ABSTRACT
The 552 amino acid vaccinia virus DNA ligase consists of three structural
domains defined by partial proteolysis: (i) an amino-terminal 175 amino acid segment that is susceptible to digestion with
chymotrypsin and trypsin; (ii) a protease-resistant central domain that contains the active site of nucleotidyl
transfer (Lys-231); (iii) a protease-resistant carboxyl domain. The two protease-resistant domains are separated by a protease-sensitive interdomain bridge from positions 296 to 307.
Adenylyltransferase and DNA ligation activities are preserved when the N-terminal 200 amino acids are deleted. However, the truncated form of
vaccinia ligase has a reduced catalytic rate in strand joining and a lower
affinity for DNA than does the full-sized enzyme. The 350 amino acid catalytic core of the vaccinia ligase is
similar in size and protease-sensitivity to the full-length bacteriophage T7 DNA ligase.
The ATP-dependent DNA ligases catalyze the joining of 5' phosphate-terminated donor strands to 3' hydroxyl-terminated acceptor strands via three sequential
nucleotidyl transfer reactions (
1
,
2
). In the first step, attack on the [alpha]-phosphate of ATP by ligase results in liberation of pyrophosphate
and formation of a covalent intermediate in which AMP is linked to the [epsilon]-amino group of a lysine. The nucleotide is then transferred to the
5' end of the donor polynucleotide to form DNA-adenylate, an inverted (5')-(5') pyrophosphate bridge structure, AppN. Attack
by the 3' OH of the acceptor strand on the DNA-adenylate joins the two polynucleotides and liberates AMP.
Animal cells contain multiple ATP-dependent DNA ligase isozymes encoded by at least three genes (
3
-
5
). ATP-dependent DNA ligases are also encoded by fungi and plants, by eukaryotic
DNA viruses, by the T-odd and T-even bacteriophages, and by archaea (
6
). The ATP-dependent DNA ligases belong to a superfamily of covalent nucleotidyl
transferases that includes the GTP-dependent eukaryotic mRNA capping enzymes (
7
).
The ligase/capping enzyme superfamily is defined by a set of six short motifs
arrayed in the same order and with similar spacing in nearly all family members
(
7
). Conserved residues within these motifs are critical for covalent nucleotidyl
transfer, as shown by mutational analysis (
8
-
13
). The recently reported crystal structure of T7 DNA ligase shows that the ATP
binding site is made up of five of the six conserved motifs (
14
).
The
Chlorella
virus DNA ligase (a 298 amino acid polypeptide) is the smallest ATP-dependent ligase described to date (
15
). Cellular DNA ligases are much larger; for example, human ligases I, III, and
IV are 919, 922, and 844 amino acid polypeptides, respectively (
4
). Vaccinia virus ligase (552 amino acids) is of intermediate size (
16
).
Sequence comparisons suggest that a catalytic domain common to all ATP-dependent ligases is embellished by additional isozyme-specific protein segments situated at their amino or carboxyl
termini. Although the function of these segments is not known, there is reason
to believe that they are not essential for catalysis. This is the case for
human DNA ligase I, which retains activity after removal of 249 amino acids
from the N-terminus and 16 amino acids from the C-terminus (
12
). Because further deletions of ligase I result in loss of function, it is
surmised that a 654 amino acid segment of ligase I represents the catalytic
core. In the case of the 755 amino acid
Saccharomyces cerevisiae
DNA ligase, a 650 amino acid C-terminal segment retains catalytic activity (
17
). These active domains of the cellular ligases are twice the size of the
Chlorella
virus DNA ligase. This raises the question of whether the catalytic core of the
cellular DNA ligases is structurally more complex than the `minimal' ligases
exemplified by the
Chlorella
virus enzyme and the bacteriophage T7 and T3 enzymes (359 and 346 amino acid
polypeptides, respectively).
We are exploring the structure and function of the eukaryotic DNA ligases using
the vaccinia virus enzyme as a model. Vaccinia DNA ligase is strikingly similar
at the amino acid sequence level to mammalian DNA ligases II and III (
5
,
18
). Because vaccinia ligase and mammalian ligases II and III are more similar to
each other than to ligases I and IV, these three enzymes can be regarded as a
distinct subgroup within the eukaryotic ligase family. The enzymatic properties
of vaccinia ligase have been studied using recombinant enzyme produced in
bacteria (
13
,
19
,
20
). Interestingly, the vaccinia enzyme requires high concentrations of ATP for
strand joining. This is also the case for DNA ligase II, its cellular
homologue. Mutational analysis identifies Lys-231 as the active site of adenylate transfer by vaccinia DNA ligase (
13
).
In this report we employ the classical approach of limited proteolysis to probe
the domain structure of vaccinia DNA ligase. We find that the enzyme consists
of three domains punctuated by protease-sensitive interdomain bridges. DNA ligase activity is preserved when the N-terminal domain is deleted. However, the truncated form of vaccinia
ligase has a reduced catalytic rate and a lower affinity for DNA than does the
full-sized enzyme. The 350 amino acid catalytic core of the vaccinia ligase is
strikingly similar in size and domain organization to the minimal DNA ligases
encoded by
Chlorella
virus PBCV-1 and the T-odd bacteriophages.
Vaccinia DNA ligase was expressed in bacteria as an N-terminal His
6
-tagged fusion protein and purified from bacterial lysates by Ni-agarose and phosphocellulose chromatography as described (
13
). Proteolysis reaction mixtures (20 [mu]l) containing 40 mM Tris-HCl (pH 8.0), 0.4 M NaCl, 2 mM DTT, 0.1 mM EDTA, 0.1% Triton X-100, 8% glycerol, 6 [mu]g of His
6
-ligase, and trypsin or chymotrypsin were incubated at 22oC for 15 min. The samples were denatured in SDS and the proteolysis
products were resolved by electrophoresis through a 12% polyacrylamide gel
containing 0.1% SDS. The gel was soaked in 100 ml of electroblotting buffer {10
mM CAPS (3-[cyclohexylamino]-1-propanesulfonic acid), pH 11, 10% methanol} for 5 min at room
temperature. Polypeptides were transferred to a PVDF membrane (Bio-Rad) in electroblotting buffer by electrophoresis at 500 mA for 2-4 h using a Hoefer Transphor apparatus (Model TE42). The membrane
was rinsed with deionized water, then saturated with methanol. The membrane was
stained with 0.1% Coomassie Blue in 40% ethanol/1% acetic acid for 1 min, followed by
destaining with 50% methanol and rinsing with deionized water at 4oC for 16 h. The membrane was air-dried and slices containing individual proteolytic products were
excised. Automated sequencing of the immobilized polypeptide was performed
using a modified model 477A microsequencer (Applied Biosytems).
An
Nde
I fragment containing the entire vaccinia ligase gene was excised from the
plasmid pBS-ligase (
13
) and inserted into the
Nde
I site of the T7-based expression plasmid pET16b (Novagen) to generate pET-His
10
-ligase. In this plasmid, a leader sequence encoding 10 tandem histidines
was fused in-frame to the 5' end of the ligase gene. (We found that the His
10
leader peptide afforded a better affinity purification on Ni-agarose than did the His
6
tag used previously.) N-terminal deletion variants were generated by PCR-amplification using oligonucleotide primers designed to introduce
Nde
I restriction sites and to create methionine substitutions at internal positions within the ligase polypeptide. The PCR
products were digested with
Nde
I and then inserted into the
Nde
I site of pET16b to yield plasmids pET-His10-ligase(176-552), pET-His10-ligase(191-552), and pET-His10-ligase(201-552). Wild type and
deleted His
10
-ligase expression plasmids were transformed into
Escherichia coli
BL21(DE3). Induction of ligase expression was performed as described previously
(
13
). Cells from 200 ml bacterial cultures were harvested by centrifugation and the pellets were resuspended in 5 ml of
lysis buffer [50 mM Tris-HCl (pH 8.0), 0.5 M NaCl, 10% sucrose]. The suspensions were adjusted to
0.2 mg/ml of lysozyme and 0.1% of Triton X-100. After incubation on ice for 30 min, the lysates were sonicated, then
centrifuged for 15 min at 15 000 r.p.m. in a Sorvall SS34 rotor. The soluble
supernatant fraction was adsorbed to Ni agarose; the resin was washed with
buffers containing 5, 25, and 50 mM imidazole (
13
). His
10
-ligase was recovered by step-elution with 100 mM imidazole. The Ni-agarose eluates were mixed with 4 vol of buffer A [50 mM Tris-HCl (pH 8.0), 1 mM EDTA, 2 mM dithiothreitol, 0.1%
Triton X-100, 10% glycerol] and then applied to 1 ml columns of phosphocellulose
that had been equilibrated with buffer A. The columns were eluted stepwise with
buffer A containing 0.25, 0.5, and 1 M NaCl. The His
10
-ligase proteins were recovered in the 0.5 M NaCl fraction. The protein
concentration of the enzyme fractions was determined using the Biorad dye
reagent with bovine serum albumin as a standard.
The standard substrate used in ligase assays was a 36 bp DNA duplex containing a
centrally placed nick. This DNA was formed by annealing two 18mer
oligonucleotides to a complementary 36mer strand (
19
). The 18mer constituting the donor strand was 5'
32
P-labeled and gel-purified as described (
13
,
19
). The labeled donor was annealed to the complementary 36mer in the presence of
a 3' OH-terminated acceptor strand in 0.2 M NaCl by heating at 65oC for 2 min, followed by slow-cooling to room temperature. The molar ratio of the
18mer donor to 36mer complement to 18mer acceptor strands in the hybridization
mixture was 1:4:4.
Reaction mixtures (20 [mu]l) containing 50 mM Tris-HCl (pH 8.0), 5 mM DTT, 10 mM MgCl
2
, 1 mM ATP (where indicated), 500 fmol of 5'
32
P-labeled DNA substrate, and enzyme were incubated at 22oC. Reactions were initiated by addition of enzyme and halted by the
addition of 1 [mu]l of 0.5 M EDTA and 5 [mu]l of formamide. The samples were heated at 95oC for 5 min and then electrophoresed through a 17% polyacrylamide
gel containing 7 M urea in TBE (90 mM Tris borate, 2.5 mM EDTA). The labeled
36mer ligation product was well-resolved from the 5'-labeled 18mer donor strand. The extent of ligation [36mer/
(18mer + 36mer)] was determined by scanning the gel using a FUJIX BAS1000
phosphorimager.
Recombinant vaccinia ligase containing an N-terminal His
6
tag was subjected to proteolysis with increasing amounts of trypsin. Initial
scission of the 65 kDa His-ligase yielded two major products, a 29 kDa polypeptide (T29) and a
polypeptide doublet of 35-36 kDa (Fig.
1
). Sequencing of these cleavage products by automated Edman chemistry after
transfer to a PVDF membrane revealed that T29 was a mixture of three species
arising via trypsin cleavage at vicinal residues Lys-303, Lys-304, and Lys-305. These sites are denoted by arrows above the polypeptide
sequence in Figure
2
. (Note that amino acid residues are numbered according to their position in the
native ligase polypeptide, not in the His-tagged version.) The T35 product arose via cleavage at an Arg residue
within the His-tag (the leader peptide is underlined in Fig.
2
). Thus, trypsin at limiting concentration cleaved the ligase into a 35 kDa N-terminal fragment and a 29 kDa carboxyl fragment.
We assayed the ability of the recombinant ligases to seal a 36mer synthetic
duplex DNA substrate containing a single nick (
19
). The structure of the substrate is shown in Figure
5
. Ligase activity was evinced by conversion of the 5'
32
P-labeled 18mer donor strand into an internally-labeled 36mer product (
19
). The standard ligase reaction mixtures contained 500 fmol of labeled DNA
substrate. The extent of strand joining by wild type ligase in the presence of
1 mM ATP increased linearly with protein up to 10 fmol (Fig.
5
, +ATP). The reaction saturated at >= 20 fmol of enzyme with 85% of the labeled donor strand converted to 36mer in
10 min. This upper limit of ligation probably reflected incomplete annealing of
all three component strands to form the nicked substrate. In the linear range
of enzyme-dependence in this experiment, the recombinant wild type ligase joined ~41 fmol of DNA ends per fmol of enzyme. For the purpose of estimating
the ratio of product to enzyme, the enzyme molarity was calculated based on
total protein concentration, assuming enzyme homogeneity. It was also assumed
that all enzyme molecules in the preparation were catalytically active.
Strand joining by pre-adenylylated ligase in the absence of ATP is a two step reaction
entailing: (i) AMP transfer to the donor strand to form DNA-adenylate and (ii) attack on DNA-adenylate by the acceptor strand with displacement of AMP. Because
reversal of the ligation step is both slow and dependent on high ( >= 5 mM) concentrations of exogenous AMP (Sekiguchi and Shuman, unpublished),
and because the forward reaction proceeds to completion at stoichiometric
levels of EpA, we regard ATP-independent ligation as an essentially irreversible reaction under the
conditions employed in our experiments. We analyzed the kinetics of ATP-independent ligation by wild type ligase and by each of the N-terminal deletion mutants. The concentration of nicked DNA substrate
in the ligation mixtures was 12.5 nM; enzyme-adenylate was included at several concentrations. Results are shown in
Figure
6
for the wild type enzyme and for ligase(176-552).
A native gel mobility shift assay was employed to examine the binding of
purified recombinant ligase to the
32
P-labeled nicked duplex DNA (
19
). Binding reactions were performed in the absence of magnesium so as to
preclude conversion of substrate to product during the incubation (
19
). Mixing the wild type ligase with nicked substrate resulted in the formation
of a discrete protein-DNA complex that migrated more slowly than the free DNA
during electrophoresis through a 6% native polyacrylamide gel (Fig.
7
). The abundance of this complex increased in proportion to the amount of input
ligase. In order to estimate binding affinity, the gel was scanned using a
phosphorimager. The dependence of protein-DNA complex formation on input ligase
is shown in Figure
8
. The apparent dissociation constant, estimated according to Riggs
et al
. (
21
), was ~15 nM.
We have employed partial proteolysis and deletion mutagenesis to identify
structural and functional domains of vaccinia virus DNA ligase. Three
significant findings emerge from the experiments presented: (i) the native
ligase consists of (at least) three structural domains; (ii) the N-terminal 200 amino acids are not essential for ligase activity; (iii) N-terminal deletions reduce affinity of the ligase for nicked DNA and
decrease the rate of strand joining at a step subsequent to enzyme-adenylate formation.
Vaccinia ligase consists of three structural domains. These are: (i) an amino-terminal domain that is susceptible to digestion with chymotrypsin and
trypsin; (ii) a protease-resistant central domain, and (iii) a protease-resistant carboxyl domain. The two protease-resistant domains are separated by a protease-sensitive interdomain bridge. The bridge extends from
Phe-296 to Tyr-307 (sites accessible to chymotrypsin), and includes a cluster of
trypsin-sensitive lysines at positions 303, 304, and 305. The interdomain bridge
of the vaccinia ligase is located between conserved sequence motifs III and
IIIa of the ligase/capping enzyme superfamily. Mutational and structural
studies of family members indicate that motifs III and IIIa form part of the
nucleotide binding pocket (
9
,
13
,
14
).
The domains defined by proteolysis of the vaccinia ligase are similar to those
demarcated for the T7 enzyme (
22
), except that the vaccinia protein contains the N-terminal domain that is absent from the T7 protein. Limited digestion of
T7 ligase with endoproteinase Arg-C yielded an N-terminal 16 kDa fragment and a 26 kDa carboxyl fragment. As with the
vaccinia enzyme, the sites of protease accessibility in the T7 ligase were
located between motifs III and IIIa (
22
). In the crystal structure of T7 DNA ligase, this polypeptide segment is a
disordered loop that extends from the surface of an otherwise tightly folded
protein (
14
). This would account for the observed protease susceptibility. The observation
that the same segment of the vaccinia DNA ligase is protease-sensitive suggests that it is also exposed on the surface. Thus, both the
T7 and the vaccinia ligases have two protease-resistant domains, one containing motifs I and III and another including
motifs IIIa, IV, V, and VI.
The size of the protease-sensitive polypeptide segment separating motifs III and IIIa varies
considerably within the covalent nucleotidyl transferase superfamily (
7
,
15
). Among the DNA ligases, the bacteriophage enzymes have the longest interval
(58, 55, and 49 amino acids for the T4, T3, and T7 enzymes, respectively),
whereas vaccinia ligase and
Chlorella
virus ligase have the shortest interval (27 and 24 amino acids, respectively).
In the capping enzymes, motifs III and IIIa are separated by only 3-12 amino
acids.
The N-terminal domain of vaccinia ligase was less stable to proteolysis than
were the central and carboxyl domains. This protein segment, extending to the
site of trypsin cleavage at position 190, has no counterpart in the smaller DNA
ligases encoded by the T-odd bacteriophages and by the
Chlorella
virus PBCV-1. Our analysis of the effects of N-terminal deletions on vaccinia ligase activity indicated that the N-terminal domain is not essential for adenylyl transfer or
strand joining, insofar as the deleted proteins retained catalytic function.
Yet, removal of the N-terminus did have consequences with respect to catalytic rate and DNA
substrate affinity.
Deletion of the N-terminal 176-200 amino acids resulted in a 10-fold decrement in the rate constant for single-turnover ligation by enzyme-adenylate. This could arise from mutational effects
either on AMP transfer from EpA to nicked DNA to form DNA-adenylate (step 2 of the overall ligation reaction) or on enzyme-mediated attack of the acceptor strand on DNA-adenylate to form a phosphodiester bond (step 3 of the
complete reaction). Previous studies had indicated that DNA-adenylate formation (step 2) is the rate-limiting step during DNA ligation catalyzed by the wild type
vaccinia ligase. DNA-adenylate is detectable only under reaction conditions that strongly
suppress the strand closure step (
19
). Because DNA-adenylate did not accumulate during the course of single-turnover ligation by the deletion mutants of vaccinia ligase
(Sekiguchi and Shuman, unpublished), we surmise that the observed effects of N-terminal truncations on
k
lig
were caused primarily by a slowing of step 2. We cannot exclude the possibility
that the N-terminal deletions affected the rate of step 3; however, potential rate
effects on step 3 are apparently not significant (i.e. not rate-determining) in the context of the single-turnover ligation reaction.
Deletion of the N-terminal domain had a definite impact on the affinity of the vaccinia
ligase for nicked duplex DNA. Kinetic analysis indicates that the wild type K
m
of nicked DNA for enzyme is 9 nM. The N-terminal deletions increased the K
m
value by a factor of 4-6. The kinetic experiments were supported by native gel-shift assays of the binding of ligase to nicked DNA. Detection of a
discrete ligase-DNA complex in the mobility shift assay requires that the protein remain
bound stably to the DNA ligand during the electrophoresis procedure. Formation
of lower levels of discrete shifted complexes by the truncated ligases, and the
presence of a diffuse smear of DNA trailing behind the unbound ligand,
presumably reflects both a lower binding affinity for nicked DNA in solution
(as revealed initially by the kinetic experiments) and reduced stability of the
mutant protein-DNA complex during the electrophoresis. The latter may account for the
observation that the mutational effects on DNA binding in the gel-shift assay were slightly greater than those determined kinetically.
In summary, we have defined at low resolution a domain structure for vaccinia
ligase and shown that a 352 amino acid derivative, ligase(201-552), is active in enzyme-adenylate formation and strand ligation. The size of the vaccinia
catalytic domain is comparable to that of the full-length 359 amino acid T7 enzyme; indeed the vaccinia and T7 enzymes have
similar domain boundaries. Vaccinia ligase(201-552) contains only 30 amino acids upstream of the active site lysine
residue in motif I. The
Chlorella
virus and T7 DNA ligases contain 26 and 33 amino acids upstream of their
respective active sites (
14
,
15
).
Although the N-terminal domain of vaccinia ligase is not essential, we suspect that it
does play a role in ligase function. This is based on the observations that
catalytic rate and DNA affinity are impacted by loss of the N-terminus, plus the fact that the amino acid sequence of the entire N-terminal domain of vaccinia ligase is conserved in mammalian DNA
ligase III and in mammalian DNA ligase I (
3
-
5
). The proximal region of similarity to the vaccinia N-terminal domain is recessed from the amino terminus in ligase III and
ligase I because these cellular ligases contain sizable protein segments at
their amino termini (175 and 292 amino acids respectively) that have no
counterpart in the vaccinia virus protein (
5
). This N-terminal domain of mammalian DNA ligase I contains two elements involved
in protein targeting: a nuclear localization signal and a signal that directs
ligase to sites of DNA replication within the nucleus (
23
). Vaccinia virus, which replicates in the cytoplasm, would have no need for
such a domain. It is tempting to speculate that the shared N-terminal domain of the mammalian and vaccinia ligases might, in addition
to facilitating catalysis and DNA binding as described herein, also participate
in protein-protein interactions with other components of the DNA repair or
replication machinery. In this respect, it is noteworthy that vaccinia DNA
ligase localizes to the cytoplasmic sites of DNA replication in virus-infected cells (
24
); the deletion mutants described in this report may prove useful in delineating
the requirement for targeting of vaccinia ligase
in vivo
.
REFERENCES
Return