Discovery of non-carbohydrate inhibitors of aminoglycoside-modifying enzymes

Article abstract of DOI:10.1016/j.bmc.2005.06.059

Chemical modification and inactivation of aminoglycosides by many different enzymes expressed in pathogenic bacteria are the main mechanisms of bacterial resistance to these antibiotics. In this work, we designed inhibitors that contain the 1,3-diamine ph

Full text of DOI:10.1016/j.bmc.2005.06.059

Bioorganic & Medicinal Chemistry 13 (2005) 6252–6263
Discovery of non-carbohydrate inhibitors
of aminoglycoside-modifying enzymes
a
Karen T. Welch, Kristopher G. Virga, Neil A. Whittemore, Can Ozen,
b
a
a,c
¨
Edward Wright,a Cynthia L. Brown,a Richard E. Leeb and Engin H. Serpersua,*
aDepartment of Biochemistry, Cellular and Molecular Biology and the Center of Excellence for Structural Biology,
University of Tennessee, Walters Life Sciences Building M407, Knoxville, TN 37996-0840, USA
bDepartment of Pharmaceutical Science, University of Tennessee Health Science Center, 847 Monroe Avenue Room 327,
Memphis TN 38163, USA
cGraduate School of Genome Science and Technology, University of Tennessee, Knoxville, TN 37996, USA
Received 6 January 2005; revised 24 June 2005; accepted 24 June 2005
Available online 1 September 2005
Abstract—Chemical modification and inactivation of aminoglycosides by many different enzymes expressed in pathogenic bacteria
are the main mechanisms of bacterial resistance to these antibiotics. In this work, we designed inhibitors that contain the 1,3-di-
amine pharmacophore shared by all aminoglycoside antibiotics that contain the 2-deoxystreptamine ring. A discovery library of
molecules was prepared by attaching different side chains to both sides of the 1,3-diamine motif. Several of these diamines showed
inhibitory activity toward two or three different representative aminoglycoside-modifying enzymes (AGMEs). These studies yielded
the first non-carbohydrate inhibitor N-cyclohexyl-N0-(3-dimethylamino-propyl)-propane-1,3-diamine (Compound G,H) that is com-
petitive with respect to the aminoglycoside binding to the enzyme aminoglycoside-200-nucleotidyltransferase-Ia (ANT(200)). Another
diamine molecule N-[2-(3,4-dimethoxyphenyl)-ethyl]-N0-(3-dimethylamino-propyl)-propane-1,3-diamine (Compound H,I) was
shown to be a competitive inhibitor of two separate enzymes (aminoglycoside-30-phosphotransferase-IIIa (APH(30)) and ANT(200))
with respect to metal–ATP. Thermodynamic and structural-binding properties of the complexes of APH(30) with substrates and
inhibitor were shown to be similar to each other, as determined by isothermal titration calorimetry and NMR spectroscopy.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
some.5,6 Three types of covalent modifications: acetyla-
tion, nucleotidylation, and phosphorylation reactions
are achieved by many different aminoglycoside-modify-
ing enzymes (AGMEs).7,8 Thus, inhibition of AGMEs
may allow aminoglycosides to regain their antibiotic
efficacy.
The therapeutic use of aminoglycoside antibiotics
(Fig. 1) began in the mid-1940s when streptomycin, iso-
lated from Streptomyces griseus, was developed as the
first effective treatment for tuberculosis.1,2 Since then,
many new aminoglycosides have been isolated and syn-
thetic derivatives prepared. These antibiotics are the
weapons of choice against certain gram-negative bacilli,
including Pseudomonas. However, as with many other
antibiotics, resistance to these drugs has developed and
now compromises their usage.3,4 Resistance to amino-
glycosides is caused primarily by plasmid-encoded en-
zymes that covalently modify certain functional groups
and prevent the aminoglycosides from interacting with
their target—the small subunit of the bacterial ribo-
Carbohydrate-based inhibitors of AGMEs, such as
those resulting from the removal of selected hydroxyl
or amino groups of the aminoglycoside scaffold, are
known. More recently, various inhibitors based on
neamine dimers,9,10 glycodiversification strategy,11 pseu-
do-pentasaccharide analogs of neomycin,12 amikacin
derivatives,13,14 2-deoxystreptamine (DOS) derivatives,15
and cyclic aminoglycosides16 have been prepared. Fewer
efforts are directed toward non-carbohydrate aminogly-
coside mimetics, such as small-molecule RNA binders,17
benzimidazoles,18 and cationic peptides.19
Keywords: Aminoglycosides; Drug design; Inhibitors; Bacterial
resistance.
Most clinically utilized aminoglycosides have a DOS as
their aglycone. In general, these aminoglycosides are
*
Corresponding author. Tel.: +1 865 974 2668; fax: +1 865 974
6306; e-mail: Serpersu@utk.edu
0968-0896/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2005.06.059

K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
6253
C ('')
A (')
C ('')
A (')
OH
HO
HO
OH
OH
H N
2
OH
HO
O
OH
O
B (unprimed)
B (unprimed)
O
HO
H N
O
NH
2
O
HO
2
NH
2
OH
O
H N
2
HO
O
O
OH
O
NH
HO
2
H N
2
NH
2
OH
H N
HO
OH
2
NH
2
HO
OH
O
OH
OH
O
HO
OH
OH
O
HO
HO
H N
2
O
OH
H N
2
NH
2
O
OH
HO
OH
O
O
O
O
HO
O
HO
NH
2
H
NH
O
NH
H N
2
2
HO
O
NH
OH
2
O
H
H N
2
HO
H N
2
O
NH
2
OH
OH
HO
HO
H N
O
H N
2
O
2
O
NH
2
OH
NH
2
HO
O
O
O
O
O
NH
2
HO
H N
2
NH
2
HO
H N
2
Figure 1. Aminoglycosides with 4,6-(left panel) and 4,5-(right panel) disubstituted 2-deoxystreptamine ring. Left panel (from top): kanamycin A,
amikacin, and tobramycin. Right panel (from top): ribostamycin, paromomycin, and lividomycin A. Letter and prime notations for the first three
rings are also shown on top.
4,5- or 4,6-substituted at the DOS ring and can be
pseudotri-, tetra-, or pentasaccharides (Fig. 1). Amino-
glycosides with a 4,5- or 4,6-substituted DOS ring can
also be classified as neomycin- and kanamycin-type anti-
biotics, respectively. In this work, we selected the sim-
plest common pharmacophore to aminoglycosides
bearing a 2-DOS aglycone: the 1,3 diamine motif
(Fig. 2) and designed an exploratory library of non-car-
bohydrate aminoglycoside analogs to evaluate the po-
tential of this functional group in producing molecules
that recognize, bind, and inhibit multiple AGMEs.
These analogs contain a 1,3-diamine unit that is
flanked by a diverse array of side-chain functionality.
This proof-of-concept study yielded the first non-
carbohydrate inhibitor that is competitive with the
aminoglycoside substrates of an aminoglycoside
nucleotidyltransferase (the aminoglycoside-200-nucleotid-
yltransferase-Ia ANT(200)).
2. Results and discussion
The diamines were prepared by the matrix synthesis
using a one-pot reaction (Scheme 1). Two equivalents
of primary amines A–I (Fig. 3) were reacted with one
equivalent of 1,3-dichloropropane in a solvent-free reac-
tion at an elevated temperature for 48 h20 Two SN2 reac-
tions took place to yield one or three diamines, based
upon starting materials, with the product secondary
amines salting out as dihydrochlorides. Thus, in one
step, we were quickly able to produce 45 compounds
containing the 1,3-x motif. The reaction mixtures were
easily deconvoluted by testing three related samples,
that is, A,A; A,B; and B,B. The chemistry of this syn-
thetic route was complicated by the tendency of amine
H to undergo side reactions. Amine H contains a tertia-
ry nitrogen atom that can act as an internal strong base
to desalt the product secondary amine and allow it to re-
act with another molecule of 1,3-dichloropropane. This
process leads to the formation of undesired high molec-
ular weight adducts. Several test wells containing amine
H were active in our primary screen; therefore, an alter-
nate synthetic route was needed to validate hits resulting
from these wells.
The alternate synthetic devised route is shown in Scheme
2. The first amine is added to acryloyl chloride, yielding
an a,b-unsaturated amide containing one of the desired
side-chains. The second amine is subsequently added by
a Michael addition to the a,b-unsaturated amide by the
Figure 2. The 1,3 diamine motifs of kanamycin A (top) and the
compound H,I (bottom).

6254
K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
Scheme 1. Synthesis of compounds A,A–I,I. (a) R-NH2 (amines A–I); (b) R0-NH2 (amines A–I).
Figure 3. Diamine compounds that showed inhibitory activity to one or more AGME.
Scheme 2. Synthesis of compounds A,H; G,H; and H,I. Reagents: (a) R-NH2, pyridine, DCM; (b) R0-NH2, EtOH; (c) THFÆBH3 complex, THF; and
(d) 6 N HCl, MeOH.
second amine. The resulting compound is purified by
flash chromatography and then reduced with borane in
tetrahydrofuran to afford the desired diamine. One
advantage of this synthesis over the previous one is the
greater control of reactivity. Another is the production
of a single diamine, as opposed to a mixture of up to
three compounds.
phosphotransferase-IIIa (APH(30)) transfers a phos-
phate group from ATP to the 30-hydroxyl group of
appropriately functionalized aminoglycosides. The
200-nucleotidyltransferase-Ia (ANT(200)) catalyzes the
transfer of adenosine monophosphate from ATP to
the 200-hydroxyl group of appropriately functionalized
aminoglycoside antibiotics. Aminoglycoside 3-acetyl-
transferase-IIIb (AAC(3)) catalyzes the transfer of an
acetyl group from acetyl CoA to the 3-amino group of
aminoglycosides containing a 2-deoxystreptamine ring.
Three representative AGMEs7 were chosen to evaluate
the activity of these compounds. Aminoglycoside 30-

K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
6255
These enzymes represent three distinct covalent modifi-
cations, two different co-substrates (Metal–ATP with
nucleotidyl- and phosphotransferases and acetylCoA
with the acetyltransferase), and different levels of amino-
glycoside substrate specificity. Moreover, they also vary
in the distance between the site of covalent modification
and the 1,3-diamine unit of the 2-deoxystreptamine ring.
scale and purity (Scheme 2). These compounds were
tested to determine the nature of inhibition with
APH(30) and ANT(200). The results of these assays indi-
cate a difference in the action of the two compounds.
Compound G,H is competitive with the aminoglycoside
substrate for ANT(200) but not for APH(30). Data shown
in Figure 4A yielded a Ki of 540 32 lM for this
inhibition.
Enzymatic screening of the synthetic diamines proceed-
ed as follows: all crude reaction mixtures were screened
for inhibition of aminoglycoside modification with
APH(30). A smaller set of samples, more than half of
the total number of those tested with APH(30), were
screened with AAC(3). From these, a subset of com-
pounds were chosen for inhibitory activity and ease of
purification, and these compounds were HPLC purified
and retested with AAC(3), ANT(200), and APH(30). The
most potent inhibitor of all three enzymes was tested to
determine the type of inhibition. Percent inhibition for
diamines that inhibit more than one AGME is shown
in Table 1.
2.1. Compound G,H is the first non-carbohydrate inhibitor
of an AGME that is competitive with the aminoglycoside
substrate
Conversely, compound H,I is competitive with metal–
ATP for both enzymes that require it for catalysis.
The Ki values for H,I are comparable—160 30 lM
for APH(30) and 125 24 lM for ANT(200). Competi-
tive inhibition of ANT(200) by H,I is shown in Figure
4B. Due to the low solubility of AAC(3) and ANT(200)
and availability of X-ray structure for APH(30), further
structural studies were performed with APH(30) and
complexes of APH(30) with ATP or compound H,I by
isothermal titration calorimetry (ITC), NMR, and dock-
ing calculations. These studies are described in the
following paragraphs.
The AGME that was most affected by these compounds
was AAC(3). Eighteen of the 45 compounds tested were
inhibitors of this enzyme. Compounds containing amine
H (3-(dimethylamino)propylamine) were inhibitors of
AAC3, decreasing the rate of catalysis to less than
30% of the uninhibited rate.
In contrast, a smaller number of compounds retarded
the rate of catalysis of APH(30). There was not as strong
a trend in the chemical structures of the active com-
pounds as there was for AAC(3); however, all com-
pounds that inhibited APH(30) contained amine H
(3-(dimethylamino)propylamine), amine I (3,4-dimeth-
oxyphenethylamine), or both. The common theme for
inhibitors of APH(30) was an aromatic or cyclohexyl
ring plus a dimethylamino group or diol.
Even fewer among these compounds seem to be inhibi-
tors of ANT(200). Only five of the purified diamines test-
ed showed significant inhibition of this enzyme. Similar
to APH(30), all the compounds that inhibit ANT(200)
contain either amine H, amine I, or both.
Compounds G,H and H,I were the most potent com-
pounds identified by the primary screen and were subse-
quently resynthesized by the second route in greater
Table 1. Inhibition of enzymatic activity by ꢀ0.5 mM diamine
compounds (% inhibition with respect to full activity. Errors ꢀ10%)a
Diamine
APH(30)
AAC(3)
ANT(200)
B,I
18
13
13
21
21
28
54
21
58
27
62
68
77
63
23
28
16
0
C,H
C,I
D,H
F,H
G,H
H,I
Figure 4. Inhibition of enzymatic activity of ANT(200) by compounds
G,H (A) and H,I (B). (A) Competitive inhibition with respect to
tobramycin: (j) No inhibitor, (m) 0.625 mM, (.) 1.25 mM, and (Ç)
2.5 mM G,H. (B) Competitive inhibition with respect to MgATP: (j)
No inhibitor, (m) 0.625 mM, and (.) 1.25 mM H,I. In both plots, the
rate is shown as the change in absorbance per unit time and data are
shown with linear regression lines.
0
22
42
a Substrate and enzyme concentrations used were: 50 lM kanamycin A
with 2 l APH(30) or 1 lM AAC(3) and 100 lM tobramycin with
0.24 lM ANT(200).

6256
K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
Table 2. ITC-derived thermodynamic parameters for CaATP and H,I
Binding to APH(30)-IIIaa
Table 3. ITC-derived thermodynamic parameters for kanamycin A
binding to APH(30)-IIIa and its complexes with CaATP and H,Ia
N
KD (lM) DH
ꢁTDS
DG
(kcal/mol) (kcal/mol) (kcal/mol)
N
KD
DH
ꢁTDS
DG
(lM) (kcal/mol) (kcal/mol) (kcal/mol)
APH(30)-IIIa
APH(30)-IIIa– 0.9 1.0
CaATP
APH(30)-IIIa– 0.9 1.3
H,I
0.8 5.4
ꢁ31.9
ꢁ11.3
24.4
2.8
ꢁ7.5
ꢁ8.5
CaATP 0.93 35
H,I
ꢁ3.5
ꢁ8.6
ꢁ2.9
ꢁ6.4
ꢁ6.1
0.98 47
2.5
a Determined at 303 K. Standard errors: N, 15%; KD, 10%; DH, 8%;
TDS, 16%; and DG, 1%.
ꢁ9.4
1.2
ꢁ8.2
a Determined at 310 K. Standard errors: N, 9%; KD, 8%; DH, 7%; TDS,
16%; and DG; 1%.
The use of isothermal titration calorimetry allowed for
a direct thermodynamic characterization of various
complexes of APH(30), substrates, and inhibitor. Bind-
ing of CaATP and compound H,I to APH(30) takes
place with similar affinity (Table 2). Binding of both
compounds to APH(30) shows a very similar free ener-
gy change; however, the binding of compound H,I to
the enzyme is entropically disfavored, which is com-
pensated by a large enthalpic term. This is not unex-
pected since compound H,I is much more flexible in
solution than free ATP and loss of rotational and
translational freedom of molecule on binding to the
enzyme is entropically more disfavored than it is with
ATP. Similarities in the binding of compound H,I and
CaATP also extend to the ternary enzyme–CaATP/
H,I–kanamycin A complexes. As shown in Figure 5,
titration of APH(30)–CaATP or APH(30)–H,I with
kanamycin A showed that affinity of kanamycin A
to both complexes was identical and was also en-
hanced when compared to its affinity to the apoen-
zyme (Table 3). These observations are consistent
with the complexes of APH(30)–H,I and APH(30)–
CaATP having similar active site organizations. When
the competitive nature of compound H,I inhibition
with respect to ATP is considered, it is possible that
APH(30) active site undergoes similar conformational
changes in the complexes of the enzyme with H,I
and CaATP.
Further evidence for similar binding patterns of a met-
al–ATP complex and compound H,I was obtained by
NMR spectroscopy. Figures 6 and 7 show a set of
15N–1H HSQC spectra obtained by using a uniformly
15
enriched N-APH(30). The spectra acquired with the
binary complexes of enzyme–CaATP or enzyme–H,I
are similar to each other but different from that acquired
with the enzyme–kanamycin A complex. Figure 6 shows
the full spectra obtained with apoenzyme, enzyme–
CaATP, enzyme–H,I, and enzyme–kanamycin A com-
plexes. These spectra also showed that the enzyme is
quite flexible in solution, even with bound CaATP or
H,I, and some of the resonances of this 31 kDa protein
are broadened under these conditions. This is not sur-
prising because this enzyme can modify many structural-
ly different aminoglycosides; hence, it must have a
flexible structure in solution to accommodate these sub-
strates. Contrary to this, a significant increase in the dis-
persion of cross-peaks is visible in the spectrum of the
binary enzyme–kanamycin A complex, suggesting the
formation of a ꢀfirmerꢁ complex with this substrate. This
Figure 5. Titration of APH(30) (A), APH(30)–CaATP (B), and APH(30)–H,I (C) with kanamycin A. Top panels show the raw ITC data, while the
bottom panels show the best fits that yielded parameters shown in Table 2.

K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
6257
Figure 6. N–1H HSQC spectra acquired with APH(30) alone (A), APH(30)–kanamycin A (B), APH(30)–CaATP (C), and APH(30)–H,I (D).
15
is in concert with the observed dissociation constants of
the binary enzyme–ligand complexes, which indicate
that kanamycin A has the tightest binding to the en-
zyme. In Figure 7, the region of each spectrum showing
the tryptophan side-chain nitrogen–proton correlations
is displayed as an expanded plot. Of the five trypto-
phanes of APH(30), only W159 is located near the ami-
noglycoside-binding site,21 while the ATP-binding
region of APH(30) does not contain any tryptophan res-
idues. Therefore, tryptophan resonances can be used as
probes to confirm the presence or absence of a substrate
in the aminoglycoside-binding site; the binding of ATP
(or metal–ATP) is not expected to cause any changes
in chemical shifts of tryptophan residues. The expanded
regions clearly show that, as expected, the spectra of
binary enzyme–CaATP and enzyme–H,I complexes are
similar to that obtained with the apoenzyme. However,
this region is significantly different in the binary en-
zyme–kanamycin A spectrum, indicating changes in
the environment of W159 and other tryptophan resi-
dues. These observations are again consistent with the
binding of compound H,I to the ATP site.
of A93. A hydrogen bond with the backbone NH of
A93 is also important for ATP binding to APH(30);
and a similar interaction has been found for isoquinoline
sulfonamide kinase inhibitors bound to protein ki-
nases22 The hydrogen bond between H,I and A93 pro-
motes stacking of the 3,4-dimethoxyphenyl ring in an
antiparallel fashion with the side-chain of Y42. In addi-
tion, there are three potential interactions between the
amino groups of compound H,I and the active site of
APH(30). These interactions take place, with D22, E24,
and D190. This final contact provides an indication of
the relatively larger size of H,I as compared to ATP,
D190 is part of the aminoglycoside subpocket and is be-
lieved to act as the catalytic base.23 Figure 9 shows a
structural comparison of the diamines containing an
aromatic ring and amine H. Thereby offering an expla-
nation for the observed lower levels of inhibition affor-
ded by compounds A,H (green); C,H (blue); and F,H
(orange). These compounds would not be able to main-
tain contact with Y42 by aromatic stacking and one or
more of the charge-based contacts simultaneously. Sim-
ilarly, although C,H and F,H contain hydrogen bond
acceptor groups, they are not situated in such a way
as to form the same network of enzyme–ligand interac-
tions available to H,I (shown in red).
Docking calculations were used to predict the conforma-
tion of compound H,I in the active site of APH(30), as
well as to predict the contacts made by H,I in the active
site. The docked conformation of compound H,I in the
active site of APH(30) is shown in Figure 8. Compound
H,I contains a set of key features that are required for
binding to the nucleotide-binding pocket of APH(30).
First, one of the methoxy groups on the aromatic ring
forms a hydrogen bond with the backbone NH group
Figure 10 shows the requirements for binding of diam-
ines to three different AGMEs based on 55 compounds
tested in this work. It appears that the amine H moiety
(colored blue for all three structures) is a key feature for
recognition by these enzymes, as the most active
diamines against all three enzymes contain amine H.

6258
K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
Figure 7. Expanded regions of the HSQC spectra shown in Figure 6; apoenzyme (A), APH(30)–kanamycin A (B), APH(30)–CaATP (C), and
APH(30)–H,I (D).
Figure 8. The compound H,I docked into the ATP-binding site of APH(30) (A). Close-up of the ATP-binding site with docked H,I (B). This figure
was created with PyMol. (DeLano, W. L. The PyMOL Molecular Graphics System (2002) DeLano Scientific, San Carlos, CA, USA. http://
www.pymol.org.)
Compound H,I (top) is competitive with ATP for
APH(30) and ANT(200). The proposed structure of H,I
in the complex with APH(30) suggests that binding to
the ATP site of APH(30) is mediated by one or more
of the amino groups along with a hydrogen bond
between the phenoxy hydrogen bond donor and A93,
and a possible aromatic stacking interaction. Com-
pounds A,H, C,H, and F,H, which lack a full comple-
ment of enzyme–ligand interactions available to H,I,
are not competitive with ATP when assayed with
APH(30). We note that conclusions about the structural
requirements for binding of inhibitors to these three en-
zymes are based on a small library of compounds. Thus,
they may represent only a small fraction of structural
units that can be recognized by these enzymes.
The relationship between inhibitory activity of diamine
compounds and multiple amino groups is not surprising,

K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
6259
control for the wells containing the ANT(200) E. coli
transformants.
3. Conclusions
A study directed at finding novel, non-carbohydrate
inhibitors of aminoglycoside-modifying enzymes yielded
several molecules that inhibit multiple aminoglycoside-
modifying enzymes. A stepwise synthesis was developed
to prepare larger amounts of the most active com-
pounds. Multiple experimental techniques, including
ITC and NMR, were used to characterize inhibitor
binding to one of the AGMEs.
Figure 9. Superpositioning of compounds A,H (green), C,H (blue),
F,H (orange), and H,I (red) docked into the active site of aminogly-
coside-30-phosphotransferase-IIIb.
Of the diamines tested, compound H,I has been identi-
fied as an inhibitor of all three AGMEs studied and is
a competitive inhibitor of ATP binding for two of the
enzymes (APH(30) and ANT(200)). It binds to APH(30)
as in a similar fashion to ATP, as evidenced by compar-
ison of the APH(30)–CaATP–kanamycin
A
and
APH(30)–H,I–kanamycin A ternary complexes using
ITC and NMR. The presence of a hydrogen bond donor
and positively charged nitrogen atoms are predicted to
be the key factors in binding. This pattern is reminiscent
of the isoquinoline sulfonamide class of protein kinase
inhibitors.22
A common theme for the active compounds is the pres-
ence of amine H, which contains a propyldimethylamino
side-chain. This functionality is reminiscent of amino-
glycosides, which contain multiple amino groups. De-
spite the fact that compound G,H lacks a hydrogen
bond donor on the cyclohexyl ring, it inhibited all three
enzymes tested and is the first competitive inhibitor of
ANT(200) with respect to the aminoglycoside substrate.
Thus, compound G,H demonstrated the potential of this
conceptually simple approach to develop non-carbohy-
drate aminoglycoside analogs. This compound will be
used as a starting point for the design of more potent,
non-carbohydrate inhibitors of AGMEs.
Figure 10. Requirements for binding of diamines to aminoglycoside-
modifying enzymes. Shared motifs among the three different enzymes
are shown in blue.
4. Materials and methods
4.1. General
however, as the amino groups of aminoglycosides are
key recognition elements for RNA, as well as AGMEs.
What is somewhat surprising is the wide variety of func-
tionality that was tolerated in the second side-chain.
Specifically, compound G,H, which inhibited all three
enzymes, has neither hydrogen bond donors nor polar
groups of any kind on the amine G side-chain. Com-
pound G,H is a competitive inhibitor of ANT(200)
with respect to the aminoglycoside substrate
(Ki = 540 32). Thus, compound G,H highlighted the
differences in the aminoglycoside-binding pockets of
APH(30) and ANT(200). Interestingly, compound G,H
was also the most potent inhibitor of in vivo growth
of Escherichia coli AGME transformants. Compound
G,H inhibited cell growth for all three transformants
at both concentrations. At 10 mM, growth was 50% to
75% that of the control wells for the E. coli transfor-
mants producing AAC(3) and APH(30). Cell growth
was inhibited to 30% of the absorbance of the growth
All chemicals used in enzyme preparation and assays
were purchased from Sigma (St. Louis, MO) unless
otherwise noted. Deuterated solvents were purchased
from Cambridge Isotope Laboratories (Andover, MA).
All starting materials for synthesis of diamines were
purchased from Aldrich Chemical Company (Milwau-
kee, WI). All reagent-grade solvents used for purifica-
tion of the diamines were purchased from Fischer
Scientific (Suwanee, GA). Product purification was
performed by HPLC on a Gilson 215 HPLC (Middle-
town, WI) equipped with an Xterraꢁ Prep MS C18
5 lm, 19 · 100 mm column, Waters (Milford, MA).
Select product identifications and characterizations
were performed first by ESI MS on a Bruker Esquire
LCMS, Bruker Daltonics (Billerica, MA). Further

6260
K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
characterization of select compounds was performed
by 1H NMR recorded on a Varian INOVA-500
NMR spectrometer. 1H–15N HSQC TROSY spectra
of various enzyme–ligand complexes were recorded
of 50–400 lM, while those of ATP in the range of 25–
400 lM.
AAC(3)—The assays were performed according to the
procedure used to determine the activity of other acetyl-
transferases28,29 with modifications described in Owston
and Serpersu.26 The assay was based on the determina-
tion of product pyridine-4-thiolate formed as the
reaction product CoA reacted with 4,40-dithio-dipyri-
dine, present in the assay mixture. Thus, the amount
of CoA formation was continuously monitored by
measuring the absorbance at 324 nm using an extinction
coefficient of 19,800 Mꢁ1 cmꢁ1 for the product pyridine-
4-thiolate.
on
equipped with triple resonance probe, using uniformly
a
Varian INOVA-600 NMR spectrometer,
15
enriched N-APH(30).
4.2. Protein preparation and purification
APH(30) was overexpressed in E. coli, isolated, and puri-
fied according to the procedure described by McKay
et al.,24 with the modifications described in Ozen and
¨
Serpersu.25 Briefly, cells were grown in Luria broth
(LB) media in New Brunswick (Edison, NJ) BioFlo
110 fermentor vessel at 37 ꢂC. Cells were harvested five
hours after induction with 1 mM IPTG and were then
lysed by passage through French press in 50 mM Tris,
5 mM EDTA, 200 mM NaCl, 1 mM PMSF, and
0.2 mM DTT solution. Cell debris was removed by cen-
trifugation and the supernatant was diluted with buffer
A (50 mM Tris, 1 mM EDTA), pH 8.0, before loading
it onto a strong anion exchanger POROS 20HQ column
(250 · 4.6 mm) attached to BIOCAD 700E perfusion
chromatography workstation from Applied Biosystems.
A linear gradient of 0–20% with buffer B in buffer A was
applied through five column volumes to elute contami-
nant proteins. The pure APH(30)-IIIa was then eluted
with two column volumes of 20% buffer B in buffer A.
Pooled fractions were dialyzed against buffer A, freeze-
dried, and stored at ꢁ80 ꢂC. AAC(3)-IIIb and
ANT(200)-Ia were also overexpressed in E. coli, isolated,
and purified according to the procedures described
earlier,26,27 respectively.
ANT(200)—The assays were performed according to the
procedure reported from this laboratory.27,30 The assay
was based on the determination of inorganic phosphate
after hydrolysis of the reaction product pyrophosphate
by a pyrophosphatase enzyme. For the kinetic assays,
the concentrations of ATP were in the range of 250–
800 lM.
4.5. General procedure for preparation of diamines
A,A–I,I in parallel
Synthesis of diamine compounds A,A–I,I was carried
out in a parallel array in 13 · 100 mm screw-capped
tubes, whereby nine primary amines were selected and
each given a letter designation A–I, respectively. This ar-
ray resulted in a total of 45 individual reactions using a
1:1:1 ratio of reactants, where a 1 mmol of each of two
primary amines was reacted with 1 mmol of 1,3-dichlo-
ropropane (95 lL). The labeling, name, and quantity
of each amine used were as follows: (A) 2-Chlorobenzyl-
amine (121 lL); (B) 3-amino-1,2-propanediol (78 lL);
(C) 2-(aminomethyl)pyridine (103 lL); (D) 2-amino-1-
butanol (94.5 lL); (E) furfurylamine (88.4 lL); (F)
2-(2-aminoethoxy)ethanol (100 lL); (G) cyclohexyl-
amine (114.4 lL); (H) 3-(dimethylamino)propylamine
4.3. Preparation of diamine samples for testing
Most of the samples prepared from crude reaction mix-
tures were dissolved in 50% DMSO–water. All samples
that were not soluble under these conditions were dis-
solved in 100% DMSO. In the first round enzymatic
screen (reaction mixtures), the concentrations of the dia-
mine samples ranged from 0.2 to 3 mM. In enzymatic
assays of purified diamines, the concentration of dia-
mine used in the assay was 0.625–2.5 mM depending
upon the enzyme tested.
(126 lL);
and
(I) 3,4-dimethoxyphenethylamine
(169 lL). The disubstitution reactions were carried out
under neat conditions at 100 ꢂC for 48 h. The reaction
mixtures were gently agitated by hand, intermittently
to ensure complete mixing of all reactants. Included as
part of the 45 reactions, compounds A,A; B,B; C,C;
D,D; E,E; F,F; G,G; H,H; and I,I were all synthesized.
This was done to provide a control for those com-
pounds, as it was realized that each of those compounds
would be formed as a product homologous addition in
each of the reactions for which they were involved. Ini-
tially, these reaction mixtures were tested in crude form.
The most active compounds from the initial screen (B,I;
C,H; C,I; D,H; F,H; G,H; and H,I) were then partially
purified by HPLC and again screened for enzyme inhibi-
tion activity (Table 1).
4.4. Enzyme assays
Inhibition pattern of enzymes by diamines was deter-
mined by Lineweaver–Burk plots of data (Fig. 4). The
lines were fitted by the least-squares analysis. Below
are given the assay conditions for each enzyme.
APH(30)—The assay for ATP-dependent phosphoryla-
tion of kanamycin A was performed, as described in
McKay et al.24 The assays were performed at 27 ꢂC.
The assay mix was adjusted to contain 10% hexylene
glycol for compounds with low water solubility. Enzy-
matic activity of APH(30) was not affected in the pres-
ence of 10% hexylene glycol. For the kinetic assays,
the concentrations of kanamycin A were in the range
4.6. General procedure for discrete preparation of
diamines A,H; G,H; and H,I
A discrete synthesis for compounds A,H; G,H; and H,I
was carried out to provide compounds as trihydrochlo-
ride salts.

K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
6261
Step 1 (formation of acrylamides). The reactions were
carried out in three discrete, previously dried 100 mL
reaction vessels. To a stirred solution of acryloyl chlo-
ride (4.9 mmol, 400 lL for compound A,H and
11 mmol, 894 ll for compounds G,H and H,I) in dichlo-
romethane (50 mL) at ꢁ70 ꢂC was added dropwise each
of the three amines (A 4 mmol, 500 ll; G 10 mmol,
1.66 mL; or I 10 mmol, 1.14 mL) to one of each of the
reaction vessels, followed by the addition of pyridine
(12 mmol, 970 lL for compound A,H and 30 mmol,
2.43 mL for G,H and H,I) to each of the reaction ves-
sels. The reaction mixtures were allowed to slowly warm
up to ambient temperature and stirred for 12 h. Each
reaction mixture was then washed with aqueous NaH-
CO3 three times, followed by washing once with saturat-
ed brine. The organic layers from each reaction mixture
were then dried over anhydrous Na2SO4 and evaporated
in vacuo to complete dryness. Each of the acrylamide
intermediates was obtained in approximately 80–90%
yield based on crude weight. The compounds produced
were sufficiently purified by TLC and NMR analysis to
be used directly for the next reaction with no further
purification.
J = 7.1 Hz, 2H), 2.11 (s, 6H), 2.11–2.10 (br s, 1H),
1.49 (qin, J = 7.1 Hz, 2H); (ESI, Pos.) m/z 338.6
(M+H)+.
Step 3 (reduction of the amides). Each of the corre-
sponding amides (840 mg, 2.8 mmol A,H; 970 mg,
3.8 mmol G,H; and 1.89 g, 5.6 mmol H,I) was dissolved
in THF (15 mL for compound A,H and 40 mL for com-
pounds G,H and H,I). To this, a solution of THF/BH3
complex was added (6 mmol for compound A,H and
18 mmol for compounds G,H and H,I) and the reactions
were refluxed for 48 h. Note. The reduction of the amide
intermediate for compound A,H was attempted first.
The reduction procedure for A,H was repeated a second
time, as it was realized that the reduction was incom-
plete. For this reason, a large excess of THF/BH3 com-
plex was used (3–5 equiv) in the subsequent two amide
reductions for compounds G,H and H,I. The reaction
mixtures were cooled to ambient temperature and then
quenched with the drop wise addition of 6 N HCl in
methanol. The reaction mixtures were allowed to stir
for an additional 3 h, and then evaporated to dryness
in vacuo. Each of the dried reaction mixtures was redis-
solved in acidic methanol (50 mL) and again evaporated
to dryness. The products were obtained as white crystal-
line solids by recrystallization from ethanol/chloroform
(1:3).
Step 2 (Michael addition of amine H). Each of the acryl-
amide intermediates was dissolved in ethanol (5 mL for
compound A,H and 10 mL for compounds G,H and
H,I) and transferred to three discrete Radleyꢁs carousel
reaction vessels. Following it, three equivalents of amine
H were added to each of the reaction vessels (9.44 mmol,
1.2 mL for compound A,H and 27 mmol, 3.4 mL for
compounds G,H and H,I) and the reaction mixtures
were stirred at 80 ꢂC for 48 h. They were then evaporat-
ed in vacuo under high heat and pressure to remove eth-
anol and excess amine H. The oily residues were purified
by flash chromatography using a linear gradient from
10% to 50% methanol in chloroform.
4.6.4. N-[3-(2-Chlorobenzylamino)-propyl]-N0,N0-dimeth-
yl-propane-1,3-diamine trihydrochloride (A,H). The
product A,H (323 mg, 21%) was obtained as a fine,
white crystalline powder, mp 210–213 ꢂC. 1H NMR
(500 MHz, D2O) d 7.47 (t, J = 7.6 Hz, 2H), 7.39 (t,
J = 7.8 Hz, 1H), 7.34 (t, J = 7.6 Hz, 1H), 4.35 (s, 2H),
3.21–3.13 (m, 4H), 3.09 (q, J = 8.1 Hz, 4H), 2.83 (s,
6H), 2.13–2.03 (m, 4H); 13C NMR (75 MHz, D2O) d
133.8, 131.7, 131.2, 129.6, 127.7, 127.4, 53.8, 48.1,
44.2, 44.0, 43.8, 42.4 (2C), 22.1, 20.7; (ESI, Pos.) m/z
284.2 (M+H)+.
4.6.1. N-(2-Chlorobenzyl)-3-(3-dimethylamino-propyla-
mino)-propionamide (A,H amide intermediate). (840 mg,
1
78%). H NMR (500 MHz, CDCl3) d 8.12 (br s, 1H),
4.6.5. N-Cyclohexyl-N0-(3-dimethylamino-propyl)-pro-
pane-1,3-diamine trihydrochloride (G,H). The product
G,H (334 mg,10%) was obtained as a fine, white crystal-
line powder, mp 239–242 ꢂC. 1H NMR (500 MHz, D2O)
d 3.16 (t, J = 8.1 Hz, 2H), 3.12–2.98 (m, 7H), 2.82 (s,
6H), 2.14–1.90 (m, 6H), 1.80–1.66 (br s, 2H), 1.56 (d,
J = 12.9 Hz, 1H), 1.30–1.15 (m, 4H), 1.14–1.01 (m,
1H); 13C NMR (75 MHz, D2O) d 57.0, 53.8, 44.3,
44.0, 42.4 (2C), 40.7, 28.3 (2C), 24.0, 23.4(2C), 22.4,
20.7; (ESI, Pos.) m/z 242.4 (M+H)+; CHN Anal. Calcd
for C14H34Cl3N3Æ2H2O: C, 43.47; H, 9.90; N, 10.86.
Found: C, 43.55; H, 9.92; N, 10.80.
7.31–7.26 (m, 2H), 7.18–7.12 (m, 2H), 4.42 (d,
J = 5.9 Hz, 2H), 3.01(t, J = 6.1 Hz, 2H), 2.90 (t,
J = 6.1 Hz, 2H), 2.61 (t, J = 5.9 Hz, 2H), 2.44 (t,
J = 6.4 Hz, 2H), 2.17 (s, 6H), 1.72 (qin, J = 6.1 Hz,
2H); (ESI, Pos.) m/z 298.3 (M+H)+.
4.6.2. N-Cyclohexyl-3-(3-dimethylamino-propylamino)-
propionamide (G,H amide intermediate). (970 mg, 38%).
1H NMR (500 MHz, CDCl3) d 7.66 (br s, 1H), 3.73–
3.64 (m, 1H), 2.80 (t, J = 6.1 Hz, 2H), 2.60 (t,
J = 6.8 Hz, 3H), 2.27 (t, J = 7.1 Hz, 4H), 2.15 (s, 6H),
1.83–1.77 (m, 2H), 1.65–1.56 (m, 4H), 1.55–1.48 (m,
1H), 1.35–1.24 (m, 2H), 1.17–1.04 (m, 3H); (ESI, Pos.)
m/z 256.2 (M+H)+.
4.6.6. N-[2-(3,4-Dimethoxyphenyl)-ethyl]-N0-(3-dimeth-
ylamino-propyl)-propane-1,3-diamine
trihydrochloride
(H,I). The product H,I (720 mg, 17%) was obtained as
a fine, white crystalline powder, mp 254–256 ꢂC. 1H
NMR (500 MHz, D2O) d 6.92 (d, J = 7.8 Hz, 1H),
6.89 (s, 1H), 6.82 (d, J = 8.3 Hz, 1H), 3.8 (d,
J = 7.8 Hz, 6H), 3.24 (t, J = 7.6 Hz, 2H), 3.20–3.14 (m,
2H), 3.12–3.04 (m, 6H), 2.89 (t, J = 7.6 Hz, 2H), 2.83
(s, 6H), 2.12–1.98 (m, 4H); 13C NMR (75 MHz, D2O)
d 147.9, 146.9, 128.9, 121.0, 112.0, 111.9, 55.3 (2C),
4.6.3. N-[2-(3,4-Dimethoxyphenyl)-ethyl]-3-(3-dimethyla-
mino-propylamino)-propionamide (H,I amide intermedi-
ate). (1.89, 56%). 1H NMR (500 MHz, CDCl3) d 7.76 (t,
J = 4.9 Hz, 1H), 6.74–6.69 (m, 1H), 6.68–6.62 (m, 2H),
3.77 (d, J = 5.9 Hz, 6H), 3.40 (q, J = 7.1 Hz, 2H), 2.73
(t, J = 6.1 Hz, 2H), 2.67 (t, J = 7.3 Hz, 2H), 2.49 (t,
J = 7.1 Hz, 2H), 2.23 (t, J = 5.6 Hz, 2H), 2.18 (t,

6262
K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
53.8, 48.3, 44.2, 44.0, 43.9, 42.7 (2C), 30.7, 22.0, 20.7;
(ESI, Pos.) m/z 324.5 (M+H)+; CHN Anal. Calcd for
C18H36Cl3N3O2: C, 49.95; H, 8.38; N, 9.71. Found: C,
49.82; H, 8.54; N, 9.70.
the indirect dimension.34 Datasets were obtained with
a spectral width of 8000 Hz in the H dimension and
1
2127.55 Hz in the 15N dimension and 128 scans of 512
real time points for each of 80 t1 increments were record-
ed. The data were processed using the Felix processing
software package (Accelerys, San Diego, CA). The 15N
dimension was zero-filled to 256 pts, with the sensitivity
enhancement option selected on the left half of the
spectrum.
4.7. Isothermal titration calorimetry
Isothermal titration calorimetry (ITC) experiments were
performed at 37 ꢂC using a VP-ITC calorimeter from
Microcal, Inc. (Northampton, MA). The enzyme and li-
gand solutions were prepared in 50 mM Tris buffer, pH
7.5, containing 50 mM KCl. The enzyme solution was
extensively dialyzed against this buffer and ligand solu-
tions were prepared in the final dialysate. Enzyme and
ligand solutions were degassed before titrations. Sample
cell and syringe contained 90 lM APH(30) and 1.5 mM
kanamycin A, respectively, in all titrations. In ternary
complex experiments, kanamycin A solution containing
1.5 mM Ca2+ and 1.0 mM ATP or 2.0 mM diamine H,I
was titrated into an enzyme solution containing matched
concentrations of Ca2+ and ATP or diamine H,I. Under
these conditions, more than 95% of APH(30) was present
either in enzyme–CaATP or enzyme–H,I complex dur-
ing the onset of titrations. Binding data were corrected
by performing appropriate ligand- to buffer-dilution
titrations. For all experiments, injection volume was
set to 10 lL and injections were separated by 240 s. Cell
stirring speed was 300 rpm. One set of binding site(s)
model of Origin software package was used for data fit-
ting to obtain binding parameters N (stoichiometry), KA
(association constant), and DH (enthalpy change). The
dissociation constant, free energy of binding (DG) and
entropy change (DS) were obtained using Eqs. 1–3.
4.9. Docking calculations
The docking calculations were performed using the
Flexidock program within the SYBYLꢁ 6.9 software
package (Tripos, Inc., St. Louis, MO). The crystal struc-
ture of aminoglycoside phosphotransferase 30-IIIa
cocrystallized with AMPPNP (PDB #1J7U) was used
for all docking calculations.35 The enzyme-binding site
was based upon the position of the bound AMPPNP;
and the binding site was prepared for calculations
according to the Flexidock setup dialog. No water mol-
ecules or metal ions were included in the calculation.
Residues within three angstroms of the ANPPNP that
contained a carboxylate group were selected as contain-
ing hydrogen bond acceptors. Alanine 93 was selected as
containing a hydrogen bond donor. The diamines were
built using the Sketch module of SYBYL 6.9. All the
torsion angles in the chain were set to 180ꢂ and then
the energy was minimized to give the starting structure
for the docking calculations. All nitrogen atoms in each
diamine molecule were modeled as positively charged,
and Gasteiger–Huckel charges were calculated for each
of the diamines. Two docking calculations were per-
formed on each diamine, representing two different
alignments in the binding pocket that were 180ꢂ apart.
The first orientation aligned the aromatic ring of the dia-
mine with the adenine ring system of ANPPNP, with the
rest of the diamine oriented to loosely match the posi-
tioning of the sugar phosphate backbone. The second
orientation aligned the amino terminal group of the dia-
mine (the tertiary amino group of amine H) with the NH
in the adenine ring of AMPPNP that forms a hydrogen
bond with the backbone carbonyl of alanine 93. The rest
of the molecule was oriented to loosely match the posi-
tioning of the sugar phosphate moiety. The charged
amino groups were selected as hydrogen bond donors
and oxygen atoms with available lone pairs were identi-
fied as hydrogen bond acceptors. The maximum number
of generations was raised to 30,000 and each calculation
produced 20 docked structures. Diamine H,I, the only
compound proven to be a competitive inhibitor of
APH(30) with respect to MgATP, was the only one to
yield an energy minimum inside the ATP-binding site.
KD ¼ 1=KA;
ð1Þ
ð2Þ
ð3Þ
DG ¼ ꢁRT ln KA;
DG ¼ DH ꢁ TDS;
CaATP, a competitive inhibitor of the enzyme,31 was
used, instead of the catalytically required cofactor
MgATP, to prevent product formation in ternary com-
plex titrations.
4.8. NMR spectroscopy
All proton spectra for the characterization of com-
pounds A,A to I,I were acquired on a 500 MHz Varian
INOVA spectrometer. All the spectra with enzyme–sub-
strate and enzyme–inhibitor complexes were acquired
on a 600 MHz Varian INOVA spectrometer using uni-
15
formly N-enriched APH(30). NMR samples contained
4.10. Whole-cell assays
219 lM APH(30) in 25 mM Tris–HCl, pH 7.5, contain-
ing 25 mM KCl. When present, kanamycin A was
450 lM, Ca2+ and ATP were each 0.5 mM, and H,I
was 1.0 mM. A sensitivity-enhanced 1H–15N HSQC
spectrum,32 with the TROSY option,33 from the Bio-
Pack set of pulse sequences for the Varian spectrometer,
was acquired in the phase-sensitive mode using the
States-Haberkorn method for quadrature detection in
Whole-cell assays were performed with the purified diam-
ines and the E. coli transformants used to produce the
AGMEs. The assays were performed in 96-well plates
with selected diamine compounds. Diamines B,I; C,H;
C,I; D,H; E,H; G,H; and H,I were assayed at concentra-
tions of 10 and 50 mM. The 96-well plates were inoculated
with the transformants to yield an initial absorbance of

K. T. Welch et al. / Bioorg. Med. Chem. 13 (2005) 6252–6263
6263
16. Asensio, J. L.; Hidalgo, A.; Bastida, A.; Torrado, M.;
Corzana, F.; Chiara, J. L.; Garcia-Junceda, E.; Canada,
J.; Jimenez-Barbero, J. J. Am. Chem. Soc. 2005, 127,
8278–8279.
17. Yu, L. P.; Oost, T. K.; Schkeryantz, J. M.; Yang, J.;
Janowick, D.; Fesik, S. W. J. Am. Chem. Soc. 2003, 125,
4444–4450.
0.03 U/well. Cell growth was monitored using a Labsys-
tems Multiskan MCC/340 Spectrophotometer manufac-
tured by Fisher Scientific. Absorbance was compared
with those of control cultures on the same plate every
8 h for 36 h by reading the absorbance at 600 nM.
18. He, Y.; Yang, J.; Wu, B.; Robinson, D.; Sprankle, K.;
Kung, P.-P.; Lowery, K.; Mohan, V.; Hofstadler, S.;
Swayze, E. E.; Griffey, R. Bioorg. Med. Chem. Lett. 2004,
14, 695–699.
19. Boehr, D. D.; Draker, K. A.; Koteva, K.; Bains, M.;
Hancock, R. E.; Wright, G. D. Chem. Biol. 2003, 10, 189–
196.
Acknowledgments
This research was supported by Grants MCB 01110741
from the National Science Foundation (E.H.S.) and
from the National Institutes of Health AI057836
(R.E.L.) and AI053796 (R.E.L.).
20. Shepherd, R. G.; Baughn, C.; Cantrall, M. L.; Goodstein,
B.; Thomas, J. P.; Wilkinson, R. G. Ann. N.Y. Acad. Sci.
1966, 135, 686–710.
References and notes
21. Burk, D. L.; Hon, W. C.; Leung, A. , K.-W.; Berghuis, A.
M. Biochemistry 2001, 40, 8756–8764.
22. Daigle, D. M.; McKay, G. A.; Wright, G. D. J. Biol.
Chem. 1997, 272, 24755–24758.
23. Hon, W.-C.; McKay, G. A.; Thompson, P. R.; Sweet, R.
M.; Yang, D. S. C.; Wright, G. D.; Berghuis, A. M. Cell
1997, 89, 887–895.
1. Schatz, A.; Bugie, E.; Walksman, S. A. Proc. Soc. Exp.
Biol. Med. 1944, 55, 66–69.
2. Feldman, W. H. a. H. H. C. Proc. Staff Meet. Mayo Clin.
1944, 19, 593–599.
3. Poole, K. Antimicrob. Agents Chemother. 2005, 49, 479–
487.
4. Yamane, K.; Wachino, J. I.; Doi, Y.; Kurokawa, H.;
Arakawa, Y. Emerg. Infect. Dis. 2005, 6, 951–953.
5. Spotts, C. R.; Stanier, R. Y. Nature 1961, 192, 633–637.
6. Moazed, D.; Noller, H. F. Nature 1987, 327, 389–394.
7. Shaw, K. J.; Rather, P. N.; Hare, R. S.; Miller, G. H.
Microbiol. Rev. 1993, 57, 138–163.
24. McKay, G. , A.; Thompson, P. R.; Wright, G. D.
Biochemistry 1994, 33, 6936–6944.
¨
25. Ozen, C.; Serpersu, E. H. Biochemistry 2004, 43, 14667–
14675.
26. Owston, M. A.; Serpersu, E. H. Biochemistry 2002, 41,
10764–10770.
8. Davies, J. E. Science 1994, 264, 375–382.
9. Sucheck, S. J.; Greenberg, W. A.; Tolbert, T. J.; Wong,
C-H. Angew. Chem. Int. Ed. 2000, 39, 1080–1084.
10. Agnelli, T.; Sucheck, S. J.; Marby, K. A.; Rabuka, D.;
Yao, S. L.; Sears, P. S.; Liang, F. S.; Wong, C. H. Angew.
Chem. Int. Ed. 2004, 43, 1562–1566.
27. Wright, E.; Serpersu, E. H. Protein Expres. Purif. 2004,
35, 373–380.
28. Wright, G. D.; Ladak, P. Antimicrob. Agents Chemother.
1997, 41, 956–960.
29. Williams, J. W.; Northrop, D. B. J. Antibiot. 1979, 32,
1147–1154.
11. Li, J.; Wang, J. H.; Czyryca, P. G.; Chang, H. W.; Orsak,
T. W.; Evanson, R.; Chang, C. W. T. Org. Lett. 2004, 6,
1381–1384.
30. Ekman, D. R.; DiGiammarino, E. L.; Wright, E.;
Witter, E. D.; Serpersu, E. H. Biochemistry 2001, 40,
7017–7024.
12. Fridman, M.; Belakhov, V.; Yaron, S.; Baasov, T. Org.
Lett, 2003, 5, 3575–3578.
31. Cox, J. R.; McKay, G. A.; Wright, G. D.; Serpersu, E. H.
J. Am. Chem. Soc. 1996, 118, 1295–1301.
32. Kay, L. E.; Keifer, P.; Saarinen, T. J. Am. Chem. Soc.
1992, 114, 10663–10665.
33. Pervushin, K.; Riek, R.; Wider, G.; Wuthrich, K. Proc.
Natl. Acad. Sci. U.S.A. 1997, 94, 12366–12371.
34. States, D. J.; Haberkorn, R. A.; Ruben, D. J. J. Mag. Res.
1982, 48, 286–292.
13. Haddad, J.; Kotra, L. P.; Llano-Sotelo, B.; Kim, C.;
Azucena, E. F.; Liu, M. Z.; Vakulenko, S. B.; Chow, C. S.;
Mobashery, S. J. Am. Chem. Soc. 2002, 124, 3229–3237.
14. Kim, C.; Haddad, J.; Vakulenko, S. B.; Meroueh, S. O.;
Wu, Y.; Yan, H.; Mobashery, S. Biochemistry 2004, 43,
2373–2383.
15. Griffey, R. H.; Hofstadler, S. A.; Swayze, E. E. Carbohy-
dr.-Based Drug Discovery 2003, 2, 483–499.
35. Burk, D. L.; Hon, W. C.; Leung, A.-W.; Berghuis, A. M.
Biochemistry 2001, 40, 8756–8764.