Exploring the substrate promiscuity of drug-modifying enzymes for the chemoenzymatic generation of N-acylated aminoglycosides

Article abstract of DOI:10.1002/cbic.200900584

Aminoglycosides are broad-spectrum antibiotics commonly used for the treatment of serious bacterial infections. Decades of clinical use have led to the widespread emergence of bacterial resistance to this family of drugs limiting their efficacy in the clinic. Here, we report the development of a methodology that utilizes aminoglycoside acetyltransferases (AACs) and unnatural acyl coenzyme A analogues for the chemoenzymatic generation of N-acylated aminoglycoside analogues. Generation of N-acylated aminoglycosides is followed by a simple qualitative test to assess their potency as potential antibacterials. The studied AACs (AAC(6′)-APH(2″) and AAC(3)-IV) show diverse substrate promiscuity towards a variety of aminoglycosides as well as acyl coenzyme A derivatives. The enzymes were also used for the sequential generation of homo- and hetero-di-N-acylated aminoglycosides. Following the clinical success of the N-acylated amikacin and arbekacin, our chemoenzymatic approach offers access to regioselectively N-acylated aminoglycosides in quantities that allow testing of the antibacterial potential of the synthetic analogues making it possible to decide which molecules will be worth synthesizing on a larger scale.

Full text of DOI:10.1002/cbic.200900584

DOI: 10.1002/cbic.200900584
Exploring the Substrate Promiscuity of Drug-Modifying
Enzymes for the Chemoenzymatic Generation of N-
Acylated Aminoglycosides
Keith D. Green,[d] Wenjing Chen,[c, d] Jacob L. Houghton,[a, d] Micha Fridman,*[b] and
Sylvie Garneau-Tsodikova*[a, c, d]
Aminoglycosides are broad-spectrum antibiotics commonly
used for the treatment of serious bacterial infections. Decades
of clinical use have led to the widespread emergence of bacte-
rial resistance to this family of drugs limiting their efficacy in
the clinic. Here, we report the development of a methodology
that utilizes aminoglycoside acetyltransferases (AACs) and un-
natural acyl coenzyme A analogues for the chemoenzymatic
generation of N-acylated aminoglycoside analogues. Genera-
tion of N-acylated aminoglycosides is followed by a simple
qualitative test to assess their potency as potential antibacteri-
als. The studied AACs (AAC(6’)-APH(2’’) and AAC(3)-IV) show di-
verse substrate promiscuity towards a variety of aminoglyco-
sides as well as acyl coenzyme A derivatives. The enzymes
were also used for the sequential generation of homo- and
hetero-di-N-acylated aminoglycosides. Following the clinical
success of the N-acylated amikacin and arbekacin, our chemo-
enzymatic approach offers access to regioselectively N-acylated
aminoglycosides in quantities that allow testing of the antibac-
terial potential of the synthetic analogues making it possible
to decide which molecules will be worth synthesizing on a
larger scale.
Introduction
Aminoglycosides are broad-spectrum antibiotics commonly
used for the treatment of serious bacterial infections.[1–3] These
antibacterial agents target the prokaryotic ribosome by bind-
ing the decoding A-site of the 16S ribosomal RNA and cause
interference with protein translation, which ultimately leads to
cell death.[1,4–6] Decades of intensive clinical use of aminoglyco-
sides has led to evolutionarily driven bacterial resistance that
compromises their clinical use.[7] A common mode of bacterial
resistance evolved through the acquisition of enzymes that
chemically modify the aminoglycoside including aminoglyco-
side acetyltransferases (AACs), nucleotidyltransferases (ANTs),
and phosphotransferases (APHs).[2,4] Other modes of resistance
include decreasing cell membrane permeability towards ami-
noglycoside uptake and structural alteration in the ribosomal
target of the drug, as well as extrusion of the aminoglycosides
from the cell by efflux pumps.[8]
tion (Scheme 1).[13] Mobashery and co-workers applied AHB
substitution at position N-1 of neamine with the rational that
AHB substitution in amikacin is responsible for the protection
against a number of aminoglycoside-modifying enzymes.[14]
Some of the structures showed considerably enhanced activity
against different pathogenic and resistant strains, and their ac-
tivities were comparable to that of amikacin. Baasov and col-
leagues also took advantage of the AHB moiety at the N-1 po-
sition of paromamine to generate the novel aminoglycoside
NB54. NB54 exhibits significant read-through activity that
offers a potential solution to overcome genetic disorders that
result from stop codon mutations.[15] While the majority of the
known N-acylated aminoglycosides have an AHB group at po-
[a] J. L. Houghton, Dr. S. Garneau-Tsodikova
Department of Medicinal Chemistry in the College of Pharmacy
University of Michigan, 210 Washtenaw Ave.
Ann Arbor, MI 48109 (USA)
While N-acetylation by the AAC family of aminoglycoside-
modifying enzymes evolved in bacteria to deactivate the ami-
noglycosides, some natural and effective aminoglycosides con-
tain an N-acylated amine (Scheme 1). In most cases, these nat-
ural N-acylated aminoglycosides show broad-spectrum activity
against strains that are resistant to non-N-acylated aminoglyco-
sides. One such example is butirosin, which is produced by Ba-
cillus circulans and has been identified as 1-N-(S)-a-hydroxy-g-
amino-n-butyryl (AHB) ribostamycin. Butirosin was found to be
active against some ribostamycin-resistant bacteria thus dem-
onstrating the efficacy of the AHB N-acylation effect.[9–11]
Fax: (+1)734-615-5521
E-mail: sylviegt@umich.edu
[b] Dr. M. Fridman
School of Chemistry, Tel Aviv University
Tel Aviv 66978 (Israel)
Fax: (+972)3-6409203
E-mail: mfridman@post.tau.ac.il
[c] W. Chen, Dr. S. Garneau-Tsodikova
Chemical Biology Doctoral Program, University of Michigan
210 Washtenaw Ave., Ann Arbor, MI 48109 (USA)
[d] Dr. K. D. Green, W. Chen, J. L. Houghton, Dr. S. Garneau-Tsodikova
Life Sciences Institute, University of Michigan
These observations led to the development of the semisyn-
thetic arbekacin, a marketed chemotherapeutic agent, and
amikacin,[12] both of which have an HB group at the 1-N-posi-
210 Washtenaw Ave., Ann Arbor, MI 48109 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.200900584.
ChemBioChem 2010, 11, 119 – 126
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
119

M. Fridman, S. Garneau-Tsodikova et al.
Scheme 1. Chemical structures of natural and synthetic 2-deoxystreptamine-derived (ring II) aminoglycosides used in this study.
sition N-1, other examples of different acyl groups with differ-
ent positioning around the rings of the aminoglycoside exist.
Sporaricin A and astromicin, which is also known as fortimi-
cin A, are pseudodisaccharide aminoglycosides containing an
N-methyl-N-glycinyl group and exhibit antibacterial activity
against a variety of bacterial strains.[16,17] Dactimicin, with an N-
methyl-N-formimidoylglycyl moiety, exhibits antibacterial activi-
ty superior to that of amikacin against resistant strains.[18]
These characteristics demonstrate the medicinal potential of N-
acylated aminoglycosides and the need to explore the antibac-
terial potential of novel analogues containing a variety of N-
acyl groups at different positions on the aminoglycoside.
The synthesis of N-acylated aminoglycoside derivatives
poses the challenge of approaching a specific amine position
on an aminoglycoside. The multiple amines about the amino-
glycoside skeleton (ranging between four on kanamycin A to
six on neomycin B) share similar chemical reactivity and make
it difficult to selectively N-acylate amines on the aminoglyco-
sides. The existing derivatives, such as amikacin and dibekacin,
rely on regioselective activation of the amine by metal chela-
tion and protecting group manipulations for their synthesis.[13]
Selecting an amine position other than the N-1, which was
chosen in the case of amikacin and arbekacin, sets a new syn-
thesis challenge. We therefore reasoned that we could exploit
the regioselectivity offered by AACs for the in vitro chemoen-
zymatic generation of novel regioselectively N-acylated amino-
glycosides to circumvent the need for multistep syntheses. In
this study we report the use of a family of enzymes that
evolved in bacteria to confer resistance to aminoglycosides,
the AACs, as a tool for the development of N-acylated antibiot-
ic analogues. Even though limited amounts of products are
generated chemoenzymatically, the method is extremely valua-
ble as it allows for an initial activity screen that can guide fur-
ther decisions as to which compounds justify large scale and
multistep syntheses.
In a previous publication we demonstrated that synthetic
acyl coenzyme A (CoA) analogues are accepted by the CouN1
and CouN7 proteins involved in transfer of the pyrrolyl moiety
during the late stage of coumermycin A1 biosynthesis. By using
a chemoenzymatic approach, we demonstrated that novel
novobiocin analogues can be produced in milligram quantities
by the action of the pair of enzymes CouN1 and CouN7.[19]
Recently, we also demonstrated the substrate and cosubstrate
promiscuity of choline acetyltransferase (ChAT) for the produc-
tion of acetylcholine analogues.[20] These studies clearly dem-
onstrate how the promiscuity of an acyltransferase system can
be used for the chemoenzymatic generation of novel mole-
cules that would otherwise require multistep syntheses.
Results and Discussion
The promiscuity of AACs towards a variety of aminoglycosides
is well established. However, the promiscuity of these AACs to-
wards acyl-CoA cosubstrates has had limited exploration. Due
to the challenges and multiple protecting group manipulations
required for the generation of a single N-acylated aminoglyco-
side analogue, we aimed to develop a methodology that will
rapidly allow access to the desired compound and will enable
an initial screen to assess the antibacterial potential of the de-
sired N-acylated aminoglycoside analogues. Two AACs, AAC(3)-
IV and AAC(6’)-APH(2’’) were chosen for over-expression, purifi-
cation, and testing for substrate and cosubstrate promiscuity
(Figure S1 in the Supporting Information and Table 1). It has
previously been reported that incorporation of an N-terminal
histidine tag to AAC(6’)-APH(2’’) affects the kinetic parameters
of both acetyl-CoA and kanamycin A.[21] We therefore tested N-
terminal hexahistidine tagged (NHis6), N-terminal decahistidine
tagged (NHis10), C-terminal hexahistidine (CHis6) tagged, and a
N-terminal tag cleaved AAC(6’)-APH(2’’) to compare kinetic pa-
rameters. In our hands all forms of AAC(6’)-APH(2’’) display a
KM of the same magnitude for acetyl-CoA (2.0Æ0.8 to 9.9Æ
120
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 119 – 126

Chemoenzymatic Generation of N-Acylated Aminoglycosides
Table 1. Substrates accepted by AAC(6’)-APH(2’’)(CHis6) and AAC(3)-IV(NHis10)[a] from pET28a by using various aminoglycosides at pH 6.6 and 5.7 (for neo-
mycin B, sisomicin, and gentamicin with AAC(3)-IV).[g]
Amikacin[b] Gentamicin Kanamycin A[b] Neomycin B Paromomycin[a] Sisomicin
Tobramycin
Substrate
Structure
6’
6’
3
6’
6’
3
3
6’
3
6’
3
[e]
[f]
ꢁ[d]
–
ꢁ
–
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
3
acetoacetyl-CoA
3
–
–
3
3
3
3
–
–
3
ꢁ
ꢁ
3
ꢁ
ꢁ
3
–
–
3
ꢁ
ꢁ
3
–
–
3
ꢁ
ꢁ
acetyl-CoA
benzoyl-CoA
butyryl-CoA
–
–
–
–
3
ꢁ
4-bromo-thiophene-
2-carbonyl-S-CoA[c]
3
3
3
3
3
3
–
ꢁ
–
ꢁ
–
5-bromo-thiophene-
3
3
–
ꢁ
–
–
ꢁ
ꢁ
ꢁ
ꢁ
–
ꢁ
ꢁ
–
–
ꢁ
ꢁ
2-carbonyl-S-CoA[c]
crotonyl-CoA
ꢁ
ꢁ
ꢁ
–
ꢁ
6-fluoro-picolinyl-S-CoA[c]
–
–
ꢁ
ꢁ
–
ꢁ
–
–
3
3
3
3
3
3
3
–
–
ꢁ
glutaryl-CoA
–
ꢁ
–
–
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
–
–
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
–
–
–
ꢁ
ꢁ
–
glycyl-CoA[c]
3
3
3
3
d,l-b-OH-butyryl-CoA
isovaleryl-CoA
malonyl-CoA
–
–
3
–
–
3
ꢁ
ꢁ
–
–
–
3
ꢁ
ꢁ
ꢁ
–
–
–
–
–
–
3
methylmalonyl-CoA
–
–
ꢁ
–
ꢁ
ꢁ
–
ꢁ
–
ꢁ
palmitoyl-CoA
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
–
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
3
3
3
3
3
3
3
3
3
3
n-propionyl-CoA
[a] Paromomycin was not tested with AAC(6’)-APH(2’’) as there is no NH2 at position 6’. [b] Amikacin and kanamycin A were not tested with AAC(3)-IV due
to their previously reported precipitous KM with acetyl-CoA. [c] Synthetic CoA analogues. The CoA analogues without symbols are commercially available.
[d] Indicates a poor or no increase in the spectrophotometric assay over 30 min. [e] Indicates a moderate increase in the spectrophotometric assay over
30 min. [f] Indicates a good increase in the spectrophotometric assay over 30 min. [g] Procedure utilized for acyl transfer by AACs: the CoA-SH released
upon the acylation reaction (200 mL reaction mixture containing 50 mm MES buffer, 20 mm aminoglycoside, 40 mm CoA derivative, 2 mm DTDP, and 5.9 mg
AAC enzyme) was monitored by a spectrophotometric assay at 324 nm, the characteristic wavelength of 4-thiopyridone formed by reaction of CoA-SH and
DTDP.
1.9 mm), however, the catalytic turnovers were drastically differ-
ent (Table 2). The NHis6-tagged enzyme has a turnover (kcat) of
0.012Æ0.003 sÀ1, the NHis10-tagged enzyme has a kcat of
0.163Æ0.019 sÀ1, the CHis6-tagged enzyme has a kcat of 0.300Æ
0.075 sÀ1, and the tag-cleaved (untagged) enzyme had an ob-
served kcat of 0.559Æ0.155 sÀ1. As the CHis6-tagged and untag-
ged enzymes have nearly identical catalytic efficiencies with
acetyl-CoA, we chose to test the remaining aminoglycoside
ChemBioChem 2010, 11, 119 – 126
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
121

M. Fridman, S. Garneau-Tsodikova et al.
Table 2. Kinetic parameters for AAC(6’)-APH(2’’) and AAC(3)-IV.
Enzyme (vector used)
pH
Aminoglycoside
CoA derivative
KM [mm]
kcat [sÀ1
]
kcat/KM [mmÀ1 sÀ1
]
CoA analogue kinetic parameters
AAC(6’)-APH(2’’)(NHis) (pET28a)
6.6
6.6
6.6
6.6
kanamycin A
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
acetyl-CoA[a]
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
malonyl-CoA
acetyl-CoA
n-propionyl-CoA
malonyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
acetoacetyl-CoA
benzoyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
acetyl-CoA
2.0Æ0.8
4.8Æ1.5
0.012Æ0.003
0.017Æ0.001
0.163Æ0.019
0.044Æ0.001
0.559Æ0.155
0.144Æ0.005
0.300Æ0.075
0.075Æ0.008
0.134Æ0.032
0.126Æ0.028
0.124Æ0.017
0.146Æ0.008
0.182Æ0.052
0.123Æ0.015
0.041Æ0.004
0.188Æ0.063
0.076Æ0.015
0.173Æ0.028
0.121Æ0.007
0.179Æ0.026
0.089Æ0.004
0.043Æ0.004
0.037Æ0.008
0.104Æ0.069
0.088Æ0.003
0.020Æ0.001
0.025Æ0.002
0.350Æ0.083
0.045Æ0.002
0.130Æ0.018
0.075Æ0.003
0.057Æ0.007
0.032Æ0.003
0.045Æ0.002
0.69Æ0.11
0.024
0.003
0.050
0.005
0.056
0.011
0.055
0.027
0.012
0.131
0.060
0.002
0.040
0.043
0.002
0.013
0.024
0.112
0.211
0.066
0.016
0.053
0.0004
0.152
0.001
0.001
0.0003
0.039
0.0018
0.035
0.0016
0.028
0.002
0.012
0.016
0.004
AAC(6’)-APH(2’’)(NHis) (Int-pET19b-pps)
AAC(6’)-APH(2’’)(no tag) (Int-pET19b-pps)
AAC(6’)-APH(2’’)(CHis) (pET22b)
kanamycin A
kanamycin A
kanamycin A
3.3Æ0.2
9.1Æ0.9
9.9Æ1.9
13.5Æ1.5
5.5Æ0.4
2.8Æ0.1
11.1Æ2.4
0.96Æ0.17
2.1Æ0.6
6.6
6.6
amikacin
71.6Æ14.6
4.6Æ0.4
gentamicin
2.9Æ0.6
18.9Æ5.0
14.8Æ6.3
3.1Æ0.9
6.6
6.6
6.6
5.7
5.7
neomycin B
sisomicin
1.5Æ0.5
0.57Æ0.16
2.7Æ0.5
tobramycin
gentamicin
gentamicin
5.4Æ1.1
AAC(3)-IV(NHis) (pET28a)
0.803Æ0.001
94.6Æ16.8
0.68Æ0.09
91.8Æ8.0
23.1Æ2.1
78.8Æ12.1
9.17Æ1.90
24.9Æ1.1
3.7Æ0.9
AAC(3)-IV(NHis) (Int-pET19b-pps)
5.7
6.6
5.7
6.6
neomycin B
paromomycin
sisomicin
46.9Æ1.7
2.05Æ0.74
15.1Æ0.9
3.76Æ1.17
43.0Æ1.1
46.5Æ12.4
n-propionyl-CoA
acetyl-CoA
n-propionyl-CoA
malonyl-CoA
tobramycin
0.189Æ0.049
Aminoglycoside kinetic parameters
AAC(6’)-APH(2’’)(CHis) (pET22b)
6.6
6.6
5.7
5.7
6.6
5.7
6.6
kanamycin A
sisomicin
gentamicin
neomycin B
paromomycin
sisomicin
acetyl-CoA
acetyl-CoA
acetyl-CoA
acetyl-CoA
acetyl-CoA
acetyl-CoA
acetyl-CoA
0.40Æ0.13
5.9Æ1.8
0.034Æ0.005
0.382Æ0.104
0.036Æ0.004
0.045Æ0.030
0.290Æ0.004
0.026Æ0.003
0.12Æ0.02
0.087
0.065
0.040
0.082
0.023
0.074
0.066
AAC(3)-IV(NHis) (Int-pET19b-pps)
0.89Æ0.12
0.56Æ0.26
1.26Æ0.19
0.35Æ0.02
1.87Æ0.51
tobramycin
[a] The reaction was performed with the Thio-Glo1 fluorescent detection method.
and CoA derivative pairs with the CHis6-tagged enzyme due to
the ease of purification and its improved catalytic activity com-
pared to the NHis-tagged forms.
AAC(3)-IV was, therefore, the construct of choice for our study.
Kinetic analysis showed that the NHis6-tagged AAC(3)-IV gave
a KM of 0.803Æ0.001 mm and catalytic turnover of kcat =0.043Æ
0.004 sÀ1 for acetyl-CoA to give a catalytic efficiency of
Based on the results from AAC(6’)-APH(2’’) we thought that
the CHis6-tagged AAC(3)-IV would be the construct of choice.
However, examination of AAC(3) X-ray crystal structures similar
to AAC(3)-IV revealed that dimerization takes place near the C-
terminal residues of the protein.[22] Therefore, the proximity of
the CHis6-tag to the dimer interface might affect the ability of
AAC(3)-IV to bind well to the NiII-NTA column, explaining the
difficulties associated with purification of the AAC(3)-IV CHis6-
tagged construct that we encountered. The NHis-tagged
0.053 mmÀ1 À1. The NHis10-tagged AAC(3)-IV gave similar results
s
for the KM (0.68Æ0.09 mm), but an increased catalytic turnover
(0.104Æ0.069 sÀ1) for acetyl-CoA resulting in an increased effi-
ciency (0.152 mmÀ1 À1). From these experiments we decided to
s
use the NHis10-tagged AAC(3)-IV enzyme to determine the
kinetic values for the remaining substrate/cosubstrate pairs.
We first tested the chosen AACs using n-propionyl-CoA in an
initial attempt to assess the substrate specificity of the catalytic
122
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 119 – 126

Chemoenzymatic Generation of N-Acylated Aminoglycosides
site towards larger groups compared with the methyl group of
acetyl-CoA. AAC(6’)-APH(2’’) and AAC(3)-IV readily transferred
n-propionyl-CoA to the N-6’ and N-3 of a wide variety
of aminoglycosides, respectively
(Table 2). The formation of 6’-N-
n-propionyl neomycin B and 3-N-
n-propionyl neomycin B were
confirmed by HRMS (Figure 1B
and C).
The relative relaxed acyl-CoA specificity of the enzyme towards
N-1 acylated amikacin demonstrates the potential of the
AAC(6’)-APH(2’’) for the preparation of doubly and hetero-N-
The formation of the products
can also be easily and clearly de-
tected by a simple TLC test as
demonstrated in Figure S2 in the
Supporting Information. With
the initial results demonstrating
the tolerance of the chosen
AACs towards more sterically de-
manding acyl groups, we tested
a selection of both commercially
available and synthetic acyl-CoAs
(Table 1). All of the synthetic
acyl-CoA analogues used in this
study were prepared according
to our previously reported meth-
odology.[19]
6’-N-Acylation of aminoglyco-
sides by AAC(6’)-APH(2’’) bi-
functional enzyme
AAC(6’)-APH(2’’) exhibited re-
laxed substrate specificity by
readily transferring a diverse set
of acyl-CoAs to a diverse set of
aminoglycoside scaffolds. Out of
the 96 combinations tested, 29
exhibited good transfer activity
and demonstrated appreciable
product increases when com-
pared to acetyl-CoA after incuba-
tion for 30 min. Reactions were
monitored at 324 nm for the re-
action of DTDP (4,4’-dithiodipyri-
dine) with the CoA released
upon product formation.[23]
Interestingly, methylmalonyl-
CoA was successfully transferred
to neomycin B while it was a
poor substrate for kanamycin A
and amikacin. This can be ration-
alized by the common pseudo-
trisaccharide structural features
of kanamycin A and amikacin
compared with the pseudopen-
tasaccharide structure of neomy-
cin B. Interestingly, the opposite
case occurred with glutaryl-CoA. Figure 1. Mass spectra for: A) neomycin B, B) 6’-N-n-propionyl neomycin B, and C) 3-N-n-propionyl neomycin B.
ChemBioChem 2010, 11, 119 – 126
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
123

M. Fridman, S. Garneau-Tsodikova et al.
acylated aminoglycosides. The potential of N-glycinylated ami-
noglycosides has been established by the active astromicin
and sporaricin A. The successful transfer of the glycyl moiety to
the 6’ position of all aminoglycosides tested demonstrates the
power of our chemoenzymatic methodology for the genera-
tion of potentially novel active N-glycinylated aminoglycosides
(Table 1).
3-N-Acylation of aminoglycosides by AAC(3)-IV
AAC(3)-IV exhibited tolerance towards a variety of combina-
tions of aminoglycosides and acyl-CoA analogues: out of the
80 combinations tested, 25 had successful transfers. Interest-
ingly, combinations that involved gentamicin exhibited the
highest percentage of success with acyl-CoAs ranging from the
natural and small acetyl group to the functionalized and steri-
cally demanding acetoacetyl and 6-fluoropicolinyl groups.
Given the vast use of gentamicin in the clinic as a last resort
treatment for some severe respiratory system infections and
the emergence of gentamicin resistant strains, the generation
of N-acylated analogues of this drug by using AAC(3)-IV will
enable a more diverse exploration of potentially potent deriva-
tives of gentamicin.
Sequential double acylation by AAC(3)-IV and AAC(6’)-
APH(2’’) yields double hetero- as well as homo-N-acylated
aminoglycosides
Based on the observed relaxed substrate specificity of a variety
of acyl-CoAs and the N-1 acylated amikacin, we tested the pos-
sibility of dual acylation of aminoglycosides using a sequential
transfer of acyl groups by the two enzymes. In order to further
extend our methodology, neomycin B was acylated at both the
6’ and 3 positions by using AAC(6’)-APH(2’’) and AAC(3)-IV se-
quentially. An increase in absorbance was observed after ad-
ding AAC(6’)-APH(2’’) to a mixture of acetyl-CoA (or n-propion-
yl-CoA) and neomycin B. After 30 min incubation, an additional
portion of acetyl-CoA (or n-propionyl-CoA) was added followed
by the addition of AAC(3)-IV. An increase in the absorbance in-
dicated the acylation at 3 position of 6’-acylated neomycin B to
form 3,6’-diacylated neomycin B (Figure 2). The formation of 6’-
N-acetyl neomycin B by AAC(6’)-APH(2’’) followed by the for-
mation of 3,6’-N’,N-diacetyl neomycin B by AAC(3) could be
visualized by TLC (Figure 2C). When sisomicin was used with a
combination of acetyl-CoA and n-propionyl-CoA, respectively,
6’-N-acetyl-3-N-n-propionyl sisomicin was obtained. These ex-
amples demonstrate the utility of our chemoenzymatic
method for the generation of aminoglycoside derivatives with
more than one N-acylated position. Aminoglycosides that were
initially acylated by AAC(3)-IV failed to go through the second
acylation by AAC(6’)-APH(2’’). However, if the same aminogly-
cosides were initially acylated by AAC(6’)-APH(2’’), the 6’-N-acy-
lated aminoglycoside was readily accepted as a substrate of
AAC(3)-IV (Figure 2) as proved to be the case for neomycin B.
When the 3 position of neomycin B was acylated by AAC(3)-IV,
AAC(6’)-APH(2’’) was unable to perform acylation on the 3-acy-
lated neomycin B. On the other hand, initial acylation by
Figure 2. Spectrophotometric assay plots. A) The formation of 6’-N-n-pro-
pionyl neomycin B by using AAC(6’)-APH(2’’) followed by the formation of
*
6’,3-N’,N-di-n-propionyl neomycin B by addition of AAC(3)-IV ( ) and the for-
mation of 6’-N-n-propionyl neomycin B by using AAC(6’)-APH(2’’) followed
by the formation of 6’-N-n-propionyl-3-acetyl neomycin B by addition of
*
*
AAC(3)-IV ( ). : 0–3600 s=1 equiv neomycin B+1.1 equiv ProCoA+
AAC(6)-APH(2’’); 3600–5400 s=add 1.1 equiv ProCoA; 5400–7200 s=add
*
AAC(3)-IV; : 0–3600 s=1 equiv neomycin B+1.1 equiv ProCoA+AAC(6’)-
APH(2’’); 3600–5400 s=add 1.1 equiv AcCoA; 5400–7200 s=add AAC(3)-IV.
B) The formation of 6’-N-acetyl neomycin B by using AAC(6’)-APH(2’’) fol-
lowed by the formation of 6’-N-acetyl-3-N-n-propionyl neomycin B by addi-
*
tion of AAC(3)-IV ( ) and the formation of 6’-N-acetyl neomycin B by using
AAC(6’)-APH(2’’) followed by the formation of 6’,3-N’,N-diacetyl neomycin B
*
*
by addition of AAC(3)-IV ( ). : 0–1800 s=1 equiv neomycin B+1.1 equiv
AcCoA+AAC(6’)-APH(2’); 1800–3600 s=add 1.1 equiv ProCoA; 3600–
*
4800 s=add AAC(3)-IV; : 0–1800 s=1 equiv neomycin B+1.1 equiv Ac-
CoA+AAC(6’)-APH(2’’); 1800–3600 s=add 1.1 equiv AcCoA; 3600–4800 s=
add AAC(3)-IV. C) TLC showing the formation of 6’-acetyl-neomycin B and
6’,3-N’,N-diacetyl neomycin B.
AAC(6’)-APH(2’’) resulted in the 6’-acylated neomycin B, which
was readily acylated by AAC(3)-IV to yield the 3,6’-diacylated
neomycin B. This proved to be the case for sisomycin as well.
We suggest that the amino group at the 3 position of these
antibacterials is crucial for the binding to the active site of
124
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 119 – 126

Chemoenzymatic Generation of N-Acylated Aminoglycosides
AAC(6’)-APH(2’’) and therefore the desired diacylation must
commence with a transfer by AAC(6’)-APH(2’’) followed by the
second transfer by AAC(3)-IV.
Conclusions
AAC(3)-IV and AAC(6’)-APH(2’’) were found to exhibit tolerance
towards a variety of aminoglycosides as well as a diverse set of
acyl-CoAs. The relaxed substrate specificity of both enzymes
proved to be a useful tool for the preparation of a variety of
regioselectively N-acylated aminoglycosides. The chemoenzy-
matic methodology proposed herein could be used for the
generation of sufficient amounts of compound for a qualitative
screen for antibacterial activity. The enzymes proved efficient
not only for catalyzing single N-acylation but could also be
used for sequential dual acylations. In view of the success of
N-acylated aminoglycosides in exhibiting superior antimicrobial
activity and reduced toxicity, our methodology sets the
ground for the rapid generation and exploration of novel
members of this family of compounds.
BioTLC-based antibacterial activity screen
With the relaxed substrate and cosubstrate specificity of both
AACs established, a suitable assay to assess the antibacterial
activity of small quantities of the N-acylated aminoglycosides
was needed. As the synthesis and purification of milligram
scales of each analogue would require a large amount of the
acyl-CoAs and time, we needed a way to perform an initial
qualitative antibacterial test on a prepurified analogue. We suc-
cessfully accomplished both the prepurification and qualitative
antibacterial test using a bioTLC protocol.[24] The bioTLCs re-
vealed that neomycin B that was acetylated or n-propionylated
at the 6’ amine showed a clear inhibition of the growth of
B. subtilis (Figure 3A). Gentamicin acylated at the 3 position
with acetyl and n-propionyl-CoA also exhibited a clear zone of
growth inhibition (data not shown). The minimum inhibitory
concentrations (MICs) of neomycin B and gentamicin against
B. subtilus are 0.1525 and 0.0549 mgmLÀ1, respectively.[25] Our
Experimental Section
Experimental details (bacterial strains, plasmids, materials, and in-
strumentation as well as methods for preparation of over-expres-
sion constructs; overproduction and purification of AAC(6’)-
APH(2’’) and AAC(3)-IV proteins; pH profiles of AAC(6’)-APH(2’’) and
AAC(3)-IV; assay for determination of CoA derivatives substrate
specificity; determination of kinetic parameters; TLC time course;
and bioTLC) are provided in the Supporting Information. Represen-
tative examples of MS spectra, FPLC traces, pH profiles, UV/Vis
assay results, kinetic curves, and TLC standards of neomycin B, gen-
tamicin, CoA and acetyl-CoA are also provided.
Acknowledgements
This work was supported by the Life Sciences Institute and the
College of Pharmacy at the University of Michigan (S.G.T.) and by
the Department of Chemistry at the University of Tel Aviv (M.F.).
This research was also supported by a grant from the United
States–Israel Binational Science Foundation (BSF), Jerusalem,
Israel (grant 2008017, S.G.T. and M.F.). The ACS Division of Medic-
inal Chemistry and Eli Lilly are acknowledged for a predoctoral
fellowship to J.L.H. We thank Dr. Mi Hee Lim (LSI, University of
Michigan) for the use of her SpectraMax M5. We thank Profs.
Timor Baasov (Israel Institute of Technology) and John S. Blan-
chard (Albert Einstein College of Medicine, NY) for providing the
plasmids encoding the AAC(6’)-APH(2’’) and AAC(3)-IV, respective-
ly. We thank Dr. Tapan Biswas (University of Michigan) for the
Int-pET19b-pps vector.
Figure 3. BioTLC showing: A) inhibition of Bacillus subtilis growth by neomy-
cin B (lanes 1 and 4), 6’-N-acetyl neomycin B (lane 2), and 6’-N-n-propionyl
neomycin B (lane 3). B) There is no inhibition of B. subtilis growth by 3-N-
acetyl neomycin B (lane 5) and 3-N-n-propionyl neomycin B (lane 6).
novel N-acylated aminoglycosides could be either more or less
active than their parent drugs. However, because neomycin B
and gentamicin are highly active against B. subtilis, the bioTLC
does not allow detecting a small reduction or increase in activ-
ity. On the other hand, neomycin B acylated at the 3 position
showed no inhibition compared to neomycin B (Figure 3B).
The bioTLC procedure successfully combines a short and
simple TLC isolation of the active compound followed by a
qualitative antibacterial assay that can indicate the antibacteri-
al potential of the tested analogue. The minimal quantity of
compounds required for the bioTLC procedure allows the pro-
duction of a large number of analogues with reasonable quan-
tities of the acyl-CoAs.
Keywords: aminoglycoside acetyltransferases · antibiotics ·
chemoenzymatic acylation
enzymes
·
coenzyme A cosubstrates
·
[1] B. D. Davis, Microbiol. Rev. 1987, 51, 341.
[2] S. Jana, J. K. Deb, Appl. Microbiol. Biotechnol. 2006, 70, 140.
[3] M. Hainrichson, I. Nudelman, T. Baasov, Org. Biomol. Chem. 2008, 6, 227.
[4] S. Magnet, J. S. Blanchard, Chem. Rev. 2005, 105, 477.
[5] J. M. Ogle, D. E. Brodersen, W. M. Clemons, Jr., M. J. Tarry, A. P. Carter, V.
Ramakrishnan, Science 2001, 292, 897.
[6] Q. Vicens, E. Westhof, Biopolymers 2003, 70, 42.
ChemBioChem 2010, 11, 119 – 126
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
125

M. Fridman, S. Garneau-Tsodikova et al.
[7] G. D. Wright, A. M. Berghuis, S. Mobashery, Adv. Exp. Med. Biol. 1998,
456, 27.
[18] Y. Matsuhashi, T. Yoshida, T. Hara, Y. Kazuno, S. Inouye, Antimicrob.
Agents Chemother. 1985, 27, 589.
[8] A. Kumar, H. P. Schweizer, Adv. Drug Delivery Rev. 2005, 57, 1486.
[9] P. W. K. Woo, H. W. Dion, Q. R. Bartz, Tetrahedron Lett. 1971, 12, 2617.
[10] P. W. K. Woo, Tetrahedron Lett. 1971, 12, 2621.
[11] P. W. K. Woo, H. W. Dion, Q. R. Bartz, Tetrahedron Lett. 1971, 12, 2625.
[12] S. Kondo, K. Hotta, J. Infect. Chemother. 1999, 5, 1.
[13] H. Kawaguchi, T. Naito, S. Nakagawa, K. I. Fujisawa, J. Antibiot. 1972, 25,
695.
[19] M. Fridman, C. J. Balibar, T. Lupoli, D. Kahne, C. T. Walsh, S. Garneau-
Tsodikova, Biochemistry 2007, 46, 8462.
[20] K. D. Green, M. Fridman, S. Garneau-Tsodikova, ChemBioChem 2009, 10,
2191.
[21] D. D. Boehr, W. S. Lane, G. D. Wright, Chem. Biol. 2001, 8, 791.
[22] E. Wolf, A. Vassilev, Y. Makino, A. Sali, Y. Nakatani, S. K. Burley, Cell 1998,
94, 439.
[14] J. Haddad, L. P. Kotra, B. Llano-Sotelo, C. Kim, E. F. Azucena, Jr., M. Liu,
S. B. Vakulenko, C. S. Chow, S. Mobashery, J. Am. Chem. Soc. 2002, 124,
3229.
[15] I. Nudelman, A. Rebibo-Sabbah, M. Cherniavsky, V. Belakhov, M. Hain-
richson, F. Chen, J. Schacht, D. S. Pilch, T. Ben-Yosef, T. Baasov, J. Med.
Chem. 2009, 52, 2836.
[23] M. L. Magalhaes, J. S. Blanchard, Biochemistry 2005, 44, 16275.
[24] B. Ostash, A. Saghatelian, S. Walker, Chem. Biol. 2007, 14, 257.
[25] M. A. Shahid, M. Siddique, S.-U. Rehman, S. Hameed, A. Hussain, Am. J.
Food Technol. 2007, 2, 457.
[16] K. Hotta, T. Ogata, J. Ishikawa, M. Okanishi, S. Mizuno, M. Morioka, H.
Naganawa, Y. Okami, J. Antibiot. 1996, 49, 682.
Received: September 18, 2009
[17] F. Kobayashi, Y. Saino, T. Koshi, Y. Hattori, Antimicrob. Agents Chemother.
1980, 17, 337.
Published online on November 6, 2009
126
www.chembiochem.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 119 – 126