NMR and amber analysis of the neamine pharmacophore for the design of novel aminoglycoside antibiotics

Article abstract of DOI:10.1016/j.bioorg.2010.10.002

The dependence of the solution structure of neamine on pH was determined by NMR and AMBER molecular dynamics methods at pD 3.3, pD 6.5, and pD 7.4 in D₂O at 25 °C. Unlike neamine structures at pD 3.3 and 6.5, which essentially showed only one conformer, slowly exchanging primary, P-state, and secondary, S-state, neamine conformers populated on the NMR time scale at ～80% and ～20%, respectively, were detected at pD 7.4 with kinetic constants k_on(P→S) = 1.9771 s^-1 and k _off(S→P) = 1.1319 s^-1. A tertiary, T-state, neamine species populated at ～3% was also detected by NMR at pD 7.4. The pKa values determined by NMR titration experiments are pKa1 6.44 ± 0.13 for N3 of ring-II, pKa2 7.23 ± 0.09 for N2′ of ring-I, pKa3 7.77 ± 0.19 for N1 of ring-II, and pKa4 8.08 ± 0.15 for N6′ of ring-I. Ring-I and ring-II of the P-state neamine and ring-I of the S and T-states of neamine possess the ⁴C₁ chair conformation between pD 3.3 and pD = 7.4. In contrast, ring-II of the S and T-states of neamine most likely adopt the ⁶rH₁ half-chair conformation. The P and S-states of neamine exhibit a negative syn-ψ glycosidic geometry. The exocyclic aminomethyl group of ring-I adopts the gt exocyclic rotamer conformation around physiological pHs while the gg exocyclic rotamer conformation predominates in acidic solutions near and below pH 4.5. Neamine exists in the P-state as a mixture of tetra-/tri-/di-protonated species between pD 4.5 and pD 7.4, while the S-state neamine exist only in a di-protonated species around physiological pDs. The existence of the S-state neamine may facilitate binding of neamine-like aminoglycosides by favorable entropy of binding to the A-site of 16S ribosomal RNA, suggesting that novel aminoglycoside compounds carrying a S-state neamine pharmacophore can be developed.

Full text of DOI:10.1016/j.bioorg.2010.10.002

Bioorganic Chemistry 39 (2011) 28–41
Contents lists available at ScienceDirect
Bioorganic Chemistry
journal homepage: www.elsevier.com/locate/bioorg
NMR and amber analysis of the neamine pharmacophore for the
design of novel aminoglycoside antibiotics
b
b
c
Cenk A. Andac a, , Thomas C. Stringfellow , Ulfert Hornemann , Ningur Noyanalpan
⇑
a Department of Pharmacology, Medical School, Dicle University, Diyarbakir 21280, Turkiye
b School of Pharmacy, University of Wisconsin, Madison, WI 53705, United States
c School of Pharmacy, Gazi University, Ankara 06330, Turkiye
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 18 September 2010
Available online 23 October 2010
The dependence of the solution structure of neamine on pH was determined by NMR and AMBER molec-
ular dynamics methods at pD 3.3, pD 6.5, and pD 7.4 in D2O at 25 °C. Unlike neamine structures at pD 3.3
and 6.5, which essentially showed only one conformer, slowly exchanging primary, P-state, and second-
ary, S-state, neamine conformers populated on the NMR time scale at ꢀ80% and ꢀ20%, respectively, were
detected at pD 7.4 with kinetic constants kon(P?S) = 1.9771 sꢁ1 and koff(S?P) = 1.1319 sꢁ1. A tertiary, T-
state, neamine species populated at ꢀ3% was also detected by NMR at pD 7.4. The pKa values determined
by NMR titration experiments are pKa1 6.44 0.13 for N3 of ring-II, pKa2 7.23 0.09 for N20 of ring-I,
pKa3 7.77 0.19 for N1 of ring-II, and pKa4 8.08 0.15 for N60 of ring-I. Ring-I and ring-II of the P-state
neamine and ring-I of the S and T-states of neamine possess the 4C1 chair conformation between pD 3.3
and pD = 7.4. In contrast, ring-II of the S and T-states of neamine most likely adopt the 6rH1 half-chair
Keywords:
Aminoglycoside
Neamine
NMR
AMBER
Molecular dynamics
Clustering
conformation. The P and S-states of neamine exhibit a negative syn-
w glycosidic geometry. The exocyclic
Exchange rate constants
pKa
Transition state analysis
Boltzmann’s distribution law
aminomethyl group of ring-I adopts the gt exocyclic rotamer conformation around physiological pHs
while the gg exocyclic rotamer conformation predominates in acidic solutions near and below pH 4.5.
Neamine exists in the P-state as a mixture of tetra-/tri-/di-protonated species between pD 4.5 and pD
7.4, while the S-state neamine exist only in a di-protonated species around physiological pDs. The exis-
tence of the S-state neamine may facilitate binding of neamine-like aminoglycosides by favorable
entropy of binding to the A-site of 16S ribosomal RNA, suggesting that novel aminoglycoside compounds
carrying a S-state neamine pharmacophore can be developed.
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
however, can be toxic to mammalian cells (nephrotoxic and oto-
toxic in humans) at high doses and are prone to undergo a variety
Aminoglycoside antibiotics, such as gentamicin and kanamycin,
have long been used as efficient antimicrobial agents active against
both gram-positive and gram-negative bacteria as well as myco-
bacteria [1]. These antibiotics are considered to exert their antimi-
crobial activity mainly by interacting with bacterial ribosomes at
the aminoacyl-tRNA binding site, the A-site, which is constituted
by specific nucleotides of 16S ribosomal RNA in the decoding
region of their 30S subunits [2]. Antibiotic binding interferes
with the fidelity of mRNA translation, results in miscoding, and
leads ultimately to bacterial cell death [3]. The aminoglycosides,
of chemical modifications by different bacterial aminoglycoside
antibiotic inactivating enzymes [4,5].
Aminoglycosides bound to bacterial resistance enzymes exhibit
different conformations involving rotations around the two bonds
that constitute the glycosidic linkage extant between rings I and II
in comparison to the glycosidic linkages occurring in aminoglyco-
sides complexed with the A site of 16S rRNA [6,7]. The rotation
that is crucially affected in enzyme bound aminoglycosides is
characterized by the torsion angles
u w
(H10–C10–O4–C4)/ (C10–O4
–C4–H4), Fig. 1, which assume values around ꢁ21/152 for
neamine- or paromamine-like aminoglycosides in complex with
resistance enzymes. This results in an anti-
w geometry for rings I
Abbreviations: nea33/65/74, neamine at pD 3.3/6.5/7.4; nea65p/s, the primary/
secondary neamine species at pD 6.5; nea74p/s/t, the primary/secondary/tertiary
neamine species at pD 7.4; nea-4/3/2H, the tetra-/tri-/di-protonated state of the
primary neamine species; nea-2h, the di-protonated state of the secondary neamine
species.
and II [7,8], where N60 of ring I and N3 of ring II are anti-axially
oriented. In contrast, aminoglycosides in complex with the A site
of 16S rRNA adopt a negative syn-
w geometry, where U/w is about
ꢁ40°/ꢁ30° [8], where N60 of ring I and N3 of ring II are syn-axially
⇑
Corresponding author. Fax: +90 412 248 8440.
E-mail address: cenk_andac@yahoo.com (C.A. Andac).
oriented.
0045-2068/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.bioorg.2010.10.002

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
29
implemented on a neamine derivative, 1,20,3,60-tetrakis-N-trifluo-
roacetyl neamine, which was synthesized using the neamineꢂ4HCl
product as the starting material according to a chemical procedure
described by Hunziker et al. [13] An m/e ratio of 707.1 Da was
found in positive ion mode (M+) for the neamine derivative
(MW = 706.1 Da) by LC–MS. The purity of freshly synthesized nea-
mine batches were confirmed by NMR. 1H NMR (D2O, d = 4.8 ppm):
5.82 ppm (d, H), 3.94–3.85 ppm (m, 3H), 3.59 ppm (t, H), 3.51–
3.32 ppm (m, 5H), 3.24 ppm (m, H), 3.16 ppm (q, H), 2.4 ppm (m,
H), 1.8 ppm (q, H). Proton-coupled 13C (D2O): 97–95 ppm (d,
CH); 78.5–76.8 ppm (d, CH); 76.2–74.5 ppm (d, CH); 73.5–72
ppm (d, CH); 72.8–70 ppm (d, CH); 70–69 ppm (d, CH); 68.6–
67.8 ppm (d, CH); 54.5–53.2 ppm (d, CH); 50.8–49.4 ppm (d, CH);
49.3–48 ppm (d, CH); 42–39 ppm (t, CH2), 29.8–26.5 ppm (t, CH2).
2.2. Titration analysis and pKa
Fig. 1. Haworth formula of neamine-/paromamine-like aminoglycosides showing
atom numbering and the torsion angles
(H10–C10–O–C4), (C10–O–C4–H4) and
(H50–C50–C60–N/O60) for rings
and II. The DAG (2,6-diamino-2,6-dideoxy
U
w
x
I
A micro NMR-pH electrode from Mettler Toledo Analytical, Inc.
was used both for potentiometric and NMR titration experiments.
0.5 M NaOD and/or 0.5 M DCl solutions (Aldrich Chemicals) were
used to adjust the pD of ꢀ10 mM neamine solutions in D2O. pKa
values for the amino groups of neamine were determined by a line
fitting procedure via Origin Software v.6 (http://www.origin-
lab.com), which utilized Eq. (1) [14], by plotting 1H NMR chemical
shifts (d in ppm) for the primary vicinal protons of the neamine
amino groups (e.g., H3 for the N3 amino group of ring II) versus
pD (from
glucosyl) group is denoted as ring I and the 2-DOS (2-deoxystreptaminyl) group is
called ring II.
The increasing concerns about bacterial resistance and toxic ef-
fects have led to the development of semisynthetic aminoglyco-
sides which are less toxic and less sensitive to enzymatic
modification. Understanding the binding modes of bound amino-
glycosides in complex with 16S rRNA [9,10] and bacterial resis-
tance enzymes [7] at the molecular level recently has been a
major strategy to develop semisynthetic aminoglycoside analogs.
Extensive research and analysis on the binding interactions be-
tween RNA and aminoglycosides have proven that neamine,
Fig. 1, is the most suitable core structure that possesses antibiotic
activity and, therefore, has been commonly applied in the develop-
ment of synthetic analogs [11,12]. However, the structure of nea-
mine in the unbound state has not been fully explored so far.
Here, the unbound solution structures of neamine at pH 3.3, 6.5
and 7.4 at 25 °C in D2O are investigated by NMR and computational
techniques. pKa values for the amino groups of neamine are deter-
mined by H1 NMR titration experiments. Kinetic and NMR proper-
ties for the most populated two different neamine species found to
exist at near physiological pH values are revealed here, to the best
of our knowledge, for the first time in literature. Information
gained from the NMR and computational data are then utilized
to propose novel aminoglycoside antibiotics.
pD ¼ pKa þ log½ðdprot ꢁ dobsÞ=ðdobs ꢁ ddeprot
Þ
ð1Þ
3.3 to 8.7). dprot and ddeprot in Eq. (1) represent upper and lower
bounds for the protonated and deprotonated states of the corre-
sponding amino groups, respectively, while dobs is a chemical shift
in between the boundary states.
2.3. NMR analysis
NMR samples were weighted out and dissolved in 0.6 ml of D2O
in Eppendorf tubes to yield a final neamine concentration of
ꢀ10 mM. A micro NMR-pH electrode was used to adjust the pD
with 0.5 M NaOD and/or 0.5 M DCl solutions (Aldrich Chemicals).
O2 was removed by bubbling N2 gas through the samples. The
air-free NMR samples were transferred to 5 mm NMR tubes, which
were then immediately flame-sealed.
2. Materials and methods
400 and 500 MHz Varian Unity INOVA NMR spectrometers
(available in the School of Pharmacy at the University of Wisconsin-
Madison, WI USA) were used to acquire homunuclear 1D/2D NMR
2.1. Synthesis and characterization of neamine
spectra (1H, 13C, gCOSY90, gDQF-COSY, and 1H,1H-NOESY {tmix
=
Several batches of neamine were synthesized by acid hydroly-
sis of neomycin B as described by Wong et al. [11]. The acid
hydrolysis reaction of neomycin B was monitored by TLC using
a solvent mixture of BuOH/AcOH/H2O (2/2/1) and a solution of
1% ninhydrin as an indicator (Rf: 0.32 and 0.37) [11,13]. The
acidic precipitate formed at the end of the reaction was filtered
off and washed with ether several times to give the desired prod-
uct (Fig. 1) in salt form (neamineꢂ4HCl) as an off-white powder
(yield: 4.45 g, 9.5 mmol, 91%). For NMR samples, the neamine
product was further recrystallized twice in fresh methanol and
filtered off as white powders, which were then washed with
diethyl ether after each crystallization step. The remaining white
powders were vacuum-aspirated for 1 day to remove residual
methanol, ether and water.
75 ms}) using quad-nucleus NMR probes (the qn4549 probe for
the V400 MHz unit and the qn6121 probe for the V500 MHz unit)
and indirect NMR data (gHMBC) using a tunable triple-resonance
probe (the hcx4765 probe for the V500 MHz unit). All NMR data
were obtained at 25 °C in D2O. FID data were processed by Varian
NMR v6.1C and Mestre-C v.1.1,2.3a,3.5.1 (http://mestrelab.com)
NMR software packages. All 1D/2D NMR spectra were referenced
to the HDO peak at d = 4.8 ppm.
Anomeric/transglycosidic NOE cross-peaks were integrated on
2D-NOESY spectra and normalized by Mestre-C NMR software to
the corresponding inverted diagonal-peaks, H10 for the primary
species and h10 for the secondary species, whose integral was set
to ꢁ105 integral units (iu).
Dihedral angles for intra-residual vicinal protons were deter-
mined by an empirically generalized Karplus-like equation devel-
oped for conformationally rigid six-membered ring systems by
Haasnoot, De Leeuw and Altona (HLA) [15].
The HPLC-MS (Agilent 1100 HPLC-MSD SL Mass Spectrometer,
available in the School of Pharmacy at the University of Wiscon-
sin-Madison, WI USA) characterization of neamine was indirectly

30
C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
2.4. Determination of conformational exchange rate constants of
neamine
Exchange rate constants between the primary (P) and second-
ary (S) conformers of neamine, generically expressed by Eq. (2),
are kPS (=1/
(=1/ S) in Hz for conversion from S to P states of neamine, which
are inversely related to the time-life of the P ( P) and S ( S) states,
respectively. When the overall exchange is at equilibrium, the ex-
change constant can be defined as Kex = kPS/kSP
sP) in Hz for conversion from P to S states and kSP
s
s
s
.
Fig. 2. Haworth formula of 4-methoxy-2-deoxystreptamine (MODOS). MODOS was
used as a model compound to study ring II of neamine by Nudged Elastic Band
(NEB) [16–18] transition state analysis.
kPSð¼1=sP
Þ
*
sS
P )ð¼1= Þ S
ð2Þ
kSP
At slow exchange rates, kPS and kSP can be determined by Eqs. (3)
and (4), respectively,
expressed by Eq. (8), where
rotamers (gg, gt or tg), Pgg + Pgt + Ptg = 1, D
cal/mol between two energy minima, one of which must be the
lowest one. R is the ideal gas constant (1987 cal/K mol) and T is
the temperature in K (298.15 K).
imum itself then becomes zero.
x
refers to one of the three staggered
U(R) is the difference in
1=T2S ¼ 1=T2 þ kPS
ð3Þ
ð4Þ
1=T2P ¼ 1=T2 þ kSP
where T 2 is the transverse (spin–spin) relaxation time [1 1 5] in
DU(R) for the lowest energy min-
seconds for a spin in fast exchange, which can be determined by
D
UðRÞ
x
D
D
UðRÞgt=RTÞ
Px ¼ ½expðꢁ
=RTÞꢃ=½expðꢁ UðRÞgg=RTÞ þ expðꢁ
Eq. (5) in which
D
m1/2 is the line-width in Hz at the half height of
the NMR peak at fast exchange. Similarly, T 2P and T 2S are the
spin–spin relaxation times in seconds for the same spin sampling
P and S states at slow exchange rates, which can be computed by
þ expðꢁ UðRÞtg=RTÞꢃ
ð8Þ
D
Eq. (6) and Eq. (7), respectively, where
corresponding line-widths at the half height of the peak in P or S
Dm1/2(P) and Dm1/2(S) are the
3. Results and discussion
states at slow exchange.
3.1. pKa values
T2 ¼ 1=
T2P ¼ 1=
T2S ¼ 1=
p
D
m1=2
m1=2ðPÞ
m1=2ðSÞ
ð5Þ
ð6Þ
ð7Þ
Potentiometric titration experiments revealed that the amino
groups of neamine are all protonated (tetra-protonated state,
nea-4H) below pH 4.5 and deprotonated above pH 9 (Fig. 3).
pKa values for the amino groups of neamine were determined
by NMR titration experiments as described in Section 2.2. Chemical
shifts for the primary proton peaks vicinal to the amino groups
(e.g., H1 of N1Hþ3 ) were plotted versus pD (from 3.3 to 8.7).
Fig. 4A–D illustrate the best-fit pKa curves fitted by Origin Software
v.6 using Eq. (1) as a function of dobs in ppm (on the y-axis) and pD
(on the x-axis). The NMR chemical shifts (dobs) for the H3, H20, H1
p
D
p
D
2.5. Calculation
AMBER v10 [16] software package was used to determine the
molecular dynamics (MD) structures of the neamine species by
Particle Mesh Ewald method [16] in explicit solvent environment
and to implement Nudged Elastic Band (NEB) [16–18] transition
state analysis in implicit water environment. AM1-BCC [19] partial
atomic charges of molecules were determined by the antechamber
[20] module of AMBER v10. Parm99SB [21] parameters were used
to neutralize neamine structures with Clꢁ ions followed by solva-
tion in TIP3 water box. After 1000 iterations of a short minimiza-
tion and 10 ps of temperature equilibration steps, the solvated
neamine structures were exposed to 10 ns of molecular dynamics.
The primary neamine species, tetra-/tri-/di-protonated states (nea-
4H/3H/2H), were not restrained during explicit solvent simula-
tions. Ring II of the secondary neamine species, di-protonated state
only (nea-2h), was restrained to a 6rH1 relaxed half-chair con-
former during explicit solvent simulations. R.m.s.d. analysis, trans-
glycosidic distance measurements, H-bonding and cluster analysis
through the trajectories of the explicit solvent simulations were
implemented by the ptraj module of AMBER v10.
UF
and H60 (UF: upfield) proton peaks were used to determine the
pKa1 of N3 (pKa1-N3, 6.44 0.13, R2 = 0.96), Fig. 4A, pKa2 of N20
(pKa2-N20, 7.23 0.09, R2 = 0.97), Fig. 4B, pKa3 of N1 (pKa3-N1,
7.77 0.19, R2 = 0.94), Fig. 4C, and pKa4 of N60 (pKa4-N60,
8.08 0.15, R2 = 0.93), Fig. 4D, respectively.
Table 1 compares the pKa values reported in this paper by 1H
NMR titration studies in D2O (Table 1d and Fig. 4A–D) to the liter-
ature pKa values of neamine-related moieties of neomycin B by 15
N
NMR by Botto and Coxon [24] (Table 1a), gentamycin C1 by 1H
NEB transition state analysis was performed on a model com-
pound, 4-methoxy-2-deoxystreptamine (MODOS), shown in
Fig. 2. A NEB transition state pathway energy plot was constructed
out of 30 frames between two end-point ring conformers, 4C1 and
6S2, of MODOS.
Three staggered gg, gt and tg exocyclic rotamer populations for
the low-energy structures of the primary and secondary neamine
species were determined by the Boltzmann distribution law
[22,23]. According to the Bolztmann’s distribution law, a fractional
probability of a rotamer population P( at a given temperature is
Fig. 3. Potentiometric base-titration between pH ꢀ3.7 and ꢀ12.0 for neamine.
Neamine looses its buffering capacity below pH 4.7 and above pH 9.0. 10
increments of 0.5 N NaOH was used to titrate the neamine solution.
lL
x
)

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
31
Fig. 4. The pKa curves were determined using Eq. (1) for (A) pKa1 of the N3 amino group, (B) pKa2 of the N20 amino group, (C) pKa3 of the N1 amino group and (D) pKa4 of the
N60 amino group. The squared regression (R2), upper (dprot) and lower (ddeprot) chemical shift boundaries and the pKa values are printed next to the curves within the boxes.
The pKa values are also presented in larger type letters to the left of the curves.
Table 1
pKa values of neamine and neamine-related moieties of aminoglycoside antibiotics.
pKa1-N3
pKa2-N20
pKa3-N1
pKa4-N60
Neamine moiety of neomycin Ba
Neamine moiety of gentamycin C1b
Neamine – Sutrisno et al.c
5.74 0.04
6.19 0.40
6.35 0.20
6.44 0.13
7.55 0.04
7.40 0.20
n.r.
8.04 0.03
7.67 0.40
7.73 0.15
7.77 0.19
8.60 0.02
9.86 0.40
8.62 0.08
8.08 0.15
Neamine – Andac et al.d
7.23 0.09
n.r.: not reported.
a
By 15N NMR in H2O/D2O (85/15) by Botto et al. [24].
b
c
By 1H NMR in D2O by Schacht et al. [25].
By potentiometric titration in H2O by Sutrisno et al. [26], where pKa values were assigned by comparison to a and b.
d
Fig. 4A–D by 1H NMR in D2O (work in this thesis).
NMR by Schacht et al. [25] (Table 1b), and neamine by potentio-
metric titration in H2O by Sutrisno et al. [26] (Table 1c). Indeed,
Sutrisno et al. published three unassigned pKa values of
6.35 0.20, 7.73 0.15 and 8.62 0.08 out of four expected pKa
values. The pKa assignment given in Table 5c for Sutrisno et al. data
was estimated in this paper by comparing to the pKa values of
Botto et al. (Table 1a) and of Schacht et al. (Table 1b).
Considering the percent error ranges given in Table 1 for the
pKa values of the neamine work in this thesis and of the work of
the other authors, the difference between the corresponding pKa
values may range up to 1.23 pH/D units, which may possibly arise
from the different techniques applied to measure the pKa points.
Also, the presence of a substituent on neamine, as in neomycin B
or gentamycin C1, may bring about intra-molecular interactions
such as an inter-residual bifurcated H-bond from an N20 hydrogen
of ring I to O400 and O500 of ring III in neomycin B [27], which should
be considered potential factors for pKa shifts. It should also be
noted that there is usually some difference between pKa values
measured in H2O and D2O [28], whose relationship is arbitrary
and it very much depends on the compound studied, the ionic

32
C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
strength of the solvent(s), and the techniques used. At present a
correct/adequate extrapolation of pKa(D2O) values to aqueous
phase pKa(H2O) values is not possible when different titration
techniques are involved, as in the present case.
rise steadily between pH 6.9 and 8.0, giving the highest concentra-
tion of ꢀ18.4% in molar concentration (averaged out of several
determinations) at pD 7.4 at 25 °C, but rapidly fade outside this
maximal region.
Fig. 6 concentrates on the spectral regions of the H(h)2proR (on
the left) and H(h)2proS (on the right) proton peaks of a nea74 sam-
ple, Fig. 6A, showing the primary and secondary neamine species,
and another spectrum of the same neamine sample kept at room
temperature for 3 years, Fig. 6B, which yielded a trace amount of
a tertiary/micro neamine species (ꢀ3%), named nea74t for this rea-
son, associated with the smallest proton peaks marked with italic
lowercase letters at the bottom spectrum. To the best of our
knowledge, this is the first time that the presence of a secondary
and tertiary neamine species in low-salt media is being reported
in literature.
3.2. Presence of primary, secondary and tertiary neamine species and
proton chemical shifts
Basically three neamine samples at pD 3.3, 6.5 and 7.4 were
studied in details by 1D/2D NMR techniques. Neamine at pD 3.3
(nea33) was chosen to ascertain that in one set of experiments
all amino groups are fully in their protonated states (nea-4H). Nea-
mine at pD 6.5 (nea65) was chosen because it gains three proton
charges at this pD and it is assumed that the N20, N60 and N1 amino
groups are then completely protonated (nea-3H). Neamine at pD
7.4 (nea74) was selected to determine the unbound conformation
of neamine under physiological conditions. Presumably neamine
gaines 2 (nea-2H, where N60 and N1 amino groups are completely
protonated) or 3 (nea-3H, similar to nea65) proton charges at pD
7.4.
At least two distinct neamine species were discovered in differ-
ent concentrations, which exhibit slow exchange rates around
physiological pH, pD ꢀ7.4, and at pD 6.5 in low-salt media. They
are denoted as nea74p and nea65p, respectively, for the primary/
major and nea74s and nea65s, respectively, for the secondary/min-
or conformers. Fig. 5 shows an overlay plot of 1H NMR spectra of
neamine between pD 3.3 and 8.7. The h10 peaks of the secondary
neamine species are marked with arrows and lowercase. The H10
peaks of the primary neamine species, marked with H10 in Fig. 5,
resonate downfield from the h10 peaks. In this region these peaks
Proton chemical shifts (d, ppm) for rings I and II of nea33,
nea65p, nea65s, nea74p, and nea74s are summarized in Table 2,
which were determined by COSY, NOESY and HMBC type NMR
(proR,proS)
experiments. Spectral assignments for H60
prochiral pro-
tons were found to be equivocal. Therefore, the upfield H(h)60 peak
UF
is denoted H(h)60
and the downfield H(h)60 peak is denoted
H(h)60DF. The stereospecific assignment of the H/h2proR and H/
h2proS peaks revealed that H(h)2proR (ꢀ2.4 ppm) resonates down-
field from H(h)2proS (ꢀ1.8 ppm).
The H4 peaks of the primary neamine species at pD 3.3 and 6.5
were confirmed by gHMBC via a transglycosidic through-bond con-
nection from H10 of ring I to C4 of ring II (Appendix A). The unam-
biguous h4 peak assignment of the secondary neamine species at
pD 6.5 and 7.4 as well as the primary H4 peak of neamine at pD
7.4 was confirmed by NOESY based on the strongest transglycosidic
Fig. 5. Some selected overlayed NMR spectra between pD 3.3 and 8.7 showing chemical shifts of the 10 proton-peaks (marked with arrows) for ring I of the secondary
neamine species (and lower letter case). H10 chemical shifts of the primary neamine species are also shown around 5.8 ppm in the figure.

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
33
Fig. 6. Spectral regions of the H(h)2proR (d = 2.30 ppm) and H(h)2proS (d = 1.62 ppm) proton peaks of (A) a nea74 sample, showing the primary (ꢀ83%) and secondary (ꢀ17%)
neamine species, and (B) the same nea74 sample kept at room temperature for 3 years, indicating the presence of a trace amount of a tertiary/micro neamine species (3.4% in
concentration), where the percent molar concentrations for the primary and secondary species are ꢀ73% and ꢀ25%, respectively. ;: peaks belonging to the secondary species
nea74s and .: visible small peaks belonging to the tertiary species nea74t. The tertiary peaks are marked with italic lowercase letters while the secondary peaks are shown
with lowercase letters.
Table 2
Proton chemical shifts determined for different neamine species chemical shifts [d
(ppm)].
Proton chemical shifts given in Table 2 together with the stack
plots shown in Fig. 5 reveal that while a multitude of peaks of the
primary neamine species exhibit drastic upfield chemical shifts as
pD is increased, the chemical shifts of the secondary neamine spe-
cies are rather pD tolerant between pD 6.5 and 8.5. For instance;
the chemical shift difference for the H4 proton peaks of nea65p
and nea74p is 0.25 ppm while this difference is only 0.04 ppm
for the h4 proton peaks of nea65s and nea74s. As the pD is in-
creased from 3.3 to 7.4, the greatest upfield shifts occur with H10,
H20, H30, H1, H2proS, H2proR, H3 and H4 of the primary neamine
species.
nea33
nea65p
nea65s
nea74p
nea74s
Ring I
10
5.99
3.51
4.05
3.54
4.08
3.32
3.53
5.82
3.37
3.94
3.48
4.08
3.27
3.49
5.44
3.40
3.91
3.37
3.93
3.14
3.48
5.68
3.18
3.83
3.44
4.08
3.24
3.46
5.39
3.29
3.86
3.36
3.93
3.12
3.47
20
30
40
50
60(UF)
60(DF)
Ring II
It is clear that the greatest shifts between pD 3.3 and pD 7.4 in-
volve protons in the vicinity of the titratible groups with the lowest
pKa values such as pKa1-N3 (6.44 0.13) for the H4/H3/H2proR/S
shifts and pKa2-N20 (7.23 0.09) for the H30/H20/H10 shifts. As
shown in Fig. 3 a neamine amino group, presumably N3 of ring II
(Fig. 1), begins to shed protons above pD 4.7 and it is suggested
that an equilibrium constituted by two or more different proton-
ated primary states of neamine that are in fast exchange on the
NMR time scale begins to become established. A fast exchange
would cause the NMR spectrum to reveal only one main compound
with varying/shifting chemical shifts causing an apparent move-
ment of single peaks. Because the pKa values for the N1 (pKa3-
N1 = 7.77 0.19) and N60 (pKa4-N60 = 8.08 0.15) amino groups
are greater than pD 7.4, these amino groups of the primary
1
3.41
1.99
2.57
3.61
4.06a
3.76
3.65
3.29
1.72
2.37
3.28
3.73a
3.68
3.56
3.68
1.63
2.28
3.28
3.51b
3.62
3.52
3.20
1.60
2.28
3.14
3.57a
3.65
3.50
3.67
1.61
2.27
3.12
3.47b
3.62
3.48
2(proS)
2(proR)
3
4
5
6
UF: upfield. DF: downfield.
a
Confirmed by HMBC.
Confirmed by NOESY. All other assignments were determined by COSY-type
experiments.
b
NOE cross-peaks from h10 to h4 and from H10 to H4 (Appendix B),
respectively, which greatly facilitated the COSY assignment of the
rest of the ring II proton peaks.

34
C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
neamine species are mainly in their protonated forms at pD 7.4 and
thus the nearby proton chemical shifts are less affected by pH
changes between pD 3.3 and pD 7.4.
determined by Eq. (5). Dm1/2 in Eq. (5) requires a high temperature
NMR experiment to allow the H10 and h10 peaks to merge into a
single peak at fast exchange. However, such an experiment with
With pKa2-N20 being 7.23 0.09, Fig. 4B, it seems likely that the
secondary neamine species reaches a maximum concentration at
pD 7.4, where the di-protonated state of the secondary neamine
species (nea-2h) is more populated than its tri-protonated counter-
part (nea-3h). Thus, it is possible that only the nea-2h, where the
N60 and N1 amino groups are protonated, exists near the transition
state (described in Section 3.4). Presumably, nea-3h is not electro-
statically stable and is immediately converted to the tri-protonated
state of the primary neamine species (nea-3H) at fast exchange
rates, leaving nea-2h as the only secondary neamine species that
is in slow exchange with the primary neamine species in aqueous
medium at pD ꢀ7,4. As pD is decreased, the concentration of nea-
3H increases, which gradually decreases the population of nea-2h
down to zero at pD 5.7. When pD is increased, the concentration
of mono-protonated primary neamine species (nea-H), where only
the N60 amino group is protonated, increases lowering the concen-
tration of nea-2h. This proposed transition state theory seems to be
the most likely explanation why the secondary proton peaks are
sheltered from drastic shifts upon pH changes.
neamine was not carried out. Instead, Dm1/2 was taken from the
measured H10 peak width of 1.71 Hz of nea33 at 25 °C, assuming
that a single H10 peak at acidic pH at 25 °C would be virtually iden-
tical to a merged H10 peak at pD 7.4 at a higher temperature.
Based on the approximated
Dm1/2(P), Dm1/2(S) and Dm1/2 values
given above, the derived kPS, kSP and Kex (=kPS/kSP) rate constants
are listed in Table 3. The frequency (Hz) of the kPS and kSP exchange
rates are inversely proportional to the life-time of the P (
sP = 1/kSP)
and S ( S = 1/kPS) states, respectively. This means that nea74p
s
exists at a lower frequency of 2.63 Hz (kPS) with a longer sP of
0.38 s on the NMR time scale as compared to nea74s which ap-
pears at a relatively higher frequency of 3.85 Hz (kSP) due to a
shorter sS of 0.25 s. The exchange rate constant at equilibrium
(Kex) is less than 1 (0.77) suggesting that the direction of the ex-
change favors nea74p in aqueous solution at 25 °C. This appears
to be the first time that these kinetics parameters have been re-
ported for neamine existing in two conformational states at pD
7.4 at 25 °C.
Exchange rates between nea74t and other neamine species
Despite the fact that only a few tertiary proton assignments
were made for nea74t and that nea74t was not studied at other
pHs, it seems likely that the micro species exist in a di-protonated
state and that the chemical shifts are pD independent, just as dis-
cussed above for the secondary species nea-2 h.
could not be determined due to a very low signal-to-noise ratio.
3.4. Neamine endocyclic dihedral angles and ring conformers
Vicinal proton-proton coupling constants, 3JHH, listed in Table 4
were determined by deconvolution analysis on high resolution 1H
NMR spectra of neamine at pD 3.3, 6.5, and 7.4. The corresponding
3.3. Exchange rate constants
3
endocyclic dihedral angles were determined by applying the JHH
values in an empirically generalized Karplus-like equation devel-
oped by Haasnoot–Leuuw–Altona (HLA) [15] which includes
b-substituent corrections optimized for three and four substituents
in conformationally rigid six-membered ring systems. The result-
ing HLA torsion angles (HLA°) for nea33, nea65p, nea65s, nea74p,
nea74s and nea74t were reported in Table 2 as t (trans) for torsion
angles around 180°, g+ (gauche+) for those around + 60°, gꢁ
(gaucheꢁ) for those around ꢁ60°. The h3–h4 torsion angle for
nea74s is not reported in Table 4 as 3Jh3–h4 could not be determined
due to severe spectral overlaps on the h3 and h4 peaks. Also the
h3–h4, h4–h5, h5–h6, h6–h1 torsion angles of nea74t are not
reported because the h1, h3, h4, h5 and h6 peaks could not be as-
signed due to low signal-to-noise ratio.
As briefly mentioned in Section 3.2 and shown in Fig. 6A, nea-
mine at pD 7.4 (nea74) initially revealed two slowly exchanging
main species in low-salt media, which are termed primary
(nea74p) for the high concentration (ꢀ%82) species and secondary
(nea74s) for the low concentration (ꢀ%18) species. Exchange rates,
kPS and kSP in Eq. (2), between nea74p (P state) and nea74s (S state)
were determined using the spin–spin (transverse) relaxation times
(T 2) of nea74p and nea74s, T 2P by Eq. (6) and T 2S by Eq. (7),
respectively, in which
Dm1/2(P) and Dm1/2(S) were measured off the
line-width at the half height of the H10 and h10 proton peaks,
respectively, on the 1H NMR spectrum of nea74 as shown in
Fig. 7.
Dm1/2(P) and Dm1/2(S) were measured as 2.54 Hz and
2.95 Hz, respectively, in the absolute intensity mode so that the
peak heights could be compared to each other by the deconvolu-
tion (line-fitting) function of the Varian v6.1C software.
Determination of the exchange rate constants, kPS in Eq. (3) and
kSP in Eq. (4), also depends on T 2 at fast exchange rates, which is
The most likely theoretical ring conformers of neamine are 4C1,
4C1, 2S0 for ring I, presented in Table 5 4C1, 4C1, 6S2, 6rH1 (‘‘relaxed
half-chair’’) for ring II, presented in Table 6, whose HLA° torsion
angles were determined upon fusing neamine rings I and II substit-
uents with the carbon frame of the 4C1 conformations for ring I and
ring II of the kanamycin X-ray structure, Cambridge Structural
Database CSD ID: 610593 [29], the 4C1 conformation for ring IV
of neomycin B in complex with 16S rRNA, X-ray structure PDB
ID: 2ET 4 [30], and the 2S0 (of ring I) and 6S2 (of ring II) conforma-
tions for the iduronate residue of an heparin derivative in complex
with a basic fibroblast growth factor, NMR structure PDB ID: 1BFC
[31]. The rH conformer is thought to be an intermediate state
between the S and H conformers. The generic HLA° torsion angles
for the 6rH1 conformation of ring II of the secondary neamine struc-
ture were determined by NEB transition state analysis between 4C1
Table 3
Rate constants and life-times characterizing neamine conformational exchange at pD
7.4.
Fig. 7. Anomeric proton peaks H10 and h10 for the P and S states of neamine at pD
kPS(=1/sS
kSP(=1/sP
Kex(=kPS/kSP
)
)
nea74p ? nea74s
nea74s ? nea74p
nea74p M nea74s
2.63 sꢁ1 (Hz)
3.85 sꢁ1 (Hz)
0.77
7.4. M in the figure marks the line-widths at the half height (Dm
1/2) of the H10 and
h10 proton peaks assigned for the P and S states of neamine at pD 7.4 and 25 °C,
respectively.
)

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
35
Table 4
Experimental JHH (Hz) values and theoretical HLA (Haasnoot, De Leeuw and Altona) [15] dihedral angles (HLA°) of different neamine species.
3
nea33
nea65p
nea65s
nea74p
nea74s
nea74t
3JHH/HLA°
3JHH/HLA°
3JHH/HLA°
3JHH/HLA°
3JHH/HLA°
3JHH/HLA°
Ring I
10–20
20–30
30–40
40–50
Ring II
1–2(R)
1–2(S)
2(R)-3
2(S)-3
3–4
3.890/g+
10.957/t
9.272/t
9.851/t
3.802/g+
10.812/t
9.118/t
9.828/t
3.769/g+
10.305/t
9.191/t
9.410/t
3.832/g+
10.628/t
9.308/t
9.764/t
3.667/g+
10.526/t
9.343/t
9.745/t
3.63/g+
7.97/t
9.5/t
9.5/t
4.299/gꢁ
12.514/t
4.215/g+
12.617/t
10.185/t
9.265/t
4.277/gꢁ
12.504/t
4.195/g+
12.402/t
9.618/t
9.263/t
9.005/t
10.253/t
4.606/gꢁ
12.677/t
4.151/g+
12.677/t
3.240/ꢀꢁ120°
9.334/t
4.152/gꢁ
12.539/t
4.314/g+
12.409/t
9.519/t
9.278/t
9.305/t
10.084/t
4.493/gꢁ
12.598/t
4.151/g+
12.635/t
nd/nd
9.175/t
9.211/t
10.202/t
4.18/gꢁ
12.37/t
4.18/g+
12.37/t
nd/nd
nd/nd
nd/nd
nd/nd
4–5
5–6
6–1
9.283/t
10.374/t
9.360/t
10.089/t
g+: 60°; gꢁ: ꢁ60°; t: 180°; nd: not determined.
3
Table 5
conformer are not consistent with the experimental Jh4h5
Averaged 3JHH (Hz) and torsion angles (HLA°) for chair (C) and skew (S) conformers of
3
(9.33 Hz) and Jh3h4 (3.24 Hz) values, respectively, given in Table
ring I.
4 for nea65s. The assumption with the HLA equations is that the
3JHH values to be used is biased towards torsion angles 60° and
180° on stable structures of rigid six-membered rings, for which
coupling constants occur in the ranges 0–5.5 Hz and 7.5–13 Hz,
respectively [15]. Considering that there is noticeable conforma-
tional as well as vibrational flexibility concerning the C3–C4–C5
bonds of ring II of the secondary neamine species, the HLA° value
listed in Table 4 for the h4-h5 (t) and h3–h4 (ꢀꢁ120°) vicinal pro-
tons of nea65s, are possibly not representing the genuine dihedral
angles, which indeed should be about 130° and ꢀꢁ130°,
respectively.
Torsion
4C1
4C1
2So
3JHH
HLA°
g+
t
t
t
3JHH
HLA°
3JHH
HLA°
10–20
20–30
30–40
40–50
ꢀ2–4
ꢀ 2–4
ꢀ 2–4
ꢀ 2–4
ꢀ 2–4
gꢁ
gꢁ
g+
g+
ꢀ 2–4
ꢀ30°
ꢀ9–12
ꢀ9–12
ꢀ9–12
ꢀ 9–12
ꢀ 9–12
ꢀ 2–4
t
t
ꢀ ꢁ120°
HLA°: Torsion angle by HLA (Haasnoot, De Leeuw and Altona) [15]; g+: ꢀ60°; gꢁ:
ꢀꢁ60°; t: ꢀ180°.
and 6S2 ring conformers of MODOS (Fig. 2), Section 2.5. An energy
plot of the NEB transition state pathway for MODOS was con-
structed out of 30 frames as shown in Fig. 8, in which frame #16,
an intermediate energy structure of MODOS with 6rH1 conforma-
tion, was found to best fit the HLA° torsion angles given for nea65s
in Table 4. Therefore, the structure coordinates of frame#16 were
used to generalize HLA° dihedral angles in Table 6 for likely 6rH1
conformation of ring II.
Comparison of the data presented in Tables 5 and 6 to those
presented in Table 4 strongly suggests that rings I and II of the pri-
mary neamine species as well as ring I of the secondary species
adopt the 4C1 chair conformation, shown in Fig. 9A for ring I and
Fig. 9B for ring II.
As seen in Table 4 3Jh3–h4 of nea74s could not be determined due
to spectral overlaps which incapacitated deconvolution analysis
and coupling constant measurements for the h3 and h4 peaks of
nea74s. Nevertheless, the ring II structure of the secondary nea-
mine species is assumed to adopt a single conformation around
the physiological pH as the chemical shifts of the secondary proton
peaks are pH tolerant. Thus NMR and structural information gained
for ring II of nea65s also applies to the ring II structure of nea74s.
Ring I of the tertiary neamine species, nea74t, possess torsion
angles similar to those of nea65s and nea74s, Table 4, suggesting
the presence of a 4C1 chair conformer. It is known that the tertiary
neamine species (nea74t) exists in the lowest concentration (ꢀ3%)
and it takes much longer time to form compared to the secondary
neamine species. In addition, it is highly likely that the ring II struc-
ture for the tertiary neamine species does not differ much from
The NMR data for nea65s presented in Table 4 seem to be most
consistent with HLA° values for a 6rH1 conformer given in Table 6,
whose structure is shown in Fig. 9C. However it should be noted
3
that of the secondary species due to similar JHH values involving
3
that the theoretical JHH range given in Table 6 for the H4–H5
the h3/h2/h1 vicinal protons as reported in Table 4 for nea65s
and nea74t. Yet, the tertiary neamine species should be energeti-
cally higher (less stable) than the 6rH1 conformer of nea65s in
(1.5–4 Hz for a 130° dihedral angle) and H3–H4 (ꢀ2–5 Hz for
ꢀꢁ130° dihedral angle) vicinal proton systems of the 6rH1 ring II
Table 6
3
Generalized JHH (Hz) and torsion angles, HLA°, for chair (C), skew (S) and hypothetical relaxed half-chair, rH, conformers of ring II of neamine.
Torsion
4C1
4C1
2So
6rH1
3JHH
3JHH
HLA°
3JHH
HLA°
3JHH
HLA°
HLA°
1–2proR
1–2proS
2proR–3
2proS–3
3–4
4–5
5–6
6–1
ꢀ1.5–5
ꢀ11–13
ꢀ2–5
gꢁ
t
ꢀ1.5–5
ꢀ1.5–5
ꢀ2–5
ꢀ2–5
ꢀ1–4
ꢀ3–5
ꢀ3–5
ꢀ3–5
g+
gꢁ
gꢁ
g+
gꢁ
g+
gꢁ
g+
ꢀ1.5–5
ꢀ1.5–5
ꢀ2–5
g+
ꢀ1.5–5
ꢀ11–13
ꢀ8–3.5
ꢀ9–12
ꢀ2–5
gꢁ
t
g-
g+
t
g-
30° -g+
150°-t
ꢀꢁ130°
130°
t
ꢀ11–13
ꢀ9–11
ꢀ7–10
ꢀ7–10
ꢀ9–11
ꢀ2–5
g+
t
t
t
t
ꢀ0.9–1.7
ꢀ7–10
ꢀ7–10
ꢀ4.9–7.9
ꢀꢁ100°
t
t
ꢀ1.5–4
ꢀ7–10
ꢀ9–11
ꢀ150°
t
HLA°: Torsion angle by HLA (Haasnoot et al.) [15]; g+: ꢀ60°; gꢁ: ꢀ ꢁ60°; t: ꢀ180°.

36
C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
minima and the energy barriers, Aà and Bà, for the proposed 6rH1
ring II conformers of nea65s/74s and nea74t, respectively. These
energy barriers are not seen in Fig. 8 because the application of
existing algorithms of quantum chemistry and molecular dynamics
to study reaction mechanisms involving changes in pyranose and
cyclohexane conformations are not straightforward in times when
there is a multitude of possible conformations/constraints along
the pathway [32].
3.5. Through-space transglycosidic NOEs
Due to spectral overlaps, intensities for many of the intra-resid-
ual NOESY cross-peaks of the neamine species could not be re-
solved and inter-residual NOESY cross-peaks not involving H(h)10
of ring I were likewise essentially inaccessible. Nevertheless, a
majority of the transglycosidic NOESY cross-peaks from the ano-
meric proton H10 of ring I to H3, H4, and H5 of ring II were unam-
biguously located by their proton chemical shifts and they were
taken into account in the determination of transglycosidic neamine
geometries.
Fig. 8. Nudged Elastic Band (NEB) transition state pathway energy plot of 4-
methoxy-2-deoxystreptamine (MODOS). Position of the high-energy starting coor-
dinates of 4C1 chair [frame 1; 0.9213 kcal/mol], low-energy 4C1 chair [frame 5;
ꢁ13.8339 kcal/mol], lowest energy 4C1 chair [frame 11; ꢁ14.6616 kcal/mol], 6rH1
relaxed half-chair frame 16 (circled); ꢁ8.8066 kcal/mol], TSà transition state [frame
19; ꢁ5.5199 kcal/mol], 6S2 skew [frame 23; ꢁ8.0708 kcal/mol], and high energy
o
*
end-point coordinates of 6S2 skew [frame 30; 9.4554 kcal/mol] conformers are
o
marked on the plot.
D
G1 is the energy difference (5.855 kcal/mol) between 4C1 and
Transglycosidic H(h)10–H(h)3/4/5 NOESY cross-peak intensities
measured in phase-sensitive mode [33] are given in Table 7. It
should be noted that the h4 and h6 peaks of the secondary nea-
mine species, nea65s and nea74s, severely overlap, see Table 2.
Thus the h10–h4 NOE intensities given in Table 7 include a negligi-
ble amount of h10–h6 NOE volume. Also, the transglycosidic the (H/
h)10–(H/h)3 NOESY cross-peaks were found to be very weak (>5 Å)
which disabled NOE measurements in some cases.
6rH1,
D
Gà is the energy difference (9.1417 kcal/mol) between 4C1 and TSà and
DG2 is
the energy difference (6.5908 kcal/mol) between 4C1 and 6S2. For simplicity only
carbon structures are shown for 4C1, 6rH1, TSà and 6S2.
the NEB transition state (TS) pathway, frame 16 in Fig. 8, but much
less than the hypothetical TS for the 2H1 half-chair conformer,
frame 19 in Fig. 8, as the population at TS is zero. Giving that the
energy barrier between the 4C1 (frame 11 in Fig. 8) and 2S0 (frame
23 in Fig. 8) minima is too high to allow the 4C1 ? 2S0 conversion,
In all cases studied the transglycosidic NOE intensities, (H/h)10–
(H/h)3, (H/h)10–(H/h)4 and (H/h)10–(H/h)5, were found to be posi-
tive, indicating some degree of flexibility about the glycosidic
bonds [34].
D
Gà = 9.1417 kcal/mol, nea74t is more likely to be somewhere at
frame 17 (ꢁ7.1016 kcal/mol) in Fig. 8, a less stable structure that
approaches a half-chair conformation at the transition state.
It should be noted that the 6rH1 conformer of the secondary
(nea65s/74s) as well as the tertiary (nea74t) neamine species do
not sit in a minimum in Fig. 8. To clearly image this, a hypothetical
TS energy plot was sketched in Fig. 10 to show the expected energy
In NMR studies of neamine/paromamine-like aminoglycosides,
cross relaxation rates or magnitudes of transglycosidic NOE inten-
sities of IH1 –H3, IH1 –H4 and IH1 –H5 for H10–H3, H10–H4 and H10–H5
0
0
0
through-space connectivities, respectively, between ring I and II
are major determinants of the syn-
W and anti-W geometries
Fig. 9. Structures of neamine rings I and II. (A) 4C1 chair conformation for ring I of nea33, nea65p, nea65s, nea74p and nea74s. (B) Ring II of nea33, nea65p and nea74p. (C)
6rH1 relaxed half-chair conformation for ring II of nea65s and nea74s. Only protons attached to carbons are shown for simplicity.

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
37
Fig. 10. Hypothetical transition state pathway energy plot of MODOS. This plot is
presented to remedy the NEB transition state pathway energy plot shown in Fig. 8
which does not exhibit intermediary energy barriers Aà and Bà for the formation of
6rH1 of the secondary neamine species nea(65/74)s (#16 in the figure, the same as
frame 16 in Fig. 8) and of the tertiary neamine species nea74t (#17 in the figure, the
same as frame 17 in Fig. 8), respectively.
Fig. 11. NOESY spectrum of nea74. The primary and secondary transglycosidic
NOESY cross-peaks from (H/h)10 of ring I to (H/h)3/4/5 of ring II are boxed between
4.2 and 3 ppm along the F2 dimension. The NOESY FID was acquired with a mixing
time of 75 ms.
Table 7
Transglycosidic NOE intensities given in integral units (iu). The NOESY cross-peaks
were normalized by Mestre-C software to the corresponding inverted diagonal-peaks,
H10 for the primary species and h10 for the secondary species, whose integral was set
to ꢁ100,000 iu.
H10–H50 NOESY cross-peaks at ꢀ4.06 ppm, see Table 2. Given that
ring I of the neamine species studied adopts a 4C1 chair conforma-
tion, an average distance of ꢀ3.8 Å may be attributed to the
H10–H30 and H10–H50 distances of ring I based on molecular
dynamics studies of neamine. Thus, the majority of the overlapping
Protons
nea33
nea65p
nea65s
nea74p
nea74s
10–3
10–4
10–5
105a
1461b
181
n.d.
n.d.
n.d.
544
57
n.d.
687
20
0
NOESY cross-peak around 4.06 ppm is due likely to IH1 –H4 of
581c
581c
543d
n.d.
nea33. The transglycosidic H10–H5 (d = 3.65 ppm) and H10–H3
(d = 3.61 ppm) NOESY cross-peaks were observed to be mild (181
iu) and weak (105 iu), respectively, a finding which supports a
n.d.: not determined due to very low signal-to-noise ratio.
a
Overlaps with H10–H6 cross-peak.
syn-
W glycosidic geometry for nea33.
b
overlaps with H10–H30 and H10–H50 cross-peaks.
H10–H4 and H10–H5 NOESY cross-peaks of nea65p severely
overlap around 3.70 ppm with a total integral volume of 581 iu.
c
H10–H4 and H10–H50 cross-peaks overlap.
overlaps with h10–h6 cross-peak.
d
0
Nevertheless, a feeble IH1 –H3 observed at 3.28 ppm was indicative
of a syn-
W geometry. Thus the NOESY cross-peak around 3.70 ppm
is mainly attributed to the H10–H4 NOE of nea65p. A strong h10–h4
NOE (543 iu) of nea65s along with feeble h10–h3 (d = 3.28 ppm)
and h10–h5 (d = 3.62 ppm) NOEs strongly suggested that nea65s
described in Section 1. In addition, NMR studies in literature re-
vealed that aminoglycosides [27,34] as well as oligosaccharides
[35] exhibit a mixture of syn-
W and anti-W transglycosidic geom-
adopt a syn-
W glycosidic geometry.
etries in varying concentrations in free form in solution, the
amount of which is governed by electrostatic configuration and
pH [34]. Based on NMR data given in the literature these NOE
A strong H10–H4 NOESY cross-peak (544 iu) at 3.57 ppm was
observed for nea74p. In addition, a feeble H10–H3 NOESY cross-
peak at 3.14 ppm and a weak H10–H5 cross-peak (57 iu) at
0
0
0
intensities are rank ordered as IH1 –H4 ꢄ IH1 –H5 > IH1 –H3 for pre-
dominant syn- geometries (>95%) as in basic neomycin B [27]
at pH 9.7 and acidic neamine [27] at pH 4.5. The rank ordering fol-
3.65 ppm strongly suggested a syn-
sidic bonds of nea74p. strong NOESY cross-peak (687 iu)
W geometry about the glyco-
W
A
observed at 3.57 ppm was mainly attributed to the h10–h4 NOE
of nea74s. Similar to the NOE pattern observed with nea33/65s/
65p/74p, the feeble h10–h3 and weak h10–h5 (20 iu) NOE intensi-
ties of nea74s are much less than that of the h10–h4 peak, again
0
0
0
lows as IH1 –H5 ꢅ IH1 –H3 P IH1 –H4 for an equilibrium, for instance;
between syn- (47%) and anti- (53%) geometries at fast ex-
W
W
change rates as in acidic kanamycin A [34] at pH 4.4.
Transglycosidic NOESY cross-peaks for the primary and second-
ary neamine species appear in two different transglycocidic NOE
regions around physiological pH, including nea65p/s (data not
shown) and nea74p/s, Fig. 11, in which the transglycosidic cross-
peak regions are boxed and marked on F2 dimension accordingly.
This can only occur if two molecular species interact at slow ex-
change rates and in this case presumably it occurs between the
4C1 and 6rH1 conformers of ring II, but is not due to a possible slow
exchange between two distinct glycosidic geometries of neamine.
Even if a slow exchange existed around the glycosidic bonds, it
would be detected as split cross-peaks in a single transglycosidic
NOESY region but not in two transglycosidic NOESY regions.
strongly suggesting a syn-
W geometry.
In addition, it was found that the inter-residual (H/h)10–(H/h)4
NOE intensities were slightly greater than the intra-residual (H/
h)10–(H/h)20 NOE volumes, providing strong evidence for syn-
geometry for all neamine species studied.
W
Table 8
Transglycosidic proton-proton distances (Å), / (H10–C10–O–C4)/
w
(C10–O–C4–H4)
(O50–C50–C60–N60)* for the best
representative clustering coordinates out of 10 ns MD.
dihedral angles (°) and exocyclic rotamers
x
H10–H3
H10–H4
H10–H5
//
w
x
nea-4H
nea-3H
nea-2H
nea-2h
4.486
4.510
4.658
4.364
2.790
2.541
2.398
2.555
3.153
3.434
3.496
2.982
ꢁ41.5/ꢁ56.6
ꢁ39.3/ꢁ26.7
ꢁ26.7/ꢁ33.5
ꢁ53.4/ꢁ14.7
gg (%96)
0
0
Transglycosidic NOESY cross-peak intensities IH1 –H3, IH1 –H4 and
gt (%99.3)
gt (%98.9)
gt (%99.5)
0
IH1 –H5 were measured in phase-sensitive mode and are listed in
integral units (iu) in Table 8 for nea33, nea65p/s and nea74p/s.
The transglycosidic H10–H4 NOESY cross-peak of nea33 (1461 iu)
was found to overlap with the intra-residual H10–H30 and
*
Populations for the exocyclic rotamers were determined by the Boltzmann dis-
tribution law [22,23], Eq. (8).

38
C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
Fig. 12.
U/W plots of the neamine species corresponding to the last 5 ns of 10 ns MD. U W
(H10–C10–O–C4) on the x-axis was plotted versus (C10–O–C4–H4) on the y-axis for
(A) nea-4H (nea33), (B) nea-3H (nea65p and nea74p), (C) nea-2H (nea74p), and (D) nea-2h (nea65s, nea74s and possibly nea74t).
As a conclusion, magnitudes of the transglycosidic NOE intensi-
the glycosidic bonds which oscillate between negative syn-
W (U/
ties (I) presented in Table 8 for nea33, nea65p/s and nea74p/s were
W
= ꢁ50/ꢁ40) and positive syn-
W (U/W
= 10/0) geometries. These
0
0
0
determined as I(H/h)1 –(H/h)4 ꢄ I(H/h)1 –(H/h)5 > I(H/h)1 –(H/h)3. This rank
findings parallel the positive transglycosidic NOE connectivities
observed with the neamine species mentioned in Section 3.5,
ordering is a strong indication of a negative syn- (where the
W
W
dihedral angle is rotated counterclockwise) glycosidic geometry
adopted in high concentrations by neamine at varying pD values,
indicating flexible rotations at fast exchange about the syn-
geometry of the neamine species. Considering the magnitudes (I)
W
0
0
suggesting that the syn-
and independent of pH.
W
geometry of neamine is very stable
of the transglycosidic NOEs, where I(H/h)1 –(H/h)4 ꢄ I(H/h)1 –(H/h)5
>
0
I(H/h)1 –(H/h)3, the glycosidic fast exchange favors the negative
syn-W geometry in D2O.
3.6. Molecular dynamics analysis
Transglycosidic proton-proton distances for the best represen-
tative cluster structures are given Table 8. As seen in the Table,
the H(h)10–H(h)4 distances are within the range of 1.8–2.8 Å for
strong NOEs, those of H(h)10–H(h)5 are within the range of 2.8–
3.3 Å for medium NOEs and the H10–H3 distances are within the
range of 3.3–5.0 Å for weak NOEs [36] which are consistent with
the transglycosidic NOE intensity pattern presented in Table 7 as
10 ns of explicit solvent MD computations were implemented
for the primary neamine species, nea-4H/3H/2H, without using re-
straints and for the secondary neamine species, nea-2h, for which
ring-II was restrained to the endocyclic carbon atom coordinates of
MODOS with 6rH1 relaxed half-chair conformer, obtained by NEB
transition state analysis (frame #16 in Fig. 8). Ring I of the primary
and secondary/tertiary neamine species and ring II of the primary
neamine species adopted a stable 4C1 chair conformer during MD
computations.
0
0
0
IH1 –H4 ꢄ IH1 –H5 > IH1 –H3
.
It can be said that the H10–H4 distance of the primary neamine
species becomes shorter upon deprotonation of the amino groups
as pD increases, which brings about clockwise rotation about the
The
U
/
W
plots shown in Fig. 12A for nea-4H (nea33), Fig. 12B
glycosidic //
equilibrium between the amino groups at different pDs. For in-
stance; H10–H4///
w dihedral angles due possibly to varying electrostatic
for nea-3H (nea65p and nea74p), Fig. 12C for nea-2H (nea74p)
and Fig. 12D for nea-2h (nea65s, nea74s and nea74t) strongly sug-
gest negative syn-
w
is 2.79(Å)/ꢁ41.5(°)/ꢁ56.6(°) for nea-4H and
W
geometries, in which
U
/W
mostly populate
2.39(Å)/ꢁ26.7(°)/ꢁ33.5(°) for nea-2H (Table 8). As seen in the best
around ꢁ60/ꢁ40, with the primary and secondary neamine spe-
representative clustered neamine structures in Fig. 13A–C for nea-
cies. Moreover nea-2H and nea-2h possess some flexibility about
4H/3H/2H, respectively, the negative syn-w geometry is stabilized

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
39
Fig. 13. Representative cluster structures of (A) nea-4H, (B) nea-3H, (C) nea-2H and (D) nea-2h, which are in good agreement with NMR data. Green lines represent inter-
residual H-bonding interactions.
observed to increase by ꢀ0.2 Å as a result of the ring II conversion
from 4C1 to 6rH1. Such an increase is compensated by //
w rotation
by ꢁ27(°)/+19(°) about the glycosidic bonds of nea-2h. This in turn
disrupts the expected inter-residual bifurcated H-bond in the sec-
ondary neamine species (Fig. 13D), which provides a more flexible
attribution about the glycosidic bonds of the secondary neamine
species, and yet stable enough to preserve the negative syn-
geometry in solution.
w
According to the Boltzmann distribution law results given in
Table 8, nea-3H and nea-2H/h were found to adopt the gt exocyclic
rotamer while nea-4H favors the gg exocyclic rotamer. It is highly
likely that, when all amino groups are protonated, the electrostatic
repulsions between N60H3þ of ring I and N3Hþ3 of ring II forces the
exocyclic group of ring I to adopt the gg rotamer orientation. Such
an electrostatic tension is relieved when N3H2 of ring II is not pro-
tonated, yielding the gt exocyclic rotamer orientation.
3.7. Rational drug design
Aminoglycoside modifying enzymes are able to surmount the
energy barrier to convert the glycosidic syn-w geometry to anti-
w
geometry, which is the preferred conformation for the adenyla-
tion of kanamycin in the active site of ANT 4 [7]. To prevent the
modification of kanamycin-like aminoglycoside antibiotics by
ANT 4, Asensio et al. [6,8] synthesized a conformationally locked
neomycin B to immobilize the glycosidic rotations between ring I
and II. The locked neomycin B succeded to evade bacterial resis-
tance, but exhibited lower binding affinity than neomycin B to-
wards the A-site of 16S rRNA. In a similar vein to the locked
aminoglycoside strategy, Fig. 14A, we computationally determined
that an ethylmethylether tether placed between the N20 atom of
ring I and the O5 atom of ring II significantly enhances the binding
affinity of locked kanamycin B-like aminoglycosides (data not
shown).
Fig. 14. Rational drug design for novel aminoglycosides. (A) Use of a linker between
N20 of ring I and O5 of ring II, and (B) introduction of a double-bond between C4 and
C5 of ring II in addition to A.
by an inter-residual bifurcated H-bond from N3H of ring II to O50 of
ring I and the glycosidic O. Compared to the H10–H4 distance nea-
2H (2.398 Å), Table 8, the h10–h4 distance of nea-2h (2.555 Å) is
It is highly likely that the existence of a half-chair conformer on
ring II would facilitate the entropy of binding to the A-site of 16S

40
C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
Fig. A1. An exemplary gHMBC spectrum showing the C4–H10 transglycosidic connectivity for nea65p. This spectrum was obtained with a delay time of 1/2.jnxh (where
jnxh = 10 Hz for three-bond correlations). A strong three-bond correlation between C4 and H10) is circled at 5.82 ppm on F2 dimension with a dashed-line pointing to the C4–
1
H4 cross-peak (split by JC4–H4 ꢀ140 Hz).
rRNA. Therefore it is proposed in this paper that the presence of a
double-bond between C4 and C5 in a conformationally locked
kanamycin B derivative, as exemplified in Fig. 14B, would mimic
a half-chair conformer on ring II and ease-up the binding process
by a more favorable entropy of binding.
References
[1] T. Korzybski, Z. Kowszyk-Gindifer, W. Kurylowicz, Antibiotics, Origin Nature,
and Properties, American society for microbiology, Washington DC, 1978.
[2] B.P. Rosen, S. Mobashery, Resolving the Antibiotic Paradox: Progress in Drug
Design and Resistance, Plenum Press, New York, 1999.
[3] H.P. Rang, M.M. Dale, J.M. Ritter, P. Gardner, Pharmacology, Churchill
Livingstone, 1995.
Acknowledgment
[4] H. Umezawa, I.R. Hooper, Aminoglycoside Antibiotics, Springer Verlag, New
York, 1982.
[5] S.S. Hegde, F. Javid-Majd, J.S. Blanchard, J. Biol. Chem. 276 (2001) 45876–
45881.
[6] A. Bastida, A. Hidalgo, J.L. Chiara, M. Torrado, F. Corzana, J.M. P-Canadillas, P.
Groves, E. G-Junceda, C. Gonzalez, J. J-Barbero, J.L. Asensio, J. Am. Chem. Soc.
128 (2006) 100–116.
We would like to thank the TR-Grid e-Infrastructure of Turkey
for allowing us to use AMBER v10.
Appendix A
[7] L.C. Pedersen, M.M. Benning, H.M. Holden, Biochemistry 34 (1995) 13305–
13311.
See Fig. A1.
[8] J.L. Asensio, A. Hidalgo, A. Bastida, M. Torrado, F. Corzana, J.L. Chiara, E.G.
ˇ
Junceda, J. Canada, J.J. Barbero, J. Am. Chem. Soc. 127 (2005) 8278–8279.
[9] D. Fourmy, M.I. Recht, S.C. Blanchard, J.D. Puglisi, Science 274 (1996) 1367–
1371.
Appendix B
[10] Q. Vicens, E. Westhof, J. Mol. Biol. 326 (2003) 1175–1188.
[11] W.K.C. Park, M. Auer, H. Kaksche, C.-H. Wong, J. Am. Chem. Soc. 118 (1996)
10150–10155.
See Fig. B1.
[12] W.A. Greenberg, E.S. Priestly, P.S. Sears, P.B. Alper, C. Rosenbohm, M. Hendrix,
S.-C. Hung, C.-H. Wong, J. Am. Chem. Soc. 121 (1999) 6527–6541.
[13] R. Tona, R. Bertolini, J. Hunziker, Org. Lett. 2 (2000) 1693–1696.
[14] M.D. Reily, L.C. Robosky, M.L. Manning, A. Butler, J.D. Baker, R.T. Winters, J. Am.
Chem. Soc. 128 (2006) 12360–12361.
[15] C.A.G. Haasnoot, F.A.A.M. De Leeuw, C. Altona, Tetrahedron 236 (1980) 2783–
2793.
[16] D.A. Case, T.A. Darden, T.E. Cheatham III, C.L. Simmerling, J. Wang, R.E. Duke, R.
Luo, M. Crowley, R.C. Walker, W. Zhang, K.M. Merz, B. Wang, S. Hayik, A.
Roitberg, G. Seabra, I. Kolossváry, K.F. Wong, F. Paesani, J. Vanicek, X. Wu, S.R.
Brozell, T. Steinbrecher, H. Gohlke, L. Yang, C. Tan, J. Mongan, V. Hornak, G. Cui,
D.H. Mathews, M.G. Seetin, C. Sagui, V. Babin, P.A. Kollman, AMBER v10,
University of California CA, San Francisco Press, 2008.
[17] H. Jónsson, G. Mills, K.W. Jacobsen, Classical and Quantum Dynamics in
Condensed Phase Simulations, World Scientific, Singapore, 1998.
[18] G. Mills, H. Jónsson, Phys. Rev. Lett. 72 (1994) 1124–1127.
[19] S.L. Dixon, K.M. Merz Jr., J. Chem. Phys. 107 (1997) 879–893.
[20] J. Wang, W. Wang, P.A. Kollman, D.A. Case, J. Mol. Graph. Mod. 25 (2006) 247–
260.
[21] V. Hornak, R. Abel, A. Okur, B. Strockbine, A. Roitberg, C. Simmerling, Proteins
65 (2006) 712–725.
[22] K.N. Kirschner, R.J. Woods, Proc. Natl. Acad. Sci. USA 98 (2001) 10541–10545.
[23] D. Eisenberg, D. Crothers, Physical Chemistry with Applications to the Life
Sciences, Addison-Wesley, 1979.
[24] R.E. Botto, B. Coxon, J. Am. Chem. Soc. 105 (1983) 1021–1028.
[25] W. Lesniak, J.M. Laren, W.R. Harris, V.L. Pecoraro, J. Schacht, Carbohydr. Res.
338 (2003) 2853–2862.
[26] Sutrisno, Y. Baran, G.A. Lawrance, E.I. von Nagy-Felsobuki, Struct. Chem. 12
(2001) 189–195.
[27] J.L. Asensio, A. Hidalgo, I. Cuesto, C. González, J. Cañada, C. Vicent, J.L. Chiara, G.
Cuevas, J. Jiménez-Barbero, Chem. Eur. J. 8 (2002) 5228–5240.
[28] K. Popov, H. Rönkkömäki, L.H.J. Lajunen, Pure Appl. Chem. 78 (2006) 663–675.
[29] Y.A. Puius, T.H. Stievater, T. Srikrishnan, Carbohydr. Res. 341 (2006) 2871–
2875.
Fig. B1. Transglycosidic/anomeric NOE traces for nea74. The overlayed spectra
show transglycosidic NOE traces for nea74s (top) and nea74p (middle) extracted
from the corresponding 2D-NOESY spectrum acquired with a mixing time of 75 ms.
The corresponding 1H NMR spectrum is given at the bottom of the figure. ‘‘iu’’
values marked on the traces are transglycosidic integral units normalized to the
corresponding reference diagonal peak (H10 for the primary species and h10 for the
secondary species) measured on the 2D-NOESY spectrum. The reference diagonal
peak was set to 100,000 iu. ‘‘*’’ = very low signal-to-noise ratio. ‘‘n.d.’’ = not
determined. The strongest transglycosidic NOE cross-peaks from h10 to h4
(positioned at 3.47 ppm on F2 dimension with 687 positive integral units, top
spectrum) and from H10 to H4 (positioned at 3.57 ppm on F2 dimension with 544
positive integral units, middle spectrum) are clearly seen for nea74s and nea74p,
respectively.
[30] B. Francois, R.J. Russell, J.B. Murray, F. Aboul-Ela, B. Masquida, Q. Vicens, E.
Westhof, Nucl. Acids Res. 33 (2005) 5677–5690.
[31] D. Mikhailov, K.H. Mayo, I.R. Vlahov, T. Toida, A. Pervin, R.J. Linhardt, Biochem.
J. 318 (1996) 93–102.

C.A. Andac et al. / Bioorganic Chemistry 39 (2011) 28–41
41
[32] A.R. Ionescu, A. Bérces, M.Z. Zgierski, D.M. Whitfield, T. Nukada, J. Phys. Chem.
A 109 (2005) 8096–8105.
L.K. Nicholson, S. Pavlopoulos, D.G. Reid, M.S. Searle, W. Schmitt, P.J. Sweeney,
G. Wickham, NMR in Drug Design, CRC Press Inc., USA, 1996.
[33] D.J. States, R.A. Haberkorn, D.J. Rune, J. Magn. Reson. 48 (1982) 286–292.
[34] F. Corzana, I. Cuesta, F. Freire, J. Revuelta, M. Torrado, A. Bastida, J.J. -Barbero,
J.L. Asensio, J. Am. Chem. Soc. 129 (2007) 2849–2865. Supporting Information.
Glossary
ˇ
[35] J. Dabrowski, T. Kozár, H. Grosskurth, N.E. Nifantiev, J. Am. Chem. Soc. 117
NMR: Nuclear magnetic resonance
NEB: Nudged elastic band
TS: Transition state
(1995) 5534–5539.
[36] P.R. Andrews, C.H. Chang, R.M. Cooke, D.J. Craik, A.J. Edwards, J. Feeney, J.T.
Gerik, C.N. Hodge, J.W. Jaroszewski, H. Kessler, G.F. King, R. Konat, J. Mackay,