Chem. Rev.

1998, 98, 1067−1087 1067

Metal Activation of Enzymes in Nucleic Acid Biochemistry

J. A. Cowan

The Evans Laboratory of Chemistry, The Ohio State University, 100 West 18th Avenue, Columbus, Ohio 43210

Received August 15, 1997 (Revised Manuscript Received January 15, 1998)

Contents
I. Introduction 1067
II. Physicochemical Properties of Magnesium 1068
III. Protein Binding Constants 1068
IV. Magnesium Analogues 1069
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

V. Kinetics/Inhibition 1070
VI. Mechanistic Issues 1072
A. Stoichiometry 1073
Downloaded via IISER BERHAMPUR on January 14, 2024 at 13:11:34 (UTC).

B. Inner-Sphere/Outer-Sphere Pathways 1074

C. Engineering the Active Site To Remove the 1075
Metal Dependency
VII. Nucleases (Restriction Enzymes) 1076
A. EcoRV and EcoRI 1076 James A. Cowan was born in Cleland, Scotland, in 1961. He earned a
B.Sc. Chemistry with first class honors from the University of Glasgow in
B. Other Restriction Nucleases 1077 1983, and a Ph.D. in Chemistry from the University of Cambridge in 1986,
C. Exonuclease III 1078 where he worked with Professor J. K. M. Sanders. His work with Professor
D. Ribonuclease H 1078 Sanders included the synthesis and study of complex porphyrin derivatives
E. General Nucleases 1079 designed as models for the electron-transfer events in the photosynthetic
reaction center. At the same time he worked with Professor A. D.
VIII. Polymerases 1080 Buckingham to develop a theory of chiral NMR. After receiving his
A. RNA Polymerases 1080 doctorate, he was granted a NATO fellowship to study with Professor H.
B. DNA Polymerases 1081 B. Gray at the California Institute of Technology. In 1988, he accepted
C. Reverse Transcriptase 1083 a position as an Assistant Professor in the Department of Chemistry at
The Ohio State University. He was promoted to Associate Professor in
IX. Phosphatases 1084 1993 and Full Professor in 1996. His research spans the fields of
X. Integrases 1085 chemistry and biology, including the reaction mechanisms of oxido-
XI. Closing Remarks 1085 reductase enzymes, the assembly of complex metal cofactors, the
XII. Acknowledgments 1085 biological chemistry of the alkali and alkaline earth ions, the chemistry of
viral and regulatory RNA response elements, and immunochemistry.
XIII. References 1085
otide binding. Such enzymes are of only passing
relevance in this review, although nucleotidyl trans-
ferases are the subject of greater focus since this
I. Introduction activity is relevant to the chemistry of RNA and DNA
polymerases. The polymerase domains of such en-
Most enzymes that participate in the biochemistry zymes also possess an additional labile metal-binding
of nucleic acids require divalent metal ion cofactors site in addition to the metal ion chelated to the
to promote activity.1,2 Magnesium is, with few excep- nucleotide di- or triphosphate. Acidic active-site
tions, the metal ion of choice, and its central role as residues can interact with this metal-phosphate
a metal cofactor in nucleic acid biochemistry empha- center and contribute to active-site chemistry, as
sizes the importance of understanding its mechanistic
exemplified by adenylate kinase.7 These kinds of
chemistry at a molecular level. This review will focus
binding interactions have been previously reviewed.8,9
attention on those enzymes that use free divalent
magnesium as a cofactor. There exists a large In this paper I will consider the functional role of
number of enzymes that utilize magnesium as a the magnesium cofactor in a number of enzyme
chelate with nucleotidyl di- or triphosphates (espe- families. As a prelude to this discussion, I will
cially ATP and ADP), where the metal cofactor serves summarize pertinent facts relating to the suitability
as a mediator of phosphoryl or nucleotidyl transfer of Mg2+ for its role as enzyme activator, required
chemistry, or hydrolytic elimination of phosphate or metal ion stoichiometry, methods for characterizing
pyrophosphate.3-6 In such enzymes the role of the its solution chemistry, and mechanistic details of
metal is fairly well understood, while association of function. This will provide an essential background
the metal to the enzyme is a consequence of nucle- for the subsequent review of specific enzyme families.
S0009-2665(96)00436-0 CCC: $30.00 © 1998 American Chemical Society
Published on Web 04/07/1998
1068 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

II. Physicochemical Properties of Magnesium Primarily, it is the tendency of Mg2+ to maintain a

moderate to high hydration state, and the involve-
The selection of magnesium as an enzyme metal- ment of these metal-bound H2O in mediating the
locofactor reflects the high natural abundance of this binding and catalytic chemistry of the magnesium
divalent cation, and a favorable combination of cofactor, that characterizes and is the hallmark of
physical and chemical properties (Table 1). These the chemical interactions of this metal ion with
include its redox inertness and small ionic radius, nucleic acid substrates.
resulting in high charge density, transport numbers,
and slow solvent exchange rates.1 There is a ten- III. Protein Binding Constants
dency to bind water molecules rather than bulkier
The design of metal binding sites on proteins is not
ligands in the inner coordination shell, which leads
ad hoc, but rather must satisfy certain constraints
to extensive hydration.1 In terms of these properties,
that relate to the physiological availability of the
the biological chemistry of divalent magnesium dif-
metal species. Clearly, proteins must bind metal ions
fers significantly from the other alkali and alkaline
with binding affinities (KDs) that are smaller than
earth metals. The distinction is clearly made by
the available concentration of these species in their
comparing the binding modes of magnesium and
respective in vivo environmentsotherwise the metal
calcium ions to biological macromolecules. There
cofactors would not bind. Creation of high-affinity
exists a large body of crystallographic data (sum-
metal binding pockets on proteins requires more
marized in ref 3) that demonstrates outer-sphere
extensive design than low-affinity sites and is there-
complexation by hexahydrated magnesium to be the
fore more demanding of evolutionary development.
normal binding mode for divalent magnesium to
If a metal cofactor is available in high concentrations
oligonucleotides (Figure 1). Similarly, binding sites
(greater than or equal to millimolar) there is no need
on proteins normally show three or fewer direct
for a high-affinity binding site since the metal bind-
binding contacts by protein side chains in order to
ing pocket would be populated under less stringent
minimize steric congestion around Mg2+,1,8,10 while
conditions, and so proteins typically bind metal ions
Ca2+ binding proteins typically show fewer bound
with KDs that match physiological availability. Se-
water molecules (e2) and an expanded coordination
lection between metal ions of comparable cellular
number of seven (Figure 2).11,12 This reflects the
concentrations can be made on the basis of ligand
larger ionic radius of Ca2+ and reduced steric barriers
type and coordination geometry.13 In the case of
to direct binding of larger protein-derived ligands.
intracellular free calcium ion, the concentration is
Table 1. Physicochemical Properties of Biologically
Relevant Alkali and Alkaline Earth Ions165
sub-micromolar and Ca2+ binding proteins show KD
e 1 µM, while extracellular Ca2+ proteins show KD
coordination transport > 1 mM, reflecting the higher extracellular concen-
no. kex (s-1) radius (Å) no.a
trations of calcium. As a result of the high cellular
Na+ 6 1010 0.95 7-13 concentration of free Mg2+ (∼0.5 mM free ion),14
K+ 6-8 1010 1.33 4-6 typical magnesium binding sites on nuclease enzymes
Mg2+ 6 5 × 105 0.65 12-14
Ca2+ 6-8 109 0.99 8-12 and ribozymes are of only moderate affinity (KD ≈
a Number of outer-sphere water molecules loosely associated
0.1-1.0 mM). Later it will be seen that this simple
fact provides a powerful constraint for analysis of
and transported with the mobile cation in solution.
metal-binding stoichiometry. Despite the similarity

Figure 1. Hexahydrated Mg2+ binds to the wobble base pair GT in the DNA strand 5′-CGCGTG-3′ (left) and a site on
tRNAPhe (right).161,162 Hydrogen bonds are shown by the dotted lines, and the distances (in Å) are oxygen-oxygen internuclear
spacings. Key: W, water molecule; OPO, a backbone phosphate.
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1069

Figure 2. Typical protein coordination modes for magnesium- and calcium-binding proteins, illustrating the tendency for
extensive hydration for the former. The Ca2+ site in staphylococcal nuclease (left) shows four direct binding contacts from
Asp21, Asp40, the backbone carbonyl of Thr41, and a phosphate from the nucleotide substrate mimic.163,167 One of two
additional bound H2O molecules is also shown. For magnesium-dependent nucleases the hexaaquo Mg2+ site in bacteriophage
E. coli T4 ribonuclease H (middle),102 and the Mg2+ site in avian sarcoma viral integrase (right) are shown.49
Table 2. Magnesium Binding Motifs metal must also vary, while maintaining a KD at a
motifa enzyme ref(s) physiologically balanced level. The availability of
-NADFDGD- RNA polymerase 19
high-resolution structural data on several metal-
-GDD- RNA polymerase 21 dependent nuclease enzymes allows a critical analy-
-D-NSLYP- and DNA polymerase φ29 129, 130 sis of the coordination chemistry of the bound cofac-
-K-NS(L/V)YG- tor, and the implications for an understanding of the
-YGDTDS- DNA polymerase R 22 functional mechanism.
-YXDD- reverse transcriptase 147
-LXDD- telomerase 166 In contrast to the situation for Ca2+ binding
a
proteins, where metal-binding motifs can be identi-
Letters in bold are metal binding ligands. fied from primary protein sequences,17,18 few such
general motifs are known for Mg2+ (Table 2, and
Figures 3 and 4). Some of these include the homolo-
gous -NADFDGD- motif found in RNA polymerases
isolated from E. coli, M. thermoautotrophicum, S.
oleracea chloroplast, and S. cerevisiae Pol I, II, and
III. By use of Fe2+ as a probe with hydroxyl radical
footprinting, and studies of Asp mutants, Mustaev
and co-workers have provided evidence that the
conserved triad of carboxylates illustrated in Figure
4 is involved in Mg2+ coordination.19 Such residues
are also observed in DNA Pol I, and HIV-RT.20 A
-GDD- motif has been observed for poliovirus RNA-
dependent polymerase, and it has been proposed that
the Asp-Asp pair is involved in metal binding.
Overall, magnesium binding sites appear to show
a greater homology in their three-dimensional struc-
ture, rather than in primary sequence. This is
Figure 3. Comparison of magnesium coordination spheres especially so in the case of the restriction enzymes,
for selected magnesium proteins (the metal binding do- and the family of bacterial and viral RNase H
mains of E. coli ribonuclease H, the chemotaxal protein domains and viral integrases, illustrated in Figure
CheY, and the DNA repair enzyme exonuclease III). 5.1,8,10
in KD, many metal-binding proteins, and magnesium-
dependent enzymes in particular, show a rich variety IV. Magnesium Analogues
of coordination motifs (Table 2 and Figures 3 and
4).8,10 These variations tailor the KD as described, Divalent magnesium is essentially spectroscopi-
but also accommodate the diverse functional roles of cally invisible and has such a low electron density
the metal cofactor. For magnesium, the importance that it is difficult to distinguish by most spectroscopic
of metal-bound H2O in the catalytic mechanism of and crystallographic experiments. For this reason,
many enzymes has already been illustrated8,15,16 and the chemistry of magnesium-dependent enzymes has
can extend to a requirement for specific coordination been studied with the use of transition metal probes
geometries for bound waters (Figure 2). Conse- and analogues. The underlying assumption in many
quently, the coordination mode for the protein-bound studies of metal cofactor requirement in nucleic acid
1070 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

Figure 4. The conserved -NADFDGD- metal-coordination motif from a selection of RNA polymerases from E. coli, M.
thermoautotrophicum, S. oleracea chloroplast, and S. cerevisiae Pol I, II, and III (adapted from ref 19). The three aspartate
residues most likely form a metal binding core. Note the different sequence spacing of the three binding residues relative
to the three binding residues shown for E. coli RNase H and Che Y in Figure 3.

biochemistry is that magnesium analogues (espe- ion which promotes activity levels that are greater
cially manganese) show similar chemistry to that than the requirements for metabolism. For example,
displayed by Mg2+. It will be seen that this is not a Myers and co-workers have explored the metal
good general assumption. In fact, the stoichiometry dependence of the ribonuclease H activity of Thermus
and coordination mode of other metals may, and thermophilus DNA polymerase and have shown that
frequently do, differ from Mg2+. In particular, it is Mn2+ promotes higher levels of RNase H activity than
usually observed that Mn2+ (or other transition Mg2+.32 This activity is manifest by 5′-3′ exonu-
metals such as Co2+, Fe2+, or Zn2+) confers higher clease digestion of RNA/DNA hybrids. Mn2+ acti-
levels of activity for both native enzyme, and Asp or vates optimally over Mg2+; however, Mg2+ is most
Glu mutants (since these typically coordinate more likely the natural cofactor. In fact, many class I
tightly than the natural Mg2+ cofactor). For example, RNases H are activated by Mn2+ and can utilize
the plant-stimulated nuclease from Fusarium solani Mn2+, but class II RNases H are inhibited by this
is stimulated by Mn2+ > Ca2+ > Co2+ > Mg2+, but is metal ion.33
inhibited strongly by Zn2+.24 These transition metal In summary, while the use of analogues can
ions may also show changes in coordination geom- provide insight on function, there are differences in
etry, or induce conformation perturbations in active- coordination chemistry that can lead to changes in
site residues that influence substrate specificity. substrate selectivity and metal ion stoichiometry.
The basis for metal ion mutagenicity and nucle- Such problems need to be recognized and accounted
otide selectivity in human DNA polymerase β has for during data analysis.
been examined.25,26 It has been demonstrated that
distinct nucleotidyl reactivity patterns arise with a V. Kinetics/Inhibition
variety of divalent metal ions, while these also
promote a change in side-chain position of Asp192 Measurement of enzyme activity as a function of
relative to Mg2+. E. coli DNA polymerase I shows solution conditions provides an effective means of
reverse transcriptase ability; however, the fidelity is characterizing the metallobiochemistry of magnesium-
lower with Mn2+ as the catalytic cofactor,27 and so dependent enzymes. The dependence of activity on
Mn2+ and other divalent analogues such as Be2+ are solution pH and cofactor concentration can be used
mutagenic and carcinogenic.28 Mn2+ mutagenicity of to report directly on the ligand environment in the
T7 DNA polymerase and E. coli DNA polymerase I active site, and on metal cofactor stoichiometry.34-36
arises as a result of reduced nucleotidyl discrimina- This is particularly pertinent to the issue of the metal
tion, even though the activity levels are higher.29 cofactor requirement to effect catalysis, where there
In keeping with the aforementioned trends for is inconsistency between crystallographic and solu-
polymerase enzymes, loss of specificity for the TaqI tion measurements.10 This issue will be addressed
restriction endonuclease is observed with Mn2+ rela- later, but can be readily approached in solution
tive to the natural Mg2+ cofactor.30 Evidence for studies by a combination of kinetic and thermody-
three types of ternary complex with Mn2+ are pre- namic measurements.
sented, presumably each reflecting distinct Mn2+ Several methods for monitoring activity have been
coordination modes. Such loss of sequence specificity reported. Often these involve the use of radiolabeled
with the use of metal cofactors other than Mg2+ is substrates and evaluation of the rate of formation of
common for EcoRI and other restriction nucleases.31 labeled fragments.37-39 A more convenient method,
It has been earlier noted that Mn2+ will often when applicable, is direct monitoring of the change
promote higher levels of enzyme activity than Mg2+. in absorbance that results from hydrolysis of a
This should not be misconstrued as an indication that nucleic acid oligo- or polynucleotide substrate.16,34,36,40
Mn2+ is in fact the natural cofactor, but simply A variety of mechanistic models have been pro-
reflects the distinct chemistry of this transition metal posed and used to rationalize the metal dependence
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1071

Figure 5. Homology in the tertiary structural arrangement of the active-site metal binding residues in some magnesium-
binding proteins. (A) (Left) Schematic illustration of the two Mn2+ ions identified in the active site of the HIV-RT RNase
H domain after doping the cofactor-free crystal with 45 mM Mn2+. The positions of key carboxylate residues are shown.
(Right) The structurally homologous sites in E. coli RNase H and avian sarcoma viral integrase, showing the positions of
conserved carboxylate residues (the numbering for ASV-IN is shown in parentheses). Only one metal-binding site (Mg2+
or Mn2+) has been identified.49,97 The metal ions in bold are the major homolgous sites common to the enzymes shown in
A and B. (B) A comparison of the metal cofactor binding domains, and other features of the catalytic pockets of EcoRI and
EcoRV, illustrating the high degree of local homology for these two restriction endonucleases.38
of nuclease activity. This issue has been most form of competitive binding between the metal ions
thoroughly addressed by the laboratories of and enzyme, to the substrate. Such a viewpoint is
Pingoud,38,41 Halford,39,42 and Cowan.17,34 A general supported by the similarity of this response over a
observation in almost all plots of enzyme activity as diverse group of enzymes,34-36,43 and the similarity
a function of metal cofactor concentration is an initial of apparent binding constants for metal-substrate
increase in activity with increasing [Mg2+], followed complexation determined from these kinetic and
by a gradual decrease in activity (Figure 6). While thermodynamic measurements. The effect is clear
direct enzyme inhibition is a possibility, no direct in the case of E. coli RNase H, where there is no
evidence has been advanced for such, and the author evidence for secondary magnesium binding sites on
of this review is inclined to see this effect as a general the enzyme,34 while metal-hybrid and metal-
example of substrate inhibition, where the metal ions enzyme binding constants (KI ≈ 14 mM, and KMET ≈
bind to the nucleic acid substrate, perturbing the 0.2 mM, respectively), obtained by kinetic measure-
charge density of the substrate and its interaction ments for E. coli RNase H34 and by independent NMR
with the enzyme. This may also be thought of as a and calorimetric titration methods, are found to be
1072 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

Figure 6. Variation of initial velocity (vo) with magnesium concentration for E. coli RNase H (left) and the exonuclease
activity of Klenow (right). The data clearly shows the inhibitory influence of high concentrations of Mg2+, and was fit to
a model that assumes metal-mediated substrate inhibition.34,62

similar.44,47,62 This also provides good evidence for

catalytic activation by the same metal ion that binds
to the enzyme alone.
An important observation is that this kind of
inhibition effect is diminished by the presence of
polyamines, which competitively displace Mg2+ ions,
but have an overall reduced charge density relative
to a number of divalent magnesium ions bearing an
equivalent charge.36 Such an observation may be
taken as a diagnostic test for this kind of metal-
mediated inhibition. The generality of a metal-
mediated inhibitory influence on substrate interac-
tions with enzyme is consistent with the observation
that polyamines (spermine, spermidine, and pu-
trescine) modulate DNA-dependent DNA polymerase
activity.45 These species shift the optimal Mg2+
concentration of polymerase from 10 mM to a more
physiological level. For example, polyamines were
found to increase (by 2-6-fold) the activity of ma-
larial R-like DNA polymerase at suboptimal concen-
trations of divalent magnesium.46 This general phe- Figure 7. (A) An alternative minimalist scheme for
nomenon can be attributed to the requisite neutral- Klenow activity, based on the theoretical analysis of
ization of charges on the nucleic acid substrate, but Aaqvist and Warshel,58 showing one essential metal cofac-
without the inhibitory influence observed with higher tor and an activated nucleophilic solvent water. (B) Sche-
concentrations of Mg2+. The higher binding affinities matic illustration of the metal binding sites identified from
of polyamines also renders these cations more effec- crystallographic studies of the Mn2+ derivative of the DNA
polymerase I Klenow fragment.48,100 Site B is weakly
tive than monovalent alkali metal ions. populated in the presence of substrate or a substrate
The observation of sigmoidal behavior, or a lag analogue. Site A is the principal site of both metal
phase, in some kinetic profiles of activity versus coordination and catalytic chemistry.
metal cofactor concentration may be taken as evi-
dence for the requirement of additional weakly bound transition metals is often viewed as the nucleophile,
metal cofactors that may either be required for the higher pKa of Mg2+-bound H2O makes metal
catalytic activity or to stimulate substrate binding. activation of H2O less favorable in the case of divalent
magnesium. Warshel and co-workers have published
VI. Mechanistic Issues an insightful theoretical analysis of the Klenow
In sections II and III, I have presented evidence fragment from DNA polymerase I, which is more
that favor outer-sphere-mediated hydrolytic path- generally applicable and identifies electrostatic sta-
ways for a large section of magnesium-dependent bilization as the principal role of the metal cofactor-
enzymes that carry out reactions on nucleic acid (s).64 These workers argued against a magnesium-
substrates. The activity of substitutionally inert bound hydroxide as the active nucleophile, but favored
complexes indicates that metal-bound H2O is not an external hydroxide that is stabilized electrostati-
necessarily required as a nucleophile, but rather a cally by the metal cofactor. Stabilization of this
free solvent molecule will serve as the hydrolytic hydroxide could be accounted for in terms of the
agent, perhaps with base catalysis from a neighbor- electrostatic influence of the high affinity metal ion
ing side chain. While H2O bound to Zn2+ and other alone (Figure 7A). The role of the putative low-
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1073

affinity metal ion (site B) was discussed in terms of However, this question can be addressed by kinetics
electrostatic stabilization of the negative charge that methods. One strategy that has been used success-
is transferred from hydroxide to phosphate. Such a fully involves determination of metal binding param-
model was also favored for one metal ion enzymes eters from kinetics experiments, and comparison of
(such as Staphylococcal nuclease) where the single the fitted binding parameters with results from
cation carries out both roles.58 In this, and other independent calorimetry experiments carried out in
cases where single metal ion cofactors have been the presence of substrate analogues.34,62 The results
identified, the metal is typically divalent calcium. are analyzed with a view to identifying specific
Inasmuch as all Ca2+-dependent nucleases act by one mechanistic models that yield consistent results from
metal ion catalytic pathways, it seems reasonable to each independent approach and, importantly, that
infer that magnesium, with a charge density that is are consistent with the physiological availability of
significantly higher than that of Ca2+,67 would also the divalent metal ions. Nature imposes well-defined
show a requirement for one metal cofactor. This is limits on metal binding stoichiometry. This is con-
indeed generally substantiated by available evidence. trolled by both the magnitude of the dissociation
constant and the bioavailability of metal cofactor in
A. Stoichiometry vivo. For example, a 40 mM binding affinity is of
The stoichiometry of metal cofactor that is required little significance if the available concentration of the
to mediate phosphate ester hydrolysis is an impor- cofactor is 0.5 mM. In effect, the dissociation con-
tant variable for mechanistic considerations. This stant for a biologically relevant divalent magnesium
problem has been the subject of debate as a result of cofactor should be ∼0.5 mM, since this is the con-
contradictions between solution and crystallographic centration of free magnesium in a cellular environ-
studies47-50 and has fueled recent efforts by chemists ment.14 This is a powerful constraint in assessing
to construct synthetic models to address this issue.51-53 metal ion stoichiometry and distinguishing kinetic
Several binuclear metallohydrolases that contain a models.
pair of bridged transition metal ions at the active site Evaluations of metal cofactor stoichiometry by
have been characterized.2,54 However, while there crystallographic and solution studies have been made
are a few examples of metabolic enzymes that appear for several enzymes (summarized in Table 3). There
to utilize two magnesium centers in close proximity, is now extensive data available for E. coli Klenow
such as xylose (glucose) isomerase,55 and glutamine fragment, RNase H, Exo III, and EcoRV. Some of
synthase,56 the evidence for binuclear magnesium in these results will be presented in greater detail in
nucleic acid hydrolysis is not so clear-cut and is the sections dealing with specific enzymes that follow;
supported by a rather small body of crystallographic however, some general comments can be made that
data with relatively few confirmatory results from shed considerable light on this dichotomy. X-ray
solution experiments. The two metal ion mechanism structures of proteins with labile Mg2+ cofactors are
has been proposed for numerous metal hydrolase,2 usually obtained by doping preexisting crystals of
nuclease,39,57,84 polymerase,58 and ribozyme reac- “apo” enzyme with very high concentrations of metal
tions;59,60 however, solution thermodynamic and ki-
cofactor to permit population of metal binding sites
netic measurements on substrate-free enzyme typi-
in a kinetically reasonable time frame. Often the
cally indicate a single metal cofactor (Table 3). An
metal used is not the natural magnesium cofactor,
important caveat to note is that simple binding
which has a low electron density and is difficult to
experiments do not directly indicate the number of
metal ions required for catalytic turnover; for ex- distinguish from oxygen in water, but rather a
ample, it is possible that an additional metal binding transition metal ion carrying a higher electron den-
site is found in the enzyme-substrate complex, as sity which is more easily observed. Increasingly,
suggested for the Klenow fragment of DNA pol I.48,61 crystallographic studies are using the natural Mg2+
cofactor and making use of the strong pattern of
Table 3. Summary of Metal Binding Constants and octahedral coordination by metal-bound H2O to dis-
Stoichiometries Evaluated from Solution tinguish Mg2+ from general solvent water. Although
Measurements and Crystallographic Analysis10 much useful information has come from studies with
crystallography solution magnesium analogues, a few simple caveats must be
no. of [M2+] no. of KD
kept in mind. (1) Coordination sites in a preexisting
enzyme M2+ used (mM) M2+ (mM) structure of an apoenzyme may not exist in the
S. nuclease 1 0 (Ca2+)a 1 0.005
structure of the active metalloenzyme. (2) Coordina-
Klenow exonuclease 2 12 to 50 (Mn2+) 1 0.40 tion preferences for Mg2+ and analogues are often
Exo III 1 20 (Mn+) 1 0.11 different. (3) Under conditions of high metal con-
EcoRV 1 or 2 10 to 30 (Mg2+) 1 NDb centration used in doping experiments, weak and
EcoRI 1 1 ND physiologically irrelevant sites may be occupied.
HIV-RNase H 2 45 (Mn2+) ND
E. coli RNase H 1 50 to 100 (Mg2+) 1 0.5 The paradigm of the two metal ion model is
Che Y 1 50 (Mn2+) 1 0.2 perhaps best defined as the requirement for two
ASV-IN 1 200 (Mg2+) ND metal ions, located in close proximity (<4 Å), that
a The Ca2+ cofactor binds sufficiently tightly that it can be are bridged by a common substrate. This allows a
isolated in the metal-bound state and requires no additional clear distinction from other enzymes that may bind
solution Ca2+(aq) to maintain saturation of this site. b ND, not two metal cofactors in the same catalytic domain, but
determined.
do not function as a coherent catalytic unit. For
1074 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

example, crystallographic analysis of T5 5′-exonu-

clease shows two Mn2+ sites for data collected in the
presence of 25 mm Mn2+.63 These sites are, however,
separated by 8.1 Å (10 Å in the case of Taq 5′-
exonuclease), which is significantly greater than the
<4 Å separation observed in the putative two-metal
nucleases. Two Mg2+ sites are separated by 7 Å in
T4 RNase H (see Figure 16, and the discussion in Figure 8. Schematic illustration of transition-state sta-
section VII.D). Accordingly, one would expect a bilization by outer-sphere complex formation to hydrated
difference in mechanism from the two-metal variety magnesium ion. Both hydrogen bonding and electrostatics
contribute to the stabilization of the increased negative
for these, and related enzymes. charge in the transition state; however, for divalent
The stoichiometry of catalytic metal ions is better magnesium, the hydrogen-bonding contribution is domi-
characterized for calcium-dependent nucleases. These nant.16,35
have been more extensively studied, for reasons that
include: (1) calcium-binding proteins typically bind sition-state stabilization is effected by hydrogen
the cofactor with higher affinity than magnesium, bonding with essentially no contribution from elec-
and so the site is more readily populated; (2) calcium trostatic stabilization through the charge on the
is a more electron-rich cation, and is therefore easier cofactor.16,35,66 The solvation state of the metal co-
to identify; and (3) there are significantly more factor is of critical importance for the proper func-
examples of well-defined structures of calcium pro- tioning of these enzymes and is defined by the
teins available for analysis. Overall, there is exten- number of protein ligand contacts. The metal bind-
sive experimental data from solution and structural ing pockets of metal-dependent nucleases have evolved
studies on a wide variety of enzymes to support a one to allow a large variation in the number of protein
metal ion model as the general mechanistic scheme ligands (and thereby solvent water), while optimizing
in considerations of magnesium- or calcium-mediated the binding affinity to physiological requirements.
hydrolysis reactions in nucleic acid biochemistry. The distinct coordination properties of Mg2+ (six
coordinate), versus the larger Ca2+ (six to eight
B. Inner-Sphere/Outer-Sphere Pathways coordinate), most likely favors inner-sphere over
outer-sphere chemistry for the latter, as described in
With the background provided by sections II and section II.
III, it is clear that protein-bound Mg2+ shows exten- A major difficulty in distinguishing inner- and
sive hydration. It might be expected, then, that these outer-sphere pathways for magnesium-promoted re-
metal-bound water molecules would be utilized in actions stems from the kinetic lability of this metal
enzyme-mediated chemistry. While direct binding of ion. One approach that has been successfully applied
a substrate molecule to a metal cofactor is common is the use of cobalt and chromium complexes that are
for transition metals, the assumption that magne- substitutionally inert within the time frame of the
sium cofactors generally serve as simple Lewis acids enzyme-catalyzed reactions.9,37,68,69 If such complexes
is inconsistent with the observation that substitu- promote a magnesium-dependent reaction, this is
tionally inert transition metal complexes, incapable evidence in favor of an outer-sphere pathway. Of
of direct binding to substrate, can promote the course the absence of activity promoted by such
reactions of a number of important magnesium- complexes does not necessarily preclude such a
dependent enzymes.16,35,37 In magnesium biochem- pathway.
istry, outer-sphere pathways that require a solvated Catalytic activation of E. coli ribonuclease H by a
metal cofactor are important, and are perhaps the series of inert chromium complexes [Cr(NH3)6-x
major mechanistic class displayed during metal- (H2O)x]3+ (x ) 0-6) that bear water and ammonia
mediated phosphate ester hydrolysis. As an aside, ligands in well-defined geometries in the inner
it is intriguing to note that crystallographic data coordination shell has been examined.16,35 This ap-
collected on zinc carboxypeptidase suggests that the proach has been applied also to E. coli exonuclease
peptide carbonyl oxygen never forms a direct coor- III,35 topoisomerase I,70 and other enzymes and
dinate bond to the zinc ion.65 The metal cofactor ribozymes9,71 to establish outer-sphere pathways.
seems to electrostatically stabilize the buildup of Such complexes are observed to function by transi-
negative charge on the carbonyl O in the transition tion-state stabilization, and the approach permits one
state, and also facilitates deprotonation of a bound to distinguish hydrogen bonding from electrostatic
nucleophilic H2O. contributions inasmuch as NH3 bound to Cr(III) or
This idea of outer-sphere activation is, therefore, Co(III) centers has negligible H-bonding propensity.66
an important one that is likely to be of wide rel- Moreover, since complexes of the general form
evance. Figure 8 illustrates outer-sphere catalysis [Cr(NH3)6-x(H2O)x]3+, with well-defined, but variable,
by stabilization of the developing negative charge in coordination geometry are synthetically available,
the transition state of the reaction. According to this these also afford a probe of the preferred structural
mechanism the metal cofactor serves principally to arrangement for hydrogen-bonding interactions. For
stabilize the transition state, either electrostatically, example, it has been found that only those complexes
and/or through hydrogen bonding from the metal- with a facial array of bound water molecules promote
bound waters. The latter appears to be the larger catalysis of E. coli ribonuclease H; as expected from
contribution in the few cases that have been exam- comparison with the ligation of the enzyme-bound
ined.16,35 Experimental data clearly show that tran- Mg2+ cofactor. Figure 9 illustrates the crystallo-
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1075

Figure 9. The crystallographic coordination modes for

magnesium binding to E. coli ribonuclease H and exonu-
clease III (compare Figure 3), illustrating the distinct
solvation states of each ion. These coordination states are
compared to inert chromium complexes that show the
minimal hydration requirements for efficient promotion of
enzyme-catalyzed phosphate ester hydrolysis for each
enzyme. For ribonuclease H, the mer-[Cr(NH3)3(H2O)3]3+
complex is inactive, while for exo III, those complexes with
one less water of hydration (namely cis- or trans-[Cr(NH3)2-
(H2O)4]3+]) show significantly reduced activity.16,35

graphically defined hydration state of the natural

magnesium cofactor and also shows the minimal Figure 10. Active site of native E. coli ribonuclease H. A
hydration states required by the inert metal probe double mutant of active-sites residues (Asp10Arg/Glu48Arg)
complex to efficiently mediate catalytic activity.16,35 yields a functional enzyme with a distinct pH profile,
There is an apparent relationship between the re- relative to native.72
quirements for metal-bound solvent interactions with Table 4. Activities of Mutants Relative to Native
the substrate, and the number of coordinating protein E. coli RNase H72
ligands, which ultimately controls the hydration state enzyme kcat (s-1)
of the bound metal. Relative to ribonuclease H,
native 28 ( 8
exonuclease III requires more extensive solvation of native (-Mg2+) a
the metal cofactor for optimal activity, and this is Glu48Asp 1.1 ( 0.3
reflected in the smaller number of protein ligands Glu48Gln a
(Figure 3). Asp70Asn a
Asp10Glu 14 ( 4
C. Engineering the Active Site To Remove the Asp10Asn a
Asp10Ser a
Metal Dependency Asp10Arg a
The preceding section has described a mechanistic Asp10Gly a
Asp10Arg/Asp70Lys a
model for enzyme-catalyzed metal-mediated hydroly- Asp10Arg/Glu48Arg (-Mg2+) 27 ( 7b
sis of a nucleotide backbone,16,35 through stabilization a Activity is below detection levels. b For the Asp10Arg/
of a transient intermediate by formation of a hydro-
Glu48Arg mutant the activity obtained at pH ≈ 5.5 is reported.
gen-bonded outer-sphere complex with the positively Other data was obtained at pH 7.3.
charged hydrated metal cofactor. Consideration of
this model led to the hypothesis that mutation of or single-point mutations of the magnesium-binding
active-site carboxylate residues to positively charged residues, Asp10 and Glu48, yield inactive enzyme,
Lys or Arg might provide sufficient positive charge either with or without added Mg2+ (Table 4), with
density and hydrogen-bonding propensity in the the expected exception of the Asp10Glu and Glu48Asp
active-site domain to mimic the role of hydrated mutants which retained their ability to bind metal
divalent magnesium. This idea has been successfully cofactor and exhibited 50% and 4% activity, respec-
demonstrated by engineering novel metal-indepen- tively. Introduction of a single positive charge (for
dent active mutants of E. coli RNase H (Figure 10).72 example, Asp10Arg) was found to be insufficient for
Of the three principal active-site carboxylate activation, while also serving to inhibit binding of
residues, crystallographic evidence suggested that magnesium cofactors. In contrast, at low pH the
both Glu48 and Asp10 are bound to Mg2+, while double mutant Asp10Arg/Glu48Arg demonstrated
Asp70 has been proposed to serve as a catalytic base almost native levels of initial activity, even in the
that is required for deprotonation of water prior to absence of added Mg2+, although significant product
hydrolysis of the backbone (Figures 2 and 10).73 inhibition was observed, presumably as a result of
Mutation of the essential catalytic base (Asp70Asn), the increase in positive charge density in the catalytic
1076 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

EcoRI is also a type II endonuclease and shows a high

degree of structural homology with the active site of
EcoRV (Figure 5B),38,41,82 although there is no general
sequence homology. Divalent magnesium is an es-
sential cofactor for type II endonucleases. A mecha-
nistic model has been suggested for EcoRI and EcoRV
where the attacking water is activated by the phos-
phate placed 3′ to the scissile phosphodiester and the
leaving group is protonated by a water molecule
associated with the Mg2+ cofactor.38,41 This effect has
been described as substrate-assisted catalysis and
has been supported by experiments that remove the
phosphate group (substituting with an H-phospho-
Figure 11. Active site of Serratia nuclease, showing the nate) and by phosphorothioate substitution. In gen-
metal cofactor and Arg residues that likely contribute to eral, activity is lost or severely diminished following
transition state stabilization.36 these substitutions, and in the latter case, activity
is only observed if the negatively charged sulfur is
pocket. Both an increase in product inhibition and
in the RP configuration. Nevertheless, there would
a requirement for a lower pH to achieve reasonable
appear to be a requirement for participation by a
levels of activity are reasons for the use of metal
carboxylate residue to provide base catalysis, since
cofactors rather than positively charged side chains
the pKa of a phosphodiester is too low (pKa < 2) for
in enzyme design. These observations are consistent
this role.
with a model of substrate activation by native
enzyme that is dominated by hydrogen bonding from Unlike EcoRV, EcoRI produces sticky end products,
waters of solvation. In the double mutant, the role since the homologous active site is offset relative to
of the hydrogen bond donor is most likely accom- the DNA recognition sequence (5′-GAATTC-3′; Figure
modated by the guanidinium centers of the mutant 12). In the presence of Mg2+, association of EcoRI
side chains. The use of guanidinium centers in with palindromic tridecadeoxynucleotides containing
natural enzymes to further enhance transition-state the recognition site occurs at an almost diffusion-
stabilization is well-documented for Serratia nuclease controlled rate.83 Figure 5B illustrates how both
and homologues (Figure 11),36 RNase T1,74 and enzymes show a similar structural arrangement in
nuclease P1.75 the vicinity of the scissile phosphate, and form a motif
that is composed of four active-site residues (Pro90,
VII. Nucleases (Restriction Enzymes) Asp91, Glu111, and Lys113 for EcoRI; and Pro73,
Asp74, Asp90, and Lys92 for EcoRV) that appear to
Few magnesium-dependent nuclease enzymes have form a binding pocket for the essential divalent metal
been studied in detail insofar as the role of the cofactor.38,41 For EcoRI, both kinetic and structural
essential metal cofactor is concerned. Noteworthy studies are consistent with one essential metal co-
exceptions are the restriction enzymes EcoRI, and factor in the active site.38,41 The role of the metal
especially EcoRV, which has been the subject of lively cofactor has been most clearly elucidated for EcoRV
competition between the laboratories of Halford and and this is now taken up, although the mechanistic
Pingoud.38,39,65 There is now an extensive body of details of metal-promoted hydrolysis are most likely
structural and mechanistic information concerning similar for each enzyme.
the role of the metal cofactors in the function of this Unlike other restriction enzymes in this class,
enzyme, and Pingoud and Jeltsch have recently which demonstrate specific binding in the absence
published a thorough review of the recognition and of Mg2+, EcoRV binds nonspecifically in the absence
cleavage chemistry of type II endonucleases, the
of the divalent cofactor.77 In 1993, on the basis of
family to which both of the aforementioned enzymes
kinetic evidence, Pingoud and co-workers proposed
belong.76
a requirement for one metal ion to effect EcoRV
catalysis,38,41 but also pointed out that binding of an
A. EcoRV and EcoRI additional high-affinity metal ion could not be ex-
Figure 12 shows that EcoRV binds and cleaves cluded since an additional residue (Glu45) was avail-
double-strand DNA in a blunt-end fashion with a able for binding and coordination of heavy metals
high degree of specificity at sites defined by a unique (such as Pb2+) had been recognized in crystallo-
sequence of six contiguous bases (5′-GATATC-3′).38 graphic experiments. The 2-Å resolution structures
of dimeric EcoRV cocomplexed with both an un-
decamer substrate and product (Figure 13),50 pub-
lished by Kostreiva and Winkler, showed one Mg2+
bound to the scissile phosphodiester group and two
carboxylate oxygens from Asp 74 and Asp 90 in the
substrate complex, while the product complex showed
two Mg2+ bound to each 5′-phosphate ester oxygen
Figure 12. Schematic illustration of the recognition (Figure 13C) (Mg32+ and Mg42+). On the basis of this
sequences for EcoRI and EcoRV. The arrows indicate the structure, and other kinetic data, Halford and Win-
cleavage sites. kler and co-workers subsequently proposed a two
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1077

Figure 13. (A) Schematic illustration of the catalytic pocket for the EcoRV-substrate complex. Note that only one metal
cofactor is identified. (B) Proposed model for the transition state, showing recruitment of a second metal cofactor. However,
in other work, the critical residue Glu45 has been mutated to residues that do not bind magnesium and the mutants were
found to be active. (C) Schematic illustration of the catalytic pocket for the EcoRV-product complex. Note that now two
metal ions are observed, although metal ion concentrations in the range of 10 to 30 mM were used. Adapted from ref 50.

metal ion transition-state model. Later, Halford and ions is required for catalysis; the other appears to
co-workers also examined the metal requirement for promote binding of the substrate molecule. This
EcoRI and EcoRV and correctly pointed out that the most likely corresponds to the additional metal
Pingoud analysis incorrectly assumed total cooper- cofactor suggested by the stopped-flow kinetic studies
ativity.39 They suggested a catalytic requirement of of Halford and co-workers.39,42
one metal cofactor for the former, and a two metal Specific recognition of DNA by EcoRV can be
ion mechanism for the latter. It was observed that promoted by Ca2+ ion. By analyzing gel shifts with
Ca2+ inhibits Mg2+ activation of EcoRV, but stimu- a permuted set of DNA fragments, the degree of DNA
lates Mn2+ activation of the same enzyme, suggesting bending was shown to be similar to that seen in the
that Ca2+ can displace a metal from one of the two crystal structure of the cognate DNA-protein com-
sites. Such a model was supported by stopped-flow plex in the presence of Mg2+.57 Calcium ion, there-
fluorescence experiments that examined the influ- fore, mimics the ability of Mg2+ to generate a specific
ence of metal ions on DNA binding to EcoRV.42 Two protein-metal-DNA complex, but is incapable of
Mg2+-dependent transitions were observed and in- inducing the cleavage reaction. This may also reflect
terpreted in terms of a model where one Mg2+ binds the tendency of Ca2+ to engage in inner-sphere rather
to the active site before phosphodiester hydrolysis, than outer-sphere interactions, if the latter are
with a second binding subsequently to a preformed required for catalysis. Outer-sphere activation is also
enzyme-DNA complex. consistent with the absence of any significant influ-
These facts notwithstanding, several problems ence on the relative cleavage rates for Mg2+- versus
nevertheless remained with this interpretation. First, Mn2+-induced activity with phosphorothioate sub-
the crystal complex with the substrate lacked activ- strates, since direct coordination to the metal cofactor
ity, even in the presence of saturating magnesium, should discriminate between a hard Mg2+ ion and
and the active site bound only one divalent magne- softer Mn2+ ion.79
sium ion (Figure 13A). Second, a Mg2+ concentration Recent results for an Ile91Leu mutant show a
of g10 mM was required to populate both sites; greater than 1000-fold decrease in activity.80 A
however, the key issue is whether these sites can be change in metal ion dependency was observed, with
populated under physiological conditions (<1 mM). a preference for Mn2+ rather than Mg2+ as for the
Third, and perhaps most importantly, the critical native enzyme. The mutant shows evidence of non-
residues that define the binding site of the second specific DNA binding in gel-shift experiments, as does
metal ion (Mg22+ in Figure 13B), especially Glu45, the native. At noncognate sites that differ from
have been mutated to nonbinding residues (such as EcoRV by one base pair, Mn2+ gives higher cleavage
Ala) and the mutants were found to be active.78 It rates than Mg2+, but the effect is reversed for the
has therefore been suggested that a second, but Ile91 Leu mutant.81 Since each mutant requires the
spatially distinct and noncatalytic site might exist,77 same carboxylate residues for binding, compared to
since the DNA binding specificity is increased by native, it is likely that the switch in metal preference
Mg2+ binding to a site that is distinct from that of arises from a structural perturbation of the metal
the catalytic center (bound by Glu 45, Asp 74, Asp binding pocket, as described in section IV.
90). Also, the Ala triple mutant which cannot bind
catalytic metal, was found to bind specifically to DNA B. Other Restriction Nucleases
in the presence of Mg2+. Phosphorothioate deriva-
tives provided further evidence for a Mg2+ site There is a paucity of structural and mechanistic
distinct from the catalytic center. Only one of these information concerning other restriction enzymes,
1078 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

especially with regard to the chemistry of the metal

cofactor. Nevertheless, the constellations of active-
site metal-binding residues in other structurally
characterized endonucleases (BamHI,84 and PvuII85)
is similar to that of EcoRI and EcoRV, summarized
in Figure 5B,86 suggesting that each of these enzymes
proceed by similar one metal catalyzed mechanisms,
and most likely by outer-sphere pathways (although
such enzymes are not activated by substitutionally
inert complexes9). Two points are particularly note-
worthy from these structural studies. First, the
similarity in tertiary structure in the magnesium
binding site is not reflected in the primary sequence;
and second, there is extensive structural similarity
in the active sites of these restriction enzymes
relative to the 3′-5′ exonuclease domain of the Klenow
fragment. The restriction nucleases are likely to
proceed by a one metal ion pathway, while the
Figure 14. (Left) Calorimetric titration of E. coli exonu-
stoichiometry for Klenow will be addressed in section clease III with Mg2+. The raw data is shown above, and
VIII.B later in the review. The key issues of blunt the integrated heats and fitted data are below. (Right) A
or sticky end cutting, and sequence recognition for similar experiment with Mn2+ as titrant.10,47
the restriction endonucleases, are most likely dictated
by the positioning of other residues at the enzyme- showed only one bound Mn2+;89 and so one class of
substrate interface, with little involvement (except site appears not to be populated, and is possibly
for a few cases such as EcoRV) by divalent metal sterically inaccessible for the crystals of native
cofactors. As a side note, it has been observed that enzyme with the doping strategy employed (compare
introduction of phosphoramide bonds at the cleavage the case of Klenow, vide supra). This result is
position of one strand blocks cleavage of that strand particularly pertinent to the issue of one metal versus
for the endonucleases EcoRII and SsoII, and in the two metal ion mechanisms in phosphate ester hy-
case of EcoRII reduces the rate of cleavage of the drolysis. Studies with substitutionally inert com-
other natural strand.87 In the presence of the Mg2+ plexes support an outer-sphere mechanism of hydrol-
cofactor, association rates are reduced 3-fold and ysis,35 and inhibition at higher metal cofactor concen-
dissociation rates are increased 1.5-fold, and so trations is again observed.35,90
recognition and specific binding is strongly influenced
by the metal cofactor. This is consistent with the D. Ribonuclease H
breakdown in recognition specificity of EcoRI and
other endonucleases (such as ApoI, AseI, BamHI, One of the best characterized examples is the
EcoRI, EcoRV, HindIII, KpnI, PstI, PvuII, and TaqI, Escherichia coli endoribonuclease H, a low molecular
among others)31 with magnesium analogues, as pre- weight enzyme (Mr ≈ 17 580) that hydrolytically
viously noted in section IV. cleaves the ribonucleotide backbone of RNA‚DNA
hybrids, producing 5′-phosphate and 3′-hydroxyl oli-
gonucleotides. Both native73,91 and mutant92-94 E.
C. Exonuclease III coli RNase H enzymes have been structurally char-
acterized, as well as other members of this general
Escherichia coli exonuclease III is a monomeric low family. Other relevant proteins include CheY, a
molecular weight (Mr ≈ 28 kDa) apurinic/apyrimi- chemotaxal protein that shows significant structural
dinic endonuclease that serves an integral role in homology with E. coli RNase H in the metal-binding
DNA repair.88 The crystal structure of the enzyme active site.95 E. coli ribonuclease H is structurally
has been elucidated in the metal-bound state (for homologous to the RNase H domain of HIV reverse
Mn2+ and Sm2+ derivatives) and shows one metal transcriptase (RT) (Figure 15) and shows conserva-
cofactor at the active site (Figure 3).89 This is a tion of key active-site residues (Figure 5A),91,96,97
particularly valuable example for illustration of the nevertheless, no consensus has been reached on the
difference in bound metal stoichiometry when evalu- stoichiometric requirement for metal cofactors in
ated for different metal ions. The binding of Mg2+, these two enzymes. The E. coli enzyme has been
Ca2+, and Mn2+ to the E. coli ribonuclease H and extensively characterized by NMR methods (both
exonuclease III enzymes has been compared in solu- structural and functional studies),98 including metal
tion, and clearly demonstrates 1:1 stoichiometry for binding.99
metal binding to ribonuclease H, but not for exonu- A crystallographic analysis of the Mn2+-doped
clease III (Figure 14).47 For both Mg2+ and Ca2+ ions, crystals of the RNase H domain of HIV-RT revealed
one binding site was determined. In contrast, the two bound Mn2+ ions (Figure 5A),97 located among
binding profile for Mn2+ in exo III is distinct, with four acidic residues (Asp443, Glu478, Asp498, and
clear evidence for (at least) two classes of metal Asp549) in a Mn2+-doped crystal.97 These acidic
binding site, one showing exothermic binding and the residues are four of the seven conserved residues
other showing endothermic binding. The X-ray struc- found in all bacterial and retroviral RNase H do-
ture of exo III obtained by Tainer and co-workers mains, including the E. coli enzyme (Figure 5A).
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1079

Figure 15. The substrate binding loop (/) of E. coli RNase

H substrate binding is missing in the p51 subunit of HIV-
RT, and is replaced by the p66 subunit (not shown) in the
case of HIV-RT. Adapted from ref 101. Figure 16. The active site of bacteriophage T4 RNase H
showing the two Mg2+-binding sites. The intermetal dis-
tance is 7 Å.
Either the HIV-RT RNase H domain does indeed bind
two divalent metal ions, or this result again reflects
metal ion model proposed for certain nucleases. The
the specific use of Mn2+ and/or the high concentra-
tions of Mn2+ used (∼45 mM) in the doping technique. Ba2+ ions were ligated by Asp19, Asp132, Asp155,
The discussion that follows for reverse transcriptase and Asp132, Asp155, Asp157, respectively. With
(section VIII.C) and retroviral integrase (section X) Mn2+, only one bound ion was observed.
provides further insight on this point. On the basis The metal cofactor regulates many facets of enzyme
of crystallographic evidence from the HIV-RT RNase chemistry. Crouch and co-workers have examined
H domain, it was proposed that the enzyme most the kinetics of hybrid binding to E. coli RNase H
likely acts in a manner analogous to the exonuclease using surface plasmon resonance.103 The presence
domain of DNA polymerase I (Figure 7B).100 By of Mg2+ appears to decrease kon by almost 30-fold, and
analogy with the two metal ion model proposed for koff by 5-fold, with an effective increase in binding
the Klenow fragment, a similar metal cofactor ar- affinity of 6-fold.
rangement was also proposed in preliminary struc-
tural studies of the E. coli ribonuclease H.96 How-
ever, more detailed crystallographic work, with E. General Nucleases
magnesium-bound enzyme,73,91 has suggested one
metal binding site in an analogous fashion to the There exist many families of endo- and exonuclease
magnesium binding pocket of the CheY chemotaxal enzymes that digest target DNA molecules in a
protein (Figure 3).95 A one metal ion model is also relatively nonspecific fashion. Many of these require
strongly indicated by solution kinetic, NMR, and Mg2+ and/or Ca2+ cofactors. Pingoud and co-workers
calorimetric studies,34,44 for both Mg2+ and Mn2+ have identified a cluster of conserved residues in a
binding. family of six nonspecific nucleases related to Serratia
Although the E. coli RNase H active-site domain nuclease, which has been crystallographically char-
is structurally similar to that of the HIV RNase H acterized (Figure 11).36 Five of these are very
domain, one distinct feature is the presence of a basic important for catalytic activity (Arg 57, Arg 87, His
loop in the former, that is absent in the HIV RNase 89, Asn 119, and Glu 127). Some influence kcat and
H domain where it is replaced by a piece of the others Km, and metal ion and pH dependence studies
polymerase domain from the other subunit (Figure led to a tentative assignment of function. As de-
15). Marqusee and co-workers have demonstrated scribed earlier in section VI.C, the Arg residue most
that the basic loop is not essential for catalytic
likely contributes to enhance transition-state stabi-
activity.101 A significantly reduced activity with a
lization.
strict Mn2+ requirement was found, while the Mn2+
affinity also decreased by 2 orders of magnitude. There are many nucleases, especially localized in
Structural studies have provided valuable insight the nucleus of eukaryotic organisms, which require
on metal cofactor chemistry for this family of en- both Mg2+ and Ca2+ for optimal activity. Typically,
zymes. The structure of bacteriophage T4 RNase H Ca2+ serves a structural role and Mg2+ is catalytic.
has been determined and shows two bound Mg2+ ions Both Mg2+ and Mg2+/Ca2+ endonucleases have been
(Figure 16).102 One is surrounded by six H2O and is isolated from rat liver nuclei,104 while several nuclear
therefore held in an outer-sphere mode, while the nucleases have been isolated from rat thymocytes.105
other is ligated by five H2O and the carboxylate of These nuclear Ca2+/Mg2+-dependent endonucleases
Asp132. The Ba2+ derivative also showed two bound have been implicated in the extensive internucleo-
ions at distinct sites from the Mg2+ sites, with somal DNA fragmentation arising during programmed
intermetal distances of 5.0 and 7.0 Å, respectively. cell death (apoptosis) and are potential targets for
In both cases, this is greater than the putative two drug design.106-110 Physiological concentrations of
1080 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

Zn2+ known to inhibit apoptosis also inhibit the clease or ribonuclease functions that are required for
activity of these Ca2+/Mg2+ nucleases, which are their overall operation. A large number of poly-
found extensively in lymphocyte tissue. Other than merases have now been structurally characterized
inhibition by zinc, the activity of these nucleases may and can be divided into three major categories;
be regulated at the transcriptional level. For ex- namely RNA and DNA polymerases, and reverse
ample, an 18 kD Ca2+/Mg2+-dependent nuclease transcriptases. Divalent magnesium is an essential
(putatively a cyclophilin) identified in lymphoid cells cofactor for these three classes of polymerase enzyme,
undergoing apoptosis, is regulated by the glucocor- and serves a variety of roles.
ticoid receptor.111,112 RNA polymerases are commonly used in transcrip-
For many nucleases the metal cofactors appear to tion, where the information encoded in a genomic
mediate not only direct catalytic activity, but also DNA (or in some retroviruses, a genomic RNA)
structural changes that regulate substrate recogni- molecule is transcribed into a single strand of mes-
tion and binding specificity. In the absence of Mg, senger RNA. DNA polymerases replicate double-
the periplasmic nuclease from Streptomyces antibi- strand DNA, and usually possess an exonuclease
oticus does not form tight specific complexes with activity as a proof-reading function. Finally, reverse-
cognate (dG) sequences (n g 4).113 Changes in site transcriptases are commonly associated with retro-
specificity of single strand specific endonucleases, viruses. This enzyme catalyzes three reactions: first
such as mung bean nuclease, have been observed the conversion of single-strand genomic RNA to a
with a change in Mg2+ concentration, and also by double-strand RNA-DNA hybrid (DNA polymerase
varying temperature and ionic strength.114 Confor- function); second, the digestion of the RNA strand of
mational changes of supercoiled DNA were proposed. the hybrid (ribonuclease H function); and third, a
As previously noted for nuclear nucleases, many second DNA polymerase reaction that converts the
fungal multisubunit nucleases require two types of single-strand DNA to double-strand DNA prior to
metal for optimal activity. Studies with metal- integration into the host cell’s genomic DNA.
selective chelating agents suggest a Ca2+ requirement
to stabilize quaternary structure and a Mg2+ require- A. RNA Polymerases
ment to promote activity.115 The fungal nuclease In addition to its well-known catalytic role, Mg2+
from Aspergillus may be related to Serratia nuclease, induces structural changes in the E. coli RNA poly-
and again Ca2+ has been found to increase the merase open complex with the λPR promoter that
thermostability of this enzyme.36,116 The acid-soluble have been examined by chemical and enzymatic
nuclease γ has been purified from U. maydis cell footprinting.119-121 Sigman and co-workers122 have
extracts.117 Either Mg2+ or Ca2+ promotes nicking demonstrated a Mg2+ requirement for E. coli RNA
of one strand of the substrate DNA duplex, while polymerase to form single-stranded DNA structures
Mn2+, Co2+, or Zn2+ promotes double-strand cleavage. that resemble open transcription complexes. The
Many nucleases serve as general hydrolytic agents influence of Mg2+ on formation of the transcription
to break down nucleic acids. One of the best-studied bubble upon binding of E. coli RNA polymerase to
examples is DNase, although this class of enzyme the T7A1 promoter has also been examined by use
uses Ca2+ as the principal catalytic site, with Mg2+ of chemical probes, including hydroxyl radical foot-
and other Ca2+ ions serving as structural cofactors. printing.19,120 These data suggest that formation of
Early work on pancreatic DNase A demonstrated the transcription active complex requires two steps:
distinct Mg2+ and Ca2+ sites, with the number of sites one Mg2+-dependent, and the other Mg2+-indepen-
varying with solution pH.118 Aaqvist and Warshel dent. The data also suggests that one or two Mg2+
have examined the effect of metal ion substitution ions activate polymerase activity, while other ions are
on the activity of staphylococcal nuclease by free responsible for enlarging the transcription bubble.
energy perturbation calculations.58 Their results Magnesium binding motifs have been identified in
suggest a protein architecture that is designed to primase, a ssDNA-dependent RNA polymerase (Table
optimize the activity of Ca2+. More electrophilic 2 and Figure 4).123 Also, the Mg2+ binding site of E.
cations with large hydration free energies increase coli RNA polymerase has been investigated by use
the activation barrier for hydrolysis as a result of of hydroxyl radical footprinting,19 and mutagenesis
overstabilization of the putative HO- nucleophile. studies have implicated a -NADFDGD- motif in Mg2+
Optimization of metal cofactors is related to the dual binding (Table 2 and Figure 4).
requirement of stabilizing both the entering nucleo- The structure of bacteriophage T7 RNA polymerase
phile, and also the transition state for the reaction. has been reported at 3.3 Å resolution.124 Both
One obvious problem in over-generalizing these is- Asp537 and Asp812 have been proposed as metal-
sues stems from the discrete coordination chemistries binding residues; however, the Asp537Asn and
and pKas for metal-bound H2O that must be consid- Asp812Asn mutants demonstrate only slightly lower
ered when comparing results from transition metals Kds for Mn2+ relative to native enzyme, although the
versus alkaline earths. activity was greatly reduced.125 Scatchard analysis
of EPR Mn2+ binding data has suggested two bound
ions;125 however, such an approach is not a reliable
VIII. Polymerases method for the estimation of metal ion stoichiometry.
Polymerases are enzymes that catalyze the replica- It is proposed that the Asp537 and Asp812 residues
tion and synthesis of strands of DNA or RNA from a bridge the metal cofactors and that subtle changes
single-strand or double-strand template polynucle- in the geometry of these residues might have a strong
otide. Some of these enzymes contain other exonu- influence on catalytic activity.
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1081

Table 5. Thermodynamic Binding Parameters from cofactor rather than Mn2+ or any other transition
Calorimetric Studies of Mg2+ Binding to Klenow62 metal ion; however, studies with Mn2+ allow a direct
Mg2+ Mn2+ comparison with crystallographic data. In the pres-
parameters Mg2+ (+ TMP2-) Mn2+ (+ TMP2-) ence of TMP2-, one Mg2+ binds at each of the active
Polymerase sites (one at the 3′-5′ exonuclease site, and one at the
no. of bound cations 1 1 1 1 polymerase site) and the binding constants are
K1 (mM) 0.16 0.08 0.02 0.03
slightly augmented relative to binding in the absence
3′-5′ Exonuclease of TMP2-. A similar result is obtained with Mn2+ and
no. of bound cations 1 1 1 1
K2 (mM) 0.40 0.20 0.60 0.07 is supported for both cofactors by a detailed kinetic
analysis.62

Mutation of the aspartic acid residues of the -GDD- Warshel and co-workers have carried out a theo-
acid sequence motif of poliovirus RNA-dependent retical analysis of the role of metal ions in phosphate
RNA polymerase results in enzymes with different ester hydrolysis catalyzed by the Klenow fragment
metal ion requirements for optimal activity.21 The of E. coli DNA polymerase I64 and argued against a
repeated carboxylate motif is a common feature of metal-bound hydroxide as the active nucleophile, but
polymerase domains (Table 2 and Figure 4). Both favored an external hydroxide that is stabilized
aspartate residues are required for Mg2+ binding; electrostatically by the metal cofactor. Stabilization
however, Asn mutations are found to be active with of this hydroxide could be accounted for in terms of
transition metal replacements. Mutation of either the electrostatic influence of the high-affinity metal
Asp to Asn knocks out activity with Mg2+ as cofactor, ion alone (Figure 7A).
however, the -GDN- mutation is active with Mn2+.21
The conserved amino acids in region I (-YGDTDG-)
Inasmuch as an amide linkage may serve as a ligand,
of human DNA polymerase R are most likely involved
especially to Mn2+, it is possible that the activity of
this mutant reflects retention of the metal cofactor. in formation of a metal binding domain (Table 2 and
Figure 4). Both Asp1002Asn and Thr1003Ser mu-
B. DNA Polymerases tants were found to utilize Mn2+ more effectively than
Mg2+;22,23 however, these mutants show enhanced
The Klenow fragment is a proteolysis product of misinsertion fidelity as a result of a 850- and 62-fold
DNA polymerase I, an important enzyme in DNA higher KD. Mutation of Asp1004 resulted in a loss
replication. This fragment contains both the 3′-5′ of activity for either metal cofactor, presumably
exonuclease and polymerase domains126 and has arising from loss of metal binding. The results are
become a paradigm for a class of metalloenzymes that consistent with metal binding to each Asp residue;
follow a two metal ion mechanism of phosphate ester more critically for Asp1004. In the case of Mn2+
hydrolysis.48,61 This is based extensively on the activation, a significant increase in misincorporation
crystallographic work of Steitz and co-workers.48,61
of nucleotide bases was observed, but not for Mg2+,
Early studies of Klenow demonstrated two metal ions
even for the active Asp/Asn and Thr/Ser mutants
(illustrated in Figure 7B) bound in the exonuclease
active site in the presence of the substrate mimic described earlier. The effect for Mn2+ was manifest
thymidine monophosphate (TMP2-).61 In the absence by the influence of this cation on KMs, and the results
of TMP2- only one metal ion (A in Figure 7B) was suggest that subtle changes in Mn2+ coordination
observed to bind. Multiple bound ions have also been chemistry resulting from these mutations gives rise
proposed on the basis of solution studies of Klenow to structural changes in the active-site pocket that
by kinetic, optical, and EPR methods with Co2+ and diminish its ability to recognize specific nucleotide
Mn2+ cofactors.127,128 After these studies, other en- bases. The consequences of these structural pertur-
zymes have been crystallographically characterized bations for increased mutagenicity were discussed in
with evidence of two active-site metal ions.97 A section IV.
common feature of each of these studies is the use of The DNA polymerase φ29 shares many of the
Mn2+, and occasionally Zn2+ and Co2+, as analogues characteristics of R-like DNA polymerases, including
for the presumed natural Mg2+ cofactor, since the two conserved regions of amino acid similarity (D-
heavier manganous ion facilitates discrimination of
NSLYP and K-NS(L/V)YG) that have been proposed
the metal ion from solvent water.49 Subsequent
to form part of the active site.129,130 Tyr to Phe
characterization of bound Mg2+ is often carried out
by difference methods, by subtracting electron den- mutations of the Tyr residues in these two domains
sity maps of an enzyme from those obtained with a were found to influence the polymerization activity
manganese or heavy metal complex of an enzyme;48 of Mg-NTP substrates, but not Mn-NTP substrates.
however, the difficulties associated with the analysis In the first domain, an Asp to Glu mutation greatly
of metal cofactor stoichiometry from crystallography influenced the polymerase activity, but not the exo-
have been alluded to earlier. nuclease function. Polymerase activity could be
The binding stoichiometries to Klenow for both partially recovered by use of Mn-NTPs. These results
Mg2+ and Mn2+ have been examined by titration are consistent with involvement of these two amino
calorimetry (Table 5), both in the presence and acid regions in the polymerase, rather than the
absence of the substrate analogue thymidine mono- exonuclease domain. Also, the recovery of partial
phosphate, TMP2-.62 Considering their physiological activity for the Asp to Glu mutation with Mn2+ is
availablity, divalent magnesium is the physiological consistent with direct interaction with the NTP-
1082 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

Figure 17. R-Carbon trace structures of rat DNA polymerase β, HIV-RT, E. coli DNA pol I, and T7 RNA polymerase,
showing the positioning of common aspartate residues. Adapted from ref 132.

Figure 18. Stereoview of the polymerase active site of rat DNA polymerase β, and a schematic model of the reaction
chemistry. Adapted from ref 131.

bound metal cofactor: a more favorable interaction reactions, one divalent cofactor is carried into the
arising in the case of Mn2+, relative to Mg2+. active site as a tight complex with dNTPs (shown
In addition to DNA polymerases R, the polymerases with ddCTP in Figure 18).131 There is an additional
β have also been the subject of intense structural magnesium binding site that is most likely stabilized
investigation. Ternary complexes of rat DNA poly- by the high charge density resulting from the mul-
merase β complexed with a DNA template primer tiple carboxylates and the triphosphate. The Mg2+
and dideoxycytidine (ddCTP) have been determined bound to the nucleotide triphosphate is well-known
at 2.9 and 3.6 Å and yield significant mechanistic to facilitate nucleophilic attack,133-136 although the
insight.131,132 Three conserved aspartates are located 3′-OH of the attacking sugar need not necessarily be
at the active polymerase site of all DNA polymerases directly coordinated to the magnesium center as
thus far characterized, while RNA polymerases show shown schematically in Figure 18, but may be
two of these (Figure 17). For nucleotidyl transferase stabilized electrostatically as an outer-sphere com-
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1083

plex.16,64 Also, the use of the metal-bound carboxylate activity and the ability to catalyze specific removal
as a catalytic base is not appealing and has been of the tRNA replication primer. These observations
discussed in the literature.137,138 Previously it has point to obvious structural and/or coordination changes
been shown that a DNA polymerase can be converted for Mg2+, relative to Mn2+ activation, and may even
to an RNA polymerase by changing the metal cofactor be indicative of distinct coordination sites.
from Mg2+ to Mn2+.139 Figure 18 shows that all three Conservative mutation Q151N in the polymerase
phosphates on ddCTP bind to one or other of the two domain of HIV-1 RT demonstrated similar activity to
metal cofactors, and so it is likely that the metal ions that of WT,143 although the Q151A mutant demon-
help to position the nucleotide. It is argued that a strated reduced activity. The preference for divalent
slight change in orientation of ddCTP would be cation changed from Mg2+ to Mn2+. The influence of
sufficient to reduce steric hindrance at the C2′ ribose each of the p66 and p51 subunits on substrate and
of ddCTP and allow RNA to become as good a inhibitor binding, and metal ion preference, has been
substrate as DNA.131,139 This could be brought about tested by subunit-selective mutagenesis studies on
by subtle changes in the coordination chemistry of Glu 89. Evaluation of Michaelis and inhibition
Mg2+ relative to Mn2+, as suggested earlier in section constants, and metal preference was evaluated for
IV on the origins of mutagenicity. E89G mutants of HIV-1 RT (for both the p66 and p51
In a later study of rat DNA pol β, using Mn2+ as domains).144 Only those mutants containing a p66
metal cofactor, two metal centers were observed even WT domain demonstrated a preference for Mg2+ (i.e.,
in the absence of a nucleotide triphosphate.140 These p51 mutants had no effect), while substrate and
two cations were held to the enzyme by a single inhibitor binding were also unchanged. However,
bridging aspartate: a mode of binding that seems absolute activity was influenced by mutation of the
unreasonable for Mg2+. It is likely that these sites p66 Glu89 residue. Clearly, the p66 domain contrib-
reflect specific binding preferences for Mn2+ under utes the most important side-chain functionality to
the high metal concentrations used. Other mecha- metal-mediated chemistry.
nistic aspects of polymerase activity, relating to The inhibitory influence of the antiviral drugs AZT
orientation of the primer template, have been de- (azidothymidylate) and N-ethylmaleimide (NEM) on
bated; however, this issue is not relevant to the focus the RNase H activity of HIV-1 RT is distinct when
of this review.137,138 using either Mg2+ or Mn2+ as cofactors.145 AZT
monophosphate is a competitive inhibitor in the
C. Reverse Transcriptase presence of Mg2+, but is noncompetitive with a Mn2+
Blain and Goff have examined the RNase H activ- cofactor. This observation is also consistent with
ity of native and mutant Moloney murine leukemia distinct metal-binding sites for Mg2+ and Mn2+ in the
virus reverse transcriptase.141 Two of these mutants, RNase H active site. Furthermore, the mode of
designated ∆5 and ∆C, contain deletions in the substrate cleavage is distinct, with endonucleolytic
RNase H domain. These were found to be active with hydrolysis of a poly(rA)/poly(dT) substrate in the
in vitro assays using Mn2+, but were found to be presence of Mg2+ and both endo- and exonucleolytic
inactive in vivo, where the physiological cofactor is hydrolysis in the presence of Mn2+.145,146 With poly-
Mg2+. These results are consistent with perturbation (rG)/poly(dC) both exo- and endonucleolytic cleavage
of the metal cofactor binding site in the RNase H are observed for Mg2+ and Mn2+. The authors sug-
domain, but apparently the deletion mutations do not gest that the RNase H domain of HIV-1 RT under-
completely remove the Mn2+ binding propensity. goes conformational changes upon binding hybrid
Three point mutants (R657S, Y598V, and S526A) substrates and that these changes depend on the
were found to show full activity with Mn2+, and divalent cofactor.
partial activity in the presence of Mg2+. Again, All RT enzymes possess a common stretch of seven
mutation of these residues are likely to perturb the amino acids termed the -YXDD- box. Amino acid
metal-binding site (Table 2 and Figure 4). Two other substitution of X results in changes in divalent metal
point mutants (Y586F and D524N) showed no activ- ion preference and enzyme activity, and so the
ity for Mg2+ activation, and only minimal levels (5%) -YXDD- box appears to play a role at the active center
for Mn2+ in the Y586F mutant. In these cases, the of RTs by binding the metal cofactor.147
metal-binding capability has clearly been more seri- The metal cofactor carries out functions in addition
ously damaged, suggestive of direct coordination by to catalysis. Mg2+ stabilizes HIV-1 RT binding to
Asp524 and Tyr586. Variations in DNA polymerase DNA-DNA substrates, with a 20-60-fold decrease
and processivity were also observed with the use of in koff, and an increase in productive interaction with
Mn2+ in these RNase H-deficient mutants, consistent DNA.148-150 Other charged species may enhance or
with the mutagenic effects of this cation described diminish the influence of magnesium. For example,
earlier in section IV. Ca2+ competitively inhibits Mg2+-promoted func-
The stimulatory effect of Mg2+ compared to Mn2+ tions.151 Spermine enhances, but cannot replace
has been examined for HIV-RT.142 Wild-type enzyme Mg2+ in reactions catalyzed by avian myeloblastosis
shows normal functional properties for both cations, virus reverse transcriptase,152 in a manner consistent
although RNase H* activity (hydrolysis of double- with the function of polyamines as described in
strand RNA) was observed with Mn2+. An E478Q section V.
mutant in the p66 domain, which participates in Finally, a reverse transcriptase motif has been
metal binding, was found to be inactive with Mg2+, identified in the catalytic subunit of telomerase, the
although Mn2+ restored both its endoribonuclease ribonucleoprotein required for replication of chromo-
1084 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

some ends, isolated from Euplotes aediculatus and contacts to Asp26 and Asp24 on each monomer. At
yeast (S. cerevisiae).166 Many metal binding residues higher metal concentrations the phosphatase is ob-
are conserved, and mutations in these motifs results served to bind up to 4.5 Mg2+ per monomer, however,
in loss of activity and telomere shortening. Interest- it is not clear that any site other than the “tight” site
ingly the -YXDD- motif described earlier for RT is physiologically relevant. The structure of Saccha-
enzymes, appears as -LXDD- in telomerase. romyces cerevisiae yeast inorganic pyrophosphatase
has also been obtained at 2.4 Å resolution (Figure
IX. Phosphatases 19) using Mn2+ as an analogue for Mg2+.154 This
Crystallographic studies of E. coli inorganic pyro- structure showed four bound Mn2+ ions and two
phosphatase, a hexameric soluble enzyme, provide bound phosphates that putatively are derived from
evidence for 1.5 Mg2+ cofactors per monomer (Figure pyrophosphate. The same group had previously
19).153 One ion resides in a “tight” binding site and structurally characterized the E. coli enzyme and
is coordinated by three aspartates (Asp65, Asp70, and identified one active-site Mn2+ ion,164 as found earlier
Asp102). The other Mg2+ is shared by two mono- by Kankare et al.153 It remains uncertain whether
mers, and appears to stabilize the interfacial contact these two enzymes do in fact follow distinct mecha-
region, and is coordinated by outer-sphere H-bond nistic pathways, or at least differ in their use of metal

Figure 19. (Top) The catalytic (left) and interfacial (right) Mg2+ sites of inorganic pyrophosphatase. Adapted from ref
153. (Bottom) Schematic illustration of the proposed transition state for the pyrophosphatase activity of the Saccharomyces
cerevisiae enzyme. The four Mn2+ ions and two phosphates, identified crystallographically, are indicated. Adapted from
ref 154.
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1085

cofactors, or if the additional metal cofactors for the phering the chemistry of this essential metal cofactor.
yeast enzyme arise from the use of Mn2+ ion in the This fact contributes to the design of magnesium
structural study. In passing, it can be noted that a binding pockets in proteins and enzymes. The im-
structural role is played by the Mg2+ ion cofactor for portance of reading the relevance of metal binding
the dinuclear zinc E. coli enzyme, alkaline phos- chemistry within the context of natural abundance
phatase.155 has been repeatedly emphasized.

X. Integrases XII. Acknowledgments

Retroviral integrase (IN) enzymes catalyze inser- This work was supported by grants from the
tion of virally encoded DNA into the genomic DNA Petroleum Research Fund (administered by the
of a host156,157 and belong to a structurally related American Chemical Society), and the National Sci-
superfamily that includes RNase H, RuvC resolvase, ence Foundation (CHE-9706904). J.A.C. is a Camille
and MuA transposase.49,158 Detailed structural in- Dreyfus Teacher-Scholar (1994-1999) and a Na-
formation is not available on the latter two, but tional Science Foundation Young Investigator (1992-
recent reports describe the structures of HIV and 1997).
avian sarcoma virus (ASV) integrase enzymes.49,97
Only for ASV-IN has the position of the divalent XIII. References
metal cofactor (Mg2+ or Mn2+) been defined. A single
(1) Biological Chemistry of Magnesium; Cowan J. A., Ed.; VCH:
metal ion was resolved and the placement of aspar- New York, 1995.
tate side chains and metal ions was found to be (2) Wilcox, D. E. Chem. Rev. 1996, 96, 2435-2458.
similar to E. coli RNase H. Figure 5A summarizes (3) Mizuguchi, H.; Cook, P. F.; Hasemann, C. A.; Uyeda, K.
Biochemistry 1997, 36, 8775-8784.
the metal binding sites for the two RNase H struc- (4) Kasho, V. N.; Stengelin, M.; Smirnova, I. N.; Faller, L. D.
tures (E. coli and HIV-RT) and ASV-IN. There Biochemistry 1997, 36, 8045-8052.
(5) Richardson, C. C. Annu. Rev. Biochem. 1969, 38, 795-840.
appears to be a common metal binding site, and only (6) Tsai, M. D.; Yan, H. Biochemistry 1991, 30, 6806-6818.
the additional Mn2+ site illustrated in the HIV-RT (7) Yan, H.; Tsai, M. D. Biochemistry 1991, 30, 5539-5546.
RNase H domain in Figure 5A is unusual. A unique (8) Black, C. B.; Huang, H.-W.; Cowan, J. A. Coordn. Chem. Rev.
1994, 135/136, 165-202.
metal binding site has also been tentatively identified (9) Cowan, J. A. Comments Inorg. Chem. 1992, 13, 293-312.
in crystallographic studies of the RuvC enzyme.159 On (10) Cowan, J. A. J. Biol. Inorg. Chem. 1997, 2, 168-176.
the basis of epitope-mapping studies with antibodies, (11) Falke, J. J.; Snyder, E. E.; Thatcher, K. C.; Voertler, C. S.
Biochemistry 1991, 30, 8690-8697.
binding of Mg2+ is suggested to regulate the confor- (12) Snyder, E. E.; Buoscio, B. W.; Falke, J. J. Biochemistry 1990,
mation of the monomeric subunits of HIV-IN.160 The 29, 3937-3943.
accessibility of epitopes (surface peptide fragments (13) Cowan, J. A. Inorganic Biochemistry. An Introduction, 2nd ed.;
Wiley-VCH: New York, 1997; pp 1-16.
recognized by antibodies) to the C-terminal domain (14) Frausto da Silva, J. J. R.; Williams, R. J. P. The Biological
and catalytic core is reduced in the presence of the Chemistry of the Elements. The Inorganic Chemistry of Life;
Clarendon Press: Oxford, 1991; pp 244-245.
magnesium cofactor, supportive of a conformational (15) Cowan, J. A. Inorg. Chim. Acta 1998, in press.
change. Such structural changes may contribute to (16) Black, C. B.; Foster, M.; Cowan, J. A. J. Biol. Inorg. Chem. 1996,
the aggregation mechanism of the monomers to form 1, 500-506.
(17) Cowan, J. A. Inorganic Biochemistry. An Introduction, 2nd ed.;
an active multimeric complex. Wiley-VCH: New York, 1997; pp 291-303.
Clearly, a unique structural and catalytic metal ion (18) Calcium in Biology; Wiley-Interscience: New York, 1983.
is the common theme that emerges from structural (19) Zaychikov, E.; Martin, E.; Denissova, L.; Kozlov, M.; Markovtsov,
V.; Kashlev, M.; Heumann, H.; Nikiforov, V.; Goldfarb, A.;
studies of this superfamily of enzymes. This provides Mustaev, A. Science 1996, 273, 107-109.
strong support for the argument presented earlier (20) Joyce. C. M.; Steitz., T. A. Annu. Rev. Biochem. 1994, 63, 777.
(21) Jablonski, S. A.; Morrow, C. D. J. Virol. 1995, 69, 1532-1539.
that the second Mn2+ site identified crystallographi- (22) Copeland, W. C.; Wang, T. S. F. J. Biol. Chem. 1993, 268, 11028-
cally for the HIV-RT RNase H is likely to be physi- 11040.
ologically irrelevant, is populated as a consequence (23) Copeland, W. C.; Lam, N. K.; Wang, T. S. F. J. Biol. Chem. 1993,
268, 11041-11049.
of the experimental conditions and that only the (24) Gerhold, D. L.; Pettinger, A. J.; Hadwiger, L. A. Physiol. Mol.
common metal binding site is of functional relevance. Plant Pathol. 1993, 43, 33-46.
(25) Pelletier, H.; Sawaya, M. R.; Wolfle, W., Samuel H.; Kraut, J.
Biochemistry 1996, 35, 12762-12777.
XI. Closing Remarks (26) Pelletier, H.; Sawaya, M. R. Biochemistry 1996, 35, 12778-
12787.
Magnesium is a truly versatile metal cofactor. It (27) Ricchetti, M.; Buc, H. EMBO J. 1993, 12, 387-396.
(28) Focher, F.; Verri, A.; Maga, G.; Spadari, S.; Huebscher, U. FEBS
mediates a large range of enzymatic reactions, by Lett. 1990, 259, 349-352.
both outer- and inner-sphere pathways and demon- (29) Tabor, S.; Richardson, C. C. Proc. Natl. Acad. Sci. U.S.A. 1989,
strates remarkable variability in coordination chem- 86, 4076-4080.
(30) Cao, W.; Mayer, A. N.; Barany, F. Biochemistry 1995, 34, 2276-
istry. Much of the biochemistry of magnesium ion 2283.
is dictated by a few key properties (namely, a high (31) New England Biolabs, Inc. Catlog 1996/97, 241.
(32) Auer, T.; Landre, P. A.; Myers, T. W. Biochemistry 1995, 34,
charge density and extensive hydration state) that 4994-5002.
discriminate it from transition metal analogues. (33) Eder, P. S.; Walder, J. A. J. Biol. Chem. 1991, 266, 6472-6479.
Accordingly, it is not surprising that coordination (34) Black, C. B.; Cowan, J. A. Inorg. Chem. 1994, 33, 5805-5808.
(35) Black, C. B.; Cowan, J. A. Eur. J. Biochem. 1997, 243, 684-
changes for the latter should give rise to mutagenic 689.
effects, loss of binding and recognition specificity, and (36) Friedhoff, P.; Kolmes, B.; Gimadutdinow, O.; Wende, W.; Krause,
typically an increase in activity. For nucleic acids, K. L.; Pingoud, A. Nucleic Acids Res. 1996, 24, 2632-2639.
(37) Jou, R.; Cowan, J. A. J. Am. Chem. Soc. 1991, 113, 6685-6686.
outer-sphere interactions through magnesium-bound (38) Jeltsch, A.; Alves, J.; Wolfes, H.; Maass, G.; Pingoud, A. Proc.
water molecules are extremely important for deci- Natl. Acad. Sci. U.S.A. 1993, 90, 8499-8503.
1086 Chemical Reviews, 1998, Vol. 98, No. 3 Cowan

(39) Vipond, I. B.; Baldwin, G. S.; Halford, S. E. Biochemistry 1995, (85) Cheng, X.; Balendiran, K.; Schildkraut, I.; Anderson, J. E. EMBO
34, 697-704. J. 1994, 13, 3927-3935.
(40) Hale, S. P.; Poole, L. B.; Gerlt, J. A. Biochemistry 1993, 32, (86) Aggarwal, A. K. Curr. Opin. Struct. Biol. 1995, 5, 11-19.
7479-7487. (87) Gromova, E. S.; Vinogradova, M. N.; Uporova, T. M.; Gryaznova,
(41) Jeltsch, A.; Alves, J.; Maass, G.; Pingoud, A. FEBS Lett. 1992, O. I.; Isagulyants, M. G.; Kosykh, V. G.; Nikol’skaya, I. I.;
304, 4-8. Shabarova, Z. A. Bioorg. Khim. 1987, 13, 269-72.
(42) Baldwin, G. S.; Vipond, I. B.; Halford, S. E. Biochemistry 1995, (88) Weiss, B. In The Enzymes, 3rd ed.; Academic Press: New York,
34, 705-714. 1981; Vol. 14, pp 203-231.
(43) Demple, B.; Johnson, A.; Fung, D. Proc. Natl. Acad. Sci. U.S.A. (89) Mol, C. D.; Kuo, C.-F.; Thayer, M. M.; Cunningham, R. P.;
1986, 83, 7731-7735. Tainer, J. A. Nature 1995, 374, 381-386.
(44) Huang, H.-W.; Cowan, J. A. Eur. J. Biochem. 1994, 219, 253- (90) Kow, Y. W. Biochemistry 1989, 28, 3280-3287.
260. (91) Katayanagi, K.; Miyagawa, M.; Matsushima, M.; Isikawa, M.;
(45) Shimamura, S.; Hibasami, H.; Kano, U.; Watanabe, S.; Suzuki, Kanaya, S.; Ikehara, M.; Matsuzaki, T.; Morikawa, K. Nature
S.; Nakashima, K. Int. J. Biochem. 1990, 22, 545-549. 1990, 347, 306-309.
(46) Bachrach, U.; Abu-Elheiga, L. Eur. J. Biochem. 1990, 191, 633- (92) Kashiwagi, T.; Jeanteur, D.; Haruki, M.; Katayanagi, K.;
637. Kanaya, S.; Morikawa, K. Protein Eng. 1996, 9, 857-867.
(47) Casareno, R. L. B.; Cowan, J. A. J. Chem. Soc., Chem. Commun. (93) Katayanagi, K.; Miyagawa, M.; Matsushima, M.; Ishikawa, M.;
1996, 1813-1814. Kanaya, S.; Nakamura, H.; Ikehara, M.; Matsuzaki, T.; Morika-
(48) Beese, L. S.; Steitz, T. A. EMBO J. 1991, 10, 25-33. wa, K. J. Mol. Biol. 1992, 223, 1029-1052.
(49) Bujacz, G.; Jaskólski, M.; Alexandratos, J.; Wlodawer, A.; (94) Katayanagi, K.; Ishikawa, M.; Okumura, M.; Ariyoshi, M.;
Merkel, G.; Katz, R. A.; Skalka, A. M. Structure 1996, 4, 89- Kanaya, S.; Kawano, Y.; Suzuki, M.; Tanaka, I.; Morikawa, K.
96. J. Biol. Chem. 1993, 268, 22092-22099.
(50) Kostrewa, D.; Winkler, F. K. Biochemistry 1995, 34, 683-696. (95) Needham, J. V.; Chen, T. Y.; Falke, J. J. Biochemistry 1993, 32,
(51) Bruice, T. C.; Tsubouchi, A.; Dempcy, R. O.; Olson, L. P. J. Am. 3363-3367.
Chem. Soc. 1996, 118, 9867-9875. (96) Yang, W.; Hendrickson, W. A.; Crouch, R. J.; Satow, Y. Science
(52) Deal, K. A.; Hengge, A. C.; Burstyn, J. N. J. Am. Chem. Soc. 1990, 249, 1398-1406.
1996, 118, 1713-1718. (97) Davies, J. F.; Hostomska, Z.; Hostomsky, Z.; Jordan, S. R.;
(53) Jenkins, L. A.; Bashkin, J. K.; Autrey, M. E. J. Am. Chem. Soc. Matthews, D. A. Science 1991, 252, 88-95.
1996, 118, 6882-6825. (98) Oda, Y.; Yamazaki, T.; Nagayama, K.; Kanaya, S.; Kuroda, Y.;
(54) Kanyo, Z. F.; Scolnick, L. R.; Ash, D. E.; Christianson, D. W. Nakamura, H. Biochemistry 1994, 33, 5275-5284.
Nature 1996, 383, 554. (99) Oda, Y.; Nakamura, H.; Kanaya, S.; Ikehara, M. J. Biomol. NMR
(55) Jenkins, J.; Janin, J.; Rey, F.; Chiadmi, M.; van Tilbeurgh, H.; 1991, 1, 247-255.
Lasters, I.; De Maeyer, M.; Van Belle, D.; Wodak, S. J.; (100) Beese, L. S.; Friedman, J. M.; Steitz, T. A. Biochemistry 1993,
Lauwereys, M.; Stanssens, P.; Mrabet, N. T.; Snauwaert, J.; 32, 14095-10141.
Matthyssens, G.; Lambeir, A.-M. Biochemistry 1992, 31, 5449- (101) Keck, J. L.; Marqusee, S. J. Biol. Chem. 1996, 271, 19883-19887.
5458. (102) Mueser, T. C.; Nossal, N. G.; Hyde, C. C. Cell 1996, 85, 1101-
(56) Almassey, R. J.; Janson, C. A.; Hamlin, R.; Xuong, X. H.; 1112.
Eisenberg, D. Nature 1986, 323, 304. (103) Haruki, M.; Noguchi, E.; Kanaya, S.; Crouch, R. J. J. Biol. Chem.
(57) Vipond, I. B.; Halford, S. E. Biochemistry 1995, 34, 1113-1119. 1997, 272, 22015-22022.
(58) Aaqvist, J.; Warshel, A. J. Am. Chem. Soc. 1990, 112, 2860- (104) Hibino, Y.; Iwakami, N.; Sugano, N. Biochim. Biophys. Acta
2868. 1991, 1088, 305-307.
(59) Pyle, A. M. Science 1993, 261, 7089. (105) Nikonova, L. V.; Beletsky, I. P.; Umansky, S. R. Eur. J. Biochem.
(60) Steitz, T. A.; Steitz, J. A. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 1993, 215, 893-901.
6498-6502. (106) Giannakis, C.; Forbes, I. J.; Zalewski, P. D. Biochem. Biophys.
Res. Commun. 1991, 181, 915-20.
(61) Ollis, D. L.; Brick, P.; Hamlin, R.; Xuong, N. G.; Steitz, T. A.
(107) Chang, M. P.; Baldwin, R. L.; Bruce, C.; Wisnieski, B. J. Science
Nature (London) 1985, 313, 762-766.
(Washington, D.C.) 1989, 246, 1165-1168.
(62) Black, C. B.; Cowan, J. A. Manuscript submitted.
(108) Beletskii, I. P.; Matyasova, J.; Nikonova, L. V.; Skalka, M.;
(63) Ceska, T. A.; Sayers, J. R.; Stier, G.; Suck, D. Nature 1996, 382,
Umanskii, S. R. Gen. Physiol. Biophys. 1989, 8, 381-398.
90-93.
(109) Nikonova, L. V.; Zotova, R. N.; Umanskii, S. R. Biokhimiya
(64) Fothergill, M.; Goodman, M. F.; Petruska, J.; Warshel, A. J. Am. (Moscow) 1989, 54, 17709-17718.
Chem. Soc. 1995, 117, 11619-11627. (110) Khodarev, N. N.; Sokolova, I. A.; Aleksandrova, S. S.; Volgina,
(65) Christianson, D. W.; Lipscomb, W. N. Acc. Chem. Res. 1989, 22, V. V.; Votrin, I. I. Byul. Eksp. Biol. Med. 1987, 103, 312-314.
62-69. (111) Caron-Leslie, L. A. M.; Schwartzman, R. A.; Gaido, M. L.;
(66) Black, C. B.; Cowan, J. A. J. Am. Chem. Soc. 1994, 116, 1174- Compton, M. M.; Cidlowski, J. A. J. Steroid Biochem. Mol. Biol.
1178. 1991, 40, 661-671.
(67) Shriver, D. F.; Atkins, P. W.; Langford, C. H.; Inorganic (112) Montague, J. W.; Hughes, F. M.; Cidlowski, J. A. J. Biol. Chem.
Chemistry; 2nd ed.; Freeman: New York, 1994. 1997, 272, 6677-6684.
(68) Rawlings, J.; Speckhard, D. C.; Cleland, W. W. Biochemistry (113) Cal, S.; Nicieza, R. G.; Connolly, B. A.; Sanchez, J. Biochemistry
1993, 32, 11204-11210. 1996, 35, 10828-10836.
(69) Cornelius, R. D.; Cleland, W. W. Biochemistry 1978, 17, 3279. (114) Kowalski, D. Nucleic Acids Res. 1984, 12, 7071-7086.
(70) Kim, S.; Cowan, J. A. Inorg. Chem. 1992, 31 3495-3496. (115) Ito, K.; Hara, C.; Matsuura, Y.; Minamiura, N. Arch. Biochem.
(71) Hampel, A.; Cowan, J. A. Chem. Biol. 1997, 4, 513-517. Biophys. 1995, 317, 25-32.
(72) Casareno, R.; Li, D.; Cowan, J. A. J. Am. Chem. Soc. 1995, 117, (116) Ito, K.; Matsuura, Y.; Minimiura, N. Arch. Biochem. Biophys.
11011-11012. 1994, 309, 160-167.
(73) Katayanagi, K.; Okumura, M.; Morikawa, K. Proteins: Struct., (117) Yarnall, M.; Rowe, T. C.; Holloman, W. K. J. Biol. Chem. 1984,
Funct. Genet. 1993, 17, 337-346. 259, 3026-3032.
(74) Heydenreich, A.; Koellner, G.; Choe, H.-W.; Cordes, F.; Kisker, (118) Price, P. A. J. Biol. Chem. 1972, 274, 2895-2899.
C.; Schindelin, H.; Adamiak, R.; Hahn, U.; Saenger, W. Eur. J. (119) Craig, M. L.; Suh, W.-C.; Record, M. T., Jr. Biochemistry 1995,
Biochem. 1993, 218, 1005-1012. 34, 15624-15632.
(75) Volbeda, A.; Lahm, A.; Sakiyama, F.; Suck, D. EMBO J. 1991, (120) Suh, W. C.; Leirmo, S.; Record, M. T., Jr. Biochemistry 1992,
10, 1607-1618. 31, 7815-1725.
(76) Pingoud, A.; Jeltsch, A. Eur. J. Biochem. 1997, 246, 1-22. (121) Suh, W. C.; Ross, W.; Record, M. T., Jr. Science (Washington,
(77) Jeltsch, A.; Maschke, H.; Selent, U.; Wenz, C.; Kohler, E.; D.C.) 1993, 259, 358-361.
Connolly, B. A.; Thorogood, H.; Pingoud, A. Biochemistry 1995, (122) Kuwabara, M. D.; Sigman, D. S. Biochemistry 1987, 26, 7234-
34, 6239-46. 7238.
(78) Thielking, V.; Selent, U.; Kohler, E.; Landgraf, A.; Wolfes, H.; (123) Griep, M. A. Indian J. Biochem. Biophys. 1995, 32, 171-178.
Alves, J.; Pingoud, A. Biochemistry 1992, 31, 3727-3732. (124) Sousa, R.; Chung, Y. J.; Rose, J. P.; Wang, B.-C. Nature 1993,
(79) Piccirilli, J. A.; Vyle, J. S.; Caruthers, M. H.; Cech, T. R. Nature 364, 593-599.
1993, 361, 85. (125) Woody, A.-Y. M.; Eaton, S. S.; Osumi-Davis, P. A.; Woody, R.
(80) Moon, B. J.; Vipond, I. B.; Halford, S. E. J. Biochem. Mol. Biol. W. Biochemistry 1996, 35, 144-152.
1996, 29, 17-21. (126) Joyce, C. M.; Grindley, N. D. F. Proc. Natl. Acad. Sci. U.S.A.
(81) Vipond, I. B.; Moon, B.-J.; Halford, S. E. Biochemistry 1996, 35, 1983, 80, 1830-1834.
1712-1721. (127) Mullen, G. P.; Serpersu, E. H.; Ferrin, L. J.; Loeb, L. A.; Mildvan,
(82) Rosenberg, J. M. Curr. Opin. Struct. Biol. 1991, 1, 104-113. A. S. J. Biol. Chem. 1990, 265, 14327-14334.
(83) Alves, J.; Urbanke, C.; Fliess, A.; Maass, G.; Pingoud, A. (128) Han, H.; Rifkind, J. M.; Mildvan, A. S. Biochemistry 1991, 30,
Biochemistry 1989, 28, 7879-7888. 11104-11108.
(84) Newman, M.; Strzelecka, T.; Dorner, L. F.; Schildkraut, I.; (129) Blasco, M. A.; Lazaro, J. M.; Bernad, A.; Blanco, L.; Salas, M. J.
Aggarwal, A. K. Science 1995, 269, 656-663. Biol. Chem. 1992, 267, 19427-19434.
Enzymes in Nucleic Acid Biochemistry Chemical Reviews, 1998, Vol. 98, No. 3 1087

(130) Blasco, M. A.; Lazaro, J. M.; Blanco, L.; Salas, M. J. Biol. Chem. (151) Tan, C.-K.; Zhang, J.; Li, Z.-Y.; Tarpley, W. G.; Downey, K. M.;
1993, 268, 24106-24113. So, A. G. Biochemistry 1991, 30, 2651-2655.
(131) Pelletier, H.; Sawaya, M. R.; Kumar, A.; Wilson, S. H.; Kraut, (152) Aoyama, H. Biochem. Int. 1989, 19, 67-76.
J. Science 1994, 264, 1891-1903. (153) Kankare, J.; Salminen, T.; Lahti, R.; Cooperman, B. S.; Baykov,
(132) Sawaya, M. R.; Pelletier, H.; Kumar, A.; Wilson, S. H.; Kraut, A. A.; Goldman, A. Biochemistry 1996, 35, 4670-4677.
J. Science 1994, 264, 1930-1935. (154) Harutyunyan, E. H.; Kuranova, I. P.; Vainshtein, B. K.; Hohne,
(133) Lowenstein, J. M.; Schatz, M. N. J. Biol. Chem. 1961, 236, 305- W. E.; Lamzin, V. S.; Dauter, Z.; Teplyakov, A. V.; Wilson, K. S.
307. Eur. J. Biochem. 1996, 239, 220-228.
(134) Lowenstein, J. M. Biochem. J. 1958, 70, 222-230. (155) Janeway, C. M. L.; Xu, X.; Murphy, J. E.; Chaidaroglou, A.;
(135) Miller, D. L.; Ukena, T. J. Am. Chem. Soc. 1969, 91, 3050-3053. Kantrowitz, E. R. Biochemistry 1993, 32, 1601-1609.
(136) Cowan, J. A. Inorg. Chem. 1991, 30, 2740-2747. (156) Katz, R. A.; Skalka, A. M. Annu. Rev. Biochem. 1994, 133-173.
(137) Pelletier, H. Science 1994, 266, 2025-2026. (157) Katz, R. A.; Skalka, A. M. Annu. Rev. Genet. 1992, 26, 527-
(138) Steitz, T. A.; Smerdon, S. J.; Jager, J.; Joyce, C. M. Science 1994, 544.
266, 2022-2025. (158) Rice, P.; Mizuuchi, K. Cell 1995, 82, 209-220.
(139) Van de Sande, J. H.; Loewen, P. C.; Khorana, H. G. J. Biol. (159) Ariyoshi, M.; Vassylyev, D. G.; Iwasaki, H.; Nakamura, H.;
Chem. 1972, 247, 6140. Shinagawa, H.; Morikawa, K. Cell 1994, 78, 1063-1072.
(140) Davies, J. F.; Almassay, R. J.; Hostomska, Z.; Ferre, R. A.; (160) Asante-Appiah, E.; Skalka, A. M. J. Biol. Chem. 1997, 272,
Hostomsky, Z. Cell 1994, 76, 1123-1133. 16196-16205.
(141) Blain, S. W.; Goff, S. P. J. Biol. Chem. 1996, 271, 1448-1454. (161) Quigley, G. J.; Teeter, M. M.; Rich, A. Proc. Natl. Acad. Sci.
(142) Cirino, N. M.; Cameron, C. E.; Smith, J. S.; Rausch, J. W.; Roth, U.S.A. 1978, 75, 64.
M. J.; Benkovic, S. J.; Le Grice, S. F. J. Biochemistry 1995, 34, (162) Gessner, R. V.; Quigley, G. J.; Wang, A. H.-J.; van der Marel,
9936-43. G. A.; van Boom, J. H.; Rich, A. Biochemistry 1985, 24, 237-
(143) Sarafianos, S. G.; Pandey, V. N.; Kaushik, N.; Modak, M. J. 240.
Biochemistry 1995, 34, 7207-7216. (163) Weber, D. J.; Libson, A. M.; Gittis, A. G.; Lebowitz, M. S.;
(144) Kew, Y.; Song, Q.; Prasad, V. R. J. Biol. Chem. 1994, 269, Mildvan, A. S. Biochemistry 1994, 33, 8017-8028.
15331-15336. (164) Harutyunyan, E. H.; Oganessyan, V. Y.; Oganessyan, N. N.;
(145) Zhan, X.; Tan, C.-K.; Scott, W. A.; Mian, A. M.; Downey, K. M.; Terzyan, S. S.; Popov, A. N.; Rubinskiy, S. V.; Vainshtein, B.
So, A. G. Biochemistry 1994, 33, 1366-1372. K.; Nazarova, T. I.; Kurilova, S. I.; Vorobyova, N. N.; Avaeva, S.
(146) Tan, C.-K.; Cival, R.; Mian, A. M.; So, A. G.; Downey, K. M. M. Crystallographia Rep. 1996, 41, 84-96 [English translation
Biochemistry 1991, 30, 4831-4834. from Krystallografiya (in Russian)].
(147) Kanda, T.; Saigo, K. Biochim. Biophys. Acta 1993, 1163, 223- (165) Burgess, J. Metal Ions in Solution; Halsted Press: New York,
226. 1993.
(148) De Stefano, J. J.; Bambara, R. A.; Fay, P. J. Biochemistry 1993, (166) Lingner, J.; Hughes, T. R.; Shevchenko, A.; Mann, M.; Lundblad,
32, 6908-6915. V.; Cech, T. R. Science 1997, 276, 561-567.
(149) Kruhoffer, M.; Urbanke, C.; Grosse, F. Nucleic Acids Res. 1993, (167) Cotton, F. A.; Hazen, E. E.; Legg, M. J. Proc. Natl. Acad. Sci.
21, 3943-3949. U.S.A. 1979, 76, 2551-2555.
(150) Hsieh, J. C.; Zinnen, S.; Modrich, P. J. Biol. Chem. 1993, 268,
24607-24613. CR960436Q