ABSTRACT
Double-stranded (ds) oligodeoxynucleotides (29mers) containing an O
6
-ethylguanine (O
6
-EtGua) flanked 5
'
and 3
'
by different bases (5
'
..TGT..3
'
; 5
'
..CGG..3
'
, 5
'
..GGT..3
'
; 5
'
..GGG..3
'
; 5
'
..GGA..3
'
) were synthesized to investigate the binding and repair characteristics of
recombinant human O
6
-alkylguanine-DNA alkyltransferase (AT)
in vitro
. The apparent association constant (K
A(app)
) of AT to the oligomers and the repair rate constant for O
6
-EtGua (k) respectively, were determined by gel retardation and a
monoclonal antibody-based filter binding assay. When ds- or single-stranded (ss) oligomers with or without O
6
-EtGua were used, no major differences in K
A(app)
values were observed with either substrate: K
A(app)
values for native AT were 7.1 and 8.4*10
5
M
-1
respectively, for unmodified and [O
6
-EtGua]-containing ds-oligomers. The corresponding values for ss-oligomers were 1.0 and 4.9*10
5
M
-1
. The N-terminal first 56 amino acids of AT only exert a limited influence on DNA
binding; the K
A(app)
values for an N-terminally truncated AT protein (1.1*10
5
M
-1
) and native AT were of the same order. Moreover, K
A(app)
was hardly affected by Cys
145
-methylated AT (2.0*10
5
M
-1
). The k-values (6.5-11.5*10
6
M
-1
s
-1
) were not significantly dependent on nucleotide sequence. k-values of 5.3 and 4.0*10
6
M
-1
s
-1
respectively, were obtained with the N-terminally truncated AT protein and for repair of the postreplicative
mispair [O
6
-EtGua]: T by native AT. The low K
A(app)
, the negligible influence on K
A(app)
guanine-O
6
of ethylation, and the minor modulation of K
A(app)
and k by varying the bases flanking O
6
-EtGua, all indicate that the binding of AT to DNA is non-specific and mediated mainly by ionic interactions [reduced K
A(app)
and k-values at increased ionic strength]. Surplus DNA reduces the rate of O
6
-EtGua repair in ds-oligomers by competitive binding of AT molecules. The reaction
mechanism of AT with DNA
in vivo
requires further investigation.
In vitro
or
in vivo
exposure to N-nitrosamides, such as N-ethyl-N-nitrosourea (EtNU) or N-methyl-N-nitrosourea (MeNU), leads to the
formation of seven N- and five O-alkylation products (including alkylphosphotriesters) in cellular
DNA (
1
-
3
). The latter include O
6
-alkylguanines (O
6
-AlkGua) which can cause G:C -> A:T transition mutations (
4
-
6
), as observed, e.g., at the central guanine of codon 12 (GGA) of the
H-ras
gene in rat mammary and skin tumors induced by MeNU (
7
,
8
). The relative rate of guanine-O
6
alkylation in DNA upon exposure to a given N-nitrosamide is dependent on sequence context and chromatin structure (
9
-
12
). A central guanine is most permissive for O
6
-alkylation by MeNU or EtNU when flanked 5' by a purine, and least permissive when flanked 5' by thymine or 5-methylcytosine (
9
,
12
).
O
6
-AlkGua in DNA is repaired by the repair protein O
6
-alkylguanine-DNA alkyltransferase (AT; EC 2.1.1.63;
13
) and, with increasing size of the alkyl residue, also via the nucleotide
excision repair pathway (
14
). AT catalyzes the transfer of the alkyl group to Cys
145
at the active site of the protein; the formation of S-alkylcysteine irreversibly inactivates the AT molecule (
13
).
The 39 kDa Ada protein, a bacterial AT homolog, contains two cysteine residues
active in DNA repair: Cys
321
for O
6
-AlkGua and O
4
-alkylthymines (
15
), and Cys
69
for alkylphosphotriesters (
16
). Unlike the human and rodent AT proteins, Ada is not inactivated by O
6
-benzylguanine (
17
), and its expression is inducible by chronic low-level exposure to alkylating agents (adaptive response;
18
,
19
). Ogt, a second non-inducible bacterial AT protein is homologous to the C-terminal domain of Ada and the known human and rodent ATs (
20
).
All species thus far examined harbor ATs with highly conserved C-terminal amino acid sequences (
13
,
21
-
23
). Complete homology is observed around the active center (Pro-Cys*-His-Arg-Val/Ile) (
24
). X-ray analysis of the crystal structure of a 178 amino acids fragment of
Ada, which includes the active center, has revealed two distinct domains with
novel topology (
25
). The C-terminal domain contains the active site sequence and, at considerable
distance, a `helix-turn-helix variant' DNA-binding motif possibly mediating non-specific binding to DNA. The reactive Cys* lies in a
buried position that requires a conformational change of the protein to perform
the nucleophilic competition for the target electrophile (
25
). Neither DNA-binding motifs nor nuclear localization signals have thus far been
described in the amino acid sequences of mammalian AT proteins (
26
).
In the present study we have investigated the possible sequence specificity of
DNA binding and O
6
-EtGua repair by recombinant human AT (21.7 kDa) before and after
mispairing of O
6
-EtGua with thymine during DNA replication. We have used synthetic double-stranded (ds) deoxynucleotides containing a single O
6
-ethylguanine (O
6
-EtGua) flanked by different bases, and active AT, AT inactivated by methyl-group transfer, as well as N-terminally truncated AT (lacking the first 56 amino acids).
These analyses complement studies by other groups, probing the immediate
vicinity of the substrate binding site and investigating the chemistry of alkyl
group transfer with substrate analogs (
27
-
32
), as well as studies that have used site-directed mutagenesis to evaluate the role of individual amino acids in the
active center of AT (
33
,
34
) and elsewhere in the protein (
35
-
37
).
Oligodeoxynucleotides (29mers) were synthesized on an ABI 381 DNA Synthesizer
(Applied Biosystems, Weiterstadt, Germany; Table
1
). Regular monomer building units were purchased from Applied Biosystems and [O
6
-EtGua]-phosphoramidite from Chemogen (Konstanz, Germany). [O
6
-EtGua]-containing oligomers were deprotected for 72 h at room temperature
to avoid the formation of 2,6-diaminopurine (
38
). Oligomers were purified by HPLC using reversed-phase cartridges (Nova Pack C-18; Waters, Eschborn, Germany). Elution was performed at a flow rate
of 1.5 ml/min (linear gradient; 90% A, 10-100% B in 45 min) with buffers A [100 mM triethylammonium acetate
(TEAAc), pH 5.0] and B (100 mM TEAAc, 35% acetonitrile). Fractions with the
highest absorption values at [lambda] = 254 nm were collected. To verify the incorporation of O
6
-EtGua, oligomers were enzymatically hydrolyzed to 2'-mono-deoxynucleosides (
39
) and analyzed by HPLC.
Ds-oligomers were generated by incubating [O
6
-EtGua]-containing single-strands with excess (10%) complementary single-strands in a buffer containing 25 mM morpholinopropane
sulfonic acid (MOPS), pH 7.4; 0.5 mM EDTA; for 5 min at 50oC, and subsequently for 20 min at 37oC. The temperature was then slowly lowered to 4oC and kept constant overnight. At least >= 90% of oligomers were present in ds-conformation. The strand opposite O
6
-EtGua contained a 5'-protruding end of five bases, to facilitate labeling with [[gamma]-
32
P]ATP (
40
).
AT cDNA was cloned from a human liver cDNA library by colony hybridization using
published cDNA sequences (
41
). Native and truncated AT cDNAs were ligated into the pQE-9 vector (Qiagen, Hilden, Germany, 40), and transfected into
Escherichia coli
strain M15 (Qiagen). The pQE-9 plasmid encodes a 6 x His-affinity tag (fused to the N-terminus of AT) which can be removed from AT enzymatically.
Bacteria were grown in 2 l of LB medium (100 [mu]g/ml ampicillin and 25 [mu]g/ml kanamycin) at 37oC until an OD
600
= 0.9-1.1 was reached (corresponding to 10-20 mg of AT-protein/l). Protein expression was induced by addition of
100 [mu]M isopropyl-D- thiogalactopyranoside for 2 h. Cells were harvested by
centrifugation at 4000
g
for 20 min and pellets were stored at -70oC.
Following resuspension in 40 ml buffer A (50 mM NaH
2
PO
4
, Na
2
HPO
4
, pH 8.0; 300 mM NaCl; 10% glycerol; 10 mM mercaptoethanol), bacteria were lysed
on ice by sonication. Subsequent operations were carried out at 4oC. To inhibit proteases, 1 mM Pefabloc SC (Boehringer Mannheim, Mannheim,
Germany) was added and the solution was centrifuged for 30 min at 30 000
g
. The supernatant was applied to a Ni
2+
-nitrilotriacetic acid (Ni-NTA)-agarose column equilibrated with buffer A. After washing with
buffer A recombinant AT was eluted with buffer B (50 mM NaH
2
PO
4
, Na
2
HPO
4
, pH 7.6; 200 mM imidazole; 300 mM NaCl; 10% glycerol; 10 mM [beta]-mercaptoethanol). Fractions containing AT were pooled, and the buffer was exchanged on PD-10 columns (Pharmacia-LKB, Freiburg, Germany) equilibrated with buffer C [25
mM MOPS, pH 8.0; 50 mM NaCl; 1 mM EDTA, 10% glycerol; 10 mM dithioerythrol
(DTT)]. To remove the His tag, factor Xa (Boehringer Mannheim) was added (20 U/15 mg AT), and the sample was incubated at
0oC for 18 h. Under these conditions >90% of the molecules were cleaved. The
sample was sterile-filtered and diluted four times with buffer D (50 mM MOPS, pH 7.1; 20 mM
NaCl; 1 mM EDTA; 10% glycerol; 10 mM DTT). The cleavage products were separated
on a Mono S column (Pharmacia-LKB). AT was eluted with a linear NaCl gradient (20-400 mM).
The specific activity of recombinant AT for repair of O
6
-EtGua was determined as the ratio of the concentration of active AT, as
resulting from the kinetic analyses, and the total AT concentration, as
measured photometrically using [epsilon] = 28 900 M
-1
cm
-1
at [lambda] = 280 nm. The specific activity of AT was calculated to be [approx]60%; it remained constant for months when the AT protein was stored
at -70oC.
Forty percent of inactive (non-methylated) AT-protein does not interfere with the repair process because AT
remains monomeric in solution (
44
; and our own results) and, according to the law of mass action and the low
concentrations of oligomers and AT (Fig.
1
), the frequency of AT-oligomer complexes is very low. Hence, the probability of two AT molecules
competing for one oligomer tends towards zero.
Constant amounts of native or modified AT (20 pM) were incubated for 1 h at 37oC with increasing amounts (10-100 pM) of
32
P-labeled, [O
6
-EtGua]-containing ds-oligomers in a total volume of 100 [mu]l [50 mM Tris-HCl, pH 7.6; 10 mM DTT; 1 mM EDTA; and 200 [mu]g/ml bovine serum albumin (BSA)]. The repair
reaction was stopped by adding 10 [mu]l 1 M NaCl and an excess (10 [mu]g) of anti-[O
6
-ethyl-2'-deoxyguanosine] monoclonal antibody (MAb) ER-6 (
42
). After incubation for 1 h at room temperature, [O
6
-EtGua]-oligomer-Mab complexes were separated from repaired oligomers by
filtration through pieces of nitrocellulose membrane BA 85 (Schleicher & Schuell, Dassel, Germany). Membranes were washed twice with 2 ml buffer (50 mM
Tris-HCl, pH 7.6; 1 mM EDTA; 100 mM NaCl). [O
6
-EtGua]-containing oligomers complexed to the MAb retained on the filter
were quantified by determining their
32
P-radioactivity in 4 ml scintillation cocktail (Ready Safe; Beckman, Munich,
Germany) in a liquid scintillation spectrometer (Wallac; Pharmacia-LKB).
Because reaction with O
6
-EtGua in DNA inactivates the AT molecule, repair by AT follows second-order kinetics. The repair rate constant, k, and the initial
concentration of active AT were determined by means of an iterative computer
program (
43
; available from P. Nehls).
To examine the influence of ionic strength (NaCl concentration) on k, the
initial repair rate constants, k
i
, which approximate k within the limits of short reaction times, were determined
(
44
). The influence on k of increasing concentrations of calf thymus DNA, plasmid
DNA (3.4 kb), or ds-oligomers was also investigated.
Constant amounts of
32
P-labeled unmodified or [O
6
-EtGua]- containing oligomers (300 pM) were incubated for 30 min at 4oC with increasing amounts of native AT, N-terminally truncated AT or Cys
145
-methylated AT (0.2-20 [mu]M), in 50 mM Tris-HCl, pH 7.6; 10 mM DTT; 1 mM EDTA; and 200 [mu]g/ml BSA. Free oligomers were separated from protein-bound oligomers by electrophoresis on 4%
polyacrylamide gels (TBE-buffer system;
40
) at 100 V for 1 h at 4oC. All buffers were cooled to 4oC before use. After electrophoresis, gels were vacuum dried, cut into
slices, and analyzed by liquid scintillation spectrometry to determine the
ratio of free to AT-bound oligomers.
Apparent dissociation constants, K
D(app)
, for the interaction of oligomers with native or modified AT were determined by
fitting the experimental data to the equation:
KD(app) = (f x B0) / (1 - f)
(
1
)
where f is the fraction of free oligomer, and B
0
is the initial AT concentration (
45
).
The number of base pairs (bp), n, involved in the binding of AT was determined
by a Scatchard plot v/L [mol/l] against v. v represents the `binding density'
[mol of bound protein/mol of total oligomer bases], i.e. mol of bound
oligomer/mol of oligomer multiplied by the number of bases. Since the
concentration of L, free AT, largely exceeded that of the oligomer, L
approximated B
0
at equilibrium.
If the ligand covers >= 2 bp and potential binding sites overlap, the slopes and intercepts of this
plot take on meanings different from those originally introduced by Scatchard (
46
). The intercept on the v/L axis is equal to the intrinsic dissociation
constant, K
i
, and within the limits of low binding density the slope of the line, b, is
equal to -K
i
(2n - 1). Analyzing a finite oligomer instead of an infinite lattice, the
correction factor is approximately (N - n + 1)/N (N being the number of base-pairs per oligomer). The dissociation constant per binding site, K
D
, is the product of K
i
and n. The apparent association constant K
A(app)
is 1/K
D(app)
. K
D(app)
values were determined by the program Sigma Plot (Jandel Scientific, Erkrath,
Germany).
Ds-oligomers (29mers) were used to determine the rate constants for repair of
O
6
-EtGua by recombinant human AT as a function of DNA sequence context. DNA
sequences were designed to represent all possible combinations of purines and
pyrimidines flanking a single O
6
-EtGua residue. Since ions reduce repair rates, measurements were carried
out at low ionic strength. For determination of the repair rate, AT was
incubated for 1 h at a constant concentration with varying concentrations of a
given oligomer. This experimental alternative to conventional kinetic analysis
has previously been used by Dolan
et al
. (
47
). Its advantage is that both k and the concentration of active AT can be
determined simultaneously by computer. A representative curve is shown in
Figure
1
.
Repair rate constants were not significantly influenced by the nature of the
bases flanking the O
6
-EtGua residue, with k-values ranging between 6.5 and 11.5 * 10
6
M
-1
s
-1
(Table
1
). When paired with thymine, O
6
-EtGua was repaired at a slightly lower rate (4.0 * 10
6
M
-1
s
-1
), but this difference is statistically not significant. The 6* His tag did not influence repair rates (Table
1
). N-terminal truncation of the AT molecule did not markedly influence the k-value (5 * 10
6
M
-1
s
-1
). Moreover, as native AT, the truncated AT was inactivated by O
6
-benzylguanine (data not shown). Increasing the NaCl concentration from 10
to 100 mM reduced the log k
i
value from 6.9 to 5.8. Similary, the addition of surplus DNA resulted in lower
log k
i
values (data not shown).
Table 1
.
The repair of O
6
-AlkGua in DNA by AT requires several steps: recognition and binding of the
alkyl group on the O
6
-atom of guanine, and its transfer to Cys
145
of AT. To identify the rate limiting step, the DNA concentration was increased
at a constant ratio of O
6
-EtGua/Gua (data not shown). At low DNA concentrations (0-60 nM deoxynucleotides) k-values increased in parallel to DNA concentration. If at
constant AT concentration alkyl group transfer were the rate-limiting step, then the apparent repair rate would be the product of k
multiplied by the substrate concentration. In contrast, when the repair rate is
reduced at DNA concentrations >60 mM, alkyl group transfer will no longer be
rate limiting. Rather, due to the law of mass action, a larger proportion of AT
molecules become bound to unmodified DNA and the overall repair activity of AT
is thus reduced. To test for this effect of competitive binding, surplus ds-DNA molecules of different lengths (29mers; plasmid DNA of 3.4 kb; calf
thymus DNA) were added to the samples. The extent of inhibition proved to be
dependent only on the molar concentration of base pairs, not on the lengths of
the DNA molecules added (data not shown). Thus, in the presence of a large
excess of DNA recognition and binding of the alkyl group become rate limiting (
48
).
The recognition of O
6
-alkyl groups could be facilitated by lateral diffusion of AT molecules on
the DNA surface; i.e., AT initially diffuses to DNA, binds nonspecifically and
then rapidly `scans' adjacent sites until the alkyl group is located, or just
dissociates from DNA. The scanning of an AT molecule of DNA would resemble the
movement along a continuous cylinder. This should be reflected by the behaviour
of k as a function of ionic strength (
48
). If there were purely electrostatic interaction, ions would be expected to
exert a shielding effect, resulting in a linear relation between log k and log
[NaCl] in the 29mers. This is indeed the case (log-log plots not shown). Plotting published data (fig. 7 in ref.
44
) for O
6
-methylguanine (O
6
-MeGua) in calf thymus DNA in the same way also results in a near-linear relation with the same slope and almost identical k-values.
The K
A(app)
for the binding of AT to DNA was determined by a quantitative gel retardation
assay (see Materials and Methods). Two methods were used for the analysis of
the data (see Materials and Methods; Fig.
2
A and B). The results were equal within a factor of 2, with differences varying
statistically. The data obtained using equation
1
are shown in Tables
2
and
3
.
Unrepaired O
6
-AlkGua in DNA can cause G:C -> A:T transition mutations by misreplication (
5
). The probability of O
6
-AlkGua formation in DNA by N-nitrosamides (MeNU, EtNU) is dependent on sequence context (
12
). Alkylation on guanine-O
6
is favored when a central guanine is flanked 5' by a purine; 5' thymine and 5-methylcytosine have the opposite effect (
9
-
12
,
50
). In the present study, we have investigated whether the efficiency of repair
of O
6
-AlkGua by AT is similarly influenced by the sequence context. Ds-oligomers (29mers) containing a single O
6
-EtGua flanked by different neighboring bases within an otherwise constant
sequence context were used as substrates for recombinant human AT.
Since short oligodeoxynucleotides (<10 bp) are repaired by AT at a lower rate (
51
,
52
) the ds-oligomers used in this study were 29 nt in length; i.e., close to three
helical turns. This is sufficient for AT binding, which requires 8-10 bp, and ensures that the oligomers remain double-stranded at 37oC (Table
1
). Because the Z-DNA conformation is inhibitary for AT (
53
), a random oligomer sequence was chosen to insure the B-conformation.
The rate constants for repair of O
6
-EtGua by AT ranged from k = 4.0-11.5 * 10
6
M
-1
s
-1
. Similar k-values have been published by others (
54
). The second-order association rate constant for a 203 bp lac DNA fragment with its
repressor is ~4 * 10
9
M
-1
s
-1
(
48
). This association is only limited by diffusion. The repair of O
6
-EtGua by AT must, therefore, be slowed down by other mechanisms. k-values were not significantly affected by sequence context. The
postreplicative [O
6
-EtGua]:T base-pair was a good substrate for recombinant human AT, too.
Using HT29 cell extracts, Dolan
et al
. (
50
) observed only a slight sequence dependence of k-values for repair of O
6
-MeGua. Repair rate constants were of the order of 0.1 * 10
6
M
-1
s
-1
. This low value is probably due to the use of shorter self-complementary oligomers (12mers). k-values for repair of O
6
-MeGua by Ada and mammalian ATs varied by a factor of 4 at most when a 5' flanking guanine was exchanged for adenine in the 12mers (
10
).
In contrast, Georgiadis
et al
. (
56
), using Ada and [O
6
-MeGua]- containing ds-oligomers, reported k-values varying by factors of <= 20 depending on sequence context. This is the only
report indicating a strong influence of nearest neighbors; even next-neighbor effects were detected. The k-values published by these authors are 1 or 2 orders of magnitude
lower than those reported by others (
13
,
50
) and those determined in the present study.
Interestingly, the Ada protein, contrary to AT, is insensitive to inactivation
by O
6
-benzylguanine (
17
,
57
,
58
). This has been attributed to a steric restriction of the Ada binding site and
may explain the observation that the differences between k-values for repair of O
6
-MeGua versus O
6
-EtGua are far greater with Ada than with mammalian ATs (
13
). Moreover, K
A(app)
for DNA binding is an order of magnitude higher for Ada than for human AT (
59
).
Gaffney and Jones (
60
) have published a thermodynamic comparison of base pairs varying the base
opposite O
6
-MeGua. The relative order of stability was C > A > G > T. Variation of
nearest-neighbors covered a much smaller range in T
max
(within 1o of T
m
of the oligomer) than the different base-pairs, and showed little sequence dependence. Hence, no significant
influence of nearest-neighbors is to be expected from a thermodynamic point of view.
The influence of Ada on mutational spectra has been measured in adapted and non-adapted
E.coli
carrying the pSV2gpt plasmid following exposure to N-methyl-N'-nitro-N-nitrosoguanidine (
61
). The distribution of mutations in the coding region of the
hprt
gene has been compared in human fibroblasts with intact AT activity and after
inhibition of AT by O
6
-benzylguanine (
62
). In all of these studies nearly identical results were obtained: (i) the same
types of mutations; (ii) the same sequence specificity of mutagenesis; and
(iii) occurence of the the majority of mutations in the coding DNA strand.
Similar results were obtained with MeNU (
63
).
The reduction of AT repair activity
in vitro
in the presence of excess of unmodified DNA is well documented (
44
,
64
). With the present [O
6
-EtGua]-containing ds-oligomers, inhibition of AT by unmodified DNA sequences is
negligible due to the high ratio of O
6
-AlkGua residues to normal base-pairs. Here, transfer of the alkyl group to Cys
145
in AT is rate-limiting in the repair process. With increasing concentrations of surplus
unmodified DNA, the recognition and binding of O
6
-AlkGua become rate-limiting due to competitive DNA binding.
How then are relatively few O
6
-AlkGua residues recognized by AT molecules in the presence of a large
excess of DNA not containing this alkylation product? Does AT diffuse directly
and freely to the target or rather act in a processive manner, i.e., by lateral
diffusion of AT molecules on DNA? Provided the recognition and binding of the O
6
-alkyl group on guanines by AT are rate-limiting, a model elaborated by Winter
et al
. (
48
) may be applied. In a two-step process, AT would bind to DNA weakly and non-specifically, and then scan DNA for O
6
-AlkGua residues by lateral diffusion. This facilitated search would
compensate for the competition by intact DNA. The effect would become less
pronounced with short DNA sequences, and disappear in the case of oligomers.
However, there is no evidence for lateral diffusion of AT (see above; and fig.
7 in ref.
44
): (i) the k-value for O
6
-MeGua in ds-DNA (
44
) is as low as the values determined here for O
6
-EtGua in oligomers. Hence, there is no kinetic gain from long stretches of
DNA; (ii) surface sliding would cause a complex dependence of k on ionic
strength as illustrated by Winter
et al
. (fig. 2 in ref.
48
). In reality, there is an approximately linear relation between log k and log
[NaCl] in the case of oligomers and ds-DNA. Hence, binding of AT to DNA appears to be random and AT molecules
presumably `jump' from one binding site to the next without lateral diffusion.
The K
A(app)
values obtained in the present study are in good agreement with those
calculated by Chan
et al
. (
65
) for the interaction of human AT with unmodified ds-DNA (4.7 * 10
5
M
-1
). A similar value (9 * 10
5
M
-1
) has been reported by these authors for the binding of AT to DNA reacted with
MeNU though `most' of the O
6
-MeGua residues were removed under their assay conditions.
The present experiments have shown that AT does not bind specifically to O
6
-EtGua residues in DNA, and that increased NaCl concentrations do not
enable AT to discriminate between modified or unmodified oligomers. Analyses of
our quantitative gel retardation data according to McGhee and von Hippel (
46
) revealed that 8-10 bp are involved in AT binding, in agreement with published results (
25
,
65
). The binding affinity of AT to ss-oligomers was slightly reduced in comparison to ds-oligomers. The preferential repair of ds-DNA (
55
) may be due to a differential presentation of O
6
-EtGua in ds- versus ss-DNA. AT inactivated by methyl group transfer to Cys
145
exhibited nearly unchanged binding affinty to native DNA (Table
3
), in agreement with earlier studies indicating that the binding of human AT to
DNA is not affected by the methylation status of the active center (
44
), and with the prediction that no major conformational changes should occur
upon alkylation of the Ada protein (
25
).
X-ray analyses of the crystal structure of the C-terminal half of Ada, which contains the active center, have
revealed that this part of the molecule consists of two domains (
25
), the C-terminal domain being highly conserved (
24
). Removal of 36, but not of 31, C-terminal amino acids from human AT led to a complete loss of repair
activity (
66
-
68
). A 14-amino acid peptide carrying the active site of AT (Lys-Val-Pro-Ile-Leu-Ile-Pro-Cys*-His-Arg-Val-Val-Cys-Ser) did not eliminate alkyl groups from the O
6
-atom of guanine in DNA (our unpublished data), suggesting that a correctly
folded C-terminal domain is a prerequisite for catalytic activity.
The N-terminally truncated AT (amino acids 1-56) used in the present study showed a 7-fold lower affinity for [O
6
-EtGua]- containing ds-oligomers than native AT, and a 2-fold lower affinity compared with Cys
145
-methylated AT. However, the repair rate constant, k, was not affected by N-terminal truncation. The N-terminal amino acid sequence is not conserved to the extent
found in the C-terminal domain (
24
). The N-terminal domain of AT may thus not be essential for repair, and its
function remains unclear. It might contain recognition sequences for other
nuclear proteins, and, in eukaryotic cells, be responsible for the
translocation of AT to the nucleus. Nuclear localization sequences of AT have
thus far not been detected (
26
); however, Ada expressed at high levels in NIH-3T3 cells is not efficiently transported into the nucleus (
69
), possibly due to differences in the N-terminal domains of Ada and AT. The simplest function of the N-terminal domain may be to keep the AT molecule soluble: a
hydrophilicity plot shows that primarily the first 45 amino acids are
hydrophilic (
70
). While in
E.coli
about half of the native recombinant AT molecule carrying the His tag are
expressed in soluble form (the other half as inclusion bodies), the N-terminally deleted AT protein forms inclusion bodies exclusively. This is
in accordance with Crone
et al.
(
68
) who reported that N-terminal deletion of 10 or 19 amino acids led to a significant reduction
of recombinant AT repair activity in
E.coli
cell extracts, due to a substantial decrease in the amount of AT protein: most
of the AT protein had probably formed insoluble inclusion bodies.
In conclusion, the DNA binding affinity of recombinant human AT is low, non-specific, and mainly due to ionic interactions, hardly permitting a
distinction between guanine alkylated in the O
6
-position and its intact counterpart. No DNA-binding motifs have thus far been detected in the primary sequence (
26
). The folding pattern of the C-terminal domain of Ada reveals a `helix-turn-helix variant' which is far away from the active center of
the protein; it is not clear whether this motif mediates generalized binding to
DNA (
25
). Binding of AT to DNA appears to be random; since AT molecules are incapable
of migrating along DNA, they presumably `jump' from one binding site to
another. If a binding site carries an O
6
-AlkGua, it is repaired at once. Since both the recognition and binding of
AT to O
6
-AlkGua in DNA are rather inefficient, they may be mediated by other
proteins
in vivo
. Binding to damage recognition proteins could be a function of the N-terminal part of AT. Such interactions might result in differential repair
of O
6
-EtGua in transcribed and nontranscribed gene sequences (
71
) and account for some sequence specificity
in vivo
. Factors targeting AT to O
6
-alkylation products in the DNA of transcriptionally active genes (e.g.,
transcription repair coupling factor;
72
) have not yet been identified in mammalian cells, but are worth looking for.
This work was supported by Dr Mildred Scheel Stiftung für Krebsforschung (W 87-92 Ra5) and Deutsche Forschungsgemeinschaft (SFB 354). We are
indebted to Mrs Anke Galhoff for excellent technical assistance; to Dipl.-Ing. Klaus Lennartz for installation of the computer program, and to a
referee for helpful comments.
Oligonucleotide sequence
a
Repair rate constant (k)
* 10
-6
[M
-1
s
-1
]
5'...T
G
T ...3'
10.6 +- 4.5
1
3'...ACA...5'
5'...C
G
G...3'
9.7 +- 1.0
2
3'...GCC...5'
5'...G
G
T...3'
6.5 +- 1.3
3
3'...CCA...5'
5'...G
G
G...3'
10.3 +- 2.3
4
3'...CCC...5'
5'...G
G
A...3'
11.5 +- 5.6
3'...CCT...5'
5'...G
G
A...3'
4.0 +- 1.7
5
3'...CTT... 5'
5'...C
G
G...3'
5.3
b
3'...GCC...5'
5'...C
G
G...3'
10.3
c
3'...GCC...5'
REFERENCES
Return