ABSTRACT
Removal of UV-induced pyrimidine dimers from the individual strands of the rDNA locus in
Saccharomyces cerevisiae
was studied. Yeast rDNA, that is transcribed by RNA polymerase I (RNA pol I),
is repaired efficiently, slightly strand-specific and independently of
RAD26
, which has been implicated in transcription-coupled repair of the RNA pol II transcribed
RPB2
gene. No repair of rDNA is observed in
rad1, 2, 3
and
14
mutants, demonstrating that dimer removal from this highly repetitive DNA is
accomplished by nucleotide excision repair (NER). In
rad7
and
rad16
mutants, which are specifically deficient in repair of non-transcribed DNA, there is a clear preferential repair of the transcribed strand of rDNA,
indicating that strand-specific and therefore probably transcription-coupled repair of RNA pol I transcribed genes does exist in yeast. Unexpectedly, the transcribed but not the non-transcribed strand of rDNA can be repaired in
rad4
mutants, which seem otherwise completely NER-deficient.
Cyclobutane pyrimidine dimers induced in DNA by irradiation with UV-light can be removed by the nucleotide excision repair (NER) system to
maintain the genetic integrity (reviewed in
1
-
3
). Removal of dimers from DNA is heterogeneous throughout the genome (
4
,
5
) because dimers can be a substrate for either of two subpathways of NER:
transcription-coupled and global genome repair (
6
). Transcription-coupled repair is a very efficient process in which lesion-stalled RNA polymerase II (RNA pol II) molecules may act as a
condensation site for the assembly of repair complexes (
7
-
9
). Specific gene products might enhance the efficiency of this process. In
Escherichia coli
, a protein called TRCF (transcription repair coupling factor) couples the NER
enzymes to a lesion-stalled RNA polymerase (
10
). Based on
in vitro
studies, the following model for transcription-coupled repair in
E.coli
has been proposed (
10
): TRCF releases the stalled polymerase together with the transcript, binds the
NER protein UvrA, thereby recruiting the NER proteins to lesions that interfere
with transcription. Subsequently these lesions are removed by the action of the Uvr enzymes. In mammalian cells the genes complementing the
hereditary recessive disorder Cockayne syndrome groups A and B are involved in transcription-coupled repair (
11
-
13
), while in
S.cerevisiae
the homolog of the Cockayne syndrome B gene,
RAD26
, is implicated in this process (
14
). It is still unknown whether these genes encode coupling factors analogous to
TRCF in
E.coli
, or are involved in transcription-coupled repair in a different way. Non-transcribed DNA obviously can not be a substrate for transcription-coupled repair. Nevertheless this DNA is repaired by NER
enzymes, although slower than transcribed strands (
4
), in a process referred to as global genome repair. Specific genes have been
shown to be essential for global genome repair. Notably, in human xeroderma
pigmentosum group C (XP-C) cells, non-transcribed DNA is not repaired while transcribed strands of active
DNA are repaired efficiently (
15
,
16
). In yeast the
RAD7
and
RAD16
genes are essential for repair of non-transcribed DNA (
17
,
18
). In
rad7
and
rad16
mutants the transcribed strand of active genes is repaired as efficiently as in
RAD
+
cells, showing that transcription-coupled repair is not hampered in these mutants (
18
). The actual repair process is conducted by a complex of enzymes called
repairosome (
19
), which contains most proteins that are essential for NER known so far. Most
likely this multiprotein complex performs the incisions and subsequent steps in
the same manner for both DNA strands. Possibly the repairosome is unable to
remove dimers in DNA that is condensed into chromatin, and therefore is
dependent on either global genome repair factors or transcription to be able to
operate
in vivo
(
6
). Transcription-coupled repair has been demonstrated in eukaryotes for genes transcribed
by RNA polymerase II (RNA pol II) (
20
-
23
), but not for genes transcribed by RNA pol I (
24
,
25
). Here we investigate the repair of ribosomal DNA (rDNA) in yeast, to find out
whether RNA pol I transcribed DNA is repaired in a similar way as the genes
transcribed by RNA pol II that have been studied so far. rDNA genes are highly
repetitive in all organisms, with yeast having 100-200 copies (reviewed in
26
,
27
). Two structurally and transcriptionally different subclasses of rDNA exist: some
of the copies are inactive and packed in nucleosomal arrays which are not
accessible for psoralen crosslinking while the other copies are
transcriptionally active and in an open non-nucleosomal chromatin conformation that can be crosslinked by psoralen (
28
,
29
). Removal of dimers from rDNA was virtually absent in hamster cells and
inefficient in human cells (
24
,
25
). It was speculated that removal of dimers from the highly repetitive rDNA
cluster could be due to recombination instead of NER (
24
), but subsequently it was shown that in XP-C and CS-B cells which are impaired in NER, repair of rDNA was inhibited (
30
). Repair of mammalian rDNA appeared to be not strand-specific (not transcription-coupled) and less efficient than repair of the genome overall (
24
,
25
). We have studied removal of dimers from the rDNA cluster of yeast in repair
proficient (
RAD
+
) cells and in various
rad
mutants that are disturbed in specific subpathways of NER. Our results reveal
marked differences between repair of rDNA in yeast compared to results
described for mammalian cells, as well as differences in repair of rDNA and
genes that are transcribed by RNA pol II. The data also have implications for
the function of Rad4p in NER, and possibly for its presumed human homolog, XPC.
All general procedures including DNA purification, restriction enzyme digestion,
cloning and gel electrophoresis were performed according to standard procedures
(
31
). Plasmids were propagated in
E.coli
strain JM101 under appropriate antibiotic selection.
The yeast strains used for this study are listed in Table
1
. All strains were kept on selective YNB (0.67% yeast nitrogen base, 2% glucose,
2% bacto agar) supplemented with the appropriate markers. Cells were grown in
complete medium (YEPD: 1% yeast extract, 2% bacto peptone, 2% glucose) at 28oC under vigorous shaking conditions.
Yeast cells were transformed by electroporation (2250 V/cm, 250 [mu]F, 200 [Omega]). Cells were plated on YNB with the necessary amino acids and
incubated at 28oC for 2-5 days. Disruption of the
RAD4
gene was accomplished by transformation of
Xba
I-digested pDG38 (gift of D. Gietz). Disruptions of the
RAD14
gene were obtained by transformation of
Sac
I/
Nco
I-digested pBM190 (gift of L. Prakash;
32
). Disruption of the
RAD7
,
RAD16
and
RAD26
genes has been described earlier (
18
,
14
).
Yeast cells diluted in chilled phosphate-buffered saline were irradiated with 254 nm UV light (Philips T UV 30W) at
a rate of 3.5 J/m
2
/s. Cells were collected by centrifugation, resuspended in growth medium and
incubated for various times in the dark at 28oC prior to DNA isolation (
33
). DNA was purified on CsCl gradients (
31
).
Construction and isolation of single stranded M13 derived probes recognizing the
RPB2
gene was as described before (
18
). For construction of strand-specific rDNA probes a 1 kb
Eco
RI-
Mlu
I rDNA fragment from plasmid pGEM3-EM1 (gift of J. Venema) was cloned in both orientations in M13.
Single-stranded DNA was isolated according to Sambrook
et al
. (
31
) and used for primer extension to generate
32
P-labeled strand-specific probes as described earlier (
14
,
18
).
Table 1
Genomic DNA was cut with restriction endonuclease
Hin
dIII, which generates a 6.4 kb rDNA fragment. DNA samples were divided in two
equal parts, one of which was incubated with T4 endonuclease V (T4endoV;
isolated as described earlier:
34
,
35
), and the other mock-treated, both were loaded on denaturing agarose gels as described by Bohr
et al
. (
4
). After electrophoresis the DNA was transferred to Hybond N
+
(Amersham) and hybridized to strand-specific probes. After hybridization and data analysis the probe was
removed by alkaline washing and subsequently the blot was hybridized to the
probe recognizing the opposite strand.
The amount of hybridized labeled probe in each band on the Southern blots was
quantified with a Betascope 603 blot analyser (Betagen) and used to calculate
the amount of dimers per fragment according to the Poisson distribution as was
described previously (
4
). After being scanned in the blot analyser, autoradiographs were prepared from
the Southern blots.
The yeast rDNA cluster consists of 100-200 repeats of 9.1 kb each (
26
,
27
). Each repeat contains genes for 18, 5.8 and 25S rRNA that are transcribed by
RNA pol I into a single 35S transcript that is post-transcriptionally processed into the separate rRNAs. Each unit also contains a 5S rRNA gene that is transcribed by RNA
pol III (
26
,
27
). We have studied removal of dimers from both strands of a 6.4 kb
Hin
dIII fragment that comprises almost the whole RNA pol I transcribed region. A
schematic map that shows relevant restriction sites, transcription units and
the probe used for repair experiments is presented in Figure
1
. Since probes recognizing rDNA detect all the repeats at the same time, it is
important to note that only some of the copies are indeed transcribed at a
given time, while the others are not active (
28
,
29
). Therefore a probe for the transcribed strand will recognize strands that are
actually transcribed as well as template strands that are not currently
transcribed. In addition, a fraction of both strands (correlating with the
active copies) is non-nucleosomal. Under the conditions of the repair experiments it can be
estimated that ~40-60% of the rDNA units are active (
29
).
We have studied repair of rDNA in
S.cerevisiae
. In this organism, rDNA is rather efficiently repaired (comparable to the non-transcribed strand of the
RPB2
gene) by NER, in contrast to the inefficient repair of rDNA in higher
eukaryotes (
24
,
25
). We report that in yeast, strand-specific and therefore probably transcription-coupled repair of this class of RNA pol I transcribed genes exists,
as is most clearly observed in
rad7
,
rad16
and especially in
rad4
mutants.
Only a small difference in repair of both rDNA strands is observed in
RAD
+
cells, probably since many of the rDNA copies are not active (
29
), thereby obscuring the more efficient repair of the transcribed strand of the
active fraction. Transcription-coupled repair of this class of genes might also be less efficient than
transcription-coupled repair of RNA pol II transcribed genes, because these processes
may be mediated by different factors. The
RAD26
gene (
14
) is not involved in transcription-coupled repair of rDNA, whereas in human cells the Rad26p homolog CSB (
13
) does play a role in removal of dimers from rDNA (
30
). This may reflect the more general repair defect in Cockayne syndrome cells,
that are disturbed in more than only transcription-coupled repair (
12
,
30
), while the yeast
rad26
mutant has a repair defect that seems to be confined to the transcribed strands
of RNA pol II transcribed genes. Specific involvement of Rad26p in RNA pol II
mediated transcription-coupled repair therefore most likely underlies the absence of an effect of
the
rad26
mutation on rDNA repair. An alternative explanation comes from our recent
observation that the effect of the
rad26
mutation is gene-specific and might depend on the level of transcription (
6
). Therefore the observation that rDNA repair is independent of
RAD26
might be due to the high level of transcription of the active rDNA copies (
28
,
29
).
The
rad7
and
rad16
mutants are completely deficient for removal of dimers from the non-transcribed strand of the
RPB2
and
GAL7
genes (
6
,
18
). In contrast, repair of non-transcribed strands of rDNA is only partly dependent on Rad7p and Rad16p.
This could be due to the non-nucleosomal structure of the active rDNA genes, that might allow NER
enzymes to exert their function on the non-transcribed strand of this DNA in the absence of Rad7p and Rad16p.
Alternatively, some transcription by RNA pol III could come from the opposite
direction (the 5S gene), causing transcription-coupled repair (mediated by RNA pol III) of the non-transcribed strand. However, we did not observe any transcripts
derived from the non-transcribed strand on Northern blots with strand-specific probes, while transcripts from the transcribed strand were
present in high amounts (data not shown). Moreover, it can be inferred from
mutation spectra that targeting of repair enzymes to transcribed strands is
probably not mediated by RNA pol III, since mutations in the
SUP4
-o gene that is transcribed by this polymerase are found mainly in the
transcribed strand (
38
).
Most surprisingly, removal of dimers from rDNA occurs in a
rad4
mutant. This repair is confined to the transcribed strand. Dimer removal from
the transcribed strand in the
rad4
mutant is conducted by
bona fide
NER, since in a
rad4rad14
mutant (Rad14p is essential for NER,
32
) this rDNA repair is completely abrogated (data not shown). As expected, this
repair is independent of Rad7p and Rad26p, since in
rad4rad7
and
rad4rad26
double mutants the transcribed strand of rDNA is still repaired (data not
shown). Until now Rad4p was considered essential for NER (both transcription-coupled repair and global genome repair), since no dimers are removed from
both strands of an active gene as well as from the genome overall in a
rad4
mutant (
18
). The analysis of protein complexes suggested that Rad4p is part of the
repairosome (
19
,
39
), but to date the molecular function of this protein has not been revealed. The
Rad4 protein is essential for NER in a reconstituted yeast repair system (
40
) and cell-free extracts of
rad4
mutants are defective for NER (
19
). Apparently in the special case of RNA pol I transcribed DNA, NER can operate
in the absence of this protein. Isogenic
rad4
,
rad14
and
rad4rad14
mutants are equally UV-sensitive (not shown), and therefore the Rad4p-independent repair of the transcribed strand of rDNA may be
fortuitous rather than reflecting a biological important function. This
phenomenon is not conserved in higher eukaryotes since an XP-C mutant, the proposed human counterpart of a
rad4
mutant on the basis of limited sequence homology between the yeast Rad4 and the
human XPC proteins (
41
), is completely defective in repair of both rDNA strands (
30
). Strikingly, yeast
rad4
mutants and human XP-C cells
in vivo
seem to have reciprocal phenotypes regarding repair of RNA pol I transcribed
versus RNA pol II transcribed genes: XP-C cells are only capable of repair of RNA pol II transcribed strands (
16
) while
rad4
mutants can only repair template strands that are transcribed by RNA pol I.
Both yeast Rad4 and human XPC proteins seem to be essential for NER, as both
are absolutely required in the respective reconstituted NER systems (
40
,
42
-
44
). Apparently, in the cell, NER can take place while these proteins are absent,
but only at sites where transcription takes place, possibly by a-as yet unknown-component of the transcription machinery. The function of Rad4p in
yeast is then supplied by RNA pol I transcription, while in human cells RNA pol
II transcription overcomes the need for the NER-function of XPC.
The molecular function of Rad4p and XPC is still unknown, but clearly these
proteins are not essential for the incision event of NER. The involvement of
transcription to bypass the need for both yeast Rad4p and human XPC for NER
in vivo
makes it tempting to speculate that a function of these proteins might be
during damage recognition, since Rad4p/XPC-independent NER seems to occur only at the site of transcription.
Alternatively these proteins might have architectural roles, e.g. in building
of a repairosome (
39
), or other important accessory functions during NER.
The findings described here reveal for the first time some similarity between
the preferential repair phenotypes of the yeast
rad4
and human XP-C mutants (
18
). Since the interaction between Rad4p or XPC with the yeast and human Rad23
proteins, respectively, is also conserved (
40
,
45
), the hypothesis that Rad4p and XPC are indeed homologs (
41
), is supported. Both proteins are essential for the NER process but can in
specific cases be replaced by components of transcription machineries. Yeast
rad7
and
rad16
mutants have a phenotype very similar to human XP-C mutants (
18
). This may be partly coincidental. In contrast to the essential function of XPC
in reconstituted NER systems (
42
,
43
), the Rad7 and Rad16 proteins seem dispensable for NER in a highly purified system
(
40
). Therefore Rad7p and Rad16p presumably have a specific function in repair of
non-transcribed DNA (
6
), whereas XPC seems to have a more general function in NER. Identification of
putative mammalian homologs of
RAD7
and
RAD16
, that seem to be the real effectors of non-transcribed DNA repair, will therefore be highly interesting.
Our studies reveal some differences in repair of rDNA in yeast versus mammalian
cells. These may be differences in efficiency rather than mechanistic
differences since NER to date has been found to be highly conserved in
eukaryotic species (
1
). Alternatively, a mechanistic divergence between the NER systems in yeast
versus higher eukaryotes may be revealed. Summarized, we report that rDNA is
repaired by NER in yeast, this repair can be strand-specific and probably transcription-coupled as revealed in specific NER mutants and finally our results
demonstrate that Rad4p is not essential for NER in the special case of the RNA
pol I transcribed strand of rDNA.
We thank Drs J. Venema, D. Gietz, R. Waters and L. Prakash for the gift of plasmids and yeast strains. Jeannette Heyn and Nanda de Groot are
acknowledged for excellent technical assistance. We are greatful to Dr Judith
Tasseron-de Jong for her advice regarding statistical analysis of the data. This
work was supported by the J.A. Cohen Institute for Radiopathology and Radiation
Protection (IRS), project 4.2.9.
Strain
Genotype
Source
a
W303-1B
MAT
[alpha]
ho can1-100 ade2-1 trp1-1
R. Rothstein
leu2-3,112 his3-11,15 ura3-1
W303236
rad16
[Delta]
::URA3
b
This laboratory
MGSC102
rad26
[Delta]
::HIS3
b
This laboratory
MGSC104
rad7
[Delta]
::LEU2
b
This laboratory
MGSC101
rad23
[Delta]
::URA3
b
This laboratory
MGSC131
rad4
[Delta]
::URA3
b
This laboratory
MGSC132
rad4
[Delta]
::URA3 rad7
[Delta]
::LEU2
b
This laboratory
MGSC133
rad4
[Delta]
::URA3 rad26
[Delta]
::HIS3
b
This laboratory
MGSC139
rad14
[Delta]
::LEU2
b
This laboratory
MGSC141
rad4
[Delta]
::URA3 rad14
[Delta]
::LEU2
b
This laboratory
MG70/X9b-7B
Mat
[alpha]
gal ade2-1 rad4-4
YGSC
c
SF RAD1o
MAT
a
gal2 leu2-1,112 his4-58
ura3-52
pep4-3 rad1
[Delta]
R. Waters
SF RAD2-/2
MAT
a
gal2 leu2-1,112 his4-580 ura3-52
pep4-3 rad2::URA3
R. Waters
YR3-3
MAT
[alpha]
leu2-3,112 ura3-52 can1 trp1
[Delta]
rad3-2
L. Prakash
REFERENCES
Return