cdc9 mutants of yeast lack detectable DNA ligase activity at restrictive temperatures. They also appear to be more sensitive than wild-type cells to ultraviolet (u.v.) radiation and it has been assumed that this is because the CDC9 ligase is needed for the final ligation step in excision repair. The fact that single-strand breaks have been demonstrated in u.v.-irradiated cdc9 mutants has been regarded as evidence for this interpretation. However, the kinetics of appearance of nicks in the DNA do not support this since maximal levels of strand breaks appear almost immediately after exposure to u.v. light and not progressively as repair events are initiated. We believe, therefore, that these strand breaks are connected with a u.v.-dependent preincision event, possibly connected with reorganization of chromatin.

This work is concerned with excision of pyrimidine dimers after ultraviolet (u.v.) irradiation of yeast and involves nine yeast mutants: radl, -2, -3, -4, -7, -10, -14 and -16 and cdc9. These eight rad mutants were originally classified as being defective in the same or related repair processes by epistasis studies and have all been shown to be defective to some degree in dimer excision (Unrau et al. 1971; Game & Cox, 1972; Prakash, 1977a,b; Reynolds, 1978; Prakash & Prakash, 1979). CDC9 is the gene for either the sole or the major DNA ligase in yeast (Barker et al. 1985). Mutations, therefore, have to be conditional lethals and appear to be somewhat more sensitive than wild-type to u.v. irradiation (Johnston, 1979).

Table 1 gives a summary of the categories of DNA found in yeast cells. We shall not be concerned with mitochondrial DNA, since dimers are not removed from it by excision repair (Moustacchi & Heude, 1982), but we use repair assays in our laboratory for the other two sets of DNA- We analyse intact chromosomal DNA on alkaline sucrose gradients (Resnick et al. 1981). Plasmid molecules are analysed on agarose gels, assaying for dimers by nicking with u.v.-endonuclease to convert covalently closed circles to open circles (McCready & Cox, 1980).

Table 1.

DNA in S. cerevisiae

DNA in S. cerevisiae
DNA in S. cerevisiae

These are essentially as described by Resnick et al. (1981) and McCready & Cox (1980). Brief outlines of the methods are given in Figs 1 and 2.

Fig. 1.

Preparation of DNA for sucrose gradients.

Fig. 1.

Preparation of DNA for sucrose gradients.

Fig. 2.

Assay for pyrimidine dimers in the 2 μm plasmid, endo, endonuclease.

Fig. 2.

Assay for pyrimidine dimers in the 2 μm plasmid, endo, endonuclease.

Fig. 3 shows a typical profile of DNA from unirradiated cells and the same DNA treated with u.v.-endonuclease. This shows the chromosomal DNA as a broad peak unaffected by u.v.-endonuclease and a minor peak of mitochondrial DNA further up the gradient. Fig. 4 shows DNA from cells irradiated at 6 J m−2 with and without u.v.-endonuclease treatment. Repair is monitored by incubating cells in unlabelled growth medium before harvesting and Fig. 5 shows a set of gradient profiles superimposed to show repair over a 1-h period after irradiation at 10 J m−2. Repair is not quite complete at the end of this period.

Fig. 3.

Sedimentation of unirradiated yeast DNA treated (- - - -) and untreated (—) with u.v.-endonuclease, kb, 101 bases.

Fig. 3.

Sedimentation of unirradiated yeast DNA treated (- - - -) and untreated (—) with u.v.-endonuclease, kb, 101 bases.

Fig. 4.

Sedimentation of DNA from cells irradiated at 6 J m−2 and treated (- - - - -) and untreated (—) with u.v.-endonuclease.

Fig. 4.

Sedimentation of DNA from cells irradiated at 6 J m−2 and treated (- - - - -) and untreated (—) with u.v.-endonuclease.

Fig. 5.

Sedimentation of u.v.-endonuclease-treated DNA from cells irradiated at 10 J m−2 and incubated for up to 90 min.

Fig. 5.

Sedimentation of u.v.-endonuclease-treated DNA from cells irradiated at 10 J m−2 and incubated for up to 90 min.

Fig. 6 shows rates of repair of chromosomal DNA in the various rad mutants. These rates vary somewhat between different alleles of some genes (data not shown) suggesting, perhaps, that the reason rad7, -10, -14 and -16 are able to remove some dimers may simply be because we are using leaky alleles. This may be so in the case of radlO since a strain with a deletion in the radio gene is as sensitive to u.v. light as our radl mutant (Friedberg, 1987). However, in the case of rad7, we find that a deletion mutant gives us an identical rate of dimer excision to that shown (unpublished observation). We do not have deletion mutants in rad14 or -16.

Fig. 6.

Rates of repair of chromosomal DNA in rad mutants.

Fig. 6.

Rates of repair of chromosomal DNA in rad mutants.

Dimers in the plasmid are assayed on gels. Total DNA is extracted carefully so as to avoid nicking. There is no selection for supercoiled DNA, or indeed for plasmid, as we want to be able to detect whether post-u.v. incubation results in nicking of circles. In this system the DNA is purified and all dimers can be cut by the micrococcal u.v.-endonuclease. Fig. 7 shows the results of such an assay. Repair of dimers after 20 J m−2 is complete after incubation for 3 h in the wild-type strain.

Fig. 7.

Disappearance of dimers from 2 urn plasmid during 3-h incubation following u.v. irradiation at 20 J m−2. Pairs of tracks represent samples untreated and treated with u.v.-endonuclease. Conversion of closed circles to open circles indicates the presence of pyrimidine dimers.

Fig. 7.

Disappearance of dimers from 2 urn plasmid during 3-h incubation following u.v. irradiation at 20 J m−2. Pairs of tracks represent samples untreated and treated with u.v.-endonuclease. Conversion of closed circles to open circles indicates the presence of pyrimidine dimers.

Fig. 8 shows what happens in a rad1 mutant incubated for 5 h, after a 20 J m−2 dose of u.v., and shows very clearly that:

Fig. 8.

Failure of a rad1 mutant to remove dimers during 5-h post-u.v. incubation.

Fig. 8.

Failure of a rad1 mutant to remove dimers during 5-h post-u.v. incubation.

(1) There is no removal of dimers.

(2) There is no accumulation of open circles in the cells during the post-u.v. incubation period, suggesting a pre-incision block.

(3) There is no increase in dimer-free circles, suggesting that not only is there no repair, but also there is no synthesis of new plasmid molecules, even using undamaged templates.

Such assays have shown that none of these rad mutants exhibits an increase in open circles during post-u.v. incubation, suggesting that all the blocks are either at or before incision. This is in agreement with work done by Reynolds & Friedberg (1981), who showed that no nicks accumulated in high molecular weight DNA in these mutants except after very high doses of u.v. light. The rates of repair of the 2μ m plasmid (Fig. 9) are similar to the rates shown for chromosomal DNA in wild type and in all mutants except rad1 6, which does not repair dimers in the plasmid at all, but repairs chromosomal DNA rather well.

Fig. 9.

Rates of repair of 2μm plasmid molecules in RAD+ and rad mutant strains.

Fig. 9.

Rates of repair of 2μm plasmid molecules in RAD+ and rad mutant strains.

CDC9 is the structural gene for the only ligase detected to date in yeast cells. In temperature-sensitive mutants there is accumulation of unligated Okazaki fragments at the restrictive temperature (Johnston & Nasmyth, 1978) and a temperature sensitive ligase activity is found (Barker et al. 1985). Mutants are apparently more sensitive than normal cells to u.v. light, though this is quite difficult to assess, since cells have to be grown at the permissive temperature to measure survival. It may be supposed that the increased sensitivity is due to a failure in the final ligation step of excision repair. So let us now see whether this is indeed the case.

Figs 10 and 11 show what happens in the 2 μ m plasmid during the first 2h of incubation at the restrictive temperature after irradiation at 20 J m−2 in a cdc9 strain. This is sufficient to introduce an average of 1·2 dimers per plasmid molecule. There are several features to notice. First, in unirradiated plasmid there is, as we would expect, an increase in open circles during the 2 h, due to aborted replication. Second, there is clearly repair of the plasmid at a rate comparable with wild-type repair. After 1 h only about a third of those plasmids originally containing dimers still do so. Third, there is an accumulation of open circles in the irradiated plasmid. ‘Phis could either be a build-up of unligated repair intermediates or of replication intermediates. The latter seems more likely because repair is being completed at normal rates in at least the majority of the DNA. This interpretation has been confirmed by coupling the cdc9 mutation with the cdc7 mutation (Fig. 12). cdc7 mutants fail to initiate DNA synthesis at the restrictive temperature, and these completely block the appearance of replication-dependent open circles. In the cdc7 cdc9 double mutant u.v.-endonuclease-sensitive sites are lost at a rate similar to the wild type during post-u.v. incubation. In other words, the open circles that appear in cdc9 mutants result from aborted replication rather than from a fault in repair.

Fig. 10.

Repair of 2 μm plasmid in a cdc9.7 mutant at 36°C. Tracks 1, 4 and 7 show the increase in open circle forms of the plasmid in the absence of u.v. irradiation. Tracks 2 and 3 show the effect of u.v.-endonuclease on the plasmid immediately after irradiation at 20 J m−2. Tracks 5 and 6 show that after 1 h most of the covalently closed molecules are now unaffected by u.v.-endonuclease.

Fig. 10.

Repair of 2 μm plasmid in a cdc9.7 mutant at 36°C. Tracks 1, 4 and 7 show the increase in open circle forms of the plasmid in the absence of u.v. irradiation. Tracks 2 and 3 show the effect of u.v.-endonuclease on the plasmid immediately after irradiation at 20 J m−2. Tracks 5 and 6 show that after 1 h most of the covalently closed molecules are now unaffected by u.v.-endonuclease.

Fig. 11.

Graph showing appearance of replication intermediates and repair of dimers in cdc9.7 mutant. The amounts of DNA in the bands on gel (Fig. 9) were estimated using a densitometer.

Fig. 11.

Graph showing appearance of replication intermediates and repair of dimers in cdc9.7 mutant. The amounts of DNA in the bands on gel (Fig. 9) were estimated using a densitometer.

Fig. 12.

The effect of introducing a cdc7 mutation (which causes failure to initiate DNA synthesis) into the cdc9 strain. There is now no accumulation of replication intermediates and it is clear that dimers in plasmids are removed at the normal rate during post-u.v. incubation.

Fig. 12.

The effect of introducing a cdc7 mutation (which causes failure to initiate DNA synthesis) into the cdc9 strain. There is now no accumulation of replication intermediates and it is clear that dimers in plasmids are removed at the normal rate during post-u.v. incubation.

We conclude that the CDC9-encoded ligase is not required for repair of the 2 μm plasmid. The plasmid appears to be repaired perfectly normally in cdc9 mutants. This implies either that there is a second DNA ligase present (but so far undetected) in yeast cells, as there is in other eukaryotes, or that another enzyme is able to seal the nicks resulting from excision repair. There is no evidence of a second ligase activity even in the most sensitive assays used to date (Barker et al. 1985). However, it is possible that the second ligase represents a very minor component of ligase activity in yeast as in mammalian cells (Soderhall & Lindahl, 1976) or that the ligase is not extracted by the methods used for making cell-free extracts to test for ligase activity.

Moving now to repair of chromosomal DNA, the results are more difficult to analyse. Fig. 13 shows a gradient of DNA from a cdc9 mutant incubated at 36°C and irradiated with 10 J m−2 compared with that from unirradiated cells both without u.v. endonuclease. It is clear that u.v.-dependent nicks are appearing on incubation. The DNA is prelabelled, so we are looking entirely at non-replicating DNA. So the CDC9 ligase may indeed be required for the final ligation step of excision repair in chromosomal DNA. However, partly because it is clear that the plasmid DNA is repaired successfully, we studied the cdc9 mutant more closely to test the validity of this interpretation.

Fig. 13.

Sedimentation of DNA from a cdc9 mutant incubated at 36°C and irradiated with u.v. at 10 J m−2.

Fig. 13.

Sedimentation of DNA from a cdc9 mutant incubated at 36°C and irradiated with u.v. at 10 J m−2.

At all doses analysed (0·5−10 J m−2) all nicks are present in the DNA 5 min after irradiation. There is no further increase after this time. In contrast to this nearly 2h of incubation are needed to complete repair after the same dose of u.v. If we were really looking at nicks arising as a result of failure in the last step of repair, we would expect the rate of appearance of nicks at a given dose to follow closely the kinetics of repair at that dose (Fig. 14). In addition, there are many fewer nicks than dimers. After 2 J m−2 there is about 1 nick per 10 dimers and at 8 J m−2 about 1 nick per 5 dimers. One way of accounting for this would be to suppose that repair complexes reach saturation at a low number of dimers and that, after that, repair ceases. In that case we would expect a higher ratio of nicks per dimer at low doses than at high doses, which is the opposite of what we observe.

Fig. 14.

Rapid appearance of cdcP-dependent nicks after irradiation at 8 J m−2 compared with the predicted rate of appearance of strand breaks derived from measurements of the rate of early repair after the same dose of u.v. light.

Fig. 14.

Rapid appearance of cdcP-dependent nicks after irradiation at 8 J m−2 compared with the predicted rate of appearance of strand breaks derived from measurements of the rate of early repair after the same dose of u.v. light.

Neither of these sets of observations (the fast rate of appearance of nicks and ratios of nicks to dimers) fits at all well with the standard interpretation of the appearance of ct/cP-dependent nicks after u.v. irradiation. For this reason, we have tried to find another interpretation of our results. Because of the very rapid appearance of the nicks after irradiation, we must consider the possibility that they are associated with some event occurring even before incision.

We know very little about preincision events, either in yeast or in higher eukaryotes, still less about what events are likely to involve nicking and immediate resealing of DNA strands. However, it is possible that these may include chromatin rearrangement, either at the nucleosome level (e.g. see Smerdon, 1985) or at the level of chromatin loops (Schor et al. 1975; McCready & Cook, 1984).

We have tried to understand more clearly what is going on in yeast, by looking at cdc9 rad double mutants. Do any of the rad blocks prevent appearance of the cdc9 nicks? The answer is yes, almost all of them. Table 2 compares the number of nicks induced in a cdc9 RAD+ strain with nicks induced in the double mutants, radl to rad4 all block appearance of nicks. Nicks appear at very reduced levels in radlO, -14 and -16. rad7 cdc9 is the only double mutant showing the same levels of nicking as a cdc9 RAD+ strain.

Table 2.

Strand breaks in rad cdc9 double mutants

Strand breaks in rad cdc9 double mutants
Strand breaks in rad cdc9 double mutants

We have two possible models, then, to explain the cdc9 nicks and the effects of the rad mutations in the double mutants.

Model 1: cdc9-dependent nicks are a result of unligated repair patches

In this model the cdc9-dependent nicks do represent incision and incomplete repair, but only at the earliest repair sites, i.e. at the first 10–20 % of dimers normally to be repaired (because we only see 10–20% as many nicks as there are dimers in the DNA). We have to infer that no further incision takes place if these remain unrepaired, perhaps because nucleosome or chromosomal domain rearrangement is required and only happens when repair is complete.

We have seen that radlO, -14 and -16 all show some accumulation of nicks in cdc9 rad double mutants. This is in agreement with the results of Wilcox & Prakash (1981), who concluded that radl4 and -16 must be blocked in post-incision events. We would interpret the result differently, however. Since we know that radlO, -14 and -16 are able to repair some dimers it would seem reasonable to expect to see some nicks as a result of attempted repair, and indeed nicks do appear at much reduced levels compared to those seen in the cdc9 RAD+ cells. These mutants are either leaky, therefore, or else they are able to recognize and repair only a subset of the dimers present. In either case they are not performing normal levels of incision and there is no indication of a post-incision rather than a preincision block.

Only the rad7 cdc9 double mutant shows the same levels of nicking as the cdc9 RAD+. The simplest interpretation of this is that in a rad7 strain, the same subset of dimers (i.e. 10–20% of the sites) is recognized and incised as in the wild type. Perhaps in rad7 it is nucleosome rearrangement or some other mechanism of making dimers accessible that is faulty. Why is there no further initiation of repair even at low doses, when the first set of incisions fail to result in complete repair? Since proportionally more nicks appear per dimer at higher than at lower doses we are clearly not looking at saturation of repair complexes. It must be concluded that if the dimers normally repaired first are incised but not completely repaired then other dimers never become accessible to the repair complexes. One explanation for this would be that the chromatin rearrangement needed to make more dimers accessible fails to occur if early repair is not completed.

Why do nicks in the chromosomal DNA appear at a rate far exceeding the rate of even very early repair (Fig. 14)? We have no ready explanation for this except that for some reason, if unligated repair patches persist in the DNA, new repair sites are incised more readily (until the critical 10–20%, after which no further incisions occur). This seems very unlikely, especially since incision is generally thought to be the rate-limiting step in excision repair.

How is it that the CDC9 ligase is required for ligation of repair patches in chromosomal DNA (in this model) but not in plasmid DNA? (We have never observed any rapidly appearing nicks in plasmid DNA comparable to that seen in chromosomal DNA even at u.v. doses up to 100 J m−2.) We have to conclude either that a very small amount of active CDC9 ligase is present and used preferentially for plasmid repair (rather than for replication or repair in chromosomes) or that another enzyme, perhaps a topoisomerase or a second, u.v.-dependent, ligase can seal nicks in the plasmid but not in chromosomal DNA.

Model 2: cdc9-dependent nicks are a result of preincision events

In this model the cifcP-dependent nicks result from a preincision event. This event would need the gene products of RADI, -2, -3 and -4 (since in radl—4 cdc9 double mutants nicks do not appear). RADIO, -14 and -16 may also be required but our alleles are leaky. The rad7 block is after this event (because rad7 cdc9 double mutants show cdc9 RAD+ levels of nicks).

This model would explain the kinetics of appearance of the nicks and, if the preincision event involved a structural rearrangement of chromatin, for example, could explain why we do not see as many nicks as there are dimers. Nicks do not even have to be at dimers. The nature of the preincision event(s) can only be guessed at, but could be attachment of chromosome domains to repair complexes or rearrangement of chromatin.

The 2 μm plasmid does not show evidence of Ci/cP-dependent nicks and, therefore, cannot require the preincision event unless the nicks are sealed by a different enzyme. However, plasmid molecules are deficient in dimer excision in the rad mutants (Fig. 9). Therefore, the RAD gene products may be involved in both very early and later repair events (as has also been suggested by White, 1985). This could be, for example, if the RAD gene products ‘flag’ the sites of damage immediately after u.v. and the flagging is essential both for the preinclusion event and for incision. Another possibility is that the first step in excision repair is attachment to a multienzyme repair complex. If a RAD gene product in this complex is defective then neither the preincision event nor later steps can be completed efficiently.

Neither model fully explains all the data. Nor are the two models mutually exclusive. It is possible that the CDC9 ligase is involved during both preincision and postincision events. Certainly a proportion of repair can be completed in its absence (e.g. repair of 2 μm plasmid) whilst the repair of other categories may require it. This is perhaps more easily explained by the second model if, for example, some DNA but not all, needs to be reorganized to be repaired. This seems likely since plasmid and chromosomal DNA are quite differently arranged.

Barker
,
D. G.
,
Johnson
,
A. L.
&
Johnston
,
L. H.
(
1985
).
An improved assay for DNA ligase reveals temperature-sensitive activity in cdc9 mutants of Saccharomyces cerevisiae.
Molec. gen. Genet.
200
,
458
462
.
Friedberg
,
E. C.
(
1987
).
The molecular biology of nucleotide excision repair of DNA: recent progress..J
.
Cell Sci. Suppl.
6
,
1
—24.
Game
,
J. C.
&
Cox
,
B. S.
(
1972
).
Epistatic interactions between four rad loci in yeast
.
Mutat. Res.
16
,
353
362
.
Johnston
,
L.
(
1979
).
The DNA repair capacity of cdc9, the Saccharomyces cerevisiae mutant defective in DNA ligase
.
Molec. gen. Genet.
170
,
89
92
.
Johnston
,
L. H.
&
Nasmyth
,
K. A.
(
1978
).
Saccharomyces cerevisiae cell cycle mutant cdc9 is defective in DNA ligase
.
Nature, Land.
274
,
891
893
.
McCready
,
S. J.
&
Cook
,
P. R.
(
1984
).
Lesions induced in DNA by ultraviolet light are repaired at the nuclear cage
.
J. Cell Sci.
70
,
189
196
.
McCready
,
S. J.
&
Cox
,
B. S.
(
1980
).
Repair of the 2 μm plasmid DNA in Saccharomyces cerevisiae.
Curr. Genet.
2
,
207
210
.
MOUSTACCHI
,
E.
&
Heude
,
M.
(
1982
).
Mutagenesis and repair in yeast mitochondrial DNA
. In
Molecular and Cellular Mechanisms of Mutagenesis
(ed. J. F. Lemontt &
W. M.
Generoso
), pp.
273
301
.
Prakash
,
L.
(
1977a
).
Repair of pyrimidine dimers in radiation sensitive mutants, rad3, rad4, rad6and rad9 of Saccharomyces cerevisiae.
Mutat. Res.
45
,
13
20
.
Prakash
,
L.
(
1977b
).
Defective thymine dimer excision in radiation sensitive mutants rad16 and radió of Saccharomyces cerevisiae.
Molec. gen. Genet.
152
,
125
128
.
Prakash
,
L.
&
Prakash
,
S.
(
1979
).
Three additional genes involved in pyrimidine dimer removal in Saccharomyces cerevisiae: RAD7, RAD 14 and MMS19. Molec. gen. Genet.
176
,
351
—359.
Resnick
,
M. A.
,
Boyce
,
J. M.
&
Cox
,
B. S.
(
1981
).
Postreplication repair in Saccharomyces cerevisiae.
J. Bact.
146
,
285
290
.
Reynolds
,
R.
(
1978
).
Removal of pyrimidine dimers from Saccharomyces cerevisiae nuclear DNA under nongrowth conditions as detected by a sensitive, enzymic assay
.
Mutat. Res.
50
,
43
46
.
Reynolds
,
R.
&
Friedberg
,
E.
(
1981
).
Molecular mechanisms of pyrimidine dimer excision in Saccharomyces cerevisiae-. incision of ultraviolet irradiated deoxyribonucleic acid in vivo.
J. Bact.
146
,
692
704
.
Schor
,
S. L.
,
Johnson
,
R. T.
&
Waldren
,
C. A.
(
1975
).
Changes in the organisation of chromosomes during the cell cycle: response to ultraviolet light
.
J. Cell Sci.
17
,
539
565
.
Soderhall
,
S.
&
Lindahl
,
T.
(
1976
).
DNA ligases of eukaryotes
.
FEBS Lett.
67
,
1
—8.
Smerdon
,
M. J.
(
1985
).
Completion of excision repair in human cells
.
J. biol. Chem.
261
,
244
252
.
Unrau
,
P. R.
,
Wheatcroft
,
R.
&
Cox
,
B. S.
(
1971
).
The excision of pyrimidine dimers from DNA of ultraviolet irradiated yeast
.
Molec. gen. Genet.
113
,
359
362
.
White
,
C. I.
(
1985
).
Ph.D. thesis, National Institute for Medical Research, Mill Hill, London
.
Wilcox
,
D. R.
&
Prakash
,
L.
(
1981
).
Incision and postincision steps of pyrimidine dimer removal in excision-defective mutants of Saccharomyces cerevisiae.
J. Bact.
148
,
618
623
.