Mutation Research, 250 (1991) 307-317 © 1991 Elsevier Science Publishers B.V. All rights reserved 0027-5107/91/$03.50 ADONIS 002751079100188G
307
MUT 02540
Revell revisited J o h n R . K . S a v a g e a n d A l i s o n N. H a r v e y MRC Radiobiology Unit, Chilton, Didcot, OXI l ORD (U.K.) (Accepted 9 May 1991)
Keywords: Chromatid breaks; Harlequin; Revell ratios
Summary The controversy of the Classic versus the Exchange theories for the origin of simple chromatid breaks is outlined. Using BrdU harlequin sister-chromatid differentiation four Revell ratios can be defined and these have been obtained and tested as a block in V79 hamster cells. The values are quite different from the simple predictions. However, values similar to those observed (taken as a block) can be readily simulated from Revell theory by making the assumption that intra-chromatid events are dominant (0.7:0.3). They can also be obtained from contingency modelling of Classic theory using the same assumption plus the additional constraint that there is no contribution from isolated single lesions (Poisson class 1). If this latter assumption is correct, then the frequency of breaks involving a colour-jump (ratio III) should not decline to zero as the dose fails. A dose-response experiment shows that it does not, but remains approximately constant at about 12%, even in the unirradiated control. An added complication arises when we discover for the TB/BB situation, that whilst neither breaks nor gaps show any excess BB involvement (sensitisation), lesions involved in interchanges show at least a 2-fold BB excess. Clearly, the chromatid discontinuities we are scoring are not behaving as would be expected of a residue of unrejoined primary breaks (Classic theory) and we infer also that they are not 'simple'.
It is a great pleasure and honour to make a contribution to the 250th volume of Mutation Research. The journal has spanned a large part of my research career and played an invaluable role for both input and output. I offer my sincere congratulations and thanks to Prof. Sobels and his team for the excellent steering that has enabled the journal to achieve this milestone (JRKS).
Correspondence: Dr. John R.K. Savage, MRC Radiobiology Unit, Chilton, Didcot, Oxon. O X l l 0RD (U.K.), Tel. Abingdon (0235) 834393, Fax (0235) 834918.
The journal was launched (1964) at the time of a lively (and often stormy) controversy over the nature and origin of chromatid 'breaks'. The 'Classic' breakage-and-reunion theory (developed by Sax, 1940 and mathematically formalised by Lea and Catcheside, 1942 and Lea, 1946) which lay at the foundation of all aberration thinking, had been seriously challenged by the 'Exchange' hypothesis (Revell, 1959). Mutation Research became an ideal forum for this debate and many of the important papers on this topic have been published in it over the years. Today, in the light of our much increased knowledge of chromosome structure and the
3(18
molecular events that lead to aberrations, some of the early concepts have to be modified (Savagc, 1989, 1990). However, the original breakage-andreunion idea still dominates and is now almost universally accepted, sincc the bulk of the accumulated evidence seems to be in its favour. We thought that it might be timely to air the controversy again, and present some new data which suggests that things may not be as simple as most people might think or wish. The controversy This p a p e r is in no sense a critical review of the topic and the information given in this section is only to provide a basic understanding; those wishing more must consult original papers and discussions (e.g. Lea, 1946; Revell, 1959, 1974; Evans, 1962; Savage, 1975, 1976, 1989). According to 'Classic" theory, the primary events of radiation damage at the chromosome level are breaks in the chromosome continuity thread ( ' c h r o m o n e m a ' - - it must be r e m e m b e r e d that these early workers had a very different picture of chromosome structurc from that current today). These breaks have one of three fates; (a) to restitute i.c. to join back to the original configuration and effectively disappear, (b) to rejoin or rearrange i.e. to join up with another broken end close in space and time and so form an (visible or invisible) aberration, (c) to remain open and appear at metaphase as a simple break at some point in the chromosome. Mathematical considerations and d o s e - r e sponse curve observations require that the initial number of primary breaks greatly exceed those detected: ' O f all the breaks primarily p r o d u c e d . . , only a fraction, and probably a small fraction, survive to the time of fixation when they are observed either as simple breaks or in structural rearrangements' (Lea, 1946, p. 253). ' T h e chromosome breaks which lead to fusions between chromosome arms must constitute only a very small proportion of the total breaks i n d u c e d . . . ' (Sax, 1940, p. 58). So, we have the picture of a large pool of breaks, a small fraction of which are removed by rejoining to form aberrations and the remainder
restitute except for a small residue of (unrejoincd or unrejoinable) simple breaks. Thus, everything starts from a break. In 1959, Revell published a radical challenge to this concept. He proposed that the primary event was not a break (i.c. no loosc joinablc cnds), but an unstable lesion which, left to itself, decays (i.e. there is no such thing as 'restitution" in the Classic sensc).Thesc lesions could, however, interact with one another and form cxchangcs ( - Classic 'rejoins'). The exchange proccss somctimes fails, lcaving unrcjoined ends, a fact that was well known from the observation of incomplete interchanges and non-union isochromatid breaks. Revell argued that in the case of intra-arm intrachanges, some of the incomplete forms, when seen in condenscd chromosomes. would look like (and gcnerally bc scored as) breaks ('discontinuities') - - but these breaks arc secondary, not primary. Since there arc no primary breaks, all discontinuities observcd must arise in this way. He re-classified the simple two-lesion chromatid intrachanges, demonstrated the forms of possiblc cxchangc using the now famous 'Revcll loop' t and how these would a p p e a r at metaphasc. He also proposed two numerical relationships that would bc expected if his hypothesis was correct, and provided confirmatory evidence from Vicia. He also pointed out that the required rarity of breaks had bccn ovcrlooked because a critical distinction had not been made between truc d i s c o n t i n u i t i e s and a c h r o m a t i c - l e s i o n s ('gaps'). Incidentally, this distinction, which is now made in all critical work, originated from this caution. Such a fundamental challenge produced a furore and a flurry of experiments to test the predictions (see Heddle ct al., 1969: Revell, 1974, for refs.). It soon became clear, certainly for
I Although it has come in for much criticism, the loop is not an integral part of Revell theory. It is purely a diagrammatic device to bring into proximity lesions separated longitudinally on a chromosome arm (Savage. 1986). That such lesions occur and that they interact, is inescapable -- the interstitial deletion ('minute') is an obvious witness. All forms of the loop can actually be seen by careful examination of a range of chromatid inter-arm intrachanges.
309
horizontal
"X"
vertical
(H)
"U"
(l-H)
.'g A
la
lb
4a
4b
C~
c*
NUp
NUd
2a
2b
3a
3b
m
c
c
I t-
"0
E
°9_ c+m
Ratio I
12.5]
( l a + l b * 2 a +3a+3b) (4a+4b)
Ratio II
[1.0]
(la+lb) Ratio III
Total 4
(la+lb) Ratio IV
{la+ lb+2a+3a+3b) [0.4]
Total 2
-l/Ratio I
(4a+4b) [I.0]
Fig. 1. The 8 incomplete 2-lesion chromatid intra-arm intrachanges upon which Revell based his theory. For diagrams and details see ReveU, 1959; Savage, 1986, 1989. The lower part of each box indicates how they are scored at metaphase: c, chromatid breaks without colour-jump; c*, break with colour-jump; m, single minute; NUp, NUd, non-union proximal or distal isochromatid breaks. The complete types are not shown but both 2 and 4 are used in ratio 11. If sister-chromatid differentiation is present, 4 ratios can be defined and their default values (all four categories equally likely) are shown in brackets. In practice various constraints operate to change relative frequencies and so modify ratios. The two most important are intra- versus inter-single chromatid events (W,(1- IV)) and ' X ' versus ' U ' type exchange (diagrammatically Horizontal versus Vertical, H,(1-H)). The effect of these two factors on the ratios are shown in Fig. 2.
mammalian cells, that the simple predictions (ratio I = 2.5; ratio II = 1; Fig. 1) were seldom realised, but there were big differences between species, cell type, growth conditions, etc. There were also a number of factors that could modify these ratios (e.g. Savage et al., 1968). Recognising that, on the basis of the Exchange hypothesis, 40% of breaks (la, lb Fig. 1) should be accompanied by an exchange between sister-
chromatids, Heddle et al. (1969) set out to test this, using second-cell division segregation of tritiated thymidine in Potorous to detect label switches. They found 38% of breaks showed them, and concluded that the majority of terminal deletions here were incomplete intrachanges. Later, in a repeat with Chinese hamster ceils (Heddle and Bodycote, 1970)only 15-18% (with evidence that S > G2) label-switch breaks were seen, which they took as being indicative of a mixture of simple and intrachange origins. With the advent of FPG harlequin staining (Wolff and Perry, 1974) which gives excellent differentiation (BrdU, TB/BB) between sisterchromatids and minimal ambiguity of sister-chromatid exchanges (colour-jumps), Wolff and Bodycote (1975) were able to perform a much cleaner test in CHO cells. Scoring over the first 8 h post-irradiation, when a changing mixture of G2 and S cells were present, and with rigorous classification of break categories, they found about 8% of breaks had a colour-jump (only 3% at the most sensitive sample time) and concluded 'that very few of the breaks are incomplete exchanges'. Classic theory does not deny the existence of breaks arising from incomplete intrachanges (they enter into some of Lea's calculations) but it insists that contribution from this source is negligible when compared with those arising from failed restitution of single breaks. In contrast, pure Exchange theory insists that there is n o contribution from single lesions. The Reveli ratios
Using the eight incomplete forms of simple, two-lesion chromatid intra-arm intrachanges 2, we can, with the added dimension of colour-jump breaks (c*) define four Revell ratios (Fig. 1). We
2 There is a range of more complex 3- and 4-lesion intrachanges, first described in full by Fox (1967) under the name 'insertion-intrachanges'. Many of these simulate simple ones when seen at metaphase, but some characteristic ones are always observed. They are seldom considered, but they may be much more common than we think, especially after high-LET radiations or chemical clastogens. These complex forms will not, of course, obey ordinary Revell rules and, if common, will warp observed ratios.
310
and 11 in an identical manner, effects on ratios 11I and IV are quite different.
30
20
A ratio test 0
O
n-
tr"
10
iI / 1 L0
W orH
g 8.0
0.8
,9~,
/
>
0
O .m
IT
i'r
0
Fig. 2. If the t a b l e o f Fig. 1 is s u b j e c t e d to v a r i o u s biases, the r e s u l t i n g Revell ratios d e p a r t d r a m a t i c a l l y f r o m the s i m p l e p r e d i c t i o n s . T w o f a c t o r s t h a t a r e k n o w n to o c c u r a r c the d o m i n a n c e , o r o t h e r w i s e , o f i n t r a - c h r o m a t i d as o p p o s e d to
inter-chromatid exchange events (factor W) and the dominance of "U' over'X" type exchanges -- referred to diagrammatically as ltorizontal versus Vertical exchange (factor H) (Savage, 1986, 1989). Breaks deriving from intrachangcs are Ix~und to be influenced by alterations in these factors.
note that these ratios arc inter-related and that their values, considered as a block, can bc changed by a number of factors. The effects of two of these factors are shown in Fig. 2. Others include incompleteness (a and b forms may not be equal) and, of course, in a BrdU T B / B B context, differential chromatid sensitivity. Equivocal results exist for this last factor (see later) but if present, its main effect will bc to boost intra-arm events (i.e. to increase W, Fig. 1). Because factor variation changes all these ratios in concert, it is important in any tests using this approach, to consider them all together and not in isolation. For example, in Fig. 2 it can bc secn that whilst changes in W or H affect ratios I
Chinese hamster cells (V79-4) were grown fl~r 17 h in MEM containing 10 / z g / m l BrdU. They were then irradiated with various doses of X-rays (250 kVp, 14 mA). Immediately after radiation, the medium was replaced with one containing 10 / z g / m l thymidine. Metaphases were sampled at 1.5 h intervals up to 7.5 h (1.5 h colcemid) and chromosomes stained by a FPG method to distinguish "FF, TB and BB chromatin. Second division cells in the first four samples were scored for all categories of chromatid-type aberrations (Savage, 1976). rl"T patches allowed unambiguous distinction between S and G2 cells. Table 1 shows the ratios we observed. There is a fair scatter, but no systematic trends. We did not find any obvious trends with sample time either, even though we would be looking at different cell mixes. The S cells nearly all had small "IYI"patches, signifying that they were mostly lateS, so it may not be surprising that the marked differences others have found for R l I I , have not been confirmed. In general, compared with the default values, R 1 and R II are bigger, R Ill smallcr and R IV slightly smaller. Inspection of Fig. 2 shows such changes to be characteristic of an increase in W, i.e. that intrachromatid events are more frequent than interchromatid ones. Changing W to 0.75, and multiplying out the Revell table of Fig. 1 gives a fair approximation of observation. Because this table deals with whole aberrations and not the interactions of individual lesions or breaks, simulations are inevitably limited. If we adopt a more realistic contingency approach (discussed in a little more detail below) an even closer approximation can be obtained. Interestingly, the same assumption about the dominance of intra-chromatid events has to be made, and, in addition, we have to eliminate any contribution from the Poisson 1class (i.e. no unrestituted isolated breaks). Of course, whilst this demonstrates that departures from Revell's original predictions do not necessarily negate his theory (which is a conclusion many have been quick to draw in the past)
311
there is an inherent danger in this approach. Having so many variables to play with, a little calculator juggling will usually come up with a tolerable fit to most data sets! However, there is a safeguard if one insists on considering all four ratios as a block that stands or falls together, for then, the number of (biologically reasonable) factor variations becomes quite limited.
Classical predictions Apart from within-experiment estimates (Heddie et al., 1969, 1970; Wolff and Bodycote, 1975) no one seems to have computed the frequency of colour-jump breaks expected on the basis of Classic theory, nor to have predicted intrachange ratios to compare with Revell frequencies. Such calculations are necessarily limited and tedious, but not excessively difficult. On the basic premise that primary breaks will be distributed at random (Lea, 1946), the frequency of 'sites' 3 with 0, 1, 2 . . . breaks will conform to the terms of a Poisson. Two or more breaks may interact and produce a limited variety of intrachange types, conditioned by the disposition of the breaks, within or between chromatids in the site. The random (?) subset of sites which will lead to interchanges are not included in these calculations as they are not expected to contribute to simple breaks. We need then (a) to consider, for each break, and each combination of 2, 3 . . . breaks every possible outcome, and, using assigned probabilities, compute its expected frequency, (b) to assess how this outcome will (or will not) appear at metaphase, and (c) how it will most likely be scored. Then, by summations, we can derive for each relevant aberration category (e.g. break, break with colour-jump, minute, etc.) a partition coefficient. These coefficients can then be used to multiply the appropriate Poisson class to provide its expected contribution to each aberration type. Fur-
3 'Site': W e d o not wish h e r e to m a k e a f o r m a l d e f i n i t i o n - t a k e it as a v o l u m e in which any n u m b e r of b r e a k s can o c c u r a n d in w h i c h they can interact in all possible ways to produce structural intrachanges.
ther summation for the whole Poisson gives the complete aberration spectrum for the 'dose'. Clearly, the zero class will contribute nothing and the one class has only a single possible outcome, a simple break. Two or more breaks can be distributed between or within chromatids and produce the whole range of complete and incomplete intrachanges (including for the three and above the more complex ones not considered in this analysis). It is obvious that the relative contribution to each category is going to change with dose. In particular, the break contribution from incomplete intrachanges (the only source of colour-jump breaks) will diminish as dose decreases, because l-break sites will become dominant as the Poisson mean falls. The principal difficulty in compiling the contingency tables and in deriving the partition coefficients, lies in deciding the probability conditions which underlie Classic theory. Equivocal and conflicting answers to a number of important questions arise when early literature is studied. For example, it is agreed that the majority of breaks restitute and do not form aberrations. Is this because restitution is intrinsically favoured (i.e. must we assign a higher probability to the restitution option, and that in all circumstances?) or, is it because breaks not 'fixed' by inter- or intrachange have nothing else left to do but restitute? (i.e. given two or more breaks, restitution becomes just one competing option.) It is necessary, then, to explore several probability models. We have considered just two cases, with their extensions: Case 1. The probability of restitution is intrinsically dominant under all conditions (i.e. in competition, return to the original configuration is always the most likely result). Case 2. The probability of restitution is not dominant in an interaction situation (i.e. free competition, on an equal basis, for broken ends within a site). Only Case 2 will concern us in this paper. For both models, we have derived partition coefficients for 2- (48 outcomes) and 3- (960 outcomes) break sites. The full contingency table for 4-break sites would require the evaluation of 26 880 outcomes and we felt that for the range of
312
doses wc were likely to consider, 4-break sites would not be frequent enough to warrant the effort. We therefore merge all Poisson classes higher than 3 into thc 3-class, applying its coefficients to this merged frequency. This naturally limits the range of Poisson means we can cover, since the accuracy of partition will decline as thc higher classes increase. Total aberration frequencies havc been computed for of Poisson means up to 2.0, and expressed in the form of Revell ratios. Fig. 3 shows a typical result for a particular set of conditions using Case 2, The conditions were chosen so that the ratios coincide fairly well with experimental observation when the Poisson mean is 1.0 (Table 1). These were (a) that the probability of restitution or rejoining is (I.75, and is unaffected by competition; (b) that there is a dominance of (a bias for) intra-chromatid, as opposcd to interchromatid events. 0.7:0.3. This bias is equivalent to an increase in W (Fig. 1 ), but its application in a contingency situation is more complex, requiring conditional probabilities to operate in 3-break sites.
In Fig. 3A, the Poisson 1-class ( - isolated single breaks) contributes to the final break score and, as expcctcd, the colour-jump break frequency (c*, R III) ~ 0 as thc mean falls. Fig. 3B is one step towards a Revell situation. There is n o contribution from unrestituted isolated single breaks. The ratio values obtained arc much morc in linc with observation if this condition obtains (Table 1). R 111 is now less affected by Poisson mean, and certainly does not approach zero as this falls. In fact, it will reach the values for a pure 2-class at the limit. Obviously these curves are just two of an infinitudc of such curve-sets satisfying all possible conditions.
Dose-response experiment One obvious test. therefore, betwccn the two theories, is to examine the change in Ratio II1 with dose. Such a test was originally suggested by Heddle ct al. (1969), but as far as we arc awarc. not performed.
TABLE 1 V79-4. O B S E R V E D R E V E I . L R A T I O S F R O M G2 O R S D A T A S U M M E D O V E R S A M P L E T I M E l)ose (Gy)
RI
RII
Rill
RIV
(I.75
02
28/13 (2.15)
41/24 (I.71)
4/28 ((I.141
4/13 ((I.31)
I.II
G2
3811/48 (7.92)
2611/91 (2.8f~)
40/380 (11.11 )
411/48 (0.831
S
¢~8/6 (11.33)
58/12 (4.83)
~/~)8 (11.(19}
i~/~ (I .(1111
(}2
459/73 (6.201
354/144 (2.461
77/459 ((I. 17 )
77/73 ( 1.115I
S
262/59 (4.44)
217/130 (1.67)
38/262 (11.15)
38/59 (11.64)
1 197/1 qq (6.02)
930/401 (2.32)
165/1 197 (0.141
165/199 (0.831
Theoretical adjusted Revell ratio ~' 5.59
3.(16
0.18
1.0
Prediction from Classic C'ase 2 "
2.31
(I.14
I.(19
1.5
Total
7.77
In both theoretical predictions, an intra-chromatid event bias of 0.7 has been assumed. For Case 2 this bias has been adjusted in the 3 + class by conditional probability considerations.
313
Because, at very low doses, a very large number of chromosomes need to be scanned to find sufficient breaks, we changed to another V79 line (379A) and developed a shake-off technique. By using a modified hypotonic treatment, we were able to obtain a large number of broken metaphases and loose chromosomes on each slide. A range of X-radiation doses up to 2.0 Gy was given to cultures that had been exposed to 10 /zg/ml BrdU for 17 h. Only one sample block was used, 1.5-3 h colcemid, and the slides were harlequin-stained and coded. Scoring continued until at least 200 chromatid breaks (terminal deletions) had been analyzed at each dose. Whilst searching for breaks, all achromatic lesions ('gaps') and all chromatid interchanges seen were also fully analyzed. Great care was taken in defining a true discontinuity (as far as this is possible with the light microscope - - Brecher, 1977) and a 'buffer' category of gap/break was provided for equivocal cases. This buffer was merged with the gap score for plotting. Breaks and gaps were assigned to light (1, = BB) or dark (d, = T B ) or as having a colour-jump at the apparent point of breakage (give or take a few
million base pairs!). Interchanges were classified for symmetry, completeness and presumptive breakpoints assigned 1/1, d / d , l / d . Tables 2 and 3 and Figs. 4 and 5 give the results. The overall ratio 111 is only about 12%, which is rather low for a simple Revell situation (though in good agreement with earlier findings for this material), but it does not change with dose (Table 2, Fig. 3). Moreover, surprisingly, the control breaks ('spontaneous' - - probably mainly BrdU-induced) show the same coiour-jump frequency. A similar picture emerged when we plotted data from the previous ratio experiment (Table 2, Fig. 4). The other surprise comes when we look at the allocation of breaks to 1 (BB) and d (TB) chromatids (Fig. 5). With the exception of the controis, there is no evidence of differential sensitivity for either breaks or gaps. This is in line with previous observations (Wolff and Bodycote, 1975). However, there is clear evidence of increased BB sensitivity for the lesions involved in interchanges, by a factor of at least two. We added, for other purposes, a "Iq'/TB protocol (8.5 h 10 /zg/ml BrdU then 10 /zg/ml
TABLE 2 ALLOCATION OF X-RAY-PRODUCED CHROMATID TERMINAL DELETIONS TO LIGHT (I, BB) OR DARK (d, TB) CHROMATIN AND THE FREQUENCY SHOWING COLOUR-JUMPS (c*, Rill) Material
Dose (Gy)
I
d
RIll
1/d
V79-379A T B / B B
0 0.3 0.6 0.9 1.2 1.5 2.0
30 107 112 236 106 211 129
17 109 116 214 117 219 138
8 34 23 50 30 72 36
0.145 0.136 0.092 0.100 0.119 0.143 0.119
1.765 0.982 0.966 1.103 0.906 0.963 0.935
901
913
245
0. I 19
0.987
202 143
153 120
50 40
0.123 0.132
1.320 1.192
345
273
99
0.127
1.264
29 89 219 304
44 105 196 309
12 25 47 118
0.141 0.114 0.102 0.161
0.659 0.848 1.117 0.984
612
610
190
0.135
1.003
Totals excluding control T'f/TB
0.9 1.5
Total V79-4 TB/BB
Totals excluding control
0 0.75 1.0 1.5
c*
314
CASE 2 Break contribution from all classes
A
B
No break contribution from 1-class
2.0
(Classic theory) 'RI
0.6
24
=
.o_
R II ]clent~cal
O
3.0
> o
=
.O
for A & B
IT
rl-
R IV identical
. . . .
"1o~
......
R
~%-
-
1.0
-
12 °.....
0.2
1.0
FI III f
I
0.4
1.2
2.0
0.4
Poisson mean
1.2
Poisson
2.0
mean
Fig. 3. Expected changes in Revell ratios as Poisson mean ( ~ dose) is varied. Model 2, free interaction without intrinsic restitution bias, is used. (A) Simple contingency derivation based on Classic theory. All Pois~m class breaks contribute to final score. Probability of rejoining is 0.75, and there is an intra-/inter-arm event bias of 0.7 (with conditional probability adjustment for the 3 + class) Note that R Iii tends to 0 as dose falls. (B) Identical to A, but there is no contribution to breaks from Poisson class 1 (restitution of isolated single breaks = I(~)~,). R Ill is much less affected by dose, and does not tend to 0 as dose is lowered.
TABLE 3 I,ESION A L L O C A T I O N TO L I G | IT AND D A R K C H R O M A T I D S Material
V79-379A T B / B B
Dose
Achromatic lesions (gaps)
Chromatid interchanges
(Gy)
1
I/1
0 0.3 ().6 0.9 1.2 1.5 2.0
Totals excluding control "Iq-/TB
0.9 1.5
Total V79-4 T B / B B
Totals excluding control
0 0.75 1.0 1.5
d
freq. of l
I/d
d/d
freq. of I
87 201 179 350 131 177 113
13 17(,~ 176 324 130 204 165
6.692 1.123 1.017 1.080 1.008 0.868 (I.685
17 62 56 83 51 69 41
7 33 42 47 33 37 38
I I1 8 27 12 21 20
4.556 2.855 2.655 2.109 2.368 2.215 1.538
1 151
1 178
0.977
362
23{1
99
2.229
587 288
625 329
0.939 0.875
76 52
62 43
44 24
1.427 1.615
875
954
0.917
128
105
68
1.498
51 390 597
113 590 827
0.451 0.661 0.722
17 35 5() 114
3 37 49 149
2 18 21 47
5.286 1.466 1.637 1.55 I
1038
1 53()
0.678
199
235
86
1.555
315 Frequency
entiai sensitivity for interchange lesions. This suggests that colour-jumps are a characteristic of aberration formation and not something produced by BrdU incorporation.
0.20, 0.16 .......
~., t ~
........ ~ t .
•
.J=
~...
Discussion
vT -3,gArB
0.06
0.041 0
V79-4 TB/BB
0
0.'3
016
0',
1'2
t5
2.0
Dose (Gy)
Fig. 4. The frequency of colour-jump breaks (c*, ratio ]11) does not fall to zero as dose (and the absolute frequency) falls but remains about 12% even in unirradiated controls. Similar values are found for both T B / B B and T T / T B in spite of differences expected in strand sensitization. The inference is that single isolated breaks do not make a contribution to the final break score (cf. Fig. 3).
thymidine until radiation and fixation) at two doses in the dose-response curve. The observed frequencies are given in Tables 2 and 3 and have been added to the figures. It will be seen that the same frequency of colour-jump breaks (R III) is found, but there is a marked reduction in differFrequency 8-
--o- Breaks lid TB/t3B ..... ] -l-
L
6~ \ \
0
,
--- Exchanges lid TB/BB I
4"
0 ~
Breaks I/d TI"/IB
~-- Gaps I/d TB/BB
I-
0.3
0.6
0.9
1.2
1.5
~
2.0
Dose (Gy)
Fig. 5. Comparing the frequencies (light/dark) with which breaks are allocated to different chromatids. There is absolutely no evidence of greater sensitization of BB for either breaks or gaps, except in unirradiated controls (where most breaks result from BrdU incorporation?), In contrast, there is a marked sensitivity difference for breaks derived from chromatid interchanges. According to Classic theory, all breaks come from a common pool and should show the same degree of sensitization.
The first, and most obvious point to be made is that on at least two counts, the aberrations which we are scoring as 'breaks' (chromatid terminal deletions, discontinuities) are not behaving in a way consistent with Classic theory, as originally defined. Over a wide range of dose and absolute frequency, and even into the unirradiated control ('spontaneous', probably BrdU-induced), the frequency of coiour-jump breaks (R III) remains more or less constant at around 12% in V79-379A for both T B / B B and (at two doses) " I ~ / T B chromatin substitution. Such an observation would be expected on Revell theory, but only on Classic theory if there was no (or negligible) contribution from unrestituted single breaks (Fig. 3). An effectively constant proportion of the break/lesion pool at each dose shows the jump, regardless of the substitution status used to distinguish sister-chromatids. This suggests a link with aberration formation rather than an artefact of, say, BrdU incorporation. It also rules out incomplete SCE as a contaminating source (i.e. SCE as defined: S-dependent and unrelated to structural aberrations) since S e E , in contrast to breaks, do not increase with dose in G2 and hardly increase in late-S cells. In any case, incompleteness in SCE is a very rare event. We have therefore to accept that either there are no such things as single breaks (which is Revells original contention) or that, under normal circumstances, restitution is so complete that no significant contribution to breaks arises from this source (which, for practical purposes, brings us almost to a Revell situation). It is interesting that successful simulation from Classic ideas (with Case model 2) of observed Revell ratios also requires no contribution from Poisson class 1. We said above 'almost to a Revell situation', because although Case 2 as applied derives all breaks from intrachanges, failed resti-
316
tution, as an option, is still allowed within a competition situation of 2 or more breaks in a site. Even this option can be removed to give a pure Revell simulation, but this is best done from a Case 1 model which we are not discussing here. Suffice it to say that so far, we have not been able to produce such good fits to observed ratios with this approach. The other interesting thing that emerges from the observed ratios and all our attempts to model them, is the requirement for the dominance of intra-chromatid events ( W = 0.7). Without this constraint, it is not possible to achieve reasonable ratio values (considered as a block). Either intrachromatid lesion interactions are favoured, or intra-chromatid strand proximity is predominant within sites. During transit of late-S and G2, chromatin is actively condensing to form two spatially discrete sister-chromatids, visualization of which signals the onset of prophase. It is inevitable that such a process will tend to favour, progressively, the formation of intra-chromatid sites at the expense of inter-chromatid ones, lowering the colour-jump probability and enhancing minute frequency. If this explanation is plausible, we might anticipate that R I I I would increase as we move earlier in S. Heddle and Bodycote (197(I) found such an increase, Wolff and Bodycotc (1975) did not. We did not look at early S cells, but it is a point worth checking, always bearing in mind the very complex proximity situations that are likely to obtain at that time arising from the very staggered and scattered programme of replication ( A g h a m o h a m m a d i and Savage, 1990). The second count on which breaks do not conform is with respect to BrdU sensitization. In line with the findings of others (e.g. Wolff and Bodycote, 1 9 7 5 ) w e can confirm that in V79379A, there is no evidence of excess BB sensitivity for breaks. We also show that there is none for achromatic lesions (gaps). A slight, but non significant increase may exist in V79-4. Controls were an exception, probably because most of the spontaneous breaks are BrdU-induced. In sharp contrast, lesions/breaks involved in interchanges (inter-arm intrachanges and triradials are not included in the analysis) show at least a two-fold increase in a T B / B B situation (slightly less in V79-4).
Now, according to Classic theory, both types of aberration use the same pool of breaks. A random (?) sample from the ptx)l arc fixed by rejoins (exchanges which include both inter- and intrachanges, Lea, 1046) and the remainder restitute (randomly?) leaving a residual of open simple breaks, if the conditions of randomness are fulfilled, then it is inescapable that all three categories should show the same BB bias. They do not, and that raises some difficult questions. ls there a common pool or two different pools'? Or, put in another way, are intrachange sites, or modes of rejoining, or kinds of lesion in them, different from interchange ones? What differences could promote a sensitivity contrast? It seems unlikely that there is a fundamental difference between the lesions. A small percentage (2-3) of interchanges are 'mixed', in the sense that one (very, occasionally, both) of the participating lesions involves a colour-jump. The simple explanation would be that such aberrations result from the interaction of a single lesion plus an intrachange. These events belong to a family of such compound interactions of which the triradial is another member. Alternatively, the explanation could lie in a differential repair or restitution favouring BrdUenhanced breakage. One could envisage a situation where interchanges are established very rapidly after lesion induction, but that intrachanges come later after a preferential removal of the sensitized break component. We are not aware of evidence for such preferential repair nor for a formation-time separation for these two aberration categories. The close similarity of gaps and breaks despite big frequency differences, is also interesting, and might indicate a closer relationship between them than we have, perhaps, allowed (cf. Brecher, 1977). To these questions (and others which will occur to thoughtful readers) we do not have ready answers, but they suggest that aberration production (as opposed to reduction; Savage, 1990) is probably more complicated than either the early workers or even we, with our vast increase in knowledge, have imagined. During the extensive scoring of breaks for this experiment we have received many indicators that this is so. How does
317
one handle morphologically 'unambiguous' la intrachanges that have no colour-jump or clear 2a that do? So, at least we can anticipate that there is still good experimental mileage on this road of structural chromosomal aberrations likely to provide many papers for Mutation Research in the years ahead.
Acknowledgements Our thanks go to David Papworth for much help and advice in probability theory and to the Unit photographers, K. Glover and E. Evans who helped prepare the figures. This work is supported in part by CEC contract Bi7-039.
References Aghamohammadi, S.Z., and J.R.K.Savage (1990) BrdU pulse/reverse staining protocols for investigating chromosome replication, Chromosoma, 99, 76-82. Brecher, S. (1977) UItrastructural observations of X-ray induced chromatid gaps, Mutation Res., 42, 249-268. Evans, H.J. (1962) Chromosome aberrations induced b~, ionizing radiations, Int. Rev. Cytol., 13, 221-321. Fox, D.P. (1967) The effects of X-rays on the chromosomes of locust embryos, 111. The chromatid aberration types, Chromosoma, 20, 386-412. Heddle, J.A., and D.J. Bodycote (1970) On the formation of chromosomal aberrations, Mutation Res., 9, 117-126. Heddle, J.A., D. Whissell and D.J. Bodycote (1969) Changes in chromosome structure induced by radiations: A test of the two chief hypotheses, Nature (London), 221, 11581160. Lea, D.E. (1946) Actions of radiations on living cells, 1st edn., Cambridge University Press.
Lea, D.E., and D.G. Catcheside (1942) The mechanism of the induction by radiation of chromosome aberrations in Tradescantia, J. Genet., 44, 216-245. Revell, S.H. (1959) The accurate estimation of chromatid breakage and its relevance to a new interpretation of chromatid aberrations induced by ionizing radiations, Proc. Roy. Soc. (London), Ser. B, 150, 563-589. Revell, S.H. (1974) The breakage-and-reunion theory and the exchange theory for chromosomal aberrations induced by ionizing radiations: A short history, Adv. Rad. Biol., 4, 367-416. Savage, J.R.K. (1975) Radiation-induced chromosomal aberrations in the plant Tradescantia: Dose-response curves, 1. Preliminary considerations, Rad. Bot., 15, 87-14(I. Savage, J.R.K. (1976) Annotation: Classification and relationships of induced chromosomal structural changes, J. Med. Genet., 13, 103-122. Savage, J.R.K. (1986) The 'Revell Loop', Clin. Cytogenet. Bull., 1, 192-196. Savage, J.R.K. (1989) The production of chromosome structural changes by radiation: An update of Lea (1946), Chapter VI, Br. J. Radiol., 62, 507-520. Savage, J.R.K. (1990) Mechanisms of chromosome aberrations, in: M.L. Mendelsohn and R.J. Albertini (Eds.), Mutation and the Environment, B, Wiley-Liss, New York, pp. 385-396. Savage, J.R.K., R.J. Preston and G.J. Neary (1968) Chromatid aberrations in Tradescantia bracteata and a further test of Revell's hypothesis, Mutation Res., 5, 47-56. Sax, K. (1940) An analysis of X-ray induced chromosomal aberrations in Tradescantia, Genetics, 25, 41-68. Wolff, S., and J. tkxtycote (1975) The induction of chromatid deletions in accord with the breakage-and-reunion hypothesis, Mutation Res., 29, 85-91. Wolff, S., and P. Perry (1974) Differential Giemsa staining of sister chromatids and the study of sister chromatid exchanges without autoradiography, Chromosoma, 48, 341353.
307
MUT 02540
Revell revisited J o h n R . K . S a v a g e a n d A l i s o n N. H a r v e y MRC Radiobiology Unit, Chilton, Didcot, OXI l ORD (U.K.) (Accepted 9 May 1991)
Keywords: Chromatid breaks; Harlequin; Revell ratios
Summary The controversy of the Classic versus the Exchange theories for the origin of simple chromatid breaks is outlined. Using BrdU harlequin sister-chromatid differentiation four Revell ratios can be defined and these have been obtained and tested as a block in V79 hamster cells. The values are quite different from the simple predictions. However, values similar to those observed (taken as a block) can be readily simulated from Revell theory by making the assumption that intra-chromatid events are dominant (0.7:0.3). They can also be obtained from contingency modelling of Classic theory using the same assumption plus the additional constraint that there is no contribution from isolated single lesions (Poisson class 1). If this latter assumption is correct, then the frequency of breaks involving a colour-jump (ratio III) should not decline to zero as the dose fails. A dose-response experiment shows that it does not, but remains approximately constant at about 12%, even in the unirradiated control. An added complication arises when we discover for the TB/BB situation, that whilst neither breaks nor gaps show any excess BB involvement (sensitisation), lesions involved in interchanges show at least a 2-fold BB excess. Clearly, the chromatid discontinuities we are scoring are not behaving as would be expected of a residue of unrejoined primary breaks (Classic theory) and we infer also that they are not 'simple'.
It is a great pleasure and honour to make a contribution to the 250th volume of Mutation Research. The journal has spanned a large part of my research career and played an invaluable role for both input and output. I offer my sincere congratulations and thanks to Prof. Sobels and his team for the excellent steering that has enabled the journal to achieve this milestone (JRKS).
Correspondence: Dr. John R.K. Savage, MRC Radiobiology Unit, Chilton, Didcot, Oxon. O X l l 0RD (U.K.), Tel. Abingdon (0235) 834393, Fax (0235) 834918.
The journal was launched (1964) at the time of a lively (and often stormy) controversy over the nature and origin of chromatid 'breaks'. The 'Classic' breakage-and-reunion theory (developed by Sax, 1940 and mathematically formalised by Lea and Catcheside, 1942 and Lea, 1946) which lay at the foundation of all aberration thinking, had been seriously challenged by the 'Exchange' hypothesis (Revell, 1959). Mutation Research became an ideal forum for this debate and many of the important papers on this topic have been published in it over the years. Today, in the light of our much increased knowledge of chromosome structure and the
3(18
molecular events that lead to aberrations, some of the early concepts have to be modified (Savagc, 1989, 1990). However, the original breakage-andreunion idea still dominates and is now almost universally accepted, sincc the bulk of the accumulated evidence seems to be in its favour. We thought that it might be timely to air the controversy again, and present some new data which suggests that things may not be as simple as most people might think or wish. The controversy This p a p e r is in no sense a critical review of the topic and the information given in this section is only to provide a basic understanding; those wishing more must consult original papers and discussions (e.g. Lea, 1946; Revell, 1959, 1974; Evans, 1962; Savage, 1975, 1976, 1989). According to 'Classic" theory, the primary events of radiation damage at the chromosome level are breaks in the chromosome continuity thread ( ' c h r o m o n e m a ' - - it must be r e m e m b e r e d that these early workers had a very different picture of chromosome structurc from that current today). These breaks have one of three fates; (a) to restitute i.c. to join back to the original configuration and effectively disappear, (b) to rejoin or rearrange i.e. to join up with another broken end close in space and time and so form an (visible or invisible) aberration, (c) to remain open and appear at metaphase as a simple break at some point in the chromosome. Mathematical considerations and d o s e - r e sponse curve observations require that the initial number of primary breaks greatly exceed those detected: ' O f all the breaks primarily p r o d u c e d . . , only a fraction, and probably a small fraction, survive to the time of fixation when they are observed either as simple breaks or in structural rearrangements' (Lea, 1946, p. 253). ' T h e chromosome breaks which lead to fusions between chromosome arms must constitute only a very small proportion of the total breaks i n d u c e d . . . ' (Sax, 1940, p. 58). So, we have the picture of a large pool of breaks, a small fraction of which are removed by rejoining to form aberrations and the remainder
restitute except for a small residue of (unrejoincd or unrejoinable) simple breaks. Thus, everything starts from a break. In 1959, Revell published a radical challenge to this concept. He proposed that the primary event was not a break (i.c. no loosc joinablc cnds), but an unstable lesion which, left to itself, decays (i.e. there is no such thing as 'restitution" in the Classic sensc).Thesc lesions could, however, interact with one another and form cxchangcs ( - Classic 'rejoins'). The exchange proccss somctimes fails, lcaving unrcjoined ends, a fact that was well known from the observation of incomplete interchanges and non-union isochromatid breaks. Revell argued that in the case of intra-arm intrachanges, some of the incomplete forms, when seen in condenscd chromosomes. would look like (and gcnerally bc scored as) breaks ('discontinuities') - - but these breaks arc secondary, not primary. Since there arc no primary breaks, all discontinuities observcd must arise in this way. He re-classified the simple two-lesion chromatid intrachanges, demonstrated the forms of possiblc cxchangc using the now famous 'Revcll loop' t and how these would a p p e a r at metaphasc. He also proposed two numerical relationships that would bc expected if his hypothesis was correct, and provided confirmatory evidence from Vicia. He also pointed out that the required rarity of breaks had bccn ovcrlooked because a critical distinction had not been made between truc d i s c o n t i n u i t i e s and a c h r o m a t i c - l e s i o n s ('gaps'). Incidentally, this distinction, which is now made in all critical work, originated from this caution. Such a fundamental challenge produced a furore and a flurry of experiments to test the predictions (see Heddle ct al., 1969: Revell, 1974, for refs.). It soon became clear, certainly for
I Although it has come in for much criticism, the loop is not an integral part of Revell theory. It is purely a diagrammatic device to bring into proximity lesions separated longitudinally on a chromosome arm (Savage. 1986). That such lesions occur and that they interact, is inescapable -- the interstitial deletion ('minute') is an obvious witness. All forms of the loop can actually be seen by careful examination of a range of chromatid inter-arm intrachanges.
309
horizontal
"X"
vertical
(H)
"U"
(l-H)
.'g A
la
lb
4a
4b
C~
c*
NUp
NUd
2a
2b
3a
3b
m
c
c
I t-
"0
E
°9_ c+m
Ratio I
12.5]
( l a + l b * 2 a +3a+3b) (4a+4b)
Ratio II
[1.0]
(la+lb) Ratio III
Total 4
(la+lb) Ratio IV
{la+ lb+2a+3a+3b) [0.4]
Total 2
-l/Ratio I
(4a+4b) [I.0]
Fig. 1. The 8 incomplete 2-lesion chromatid intra-arm intrachanges upon which Revell based his theory. For diagrams and details see ReveU, 1959; Savage, 1986, 1989. The lower part of each box indicates how they are scored at metaphase: c, chromatid breaks without colour-jump; c*, break with colour-jump; m, single minute; NUp, NUd, non-union proximal or distal isochromatid breaks. The complete types are not shown but both 2 and 4 are used in ratio 11. If sister-chromatid differentiation is present, 4 ratios can be defined and their default values (all four categories equally likely) are shown in brackets. In practice various constraints operate to change relative frequencies and so modify ratios. The two most important are intra- versus inter-single chromatid events (W,(1- IV)) and ' X ' versus ' U ' type exchange (diagrammatically Horizontal versus Vertical, H,(1-H)). The effect of these two factors on the ratios are shown in Fig. 2.
mammalian cells, that the simple predictions (ratio I = 2.5; ratio II = 1; Fig. 1) were seldom realised, but there were big differences between species, cell type, growth conditions, etc. There were also a number of factors that could modify these ratios (e.g. Savage et al., 1968). Recognising that, on the basis of the Exchange hypothesis, 40% of breaks (la, lb Fig. 1) should be accompanied by an exchange between sister-
chromatids, Heddle et al. (1969) set out to test this, using second-cell division segregation of tritiated thymidine in Potorous to detect label switches. They found 38% of breaks showed them, and concluded that the majority of terminal deletions here were incomplete intrachanges. Later, in a repeat with Chinese hamster ceils (Heddle and Bodycote, 1970)only 15-18% (with evidence that S > G2) label-switch breaks were seen, which they took as being indicative of a mixture of simple and intrachange origins. With the advent of FPG harlequin staining (Wolff and Perry, 1974) which gives excellent differentiation (BrdU, TB/BB) between sisterchromatids and minimal ambiguity of sister-chromatid exchanges (colour-jumps), Wolff and Bodycote (1975) were able to perform a much cleaner test in CHO cells. Scoring over the first 8 h post-irradiation, when a changing mixture of G2 and S cells were present, and with rigorous classification of break categories, they found about 8% of breaks had a colour-jump (only 3% at the most sensitive sample time) and concluded 'that very few of the breaks are incomplete exchanges'. Classic theory does not deny the existence of breaks arising from incomplete intrachanges (they enter into some of Lea's calculations) but it insists that contribution from this source is negligible when compared with those arising from failed restitution of single breaks. In contrast, pure Exchange theory insists that there is n o contribution from single lesions. The Reveli ratios
Using the eight incomplete forms of simple, two-lesion chromatid intra-arm intrachanges 2, we can, with the added dimension of colour-jump breaks (c*) define four Revell ratios (Fig. 1). We
2 There is a range of more complex 3- and 4-lesion intrachanges, first described in full by Fox (1967) under the name 'insertion-intrachanges'. Many of these simulate simple ones when seen at metaphase, but some characteristic ones are always observed. They are seldom considered, but they may be much more common than we think, especially after high-LET radiations or chemical clastogens. These complex forms will not, of course, obey ordinary Revell rules and, if common, will warp observed ratios.
310
and 11 in an identical manner, effects on ratios 11I and IV are quite different.
30
20
A ratio test 0
O
n-
tr"
10
iI / 1 L0
W orH
g 8.0
0.8
,9~,
/
>
0
O .m
IT
i'r
0
Fig. 2. If the t a b l e o f Fig. 1 is s u b j e c t e d to v a r i o u s biases, the r e s u l t i n g Revell ratios d e p a r t d r a m a t i c a l l y f r o m the s i m p l e p r e d i c t i o n s . T w o f a c t o r s t h a t a r e k n o w n to o c c u r a r c the d o m i n a n c e , o r o t h e r w i s e , o f i n t r a - c h r o m a t i d as o p p o s e d to
inter-chromatid exchange events (factor W) and the dominance of "U' over'X" type exchanges -- referred to diagrammatically as ltorizontal versus Vertical exchange (factor H) (Savage, 1986, 1989). Breaks deriving from intrachangcs are Ix~und to be influenced by alterations in these factors.
note that these ratios arc inter-related and that their values, considered as a block, can bc changed by a number of factors. The effects of two of these factors are shown in Fig. 2. Others include incompleteness (a and b forms may not be equal) and, of course, in a BrdU T B / B B context, differential chromatid sensitivity. Equivocal results exist for this last factor (see later) but if present, its main effect will bc to boost intra-arm events (i.e. to increase W, Fig. 1). Because factor variation changes all these ratios in concert, it is important in any tests using this approach, to consider them all together and not in isolation. For example, in Fig. 2 it can bc secn that whilst changes in W or H affect ratios I
Chinese hamster cells (V79-4) were grown fl~r 17 h in MEM containing 10 / z g / m l BrdU. They were then irradiated with various doses of X-rays (250 kVp, 14 mA). Immediately after radiation, the medium was replaced with one containing 10 / z g / m l thymidine. Metaphases were sampled at 1.5 h intervals up to 7.5 h (1.5 h colcemid) and chromosomes stained by a FPG method to distinguish "FF, TB and BB chromatin. Second division cells in the first four samples were scored for all categories of chromatid-type aberrations (Savage, 1976). rl"T patches allowed unambiguous distinction between S and G2 cells. Table 1 shows the ratios we observed. There is a fair scatter, but no systematic trends. We did not find any obvious trends with sample time either, even though we would be looking at different cell mixes. The S cells nearly all had small "IYI"patches, signifying that they were mostly lateS, so it may not be surprising that the marked differences others have found for R l I I , have not been confirmed. In general, compared with the default values, R 1 and R II are bigger, R Ill smallcr and R IV slightly smaller. Inspection of Fig. 2 shows such changes to be characteristic of an increase in W, i.e. that intrachromatid events are more frequent than interchromatid ones. Changing W to 0.75, and multiplying out the Revell table of Fig. 1 gives a fair approximation of observation. Because this table deals with whole aberrations and not the interactions of individual lesions or breaks, simulations are inevitably limited. If we adopt a more realistic contingency approach (discussed in a little more detail below) an even closer approximation can be obtained. Interestingly, the same assumption about the dominance of intra-chromatid events has to be made, and, in addition, we have to eliminate any contribution from the Poisson 1class (i.e. no unrestituted isolated breaks). Of course, whilst this demonstrates that departures from Revell's original predictions do not necessarily negate his theory (which is a conclusion many have been quick to draw in the past)
311
there is an inherent danger in this approach. Having so many variables to play with, a little calculator juggling will usually come up with a tolerable fit to most data sets! However, there is a safeguard if one insists on considering all four ratios as a block that stands or falls together, for then, the number of (biologically reasonable) factor variations becomes quite limited.
Classical predictions Apart from within-experiment estimates (Heddie et al., 1969, 1970; Wolff and Bodycote, 1975) no one seems to have computed the frequency of colour-jump breaks expected on the basis of Classic theory, nor to have predicted intrachange ratios to compare with Revell frequencies. Such calculations are necessarily limited and tedious, but not excessively difficult. On the basic premise that primary breaks will be distributed at random (Lea, 1946), the frequency of 'sites' 3 with 0, 1, 2 . . . breaks will conform to the terms of a Poisson. Two or more breaks may interact and produce a limited variety of intrachange types, conditioned by the disposition of the breaks, within or between chromatids in the site. The random (?) subset of sites which will lead to interchanges are not included in these calculations as they are not expected to contribute to simple breaks. We need then (a) to consider, for each break, and each combination of 2, 3 . . . breaks every possible outcome, and, using assigned probabilities, compute its expected frequency, (b) to assess how this outcome will (or will not) appear at metaphase, and (c) how it will most likely be scored. Then, by summations, we can derive for each relevant aberration category (e.g. break, break with colour-jump, minute, etc.) a partition coefficient. These coefficients can then be used to multiply the appropriate Poisson class to provide its expected contribution to each aberration type. Fur-
3 'Site': W e d o not wish h e r e to m a k e a f o r m a l d e f i n i t i o n - t a k e it as a v o l u m e in which any n u m b e r of b r e a k s can o c c u r a n d in w h i c h they can interact in all possible ways to produce structural intrachanges.
ther summation for the whole Poisson gives the complete aberration spectrum for the 'dose'. Clearly, the zero class will contribute nothing and the one class has only a single possible outcome, a simple break. Two or more breaks can be distributed between or within chromatids and produce the whole range of complete and incomplete intrachanges (including for the three and above the more complex ones not considered in this analysis). It is obvious that the relative contribution to each category is going to change with dose. In particular, the break contribution from incomplete intrachanges (the only source of colour-jump breaks) will diminish as dose decreases, because l-break sites will become dominant as the Poisson mean falls. The principal difficulty in compiling the contingency tables and in deriving the partition coefficients, lies in deciding the probability conditions which underlie Classic theory. Equivocal and conflicting answers to a number of important questions arise when early literature is studied. For example, it is agreed that the majority of breaks restitute and do not form aberrations. Is this because restitution is intrinsically favoured (i.e. must we assign a higher probability to the restitution option, and that in all circumstances?) or, is it because breaks not 'fixed' by inter- or intrachange have nothing else left to do but restitute? (i.e. given two or more breaks, restitution becomes just one competing option.) It is necessary, then, to explore several probability models. We have considered just two cases, with their extensions: Case 1. The probability of restitution is intrinsically dominant under all conditions (i.e. in competition, return to the original configuration is always the most likely result). Case 2. The probability of restitution is not dominant in an interaction situation (i.e. free competition, on an equal basis, for broken ends within a site). Only Case 2 will concern us in this paper. For both models, we have derived partition coefficients for 2- (48 outcomes) and 3- (960 outcomes) break sites. The full contingency table for 4-break sites would require the evaluation of 26 880 outcomes and we felt that for the range of
312
doses wc were likely to consider, 4-break sites would not be frequent enough to warrant the effort. We therefore merge all Poisson classes higher than 3 into thc 3-class, applying its coefficients to this merged frequency. This naturally limits the range of Poisson means we can cover, since the accuracy of partition will decline as thc higher classes increase. Total aberration frequencies havc been computed for of Poisson means up to 2.0, and expressed in the form of Revell ratios. Fig. 3 shows a typical result for a particular set of conditions using Case 2, The conditions were chosen so that the ratios coincide fairly well with experimental observation when the Poisson mean is 1.0 (Table 1). These were (a) that the probability of restitution or rejoining is (I.75, and is unaffected by competition; (b) that there is a dominance of (a bias for) intra-chromatid, as opposcd to interchromatid events. 0.7:0.3. This bias is equivalent to an increase in W (Fig. 1 ), but its application in a contingency situation is more complex, requiring conditional probabilities to operate in 3-break sites.
In Fig. 3A, the Poisson 1-class ( - isolated single breaks) contributes to the final break score and, as expcctcd, the colour-jump break frequency (c*, R III) ~ 0 as thc mean falls. Fig. 3B is one step towards a Revell situation. There is n o contribution from unrestituted isolated single breaks. The ratio values obtained arc much morc in linc with observation if this condition obtains (Table 1). R 111 is now less affected by Poisson mean, and certainly does not approach zero as this falls. In fact, it will reach the values for a pure 2-class at the limit. Obviously these curves are just two of an infinitudc of such curve-sets satisfying all possible conditions.
Dose-response experiment One obvious test. therefore, betwccn the two theories, is to examine the change in Ratio II1 with dose. Such a test was originally suggested by Heddle ct al. (1969), but as far as we arc awarc. not performed.
TABLE 1 V79-4. O B S E R V E D R E V E I . L R A T I O S F R O M G2 O R S D A T A S U M M E D O V E R S A M P L E T I M E l)ose (Gy)
RI
RII
Rill
RIV
(I.75
02
28/13 (2.15)
41/24 (I.71)
4/28 ((I.141
4/13 ((I.31)
I.II
G2
3811/48 (7.92)
2611/91 (2.8f~)
40/380 (11.11 )
411/48 (0.831
S
¢~8/6 (11.33)
58/12 (4.83)
~/~)8 (11.(19}
i~/~ (I .(1111
(}2
459/73 (6.201
354/144 (2.461
77/459 ((I. 17 )
77/73 ( 1.115I
S
262/59 (4.44)
217/130 (1.67)
38/262 (11.15)
38/59 (11.64)
1 197/1 qq (6.02)
930/401 (2.32)
165/1 197 (0.141
165/199 (0.831
Theoretical adjusted Revell ratio ~' 5.59
3.(16
0.18
1.0
Prediction from Classic C'ase 2 "
2.31
(I.14
I.(19
1.5
Total
7.77
In both theoretical predictions, an intra-chromatid event bias of 0.7 has been assumed. For Case 2 this bias has been adjusted in the 3 + class by conditional probability considerations.
313
Because, at very low doses, a very large number of chromosomes need to be scanned to find sufficient breaks, we changed to another V79 line (379A) and developed a shake-off technique. By using a modified hypotonic treatment, we were able to obtain a large number of broken metaphases and loose chromosomes on each slide. A range of X-radiation doses up to 2.0 Gy was given to cultures that had been exposed to 10 /zg/ml BrdU for 17 h. Only one sample block was used, 1.5-3 h colcemid, and the slides were harlequin-stained and coded. Scoring continued until at least 200 chromatid breaks (terminal deletions) had been analyzed at each dose. Whilst searching for breaks, all achromatic lesions ('gaps') and all chromatid interchanges seen were also fully analyzed. Great care was taken in defining a true discontinuity (as far as this is possible with the light microscope - - Brecher, 1977) and a 'buffer' category of gap/break was provided for equivocal cases. This buffer was merged with the gap score for plotting. Breaks and gaps were assigned to light (1, = BB) or dark (d, = T B ) or as having a colour-jump at the apparent point of breakage (give or take a few
million base pairs!). Interchanges were classified for symmetry, completeness and presumptive breakpoints assigned 1/1, d / d , l / d . Tables 2 and 3 and Figs. 4 and 5 give the results. The overall ratio 111 is only about 12%, which is rather low for a simple Revell situation (though in good agreement with earlier findings for this material), but it does not change with dose (Table 2, Fig. 3). Moreover, surprisingly, the control breaks ('spontaneous' - - probably mainly BrdU-induced) show the same coiour-jump frequency. A similar picture emerged when we plotted data from the previous ratio experiment (Table 2, Fig. 4). The other surprise comes when we look at the allocation of breaks to 1 (BB) and d (TB) chromatids (Fig. 5). With the exception of the controis, there is no evidence of differential sensitivity for either breaks or gaps. This is in line with previous observations (Wolff and Bodycote, 1975). However, there is clear evidence of increased BB sensitivity for the lesions involved in interchanges, by a factor of at least two. We added, for other purposes, a "Iq'/TB protocol (8.5 h 10 /zg/ml BrdU then 10 /zg/ml
TABLE 2 ALLOCATION OF X-RAY-PRODUCED CHROMATID TERMINAL DELETIONS TO LIGHT (I, BB) OR DARK (d, TB) CHROMATIN AND THE FREQUENCY SHOWING COLOUR-JUMPS (c*, Rill) Material
Dose (Gy)
I
d
RIll
1/d
V79-379A T B / B B
0 0.3 0.6 0.9 1.2 1.5 2.0
30 107 112 236 106 211 129
17 109 116 214 117 219 138
8 34 23 50 30 72 36
0.145 0.136 0.092 0.100 0.119 0.143 0.119
1.765 0.982 0.966 1.103 0.906 0.963 0.935
901
913
245
0. I 19
0.987
202 143
153 120
50 40
0.123 0.132
1.320 1.192
345
273
99
0.127
1.264
29 89 219 304
44 105 196 309
12 25 47 118
0.141 0.114 0.102 0.161
0.659 0.848 1.117 0.984
612
610
190
0.135
1.003
Totals excluding control T'f/TB
0.9 1.5
Total V79-4 TB/BB
Totals excluding control
0 0.75 1.0 1.5
c*
314
CASE 2 Break contribution from all classes
A
B
No break contribution from 1-class
2.0
(Classic theory) 'RI
0.6
24
=
.o_
R II ]clent~cal
O
3.0
> o
=
.O
for A & B
IT
rl-
R IV identical
. . . .
"1o~
......
R
~%-
-
1.0
-
12 °.....
0.2
1.0
FI III f
I
0.4
1.2
2.0
0.4
Poisson mean
1.2
Poisson
2.0
mean
Fig. 3. Expected changes in Revell ratios as Poisson mean ( ~ dose) is varied. Model 2, free interaction without intrinsic restitution bias, is used. (A) Simple contingency derivation based on Classic theory. All Pois~m class breaks contribute to final score. Probability of rejoining is 0.75, and there is an intra-/inter-arm event bias of 0.7 (with conditional probability adjustment for the 3 + class) Note that R Iii tends to 0 as dose falls. (B) Identical to A, but there is no contribution to breaks from Poisson class 1 (restitution of isolated single breaks = I(~)~,). R Ill is much less affected by dose, and does not tend to 0 as dose is lowered.
TABLE 3 I,ESION A L L O C A T I O N TO L I G | IT AND D A R K C H R O M A T I D S Material
V79-379A T B / B B
Dose
Achromatic lesions (gaps)
Chromatid interchanges
(Gy)
1
I/1
0 0.3 ().6 0.9 1.2 1.5 2.0
Totals excluding control "Iq-/TB
0.9 1.5
Total V79-4 T B / B B
Totals excluding control
0 0.75 1.0 1.5
d
freq. of l
I/d
d/d
freq. of I
87 201 179 350 131 177 113
13 17(,~ 176 324 130 204 165
6.692 1.123 1.017 1.080 1.008 0.868 (I.685
17 62 56 83 51 69 41
7 33 42 47 33 37 38
I I1 8 27 12 21 20
4.556 2.855 2.655 2.109 2.368 2.215 1.538
1 151
1 178
0.977
362
23{1
99
2.229
587 288
625 329
0.939 0.875
76 52
62 43
44 24
1.427 1.615
875
954
0.917
128
105
68
1.498
51 390 597
113 590 827
0.451 0.661 0.722
17 35 5() 114
3 37 49 149
2 18 21 47
5.286 1.466 1.637 1.55 I
1038
1 53()
0.678
199
235
86
1.555
315 Frequency
entiai sensitivity for interchange lesions. This suggests that colour-jumps are a characteristic of aberration formation and not something produced by BrdU incorporation.
0.20, 0.16 .......
~., t ~
........ ~ t .
•
.J=
~...
Discussion
vT -3,gArB
0.06
0.041 0
V79-4 TB/BB
0
0.'3
016
0',
1'2
t5
2.0
Dose (Gy)
Fig. 4. The frequency of colour-jump breaks (c*, ratio ]11) does not fall to zero as dose (and the absolute frequency) falls but remains about 12% even in unirradiated controls. Similar values are found for both T B / B B and T T / T B in spite of differences expected in strand sensitization. The inference is that single isolated breaks do not make a contribution to the final break score (cf. Fig. 3).
thymidine until radiation and fixation) at two doses in the dose-response curve. The observed frequencies are given in Tables 2 and 3 and have been added to the figures. It will be seen that the same frequency of colour-jump breaks (R III) is found, but there is a marked reduction in differFrequency 8-
--o- Breaks lid TB/t3B ..... ] -l-
L
6~ \ \
0
,
--- Exchanges lid TB/BB I
4"
0 ~
Breaks I/d TI"/IB
~-- Gaps I/d TB/BB
I-
0.3
0.6
0.9
1.2
1.5
~
2.0
Dose (Gy)
Fig. 5. Comparing the frequencies (light/dark) with which breaks are allocated to different chromatids. There is absolutely no evidence of greater sensitization of BB for either breaks or gaps, except in unirradiated controls (where most breaks result from BrdU incorporation?), In contrast, there is a marked sensitivity difference for breaks derived from chromatid interchanges. According to Classic theory, all breaks come from a common pool and should show the same degree of sensitization.
The first, and most obvious point to be made is that on at least two counts, the aberrations which we are scoring as 'breaks' (chromatid terminal deletions, discontinuities) are not behaving in a way consistent with Classic theory, as originally defined. Over a wide range of dose and absolute frequency, and even into the unirradiated control ('spontaneous', probably BrdU-induced), the frequency of coiour-jump breaks (R III) remains more or less constant at around 12% in V79-379A for both T B / B B and (at two doses) " I ~ / T B chromatin substitution. Such an observation would be expected on Revell theory, but only on Classic theory if there was no (or negligible) contribution from unrestituted single breaks (Fig. 3). An effectively constant proportion of the break/lesion pool at each dose shows the jump, regardless of the substitution status used to distinguish sister-chromatids. This suggests a link with aberration formation rather than an artefact of, say, BrdU incorporation. It also rules out incomplete SCE as a contaminating source (i.e. SCE as defined: S-dependent and unrelated to structural aberrations) since S e E , in contrast to breaks, do not increase with dose in G2 and hardly increase in late-S cells. In any case, incompleteness in SCE is a very rare event. We have therefore to accept that either there are no such things as single breaks (which is Revells original contention) or that, under normal circumstances, restitution is so complete that no significant contribution to breaks arises from this source (which, for practical purposes, brings us almost to a Revell situation). It is interesting that successful simulation from Classic ideas (with Case model 2) of observed Revell ratios also requires no contribution from Poisson class 1. We said above 'almost to a Revell situation', because although Case 2 as applied derives all breaks from intrachanges, failed resti-
316
tution, as an option, is still allowed within a competition situation of 2 or more breaks in a site. Even this option can be removed to give a pure Revell simulation, but this is best done from a Case 1 model which we are not discussing here. Suffice it to say that so far, we have not been able to produce such good fits to observed ratios with this approach. The other interesting thing that emerges from the observed ratios and all our attempts to model them, is the requirement for the dominance of intra-chromatid events ( W = 0.7). Without this constraint, it is not possible to achieve reasonable ratio values (considered as a block). Either intrachromatid lesion interactions are favoured, or intra-chromatid strand proximity is predominant within sites. During transit of late-S and G2, chromatin is actively condensing to form two spatially discrete sister-chromatids, visualization of which signals the onset of prophase. It is inevitable that such a process will tend to favour, progressively, the formation of intra-chromatid sites at the expense of inter-chromatid ones, lowering the colour-jump probability and enhancing minute frequency. If this explanation is plausible, we might anticipate that R I I I would increase as we move earlier in S. Heddle and Bodycote (197(I) found such an increase, Wolff and Bodycotc (1975) did not. We did not look at early S cells, but it is a point worth checking, always bearing in mind the very complex proximity situations that are likely to obtain at that time arising from the very staggered and scattered programme of replication ( A g h a m o h a m m a d i and Savage, 1990). The second count on which breaks do not conform is with respect to BrdU sensitization. In line with the findings of others (e.g. Wolff and Bodycote, 1 9 7 5 ) w e can confirm that in V79379A, there is no evidence of excess BB sensitivity for breaks. We also show that there is none for achromatic lesions (gaps). A slight, but non significant increase may exist in V79-4. Controls were an exception, probably because most of the spontaneous breaks are BrdU-induced. In sharp contrast, lesions/breaks involved in interchanges (inter-arm intrachanges and triradials are not included in the analysis) show at least a two-fold increase in a T B / B B situation (slightly less in V79-4).
Now, according to Classic theory, both types of aberration use the same pool of breaks. A random (?) sample from the ptx)l arc fixed by rejoins (exchanges which include both inter- and intrachanges, Lea, 1046) and the remainder restitute (randomly?) leaving a residual of open simple breaks, if the conditions of randomness are fulfilled, then it is inescapable that all three categories should show the same BB bias. They do not, and that raises some difficult questions. ls there a common pool or two different pools'? Or, put in another way, are intrachange sites, or modes of rejoining, or kinds of lesion in them, different from interchange ones? What differences could promote a sensitivity contrast? It seems unlikely that there is a fundamental difference between the lesions. A small percentage (2-3) of interchanges are 'mixed', in the sense that one (very, occasionally, both) of the participating lesions involves a colour-jump. The simple explanation would be that such aberrations result from the interaction of a single lesion plus an intrachange. These events belong to a family of such compound interactions of which the triradial is another member. Alternatively, the explanation could lie in a differential repair or restitution favouring BrdUenhanced breakage. One could envisage a situation where interchanges are established very rapidly after lesion induction, but that intrachanges come later after a preferential removal of the sensitized break component. We are not aware of evidence for such preferential repair nor for a formation-time separation for these two aberration categories. The close similarity of gaps and breaks despite big frequency differences, is also interesting, and might indicate a closer relationship between them than we have, perhaps, allowed (cf. Brecher, 1977). To these questions (and others which will occur to thoughtful readers) we do not have ready answers, but they suggest that aberration production (as opposed to reduction; Savage, 1990) is probably more complicated than either the early workers or even we, with our vast increase in knowledge, have imagined. During the extensive scoring of breaks for this experiment we have received many indicators that this is so. How does
317
one handle morphologically 'unambiguous' la intrachanges that have no colour-jump or clear 2a that do? So, at least we can anticipate that there is still good experimental mileage on this road of structural chromosomal aberrations likely to provide many papers for Mutation Research in the years ahead.
Acknowledgements Our thanks go to David Papworth for much help and advice in probability theory and to the Unit photographers, K. Glover and E. Evans who helped prepare the figures. This work is supported in part by CEC contract Bi7-039.
References Aghamohammadi, S.Z., and J.R.K.Savage (1990) BrdU pulse/reverse staining protocols for investigating chromosome replication, Chromosoma, 99, 76-82. Brecher, S. (1977) UItrastructural observations of X-ray induced chromatid gaps, Mutation Res., 42, 249-268. Evans, H.J. (1962) Chromosome aberrations induced b~, ionizing radiations, Int. Rev. Cytol., 13, 221-321. Fox, D.P. (1967) The effects of X-rays on the chromosomes of locust embryos, 111. The chromatid aberration types, Chromosoma, 20, 386-412. Heddle, J.A., and D.J. Bodycote (1970) On the formation of chromosomal aberrations, Mutation Res., 9, 117-126. Heddle, J.A., D. Whissell and D.J. Bodycote (1969) Changes in chromosome structure induced by radiations: A test of the two chief hypotheses, Nature (London), 221, 11581160. Lea, D.E. (1946) Actions of radiations on living cells, 1st edn., Cambridge University Press.
Lea, D.E., and D.G. Catcheside (1942) The mechanism of the induction by radiation of chromosome aberrations in Tradescantia, J. Genet., 44, 216-245. Revell, S.H. (1959) The accurate estimation of chromatid breakage and its relevance to a new interpretation of chromatid aberrations induced by ionizing radiations, Proc. Roy. Soc. (London), Ser. B, 150, 563-589. Revell, S.H. (1974) The breakage-and-reunion theory and the exchange theory for chromosomal aberrations induced by ionizing radiations: A short history, Adv. Rad. Biol., 4, 367-416. Savage, J.R.K. (1975) Radiation-induced chromosomal aberrations in the plant Tradescantia: Dose-response curves, 1. Preliminary considerations, Rad. Bot., 15, 87-14(I. Savage, J.R.K. (1976) Annotation: Classification and relationships of induced chromosomal structural changes, J. Med. Genet., 13, 103-122. Savage, J.R.K. (1986) The 'Revell Loop', Clin. Cytogenet. Bull., 1, 192-196. Savage, J.R.K. (1989) The production of chromosome structural changes by radiation: An update of Lea (1946), Chapter VI, Br. J. Radiol., 62, 507-520. Savage, J.R.K. (1990) Mechanisms of chromosome aberrations, in: M.L. Mendelsohn and R.J. Albertini (Eds.), Mutation and the Environment, B, Wiley-Liss, New York, pp. 385-396. Savage, J.R.K., R.J. Preston and G.J. Neary (1968) Chromatid aberrations in Tradescantia bracteata and a further test of Revell's hypothesis, Mutation Res., 5, 47-56. Sax, K. (1940) An analysis of X-ray induced chromosomal aberrations in Tradescantia, Genetics, 25, 41-68. Wolff, S., and J. tkxtycote (1975) The induction of chromatid deletions in accord with the breakage-and-reunion hypothesis, Mutation Res., 29, 85-91. Wolff, S., and P. Perry (1974) Differential Giemsa staining of sister chromatids and the study of sister chromatid exchanges without autoradiography, Chromosoma, 48, 341353.
