ABSTRACT
Hdf1 is the yeast homologue of the mammalian 70 kDa subunit of Ku-protein, which has DNA end-binding activity and is involved in DNA double-strand break repair and V(D)J recombination. To examine
whether Hdf1 is involved in illegitimate recombination, we have measured the
rate of deletion mutation caused by illegitimate recombination on a plasmid in
an
hdf1
disruptant. The
hdf1
mutation reduced the rate of deletion formation by 20-fold, while it did not affect mitotic and meiotic homologous
recombinations between two heteroalleles or homologous recombination between
direct repeats. Hence Hdf1 participates in illegitimate recombination, but not
in homologous recombination, in contrast to Rad52, Rad50, Mre11 and Xrs2, which
are involved in both homologous and illegitimate recombination. The
illegitimate recombination in the
hdf1
disruptant took place between recombination sites that shared short regions of
homology (1-4 bp), as was observed in the wild-type. Based on the DNA end-binding activity of Hdf1, we discuss models in which Hdf1
plays an important role in the late step of illegitimate recombination.
Chromosome rearrangements are often caused by illegitimate recombination, which
occurs between non-homologous DNA sequences or very short regions of homology. Whether Rad
proteins, which are involved in DNA double-strand break repair and homologous recombination, are also involved in
illegitimate recombination was investigated in
Saccharomyces cerevisiae
. Schiestl
et al.
showed that the frequency of illegitimate recombination during integration of a
DNA fragment was reduced by
rad50
,
51
,
52
and
57
mutations (
1
). We have also indicated that the rate of deletion formation caused by
illegitimate recombination was reduced by
rad50
,
52
,
mre11
and
xrs2
mutations, but not by
rad51
,
54
,
55
and
57
mutations (
2
). Some
rad
mutations exhibited different effects in integration and deletion analyses,
implying different mechanisms for the two recombination events and/or different
states of DNA molecules in chromosomes and plasmids.
One of the interesting mammalian factors involved in DNA double-strand break repair and V(D)J recombination during rearrangement of
immunoglobulin genes is the Ku-protein. The Ku-protein, which consists of a 70 kDa subunit (Ku70) and an 80 kDa
subunit (Ku80), binds to double-stranded DNA ends and has DNA helicase activity (
3
-
5
). The Ku-protein is a component of a DNA-dependent protein kinase (DNA-PK), a serine/threonine protein kinase whose targets are p53, c-Myc, Sp1, simian virus 40 T-antigen, RNA polymerase II and Ku-protein itself (
6
-
10
). It is also known that the
XRCC5
,
XRCC6
and
XRCC7
mutants, which lack Ku80, Ku70 and the catalytic subunit of DNA-PK, respectively, are defective in DNA double-strand break repair and V(D)J recombination (
11
-
19
).
In yeast, a Ku-protein homologue, designated Hdf (high affinity DNA binding factor), was
purified by its DNA binding activity (
20
). Hdf consists of 70 and 85 kDa subunits and has binding activity to the end of
double-stranded DNA. The
HDF1
gene, which codes for the 70 kDa subunit, was cloned and sequenced. The amino
acid sequence of Hdf1 shows a limited but significant homology with that of
mammalian Ku-protein. Hdf1 and the 85 kDa subunit form a complex that has DNA end-binding activity. These properties are the same as that of the
mammalian Ku-protein, but it has not been shown whether Hdf1 functions as a DNA
helicase or in an interaction with a catalytic subunit of DNA-PK, as was observed for mammalian Ku-protein. Recently, Siede
et al
. showed that, though an
hdf1
single mutant did not exhibit any radiation sensitivity, it exhibited
additional radiation sensitivity in a
rad52
background, indicating that Hdf1 is also responsible, at least partially, for
repair of DNA damage (
21
).
To determine whether Hdf1, which has DNA end-binding activity, plays a role in recombination, we have examined the
effect of a
hdf1
mutation on illegitimate recombination using the plasmid system for
quantitative analysis of deletion formation. The rate of illegitimate
recombination is shown to be reduced in the
hdf1
disruptant. We also show that the
hdf1
mutation does not affect mitotic and meiotic homologous recombination. These
results indicate that Hdf1 is involved in illegitimate recombination, but not
in homologous recombination.
Escherichia coli recA
strain DH10B was used for rescue of plasmid DNA isolated from
S.cerevisiae
(
22
). The yeast strains used are listed in Table
1
. The
hdf1
mutants were constructed by the one-step gene replacement method using a
Sac
I-
Hin
dIII fragment containing
hdf1
::
LEU2
of the plasmid pGEM4ZS-H/LEU (
20
,
23
). MR93-28C and MR966 were kindly supplied by A. Sugino (Osaka University). MR93-28C, MR966 and their derivatives bear the same genetic background as
SK-1 and sporulate rapidly and efficiently to produce dyad spores (
24
,
25
).
Determination of the rate of illegitimate recombination and structural analysis
of recombinants were carried out as described previously (
2
). Briefly, YCpL2, which carries two negative selection markers, the
CAN1
and
CYH2
genes, and three positive selection markers, the
URA3
,
TRP1
and
LEU2
genes, on a YCp plasmid was introduced into a haploid
can1 cyh2
strain. Because the wild-type
CAN1
and
CYH2
genes are dominant to the
can1
and
cyh2
mutations, a deletion mutation which simultaneously inactivates the
CAN1
and
CYH2
genes on the plasmid makes the transformant resistant to both canavanine (Can) and cycloheximide (Cyh). Our previous work showed that there were deletion mutations by
illegitimate recombination on plasmids obtained from the Can
R
Cyh
R
cells (
2
). The rate of deletion mutation was determined by fluctuation analysis (
26
,
27
).
Mitotic and meiotic homologous recombinations between two
his1
heteroalleles were measured before and after shifting from YPA medium (1% yeast
extract, 2% polypeptone, 2% potassium acetate) to sporulation medium (1%
potassium acetate, 0.02% raffinose). Cells were grown in liquid YPA medium to a
concentration of ~1 * 10
7
cells/ml. Then the cells were washed twice in sterile distilled water and
resuspended in 2 ml sporulation medium. The cells were grown with shaking and
plated onto SD and SD minus histidine plates at 24 h after the shift to
sporulation medium.
The rate of recombination between direct repeats on the YCpD2 plasmid was
determined using fluctuation analysis (
26
,
27
). YCpD2 carries a negative selection marker, the
CYH2
gene, flanked by the
CAN1
gene repeats on the YCp plasmid, which has three positive selection maker, the
URA3
,
TRP1
and
LEU2
genes. YCpD2 is an intermediary product in the course of construction of YCpL2,
which was described in a previous report (
2
). The haploid
cyh2
strain, which carries YCpD2, is sensitive to cycloheximide. When a deletion
between the
CAN1
repeats occurs, the cell becomes resistant to cycloheximide. The rate of
deletion can be measured by plating YCpD2 transformants onto SD medium
containing cycloheximide (10 [mu]g/ml).
To investigate the effect of mutation of the
HDF1
gene on illegitimate recombination, the
hdf1
disruptant YT444 was constructed from DH6.61D, a
can1 cyh2
mutant, and YCpL2, which is a YCp plasmid carrying the
CAN1
,
CYH2
and
URA3
genes (see Materials and Methods), was introduced into the YT444 cells. The Ura
+
transformants grown in liquid SD minus uracil medium were plated on SD plates
containing canavanine and cycloheximide. It was found that cells resistant to
both canavanine and cycloheximide appeared at the rate of 4.4 * 10
-9
/ cell/division cycle in the
hdf1
disruptant (Fig.
1
). The rate in the
hdf1
disruptant was ~5% of the rate in the isogenic
HDF1
strain (8.5 * 10
-8
/cell/division cycle). The result indicates that Hdf1 is involved in
illegitimate recombination.
To analyse deletion mutations formed in the
hdf1
disruptant, the recombinant plasmids rescued from five Can
R
Cyh
R
colonies of the disruptant were analysed by PCR as described previously (
2
). Four out of the five plasmids were found to have various sizes of deletions,
but one of them, 1HX2, was indistinguishable from the parental plasmid YCpL2 by
0.7% agarose gel electrophoresis (Fig.
2
A). When 1HX2 was again introduced into the
can1 cyh2
strain YT444, the transformant was resistant to both canavanine and
cycloheximide and, therefore, 1HX2 may have point mutations or small
rearrangements in both the
CAN1
and
CYH2
genes. The nucleotide sequences of the recombination junctions of the four
deletion plasmids were determined and those of the parental recombination sites were estimated. There were short regions of
homology (1-4 bp) between the parental recombination sites of the deletion mutation
formed in the
hdf1
disruptant (Fig.
2
B). This result was comparable with those obtained from the wild-type strain (see also
2
). From sequence analysis of the recombination junctions in the present and
previous works, the sequences of the parental recombination sites in the
hdf1
disruptant were indistinguishable from those in the wild-type strain.
Table 2
To observe the effect of the
hdf1
mutation on homologous recombination during both mitosis and meiosis, we
measured recombination frequency between two heteroalleles at the
his1
locus. The frequency of mitotic recombination in the
hdf1
diploid was comparable with that in the wild-type strain (Table
2
, 0 h). A similar result was obtained for recombination during meiosis (Table
2
, 24 h). Sporulation efficiency of the wild-type and the
hdf1
diploids was nearly 90% at 24 h after the shift to sporulation medium. We also
measured UV-induced homologous recombination between two
his1
heteroalleles. When the cells plated on SD minus histidine were irradiated by
UV (20 J/m
2
), the frequency of His
+
recombinants was increased by ~20 times in the
hdf1
diploids, an elevated frequency that was comparable with that in the wild-type strain (data not shown).
It is known that the factors involved in homologous recombination between direct
repeats are different from those involved in recombination between two
heteroalleles (
28
). We have constructed a plasmid system for detecting homologous recombination
between direct repeats. A plasmid YCpD2, which carries the
CYH2
gene between
CAN1
gene repeats on aYCp plasmid, was introduced into haploid
cyh2
strains (Fig.
3
A). The transformants were sensitive to cycloheximide. When a deletion mutation
is formed by homologous recombination between the
CAN1
repeats, the cell will become resistant to cycloheximide. The rate of deletion
mutation can be measured by plating YCpD2 transformants on SD plates containing
cycloheximide. Structural analysis of the recombinant plasmids rescued from Cyh
R
cells revealed that five out of the five plasmids rescued had a deletion of the
CYH2
-
CAN1
segment. It was found that Cyh
R
cells of the
hdf1
disruptant appeared at the rate of 9.0 * 10
-5
/cell/division cycle, which was comparable with that in the wild-type strain (1.91 * 10
-4
/cell/division cycle) (Fig.
3
B), while the rate of deletion was reduced by 10-fold in a
rad52
mutant (2.1 * 10
-5
/cell/division cycle). These results indicate that Hdf1 is not involved in
mitotic and meiotic homologous recombination between two heteroalleles or
homologous recombination between direct repeats.
The present study shows that Hdf1 is involved in the formation of deletions
caused by illegitimate recombination. However, Hdf1 is not involved in mitotic
and meiotic homologous recombination between two heteroalleles or homologous
recombination between direct repeats and a mutant is not sensitive to UV
irradiation or MMS treatment. In a previous paper, we showed that Rad52, Rad50,
Mre11 and Xrs2 participate in illegitimate recombination (
2
). Rad52 is known to be involved in both mitotic and meiotic homologous
recombination and in DNA double-strand break repair (
30
-
34
). Rad50, Mre11 and Xrs2 are involved in meiotic homologous recombination,
though they do not participate in mitotic homologous recombination (
32
,
35
-
38
). All these proteins but Hdf1 are involved in both illegitimate recombination
and homologous recombination, with the Hdf1 protein having a unique function
specifically involved in illegitimate recombination in yeast.
Hdf1 is the 70 kDa subunit of the yeast Ku-protein homologue, Hdf, which consists of 70 and 85 kDa subunits and has
DNA end-binding activity (
20
). In mammalian cells, the Ku-protein is known to play roles in double-strand break repair and both RS and coding joint formation during
V(D)J recombination (
12
-
14
,
17
,
18
). Ku-protein might mediate a common reaction during double-strand break repair and V(D)J recombination because it is known that
Ku-protein has strong binding activity to the ends of double-stranded DNA produced by double-strand breaks and the hairpin structure produced by Rag1 and
Rag2 during V(D)J recombination (
4
,
39
-
44
). It is thought that Ku-protein might be involved in DNA end-protection in both processes. In
xrs6
mutants, which lack the Ku80 subunit, the frequencies of both RS and coding
joints were reduced during V(D)J recombination. RS junctions contain various
sizes of deletion in
xrs6
mutants, while most RS were joined precisely without loss of nucleotides in the
wild-type strain (
12
). Another model is that Ku-protein might be involved in regulating a cell cycle checkpoint mediated
by DNA-PK in double-strand break repair and V(D)J recombination (
14
). Based on the present results and these considerations, we constructed models
in which Hdf1 plays a role in the illegitimate recombination process in yeast
(Fig.
4
).
We thank Drs H. Feldmann, A. Sugino and J. W. Szostak for providing plasmid and
yeast strains. This work was supported in part by grants to HI and JK from the
Ministry of Education, Science, Sports and Culture of the Japan.
Strain
Culture
Frequency of His
+
cells
Fold
0 h
24 h
induction
YT511 (
HDF1
/
HDF1
)
1
2.5 * 10
-5
2.4 * 10
-3
96
2
2.7 * 10
-5
3.7 * 10
-3
140
3
1.4 * 10
-5
3.2 * 10
-3
230
4
3.5 * 10
-5
3.1 * 10
-3
89
5
1.6 * 10
-5
2.7 * 10
-3
170
YT521 (
hdf1
/
hdf1
)
1
3.2 * 10
-5
3.7 * 10
-3
120
2
0.6 * 10
-5
2.1 * 10
-3
350
3
1.3 * 10
-5
2.5 * 10
-3
190
4
1.8 * 10
-5
1.4 * 10
-3
78
5
3.9 * 10
-5
4.0 * 10
-3
100
REFERENCES
Return