ABSTRACT
The
Escherichia coli
endonuclease III (Nth-Eco) protein is involved in the removal of damaged pyrimidine residues
from DNA by base excision repair. It is an iron-sulphur enzyme possessing both DNA glycosylase and apurinic/apyrimidinic
lyase activities. A database homology search identified an open reading frame
in genomic sequences of
Schizosaccharomyces pombe
which encodes a protein highly similar to Nth-Eco. The gene has been subcloned in an expression vector and the protein
purified to apparent homogeneity. The
S.pombe
Nth homologue (Nth-Spo) is a 40.2 kDa protein of 355 amino acids. Nth-Spo possesses glyco- sylase activity on different types of DNA substrates with
pyrimidine damage, being able to release both urea and thymine glycol from
double-stranded polymers. The eukaryotic protein removes urea more efficiently
than the prokaryotic enzyme, whereas its efficiency in excising thymine glycol
is lower. A nicking assay was used to show that the enzyme also exhibits an AP
lyase activity on UV- and
[gamma]-irradiated DNA substrates. These findings show that Nth protein is
structurally and functionally conserved from bacteria to fission yeast.
Endonuclease III (Nth) from
Escherichia coli
was originally identified as an apparent endonucleolytic activity that degrades
heavily UV-irradiated DNA (
1
). Subsequent studies showed that Nth protein is unable to catalyse hydrolysis
of phosphodiester bonds in intact or damaged DNA. Instead, the enzyme acts as a
DNA glycosylase, removing oxidized pyrimidines from DNA, and also as an
apurinic/apyrimidinic (AP) lyase, which cleaves the phosphodiester backbone by [beta]-elimination at the site where a damaged base has been removed (
2
-
5
). Nth protein can excise a wide range of damaged pyrimidine derivatives that
result from ring saturation, ring fragmentation or ring contraction. These
include thymine glycol, urea, 5,6-dihydrothymine, methyltartronylurea, [beta]-ureidoisobutyric acid, 5-hydroxy-6-hydrothymine, 5,6-dihydroxyuracil and 5-hydroxy-5-methylhydantoin (
2
-
8
).
Endonuclease III from
E.coli
(Nth-Eco) is encoded by the
nth
gene at 36 min on the bacterial chromosome and has been cloned (
9
) and sequenced (
10
). Overexpression of
nth
facilitated purification of the protein to physical homogeneity (
10
), allowing its characterization as an iron-sulphur protein (
11
) and its crystallization with subsequent establishment of a three-dimensional structure (
12
,
13
).
Escherichia coli
Nth is a monomeric 23.4 kDa protein with 211 amino acid residues. The enzyme is
elongated and bilobal with a deep cleft separating similarly sized domains, one
of which includes an iron-sulphur [4Fe-4S] cluster and the other a helix-hairpin-helix motif. The [4Fe-4S] cluster has been shown to have a DNA
binding role rather than a catalytic function (
13
). The interhelical turn in the helix-hairpin-helix structure has been identified as the binding site for free
thymine glycol (
13
) and has also been proposed as a DNA binding motif (
14
). Enzymes with catalytic properties similar to Nth-Eco have been identified in many other organisms, including
Micrococcus luteus
,
Drosophila melanogaster
and bovine and human cells (
15
-
22
).
Two bacterial genes encoding putative homologues of Nth-Eco have been sequenced recently in
Bacillus subtilis
(
23
) and
Haemophilus influenzae
(
24
), but no eukaryotic counterparts of the
nth
gene have been cloned. Here we describe the identification of a gene from
Schizosaccharomyces pombe
which encodes a protein with strong sequence similarity to Nth-Eco. The functional analysis of this eukaryotic enzyme indicates that it is a homologue of
E.coli
endonuclease III.
Nth-Eco was prepared using the overproducing
E.coli
strain [lambda]N99 cI857/pHIT1 (
10
) and was kindly provided by L. Vilpo and R. D. Wood. Urease (
Canavalia ensiformis)
and terminal deoxynucleotidyltransferase (calf thymus) were purchased from
Boehringer Mannheim. [2-
14
C]thymidine 5'-triphosphate ([2-
14
C]dTTP) was purchased from ICN. Osmium tetroxide was from Sigma-Aldrich and potassium permanganate from BDH.
Identification of potential homologues of Nth-Eco was carried out using the BLAST (
25
) network service at the National Center for Biotechnology Information (NCBI).
Contiguous peptide sequences in the non-redundant protein database were scored against the inquiry sequence using
the BLOSUM62 homology matrix (
26
). Homology of the sequences retrieved from the BLAST search was analysed using
multiple sequence alignments (
27
).
The
E.coli
strain ED8 767 containing the cosmid clone ICRFc60D1130 from
S.pombe
was obtained from the Reference Library, ICRF (
28
). The cosmid DNA was purified using a Maxi-plasmid purification kit (Qiagen Ltd) and added as a template in a PCR
reaction to amplify the putative
S.pombe
nth
gene. The oligonucleotides 5'-CCATCCCTCATATGAGTAAAGACTACGGAAC-3' and 5'-CTATCTGGATCCTTGTCCAAAATTTACGGTC-3' were used to engineer
Nde
I and
Bam
HI restriction sites at the beginning and the end of the
nth
gene respectively. Forty amplification cycles were carried out using Pfu DNA
polymerase (Stratagene). In order to add a His
10
tag at the N-terminal end of the Nth-Spo protein, the amplified fragment was inserted into the pET-16b vector by digestion with
Nde
I and
Bam
HI restriction enzymes and ligation. The construct was designated pF2-NthP and the insert was sequenced on both strands by the Sanger method (
29
). The products of two independent PCR reactions were subcloned and sequenced to
detect any possible polymerase error during the amplification.
Plasmid pF2-NthP was used to transform
E.coli
expression strain BL21(DE3) (
30
) and a single transformant colony was inoculated into 1 l LB medium containing
carbenicillin (50 [mu]g/ml). The culture was incubated overnight at 30oC without shaking and then placed on a shaker at 30oC until the absorbance at 600 nm was 0.6. The culture was induced
by adding isopropyl- 1-thio-[beta]-D-galactopyranoside (IPTG) to 1 mM. After 2 h
induction, the cells were collected by centrifugation at 4000
g
for 30 min and the pellet frozen at -80oC.
The stored pellet was thawed and resuspended in 10 ml sonication buffer (SB; 50
mM HEPES-KOH, pH 8.0, 10% glycerol, 0.5 M NaCl, 0.1 mM EDTA, 8 mM [beta]-mercaptoethanol, 1 mM PMSF, 1.6 mM imidazole). Cells were
disrupted by sonication and the lysate was clarified by centrifugation. The
supernatant was mixed with 2 ml Ni
2+
-NTA resin (Qiagen Ltd) pre-equilibrated with SB buffer and stirred gently for 1 h. The resin
was then packed into a column and washed three times with 10 ml SB and four
times with 10 ml wash buffer (WB; 50 mM HEPES-KOH, pH 8.0, 10% glycerol, 0.1 M NaCl, 0.1 mM EDTA, 8 mM [beta]-mercaptoethanol) supplemented with 80 mM imidazole. Proteins
were eluted with 16 ml WB, 250 mM imidazole and 16 ml WB, 500 mM imidazole and collected in 1 ml fractions. An aliquot of each
fraction was analysed by SDS-PAGE and those containing a single band of the overexpressed protein were pooled and dialyzed for 8 h against 1 l dialysis buffer
(DB; 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, 1 mM DTT, 50% glycerol, 30 mM NaCl). The protein
preparation was aliquoted, frozen on dry ice and stored at -80oC.
All steps were carried out at 4oC or on ice. Protein concentrations were determined by the Bradford assay.
Denatured proteins were analysed by (SDS-)PAGE (10%) using low-range molecular weight standards (BioRad).
Activities similar to Nth-Eco have been found in most organisms investigated, including human cells
(
15
-
22
), but a eukaryotic
nth
gene has not yet been cloned. To identify potential eukaryotic Nth homologues,
the amino acid sequence of Nth-Eco (211 residues) was compared with the available databases using the
BLAST algorithm. An amino acid sequence of 355 residues from
S.pombe
(GenPept accession no. Z67961), having 27% identity and 49% similarity to Nth-Eco, was identified. This hypothetical protein would be encoded by a 1068
bp open reading frame in a genomic DNA sequence from
S.pombe
chromosome I, found in the cosmid ICRFc60D1130 (
28
). Two other protein sequences, one from
Saccharomyces cerevisiae
(Swiss-Prot accession no. P31378) and another from
Caenorhabditis elegans
(GenPept accession no. Z50874) (
22
) also showed high similarity to Nth-Eco.
Alignment of the
S.pombe
,
S.cerevisiae
and
C.elegans
sequences to the prokaryotic members of the Nth family is shown in Figure
1
. The eukaryotic sequences are longer than the prokaryotic ones, with conserved
residues through the whole length of the bacterial proteins. In addition to the
overall sequence similarity, the helix-hairpin-helix motif at the active site region of Nth-Eco is strongly conserved. The [4Fe-4S] cluster loop motif (Cys-X
6
-Cys-X
2
-Cys-X
5
-Cys) appears near the C-terminus in all the prokaryotic proteins, as well as in
C.elegans
. The
S.pombe
protein has a similar motif, although with seven residues instead of five
between the last two Cys residues. In contrast, the
S.cerevisiae
protein lacks this motif.
Nth protein from
E.coli
has been shown to have DNA glycosylase activity for a variety of thymine
derivatives, including urea and thymine glycol (
3
,
5
,
7
). To determine whether the Nth-Spo protein has a similar activity, we prepared double-stranded polydeoxynucleotide substrates containing rare scattered
urea or thymine glycol residues. Urea residues were generated from [2-
14
C]thymine residues by incubating a single-stranded poly(dA[middot][2-
14
C]dT) copolymer with KMnO
4
at basic pH (
4
), while thymine glycol formation was induced by treatment with OsO
4
(
32
). These treatments were followed by the addition of an equimolar amount of a
complementary poly(dT) strand. Figure
3
shows the activity of Nth-Eco and Nth-Spo on urea- and thymine glycol-containing DNA. Comparison of the prokaryotic and
eukaryotic enzymes with both types of lesions demonstrated that Nth-Spo protein is slightly more efficient on urea-containing substrates than Nth-Eco, whereas its efficiency in removing thymine glycol is
lower. Nth-Spo protein, like Nth-Eco, requires a double-stranded polydeoxyribonucleotide as substrate, since no DNA
glycosylase activity was detected with single-stranded oxidized polymers (data not shown). The released material from the KMnO
4
-treated poly(dA[middot][2-
14
C]dT) polymer with a complementary (dT) strand was identified as urea through
degradation to volatile material by urease (
5
). In contrast, all the ethanol-soluble radioactive material released from OsO
4
-treated polydeoxynucleotide was resistant to urease (data not shown).
The capacity of Nth-Spo to generate strand breaks in a variety of damaged plasmid DNA
substrates was investigated. Figure
4
shows the activity of Nth-Spo and Nth-Eco in incising supercoiled damaged DNA. None of the proteins caused
breaks in undamaged DNA. Both Nth-Spo and Nth-Eco enzymes showed similar incision activity with plasmid DNA
containing apurinic sites introduced by heat treatment (data not shown). Nth-Spo and Nth-Eco were also able to incise UV-irradiated plasmid DNA. After 30 min incubation, 10 nM of each
protein induced ~0.5 strand breaks/plasmid molecule in the substrate used. At lower protein
concentration Nth-Eco was slightly more efficient than Nth-Spo at incising the UV-irradiated plasmid (Fig.
4
, insert). In contrast, neither of the two proteins was able to release
detectable amounts of [
14
C]thymine derivatives from UV-irradiated poly(dA[middot][2-
14
C]dT)[middot]poly(dT) treated with a wide range of UV doses (0-500 kJ/m
2
; data not shown). This suggests that the eukaryotic Nth, as well as the
prokaryotic enzyme (
6
,
33
,
34
), incises UV-irradiated DNA mainly at cytosine and uracil hydrates. The relative
activity of the enzymes appeared different when [gamma]-irradiated plasmid DNA was used as substrate. At high protein
concentrations both enzymes were able to generate ~0.3 nicks/molecule in the [gamma]-irradiated plasmid, but at lower amounts Nth-Eco was clearly more efficient than Nth-Spo at incising the substrate (0.21 versus 0.05
nicks/molecule with 2.5 nM protein). The nicking activity of 1 pmol Nth-Spo protein on the different substrates was abolished by heating at 90oC for 5 min. All these enzymatic assays were repeated in at least two
separate experiments with reproducible results.
Nth protein sequences from several bacteria exhibit highly conserved regions in
their primary structure. They also share significant homology with the MutY
type of DNA glycosylases, including the Cys-X
6
-Cys-X
2
-Cys-X
5
-Cys sequence that ligates the [4Fe-4S] cluster (
35
). However, this domain has been shown to have a DNA binding role rather than an
obligatory catalytic function (
13
,
14
). Multiple alignment and clustering analysis show that the three eukaryotic
sequences reported here are more closely related to the Nth family than to the
MutY family. Their homology to the Nth family is particularly relevant in a
well-conserved region which includes amino acids known to participate in the
formation of the active site in Nth-Eco (Fig.
1
). One of those critical residues is Lys
120
(Nth-Eco), which is the most likely candidate for formation of the Schiff base
associated with AP lyase activity (
13
,
14
). This residue is absent in the MutY DNA glycosylase, which lacks such [beta]-elimination activity (
35
).
The present results indicate that the Nth-Spo protein, similar to its Nth-Eco counterpart, possesses both glycosylase and AP lyase activities.
The substrate specificity indicated that the eukaryotic enzyme recognizes
several types of pyrimidine damage, such as thymine glycol and urea. With
double-stranded poly(dA[middot][2-
14
C]dT)[middot]poly(dT) polymers containing damaged thymine residues, both enzymes were able to release thymine glycol and urea. The
eukaryotic protein was more efficient in removing urea than the prokaryotic
enzyme, whereas it was clearly less efficient in removing thymine glycol (Fig.
4
). These data agree well with the relative activity of the purified mammalian Nth
homologue, which is much more efficient at excising urea than thymine glycol
from DNA, although the latter base derivative is the more abundant DNA lesion
induced in mammalian cells after [gamma]-irradiation (
15
,
36
).
The presence of thymine glycol induces a significant and highly localized
alteration in the structure of DNA. In contrast to 5,6-dihydrothymine or aldehydic abasic sites, which induce a relatively small
distortion in the DNA structure (
37
-
39
), thymine glycol and the opposite base are both extrahelical (
39
,
40
). It is likely that this structural perturbation provides a readily
recognizable target for different DNA repair enzymes, including the UvrABC
nuclease complex (
41
,
42
), which particularly removes bulky groups that produce extensive distortion in
DNA structure (
43
). Recently another
E.coli
enzyme has been identified, endonuclease VIII, which acts as a DNA glycosylase
and cleaves DNA containing pyrimidine oxidation products, such as thymine
glycol, urea, dihydrothymine, 5-hydroxycytosine, 5-hydroxyuracil and abasic sites (
44
). This enzyme shares homology with the faPy DNA glycosylase family but not with
the Nth family (
45
). Interestingly, endonuclease VIII excises thymine glycol with a much higher
efficiency than the Nth-Eco protein (
45
). The Nth-Eco protein may play a main role in removing ring fragmented pyrimidine
oxidation products such as urea and
N
-substituted urea derivatives rather than 5-6 ring saturated pyrimidine derivatives, such as thymine glycol,
which generate major distortion of the double helix.
We have shown that
S.pombe
has a protein structurally and functionally homologous to Nth-Eco by subcloning the gene and analysing the enzymatic function of the
product
in vitro
. A sequence very similar to the
S.pombe
nth
gene has also been found in
S.cerevisiae
and in the higher eukaryote
C.elegans
. According to the present results and given the strong similarity of the
relevant
C.elegans
(61% similarity, 45% identity) and
S.cerevisiae
(58% similarity, 37% identity) sequences to the
S.pombe
protein, we predict that they also have a Nth-like activity. The high degree of similarity found between the
C.elegans
,
S.cerevisiae
,
S.pombe
and bacterial Nth proteins, and the presence in mammalian cells of a very
similar enzyme, provide strong evidence for the general importance of Nth-mediated base excision repair against DNA damage generated by active
oxygen.
We thank The Reference Library System, ICRF for providing the ICRFc60D1130
clone. We also thank Barbara Sedgwick and Rick Wood for critical reading of the
manuscript. T.R.A. was supported by an EC training fellowship under the Human
Capital and Mobility Programme and C.A. by an EMBO long-term fellowship.
Polydeoxyribonucleotide substrates.
The poly(dA[middot][2-
14
C]dT) copolymer containing 97% dAMP residues (
4
) was synthesized from dATP and [2-
14
C]dTTP (63 Ci/mmol) with terminal deoxynucleotidyltransferase, filtered through
a Centricon C3 (Amicon) and washed three times with 10 mM Tris-HCl, 1 mM EDTA, pH 8.0 (TE) or 0.1 M NaHCO
3
, pH 9.0. A poly(dA) chain containing scattered
14
C-labelled urea residues was prepared by KMnO
4
treatment of a poly(dA[middot][2-
14
C]dT) copolymer as described (
4
). To produce thymine glycol residues in poly(dA[middot][2-
14
C]dT), the single-stranded copolymer (25-50 [mu]g, 10
6
c.p.m.) in 0.1 ml TE was incubated at 0oC for 15 min in the presence of 1.2% OsO
4
and
0.1 N NH
3
. The osmium tetroxide was then removed by passing the polymer twice through a
Sephadex G-25 spun column. Urea- or thymine glycol-containing poly(dA) were mixed with an equimolar amount of
poly(dT) to generate double-stranded polydeoxyribonucleotides.
Plasmid substrates.
Plasmid pBluescript KS+ (Stratagene) was purified from the
E.coli
DH5[alpha] host strain using a Maxi-plasmid purification kit (Qiagen Ltd). For [gamma]-irradiated plasmid DNA, pBluescript (800 [mu]g/ml in TE) was exposed on ice to a
137
Cs [gamma]-ray source at a dose rate of 2.8 Gy/min (total dose 100 Gy). For a
UV-damaged substrate, pBluescript (50 [mu]g/ml in TE) was irradiated (254 nm) with a dose of 4000 J/m
2
under a G15T8 15W germicidal lamp (Sankyo Denki, Japan). After UV- and [gamma]-irradiation, the covalently closed and open circular forms of DNA were
separated by ethidium bromide/CsCl density gradient centrifugation. Non-irradiated plasmid was subjected in parallel to the same procedure and
used as a control in assays
DNA glycosylase assay.
A reaction mixture (100 [mu]l) containing 40 mM HEPES-KOH, pH 8.0, 0.1 M KCl, 0.5 mM EDTA, 0.5 mM DTT, 0.2 mg/ml BSA and
0.4 [mu]g double-stranded polydeoxyribonucleotide with
14
C-labelled modified thymine residues (1200 d.p.m.) was incubated at 37oC for 0-30 min with different concentrations of purified proteins.
After incubation, 10 [mu]l 2 M NaCl, 10 [mu]l heat-denatured calf thymus DNA (2 mg/ml) and 300 [mu]l cold ethanol were added. The DNA was precipitated at -20oC, centrifuged at 17 500
g
for 15 min and the ethanol-soluble radioactive material was measured in 300 [mu]l supernatant by scintillation counting. Degradation of urea was
achieved with urease (2.5 U) at 37oC for 30 min.
Nicking assay.
A reaction mixture (20 [mu]l) containing 40 mM HEPES-KOH, pH 8.0, 0.1 M KCl, 0.5 mM EDTA, 0.5 mM DTT, 0.2 mg/ml BSA and
500 ng purified closed circular plasmid DNA was incubated at 37oC for 30 min with increasing concentrations of purified proteins. Reactions
were stopped by heating at 90oC for 5 min and the mixtures loaded onto a 1% agarose gel. Photographic
negatives of the ethidium bromide stained agarose gel were scanned with a
Molecular Dynamics densitometer. The average number of nicks per plasmid
molecule made by each enzyme was estimated from the fraction of remaining
covalently closed circular DNA by the Poisson distribution. The greater
fluorescence of nicked circular DNA over closed circular DNA was taken into
account in all quantifications (
31
).
REFERENCES
Return