|
|
|
|||||||
|
|||||||||
(Received for publication, November 3,
1995; and in revised form, January 31, 1996) From the
A set of phosphorothioate-containing oligonucleotides based on
pGACGATATCGTC, a self-complementary dodecamer that contains the EcoRV recognition sequence (GATATC), has been prepared. The
phosphorothioate group has been individually introduced at the central
nine phosphate positions and the two diastereomers produced at each
site separated and purified. The K
The specific recognition of DNA sequences by proteins involves
the formation of a very precise and intimate interface between the two
macromolecules. Direct contacts are observed between the protein and
both the bases and phosphates, and in addition complicated networks
bind the interacting elements
together(1, 2, 3, 4) . Much of the
emphasis has been on the study of contacts between the protein and the
DNA bases. This, known as direct readout, is thought to contribute much
of the selectivity(5, 6, 7) . The phosphate
backbone has received less attention. However, as well as a passive
role in endowing a basal DNA binding affinity to a DNA-binding protein,
it is clear that the phosphates normally play additional, more active,
roles in the generation of specificity. Work with the trp repressor has led to the concept of indirect readout(8) .
Most simply, this takes place when a protein binds specifically to the
phosphates in a DNA sequence that has an unusual conformation, with an
altered phosphate backbone, that differs from ``ideal''
B-DNA. In addition, almost all DNA sequences are distorted on binding
to proteins and this is often required to match up protein-phosphate
contacts. Thus phosphates can play a role in selectivity following any
DNA distortion that takes place on binding. This coupling of the
recognition of specific bases, DNA distortion, and specific phosphate
binding occurs with the restriction endonuclease EcoRI(9, 10) . The importance of DNA
phosphate-protein contacts provides a compelling reason for their
study. One of the commonest methods for evaluating protein-substrate
interactions is to make alterations to the partners involved and
observe the consequences. This has been widely carried out for protein
amino acids using site-directed mutagenesis and for DNA with modified
bases(11, 12, 13, 14, 15, 16) .
For DNA phosphates the most useful analogues are the phosphorothioates (17, 18, 19, 20, 21) ,
illustrated in Fig. 1, which exist as a pair of diastereomers.
Phosphorothioates probably represent the most conservative change that
can be made to a phosphate. The sulfur atom is slightly bigger than the
oxygen it replaces and the P-S bond length a little longer than the
P-O(22) . Phosphorothioates are also slightly more acidic than
phosphates (23) and may be differently solvated. With
phosphates the negative charge is evenly distributed over the two
non-bridging oxygens, but in the case of phosphorothioates, current
evidence favors negative charge localization on
sulfur(22, 24, 25) . Importantly
Mg
Figure 1:
The
structure of oligodeoxynucleotide phosphorothioates, which exist as a
pair of diastereomers.
The EcoRV endonuclease is a well characterized enzyme (35, 36) for which crystallographic data is
available(37, 38, 39) . A large body of
kinetic and binding data has revealed that, in the absence of
Mg The
endonuclease DNA contacts seen with cognate but not non-cognate
sequences provide the energy required for the energetically unfavorable
distortion of the bound DNA. This distortion is essential for
hydrolysis, and so it is these interactions that are ultimately
responsible for the discrimination of
Figure 2:
The reverse phase HPLC trace of crude
pGACGATATCsGTC (blue line). The two diastereomers are clearly
visible as the two largest peaks between 10 and 12 min. The R
Many examples (17, 19, 30, 61) have shown that R
Figure 3:
Determination of the K
Figure 4:
A summary of the specificity constants (V
Figure 5:
The interactions seen between the DNA
phosphate groups and the EcoRV restriction endonuclease. This
is taken from the complex of EcoRV with AAAGATATCTT that
contains Mg
In addition to the above difficulties, the tacit
assumption that the analogue does not change the overall global
structure of the macromolecule is usually made. This is hard to
establish unequivocally. Only one crystal structure, for all R
The three phosphates GACGpATATCGTC,
GACGApTATCGTC, and GACGATpATCGTC, which fall within the recognition
site and include the scissile phosphodiester can be considered
together. These phosphates are characterized by extensive direct
contacts to the protein and participation in an extended interconnected
network of hydrogen bonds. Much of this network is mediated by highly
structured and ordered water molecules(39) . This network
performs the following roles: 1) it interconnects the three phosphates,
2) it joins the phosphates to the amino acids in the recognition
(R)-loop (these comprise amino acids 182-187, which interact with
the GATATC bases via the major groove), 3) it links the phosphates to
the Q-loop (this is centered on amino acid 70 and interacts with the
DNA via the minor groove), and 4) it assembles the catalytic components
(Mg The S Both
phosphorothioates of GACGATATpCGTC were poor substrates with an
extremely low rate being observed with the R
The important phosphates, as assessed using
phosphorothioates, comprise GACGpApTpApTpCpGTC. This agrees very well
with the site-directed mutagenesis results reported in a related
paper(75) . Furthermore, a confirmation of the critical nature
of these phosphates has come from an ethylation interference study with
the RV endonuclease. ( Under physiological
conditions the EcoRV endonuclease must distinguish between
GATATC and all other sequences. One-base pair alterations to the GATATC
recognition site give substrates that, under optimal conditions, are
cut at vanishingly small rates(40, 42, 49) .
Using base analogues, we and others have shown that even small changes
to the bases in the recognition site frequently give very poor
substrates(13, 14, 15, 50, 51, 52) .
Often the reduction in turnover is much greater than can be accounted
for in terms of simple loss of protein-DNA contacts, exactly as is seen
with some of the phosphates. Just as alterations to particular
phosphates lead to much poorer substrates than expected, because of
weakening of other protein-DNA interactions, it is clear that changes
to the GATATC sequence do not simply result in the deletion of the
contact in question. Instead, concerted movements lead to further
misalignments that include incorrect protein-phosphate interactions.
This paper begins to identify the phosphates involved in this process.
Only with a fully cognate GATATC sequence are the proper
protein-phosphate contacts made, and only this provides enough energy
to bend the DNA, create the metal binding site, and give rapid
hydrolysis.
Volume 271,
Number 15,
Issue of April 12, 1996 pp. 8855-8862
©1996 by The American Society for Biochemistry and Molecular Biology, Inc.
A STUDY USING OLIGODEOXYNUCLEOTIDE PHOSPHOROTHIOATES (*)
and V
values found for each of these modified DNA
molecules with the EcoRV restriction endonuclease have been
determined and compared with those seen for the unmodified
all-phosphate-containing dodecamer. This has enabled an evaluation of
the roles that both of the non-esterified oxygen atoms in the
individual phosphates play in DNA binding and hydrolysis by the
endonuclease. The results have also been compared with crystal
structures of the EcoRV endonuclease, complexed with an
oligodeoxynucleotide, to allow further definition of phosphate group
function during substrate binding and turnover. For further study, see
the related article ``Probing the Indirect Readout of the
Restriction Enzyme EcoRV: Mutational Analysis of Contacts to
the DNA Backbone'' (Wenz, A., Jeltsch, A., and Pingoud, A. (1996) J. Biol. Chem. 271, 5565-5573).
, an essential cofactor for the EcoRV
endonuclease, strongly favors co-ordination to oxygen atoms in
phosphorothioates(26, 27) . Despite these differences,
the single known structure of a phosphorothioate-containing
oligonucleotide shows minimal structural perturbations as compared to
the all-phosphate parent(28) . Oligonucleotide
phosphorothioates have been used to determine the stereochemical course
of the reactions catalyzed by EcoRI (29) and EcoRV (30) and to probe details of protein-DNA
interactions(31, 32, 33, 34) .
, the endonuclease binds to all DNA sequences with
equal affinity(40, 41, 42, 43) .
However, specific contacts, made only to the bases in the cognate
GATATC sequence cause severe distortion of the DNA, and this
concomitantly creates a high affinity Mg binding
site, allowing hydrolysis to take place. The observation that specific
binding to GATATC sequences occurs in the presence of the essential
metal ion, and that the DNA in these complexes is bent to the same
extent as is observed in the crystal structure, lends support to the
above model(44, 45, 46) . Recent data (39, 47, 48) has suggested that the
endonuclease might use two metal ions for catalysis.
10
(40) shown by the enzyme. The interactions, between the
protein and the GATATC bases, have been probed using alternative
sequences(40, 42, 49) , base
analogues(13, 14, 15, 50, 51, 52, 53) ,
and site-directed mutagenesis(43, 54) , and many have
been shown to be essential for efficient catalysis. The endonuclease
also makes extensive contacts to the phosphate backbone (38, 39) . However, there has been little systematic
investigation into the role that these interactions play in DNA
recognition. M13 DNA, containing R
-phosphorothioates, showed that GATsATC sequences
were very refractory to hydrolysis and substitution elsewhere tended to
reduce cutting rates(32) . Very recently, the phosphorothioate
oligonucleotides used in this study have been used to show the presence
of a metal ion binding site distinct from the catalytic
center(55) . In this publication, we examine the effects of
both isomers of phosphorothioates within and immediately flanking the EcoRV restriction site on endonuclease-catalyzed hydrolysis
and relate the results found to the available crystal structures.
EcoRV Endonuclease Purification
The purification
of the EcoRV restriction endonuclease from overproducing Escherichia coli strains (56) by a slight modification
of the method originally described (57) has been
published(30) . SDS-gel electrophoresis and Coomassie Blue
staining showed >95% purity. The concentration of stock solutions
was determined using an E
value of
18.0(57) .Oligodeoxynucleotide Preparation
All
oligonucleotides were prepared on a 1-µmol scale using
phosphoramidite chemistry using an Applied Biosystems 381A DNA
synthesizer. Reagents were purchased from Cruachem Ltd. (Glasgow,
Scotland). Phosphorothioates were introduced using a 0.05 M acetonitrile solution of the sulfurizing agent
3H-1,2-benzodithiol-3-one 1,1-dioxide (58, 59) which
was purchased from Cambio Ltd. (Cambridge, United Kingdom). The
synthesis were interrupted immediately prior to the iodine oxidation
step of the appropriate intermediate phosphite. The resin was washed
with acetonitrile (2 min) and the sulfurizing agent applied for 20 s.
After a 30-s pause, the resin was washed with acetonitrile (2 min) and
normal synthesis resumed. 5`-Phosphates were added during the synthesis
using the reagent 5`-phosphate-on (Cruachem Ltd.). All synthesis were
performed trityl-off and the oligonucleotides removed from the resin
and deblocked with ammonia in the normal way.Oligonucleotide Purification and Phosphorothioate
Diastereomer Separation
Purification and diastereomer separation
was achieved by reverse phase HPLC (
)on Apex-I
octadecylsilyl (C-18) columns (5-µm particle size, 25 
0.45
cm) purchased from Jones Chromatography (Llanbradach, Wales). For most
oligonucleotides, triethylammonium-acetate buffers with an acetonitrile
gradient were used (buffer A, 0.1 M acetic acid adjusted to pH
6.5 with triethylamine containing 2.5% acetonitrile; buffer B, 0.1 M acetic acid adjusted to pH 6.5 with triethylamine containing
65% acetonitrile). Gradients consisting of t = 0, 0% B, t = 5 min 0% B, t = 30 min 20% B were
used. For the separation of the isomers of pGACGsATATCGTC and
pGACGATAsTCGTC, buffers based on morpholine-acetate were used (buffer
C, 0.1 M acetic acid adjusted to pH 6.5 with morpholine
containing 2.5% acetonitrile; buffer D, 0.1 M acetic acid
adjusted to pH 6.5 with morpholine containing 50% acetonitrile). The
same gradient profile as above was used. All columns were run at 50
°C at 1 ml min
. Following purification all
oligonucleotides were desalted using disposable NAP-25 gel filtration
columns (Pharmacia, St. Albans, United Kingdom). The concentrations of
the oligonucleotides were determined using an E
of 1.66 10
M cm for the double-stranded form(13) .Assignment of the Absolute Configuration of the
Oligonucleotide Phosphorothioates by Base Composition
Analysis
About 1 OD of the purified
phosphorothioate was treated either with 10 µg of snake venom
phosphodiesterase and 5 µg of alkaline phosphatase in 100 µl of
50 mM Hepes (pH 7.5), 100 mM NaCl, and 10 mM MgCl
; or with 10 µg of nuclease P1 and 5 µg of
alkaline phosphatase in the same buffer. After a 2-h incubation at 30
°C, the products formed were analyzed by reverse phase HPLC
(columns and HPLC buffers A and B as above) using a gradient t = 0 min 0% B, t = 25 min 25% B, t = 35 min 50% B. Columns were run at 1 ml min and room temperature. The standard deoxynucleosides eluted in the
order dC, dG, T, dA. dCMPs eluted between dC and dG, dGMPs and dTMPs
between dG and T, and dAMPs between T and dA. All undigested
dinucleotides, NsN, eluted after dA. The products formed in the
enzymatic digests were identified by co-elution with standards. The
dNTPs standards were a kind gift from Prof. F. Eckstein (Max Plank
Institut für experimentelle Medizin,
Göttingen, Germany). A few NsN standards were also
obtained from this source, but most of the dinucleotide assignments are
tentative due to a lack of reference material.Cleavage of Oligonucleotide Phosphorothioates
To a
12 µM solution of oligonucleotide in 300 µl of 50
mM Hepes, pH 7.5, 100 mM NaCl, and 10 mM MgCl was added EcoRV restriction endonuclease
to a final concentration of 0.9 µM. The mixture was
incubated at 20 °C and samples analyzed by HPLC at 1, 3, and 24 h.
In cases where the digestion was incomplete, an additional aliquot of
the endonuclease was added and the digestion continued for another 24
h. The HPLC protocol used the triethylammonium-acetate buffer system
detailed under ``Oligonucleotide Purification and Phosphorothioate
Separation.'' On completion of hydrolysis, the product peaks were
purified by this HPLC method and analyzed for base composition as
above.K
Kinetic constants were evaluated using a
continuous UV assay (60) at 30 °C. A buffer consisting of
50 mM Hepes, pH 7.5, containing 55 mM NaCl and 25
mM MgCl
and V Determination was used with volumes of either 1 ml (1-cm
path length) or 2 ml (2-cm path length). Oligonucleotide concentrations
varied between 0.15 and 7.5 µM. The reaction was initiated
by the addition of endonuclease (concentrations between 5 and 50
nM), and in almost every case the oligonucleotide
concentration was 10 times the level of enzyme. The increase in
absorbance at 254 nm was monitored using a Uvikon 930 spectrophotometer
and the change in absorbance related to the amount of oligonucleotide
hydrolyzed using the relationship: 
absorbance (254 nm)/nmol
oligonucleotide hydrolyzed = 0.034(60) . V and K values were
evaluated from plots of [S]/v against [S],
using eight different concentrations of the substrate. Each
determination was carried out at least four times.
Synthesis of Phosphorothioate-containing
Oligonucleotides
The dodecamers listed in Table 1were
prepared using phosphoramidite chemistry and replacing the normal
iodine/water oxidation step with a sulfurization using
3H-1,2-benzodithiol-3-one 1,1-dioxide(58, 59) . In all
cases the 5`-phosphate was attached to the oligonucleotide as part of
the chemical synthesis. Following deblocking of the oligonucleotides,
analysis by reverse phase HPLC showed the presence of the two
phosphorothioate diastereomers, which usually comprised more than 80%
of the UV absorbing material (Fig. 2). The most demanding step
in the synthesis was the separation of the two diastereomers. All the
diastereomer pairs could be separated by reverse phase HPLC on C-18
columns, but the ease with which this could be achieved showed great
variability. This has been observed in several previous
studies(19, 61) . High temperatures (50 °C) were
necessary for good separation, probably by melting the
self-complementary strands. Most of the phosphorothioate isomers could
be separated using acetonitrile gradients in triethylammonium acetate,
pH 6.5, buffers. A few (pGACGsATATCGTC and pGACGATAsTCGTC) were poorly
resolved in this buffer, and in these cases replacing the
triethylammonium acetate with morpholine acetate, pH 6.5, resulted in
better separation. Variations were also made to other HPLC parameters
that included altering the organic component (methanol, ethanol,
2-propanol), changing the column material (C-4, C-8, phenyl) and using
different pH values but none of these improved separation. Negatively
charged oligodeoxynucleotides run as their salts during reverse phase
purification and this may explain why the counter ion (triethylamine or
morpholine) can alter separation parameters. However, the resolution of
oligonucleotide phosphorothioates on C-18 HPLC is highly unpredictable.
We are unable to forecast whether a particular pair of phosphorothioate
isomers will be easy or difficult to resolve or to state the counter
ion that will give the best separation for an individual case.
Following optimization of the separation each individual
phosphorothioate could be obtained at purities of greater than 97% with
less than 3% contamination with the other isomer (Fig. 2).
isomer elutes before the S.
The traces of the purified diastereomers following HPLC separation are
also shown. Green line, R isomer; red
line, S isomer.
-oligonucleotide phosphorothioates elute before
the S on reverse phase columns using
triethylammonium acetate buffers. This was found here in every case,
and the faster elution of R-isomers was also
maintained when the buffer was altered to morpholine acetate. The
configuration of fully purified oligonucleotide phosphorothioates can
be unequivocally assigned using digestion with enzymes of known
stereospecificities. Snake venom phosphodiesterase cuts oligonucleotide
phosphorothioates having the R conformation but
does not digest those of S(62) . These
stereospecificities are reversed with nuclease P1(63) .
Therefore the phosphorothioates were separately digested with nuclease
P1 and snake venom phosphodiesterase and the resulting deoxynucleoside
products were analyzed by reverse phase HPLC(17, 30) .
When ``early'' isomers were treated with the snake venom
enzyme, a peak corresponding to dNMPs was observed by HPLC, whereas
reaction with P1 gave a NsN unhydrolyzed dinucleotide peak (not shown).
``Late'' isomers gave NsN with venom phosphodiesterase and
dNMPs with nuclease P1 (not shown). This confirms that the early
eluting isomers have the R configuration and the
late the S.Hydrolysis of Phosphorothioate Oligonucleotides by the
EcoRV Endonuclease
Prior to the determination of kinetic
constants, all the oligonucleotides were incubated with a large
quantity of the endonuclease (12 µM oligonucleotide, 0.9
µMEcoRV endonuclease) and any reaction monitored
by HPLC after 1, 3, and 24 h. Most of the oligonucleotides were
completely hydrolyzed to two new products (not shown). In some cases
this required only 1 h, whereas in others the full 24 h was needed.
Purification and base composition analysis of the two products, showed
that, in every case, the endonuclease had cut at its true site
(GAT
ATC). These results show that the presence of a
phosphorothioate does not change the cutting site of the enzyme.
Furthermore, even at high endonuclease levels, no further decomposition
of the products took place, indicating a high level of purity and the
lack of any contaminating nucleases. R-pGACGATsATCGTC was only partially hydrolyzed
after 24 h, but the addition of another aliquot of enzyme and a second
24-h incubation gave complete cutting. We have previously used R-GACGATsATCGTC to determine the stereochemical
course of the endonuclease-catalyzed reaction and shown it to be a slow
substrate(30) . We did not observe any products when the S isomers of both pGACGATsATCGTC and
pGACGATAsTCGTC were treated with two aliquots of the endonuclease. Thus
both of these appear to be non-substrates and, based on the detection
limits of this assay, are cut at 
0.1% of the rate shown by the
parent, all-oxygen-containing oligonucleotide.Kinetic Constants for the Hydrolysis of the
Phosphorothioates by the Endonuclease
The K and V values for the phosphorothioates and
the all-phosphate parent were determined using a continuous UV
absorbance assay(60) . This is based on the hyperchromic effect
and the increase in absorbance at 260 nm when a double-stranded
dodecamer is cut to single-stranded hexameric products. Kinetic
constants could not be determined for both the isomers of
pGACGATsATCGTC and the S isomer of pGACGATAsTCGTC,
because of the very low, or zero, hydrolysis rates. For all the other
oligonucleotides, the determination of K and V values, using plots of
[substrate]/velocity against [substrate], was
relatively straightforward, and a representative example is given in Fig. 3. The data set is summarized in Table 1and the (V/K)
/(V/K)
shown pictorially in Fig. 4. Phosphorothioate substitution
invariably caused a reduction in V values, and
the spread that was obtained ranged from very similar to the control (i.e. only a slight reduction) to zero or near zero. These
reductions are relatively easy to explain in terms of disturbance to
endonuclease-phosphate contacts and further alterations to the
protein-DNA interface leading to poor assembly of a catalytically
competent complex (see ``Discussion''). Most of the
phosphorothioates bound more tightly to the endonuclease as shown by
the decrease in K values, which in one case (S-pGACGAsTATCGTC) was reduced 10-fold. However,
smaller reductions were much more common. Only three sulfur-containing
oligonucleotides bound more weakly to the enzyme, but in all these
cases the K increase was very small. Better
binding of phosphorothioates is more difficult to rationalize than
drops in V, but similar results have been seen
with the EcoRI endonuclease (33) and the R17 coat
protein with phosphorothioate-substituted RNA(64) . A possible
rationalization involves the preferred negative charge location on
sulfur in phosphorothioates. Charge localization might strengthen
binding when the sulfur is involved in an ionic interaction with a
positively charged amino acid. However, a comparison of the results
given in Table 1with the actual protein-DNA contacts (Fig. 5) shows this simple explanation does not hold. The
results with EcoRI were explained by subtle adaptations to the
protein-DNA interface, which improved interactions at other sites. Base
analogues may also improve interaction, as seen with the trp repressor (65) and the EcoRI endonuclease (12) . In these two cases, the base analogue used appeared to
facilitate DNA distortion necessary for tight binding. Both of these
explanations could explain the results we observe, although there is no
strong evidence that phosphorothioate DNA is especially flexible. The
discrimination between substrates is best measured using V/K (the specificity
constant) and, as shown in Table 1and Fig. 4, these
showed a considerable spread of values when referenced to the control.
Two of the modified oligonucleotides, the R-isomers of pGAsCGATATCGTC and pGACsGATATCGTC,
had larger V/K values
mainly due to their lowered K values. The other
phosphorothioates had specificity constants that ranged in value from
that seen with the parent down to zero. In general the six phosphates
GACGpApTpApTpCpGTC, within and immediately 3` to the recognition site,
seem to be important in the endonuclease-catalyzed hydrolysis.
Phosphates farther out, in either direction, play little role.
and V
values for both the R and the S diastereomers of pGACGATATCsGTC. All the other phosphorothioates
gave results of similar quality.
/K) for
all the phosphorothioate-containing oligonucleotides. These are
referenced to the all-phosphate-containing dodecamer control, for which
this value is set to 100%.
bound to one of the DNA strands (1RVB
co-crystal structure in the Protein Data Bank, Brookhaven National
Laboratory, Upton, NY)(39) . ASN185 is asparagine 185
from one subunit of the dimeric endonuclease (the A subunit) and ASN*185 asparagine from the other (B) subunit etc. MC, main chain; SC, side chain. 
, water
molecule. All the contacts illustrated are 3.5 Å in length.
The sequence used for crystallography has been substituted, at the
appropriate positions, by the sequence used in this study, and the
GATATC recognition site is shown in uppercase letters. Only
the contacts made to the phosphates that have been investigated in this
paper are shown.
Phosphorothioates as Phosphate
Analogues
Investigators using analogues aim to delete a
protein-substrate contact, most simply by the replacement of an
interacting group with a hydrogen atom. In some cases this can be
cleanly achieved, and good examples include the replacement of serine
or cysteine with alanine and tyrosine with phenylalanine in
site-directed mutagenesis. With oligonucleotides the use of the 7-deaza
derivatives of dA and dG, where the 7-ring nitrogen of the purines is
replaced by a CH function, also approaches this ideal(12) .
However, most functions are not amenable to this straightforward
manipulation, and the phosphate group of DNA falls into this category.
DNA-binding proteins make both salt bridges and hydrogen bond
interactions to the two non-bridging oxygen atoms of the phosphate
group. Oligonucleotide derivatives in which one non-bridging oxygen is
replaced with hydrogen (H-phosphonates,
O=P[OR]-H) have been used in some
applications (66) but are difficult to prepare, in chirally
pure form, and additionally exist in equilibrium with trivalent
tautomers (HO-P[OR]). It should also be
noted that these derivatives, and indeed all phosphate analogues, not
only perturb the non-bridging oxygen substituted but also alter the
nature of the other non-bridging oxygen. In the absence of simple,
replacement derivatives, other analogues must be used. Perhaps the most
suitable are the phosphorothioates, which represent a very subtle
phosphate modification with changes to size, acidity, solvation, charge
distribution, and metal ion binding, as mentioned in the Introduction.
However, the advantage of completely deleting an interaction is that it
is possible to predict the consequence, providing that no additional
effects come into play. Thus a single phosphate contact has been
estimated to contribute 1.3 kcal/mol to a protein-DNA
complex(33, 67) , and this should result in a 10-fold
reduction in V/K (or any
other appropriate metric) when deleted(68) . A knowledge of
expected energetic consequences is important, as deviations from it are
of great interest and usually indicate further adjustments to the
protein-DNA interface. The similarity between phosphate and
phosphorothioates mean that protein-DNA interactions are much more
likely to be weakened rather than completely abolished. Thus one might
not expect the full penalty of 1.3 kcal/mol. As pointed out
previously(33) , the exact consequence of phosphorothioate
substitution will depend on the nature of the protein-phosphate
interaction under consideration. In cases of an ionic interaction
between a flexible lysine or arginine side chain and one non-bridging
oxygen, these authors predict very little effect for phosphorothioate
substitution. Even in the most critical examples, where both
non-bridging oxygen atoms form very precise and constrained hydrogen
bonds with the protein, an energetic penalty of 0.7 kcal/mol, rather
than 1.3 kcal/mol, was observed. This would be associated with a drop
in V/K to about 30% of the
wild type rate.-GsCGsCGsC, of an oligonucleotide
phosphorothioate is known(28) . Comparison with (GC)
runs of other sequences indicates that the phosphorothioates
cause no structural perturbations in this instance. However, it remains
to be established whether this will hold generally for other
phosphorothioate-containing oligonucleotides. Many synthetic
oligonucleotides containing one or a small number of phosphorothioates,
analogous to those used in this study, have been prepared. However,
very little structural characterization has been reported. The low
resolution methods used (usually based on T or circular
dichroism spectroscopy) have shown near identity with parent,
all-phosphate oligomers. It is also possible to model phosphorothioates
into ideal B-DNA without altering structural parameters. In R-isomers the sulfur points into the major grove,
and for S it is located directly on the face of
the sugar phosphate backbone. Thus we have made the usual simplifying
assumption that phosphorothioates do not alter overall DNA structure.Endonuclease DNA-Phosphate Contacts Observed by
Crystallography
Several EcoRV
endonuclease-oligonucleotide co-crystal structures have been solved (38, 39) . The highest resolution structures contain
an 11`mer, AAAGATATCTT, bound to the enzyme and three different
complexes, enzyme-DNA, enzyme-DNA-Mg, and
enzyme-product, have been elucidated. We have based the discussion of
our results on the endonuclease-DNA-Mg complex, as
this best represents the catalytically significant ternary complex. It
contains the greatest number of protein-phosphate contacts and these
are highly organized, i.e. involved in extensive networks. The
relevant enzyme-phosphate contacts seen in this structure are
illustrated in Fig. 5. As can be seen, the catalytically
essential cofactor Mg is only bound to one of the
strands. Additionally the contacts between the two macromolecules
differ slightly between the two DNA strands, and slight variations in
these interactions are also seen in the other two complexes with this
11`-mer and in the earlier structures with different oligonucleotides.
It is important to realize that the Mg-containing
strand, shown in Fig. 5, is not cut in the crystal, despite
having all the components required for hydrolysis in solution. Thus
further conformational changes, and associated changes to
endonuclease-phosphate contacts, may still be required to reach the
transition state.Relationship of Results Seen with Phosphorothioates to
Crystal Structure
For three positions GApCGATATCGTC,
GACpGATATCGTC, and GACGATATCGpTC, which all flank the GATATC
recognition site, the introduction of a phosphorothioate has little
effect and both diastereomers are cleaved almost as well or better than
the parent. Examination of Fig. 5shows that these phosphates
interact rather weakly with the endonuclease. The phosphate
GACpGATATCGTC makes no contacts, either directly or via water, to the
protein. With GACGATATCGpTC one of the phosphate oxygens, the pro-S,
interacts with the flexible side chain of Gln
, but no
contacts are seen to the pro-R. More extensive protein-DNA interactions
are seen with GApCGATATCGTC with a direct interaction with Ser
and some water-mediated contacts to other amino acids. Most
importantly these three phosphates do not take part in the extensive
network that links the more critical phosphates to important catalytic
and recognition elements of the endonuclease. It is likely that the
combination of a low level of direct interactions and a lack of
extensive networking accounts for the tolerance of these positions to
phosphorothioate substitution. and three acidic residues: Glu
,
Asp
, and Asp
) in a manner competent for
catalysis. The pro-S oxygen of GACGpATATCGTC makes the fewest contacts,
an isolated charge/charge interaction with the flexible side chain of
Lys
, and this is the only position where sulfur
substitution gives a reasonable substrate. Alterations to all the other
five oxygen atoms give poor substrates. This undoubtedly arises from
disturbances to the elaborate network illustrated in Fig. 5,
which gives the impression of a very high degree of co-operativity,
whereby these three phosphates are connected to the amino acids
responsible for both specific base recognition and for catalysis. Thus
changes to these phosphate oxygens, even the very minor one of sulfur
substitution, are likely to lead to movements throughout the
hydrogen-bonded network and to the weakening of further protein-DNA
interactions. This concerted breakdown of the macromolecular interface
will reduce the binding energy available for DNA distortion and thus
prevent efficient catalysis. Two further points deserve mentioning.
First, by comparing the Mg- and
Mn-catalyzed cleavage of the phosphorothioates, an
additional metal ion, remote from the active site, was proposed to bind
to the GACGpATATCGTC phosphate(55) . The exact role of this
metal ion in selectivity remains obscure, but it could certainly be
incorporated as an additional structural element in the network
illustrated. Second, in the case of the scissile bond GACGATpATCGTC,
the endonuclease is required to cleave a phosphorothioate diester
rather than a phosphodiester. These two diesters show approximately
similar hydrolysis rates, using model compounds in
solution(69) , and so the very low enzyme-catalyzed rate cannot
be due to intrinsic lower chemical reactivity. Rather, as above, it is
likely to arise because of network disturbance and in the case of the S-phosphorothioate (a non-substrate) to reduced
Mg binding. This atom provides one of the ligands for
the metal ion, and it is well documented that phosphorothioates bind
Mg more poorly than do
phosphates(26, 27) . phosphorothioate of GACGATApTCGTC is also not a substrate,
whereas the R is well cut. A similar result has
been observed with the EcoRI endonuclease (34) . Based
on these results, and a number of other observations, it has been
proposed that the pro-R oxygen of this phosphate is the base that
deprotonates the hydrolytic water molecule for both
nucleases(66, 70) . The location of the negative
charge on sulfur means that, with the R phosphorothioate an S
atom replaces the pro-R
oxygen, and this is able to deprotonate the attacking water molecule,
giving turnover. In contrast with the S phosphorothioate, an uncharged double-bonded oxygen is placed in
the R position. This cannot abstract a proton from
the water, and so no hydrolysis is seen. This proposal has some
difficulties, such as phosphodiesters having (at least in free
solution) the wrong pK
value to deprotonate water.
Nevertheless, it is clear from our results that the pro-R oxygen of
this phosphate has a very important function. However, it is possible
that the loss of catalysis could also be due to disruptions to the
interactions shown in Fig. 5. These differ slightly between the
two strands and involve direct contacts to the side chains of
Lys
and Thr
, as well as a set of water
mediated contacts that interconnects this phosphate with the adjacent
scissile phosphate and the catalytic apparatus. isomer. The pro-R oxygen of GACGATATpCGTC contacts the side chain
of Thr
and is also in contact, via two water molecules,
with the important pro-R oxygen of the neighboring 5` phosphate. One
assumes that the large reductions in rate seen with the R phosphorothioate of GACGATATpCGTC arise from
both alterations to its immediate protein contacts and to disturbances
to the critical preceding phosphate. The pro-S oxygen of GACGATATpCGTC
makes a hydrogen bond to the side chain of Thr
. In a
related article(75) , it is shown, using site-directed
mutagenesis, that Thr is the most critical of all
phosphate binding residues. As pointed out in this report, Thr is at one end of an 
-helix that also contains a critical
catalytic residue Glu. Furthermore, Fig. 5shows
that Thr interacts with Gln
on the other
subunit. This Gln is in the Q loop and approaches the minor groove of
the DNA. The two oxygen atoms of GACGATATCpGTC interact with the side
chains of Tyr
and Arg
, and this explains
simply the phosphorothioate results. At least one of the contacts will
be weakened by phosphorothioate substitution. We also note that the
pro-S oxygen forms a solvent-mediated contact with the side chain of
Gln. The main chain carbonyl oxygen of this amino acid is
a Mg ligand in the enzyme product
complex(39) .Role of Phosphates in the EcoRV Endonuclease-catalyzed
Reaction
The distinguishing feature of all restriction
endonucleases is their extraordinary accuracy, and the key to this is
shown in Fig. 5. The protein does not make isolated,
non-interacting, contacts to the bases in its recognition sequence and
the sugar-phosphate backbone. Rather, all the protein-DNA contacts are
intimately connected to each other and additionally to the metal ion
binding site and to important catalytic residues. Previously it has
been shown that DNA recognition is closely coupled to both metal ion
binding (42) and catalysis(43) , and this study
demonstrates that phosphate groups are also critically involved in
these processes.)At present we are unable to comment
in depth as to what each phosphate contributes, quantitatively, to
catalysis and specificity. This is because the steady state kinetics we
have used only report on the slowest step of the reaction, which for
these oligonucleotides is a mixture of the cleavage step and product
release(47) . Thus the results in Table 1report on the
important cutting step but may not provide a full and accurate
description of it. In addition, the symmetrical oligonucleotides used
place two phosphorothioates in the double-stranded 12`-mer, and in some
cases situations like this can give rise to non-additive
effects(33) . Nevertheless, we note that many of the
phosphorothioates have V/K values that are much less than 30% of the rate seen with the
control. This occurs not only for phosphates that may be directly
involved in catalysis (the scissile phosphate and the potential
phosphate base removed one step in the 3` direction) but also for
several others, removed from the cleavage site, such as GACGpATATCGTC,
GACGATATpCGTC, and GACGATATCpGTC. It has previously been suggested that
a 0.7 kcal/mol penalty, i.e. a drop in V/K to 30% of the control
value, is the most one might expect from a single phosphorothioate
substitution(12, 33) . In our case we have two
phosphorothioates, one per strand, and this would translate to a
reduction to 15%. We propose that the larger V/K reductions seen, arise
from an initial perturbation of the protein-phosphate interaction,
which is followed by changes in the network of hydrogen bonds outlined
in Fig. 5. These concerted alterations to the protein-DNA
interface cause the disruption of additional protein-DNA contacts, and
so result in poorer turnover than might be expected from simple
phosphorothioate substitution. Of the important phosphates, only
GACGApTATCGTC gives V/K reductions consistent with straightforward phosphorothioate
replacement. However, these preliminary suggestions await confirmation
using more sophisticated kinetic measurements.Relationship to Other Endonucleases
The idea that
protein-phosphate interactions change with non-cognate sequences and
that this contributes to selectivity is not new, and certainly we are
not the first to point it out. Using ethylation interference methods
with the EcoRI
endonuclease(9, 10, 12) , it has been
demonstrated that the phosphates contacted are dependent on base
sequence. With the cognate (GAATTC) sequence, a particular set of
endonuclease-phosphate contacts is seen and this pattern changes with
non-canonical sequences. These alterations in phosphate contacts
contribute significantly to the selection of cognate sites.
Furthermore, using single-turnover kinetics and binding studies with
oligonucleotides containing a single phosphorothioate of defined
stereochemistry, this group (33) also demonstrated that
phosphates played an important role in transition state stabilization
and began to dissect out the exact energetic contributions of the
substitution. The importance of phosphates, away from the cleavage, has
also been shown for this endonuclease by steady state
methods(34) . A related study with TaqI endonuclease (67) used S-methyl phosphorothioate triesters, formed
by the reaction of oligonucleotide phosphorothioates with methyl
iodide. This study both identified important TaqI phosphate
contacts but also led to the conclusion that sequence-specific
phosphate contacts were utilized in the transition state to amplify
selectivity and that base and phosphate recognition were tightly
linked. Thus, with EcoRV, EcoRI, TaqI, and
probably all restriction endonucleases, indirect readout of phosphates,
in a specific manner that is critically dependent on the presence of
the cognate site, plays a vital role in generating high selectivity.
There is no doubt that these protein-phosphate interactions deserve
further consideration and analogues such as the phosphorothioates and S-methylphosphorothioates as well as O-alkyl
phosphotriesters (34, 71, 72) and
methylphosphonates (73, 74) are available for this
purpose. At present, the most difficult step is the resolution and
purification of the two diastereomers seen with all these analogues,
and this is what is currently limiting further enzymological studies.
))
©1996 by The American Society for Biochemistry and Molecular Biology, Inc.
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
ABSTRACT
This article has been cited by other articles:
|
|
 |
|
  S. An, G. Barany, and K. Musier-Forsyth Evolution of acceptor stem tRNA recognition by class II prolyl-tRNA synthetase Nucleic Acids Res., May 1, 2008; 36(8): 2514 - 2521. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Armalyte, J. M. Bujnicki, J. Giedriene, G. Gasiunas, J. Kosinski, and A. Lubys Mva1269I: A Monomeric Type IIS Restriction Endonuclease from Micrococcus Varians with Two EcoRI- and FokI-like Catalytic Domains J. Biol. Chem., December 16, 2005; 280(50): 41584 - 41594. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Pingoud and A. Jeltsch Structure and function of type II restriction endonucleases Nucleic Acids Res., September 15, 2001; 29(18): 3705 - 3727. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Lanio, A. Jeltsch, and A. Pingoud On the possibilities and limitations of rational protein design to expand the specificity of restriction enzymes: a case study employing EcoRV as the target Protein Eng. Des. Sel., April 1, 2000; 13(4): 275 - 281. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. C. Horton, K. J. Newberry, and J. J. Perona Metal ion-mediated substrate-assisted catalysis in type II restriction endonucleases PNAS, November 10, 1998; 95(23): 13489 - 13494. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. C. Horton and J. J. Perona Recognition of Flanking DNA Sequences by EcoRV Endonuclease Involves Alternative Patterns of Water-mediated Contacts J. Biol. Chem., August 21, 1998; 273(34): 21721 - 21729. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. E. Johansson, D. Dertinger, K. A. LeCuyer, L. S. Behlen, C. H. Greef, and O. C. Uhlenbeck A thermodynamic analysis of the sequence-specific binding of RNA by bacteriophage MS2 coat protein PNAS, August 4, 1998; 95(16): 9244 - 9249. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| All ASBMB Journals | Molecular and Cellular Proteomics |
| Journal of Lipid Research | ASBMB Today |