Theoretical analysis of mixed irradiation (3).

As the model we proposed last year was contradictory to experimental data, we revised again the models for mixed irradiation by Zaider and Rossi and by Suzuki, substituting a 'reciprocal-time' pattern of repair function for a first-order one in reduction and interaction factors of the models, although we used a second order repair function last year. The reduction factor, which reduces the contribution of the square of a dose to cell killing in the models, and the interaction factor, which also reduces the contribution of the interaction of two or more doses of different types of radiation, were formulated by using the 'reciprocal-time' pattern of repair function. These newly modified models for mixed irradiation could express or predict cell survival more accurately than the older ones, especially when irradiation is prolonged at low dose rates. We present survival curves of cells calculated from the newly and the older models of assumptive simultaneous mixed irradiation with two or three types of radiation.     

Theoretical analysis of simultaneous mixed irradiation with multiple types of radiation.

A survival model Eq.1 was presented for cells irradiated simultaneously with multiple types of radiation using the extended Zaider-Rossi model, which is model for mixed irradiation with two types of radiation. [equation : see text] Eq.1. Where q(t)=2t0/t-2(t0/t)2 ?1-exp(-t0/t)? Eq.2. Eq.1 was proved by mathematical induction using the concept that mixed irradiation with n types of radiation is considered as mixed irradiation with two types of radiation regarding n-1 types as one type of radiation. The model is not limited by the dose rate of radiation, because its effect is corrected by reduction factor Eq.2. The problem of the model is that Eq.2 was led assuming repair function to be exponential given by Eq.3. tau(t)=exp(-t/t0) Eq.3. However, the repair function is usually expressed by biphasic Eq.4 rather than monophasic Eq.3. tau(t)=Aexp(-t/01)+(1-A)exp(-t/t02) Eq.4. It is, therefore, important to keep in mind that Eq. 4 should be used instead of Eq.3.     

UV endonuclease-mediated enhancement of UV survival in Micrococcus luteus: evidence revealed by deficiency in the Uvr homolog.

Unlike its phage T4 counterpart (also known as endonuclease V), Micrococcus luteus UV endonuclease (pyrimidine dimer DNA glycosylase/apurinic-apyrimidinic endonuclease) has suffered from lack of genetic evidence to implicate it in the promotion of UV survival of the cell, i.e., mutants with its deficiency are no more UV-sensitive than the wild type. On the assumption that the contribution of UV endonuclease is obscured by the presence of a homolog of Escherichia coli UvrABC endonuclease, which has recently been identified in this bacterium, survival studies were carried out in its absence. With 254-nm UV irradiation, which generates not only pyrimidine dimers but also 6-4 photoproducts as lethal lesions, a double mutant defective in both UV endonuclease and the Uvr homolog was shown to be more sensitive than a single mutant defective only in the latter, with a dose reduction factor of approximately 2 at the survival level of 37%. Furthermore, molecular photosensitization, which produces only pyrimidine dimers, revealed an even greater difference in sensitivity, the dose reduction factor being about 3.4. These results indicate that the contribution to cell survival of UV endonuclease, an enzyme specific for pyrimidine dimers, is manifest if the backup by the Uvr homolog is absent.     

Lethal and potentially lethal lesions induced by radiation--a unified repair model

A model of radiation action is described which unifies several of the major existing concepts which have been applied to cell killing. Called the lethal and potentially lethal (LPL) model, it combines the ideas of lesion interaction, irreparable lesions caused by single tracks, linear lesion fixation, lesion repair via first-order kinetics, and binary misrepair. Two different kinds of lesions are hypothesized: irreparable (lethal) and repairable (potentially lethal) lesions. They are tentatively being identified with DNA double-strand breaks of different severity. Two processes compete for depletion of the potentially lethal lesions: correct repair following first-order kinetics, and misrepair following second-order kinetics. Fixation of these lesions can also occur. The model applies presently only to plateau (stationary)-phase cells. Radiobiological phenomena described include effects of low dose rate, high LET, and repair kinetics as measured with repair inhibitors such as hypertonic solution and beta-arabinofuranosyladenine (beta-araA). One consequence of the model is that repair of sublethal damage and the slow component of the repair of potentially lethal damage are two manifestations of the same repair process. Hypertonic treatment fixes a completely new class of lesions which normally repair correctly. Another consequence of the model is that the initial slope of the survival curve depends on the amount of time available for repair after irradiation. The "dose-rate factor" occurring in several linear-quadratic formulations is shown to emerge when appropriate low-dose and long-repair-time approximations are made.     

Incorporating dose-rate effects in Markov radiation cell survival models

Markov models for the survival of cells subjected to ionizing radiation take stochastic fluctuations into account more systematically than do non-Markov counterparts. Albright's Markov RMR (repair-misrepair) model (Radiat. Res. 118, 1-20, 1989) and Curtis's Markov LPL (lethal-potentially lethal) model [in Quantitative Mathematical Models in Radiation Biology (J. Kiefer, Ed.), pp. 127-146. Springer, New York, 1989], which assume acute irradiation, are here generalized to finite dose rates. Instead of treating irradiation as an instantaneous event we introduce an irradiation period T and analyze processes during the interval T as well as afterward. Albright's RMR transition matrix is used throughout for computing the time development of repair and misrepair. During irradiation an additional matrix is added to describe the evolving radiation damage. Albright's and Curtis's Markov models are recovered as limiting cases by taking T----0 with total dose fixed; the opposite limit, of low dose rates, is also analyzed. Deviations from Poisson behavior in the statistical distributions of lesions are calculated. Other continuous-time Markov chain models ("compartmental models") are discussed briefly, for example, models which incorporate cell proliferation and saturable repair models. It is found that for low dose rates the Markov RMR and LPL models give lower survivals compared to the original non-Markov versions. For acute irradiation and high doses, the Markov models predict higher survivals. In general, theoretical extrapolations which neglect some random fluctuations have a systematic bias toward overoptimism when damage to irradiated tumors is compared with damage to surrounding tissues.     

A simple model of radiation action in cells based on a repair saturation mechanism.

This paper describes a new theoretical model for the response of cells to radiation. This model is based on the existence of a lesion interaction mechanism in the cell, along with processes of recovery and repair that are able to repair the damage produced by radiation in the cells. Such a mechanism makes the cells evolve from a sublethal state to a normal one. Repair and recovery are not instantaneous, but are produced over an average period that we suppose is represented by an exponential function. The probability of cellular recovery and repair is also affected by radiation. These mechanisms become less probable as the dose administered to the cell increases (repair saturation mechanism). This model is suitable for instantaneous doses as well as for arbitrary dose rates. Results obtained from the model for normal tissues and low doses are approximately equal to those obtained by the linear-quadratic model or by the incomplete repair model. The model yields a survival curve with an exponential tail for high doses and for long periods of irradiation.     

PLD repair in rat rhabdomyosarcoma tumor cells irradiated in vivo and in vitro with high-LET and low-LET radiation

Results are reported of studies to measure the extent of recovery of potentially lethal damage (PLD) in rat rhabdomyosarcoma tumor cells after irradiation both in vivo and in vitro with either high-LET or low-LET radiation. Stationary-phase cultures were found to exhibit repair of PLD following irradiation in vitro either with low-LET X rays or with high-LET neon ions in the extended-peak ionization region. Following a 9-Gy dose of 225-kVp X rays or a 3.5-Gy dose of peak neon ions, both of which reduced the initial cell survival to 6-8%, the maximum PLD recovery factors were 3.4 and 1.6, respectively. In contrast, the standard tumor excision assay procedure failed to reveal any recovery from PLD in tumors irradiated in situ with either X rays or peak neon ions. PLD repair by the in vivo tumor cells could be observed, however, when the excision assay procedure was altered by the addition of a known PLD repair inhibitor beta-arabinofuranosyladenine (beta-ara-A). When a noncytotoxic 50 microM concentration of beta-ara-A was added to the excised tumor cells immediately following a 14.5-Gy in situ dose of X rays, cell survival in the inhibitor-treated cells was lower than in the untreated cells (0.018 compared to 0.056), resulting in a PLD repair inhibition factor of 3.1. Delaying the addition of beta-ara-A for 1, 2, or 3 h following tumor excision reduced the PLD repair inhibition factor to 1.6, 1.5, and 0.9, respectively. Following tumor irradiation in situ with neon ions in the extended-peak ionization region (median LET = 145 keV/micron), less PLD repair was observed than after X irradiation. For 5.8 Gy of peak neon ions, the PLD repair inhibition factors were 2.1, 1.5, 1.3, and 1.1 at 0, 1, 2, and 3 h, respectively. We interpret the absence of measurable PLD repair using the standard tumor excision assay procedure as resulting from undetectable repair occurring during the long interval (about 2 h) required for the cell dissociation and plating procedures. We conclude that at least for our tumor system, PLD repair does occur after irradiation of tumors in situ, even though it is not detectable using the standard tumor excision assay procedure. Thus a failure to measure such repair by this assay in a given tumor system does not necessarily mean the cells are incapable of PLD repair.     

Discrepancies between patterns of potentially lethal damage repair in the RIF-1 tumor system in vitro and in vivo

Repair of potentially lethal damage (PLD) was studied in the RIF-1 tumor system in several different growth states in vivo and in vitro. Exponentially growing, fed plateau, and unfed plateau cells in cell culture as well as small and large subcutaneous or intramuscular tumors were investigated. Large single doses of radiation followed by variable repair times as well as graded doses of radiation to generate survival curves immediately after irradiation or after full repair were investigated. All repair-promoting conditions studied in vitro (delayed subculture, exposure of cells to depleted growth medium after irradiation) increased surviving fraction after a single dose. The D0 of the cell survival curve was also increased by these procedures. No PLD repair was observed for any tumors irradiated in vivo and maintained in the animal for varying times prior to assay in vitro. The nearly 100% cell yield obtained when this tumor is prepared as a single-cell suspension for colony formation, the representative cell sample obtained, and the constant cell yield per gram as a function of time postirradiation suggest that this discrepancy is not an artifact of the assay system. The most logical explanation of these data and information on radiocurability of this neoplasm is that PLD repair, which is so frequently demonstrated in vitro, may not be a major factor in the radioresponse of this tumor when left in situ.     

Effects of irradiation on the nonlymphoid thymus in vitro

Thymic explant cultures were used to study the radiosensitivity of nonlymphoid thymic components in dogs. Thymic fragments from fetal (50 days gestation), newborn, and juvenile (70 days old) dogs were irradiated in vitro at 0, 0.5, 1, 2, or 4 Gy prior to culture. Colonies were classified as epithelial, spindle, or mixed cell type, and colony numbers were counted and diameters measured. Radiation caused a significant dose-related decrease in the number of spindle cell colonies from all ages. There was a corresponding, but smaller, dose-related increase in the number of epithelial colonies. The diameter of spindle cell colonies also decreased with dose, and this was accompanied by a reduction in cell density. While epithelial colony diameters did not change consistently with dose, there was an overall reduction in cell density in these colonies. This was more severe in the fetal than in the juvenile cultures. These results indicate that the putative mesenchymal (spindle cell) components of the thymus are significantly more radiosensitive than the epithelium at all ages and that fetal epithelium is more sensitive than epithelium from postnatal animals. This suggests that radiation-induced thymic epithelial defects reported after prenatal irradiation could be due to a combination of direct epithelial injury and defective inductive interaction between epithelium and the more radiosensitive mesenchyme.     

Radiation protection of stimulated human lymphocytes by nicotinamide

Nicotinamide (NA) when added to human lymphocytes in vitro together with a mitogen, protected against the inhibition by gamma and UV radiation of stimulated cell growth. When stimulated by phytohemagglutinin (PHA), concanavalin A (Con A) or pokeweed mitogen (PWM) maximum protection has been observed with approximately 1 mM NA (dose reduction factor of 2-3). To obtain protection the cells had to be stimulated immediately after irradiation in the presence of NA. It is suggested that the intracellular level of NAD+ may be rate limiting for excision repair in human lymphocytes irradiated in the G0 phase. This level is presumably increased by exogenously supplied NA, leading to enhanced repair of DNA damage and increased survival.     

Reduction of spontaneous somatic mutation frequency by a low-dose X irradiation of Drosophila larvae and possible involvement of DNA single-strand damage repair

The third instar larvae of Drosophila were irradiated with X rays, and the somatic mutation frequency in their wings was measured after their eclosion. In the flies with normal DNA repair and apoptosis functions, 0.2 Gy irradiation at 0.05 Gy/min reduced the frequency of the so-called small spot (mutant cell clone with reduced reproductive activity) compared with that in the sham-irradiated flies. When apoptosis was suppressed using the baculovirus p35 gene, the small spot frequency increased four times in the sham-irradiated control group, but the reduction by the 0.2-Gy irradiation was still evident. In a non-homologous end joining-deficient mutant, the small spot frequency was also reduced by 0.2 Gy radiation. In a mutant deficient in single-strand break repair, no reduction in the small spot frequency by 0.2 Gy radiation was observed, and the small spot frequency increased with the radiation dose. Large spot (mutant cell clone with normal reproductive activity) frequency was not affected by suppression of apoptosis and increased monotonically with radiation dose in wild-type larvae and in mutants for single- or double-strand break repair. It is hypothesized that some of the small spots resulted from single-strand damage and, in wild-type larvae, 0.2 Gy radiation activated the normal single-strand break repair gene, which reduced the background somatic mutation frequency.     

Role of p21WAF1 in the cellular response to UV

UV or g irradiation mediated DNA damage activates p53 and induces cell cycle arrest. Induction of cyclin dependent kinase inhibitor p21WAF1 by p53 after DNA damage plays an important role in cell cycle arrest after gamma irradiation. The p53 mediated cell cycle arrest has been postulated to allow cells to repair the DNA damage. Repair of UV damaged DNA occurs primarily by the nucleotide excision pathway (NER). It is known that p21WAF1 binds PCNA and inhibits PCNA function in DNA replication. PCNA is also required for repair by NER but there have been conflicting reports on whether p21WAF1 can inhibit PCNA function in NER. It has therefore been difficult to integrate the UV induced cell cycle arrest by p21 in the context of repair of UV damaged DNA. A recent study reported that p21WAF1 protein is degraded after low but not high doses of UV irradiation, that cell cycle arrest after UV is p21 independent, and that at low dose UV irradiation p21WAF1 degradation is essential for optimal DNA repair. These findings shed new light on the role of p21 in the cellular response to UV and clarify some outstanding issues concerning p21WAF1 function.     

Role of p21WAF1 in the Cellular Response to UV

UV or gamma irradiation mediated DNA damage activates p53 and induces cell cycle arrest. Induction of cyclin-dependent kinase inhibitor p21WAF1 by p53 after DNA damage plays an important role in cell cycle arrest after gamma irradiation. The p53 mediated cell cycle arrest has been postulated to allow cells to repair the DNA damage. Repair of UV damaged DNA occurs primarily by the nucleotide excision pathway (NER). It is known that p21WAF1 binds PCNA and inhibits PCNA function in DNA replication. PCNA is also required for repair by NER but there have been conflicting reports on whether p21 can inhibit PCNA function in NER. It has therefore been difficult to integrate the UV induced cell cycle arrest by p21 in the context of repair of UV damaged DNA. A recent study reported that p21WAF1 protein is degraded after low but not high doses of UV irradiation, that cell cycle arrest after UV is p21 independent, and that at low dose UV irradiation p21 degradation is essential for optimal DNA repair. These findings shed new light on the role of p21 in the cellular response to UV and clarify some outstanding issues concerning p21 function.     

Mitotic viability and metabolic competence in UV-irradiated yeast cells

Colony formation is the classic method for measuring survival of yeast cells. This method measures mitotic viability and can underestimate the fraction of cells capable of carrying out other DNA processing events. Here, we report an alternative method, based on cell metabolism, to determine the fraction of surviving cells after ultraviolet (UV) irradiation. The reduction of 2,3,5-triphenyl tetrazolium chloride (or TTC) to formazan in mitochondria was compared with cell colony formation and DNA repair capacity in wt cells and two repair-deficient strains (rad1Delta and rad7Delta). Both TTC reduction and cell colony formation gave a linear response with different ratios of mitotically viable cells and heat-inactivated cells. However, monitoring the formation of formazan in non-dividing yeast cells that are partially (rad7Delta) or totally (wt) proficient at DNA repair is a more accurate measure of cell survival after UV irradiation. Before repair of UV photoproducts (cis-syn cyclobutane pyrimidine dimers or CPDs) is complete, these two assays give very different results, implying that many damaged cells are metabolically competent but cannot replicate. For example, only 25% of the rad7Delta cells are mitotically viable after a UV dose of 12 J/m(2)75% of these cells are metabolically competent and remove over 55% of the CPDs from their genomic DNA. Moreover, repair of CPDs in wt cells dramatically decreases after the first few hours of liquid holding (L.H.; incubation in water) and correlates with a substantial decrease in cell metabolism over the same time period. In contrast, cell colony formation may be the more accurate indicator of cell survival after UV irradiation of rad1Delta cells (i.e., cells with little DNA repair activity). These results indicate that the metabolic competence of UV-irradiated, non-dividing yeast cells is a much better indicator of cell survival than mitotic viability in partially (or totally) repair proficient yeast cultures.     

Evidence for reduced capacity for damage accumulation and repair in plateau-phase C3H 10T1/2 cells following multiple-dose irradiation with gamma rays

The effects of multiple-dose gamma irradiation on the shape of survival curves were studied with mouse C3H 10T1/2 cells maintained in contact-inhibited plateau phase. The dose-fractionation intervals included 3, 6, and 24 h. Following three fractionated doses (5 Gy per dose) of exposures, cells responded to further irradiation by displaying a survival curve with a much reduced shoulder width (Dq) compared to that of the survival curve measured in cells irradiated with single-graded doses alone. The effect on the mean lethal dose (D0) was small and appeared to be significant. The effect on reduction of Dq could not be completely overcome by lengthening the fractionation intervals from 3 to 6 h or 24 h, times in which repair of sublethal damage (SLD) measured by simple split-dose scheme and potentially lethal damage (PLD) measured by postirradiation incubation was completed. Other experiments showed that pretreatments of cells with fractionated irradiation appeared to slow down the cellular repair processes of SLD and PLD. Therefore, the observed change in the shape of survival curves after fractionation treatments may be attributed to a reduction of the cells' capacity for damage accumulation by an enhancement of the lethal expression of SLD and PLD. Although the molecular mechanism(s) is not known, the results of this study indicate that the acute graded dose-survival curve cannot be used a priori to extrapolate and reliably predict results of hyperfractionation. It is probable that for a nondividing or slowly dividing cell population, such an extrapolation may lead to an underestimation of cell killing. Furthermore, the findings of this investigation appear to support an interpretation, alternative to the high-linear energy transfer (LET) track-end postulate, for the effects on cell survival seen at low doses or low dose rates.     

Is repair of DNA strand break damage from ionizing radiation second-order rather than first-order? A simpler explanation of apparently multiexponential repair.

The purpose of this paper is to suggest the hypothesis that repair of radiation damage might be largely a second-order process (binary), as well as or instead of first-order (monoexponential). Second-order means that the rate of repair is proportional to n(2) instead of to n, where n is the number of repairable breaks. Integrating this equation gives a linear plot of the reciprocal proportion of unrepaired lesions, n(0)/n(t), as a function of repair time. This is in contrast to mono- or biexponential processes which give rise to reciprocal plots not consistent with such linearity, except with specially selected distributions with multiple T((1/2))'s. There is the advantage of only one parameter (the first half-time) instead of (2n - 1) parameters for n components. At times greater than 2tau of the longest exponential component, a larger proportion of damage would be incompletely repaired than in a mono- or biexponential model of repair. Data on DNA repair from published laboratory experiments were reanalyzed. Results are presented as graphs of the reciprocal of the proportion of damage remaining as a function of time after irradiation of DNA. If the second-order process is correct, these graphs should be straight lines, even though traditional semilog plots of the same data are markedly concave upward, showing the well-noted slowing down of repair with time after irradiation. All the data sets found in the literature showed a good fit to a straight line representing reciprocal repair. Repair of single-strand breaks in DNA fitted very well, from 1.0 down to 1/40 of the initial damage remaining, with tau values of 5-10 min. Repair of DSBs fitted almost as well. One set of data showed a strong dependence on temperature in the range 10-37 degrees C, with each curve fitting the straight reciprocal plot. The tau values for DSBs were 10-100 min, of similar magnitude to those for repair of animal tissues. The second-order process with a single time parameter could explain the data showing "apparently slowing down" repair previously analyzed by multiexponential formulae requiring more parameters. It appears that second-order repair may play a larger part in repair processes than has usually been assumed. It is suggested that analysis of data on repair of radiation-induced damage could test the second-order (one-parameter reciprocal) analysis, as well as using bi-or multiexponential analyses. If repair in DNA is relevant to recovery in mammalian tissues, there may be serious clinical implications, to be discussed elsewhere.     

Relative sensitivities of repair-deficient mammalian cells for clonogenic survival after alpha-particle irradiation

The clonogenic survival of cells of the radiation-sensitive hamster cell lines irs1, irs2, irs3 and xrs5, representing different DNA repair pathways, was compared to that of their parent lines after alpha-particle irradiation. Measurements of nuclear area were made to calculate the probability of surviving a single alpha-particle traversal, the average number of lethal lesions per track and per unit dose, along with the "intrinsic radiosensitivity" of these cells, allowing for the potential of multiple lethal lesions per traversal. For all cell lines studied, alpha particles were found to be more biologically effective per unit absorbed dose than X rays at inducing cell inactivation. The repair-deficient cells showed an enhanced sensitivity to alpha particles compared to their parent line, but the degree of enhancement was less than for X rays. The reduction in additional sensitivity for alpha-particle irradiation was shown not to be due predominantly to differences in cell geometry limiting the probability of a cell nucleus being traversed. The results suggest that both the nonhomologous end-joining pathway and to a lesser extent the homologous recombination repair pathway play a role in successful repair of alpha-particle-induced damage, although a large proportion of damage is not repaired by either pathway.     

A Model of Photon Cell Killing Based on the Spatio-Temporal Clustering of DNA Damage in Higher Order Chromatin Structures

We present a new approach to model dose rate effects on cell killing after photon radiation based on the spatio-temporal clustering of DNA double strand breaks (DSBs) within higher order chromatin structures of approximately 1–2 Mbp size, so called giant loops. The main concept of this approach consists of a distinction of two classes of lesions, isolated and clustered DSBs, characterized by the number of double strand breaks induced in a giant loop. We assume a low lethality and fast component of repair for isolated DSBs and a high lethality and slow component of repair for clustered DSBs. With appropriate rates, the temporal transition between the different lesion classes is expressed in terms of five differential equations. These allow formulating the dynamics involved in the competition of damage induction and repair for arbitrary dose rates and fractionation schemes. Final cell survival probabilities are computable with a cell line specific set of three parameters: The lethality for isolated DSBs, the lethality for clustered DSBs and the half-life time of isolated DSBs.By comparison with larger sets of published experimental data it is demonstrated that the model describes the cell line dependent response to treatments using either continuous irradiation at a constant dose rate or to split dose irradiation well. Furthermore, an analytic investigation of the formulation concerning single fraction treatments with constant dose rates in the limiting cases of extremely high or low dose rates is presented. The approach is consistent with the Linear-Quadratic model extended by the Lea-Catcheside factor up to the second moment in dose. Finally, it is shown that the model correctly predicts empirical findings about the dose rate dependence of incidence probabilities for deterministic radiation effects like pneumonitis and the bone marrow syndrome. These findings further support the general concepts on which the approach is based.     

Interaction of far- and near-ultraviolet radiation. The occurrence of photo-augmentation and photo-recovery in cultured mammalian cells

Guided by the phenomena of photo-augmentation and photo-recovery, which have been described with respect to the induction of erythema in human skin, experiments were undertaken with cultured mammalian cells to study whether irradiation with far- and near-ultraviolet radiation results in an interaction at the cellular level with respect to cell survival and induction of mutations. Evidence was found for both photo-augmentation and photo-recovery. Photo-augmentation (more than an additive effect) was observed for cell survival when the long-wave ultraviolet irradiation (UVA) preceded the short-wave ultraviolet irradiation (UVB). Photo-recovery (less than an additive effect) was observed for cell survival if the UVA was given after or simultaneously with the UVB. The latter effect, however, was strongly influenced by dose: doses of UVA higher than 20 000 J/m2 no longer lead to photo-recovery in cell survival. For mutation induction, reduction in mutant frequency appears indicated for both combinations of UVA and UVB and for high and low doses of UVA.     

Radiation damage repair capacity of a human germ-cell tumour cell line: inhibition by 3-aminobenzamide

The capacity of a human germ-cell tumour line to repair radiation damage has been investigated by means of a clonogenic assay. Dose-rate dependence studies, split-dose experiments and experiments designed to measure repair of potentially lethal damage have been performed. The cells showed some ability to repair radiation-induced damage in all three types of experiment. An attempt has been made to understand the possible cellular mechanisms of these repair processes by the use of 3-aminobenzamide (3-AB), an agent thought to act by inhibition of ADP-ribosylation. 3-AB added 2 h prior to and removed 18 h after irradiation at a non-toxic dose to unirradiated cells caused a small but consistent increase in cell kill with acute (150 cGy min-1) irradiation, largely involving a reduction in the shoulder region of the survival curve, but had a greater effect in increasing cell kill at a dose rate of 7.6 cGy min-1 and an even greater effect at a dose rate of 1.6 cGy min-1. When 3-Ab was present 2 h prior to the first dose and between two equal doses in a split-dose experiment, inhibition of split-dose recovery was observed. In addition, some inhibition of potentially lethal damage recovery was observed with 3-AB. A possible role for poly(ADP-ribosylation) is thus implicated in the repair of radiation-induced damage of this human tumour cell line during continuous low dose rate or fractionated radiation schedules, although other effects of 3-AB on respiratory metabolism and/or purine synthesis cannot be eliminated as the cause of the observed inhibitory effects.