Biophysical characterization of zinc ejection from HIV nucleocapsid protein by anti-HIV 2,2'-dithiobis[benzamides] and benzisothiazolones

Article abstract of DOI:10.1021/jm960253w

HIV nucleocapsid protein (NCp7) has been suggested as a possible target for 2,2'-dithiobis[benzamide] and benzisothiazolone agents that inhibit viral replication in infected cells (Rice et al. Science 1995, 270, 1194-1197). The solution behavior of these

Full text of DOI:10.1021/jm960253w

J . Med. Chem. 1996, 39, 4313-4320
4313
Biop h ysica l Ch a r a cter iza tion of Zin c Ejection fr om HIV Nu cleoca p sid P r otein
by An ti-HIV 2,2′-Dith iobis[ben za m id es] a n d Ben zisoth ia zolon es
J oseph A. Loo, Tod P. Holler, J oseph Sanchez, Rocco Gogliotti, Lisa Maloney, and Michael D. Reily*
Parke-Davis Pharmaceutical Research, Division of Warner Lambert Company, 2800 Plymouth Road,
Ann Arbor, Michigan 48105
Received April 1, 1996X
HIV nucleocapsid protein (NCp7) has been suggested as a possible target for 2,2′-dithiobis-
[benzamide] and benzisothiazolone agents that inhibit viral replication in infected cells (Rice
et al. Science 1995, 270, 1194-1197). The solution behavior of these compounds and the
mechanistic events leading to removal of Zn from HIV nucleocapsid protein in vitro has been
studied by electrospray ionization mass spectrometry, 500 MHz one- and two-dimensional
nuclear magnetic resonance spectroscopy, and circular dichroism spectroscopy. We demonstrate
that (1) Zn ejection is accompanied by formation of covalent complexes formed between the
2,2′-dithiobis[benzamide] monomers and Cys residues of Zn-depleted NCp7, (2) the rate of Zn
ejection is faster for the C-terminal Zn finger and slower for the N-terminal finger, (3) Zn
ejection results in a loss of structural integrity of the NCp7 protein, and (4) there is no
appreciable interaction between a nonreactive isostere of the lead 2,2′-dithiobis[benzamide]
and NCp7 in buffered aqueous solution. These findings are discussed in terms of the mechanism
of action of Zn ejection by aromatic 2,2′-dithiobis[benzamides].
Ch a r t 1
In tr od u ction
Processing products of retroviral gag polypeptides
invariably include nucleocapsid (NC) proteins that
contain one or two C-X2-C-X4-H-X4-C motifs which are
involved in encapsulation of the genomic RNA prior to
or during viral particle assembly. These motifs are
known to bind Zn2+ with sub-picomolar affinities.1 In
HIV, the first 55 amino acids of the mature form of
nucleocapsid protein, NCp7, have been shown to form
Zn fingers in the presence of stoichiometric Zn2 (see
Chart 1). Certain oxidizing agents, including 3-ni-
trosobenzamides and cupric ions, cause removal of zinc
from NCp7 in vitro.3 Recently, it has also been shown
that certain aromatic 2,2′-dithiobis[benzamides] and
benzisothiazolone compounds, bearing the general struc-
tures shown (2 and 3), also induce extraction of Zn from
the peptide4-7 and possess anti-HIV activity.8 A com-
mon structural feature of these agents is the presence
of electron-poor sulfur atom(s), making it likely that the
first step in Zn ejection is nucleophilic attack of the 2,2′-
dithiobis[benzamide] or benzisothiazolone sulfur by a
Zn-coordinated cysteine sulfur atom. Extensive bio-
chemical characterization of structure-activity relation-
ships (SAR) within this class of compounds has provided
evidence that a recognition event may precede Zn
ejection.4-8 For example, the meta- and para-isomers
of active compounds are inactive,4,7,8 suggesting some
importance of the three-dimensional structure of the
active compounds. If such a recognition complex ex-
isted, it would provide an obvious starting point for
structure-based drug design. It has also been suggested
that NCp7 may be the in vivo target of these compounds
based primarily on whole virion cross-linking studies,
the lack of resistance development to any of these
compounds, and the observation that 2,2′-dithiobis-
[benzamides] which do not inhibit HIV infectivity also
do not eject Zn in vitro.
Work in this and other laboratories has recently been
focused on elucidating the mechanism of action of these
compounds.4-8 This has been complicated by several
factors including the potential for the 2,2′-dithiobis-
[benzamides] to disproportionate into free sulfhydryl
and benzisothiazolone forms and their ability to form
mixed disulfides with endogenous thiols. In this report
we investigate two active compounds, 2 and 3, and a
nonreactive thioether, 6 (see Chart 1). These com-
pounds were evaluated for stability and their ability to
remove Zn from HIV nucleocapsid protein in vitro using
electrospray ionization mass spectrometry (ESI-MS),
500 MHz one- and two-dimensional nuclear magnetic
resonance (NMR) spectroscopy, and circular dichroism
(CD) spectroscopy.
* To whom correspondence should be addressed.
X Abstract published in Advance ACS Abstracts, September 1, 1996.
S0022-2623(96)00253-1 CCC: $12.00 © 1996 American Chemical Society

4314 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21
Loo et al.
F igu r e 2. Effect of pH on zinc binding measured by ESI-MS.
A 15 µM NCp7 solution containing 30 µM ZnCl2 was prepared
in 2:1 MeOH/H2O with 2.5% (v/v) acetic acid (pH 2.5) (A) or
25 mM ammonium acetate, pH 6.9 (aqueous) (B) and analyzed
by ESI-MS. Both the ESI-MS and the deconvoluted spectra
(converted to the mass domain, inset) are shown. The spectrum
acquired at pH 2.5 showed a measured mass of 6369.5 ( 0.4
Da, consistent for the apoprotein (theory 6369.5 Da). At pH
6.9, the dominant species is consistent for two zinc ions bound
to the protein.
pH 7.4, 2 is still the major species with small amounts
of the benzisothiazolone 3 and the thiol 7 present in
solution. At pH 8.5, the equilibrium is pushed further,
and 2 is the minor species in solution. Also, as the
concentration is decreased at pH 7.4, the amount of 2
decreases with a concomitant increase in species 3 and
7. Under all of the conditions studied, both 3 and 7 were
present in approximately equal proportions.
Ch a r a cter iza tion of Zn Ejection by Ma ss Sp ec-
tr om etr y. The changes in mass of Zn-loaded NCp7
proved to be useful in characterizing Zn ejection. Elec-
trospray ionization mass spectrometry has recently been
used to study the metal-binding behavior of metallo-
proteins.9 In particular, the zinc-binding characteristics
of zinc finger protein glucocorticoid receptor DNA-
binding domain10 and nucleocapsid protein11,12 have
been studied using ESI-MS. Our studies with NCp7 are
shown in the following examples. The ESI mass spec-
trum of a zinc-containing solution of NCp7 in 2:1
methanol/water with 2.5% (v/v) acetic acid (pH ∼2.5)
showed a measured molecular weight of 6370 (Figure
2A), consistent for the apoprotein mass. The protein is
not expected to bind zinc at this acidic pH due to the
protonated state of the histidine ligands. NCp7 contains
two zinc fingers, and at pH 6.9 in ammonium acetate
(aqueous), the expected protein-Zn2 species was de-
tected as the major species (Figure 2b).
F igu r e 1. Effect of pH and concentration on the aromatic
spectral regions of the 500 MHz 1H NMR spectra of H2O
solutions of 2. The H6 signal for each species is identified, and
all other signals arise from the other aromatic and amide
protons in each species. All concentrations are adjusted to
reflect [monomer]: (A) 0.459 mM 2 at pH 7.4, 20 mM sodium
phosphate buffer; (B) Sample in A raised to pH 8.5; (C) sample
in B lowered to pH 6.6, the predominant signals in C are from
the 2,2′-dithiobis[benzamide] 2; (D) 2.93 mM 3 at pH 6.9, 20
mM sodium phosphate buffer, since 3 is stable in the absence
of reducing agents, spectrum D provides a reference spectrum
for 3 for purposes of comparison; (E) 2.16 mM 2 at pH 7.4,
equilibrated for 40 min; (F) 4-fold dilution of E with H2
O; (G)
16-fold dilution of E with H2O. Top: Relative amounts of
dimeric and monomeric species present at various concentra-
tions of 2 at pH 7.4. These data were extracted from aromatic
signal peak heights as a function of concentration.
Resu lts
Equ ilibr iu m Beh a vior of th e 2,2′-Dith iobis[ben -
za m id es] in Solu tion . When dissolved in aqueous
solution, many 2,2′-dithiobis[benzamides] reversibly
disproportionate to form one molecule of mercaptan and
one molecule of benzisothiazolone. The effect of pH and
concentration on this equilibrium between 2 and its
monomeric relatives (3 and 7), as monitored by the
proton NMR spectrum, is shown in Figure 1. For these
experiments, 2 was dissolved in buffer at various
concentrations and pH’s and the relative amounts of
each species were determined by peak height after
equilibrium was established. At pH 6.6 and 0.5-1.0
mM concentrations of 2, the predominant species in
solution is the dimer 2. At similar concentrations at
ESI-MS experiments were performed to monitor
products from the reaction between NCp7-Zn2 and
selected 2,2′-dithiobis[benzamide] compounds. Two min-
utes after addition of 4 mol equiv of 2, the major
products were the [NCp7 + 2Zn] and [NCp7 + 266 +

Biophysical Characterization of Zn Ejection from NCp7
J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21 4315
F igu r e 3. Time course study of Zn ejection from NCp7 by 2
as monitored by ESI-MS. Deconvoluted mass spectra taken
17 min (A) and 60 min (B) after 60 µM 2 was added to a
solution containing 15 µM NCp7, 30 µM ZnCl2, and 25 mM
ammonium acetate, pH 6.9.
F igu r e 4. Time course study of Zn ejection from NCp7 by 3
as monitored by ESI-MS. Deconvoluted mass spectra taken
10 min (A) and 65 min (B) after addition of 60 µM 3 to a
solution containing 15 µM NCp7, 30 µM ZnCl2, and 25 mM
ammonium acetate, pH 6.9.
Zn] molecules, where 266 Da represents one-half of the
2,2′-dithiobis[benzamide] drug molecule. Some of the
apoform (zinc-free protein) was also found at this time
point. At later time periods, the apoprotein and its
various covalently modified forms, e.g., [NCp7 + 2(266)],
are the major products. Figure 3 shows the zinc ejection
profile induced by 2 after 17 and 60 min.
Accurate high-resolution mass measurements indi-
cate that up to a total of three intramolecular Cys-Cys
bonds are formed upon zinc ejection. Each disulfide
bond formation results in a net decrease of 2 Da relative
to the fully reduced apoprotein. The measured molec-
ular weight of some of the apoprotein formed by 2,2′-
dithiobis[benzamide]-induced zinc ejection (6363.4 Da)
was 6 Da less than the theoretical mass of the native
protein (6369.5 Da). This is consistent with the report
by Fenselau and co-workers,3 who found that a total of
three disulfide bonds formed upon oxidation of NCp7
by 3-nitrobenzamide.
For 3, the benzisothiazolone form of 2, only one adduct
species was found in relatively large quantities, with
release of zinc (Figure 4), whereas the 2,2′-dithiobis-
[benzamide] 2 gave many more adduct species with
different numbers of bound drug molecules (e.g., +1
drug molecule, +2, ...+6 drug molecules) (Figure 3).
Figure 5 shows a time course plot of zinc ejection from
NCp7 upon addition of compounds 2, 3, and 6 measured
by ESI-MS. All of the protein species containing zero,
one, and two zinc ions were added and normalized to
the total protein signal. Both 2 and 3 showed moderate
activity for zinc ejection in less than a 1 h time period,
with the 2,2′-dithiobis[benzamide] form being slightly
more active than the benzisothiazolone.
F igu r e 5. Time course study of total zinc ejection from NCp7
measured by MS. The absolute abundances (peak intensity)
for each protein-zinc species were summed in each mass
spectra acquired at each time point and normalized to the
total protein ion signal. For example, all of the ion signals
for the NCp7-Zn1 species were counted together (i.e.,
[NCp7 + Zn] + [NCp7 + Zn + drug] + [NCp7 + Zn + 2drug]
+ etc.).
tandem mass spectrometry (MS/MS), and liquid chro-
matography-MS (LC-MS), the C-terminal zinc finger
region was found to be more reactive with 2 than the
N-terminal zinc finger. From limited trypsin digestion
of drug-bound NCp7-Zn2, separate tryptic peptides
containing Cys36 and Cys49 were observed to be co-
valently modified by the drug. The MS data were not
able to distinguish which of these two Cys residues
reacts first with the drug. It is possible that Cys36 and
Cys49 have nearly equal reactivity.
The sites of these covalent modifications during the
early reaction stages (i.e., initial 10 min of reaction)
were identified by analyzing proteolyzed fragments of
NCp7-Zn2 obtained by exposing the drug-treated pro-
tein to trypsin. From ESI-MS mass measurements,
Ch a r a cter iza tion of Zn Ejection by NMR Sp ec-
tr oscop y. The utility of NMR spectroscopy to monitor

4316 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21
Loo et al.
F igu r e 6. Titration of ZnCl2 into a 0.5 mM solution of apo-
NCp7 in 60 mM sodium phosphate, pH 7.4. The downfield
amide proton region of the 500 MHz 1H NMR spectrum is
shown as a function of Zn/NCp7 ratio. Spectral assignments
are indicated.
F igu r e 7. Time course of Zn ejection from NCp7 by 3 as
monitored by 1H NMR (downfield amide region). Spectra were
run on 0.6 mM NCp7 samples dissolved in 60 mM sodium
phosphate buffer, pH 7.4, at 25 °C. The bottom trace shows
NCp7 in the absence of compound. From left to right, peak
assignments are Trp37 indole, Asn17 NH, and Lys38 NH.
Traces 2-6 show spectra recorded at the indicated times (in
minutes) after addition of 2 equiv of 3. Immediately after the
27 min spectrum was recorded, an additional 2 equiv of the
compound was added, and the 32 and 39 min time points were
acquired.
the ejection of zinc from NCp7 arises from the spectral
dispersion in the proton NMR signals caused by the zinc
finger structure. In particular, two amide proton sig-
nals are highly deshielded from the envelope of other
peptide amide signals in the Zn-loaded peptide but not
in apo-NCp7. These signals have been previously
assigned to Asn17 (9.49 ppm) and Lys38 (9.28 ppm) in
NCp7-Zn2.2,13,14 In order to better understand the
NMR spectral changes occurring during Zn ejection, the
reverse reaction was studied by titrating ZnCl2 into a
solution of apo-NCp7. Figure 6 shows the development
of amide proton signals arising from the singly and
doubly coordinated NCp7 as Zn is added. Two-dimen-
sional TOCSY and NOESY experiments were used to
make assignments. The observation of distinct reso-
nances for the mono-Zn complex indicates that Zn
coordination is noncooperative or negatively cooperative
and is qualitatively consistent with the recently reported
results of Me´ly et al.15 Cooperative addition of Zn would
require that occupation of a single Zn site in apo-NCp7
would increase the affinity for the second site. This
would require that apo- or dicoordinated species would
predominate at all Zn/peptide ratios e 2. Crosspeaks
at the Asn17 HR and Hâ chemical shifts in the two-
dimensional TOCSY spectrum recorded on a partially
Zn-loaded sample of NCp7 (not shown) showed that the
amide resonance assigned to the monocoordinated spe-
cies is from Asn17. It can therefore be concluded that
the N-terminal Zn finger of NCp7 is preferentially
coordinated to generate a mono-Zn finger form of NCp7
at low Zn/peptide ratios. The chemical shift of the
Trp37 indole proton is slightly different and shifted
upfield on going from apo-NCp7 to NCp7-Zn to NCp7-
Zn2, and the chemical shift of Asn17 NH is slightly
further downfield in NCp7-Zn than in NCp7-Zn2.
Since these two protons report on different Zn fingers,
these results suggest some interaction does occur be-
tween the two Zn finger moieties under these conditions,
as previously suggest by Morlett et al.14,15 However, the
observed chemical shift differences are small, and little
can be ascertained about the nature of the interaction
from this data.
The observation of spectral signals from each Zn
finger allowed us to measure the relative sensitivity to
2,2′-dithiobis[benzamide]- and benzisothiazolone-in-
duced metal ejection of each of the Zn fingers in intact
NCp7. In one NMR experiment, 2 equiv of 3 were added
to a solution of NCp7 and the ejection of Zn from each
finger was followed by monitoring the loss of intensity
of marker resonances from each finger (Figure 7).
Experiments performed with 2 yielded similar results.
In a different approach, 3 was titrated into a solution
of NCp7 and the state of the peptide was monitored
after each titration point had equilibrated, Figure 8. In
all cases, we observed a loss in intensity of the reso-
nances from Asn17 and Lys38 amide protons with
concomitant growth of a resonance at a position previ-
ously assigned to Asn17 NH in mono-Zn NCp7, sug-
gesting that Zn is ejected from the C-terminal Zn finger
much faster than from the N-terminal Zn finger.
The expected structural changes upon Zn ejection
based on the NMR spectroscopic observations described
above led us to postulate that circular dichroism (CD)
could be used to monitor Zn ejection from NCp7. The
CD spectra of NCp7 before and after equilibration with
2 are shown in Figure 9. The pronounced changes
observed in the CD spectrum are consistent with a loss
of structural integrity in NCp7 upon treatment with 2
and removal of the Zn. The large ∆ꢀ at 220 nm allowed
the time course of Zn ejection to be followed as a function
of solution conditions. Temporal changes in ꢀ220 of
NCp7 upon addition of 2 are shown in Figure 9 at two

Biophysical Characterization of Zn Ejection from NCp7
J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21 4317
F igu r e 8. Titration of NCp7 with 3 as monitored by 1H NMR
(downfield amide region). Spectra were run on 0.3 mM NCp7
samples dissolved in 60 mM sodium phosphate, pH 7.4, at 25
°C. The bottom trace shows NCp7 in the absence of compound.
From left to right, peak assignments are Trp indole, Asn17
NH, and Lys38 NH. Traces 2-6: after equilibrating with the
indicated number of equivalents of 3.
F igu r e 9. CD spectral changes in NCp7 as a result of
treatment with 2: (A) 30 µM NCp7 in 50 mM sodium
phosphate, pH 7.0, at 25 °C before (s) and after (- - -)
equilibration with 4.5 equiv of 2 and (B) time dependence of
the ellipticity (at 220 nm) of NCp7 in phosphate buffer after
addition of 4.5 equiv of 2 under the following conditions (a)
9.8 µM NCp7, pH 7.0, and (b) 73.4 µM NCp7, pH 7.0. Double-
exponential fits to the time-dependent data are shown.
concentrations with a fixed 2,2′-dithiobis[benzamide]/
NCp7 ratio of about 4. As indicated in Figure 9, these
data required a minimum of two exponentials for a good
fit.
Non r ea ctive 2,2′-Dith iobis[ben za m id e] Isoster e.
On the basis of the SAR of existing 2,2′-dithiobis-
[benzamide] compounds,4,7,8 it has been hypothesized
that interactions of these agents with NCp7 could
involve a prereaction complex in which the agent binds
in a specific manner to a site on the peptide near the
attacking Zn-liganded cysteine. One problem with
identifying such an adduct is the relatively high reactiv-
ity of the 2,2′-dithiobis[benzamide] compounds. Because
of the structural changes that NCp7 undergoes, the
recognition complex would not be expected to survive
Zn ejection. To see if a pre-ejection complex is formed,
the interaction of a nonreactive thioether isostere of 2
(6) has been studied. Since peptide chemical shifts are
sensitive to structural environs and because of the
anisotropic nature of the aromatic groups in the com-
pound, proton resonances from amino acids that interact
with the compound should be shifted. The NH and HR
chemical shifts in NCp7 were recorded as a function of
added 6 using TOCSY. In phosphate buffer at pH 5.9,
there were no detectable shifts in the resonances of
either 6 or Zn-loaded NCp7, even at a compound/NCp7
ratio of 6, suggesting that there is no substantial
interaction between 6 and NCp7 under the NMR
conditions. Furthermore, no new proton signals indica-
tive of Zn ejection (e.g., Trp37 indole) appeared, sug-
gesting that no Zn was ejected in solution.
the NCp7-Zn2 species that persisted over a 2 h period
with very little ejection of zinc observed. The relative
abundance of the [NCp7 + 2Zn + drug] ion was less
than 20% of the [NCp7 + 2Zn] ion signal. Less than
0.2 equiv of zinc was ejected over the entire time
monitored (Figure 5). On the basis of the lack of Zn
ejection observed in the NMR solutions (see above), and
the fact that 6 cannot act as an electrophile for Zn
ejection, it is most likely that the small amount of Zn
ejected in this experiment occurs due to interactions in
the gas phase and cannot be interpreted as evidence for
solution phase interactions. ESI-MS has demonstrated
the capabilities to detect weakly bound macromolecular
complexes. The ESI-MS experimental conditions used
for this study were such that any specific noncovalent
complexes would have been more prominent in the
spectrum. The presence of the drug adduct is therefore
most likely due to nonspecific interactions that may
occur from the electrospray ionization process, e.g.,
adduct formation from the anionic 6 and the cationic
NCp7 in the gas phase.
Discu ssion
NMR studies with the 2,2′-dithiobis[benzamide] 2
suggest that at physiologically relevant concentrations
and pH, the monomeric thiol and benzisothiazolone
forms of the compound represent significant populations
Electrospray ionization mass spectra of NCp7 with 6
showed only a weak noncovalent adduct formed with

4318 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21
Loo et al.
F igu r e 10. Possible reaction pathways in the interaction of 2,2′-dithiobis[benzamides] with NCp7.
in solution and that equilibrium between the various
species is rapidly established. The benzisothiazolone 3,
on the other hand, is stable under these conditions in
the absence of reducing agents. Both the 2,2′-dithiobis-
[benzamide] and the benzisothiazolone are potential
targets for nucleophilic attack, but the presence of
benzisothiazolone in solutions of 2,2′-dithiobis[benza-
mide] makes it difficult to ascertain which is better at
ejecting Zn. The different distribution of products
observed in the mass spectra upon treatment with 2 and
3 suggests that both oxidized species do react with NCp7
via distinct mechanisms.
On the basis of Zn ejection behavior as monitored by
N-(6-methyl-8-quinolyl)-p-toluenesulfonamide (TSQ) and
tryptophan fluorescence, it has been proposed that the
kinetics of Zn ejection from the C-terminal finger are
approximately 8-fold faster than from the N-terminal.5
The similar biexponential rates of Zn ejection as mea-
sured by CD suggest that Zn ejection and protein
unfolding are simultaneous events, since CD provides
a direct measure of the rate of structural changes,
whereas the TSQ fluorescence data measure the rate
of liberation of Zn. That the C-terminal Zn finger ejects
Zn faster is supported by the differential rates of
disappearance of the amide protons associated with
individual Zn fingers. By monitoring the same reso-
nances going in the other direction, that is, adding Zn
to apo-NCp7, it is clear that the N-terminal Zn site is
occupied with higher affinity. This relative binding
affinity was also observed with peptides that contained
the individual Zn fingers.11 In addition, the mass
spectrometry data support the hypothesis that the
C-terminal zinc finger is initially attacked by 2 or 3.
Thus it is clear that Zn is ejected slower from the more
thermodynamically stable (N-terminal) Zn finger. It
seems unlikely, however, that thermodynamic param-
eters drive the observed kinetics because once disulfide
exchange occurs, the Zn finger cannot reform. There-
fore, equilibrium cannot be established, and thermody-
namic arguments are not applicable. An alternative
explanation is that one of the C-terminal cysteine
sulfurs is the most reactive nucleophile, which may or
may not be related to the Zn-binding affinity for the
individual fingers. The nucleophilicity of a particular
cysteine could be due to electronic and/or steric consid-
erations mandated by the secondary or tertiary struc-
ture of NCp7. Alternatively, if a recognition step is
involved prior to electron transfer, then a C-terminal
Zn finger recognition complex is expected to be more
disposed to the initial electrophillic attack than an
N-terminal complex. However, the data presented here
with the nonreactive isostere 6, while not conclusive
evidence for the lack of a recognition complex, if taken
with the apparent nonsaturable kinetic profile for
similar compounds,5 suggest that the existence of a
Our results clearly indicate that anti-HIV 2,2′-dithio-
bis[benzamides] cause Zn ejection from NCp7 in a
stoichiometric, oxidative manner in vitro. Prominent
among the final products of 2,2′-dithiobis[benzamides]
or benzisothiazolones interacting with Zn-loaded NCp7
are drug-apoprotein adducts and internally disulfide-
bonded apoprotein molecules. In terms of possible
reaction mechanisms, the process can be viewed as an
intermolecular disulfide formation followed by Zn ejec-
tion with concerted or subsequent intramolecular dis-
ulfide bond formation and/or rearrangement (Figure 10).
First, a relatively slow nucleophilic attack of the oxi-
dized (aromatic) sulfur by a Zn-coordinated cysteine
sulfur atom results in a destabilization of the Zn
coordination sphere. This is followed by (1) rapid
formation of intramolecular disulfides within the protein
and ejection of Zn or (2) rapid diffusion of Zn away from
the protein followed by disulfide exchange. This view
of the reaction mechanism is supported by the observa-
tion that peaks in the mass spectrum corresponding to
[NCp7 + 2Zn + drug] are very low in abundance. The
subsequent disulfide bond rearrangement stage occurs
by formation of intramolecular (and possibly inter-
molecular) disulfide bonds involving cysteine sulfurs
which are (or were previously) complexed to Zn. This
disulfide formation could include attack of the mixed
aromatic cysteine disulfide formed in the initial reac-
tion by another electrophilic sulfur, with the aromatic
sulfhydryl acting as a leaving group, or alternatively,
attack of another aromatic 2,2′-dithiobis[benzamide],
Figure 10.

Biophysical Characterization of Zn Ejection from NCp7
J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21 4319
remove excess bromine. The residue was partitioned between
dichloromethane/5% sodium bicarbonate (200 mL of each). The
aqueous layer was washed with dichloromethane and acidified
to pH 1.5 with 6.0 M hydrochloric acid. After extracting with
dichloromethane (2 × 75 mL), the combined organic layers
were washed with water, dried (MgSO4), filtered, and evapo-
rated in vacuo to give 4.8 g (91%) of 3: mp 50-52 °C; NMR
(MeSO-d6) δ 0.84 (t, 3H), 1.01 (d, 3H), 1.18 (m, 1H), 1.28 (m,
1H), 2.12 (m, 1H), 5.00 (d, 1H), 7.48 (t, 1H), 7.73 (t, 1H), 7.92
(dd, 2H); MS (m/ z relative intensity) 264 (100, M - 1). Anal.
(C13H15NO3S‚0.25H2O) C, H, N.
specific complex is unlikely. This being said, all of the
Zn ejection studies mentioned heretofore have been
conducted on naked NCp7 in the absence of RNA. It is
entirely possible that recognition could be a factor
within the NCp7-RNA-drug ternary complex.
In summary, we have shown that Zn ejection is
associated with initial intermolecular disulfide bond
formation between a Zn-liganded cysteine and the 2,2′-
dithiobis[benzamide] compound. Large conformational
changes in the peptide accompany Zn ejection, and the
rate of ejection is faster for the C-terminal finger than
the N-terminal finger. The fast Zn ejection rate is
strongly concentration dependent, but both rates are
independent of pH in the range of 6-7. It is clear that
2,2′-dithiobis[benzamide] and benzisothiazolone agents
disrupt the structure of NCp7 by coordinating to Zn
ligands within the peptide. It follows that the functional
properties of the nucleocapsid are compromised by this
perturbation, and this could provide the basis for
inhibition of HIV infectivity. Although it is tempting
to speculate that NCp7 is the relevant target for these
compounds, their therapeutic efficacy and the absolute
identification of their site of action await more rigorous
cellular and in vivo testing.
2-[[(1-Ca r boxy-2-p h en yl)su lfa n yl]m eth yl]ben zoic Acid
(4). A solution of sodium ethoxide (6.8 g, 75 mmol) in ethanol
(25 mL) and methyl 2-mercaptobenzoate (12.6 g, 75 mmol) was
allowed to mix together for a period of 5 min, phthalide (10.0
g, 75 mmol) was added, and the solution was refluxed for an
18 h period. The resulting slurry was cooled, treated with a
solution of KOH (10.0 g) in 33 mL of 10% aqueous ethanol,
and heated to reflux for a 2 h period. The mixture was cooled
and filtered, and the solid was washed with ethanol. The solid
was dissolved in water, dilute HCl was added to pH 2, and
the product was recovered by filtration. Recrystallization from
hot acetic acid afforded 7.6 g (37%) of 4: mp 256-257 °C; NMR
(Me2SO-d6) δ 4.59 (s, 2H, S-CH2-Ar), 7.19-7.23 (m, 1H, Ar),
7.37-7.39 (m, 1H, Ar), 7.41-7.56 (m, 5H, Ar), 7.85-7.89 (m,
1H, Ar); MS (m/ z) 289 (M + H)+. Anal. (C15H12O4S) C, H, N.
[2S-[1[R*(R*)],2R*,3R*]]-2-[[2-[[[2-[(1-Ca r boxy-2-m eth -
ylb u t yl)ca r b a m oyl]p h e n yl]su lfa n yl]m e t h yl]b e n zoyl]-
a m in o]-3-m eth ylp en ta n oic Acid (6). A solution of 4 (1.6
g, 5 mmol) in thionyl chloride (20 mL) was heated to reflux
for 4 h. The solution was cooled, concentrated under reduced
pressure, and dried in vacuo at 40 °C to afford 1.6 g of crude
acid chloride 5. An ice-cooled solution of L-isoleucine tert-butyl
ester, monohydrochloride (1.3 g, 5.0 mmol) and N-methylmor-
pholine (1.4 mL, 13 mmol) in dichloromethane (30 mL) was
treated dropwise with the crude acid chloride 5 (0.8 g, 2.7
mmol) in dichloromethane (10 mL). The reaction mixture was
allowed to warm to ambient temperature and mix for an
additional 18 h. The solution was extracted with 5% citric
acid, 8% NaHCO3, and brine, dried with MgSO4, and concen-
trated under reduced pressure. The residue was dissolved in
dichloromethane (15 mL) containing anisole (1.5 mL) and
treated with trifluoroacetic acid (15 mL). The solvents were
removed after 6 h under reduced pressure and recrystallization
from dichloromethane/ether/hexane afforded 0.9 g (77%) of 6:
mp 165-168 °C; NMR (Me2SO-d6) δ 0.79-0.92 (m, 12Η, 4
CΗ3), 1.18-1.33 (m, 2H, CH), 1.40-1.53 (m, 2H, CH), 1.85-
1.88 (m, 2H, CH), 4.23-4.40 (m, 4H, S-CH2, CH-CO2), 7.23-
7.40 (m, 8H, Ar), 8.44-8.47 (d, 1H, J ) 3 Hz, NH), 8.52-8.55
(d, 1H, J ) 3 Hz, NH), 12.6 (broad peak, 2H, CO2H); MS (m/
z) 513 (M - H)-. Anal. (C27H34N2O6S) C, H, N.
NMR Sp ectr oscop y. All NMR spectra were collected on
a Bruker AMX500 spectrometer equipped with a thermostati-
cally controlled 1H-observed triple resonance probe. Two-
dimensional total correlation (TOCSY) spectra17 with 60 ms 7
kHz mlev-1618 spin-lock fields and two-dimensional nu-
clear Overhauser (NOESY) spectra19 were acquired at 298 or
288 K with 250 ms mixing times. All NMR spectra were
processed using UXNMR (Bruker Instruments). For two-
dimensional spectra, 1024 complex points were acquired in the
t2 dimension and 512 or 700 points in t1. The data were zero-
filled to a final matrix size of 2048 × 1024 points. Signal-to-
noise was enhanced, and truncation artifacts were minimized
by application of a cos2 function in both dimensions prior to
Fourier transformation. One-dimensional spectra contained
16K complex points. Quadrature detection in the t1 dimension
was achieved by the TPPI method20 in two-dimensional
spectra.
Exp er im en ta l Section
The expression and isolation of NCp7 protein was reported
previously.5 Synthesis of the three sulfur-containing com-
pounds are described below.
[S-(R*,R*)]-2-[[2-[[2-[(1-Car boxy-2-m eth ylbu tyl)car bam -
oyl]p h en yl]d isu lfa n yl]ben zoyl]a m in o]-3-m eth ylp en ta n o-
ic Acid ter t-Bu tyl Ester (1). A solution of 10.0 g (53.2 mmol)
of L-isoleucine tert-butyl ester in 100 mL of dichloromethane
was treated with 5.6 g (55.0 mmol) of N-methylmorpholine.
The resulting solution was cooled to 0 °C and treated rapidly
dropwise with a solution of 8.3 g (24.2 mmol) of 2,2′-dithiobis-
[benzoyl chloride]15 in 100 mL of dichloromethane, keeping the
temperature below 0 °C. The mixture was stirred at 0 °C for
1 h and allowed to come to room temperature over 18 h. The
solid was removed by filtration, washed with water, and dried
in vacuo to give 6.5 g of 1, mp 189-191 °C. The filtrate was
washed with water, 0.5 M hydrochloric acid, and water, dried
(MgSO4), filtered, and evaporated in vacuo to give an ad-
ditional 6.9 g of 1 of comparable purity. The combined
fractions, 14.4 g, gave an overall yield of 86%: NMR (MeSO-
d6) δ 0.89 (t, 6H), 0.95 (d, 6H), 1.30 (m, 2H), 1.44 (s, 18H),
1.52 (m, 2H), 1.92 (m, 2H), 4.22 (t, 2H), 7.32 (t, 2H), 7.45 (t,
2H), 7.63 (dd, 4H), 8.76 (d, 2H); MS (m/ z relative intensity)
645 (46, M+), 322 (100, 1/2M+). Anal. (C34H48N2O6S2‚0.25H2O)
C, H, N.
[S-(R*,R*)]-2-[[2-[[2-[(1-Car boxy-2-m eth ylbu tyl)car bam -
oyl]p h en yl]d isu lfa n yl]ben zoyl]a m in o]-3-m eth ylp en ta n o-
ic Acid (2). A solution of 13.2 g (20.5 mmol) of the tert-butyl
ester 1 in 50 mL of trifluoroacetic acid was stirred at room
temperature for 18 h. The solvent was removed in vacuo, and
the residue was dissolved in 50 mL of dichloromethane which
was also removed in vacuo. The residue was triturated with
150 mL of ether/pentane (2:1), and the resulting solid was
removed by filtration. After washing with 50 mL of ether/
pentane (2:1) and pentane, the solid was dried in vacuo to give
9.9 g (91%) of 2: mp 211-213 °C; NMR (MeSO-d6) δ 0.82 (t,
6H), 0.90 (d, 6H), 1.25 (m, 2H), 1.45 (m, 2H), 1.89 (m, 2H),
4.29 (t, 2H), 7.25 (t, 2H), 7.40 (t, 2H), 7.59 (d, 4H), 8.68 (d,
2H), 12.63 (s, 2H); MS (m/ z relative intensity) 531 (15, M -
1
1), 266 (100, /2M+). Anal. (C26H32N2O6S2) C, H, N.
Protein and thio compound samples were prepared by
dissolving in 20-50 mM sodium phosphate buffer with warm-
ing at 37 °C and agitation if necessary. Solution pH’s were
checked and adjusted with dilute NaOH or HCl when neces-
sary. Stock solutions of 2 and 3 (1-4 mM) were prepared by
agitating and warming the buffer/compound mixture for 15
min at 37 °C, and these were monitored by 1H NMR. The
solution of the benzisothiazolone contained one species and
[S-(R*,R*)]-3-Met h yl-2-(3-oxo-3H-b en z[d ]isot h ia zol-2-
yl)p en ta n oic Acid (3). A stirred, room temperature suspen-
sion of 5.3 g (10.0 mmol) of 2 in 200 mL of dichloromethane
was treated dropwise with 2.4 g (15.0 mmol) of liquid bromine.
The reaction mixture was stirred at room temperature for 2 h
and evaporated in vacuo. The residue was triturated with
dichloromethane which was also evaporated in vacuo to

4320 J ournal of Medicinal Chemistry, 1996, Vol. 39, No. 21
Loo et al.
(6) Reily, M. D.; Loo, J . A. Biophysical Characterization of Zn
Ejection from HIV Nucleocapsid Protein by Aromatic Disulfides.
3rd Conference on Retroviruses and Opportunistic Infections,
Washington, DC, J an. 28-Feb. 1, 1996; Abstract 339.
(7) Domagala, J .; Gogliotti, R.; Sanchez, J .; Stier, M.; Tummino, P.;
Gracheck, S.; Hupe, D.; Bader, J .; Schultz, R.; Rice, W. Two
Novel Classes of HIV-1 Inhibitors Which Interact With Nucleo-
capsid Protein p7NC: Structure Activity Relationships vs.
Antiviral Activity and Correlation with Zn Ejection. 3rd Confer-
ence on Retroviruses and Opportunistic Infections, Washington,
DC, J an. 28-Feb. 1, 1996; Abstract 338.
(8) Rice, W. G.; Supko, J . G.; Malspeis, L.; Buckheit, R. W., J r.;
Clanton, D.; Bu, M.; Graham, L.; Schaffer, C. A.; Turpin, J . A.;
Domagala, J .; Gogliotti, R.; Bader, J . P.; Halliday, S. M.; Coren,
L.; Sowder, R. C., II; Arthur, L. O.; Henderson, L. E. Inhibitors
of HIV Nucleocapsid Protein Zinc Fingers as Candidates for the
Treatment of AIDS. Science 1995, 270, 1194-1197.
remained unchanged during the course of the experiment,
whereas the 2,2′-dithiobis[benzamide] rapidly equilibrated
to form monomer and benzisothiazolone (see above). The
pH of reaction mixtures was verified before and after the
reactions.
Ma ss Sp ectr om etr y. Positive ion ESI-MS analyses were
performed with a Finnigan MAT 900Q forward geometry
hybrid mass spectrometer (Bremen, Germany)21 equipped with
a position- and time-resolved ion-counting (PATRIC) focal-
plane array detector22 after the magnet and before the quad-
rupole analyzer. Mass spectra were acquired at full acceler-
ating potential (5 kV) and a scan rate of 10 s decade-1
.
The electrospray ionization interface used was based on a
heated glass capillary inlet.23 Droplet desolvation was ac-
complished by a countercurrent stream of warm nitrogen (∼60
°C) and by energetic collisions in the electrospray interface.
Sulfur hexafluoride coaxial to the spray was used to suppress
corona discharges. Protein solutions for ESI-MS measure-
ments were prepared at a concentration of 10-20 µM in 25
mM ammonium acetate (pH 6.9) with 2 times the protein
molar concentration of ZnCl2 in water. Stock solutions of the
sulfur-containing compounds were prepared at a concentration
of ca. 10 mM in MeOH. Sample solutions were infused through
(9) Hu, P.; Ye, Q.-Z.; Loo, J . A. Calcium stoichiometry determination
for calcium binding proteins by electrospray ionization mass
spectrometry. Anal. Chem. 1994, 66, 4190-4194.
(10) Witkowska, H. E.; Shackleton, C. H. L.; Dahlman-Wright, K.;
Kim, J . Y.; Gustafsson, J .-A. Mass spectrometric analysis of a
native zinc-finger structure: the glucocorticoid receptor DNA
binding domain. J . Am. Chem. Soc. 1995, 117, 3319-3324.
(11) Surovoy, A.; Waidelich, D.; J ung, G. Nucleocapsid protein of
HIV-1 and its Zn2+ complex formation analysis with electrospray
mass spectrometry. FEBS Lett. 1992, 311, 259-262.
(12) Fenselau, C.; Yu, X.; Bryant, D.; Bowers, M. A.; Sowder, R. C.,
II; Henderson, L. E. Characterization of processed gag proteins
from highly replicating HIV-1MN. In Mass Spectrometry for the
Characterization of Microorganisms; Fenselau, C., Ed.; ACS
Symposium Series; American Chemical Society: Washington,
DC, 1994; pp 159-172.
the ESI source at a flow rate of 1 µL min-1
Microbore LC-MS experiments with electrospray ionization
were performed with VG Micromass Trio-2000 single-
.
a
quadrupole mass spectrometer and a Michrome Bioresources
HPLC unit (Vydac C18 1 mm × 150 mm column).
Limited trypsin digestion of drug-bound NCp7 was per-
formed by incubating the drug compound with NCp7-Zn2 for
10 min in 25 mM ammonium acetate (pH 6.9) followed by the
addition of trypsin (Promega, Madison, WI) to the 20% level
(w/w). The reaction was quenched after 10 min with 0.1% (v/
v) trifluoroacetic acid.
(13) Omichinski, J . G.; Clore, G. M.; Sakaguchi, K.; Appella, E.;
Gronenborn, A. M. Structural Characterization of a 39-residue
Synthetic Peptide Containing the Two Zinc Binding Domains
From the HIV-1 p7 Nucleocapsid Protein by CD and NMR
Spectroscopy. FEBS Lett. 1991, 292, 25-30.
(14) Morellet, N.; de Rocquigny, H.; Me´ly, Y.; J ullian, N.; De´me´ne´,
H.; Ottmann, M.; Ge´rard, D.; Darlix, J . L.; Fournie-Zaluski, M.
C.; Roques, B. P. Conformational Behavior of the Active and
Inactive Forms of the Nucleocapsid NCp7 of HIV-1 Studied by
1H NMR. J . Mol. Biol. 1994, 235, 287-301.
(15) Me´ly, Y.; De Rocquigny, H.; Morellet, N.; Roques, B. P.; Gerard,
D. Zinc Binding to the HIV-1 Nucleocapsid Protein: A Thermo-
dynamic Investigation by Fluorescence Spectroscopy. Biochem-
istry 1996, 35, 5175-5182.
(16) Collman, J . P.; Groh, S. E. “Mercaptan-Tail” Porphyrins: Syn-
thetic Analogues for the Active Site of Cytochrome P-450. J . Am.
Chem. Soc. 1982, 104, 1391-1403.
(17) Braunschweiler, L.; Ernst, R. R. Coherence Transfer by Isotropic
Mixing: Application to Proton Correlation Spectroscopy. J .
Magn. Reson. 1983, 53, 521-528.
Ack n ow led gm en t. We gratefully acknowledge Drs.
J ohn Domagala, Donald Hupe, David Mack, David
Moreland, and Peter Tummino for numerous insightful
discussions and the assistance of Dr. Scott Buckel and
Dana DeJ ohn in the proteolysis experiments.
Su p p or tin g In for m a tion Ava ila ble: Tabulated chemical
shifts for the aromatic and amide protons of 2, 3, and 7 (1
page). Ordering information is given on any current masthead
page.
Refer en ces
(18) Bax, A.; Davis, D. G. MLEV-17-Based Two-Dimensional Homo-
nuclear Magnetization Transfer Spectroscopy. J . Magn. Reson.
1985, 65, 355-360.
(19) Kumar, A.; Ernst, R. R.; Wu¨thrich, K. A two-dimensional nuclear
Overhauser enhancement (2D NOE) experiment for the elucida-
tion of complete proton-proton cross-relaxation networks in
biological macromolecules. Biochem. Biophys. Res. Commun.
1980, 95, 1-6.
(20) Marion, D.; Wu¨thrich, K. Application of phase sensitive two-
dimensional correlated spectroscopy (COSY) for measurements
of proton-proton spin-spin coupling constants in proteins. Bio-
chem. Biophys. Res. Commun. 1983, 113, 967-974.
(21) Loo, J . A.; Ogorzalek Loo, R. R.; Andrews, P. C. Primary to
quaternary protein structure determination with electrospray
ionization and magnetic sector mass spectrometry. Org. Mass
Spectrom. 1993, 28, 1640-1649.
(22) Loo, J . A.; Pesch, R. Sensitive and selective determination of
proteins with electrospray ionization magnetic sector mass
spectrometry and array detection. Anal. Chem. 1994, 66, 3659-
3663.
(23) Fenn, J . B.; Mann, M.; Meng, C. K.; Wong, S. F.; Whitehouse,
C. M. Electrospray ionization for mass spectrometry of large
biomolecules. Science 1989, 246, 64-71.
(1) Me´ly, Y.; Cornille, F.; Fournie-Zaluski, M. C.; Darlix, J . L.;
Roques, R. P.; Gerard, D. Investigation of Zinc-Binding Affinities
of Moloney Murine Leukemia Virus Nucleocapsid Protein and
its Related Zinc Finger and Modified Peptides. Biopolymers
1991, 31, 899-906.
(2) Summers, M. F.; Henderson, L. E.; Chance, M. R.; Bess, J . W.,
J r.; South, T. L.; Blake, P. R.; Sagi, I.; Perez-Alvarado, G.;
Sowder, R. C., III; Hare, D. R.; Arthur, L. O. Nucleocapsid Zinc
Fingers Detected in Retroviruses: EXAFS Studies of Intact
Viruses and the Solution-State Structure of the Nucleocapsid
Protein From HIV-1. Protein Sci. 1992, 1, 563-574.
(3) Yu, X.; Hathout, Y.; Fenselau, C.; Sowder, R. C., II; Henderson,
L. E.; Rice, W. G.; Mendeleyev, J .; Kun, E. Specific Disulfide
Formation in the Oxidation of HIV-1 Zinc Finger Nucleocapsid
p7. Chem. Res. Toxicol. 1995, 8, 586-590.
(4) Domagala, J . M.; Bader, J . P.; Gogliotti, R. D.; Sanchez, J . P.;
Stier, M. A.; Song, Y.; Vara Prasad, J . V. N.; Tummino, P. J .;
Harvey, P.; Holler, T. P.; Gracheck, S.; Hupe, D.; Rice, W. G.;
Schultz, R. The Structure Activity Relationships for a New
Class of Anti HIV-1 Agents Targeted Toward Nucleocapsid
Protein NCp7: The 2,2′-Dithiobisbenzamides. Unpublished
manuscript.
(5) Tummino, P. J .; Scholten, J .; Harvey, P.; Holler, T. P.; Maloney,
L.; Gogliotti, R.; Domagala, J . M.; Hupe, D. The In Vitro Ejection
of Zinc From Human Immunodeficency Virus Type 1 Nucleo-
capsid Protein by Disulfide Benzamides With Cellular Anti-HIV
Activity. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 969.
J M960253W