int. j. radiat. biol 2002, vol. 78, no. 11, 981 ± 990

EVect of subsequent acute-dose irradiation on cell survival in vitrofollowing low dose-rate exposures

C. R. MITCHELL and M. C. JOINER*

(Received 20 August 2001; accepted 24 June 2002)

Abstract. Marples et al. 1997, Joiner et al. 1999, Short et al.Purpose: Following acute irradiation, excess radiosensitivity is 1999a, b). Cells showing the HRS response exhibitgenerally seen at doses <1 Gy, a phenomenon termed ‘low-dose decreased cell survival compared with the predictionhyper-radiosensitivity’ (HRS). A very strong, HRS-like inverse by the linear-quadratic (LQ) response model baseddose-rate eVect has also been described following continuous low

on extrapolation from higher dose (1–5 Gy) datadose-rate (LDR) irradiation at <30 cGy h - 1 . We report on thesequential irradiation of a cell line by such LDR exposures (reviewed in Joiner et al. 2001) . A possible explanationfollowed by low acute doses, where either treatment individually is that at very low-acute doses, cells ‘choose’ not towould elicit a hypersensitive response. The aim was to determine upregulate radioprotective repair mechanisms toif a prior LDR exposure would remove the HRS normally seen repair damage and therefore eYcient cell kill results.in response to very small acute radiation doses.

As the dose is increased, the cells’ reaction to damageMaterials and methods: T98G human glioma cells were given singlecontinuous LDR exposures of 5–60 cGy h - 1 using a 60Co c- increases resulting in eYcient repair of DNA damagesource. At intervals of 0 or 4 h following LDR irradiation, cells and an enhanced survival response, termed ‘increasedwere further irradiated with a range of acute doses using 240-kVp radioresistance’ (IRR). It is possible that HRS/IRRX-rays. The response to the combined treatment was assessed

exist to promote the removal of small numbers ofusing high-precision clonogenic cell survival assays, and thepotential radiation transformants from cell popula-amount of HRS at acute doses <1 Gy was determined.

Results : LDR at 60 cGy h - 1 to total doses up to 5 Gy in tions following very low or background radiationasynchronously growing cells did not remove HRS in the sub- exposures, yet allowing maximum recovery from asequent acute-dose survival curve. In con� uent cultures, sub- large cytotoxic insult that might threaten the popula-sequent acute-dose HRS was not present after an LDR dose of

tion integrity. It is easy to see how this arrangement5 Gy at either 60 or 30 cGy h - 1 , but returned if a 4-h intervalwould add stability to organized tissues in a widewas left between LDR and acute-dose irradiation. In con� uent

cultures, acute-dose HRS remained for LDR treatments at 5 or variety of plant and animal species. The phenomenon10 cGy h - 1 or if the total dose was 2 Gy. Taking all cultures and had not been studied until methods of measuringdose-rates together, the ‘degree’ of acute-dose HRS, as measured small changes in survival accurately at very low dosesby as , was signi� cantly greater in cells irradiated at LDR to a

became easily accessible, with the advent of � uores-total dose of 2 than of 5 Gy.cence-activated cell sorting and computer-basedConclusions : Initial LDR exposure can aVect a subsequent HRS

response. HRS is reduced after LDR exposures at greater dose microscopy. Out of more than 45 cell lines that haveintensity, but can recover again within 4 h of completion of LDR been tested, 76% show HRS/IRR ( Joiner et al. 2001).exposure. This suggests that processes determining increased HRS has also been seen after fractionated exposures,resistance to small acute doses (removal of HRS) might be

where 0.4-Gy doses were given three times a day forgoverned by the level of repairable DNA lesions.5 days resulting in greater cell kill than a 1.2-Gydose given once a day for 5 days (Short et al.

1. Introduction 2001) . Furthermore, an inverse dose-rate eVect hasbeen observed at very low dose-rates (LDR,An increase in radiosensitivity has been detected<100 cGy h - 1 ), whereby a decrease in dose-rate hasfor acute doses <1 Gy in a number of cell linesresulted in an increase in cell kill (Mayes 2001,in vitro. This phenomenon is termed ‘low-dose hyper-Mitchell et al. 2002) ; it is possible that this phenom-radiosensitivity’ or HRS (Lambin et al. 1993, 1994,enon is due to an HRS-like eVect occurring at LDR.1996, Marples and Joiner 1993, Joiner 1994, Singh

The mechanisms behind the HRS/IRR eVect haveet al. 1994, Skarsgard et al. 1996, Wouters et al. 1996,yet to be elucidated. The eVect cannot be due tosensitive subpopulations of cells which, when irradi-

*Author for correspondence at: Karmanos Cancer Institute, ated, are killed oV resulting in a greater than expectedWayne State University, Hudson-Webber Bldg, 4100 John R, cell kill in the low-dose region. This is becauseDetroit, MI 48201-2013, USA; e-mail: joinerm@kci.wayne.edu

HRS/IRR responses have been observed in syn-Gray Cancer Institute, Mount Vernon Hospital, Northwood,UK. chronized (Marples and Joiner 1993) , con� uent

International Journal of Radiation Biology ISSN 0955-3002 print/ISSN 1362-3095 online © 2002 Taylor & Francis Ltdhttp://www.tandf.co.uk/journals

DOI: 10.1080/095530002100658 9

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

982 C. R. Mitchell and M. C. Joiner

(Short et al. 1999a) and single cell-cycle phase (Short to allow HRS to recur is much less than the intervalrequired to remove an adpative response, againet al. 2002) populations. Mathematical modelling of

the low-dose region has also shown that varying suggesting diVerent mechanistic bases for the twophenomena (Short et al. 2001).sensitivities throughout the cell cycle cannot account

for the large eVect observed (Lambin et al. 1996). Whether IRR is ‘constitutive’ or induced, a dosecertainly exists at which increased radioresistanceAnother explanation for HRS is that cells do not

reliably recognize the low damage intensity in the occurs. This may correspond to a � xed thresholdbelow which DNA repair mechanisms are not trig-low-dose region and therefore repair is ineYcient.

IRR occurs when the greater damage intensity is gered, or there may be an interaction betweendamage levels and repair ability, making repair lessmore easily recognized and repair is upregulated.

There is evidence to suggest the involvement of eYcient at low doses. If there is a damage thresholdabove which constitutive processes start operating orDNA-repair complexes in IRR. Many studies have

shown a correlation between radiosensitivity (meas- DNA repair is actively induced, then it should bepossible to irradiate continuously at a low enoughured as SF2 ) and the ability of DNA–repair complexes

to mend double-strand breaks (DSB) in both normal dose-rate to maintain accumulating damage belowthis threshold, so that a subsequent acute low-doseand tumour cells (Kiltie et al. 1997, Herring et al.

1998, Ward and Marples 2000) . Cell lines with challenge would still elicit HRS. If the dose-rate ofthe initial exposure were raised, there should comediVerent repair abilities have been tested for the

presence of HRS at low doses and the results imply a point at which a subsequent acute low-dose chal-lenge would instead elicit an IRR response. Thethat DSB repair and nucleotide excision repair, but

not base excision repair, are involved in IRR (Skov object of the experiments described here was toprobe this hypothesis.et al. 1994). Inhibitors of poly (ADP ribose) poly-

merase (PARP) have also been shown to block IRRsuggesting that pathways encompassing PARP may

2. Materials and methodsbe involved (Marples et al. 1997) and further studieshave shown evidence of protein synthesis (Marples 2.1. Cell cultureand Joiner 1995) and DNA-PK (Skov et al. 1994,

T98G cells were taken from monolayer cultureJeggo 1997) involvement in IRR.and plated into 25-cm2 tissue culture � asks (orange-While DNA repair is involved in IRR, the triggercapped, Corning, UK) containing 5 ml medium. Aand the mechanisms by which it acts are as yettotal of 5×103 cells were plated to determine survivalunknown. There are two principal hypotheticalin asynchronous growth and 5×105 cells were platedframeworks describing the involvement of DNAto determine the survival of cells grown to con� uence.repair in HRS/IRR. The simplest idea is that IRRIn all experiments, � asks were incubated at 37°Ccorresponds to constitutive DNA-repair processes,with 5% CO2 +5% O2 (balance N2 ) overnight toi.e. no proteins or genes are actually induced byallow cells to attach. The medium was then removedradiation as they are already present. However, thisand the � ask fully � lled with fresh medium. Strips ofconstitutive repair would only operate when the cellpara� lm were wound round the lid to form a tight‘recognized’ damage above a certain putative thresh-seal. Flasks to be used in con� uence survival experi-old, thus producing a more resistant response abovements were then placed in a 37°C incubator untilabout 0.5 Gy. A possible mechanism for this thresholdobserved to be in con� uence for 4 days.would be an increased access by repair complexes

to the DNA as chromatin structure was relaxed bystrand breakage. The second framework embodies 2.2. Irradiationsthe idea that IRR in high-dose-resistant humancell lines re� ects the level of induced DNA repair or Flasks were then fully immersed in a water tank

and irradiated in the dark with 60Co c-rays at LDRDNA–repair complexes, with induction actively con-trolled by a process that monitors increasing levels as described previously (Mayes 2001, Mitchell et al.

2002) . Control (unirradiated) � asks remained inof DNA damage (Marples et al. 1997). This processwas thought to be analogous to the adaptive response. the incubator. Following LDR irradiation, � asks

were removed from the water tank and placed inHowever, Wouters and Skarsgard (1997) suggestedthat IRR and the adaptive response are diVerent an incubator for 0 or 4 h at 37°C. They were

then irradiated with acute exposures of 240-kVpphenomena; their data showed that priming dosesprevent HRS but have no eVect on the high-dose X-rays at doses between 0.05 and 5 Gy, at dose-

rates of 0.23 Gy min - 1 for doses <1 Gy, and atregion of the cell-survival curve. Time-course studieshave also shown that the interval between two doses 0.44 Gy min - 1 for doses >1 Gy (dose-rates were

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

983E Ú ect of acute doses following low dose-rate

chosen to achieve an accuracy of dose delivered to termed the induced-repair (IR) equation:be 1%). Following incubation at 37°C for a further3 h to allow cells to recover, the medium was removed S=expA - arA1+Aas

ar- 1Be - d/dCBd - bd 2B, (1)

and each � ask gently washed twice with phosphate-buVered saline (PBS). Cells were then harvested using where d is the dose, as is the limiting slope, - (ln S )/d,EDTA/trypsin and centrifuged at 1000 rpm for at very low doses, and ar is the value of a applying5 min. The pellet was resuspended in 5 ml medium to the conventional high-dose response (� gure 1). dCbefore being passed through a 21-gauge needle to is a measure of the dose range over which HRSminimize cell clumping. A cell sorter (FACSVantage occurs (at dC the change from as to ar is 63%equipped with a short-enhancement module, Becton complete) and b is a constant as in the LQ equation.Dickinson) was then used to dispense exact numbers Data were � tted to the IR model using non-linearof cells into 25-cm2 � asks each containing 5 ml least-squares regression. If non-overlapping 95%medium, according to the protocol described in detail con� dence limits on the as and ar parameters wereby Short et al. (1999a). SuYcient cells were plated to obtained and 95% con� dence limits on the dCgive 200–300 colonies per � ask, three to six � askswere plated per dose point and six to 12 � asks wereplated from each of the controls. Flasks were thenincubated at 37°C for 10–14 days then � xed andstained with crystal violet and colonies >50 cellswere scored as survivors using a manual colonycounter.

In the experiments with asynchronously growingT98G cells, � asks were irradiated at low dose-ratesof 5, 10, 30 and 60 cGy h - 1 to total doses of either2 or 5 Gy, and no time was left between LDR andsubsequent HDR exposures. In the experiments withcon� uent T98G cells, � asks were irradiated at thesame dose-rates to total doses of 2 or 5 Gy, but twosets of experiments were carried out where intervalsof either 0 or 4 h were left between LDR andsubsequent HDR exposures. As only eight � askscould be irradiated in one dose-rate position at anyone time, two sets of experiments were carried outper dose and dose-rate combination to obtain morethan eight dose points in the � nal HDR acute-dosesurvival curve. The � rst set of experiments consistedof groups each comprising � asks irradiated withHDR X-rays to doses of 0.05, 0.2, 0.4, 0.7, 1, 3 and5 Gy with one control � ask irradiated at LDR only.The second set of experiments consisted of groupseach comprising the remaining intermediary datapoints, 0.1, 0.3, 0.6, 0.5, 0.8, 2 and 4 Gy and onecontrol � ask irradiated at LDR only.

2.3. Data analysis Figure 1. Survival curve obtained after irradiating T98G cellsat acute doses (reproduced from Joiner et al. 2001). Data

Plating eYciency (PE) was measured in each � ask are the mean ± SEM and � tted with the induced-repairas the number of colonies counted divided by the (IR) equation (solid line, equation 1). The broken line

shows the linear-quadratic (LQ) equation derived fromnumber of cells plated. Surviving fraction (SF) wasthe high-dose IR paramaters. At doses <1 Gy the LQcalculated by the ratio of PE in each HDR-irradiatedmodel, using initial slope ar , substantially underestimates� ask to the mean PE in zero HDR-dose (LDR the radiation response and this HRS domain is better

exposure only) controls. The SF data were analysed described by the IR model with a much steeper initialslope, as .initially by � tting to a modi� cation of the LQ model,

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

984 C. R. Mitchell and M. C. Joiner

parameter did not include zero, signi� cant HRS was and 30 and 5 cGy h - 1 to a total dose of 2 Gy) had asigni� cant number of points in the low-dose regionconsidered to be present using this analysis. Either

as or the ratio as/ar could give an estimate of the below the LQ � t. Of these � ve, four (60, 30 and10 cGy h - 1 to a total dose of 5 Gy, and 30 cGy h - 1‘degree’ of HRS. This fact may be relevant if the

eVect of a pretreatment is not simply to � ip the low- to a total dose of 2 Gy) also had a signi� cant numberof points below the LQ � t obtained using all datadose response from the presence to the absence of

HRS, but to transit gradually between the two states where the low-dose data points were included. Thus,all of the data sets showed signi� cant HRS using atas damage in� icted by the pretreatment increases.

Data were also tested with an alternative modi� ca- least one of the four statistical methods, indicatingthat IRR is not triggered after priming with dose-tion of the LQ equation:rates up to 60 cGy h - 1 to total doses of up to 5 GySF=C exp ( - ad - bd 2 ). (2) in asynchronously growing T98G cells.

Table 2 lists the results of analyses from experi-This takes into account how excess low-dose cellkilling ‘pulls down’ a survival-axis intercept, C, to ments where cells were held in con� uence arrest. In

these experiments, a 0-h interval was used betweenbelow <1.0, which would be the value of C if thedata were well-modelled by the standard LQ equa- LDR and HDR irradiations. HRS was con� rmed in

� ve out of eight of the dose/dose-rate combinations,tion in the absence of HRS. In this analysis, signi� c-ant HRS was considered to be present if the upper as tested using the four statistical methods. However,

HRS was not detectable in cells following LDR95% con� dence limit on C was <1.0.In a third method of analysis, data were � tted with irradiation with 5 Gy at dose-rates of 30 and

60 cGy h - 1 (� gure 2) and a � t of the data to the IRa standard LQ equation (i.e. equation 2 without theparameter C) using two approaches. The � rst � t was model was not possible, nor was HRS signi� cant

following LDR irradiation with 2 Gy at a dose-ratecarried out using all data in the survival curve toobtain an LQ � t and the second � t omitted the low- of 60 cGy h - 1 . Of the remaining � ve dose/dose-rate

combinations where IR � ts were achieved, fourdose data (<1 Gy), using only the high-dose data toobtain a curve � t. Data points in the low-dose region showed signi� cant HRS (10 and 5 cGy h - 1 to a total

dose of 5 Gy, and 30 and 10 cGy h- 1 to a total dose(<1 Gy) were then assessed to be above or beloweach LQ curve. A student’s t-test was used to deter- of 2 Gy). Only one data set showed signi� cant HRS

with the C-intercept analysis (10 cGy h - 1 to a totalmine whether the number of data points below thecurve, in each case, was signi� cant. dose of 5 Gy) although in all six of the eight

dose/dose-rate combinations where IR � ts were alsoachieved, the C-parameterized all-data LQ � ts3. Results crossed the surviving fraction axis at <1 (60, 30, 10and 5 cGy h - 1 to a total dose of 2 Gy, and 10 andTable 1 shows the outcome of statistical analyses

of survival curves obtained from irradiating asyn- 5 cGy h- 1 to a total dose of 5 Gy). Three dose/dose-rate combinations showed a signi� cant number ofchronous T98G cells with 60Co c-rays at various

LDRs to total doses of 2 or 5 Gy, then exposing them low-dose data points below the ‘high-dose only’ LQ� t (10 and 5 cGy h - 1 to a total dose of 5 Gy, andimmediately to HDR X-rays. Signi� cant HRS/IRR

was observed for each dose/dose-rate combination 5 cGy h- 1 to a total dose of 2 Gy) with two of thesealso indicating HRS by the ‘all data’ LQ � tusing at least one test when analysing the data using

the four statistical tests described above. IR curve (5 cGy h - 1 to a total dose of both 2 and 5 Gy). Sincesigni� cant HRS was not detectable using any of the� ts were obtained for all but one dose/dose-rate

combination (60 cGy h- 1 to 2 Gy). Signi� cant 95% statistical methods following 60 cGy h - 1 to 2 or 5 Gy,or 30 cGy h- 1 to 5 Gy (� gure 2), this would suggestcon� dence limits on the IR parameters were obtained

from � ve of the eight dose/dose-rate combinations that IRR is only triggered at dose-rates above30 cGy h - 1 to a total dose of 2–5 Gy in con� uent(60, 30 and 10 cGy h - 1 to a total dose of 5 Gy and

10 and 5 cGy h - 1 to a total dose of 2 Gy). The T98G cells when the acute dose is given immediatelyafter the priming dose. If the LDR dose intensity isoutcome of assessing the surviving fraction axis inter-

cept (C) after � tting the modi� ed LQ equation is that less than this, then HRS remains in the subsequentacute-dose HDR survival curve.all dose/dose-rate combinations show signi� cant

HRS, with the upper 95% con� dence limit on C The analysis of data from the experiments wherecells held in con� uence arrest with a 4-h intervalbeing <1.0. When an LQ � t was obtained for the

high-dose data only (dose 1 Gy) in each of the between LDR and HDR irradiations is shown intable 3. For all dose/dose-rate combinations, an IRdose/dose-rate combinations, � ve of the eight data

sets (60, 30 and 10 cGy h- 1 to a total dose of 5 Gy, � t was obtained with seven of the eight being

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

985E Ú ect of acute doses following low dose-rate

Table 1. Presence or absence of signi� cant HRS as determined by four diVerent methods of analysis of data acquired by asynchronousT98G cells.

Signi� cantIR model Upper CL data below Signi� cant data

Dose-rate Dose with of C intercept all-data below high-dose(cGy h - 1 ) (Gy) signi� cant CL <1 LQ � t LQ � t as as /ar dc

60 2 – – –

60 5 3.14 9.10 0.291.77–6.12 4.65–92.1 0.14–0.56

30 2 1.77 11.3 0.681.26–2.55 – –

30 5 3.87 15.0 0.131.81–7.97 6.93–30.3 0.083–0.22

10 2 17.1 30.7 0.0282.25–2 4.01–2 0.0068–0.096

10 5 7.16 8.77 0.153.55–17.1 5.00–18.2 0.073–2

5 2 0.823 12.9 0.390.290–2.16 2.05–2 0.018–2

5 5 0.603 2.43 0.610.290–2 – –

Cells were irradiated with 60Co c-rays at a � xed LDR to a � xed total dose, then immediately following, cultures were exposed tograded doses of X-rays at HDR to obtain a cell survival versus dose relationship. Ranges in italic represent 95% con� dence intervals.

Table 2. Results on con� uent T98G cells with a 4 h interval between LDR at HDR exposure (cf. table 1).

Signi� cantIR model Upper CL data below Signi� cant data

Dose-rate Dose with of C intercept all-data below high-dose(cGy h - 1 ) (Gy) signi� cant CL <1 LQ � t LQ � t as as /ar dc

60 2 0.860 2.93 1.17

60 5 – – –

30 2 6.41 16.1 0.0651.21–20.9 2.97–52.6 0.027–0.16

30 5 – – –

10 2 4.45 23.8 0.120.617–12.5 2.27–2 0.052–0.41

10 5 1.19 4.22 0.280.482–2 1.65–12.4 0.032–1.59

5 2 2.85 8.47 0.140.580–6.89 0.170–19.7 0.015–0.64

5 5 2.05 2 0.320.610–4.56 2.32–2 0.16–1.03

signi� cant (60, 30, 10 and 5 cGy h- 1 to a total dose analysis (60 and 10 cGy h - 1 to a total dose of 5 Gy,and 30 and 5 cGy h - 1 to a total dose of 2 Gy) andof 2 Gy, and 60, 30 and 5 cGy h - 1 to a total dose of

5 Gy). In all eight dose/dose-rate combinations, the two of these (60 cGy h - 1 to a total dose of 5 Gy, and30 cGy h - 1 to a total dose of 2 Gy) were signi� cantC-parameterized LQ curves intercepted the survival

axis at <1.0, although only � ve cases were signi� cant when analysed with the ‘all-data’ LQ method. All ofthe dose/dose-rate combinations showed signi� cant(60, 30 and 5 cGy h- 1 to a total dose of 5 Gy, and

10 and 5 cGy h- 1 to a total dose of 2 Gy). Four data HRS by at least one of the four statistical methods,including the three (60 cGy h- 1 to 2 or 5 Gy, andsets showed signi� cant HRS with the ‘high-dose LQ’

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

986 C. R. Mitchell and M. C. Joiner

30 cGy h - 1 to 5 Gy) (� gure 3) that did not show HRSwhen irradiated immediately after LDR priming.Thus, for LDR exposures in con� uent T98G cells atthe highest dose intensities tested (30–60 cGy h - 1 toa total dose of 2–5 Gy), which remove HRS in thesubsequent low-dose acute HDR response, a 4-hinterval left following LDR exposure allows the‘recovery’ from IRR and a reinstatement of HRS inthe low-dose acute HDR response.

The degree of HRS was compared across protocols(i.e. asynchronous growth zero interval LDR–HDR,con� uent growth zero or 4-h interval LDR–HDR)also between dose rate of LDR and total dose ofLDR, using the parameters as and as /ar (tables 1–3).No signi� cance of as/ar was found, although therewas a strong trend for higher as /ar in the acute-dosesurvival curve following LDR exposure to 2 Gycompared with 5 Gy ( p<0.07, paired t-test). If thevery high as /ar for 5 cGy h- 1 to 5 Gy (table 2)was excluded, this comparison became signi� cant( p<0.039, paired t-test). although there is no proced-ural reason for excluding this value except for itssize. However, as was signi� cantly greater in the 2 Gycompared with the 5 Gy LDR-irradiated groups,using all the available data ( p<0.048, paired t-test).These data may re� ect greater repair inductionfollowing increased damage at 5 Gy compared with2 Gy LDR exposures. For either the 2- or 5-Gy LDR-irradiated groups, no signi� cant dependence of eitheras /ar or as on either dose-rate or protocol wasdetected, although there was a trend towards higheras and as/ar in con� uent cells allowed 4 h comparedwith 0 h before HDR irradiation (tables 2 and 3)and this was more marked in the groups given LDRto 2 Gy compared with 5 Gy. These data mightsuggest repair occurring during the 4-h interval thatreduced damage with correspondingly greater HRSexpression.

4. Discussion

There is considerable circumstantial evidenceindicating involvement of DNA-repair processes inHRS/IRR ( Joiner et al. 2001) . Whether this repairis constitutive or actively induced is unknown, but in

Figure 2. Cell-survival curves obtained by irradiating con� uentT98G human glioma cells at low dose-rate followedimmediately by graded acute doses at high dose-rate. Cellsurvival was assessed using the cell-sort protocol. Dataare the mean ± SEM and � tted with the linear-quadraticmodel (solid line). Insets show the low-dose regions ingreater detail. (A) LDR of 30 cGy h - 1 irradiated to atotal dose of 5 Gy; (B) LDR of 60 cGy h - 1 irradiated toa total dose of 5 Gy.

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

987E Ú ect of acute doses following low dose-rate

either case there could be two ways in which thesystem responds to increasing damage. First, therecould be a � xed threshold level of damage requiredto be exceeded for repair to take place. Second, therecould be a continuous transition of increasing repairactivity as damage levels increase. To test these twopossibilities, the experiments described here weredesigned to investigate the combination of ‘priming’dose and dose-rate, given at LDR, that would aVectresponse to a subsequent HDR exposure and boththe presence or absence of HRS in the HDR expo-sure, along with the ‘degree’ of HRS as estimated byas or as /ar , were assessed.

HRS remained present in asynchronously growingT98G cells irradiated with LDR 60Co c-rays to totaldoses of 2 and 5 Gy and then immediately irradiatedwith HDR X-rays. Thus, radioprotective mechanismswere apparently not triggered at dose intensities lessthan the equivalent of 60 cGy h - 1 to 5 Gy, whichsuggests this as a minimum threshold level for theinduction of radioresistance. However, con� uentT98G cells irradiated under similar conditions didnot always show HRS in subsequent HDR exposures;thus with the highest dose intensities (30 cGy h - 1 to5 Gy, and 60 cGy h - 1 to 2 or 5 Gy) there was nosigni� cant HRS using any of the four analytical tests.Even visually assessing the data (e.g. � gure 2), therewas no indication of HRS at this level of pretreatmentdamage and, in these cases, a putative threshold forIRR was exceeded.

However, when con� uent cells were irradiated invirtually identical experiments, but with a 4-h intervalleft between LDR and HDR irradiations, HRS wasobserved in all cases, including the highest doseintensities (� gure 3). Thus DNA-repair processes, atincreased or background level, could have reducedthe amount of damage within those 4 h to below aputative threshold, and therefore when cells wereirradiated at HDR, HRS was again observed. Thefact that HRS was observed following a total dose of2 Gy at 30 cGy h - 1 and not following 5 Gy and notfollowing 2 or 5 Gy at 60 cGy h - 1 (table 2) suggeststhat the threshold for IRR in con� uent cells corre-

Figure 3. Cell-survival curves obtained by irradiating con� uentT98G human glioma cells at low dose-rate then followed4 h later by graded acute doses at high dose-rate. Cellsurvival was assessed using the cell-sort protocol. Dataare the mean ± SEM and � tted with the linear-quadraticequation (broken line) and the induced-repair equation(solid line). Insets show the low-dose regions in greaterdetail. (A) LDR of 30 cGy h - 1 irradiated to a total doseof 5 Gy; (B) LDR of 60 cGy h - 1 irradiated to a total doseof 5 Gy.

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

988 C. R. Mitchell and M. C. Joiner

Table 3. Results on con� uent T98G cells with a 4 h interval between LDR at HDR exposure (cf. table 1).

Signi� cantIR model Upper CL Signi� cant data below

Dose-rate Dose with of C data below high-dose(cGy h - 1 ) (Gy) signi� cant CL intercept <1 all-data LQ � t LQ � t as as /ar dc

60 2 28.4 54.6 0.0325.74–4000 10.7–4630 0.0076–0.063

60 5 10.6 20.0 0.0702.32–84.2 4.19–149 0.019–0.16

30 2 7.05 26.4 0.0802.59–14.5 9.25–52.2 0.043–0.13

30 5 7.39 18.3 0.0652.00–27.4 4.93–66.8 0.026–0.15

10 2 18.2 37.8 0.0395.19–63.2 10.6–132 0.020–0.072

10 5 1.59 9.38 0.100.0480–5.66 0.530–33.8 0.025–0.21

5 2 4.60 8.34 0.101.21–14.8 2.16–135 0.013–0.30

5 5 4.08 14.2 0.281.79–9.10 4.49–2 0.12–0.63

sponds to damage produced by a dose between 2 continuous transition between HRS and IRR asdamage increases, rather than a simple ‘damageand 5 Gy for 30–60 cGy h - 1 exposures.

As shown in tables 1–3, HRS was rated initially threshold’ below and above which HRS is presentand absent.in the study as ‘present’ or ‘not present’. This crude

assessment gave maximum chance of detecting HRS, There was a trend towards higher as/ar and as forcon� uent T98G cells given a 4-h interval betweenas each data set could be tested with four diVerent

methods, three of which did not assume an under- LDR and HDR irradiations, although this diVerencediid not reach signi� cance at p<0.05. As a specula-lying mathematical description of the HRS/IRR

phenomenon. This is a robust approach, but a tion, higher as following a ‘recovery’ interval mightbe expected on the basis of damage levels falling todisadvantage is that it can mask any variation in the

actual amount of HRS which may depend on the dose below the threshold level for the induction of repair,or falling to a level where there are fewer repairintensity of the LDR exposure. To approach this

problem does require a mathematical framework for processes taking place, and might result from back-ground DNA repair. Restitution of HRS within afully describing HRS in the HDR cell survival curve;

such a formulation is given in equation (1). The timescale of about 4 h, consistent with DNA repair,is supported by Short et al. (2001) who showed thatparameters in this model were determined where

possible for each data set and then the whole study by separating two doses of 0.4 Gy in time, an increasein cell kill was observed compared with giving a(tables 1–3) was tested for any dependence of any of

the model parameters on protocol, dose-rate or dose. single 0.8-Gy dose and that the maximum increasein cell kill was obtained with a separation of 4–6 h.Both as/ar and as varied with the dose delivered in

the LDR exposure. This was signi� cant in the case Shifts in the balance between HRS and IRR on thisshort timescale also indicate that HRS/IRR isof as with a smaller as corresponding to the higher

dose (5 Gy). The parameter as is a measure of the probably not a manifestation of a classical adaptiveresponse which typically operates over a longer time-initial slope in the low-dose region of the cell-survival

curve, and in HRS-expressing situations, a lower scale, as was pointed out by Wouters and Skarsgard(1997).value is associated with a more radioresistant low-

dose response ( Joiner et al. 2001). In the present We have reported that an inverse dose-rate eVectis present in both asynchronously growing and con-context, it is evidence that even when HRS is not

abolished by LDR pretreatment, it is nevertheless � uent T98G cultures (Mayes 2001, Mitchell et al.2002) (� gure 4). However, there are diVerences inreduced in the higher dose LDR exposure compared

with the lower dose. This supports the idea of a the radiation response between these two growth

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

989E Ú ect of acute doses following low dose-rate

conditions; con� uent cells are (not unexpectedly)more radioresistant and also exhibit a change fromdose-rate sparing to inverse dose-rate eVect at alower dose rate (about 30 cGy h - 1 ) compared withcells in asynchronous culture (about 100 cGy h - 1 ).Some of this diVerence undoubtedly re� ects diVer-ences in cell-cycle distributions under the diVerentgrowth conditions. However, Mitchell et al. (2002)have argued a strong case for a homology betweenthis inverse dose-rate eVect (IDRE) and the HRSseen in small, acute single-dose exposures. In � gure 4,>5 Gy the 30 and 60 cGy h- 1 exposures are in theregion of IDRE (HRS-like) in asynchronous cells,but are in the region of dose-rate sparing (IRR-like)for the con� uent cells. This parallels exactly thestrong presence and absence of HRS reportedhere in asynchronous and con� uent cell cultures,respectively, following 5-Gy exposures at 30 and60 cGy h - 1 (compare rows 2 and 4 in tables 1 and 2).

In conclusion, the experiments described here werecarried out to assess the prevalence and degree ofHRS/IRR in T98G cells after LDR ‘priming’ expo-sures. The data demonstrate that previous exposureto LDR c-rays may remove the HRS response tosubsequent acute-dose X-rays, if the LDR exposureis at a high enough dose intensity. It is speculatedthat this may re� ect either increased DNA repair inproportion to damage in� icted by the LDR irradi-ation, or there may be a threshold of damage, doseand dose-rate whereby IRR is triggered. On balance,the former seems more likely. An interval of 4 hbetween LDR and acute irradiation allows the ‘recov-ery’ of HRS, possibly re� ecting a decrease in therate of repair in proportion to damage reductionor by damage levels falling below a putative thresh-old for IRR triggering. Further investigations arerequired to examine these processes at themolecular level.

Acknowledgements

This work was supported by Amersham plc andthe Cancer Research Campaign.

ReferencesFigure 4. Cell-survival curves obtained after using low dose-rate 60Co c-rays to irradiate asynchronous (A) and con- Herring, C. J., West, C. M., Wilks, D. P ., Davidson, S. E.,� uent (B) T98G cells (from Mayes 2001, Mitchell et al. Hunter, R. D., Berry, P ., Forster, G., MacKinnon,2002) . Curves are labelled with the dose-rate in cGy h - 1 ; J., Rafferty, J. A., Elder, R. H., Hendry, J. H . andthe acute dose-rate was 33 cGy h - 1 . The relative clono- Margison, G. P., 1998, Levels of the DNA repair enzymegens per � ask were calculated by multiplying the surviving human apurinic/apyrimidinic endonuclease (APE1,fraction by the relative cell yield following irradiation. APEX, Ref-1) are associated with the intrinsic radio-Data are the mean ± SEM. sensitivity of cervical cancers. British Journal of Cancer, 78,

1128–1133.Jeggo, P. A., 1997, DNA-PK: at the cross-roads of biochemistry

and genetics. Mutation Research, 384, 1–14.Joiner, M. C., 1994, Induced radioresistance: an overview and

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.

990 E Ú ect of acute doses following low dose-rate

historical perspective. International Journal of Radiation of low dose-rate 60Co gamma ray exposure on humantumor cells in vitro. Radiation Research, 158, 311–318.Biology, 65, 79–84.

Short, S. C., Kelly, J., Mayes, C. R., Woodcock, M. andJoiner, M. C., Lambin, P . and Marples, B., 1999, AdaptiveJoiner, M. C., 2001, Low-dose hypersensitivity afterresponse and induced resistance. Comptes Rendus defractionated low-dpse irradiation in vitro. Internationall’Academie des Sciences Serie III, 322, 167–175.Journal of Radiation Biology, 77, 655–664.Joiner, M. C., Marples, B., Lambin, P ., Short, S. C. and

Short, S., Mayes, C., Woodcock, M., Johns, H. andTuresson, I., 2001, Low-dose hypersensitivity: currentJoiner, M. C., 1999a, Low dose hypersensitivity in thestatus and possible mechanisms. International Journal ofT98G human glioblastoma cell line. International Journal ofRadiation Oncology, Biology, Physics, 49, 379–389.Radiation Biology, 75, 847–855.Kiltie, A. E., Orton, C. J., Ryan, A. J., Roberts, S. A.,

Short, S. C., Mitchell, S. A., Boulton, P ., Woodcock, M.Marples, B., Davidson, S. E., Hunter, R. D.,and Joiner, M. C., 1999b, The response of human gliomaMargison, G. P ., West, C. M. and Hendry, J. H., 1997,cell lines to low-dose radiation exposure. InternationalA correlation between residual DNA double-strand breaksJournal of Radiation Biology, 75, 1341–1348.and clonogenic measurements of radiosensitivity in

Short, S. C., Woodcock, M., Marples, B. and Joiner, M. C.,� broblasts from preradiotherapy cervix cancer patients.2002, The eVects of cell cycle phase on low dose hyper-International Journal of Radiation Oncology, Biology, Physics, 39,radiosensitivity. International Journal of Radiation Biology1137–1144.(in press).Lambin, P ., Malaise, E. P . and Joiner, M. C., 1994, The eVect

Singh, B., Arrand, J. E. and Joiner, M. C., 1994, Hypersensitiveof very low radiation doses on the human bladder carcin-response of normal human lung epithelial cells at lowoma cell line RT112. Radiotherapy and Oncology, 32, 63–72.radiation doses. International Journal of Radiation Biology,Lambin, P ., Malaise, E. P . and Joiner, M. C., 1996, Might65, 457–464.intrinsic radioresistance of human tumour cells be induced

Skarsgard, L. D., Skwarchuk, M. W., Wouters, B. G. andby radiation? International Journal of Radiation Biology, 69,Durand, R. E., 1996, Substructure in the radiation279–290.survival response at low dose in cells of human tumorLambin, P ., Marples, B., Fertil, B., Malaise, E. P . and cell lines. Radiation Research, 146, 388–398.

Joiner , M. C., 1993, Hypersensitivity of a human tumour Skov, K., Marples, B., Matthews, J. B., Joiner, M. C. andcell line to very low radiation doses. International Journal of Zhou, H., 1994, A preliminary investigation into theRadiation Biology, 63, 639–650. extent of increased radioresistance or hyper-radiosensitiv-

Marples, B. and Joiner, M. C., 1993, The response of Chinese ity in cells of hamster cell lines known to be de� cient inhamster V79 cells to low radiation doses: evidence of DNA repair. Radiation Research, 138, S126–S129.enhanced sensitivity of the whole cell population. Radiation Ward, T . H. and Marples, B., 2000, Technical report: SYBRResearch, 133, 41–51. Green I and the improved sensitivity of the single-

Marples, B. and Joiner, M. C., 1995, The elimination of cell electrophoresis assay. International Journal of Radiationlow-dose hypersensitivity in Chinese hamster V79-379A Biology, 76, 61–65.cells by pretreatment with X rays or hydrogen peroxide. Wouters, B. G. and Skarsgard , L. D., 1997, Low-dose radi-Radiation Research, 141, 160–169. ation sensitivity and induced radioresistance to cell killing

Marples, B., Lambin, P ., Skov, K. A. and Joiner, M. C., 1997, in HT-29 cells is distinct from the ‘adaptive response’Low dose hyper-radiosensitivity and increased radio- and cannot be explained by a subpopulation of sensitiveresistance in mammalian cells. International Journal of cells. Radiation Research, 148, 435–442.Radiation Biology, 71, 721–735. Wouters, B. G., Sy, A. M. and Skarsgard, L. D., 1996, Low-

Mayes, C. R., 2001, EVects of low dose-rate irradiation on dose hypersensitivity and increased radioresistance in ahuman tumour cells. PhD, University of London. panel of human tumor cell lines with diVerent radio-

sensitivity. Radiation Research, 146, 399–413.Mitchell, C. R., Folkard, M. and Joiner, M. C., 2002, EVects

Int J

Rad

iat B

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Wat

erlo

o on

10/

31/1

4Fo

r pe

rson

al u

se o

nly.