these assumptions rest on weak grounds. A number of experiments indicate deviations from linearity at low doses or low dose rates. "Humped" dose-response curves, i.e. with a mutagenic effectiveness higher than expected from linear extrapolation, have been observed, e.g. in Drosophila (OFIEDAL. 1964a, b, 1968. in barley and maize after irradiation of meiotic stages 'TANCOURT et al. 1977) and in yeast ( E K L U N D 1977), and it has been suggested that the mutagenic effectiveness of low-LET radiation is again higher at dose rates below some critical value, around lo-' Gy min-' (LYON et al. 1972). In other materials, no deviation from linearity has been observed (e.g., for somatic mutation in Tradescnntia stamen hairs at doses down t o 2.5 mGy of X-rays or 0.1 mGy of neutrons (SPARROW et al. 1972). and in still other experiments even a beneficial effect of low-dose radiation has been observed (see refs. in EHRF.NBERG 1978). A number of indications are further at hand that the carcinogenic effectiveness of radiations in man might increase with decreasing dose or dose rate (BAUM 1973: BROWN 1977EHRF.NBERG 1978). It is evident that the evaluation of the genetic risks from environmental chemicals encounters the same problems.
Especially in mammalian systems, experiments (EHRENBERG and ERIKSSON 1966; Cf. also DE Nt1-* 1 Gy (gray)= I J kg-'= 100 rad: I milligray of low LET radiation corresponds to an average annual dose of background radiation received by human beings.
at those low doses or dose rates that are of at 40 Gy min-I. Doses in the range of 0.7-100 Gy practical interest become prohibitively expensive, were given.
if not impossible. For improved risk estimations at these low doses/dose rates, an understanding of the mechanisms of deviations from linearity is required. Research aiming at such clarification has to be carried out in suitable microbial systems and the validity of the obtained results for estimation of risk to man must be investigated. I n studies with such aims, it is important to realize the quantization of dose, as a basis of a definition of "low dose" and "low dose rate". If, during X-irradiation, a mammalian cell o r cell nucleus is hit by a photo-or Compton electron, it receives a dose of the order of lo-" Gy (0. I rad) or Gy ( I rad), respectively. If one hit in any of these structures changes the sensitivity to further hits, doses below the given values may be considered "low". If the sensitivity change is reversible, say. within 24 h, then dose rates below about 4 x or 4 x lo-' Gy h-I, respectively, would have to be considered low. A deviating effectiveness of low doses and low dose rates might evidently have the same mechanism.
The present paper describes some preliminary studies of the dose response of induced mutation in E . c d i . The studies aim exploratorily at an evaluation of the suitability of this system for a clarification of mechanisms of deviations from linearity at low doses/dose rates. Already 20 years ago, DIMFREC and SAMS (1959) obtained strong indication that reversion to prototrophy at three loci of E . c d i was more effectively induced by lower doses of X-rays. Due to the smaller size of the bacterial "nucleus", one hit is associated with the absorption of a dose in the range of 0.2-1 Gy. In the present studies the effectiveness of 0.75 Gy was compared with that of higher doses.

Material and methods
Bacteria. -Eschrrichiu coli Sd-4, a streptomycin dependent strain, obtained from Professor G. Bertani, Stockholm, was employed; mutation to streptomycin non-dependence was studied. A brief review of the streptomycin dependence system is presented as an Appendix.
Raditrrion. -X-irradiation was carried out at a Siemens X-ray apparatus operated at 180 kV and IS mA; the dose rate being 1.05 Gy* min-I. y-irradiation was performed in a ' W o gamma-cell Media. -Culturing broth contained per litre the following components: Bacto trypton, 10 g; Bacto yeast extract, 5 g; NaCI, 10 g; glucose, I g; streptomycin, 50 mg.
For plating, IS g Bacto agar was added to the above broth which contained streptomycin when the purpose was determination of survival; streptomycin was omitted for the mutation count. Buffer: 0. I M phosphate buffer, pH 7.
Culruritig arid treutmcnt. -The overnight culture was washed twice with buffer and suspended in the same at around lo9 cells ml-I for irradiation of cells in stationary phase. After irradiation the cells were plated directly for mutation and survival count. In studies of bacteria in log phase, the overnight culture was diluted 1:50 in fresh broth and cultured to log phase.
Stutisticul rvcrluution i f t h t dirtu. -Since we are dealing with small effects, any evaluation of the data would have to include errors of mutation frequency and survival (or the number of bacteria plated) of control and treated groups. Details of the method for the statistical treatment will be given elsewhere. The following equations for the calculation of the standard deviation, (rllbl, of the difference, d$l, between irradiated and control samples was used: In each experiment, Poisson distribution of x and y was certified by the demonstration that X/s2 and y/s2 were not significantly different from I . The variances were weighted and pooled for all the experiments, and then coefficients of variation were calculated for the doses in the 0.75 Gy range and those in the 3 to 12 Gy range. The confidence limits of the quotient between mutations per Gy at 0.75 Gy and mutations per Gy at 3 to 12 Gy were calculated for the pooled data.  Fig. 1 presents the pooled data from six tests with Xor y-rays. Though the points bear large 95% confidence intervals, the mutagenic effectiveness (mutations per Gy per 10" bacteria) is indicated to be highest at 0.75 Gy and tends to decrease continuously to about 5 Gy. At doses above 50 Gy it rises again. The quotient between the mutagenic effectiveness at 0.75 Gy and that in the intermediate dose region -3 to 12 Gywas calculated to be 1.692 with the 95% confidence limits 1.265 and 2.119. In agreement herewith, the doubling dose for mutation to streptomycin nondependence on 8 the basis of the effectiveness at 0.75 Gy would be about 3 Gy, whereas on the basis of the effectiveness at 12 Gy it would be about 5 Gy.

Results and discussion
It seems reasonable to assume that the increasing effectiveness at high doses (see Fig. I ) could be a result of the saturation of the repair systems and partly due to other effects, but a high effectiveness at low doses is more difficult to explain. A general stimulatory effect of radiations has been often observed in plant materials, under conditions so far not well defined (SALEH et al. 1978). Perhaps a higher mutagenicity at low doses should also be seen in the same context. Several explanations for such an observation with respect to other systems have been forwarded: ( I ) A mathematical model based on the correlation between sensitivity to killing and mutational response in heterogeneous populations has been applied by 0 w t : n A t . (1968,1974) and E K I U N I ) (1977) to explain the humped curves in their materials (Drosophi/o and yeast). The model as originally forwarded by OFI tii)Ai. (1968) assumes that the population consists of a sensitive and a resistant fraction; a low dose evokes response from both the sensitive and the resistant populations, whereas at higher doses the sensitive fraction is eliminated, with a decrease of the average mutational response per survivor in consequence. Though this model suits their data. it is difficult to apply to our system due to the lack of resolution of the population into sensitivity fractions in the dose region under consideration. It seems plausible that part at least of a supralinear effectiveness of low doses or low dose rates in cancer initiation, could have a mechanism of this kind (cf. B A U M 1973; BROWN 1977; E H r w N B m G 1978, and quoted papers). ( 2 ) It has been suggested ( EHRFNBFRG and ERIKSSON 1966) that the supralinear hump observed in higher organisms could be a consequence of a radiationinduced delay of development, different for cell fractions of different radiosensitivity. The hump around 0.1 Gy, reported by J1:NSSt:N and RAMI..I (1976) to characterize the dose response of micronuclei in polychromatic erythrocytes could, in later studies (JENSSFN and RAMFI. 1978). be given an explanation of this kind, the effect disappearing if "cumulative" fixation was done at different times. This explanation cannot be valid for the similar hump of the dose response curve of waxy mutations in pollen grains of barley. In these studies all pollen grains reaching maturity were collected (ilk. NF. TTANCOURT et al. 1977); besides, the effect was pronounced also at low dose rates of chronic irradiation ( E H R E N R E R G and ERIKSSON 1966).
In the bacterial system of the present study, it is most probable that every mutated bacterium, even if its development is delayed, would have equal chance of developing a colony in the mutation test.
(3) With reference to biochemical changes observed at doses below lo-' Gy, EHRENBERG and ERIKSSON (1966) suggested. as an additional possibility, that such changes could include a changed radiation sensitivity. Evidence is now accumulating that repair enzymes are inducible; this concerns error-free repair enzymes induced by alkylating and/or arylating agents ( Wi.rKiN 1967). A supralinear hump of dose-response curves for mutation, as found in the present study would be compatible with inducibility of enzymes involved in error-free excision repair. Similar conclusions were recently drawn by Ai.1 M A N N and TUS(. HI (1978) with regard to genetic effects in man. In the case of inducibility of error-prone repair, increased survival is gained at the cost of increased mutation frequency at doses above the one required for enzyme induction.
The E. c d i system of the present study is suitable for an analysis of the dose-response curve for mutation at low doses, and of the mechanism of a positive deviation from linearity at these doses.
Ar.know,/edgmenrs. -Technical assistance of Anna Markowska and Bjdrn Anderstam is appreciated with thanks. The investigation was supported financially by the Swedish Natural Science Research Council.