American Journal of Anal yt ical Chemistry, 2011, 2, 639-649
doi:10.4236/ajac.2011.26073 Published Online October 2011 (http://www.SciRP.org/journal/ajac)
Targeting Divalent Metal Ions at the Active Site of the
HIV-1 RNase H Domain: NMR Studies on the Interactions
of Divalent Metal Ions with RNase H and Its Inhibitors
Jiangli Yan1*, Haihong Wu1, Tiffany Tom1, Oleg Brodsky1, Karen Maegley2
1Oncology Chemistry Pfizer Global Research and Development, La Jolla Laboratories,
Science Center Drive, San Diego, USA
2Oncology Research Unit, Pfizer Global Research and Development, La Jolla Laboratories,
Science Center Drive, San Diego, USA
Received June 27, 2011; revised July 3, 2011; accepted August 22, 2011
Abstract
HIV-1 reverse transcriptase (RT) RNase H (HIV-RH) is a key target of anti-AIDS drugs. Metal-chelating
compounds are an important class of chemicals in pharmacological drug discovery, especially in relation to
HIV-RT and the highly-related HIV-integrase. The correlation between the metal-chelating properties and
enzyme activities of the metal chelators is always of high scientific interest, as an understanding of this may
accelerate the rational optimization of this class of inhibitors. Our NMR data show that Mg2+ and Ca2+ bind
specifically to the active site of the RNase H domain and two Mg2+ ions sequentially bind one molecule of
RNase H. We also demonstrate here, using saturated and unsaturated tricyclic N-hydroxypyridones designed
to block the active site, that the primary binding sites and affinities of divalent metal ions are correlated with
the structures of the chelating motifs. Chemical shift perturbation studies of protein/metal-ion/compound
ternary complexes also indicate that divalent metal ions play important roles for the specific interaction of
the compounds with the RNase H active site.
Keywords: Metal Chelation, HIV-1, Reverse Transcriptase, RNase H, NMR Spectroscopy
1. Introduction
HIV-1 reverse transcriptase [1] converts single-stranded
retroviral RNA into double-stranded DNA, which is in-
tegrated into the cellular genome [2-4]. HIV-1 RT is a
multifunctional enzyme that has RNA-directed DNA
polymerase, DNA-directed DNA polymerase and ribo-
nuclease H (RNase H) activities [1]. HIV-1 RT is a het-
erodimer composed of two peptide subunits, p66 and p51.
The polymerase active site is located at the N-terminus
of the p66 subunit, whereas its C-terminal end contains
the RNase H active domain [5]. The RNase H domain of
HIV-1 RT (HIV-RH) plays a role in many steps of re-
verse transcription, such as the generation of an RNA
primer for synthesis of the (+)-strand DNA, the degrada-
tion of the viral genomic RNA in the intermediate
RNA·DNA hybrid, and the removal of host tRNA and
plus-strand primers [2,6,7]. HIV-RH forms a central
five-stranded β-sheet surrounded by four α-helices. The
core domain of the RNase H active site contains a
highly-conserved DEDD motif that consists of four
acidic residues, D443, E478, D498, and D549 [5,7]. The
hydrolysis of the scissile phosphodiester bonds catalyzed
by RNase H requires divalent metal ions, preferably
Mg2+ [6,8]. However, there has been some controversy
regarding the number of metal ions (one or two) involved
in the catalysis. Crystallographic studies of both E. coli
RNase H and HIV-RH have shown that the active site
can bind two Mn2+ ions separated by approximately 4 Å
and the authors proposed a two-metal-ion catalytic
mechanism [9,10]. The ability of HIV-RH to bind to two
metal ions, either two Mn2+ or two Mg2+, has also been
confirmed by calorimetry and NMR, respectively [9,11].
However, this result was not supported by the crystal
structure of the E. coli enzyme, in which only a single
bound Mg2+ ion was identified (even though the crystal
was obtained in high concentration of MgSO4) [12]. Re-
cently, two Mg2+ ions were also observed ~4 Å apart in

J. L. YAN ET AL. 640
Bh-RNase HC-substrate complexes by Nowotny et al.
[6,8]. Later analysis by the same group also strongly
supported the two-metal-ion catalytic mechanism [6,13].
The RNase H activity of HIV-RH plays a crucial role
in the retroviral life cycle [14]. Defective mutations of
two key residues (E478Q and H539F) in the RNase H
domain induce a marked reduction in viral proliferation
[15], thus making HIV-RH an attractive chemothera-
peutical target for anti-HIV drugs [16,17]. Designing
metal-chelating compounds that are able to bind two
divalent metal ions at the active site is one strategy that
enables the direct blocking of the active site. Small
N-hydroxyimide analogs that are optimized to bind two
divalent metal ions at a 4 Ǻ distance between the ions
inhibit HIV-RT activity in vitro with an IC50 < 1.0 uM
[10]. Metal-chelating compounds are an important class
of chemicals in pharmacological drug design, especially
when targeted against proteins with metal ions at their
active sites [4]. Metal-binding property of the metal-
chelators affects the potency of enzyme inhibitors, and
the correlation between metal-chelation affinities and
enzyme inhibition is always of interest in the design of
metal-chelating inhibitors.
In this report, we applied NMR analytical methodol-
ogy to study the interactions of the divalent metal ions
Mg2+ and Ca2+ with HIV-RH and chemically-engineered
metal-chelating compounds. When the pH and ionic
strength are carefully maintained, we observed clean
chemical shift perturbations at the active site of RNase H
with both Mg2+ and Ca2+. We also used 1D 1H-NMR to
characterize the interactions between the divalent metal
ions and HIV-RH inhibitors. Two chemically similar
series, saturated and unsaturated tricyclic N-hydroxypy-
ridones have different chelating affinities depending on
the structure of the primary metal-binding sites. To un-
derstand the correlation between the metal-chelating
properties and the inhibition of the enzyme by metal-
chelating inhibitors, we also studied the HIV-RH/Mg2+/
inhibitor ternary complexes using a saturated tricyclic
N-hydroxypyridone and -thujaplicinol as the tool com-
pounds.
2. Material and Methods
2.1. Protein Preparation
A cDNA fragment encoding the C-terminal domain of
HIV-1 p66 protein (W426 to L560) was cloned into
pET-28a in fusion with a C-terminal His6-tag. BL21
(DE3)-AI cells transformed with the recombinant plas-
mid were grown at 37˚C in 2 L Celtone-N or Celtone-CN
media (Cambridge Isotope Laboratories, Andover, MA)
to produce 15N- or 13C/15N-enriched proteins, respec-
tively. After three hours, the cultures were allowed to
equilibrate to 15˚C - 18˚C. Expression of HIV-RH was
then induced using 100 uM IPTG and 0.01% arabinose at
0.7 - 1.0 OD600 cell density. The cells were grown at
15˚C - 18˚C for 16 hours and harvested by centrifugation.
The frozen cells were resuspended and lysed in 50 mM
Tris-HCl buffer (pH 8.0) containing 250 mM NaCl, 100
ul/L benzonase (Novagen), 0.2 mg/ml lysozyme (Sigma,
L-6876), and 0.25 mM TCEP (Pierce). The His6-tagged
protein was purified from the cell lysate using immobi-
lized metal affinity chromatography (ProBond resin, In-
vitrogen). The eluate fraction (step elution, OmniPrep
gravity column, BioRad) was further purified using
size-exclusion chromatography (100 mL Phenomenex
S3000 column) with a mobile phase containing 25 mM 1,
3-bis(tris(hydroxymethyl)methylamino)propane (bis-Tris
propane, pH 6.5), 150 mM NaCl, 2% glycerol, and 0.25
mM TCEP. The desired fractions were pooled and con-
centrated to 15 - 20 mg/ml using centrifugal concentra-
tors (Millipore).
2.2. Backbone Resonance Assignment of
HIV-RH
All of the experiments for the resonances assignment
were recorded at 30˚C on a Bruker Avance 700 MHz
spectrometer equipped with a TCI cryo-probe. The NMR
sample contained 0.9 mM 13C,15N-labeled HIV-RH in 25
mM bis-Tris-d14 (pH 6.5), 150 mM NaCl, 1 mM dithio-
threitol-d6 (DTT) and 5% 2H2O. The backbone reso-
nances were assigned using standard triple resonance
experiments, including CBCA(CO)NH, HNCACB, HNCA,
[18] and HN(CO)CA[19]. Sequential assignments were
obtained using TopSpin 2.0 (Bruker Biospin) for spectral
processing and SPARKY for sequential analysis [20].
2.3. Metal Titration with HIV-RH and Its
Inhibitors
In the HIV-RH titration experiments, 0.5 M MgCl2 or
Ca(NO3)2 was titrated into the protein solution. HIV-RH
protein (70 uM) was prepared in buffer containing 25
mM bis-Tris-d14 (pH 6.5), 150 mM salt (NaCl for Mg2+
titration and NaNO3 for Ca2+ titration, respectively), 1
mM DTT, and 5% 2H2O. The pH of all reagents used in
the titrations was carefully monitored and readjusted to
6.5 (if required) at each step of the titration. 1H-15N
HSQC spectra were acquired on an Inova 600 MHz
spectrometer (Varian Inc.) at 25˚C to monitor the chemi-
cal shift changes as the metal ion concentrations increased.
In the titrations with the inhibitors, each sample was
prepared independently, using identical buffer conditions
(except for the concentration of the metal ions). The final

J. L. YAN ET AL. 641
observed, most likely as a result of regional flexibility
and/or signal overlapping. Most of those amino acids, i.e.,
N474-K476, E514-L517, A538-K540, and V548-L560,
were located in loop regions and at the C-terminus. We
observed and assigned three amino acids (D443, E478,
and D498) of the DDED motif involved in metal binding,
except D549 which is located in the flexible C-terminus
of the protein. This is consistent with previous reports
[22] that the C-terminus of isolated HIV-RH is highly
dynamic and can adopt either an α-helical or random coil
conformations in crystal structures, depending on the
crystallization conditions and the space group [23]. In
solution, the C-terminus is usually disordered, but can be
stabilized at high concentration of Mg2+ (80 mM)
[11,24].
concentration of the compound was 0.20 to 0.50 mM. 1D
1H-NMR experiments were performed using an Inova
600 MHz spectrometer (Varian Inc.) at 25˚C. All spectra
were recorded with 128 transients, a 16-ppm sweepwidth
using presaturation for water suppression.
The NMR data were analyzed using ACD NMR Proc-
essor (ACD-Labs, Inc) and the total chemical shift
change Δ

obs of the 1H-15N cross peak was calculated
according to the formula,
 
22
0.17
obs NH
   (1)
where
N

 and
H

 were the chemical shift changes
in 15N and 1H dimensions, respectively. The co-crystal
structure of HIV-RH domain protein [21] was used for
chemical shift mapping. We observed that the conformation, stability and ac-
tivity of HIV-RH are very sensitive to pH, ionic strength
and divalent metal-ion concentration. A drop in the pH
of the buffer from 7.0 to 5.0 induced dramatic and global
shifts in the 1H-15N HSQC spectrum, and more HSQC
signals were detectable at lower pH. Interestingly, most
early structural studies of HIV-RH by NMR [25-28] and
x-ray crystallography [29-31] were carried out in acidic
conditions, suggesting that the isolated HIV- RH domain
is more stable at lower pH.
2.4 Calculation of the Dissociation Constants
(Kd)
In the single binding mode, one metal ion binds with one
molecule of protein or metal-chelating inhibitor. When
the reaction occurs under the fast exchange conditions,
the observed chemical shift (

obs) is the weighted average
of the chemical shifts of the free and bound species.
Therefore, the observed chemical shift change (Δ

obs) is
a function of the dissociation constant Kd [2] (Equation (2)), 3.2. Determination of the Binding Constants (Kd)
of Divalent Metal Ions with HIV-RH
where 
f and

b are the fractions and

f and

b are the
chemical shifts of the free and bound protein or compound,
respectively. M0 and L0 are the initial concentrations of
the metal ion and the chelator (protein or ligand). The
data were fit using GraphPad Prism (GraphPad Software)
to determine the Kd from the chemical shift changes.
Divalent metal ions are crucial for RNase H activity and
contribute to the conformation and stability of the protein,
especially at the C-terminus [11,28]. In our NMR bind-
ing study of the divalent metal ions to HIV-RH, we
chose 25 mM bis-Tris (pH 6.5) containing 150 mM NaCl
as our buffer to maintain pH and ionic strength, therefore
to mitigate the effects on conformation, stability and ac-
tivity of HIV-RH. The buffer was also compatible with
those used in the biochemical assays. Every reagent used
in the experiment was prepared in the buffer described,
and the pH was checked in each step and adjusted when
necessary.
3. Results and Discussion
3.1. NMR Assignments of the 1H-15N HSQC
Spectra and Flexibility at the C-Terminus of
HIV-RH
An isolated, C-terminal His6-tagged protein that contains
the C-terminal domain of HIV-RT p66 protein (W426-
L560) was expressed and purified in its 15N-labeled or
13C/15N-labeled forms for our NMR studies. The protein
was mostly well-folded as the 1H-15N HSQC peaks were
well resolved and dispersed. Backbone resonance as-
signments of most amino acids were obtained from the
standard triple resonance experiments (data available if
required). The 1H- 15N HSQC peaks of 26 amino acids,
not including those of the His6-tag and prolines, were not
In the metal titration experiments, we dialyzed 15N-
labeled HIV-RH in the described buffer. MgCl2 (1.0 M,
prepared in the same buffer) was titrated into the protein
solution to increase the Mg2+ concentration from 0 to 80
mM. Approximately 24 peaks showed obvious shifts on
the 1H-15N HSQC spectra (Figure 1(a)). The Mg2+ dose-
dependent chemical shift changes (∆
 ), calculated as
weighted sums of the changes in both the 15N

2
0000 00
4
obs obs fb fdd0
2
M
KLMKLML L 


 



(2)

J. L. YAN ET AL. 642

(a)
010 20 30 40 50 60 70 80 9
0
0
20
40
60
80
100
120
140
160
180
G444
A4 4 5
T477
E4 7 8
Q500
M
g
++ Conce ntr ation
(
mM
)
Chemical Shift Change (Hz)

(b)
D549
D443
E478
D498
(c)
Figure 1. (a) Overlay of the HSQC spectra shows the
chemical shift perturbations of HIV-RH by Mg2+ at 0 mM
(purple), 2.5 mM (pink), 5 mM (cyan), and 10 mM (blue).
(b) The Mg2+ titration curve of the chemical shift changes
vs. the Mg2+ concentration (0, 2.5, 5, 10, 20, 40, and 80 mM)
for five selected amino acids. (c) Mapping of the shift per-
turbations of HIV-RH by Mg2+ at a concentration of 5.0
mM on the HIV-RH structure. Red: ∆ 50 Hz, pink: ∆
 =
35 - 50 Hz, light pink: ∆  ≤ 35 Hz, Yellow: not observed or
assigned.
and 1H dimensions (Equation (1)), were fitted by Equa-
tion (2) using GraphPad Prism (GraphPad Software) as
shown in Figure 1(b). It was clear that the titration
curves reached a maximum plateau at about 80 mM of
Mg2+. The binding constant (Kd) of Mg2+/HIV-RH were
determined by the titration data of 17 amino acids that
showed ∆
 > 10 Hz with at least 6 data points. The cal-
culated Kd ranged from 8.4 to 16.6 mM with a goodness
of fit (R2) from 0.984 to 0.999. The averaged Kd was 13
 4 mM, very similar to the value reported previously
[11].
We also carried out titration experiments with step-
wise increases in Ca2+ concentrations from 0 to 15 mM
in a similar fashion to those using Mg2+. We obtained a
very similar pattern of chemical shift perturbations. In
the plots of ∆
 vs. the Ca2+ concentration, the titration
curves reached plateaus at 15 mM Ca2+. The Kd values of
21 amino acids were calculated to be in the range from
1.5 to 3.3 mM, yielding reasonable values of R2 from
0.992 to 1.000. The average Kd for Ca2+ to HIV-RH is
2.9  0.5 mM, indicating that the binding of Ca2+ to
HIV-RH is four times tighter than that of Mg2+.
Figure 1(c) showed the chemical shift perturbation
map of Mg2+ on the HIV-RH domain protein. Amino
acids at the catalytic center showed the greatest chemical
shift perturbations. All three observable residues in the
metal chelating DDED motif, i.e., D443, E478, and
D498 stood out in the chemical shift perturbations in-
duced by Mg2+. Ca2+ binds to the same site as Mg2+ and
amino acids located at the active site also showed the
largest chemical shift changes in response to Ca2+. Our
NMR titration data demonstrated small and localized
chemical shift perturbation patterns in HIV-RH in re-
sponse to the divalent metal ions, Mg2+ and Ca2+, when
the buffer pH and ionic strength were carefully main-
tained. It is clear that Mg2+ and Ca2+ bind specifically to
the active site of the RNase H domain, but do not sig-
nificantly change the global conformation and dynamics
of the protein. However, the C-terminus of our construct
is still mostly disordered even at high concentrations of
divalent metal ions at the testing buffer conditions. Most
of the C-terminal residues of HIV-RH were not stable
enough to produce measurable signals in the HSQC
spectra in the presence of 80 mM Mg2+ or 15 mM Ca2+.
3.3. Sequential Binding of Two Mg2+ Ions to the
Active Site
In the Mg2+ titration experiments, a few peaks, including
G444, Q500, W535ω, and V536, became broader and
broader as the Mg2+ concentration increased. More inter-
estingly, they split into two sets of peaks at 80 mM Mg2+
(Figure 2(a)). The splitting of these signals indicated the

J. L. YAN ET AL.643
existence of two slowly-exchanging protein conforma-
tions, and the occupancy of the second metal binding site
at high Mg2+ concentrations (80 mM or higher). At 80
mM Mg2+, the second metal-binding site was approxi-
mately half-occupied by Mg2+, whereas the first
metal-binding site was almost fully occupied. Interest-
ingly, the affinity of the second Mg2+ ion for HIV-RH
was reported to be ~35 mM [11]. Figure 2(b) shows four
residues with split HSQC signals at 80.0 mM Mg2+ in the
HIV-RH structure. Three of these, Q500, W535ω, and
V536, are close to D498, while D444 is on the opposite
side of the active site, adjacent to D443. When the con-
centration of Mg2+ was 5.0 mM, large chemical shift
changes were observed for D443 (Figure 2(c)). Our data
suggest that two Mg2+ ions bind sequentially to the active
site of HIV-RH. The first Mg2+ binds to site 1 and inter-
acts with D443 and E478, whereas the second Mg2+
binds to site 2 and interacts with D498 and D549 with a
slightly lower (~3 times) affinity (Figure 2(c)). The
weaker metal binding affinity of site 2 is probably due to
the flexibility of the C-terminus, where D549 is located.
However, the peak splitting was not observed in the Ca2+
titration experiments. The distance between the two Mg2+
ions in the active site is 4 Å, as shown in the crystal
structure [8,10,11]. Ca2+ has a much larger atomic radius
than Mg2+; therefore, it is unlikely that two Ca2+ ions can
8.40 8.35 8.30 8.25 8.20
108.5
109.0
109.5
110.0
110.5
111.0
7.8 7.7 7.6 7.5
123.5
124.0
124.5
125.0
125.5
126.0
126.5
127.0
E478
1
H, ppm
15
N, ppm
1
H, ppm
15
N, ppm

(a)
D549
D443
E478
D498

D549
D443
E478
D498

(b) (c)
Figure 2. (a) Regions of the HSQC spectra of HIV-RH showing peak shifting with increasing Mg2+ concentrations, at 0, 2.5,
5.0, 10, 20, 40, and 80 mM, in the direction indicated by the arrows. The dashed circles show peak splitting for V536 and
G444 and broadening for E478 and G453 at 80 mM Mg2+. (b) Mapping of splitting HSQC signals on the HIV-RH structure.
Red: AA with splitting HSQC signals, Yellow: not observed or assigned. (c) Mapping of the shift perturbations of HIV-RH by
.0 mM Mg2+ on the HIV-RH structure as shown in Figure 2(c) in a different orientation. 5

J. L. YAN ET AL. 644
fit simultaneously into the active site. Although we saw
evidence of the sequential binding of two Mg2+ ions in
our experiments, we were not able to obtain enough data
points to determine the binding constant of the second
Mg2+ ion through curve fitting.
3.4. Interaction of Metal-Ions to Metal-Chelating
Compounds
Divalent metal ions, especially Mg2+, are directly in-
volved in the catalytic activities of HIV-RH, and two
Mg2+ ions bind to one molecule of HIV-RH at the active
site. To target the active site, it is a common strategy to
design metal-chelating compounds binding to two diva-
lent metal ions at the active site [2,16]. The effect of
compound/metal binding property on enzyme inhibition
is always of interest in the lead generation and optimiza-
tion of metal-chelating compounds. In this study, we
utilized 1D 1H-NMR experiments to identify the metal-
binding sites of our lead compounds through chemical
shift perturbations induced by metal ions. Again, similar
to the protein/metal interaction, protons that are near the
metal-binding site generally experience larger perturba-
tions. Based on Equation (2), we determined the binding
affinities of the lead compounds to metal ions through
chemical shift changes. The sample preparation, data
acquisition and data processing were all automated for a
rapid measurement of Kd.
Derivatives of unsaturated and saturated tricyclic
N-hydroxypyridones, shown in Figures 3(a) and (b) re-
spectively, were designed and synthesized in-house for
targeting HIV-RH. The unsaturated tricyclic N-hydroxy-
pyridone derivatives contain a flat and rigid three-ring
core and two hypothetical metal-binding motifs (Figure
3(a)). In the Mg2+ titration experiments, all protons on
the tricyclic N-hydroxypyridone core showed significant
and incremental chemical shift changes as the Mg2+ con-
centration increased; the example of compound U-1 is
shown in Figure 3(c). Meanwhile, only slight chemical
shift changes were observed with the other protons,
which were located in the R1 and R2 substitute groups.
H-13 gave the largest chemical shift changes (up to 130
Hz) when the Mg2+ concentration changed. Therefore,
H-13 is the proton nearest to the Mg2+ ion (Figure 3(a)).
Three more unsaturated tricyclic N-hydroxypyridone
compounds were tested and H-13 showed the largest
chemical shift change in all cases. The chemical shift
changes of the core protons at the compound/Mg2+ ratio
of 1:20 when the titration curves reached the plateaus
were listed in Table 1. The chemical perturbations of the
other three core protons H-5, H-8 and H-12 showed only
half of the change of H-13. This indicates that Mg2+
preferably binds unsaturated tricyclic N-hydroxy-pyri-
done derivatives at site I by interacting with the two
oxygen atoms of the O14∸N1∸C2∸O15 motif (Fig-
ure 3(a)). The Kd values of the two unsaturated com-
pounds (U-1 and U-4) were determined through chemical
shift changes, and equivalent with each other within the
measurement deviation range. The chelator-Mg2+ com-
plexes of the other two compounds (U-2 and U-3) had
poor solubility. As the Mg2+ concentration was approxi-
mately equivalent to the compound concentration, we
saw precipitates in the NMR tubes and the proton signals
became too weak to measure. Consequently, we were not
able to calculate the Kd due to the lack of points. Inter-
estingly, when [Mg2+] was higher than 20 times of the
compound concentration, the compound proton signals
became stronger and therefore measurable. The similar
phenomena were also observed with compound U-1 and
U-4. As shown in Figure 3(c), proton signals of U-1
became broader and broader and then sharper and sharper
as the Mg2+ concentration increased gradually from 0 to
80 mM. In addition, when [Mg2+] was over 40 times of
the compound concentration, the compound proton sig-
nals shifted to an opposite direction. We believe this was
related to the second metal-ion binding and the ternary
complexes (compound/(Mg2+)2) had better solubility than
the binary complexes (compound/Mg2+). The chemical
shift change “turn-over” occured all at a Mg2+ concentra-
tion of 16 - 20 mM, suggesting that with ~20 mM Mg2+,
the second metal-binding site is approximately half-oc-
cupied by Mg2+, whereas the first metal-binding site was
almost fully occupied and therefore the Kd of the second
Mg2+ to this series compound was estimated as ~10mM.
In our NMR study, we could not obtain sufficient data to
determine Kd of the second Mg2+ binding using the dou-
ble binding curve fit due to the maximum concentration
of MgCl2 in stock solution of 0.5 M.
The observations were different when the tricyclic
N-hydroxypyridone core was saturated at the car-
bon-carbon bond between C-12 and C-13 (Figure 3(b) ).
Saturation of the C-C bond makes the six-member-ring
more flexible and no longer flat. Figure 3(d) shows the
shift of the proton signals of 0.20 mM compound S-5 in
the Mg2+ titration measurements. Table 1 lists the
chemical shift perturbations of the core protons of four
saturated tricyclic N-hydroxypyridones at a concentration
ratio of compound: Mg2+ of 1:100 when the ∆
 vs. [Mg2+]
titration curves reached the plateaus. All four saturated
derivatives showed the same interaction mapping, which
obviously was different from the mapping of the unsatu-
rated analogs. Instead of H-13 in the unsaturated cases,
proton H-5 of saturated compounds presented the largest
chemical shift change, indicating Mg2+ preferably boun-
dat site II and interacts with the nitrogen and oxygen at-

J. L. YAN ET AL.645
oms of the N4∸C3∸C2∸O15 motif when the C12-C13 bond
was saturated. In the NMR spectra, only one set of triplet
peaks was observed for either of the two H-12 or two
H-13 (Figure 3(d)), indicating the equivalence of the
12
2
N
1
13
3
11
5N 4
N
7
6
10
9
8O
15
O
14
R1
R2
Mg2+
I
II
12
2
N
1
13
3
11
5N
4 N
7
6
10
9
8O 15
O
14
R1
R2
Mg2+
I
II
(a) (b)
7.30 7.25 7.20 7.15 7.10
Chemical Shift (ppm)
7.75 7.70
Chemical Shift (ppm)
8.825 8.775
Chemical Shift (ppm)
1
H, ppm
1
H , p pm
1
H , p pm
H*
H-13
H-12
H*
H-5
(c)
8.775 8.725
Chemical Shift (ppm)3.0503.000 2.950 Chemical Shift (ppm)3.450 3.400 Chemical Shift (p...
1
H , p p m
1
H , pp m
1
H , p p m
H*
H-5 H-13
H-12
(d)
Figure 3: (a) The core structure of unsaturated tricyclic N-hydroxypyridones. (b) The core structure of saturated tricyclic
N-hydroxypyridones. (c) 1D 1H-NMR spectra of 0.50 mM U-1 with Mg2+ at concentrations of 0.0, 0.05, 0.10, 0.25, 0.50, 1.00,
2.5, 5.0, 10.0, 35.0, and 50.0 mM. (d) 1D 1H-NMR spectra of 0.20 mM S-5 with Mg2+at concentrations of 0.0, 0.2, 0.4, 0.8, 1.6,
3.2, 5.0, 10.0, 20.0, 40.0, and 80.0 mM. R1 and R2 are substituent groups. Two potential metal binding motifs are labeled as I
O14∸N1∸C2∸O15 and II N4∸C3∸C2∸O15. Mg2+ prefers binding to saturated tricyclic N-hydroxypyridones at motif II, but to the
saturated tricyclic N-hydroxypyridones at I. The arrows indicate the shift of the proton signals as the Mg2+ concentration
increases. H* are protons on the R1 or R2 substituents.

J. L. YAN ET AL. 646
-1 to U-4) and saturated (S-5 to S-8) tricyclic N-hydroxypyridones by
Chem shift changes (  ), Hz
Table 1. Chemical shift perturbations of unsaturated (U
Mg2+ chelating.
Cpd Cpd: Mg2+ ratio
H5 H8 H12 H13
Mg2+ binding Kd, mM RNase H IC50, μM
U-1 1:20 38 36 34 130 0.2 0.46
U-2 1:20 46 64 42 152 N.A. N.A.
U-3 1:20 56 -- 43 137 N.A. 2.2
U-4 1:20 41 45 38 126 0.2 0.49
S-5 1:100 31 3 15 12 4.4 2.2
S-6 1:100 26 19 21 10 4.5 0.22
S-7 1:100 26 19 20 9 4.4 0.038
13
*
*
S-8 1:100 33 6 10 2.2 0.29
o protons at C-12 or C-13. The saturated N-hydro-
3.5. Ternary Complex of HIV-RH, Mg2+, and
The use of metal chelation as an anchor to the active site
-thujaplicinol to the first Mg2+ ion, as determined from
und s-7), which has a weak bind-
in
tw
xypyridone ring is relatively flexible and exchanges be-
tween chair and boat conformations. Four saturated N-
hydroxypyridones had very similar Kd values, in a tight
range of 2 - 5 mM (Table 1) which was 10 to 20 times
larger than those of the unsaturated analogs. The overall
chemical shift changes and the differences between the
largest and the other perturbations of the core protons of
the saturated compounds were both smaller than those
observed with the unsaturated ones. We also observed
the sharp-broad-sharp line-shape changes as the Mg2+
concentration increased, but didn’t see the second bind-
ing of Mg2+ as well, probably because the Mg2+ concen-
tration was not high enough.
Metal-Chelating Compounds
is a common strategy in designing inhibitors to target
HIV-RH and the highly- related HIV integrase enzymes
[2,16,32]. The final product of the design is a ternary
complex (compound/ metal-ion/protein). Therefore, the
correlation between compound/metal or protein/metal
binding affinity and enzyme inhibition is always of high
interest. Sometimes, metal ions are critical for inhibition
by the designed compound. -thujaplicinol, a selective
inhibitor of HIV RT, is a typical metal chelator [33]. Our
study demonstrated that it binds to two Mg2+ions sequen-
tially. Its proton signals incrementally shifted downfield
and became broader and broader as the concentration of
Mg2+ increased. When the concentration of Mg2+ reached
8 mM, the signals started to shift in the opposite direc-
tion. As the Mg2+ concentration increased further, the
signals shifted upfield and became narrower. The Kd of
the proton chemical shift changes, is 0.64  0.10 mM.
Interestingly, -thujaplicinol introduced no chemical
shift perturbations in the HSQC signals of the HIV-RH
domain when metal ions were absent. On the contrary, in
the presence of 8 mM Mg2+, significant chemical shift
perturbations were observed in the HSQC spectrum of
the HIV-RH protein attributable to 500 M -thujap-
licinol. The experiments indicated that the presence of a
metal ion is necessary for -thujaplicinol to interact with
the HIV-RH domain, in agreement with the crystallo-
graphic results [21].
On the other hand, 500 M saturated tricyclic N-hy-
droxypyridone (compo
g affinity (Kd 4.4 mM) to Mg2+, induced less than 25
Hz of chemical shift changes in the signals of 14 residues
of HIV-RH in the presence of 8 mM Mg2+. Three de-
tectable chelating residuals of the DEDD motif, i.e.,
D443, E478 and D498, were among them. The chemical
shift changes were 14, 10 and 13 Hz, respectively. We
also observed line-shape-broadening of the E478 signal.
In comparison with Figure 1(c), Figure 4 indicates that
the compound interacted with the active site of HIV-RH
through Mg2+ chelation in the buffer containing 8 mM
Mg2+. Interestingly, in the absence of divalent metal ions,
this compound was found to bind HIV-RH at an alterna-
tive binding site through our NMR and crystallographic
studies (unpublished data). Although compound s-7 in-
duced small chemical shift perturbations in the HIV-RH
domain, it showed strong inhibition against HIV-RH
with an IC50 of 0.038 M. One reason could be that the
experimental conditions used for the IC50 measurement
were very different from those used in the NMR experi-
ments. In our biochemical assay, we used substrate

J. L. YAN ET AL.647
Figure 4. Mapping of the shift perturbations on the HIV-
RH structure by compound s-7 in the presence of 8 mM
tides and full-length HIV-RT protein. Iso-
ted HIV-RH domains have been either inactive or
nd/metal chelating affinity and the
an
To understand the correlation between metal-chelating
me inhibition by metal-chelating in-
hibitors, we characterized the interactions of divalent
ompared. Compounds that contain
th
ly,
re
The authors thank Dr. Cathy Moore for her valuable
Mg2+. The protein and compound concentrations are 72 and
500 M, respectively. Dark red: Δ ≥ 15 Hz; light red: Δ =
10 - 15 Hz.
oligonucleo
la
much less active in RNA hydrolysis than intact HIV-RT
[22,23,28,30]. In addition, the interactions between RT
and oligonucleotide duplexes may affect the conforma-
tion of the protein, thereby altering the affinity of RNase
H for metal ions.
As shown in Table 1, we did not observe a correlation
between the compou
ti-HIV-1 activity of the inhibitors. The interactions of
Mg2+ with either HIV-RH or the tricyclic N-hydroxy-
pyridones are weak, with Kd values in the mM range.
However, the HIV-RH IC50 of the tricyclic N-hydroxy-
pyridone compounds range from 2.2 μM to 0.038 μM.
We believe that the metal-binding ability of this category
of compounds plays an important role in anchoring the
compound to the active site of HIV-RH specifically.
Since the binding affinity of metal/compound and metal/
protein are weak, the inhibitory potency of the compound
is not driven by metal chelation, but by direct interac-
tions between the compound and the protein instead.
Therefore, the goal of chemical design is to improve the
inhibitory potency rather than to increase the metal che-
lation affinity. In fact, the strategy of focusing on pro-
tein/compound interaction is practically more efficient.
Firstly, the protein/compound direct interaction is also
the key to improve the selectivity of designed com-
pounds. Secondly, very strong metal chelation property
(for instance, in the low nM range) is not preferable be-
cause it may cause toxicity by chelating metal ions in the
blood or in tissues.
4. Conclusions
properties and enzy
metal ions with HIV-RH and two series of HIV-RH in-
hibitors using NMR chemical shift perturbations. We
noticed that the chemical shift perturbation maps of the
HIV-RH protein and the compounds are sensitive to the
pH, ionic strength and concentration of divalent metal
ions. A neutral pH of 6.5 was chosen and 150 mM salt
was used to maintain the ionic strength in the metal ion
binding studies. When the buffer pH and ionic strength
were carefully maintained, the HSQC titration data
demonstrated that Mg2+ and Ca2+ ions introduce small
and localized chemical shift perturbation patterns in
HIV-RH. Both Mg2+ and Ca2+ bind specifically to the
active site of the HIV-RH domain, with binding con-
stants of 13 and 3 mM, respectively. We also observed
that two Mg2+ ions bind sequentially to the RNase H
domain, whereas only a single binding event was ob-
served for Ca2+ ions.
The metal-chelating affinities of the saturated or un-
saturated tricyclic N-hydroxypyridone inhibitors were
also measured and c
e same tricyclic core showed similar binding affinity to
metal ions at the same primary chelating site. Metal ions
interact primarily at site I with the unsaturated tricyclic
N-hydroxypyridones, but at site II with the saturated
analogs (Figures 3(a) and (b)). The flexibility of the
N-hydroxypyridone ring gained from the saturation re-
duced the metal-binding affinity from 0.2 to 4.0 mM.
With 8 mM Mg2+ present, our NMR mapping data
demonstrated that the tricyclic N-hydroxypyridone com-
pounds bind to the active site of HIV-RH. Interesting
gardless of the weak interaction of Mg2+ with either
HIV-RH target or the tricyclic N-hydroxypyridones (the
Kd’s are in the mM range), the HIV-RH IC50 of those
compounds were much higher, from 2.2 μM to 0.038 μM.
The poor correlation between the compound/metal che-
lating affinity and the anti-HIV-1 activities of the inhibi-
tors reveals the direct protein/compound interactions of
the series of inhibitor. The metal-binding property is
necessary for the metal chelating inhibitors to bind spe-
cifically to the active site through a metal ion bridge, but
the design should focus on the direct interactions of the
compound with the protein for inhibitory potency and
selectivity.
5. Acknowledgements
discussions and comments.

J. L. YAN ET AL. 648
[1] E. J. Artsa and S. F. Le Grice, “Interaction of Retrovir
criptase with Template-Primer Duplexes
Progress in Nucleic Acid Research
6. References
al
Reverse Trans
during Replication,”
and Molecular Biology, Vol. 58, 1997, pp. 339-393.
doi:10.1016/S0079-6603(08)60041-0
[2] V. Goldschmidt, J. Didierjean, B. Ehresmann, et al.,
“Mg2+ Dependency of HIV-1 Reverse Transcription
hibition by Nucleoside Analogues an
, In-
d Resistance,” Nu-
cleic Acids Research, Vol. 34, No. 1, 2006, pp. 42-52.
doi:10.1093/nar/gkj411
[3] M. Götte, “Inhibition of HIV-1 Reverse Transcription:
Basic Principles of Drug Action and Resistance,” Exp
Review of Anti-Infectivert e Therapy, Vol. 2, 2004, pp.
707-716. doi:10.1586/14789072.2.5.707
[4] M. Götte, S. Fackler, T. Hermann, et al., “HIV-1 Reverse
Transcriptase-Associated Rnase H Cleaves RNA/RNA in
Arrested Complexes: Implications for the Mechanis
us Research, Vol. 134, No. 1-2, 2008, pp
m by
Which Rnase H Discriminates between RNA/RNA and
RNA/DNA,” EMBO Journal, Vol. 14, No. 4, 1995, pp.
833-841.
[5] S. J. Schultz and J. J. Champoux, “RNase H Activity:
Structure, Specificity, and Function in Reverse Transcrip-
tion,” Vir .
86-103. doi:10.1016/j.virusres.2007.12.007
[6] M. Nowotny and W. Yang, “Stepwise Analyses of Metal
Ions in Rnase H Catalysis from Substrate Destabilization
to Product Release,” EMBO Journal, Vol. 25, No. 9,
2006, pp. 1924-1933. doi:10.1038/sj.emboj.7601076
[7] W. Yang, J. Y. Lee and M. Nowotny, “Making and Break-
ing Nucleic Acids: Two-Mg2+-Ion Catalysis and Substrate
Specificity,” Molecular Cell, Vol. 22, No. 1, 2006, pp.
5-13. doi:10.1016/j.molcel.2006.03.013
[8] M. Nowotny, S. A. Gaidamakov, R. J. Crouch, et al.,
“Crystal Structures of RNase H Bound to an RNA/DNA
Hybrid: Substrate Specificity and Metal-Dependent Ca-
talysis,” Cell, Vol. 121, No. 7, 2005, pp. 1005-1016.
doi:10.1016/j.cell.2005.04.024
[9] J. A. Cowan, T. Ohyama, K. Howard, et al., “Metal-Ion
Stoichiometry of the HIV-1 RT Ribonuclease H Dom
Evidence for Two Mutually E
ain:
xclusive Sites Leads to
New Mechanistic Insights on Metal-Mediated Hydrolysis
in Nucleic Acid Biochemistry,” Journal of Biological In-
organic Chemistry, Vol. 5, No. 1, 2000, pp. 67-74.
doi:10.1007/s007750050009
[10] J. Q. Hang, S. Rajendran, Y. L. Yang, et al., “Activity of
the Isolated HIV Rnase H Domain and Specific Inhi
by N-Hydroxyimides,” Bioch bition emical and Biophysical Re-
criptase in the Presence of Magnesium,” Biochem
search Communications, Vol. 317, No. 2, 2004, pp. 321-
329.
[11] K. Pari, G. A. Mueller, E. F. Derose, et al., “Solution
Structure of the Rnase H Domain of the HIV-1 Reverse
Trans ,
Vol. 42, No. 3, 2003, pp. 639-650.
doi:10.1021/bi0204894
[12] K. Katayanagi, M. Okumura and K. Morikawa, “Crystal
Structure of Escherichia Coli Rnase
Mg2+ at 2.8 a Reso
HI in Complex with
lution: Proof for a Single
Mg(2+)-Binding Site,” Proteins, Vol. 17, No. 4, 1993, pp.
337-346.
doi:10.1002/prot.340170402
[13] M. Nowotny, S. A. Gaidamakov, R. Ghirlando, et al.,
“Structure
RNA/DNA Hybrid: Insight in
of Human Rnase H1 Complexed with an
to HIV Reverse Transcrip-
tion,” Molecular Cell, Vol. 28, No. 2, 2007, pp. 264-276.
doi:10.1016/j.molcel.2007.08.015
[14] O. Schatz, F. V. Cromme, F. Grüninger-Leitch, et al.,
“Point Mutations in Conserved Amino Acid Residues
within the C-Terminal Domain of HIV-1 Reverse Tran-
scriptase Specifically Repress RNase H Function,” FEBS
Letters, Vol. 257, No. 2, 1989, pp. 311-314.
doi:10.1016/0014-5793(89)81559-5
[15] S. F. Le Grice, T. Naas, B. Wohigensinger, et al., “Sub-
unit-Selective Mutagenesis Indicates Min
merase Activity in Heterodimer-Ass
imal Poly-
ociated P51 HIV-1
esign, Vol. 12, No. 15, 2006,
Reverse Transcriptase,” The EMBO Journal, Vol. 10, No.
12, 1991, pp. 3905-3911.
[16] K. Klumpp and T. Mirzadegan, “Recent Progress in the
Design of Small Molecule Inhibitors of HIV RNase H,”
Current Pharmaceutical D
pp. 1909- 1922. doi:10.2174/138161206776873653
[17] E. Tramontano, “HIV-1 RNase H: Recent Progress in an
Exciting, Yet Little Explored, Drug Target,” Mini-Reviews
in Medicinal Chemistry, Vol. 6, No. 6, 2006, pp. 727-737.
doi:10.2174/138955706777435733
[18] G. M. Clore and A. M. Gronenborn, “Multidimensional
Heteronuclear Nuclear Magnetic Resonance of Proteins,”
Methods in Enzymology, Vol. 239, 1994, pp. 349-363.
doi:10.1016/S0076-6879(94)39013-4
[19] S. Grzesiek and A. Bax, “Amino Acid Type Determina-
tion in the Sequential Assignment Procedure of U
formly 13C/15N-Enriched Proteins,”
ni-
Journal of Bio-
et al.,
t the RNase H Active
molecular NMR, Vol. 3, No. 2, 1993, pp. 185-204.
[20] D. Goddard and D. G. Kneller, “SPARKY 3,” University
of California, San Francisco, 2000.
[21] D. M. Himmel, K. A. Maegley, T. A. Pauly,
“Structure of HIV-1 Reverse Transcriptase with the In-
hibitor Beta-Thujap- licinol Bound a
Site,” Structure, Vol. 17, No. 12, 2009, pp. 1625-1635.
doi:10.1016/j.str.2009.09.016
[22] A. Jacobo-Molina, J. Ding, R. G. Nanni, et al., “Crystal
Structure of Human Immunodeficiency Virus Type 1
Reverse Transcriptase Comple xed with Double-Stranded
al of Biological Chem-
DNA at 3.0 a Resolution Shows Bent DNA,” Proceed-
ings of the National Academy of Sciences of the USA, Vol.
90, No. 13, 1993, pp. 6320-6324.
[23] J. L. Keck and S. Marqusee, “The Putative Substrate
Recognition Loop of Escherichia Coli Ribonuclease H Is
Not Essential for Activity,” Journ
istry, Vol. 271, No. 33, 1996, pp. 19883-19887.
doi:10.1074/jbc.271.33.19883
[24] G. A. Mueller, K. Pari, E. F. Derose, et al., “Backbone
Dynamics of the Rnase H Domain of HIV-1
Transcriptase,” Biochem, Vol
Reverse
. 43, No. 29, 2004, pp.

J. L. YAN ET AL. 649
r 2D NMR,” Journal of Biomolecular NMR,
9332-9342.
[25] Y. Oda, H. Nakamura, S. Kanaya, et al., “Binding of
Metal Ions to E. coli Rnase HI Observed by 1H-15N
Heteronuclea
Vol. 1, No. 3, 1991, pp. 247-255.
doi:10.1007/BF01875518
[26] Y. Oda, H. Nakamura and S. Kanaya, “Role of Histidine
124 in the Catalytic Function of R
Escherichia coli,” Journalibonuclease HI from of Biological Chemistry, Vol.
se Transcriptase in Solution
n Immunodeficiency Virus Reverse Tran-
268, No. 1, 1993, pp. 88-92.
[27] R. Powers, G. M. Clore, A. Bax, et al., “Secondary Struc-
ture of the Ribonuclease H Domain of the Human Im-
munodeficiency Virus Rever
Using Three-Dimensional Double and Triple Resonance
Heteronuclear Magnetic Resonance Spectroscopy,”
Journal of Molecular Biology, Vol. 221, No. 4, 1991, pp.
1081- 1090.
[28] R. Powers, G. M. Clore, S. J. Stahl, et al., “Analysis of
the Backbone Dynamics of the Ribonuclease H Domain
of the Huma
scriptase Using 15N Relaxation Measurements,” Biochem,
Vol. 31, No. 38, 1992, pp. 9150-9157.
doi:10.1021/bi00153a006
[29] D. Chattopadhyay, B. C. Finzel, S. H. Munson, et al.,
“Crystallographic Analyses of an Active HIV-1 Ribonu-
clease H Domain Show Structural Features That Distin-
guish It from the Inactive Form,” Acta Crystallographica
Section D: Biological Crystallography, Vol. 49, 1993, pp.
423-427. doi:10.1107/S0907444993002409
[30] J. F. Davies, Z. Hostomska, J. Hostomsky, et al., “Crystal
Structure of the Ribonuclease H Domain of HIV-1 Re-
verse Transcriptase,” Science, Vol. 252, No. 5002, 1991,
pp. 88-95. doi:10.1126/science.1707186
[31] J. Jäger, S. J. Smerdon, J. M. Wang, et al., “Comparison
of Three Different Crystal Forms Shows HIV-1 Reverse
Transcriptase Displays an Internal Swivel Motion,”
Structure, Vol. 2, No. 9, 1994, pp. 869-876.
doi:10.1016/S0969-2126(94)00087-5
[32] J. A. Grobler, K. Stillmock, B. Hu, et al., “
Inhibitor Mechanism and HIV-1 Inte
Diketo Acid
grase: Implications
for Metal Binding in the Active Site of Phosphotrans-
ferase Enzymes,” Proc. Natl. Acad. Sci. USA, Vol. 99,
No. 10, 2002, pp. 6661-6666.
doi:10.1073/pnas.092056199
[33] S. R. Budihas, I. Gorshkova, S.
lective Inhibition of HIV-1 R
Gaidamakov, et al., “Se-
everse Transcriptase-Asso-
ciated Ribonuclease H Activity by Hydroxylated Tro-
polones,” Nucleic Acids Research, Vol. 33, No. 4, 2005,
pp. 1249-1256. doi:10.1093/nar/gki268