Modeling interactions of erythromycin derivatives with ribosomes

Article abstract of DOI:10.1134/S0006297915110127

Using a method of static simulation, a series of erythromycin A analogs was designed with aldehyde functions introduced instead of one of the methyl substituents in the 3′-N-position of the antibiotic that was potentially capable of forming a covalent bond with an amino group of one of the nucleotide residues of the 23S rRNA in the ribosomal exit tunnel. Similar interaction is observed for antibiotics of the tylosin series, which bind tightly to the large ribosomal subunit and demonstrate high antibacterial activity. Binding of novel erythromycin derivatives with the bacterial ribosome was investigated with the method of fluorescence polarization. It was found that the erythromycin analog containing a 1-methyl-3oxopropyl group in the 3′-N-position demonstrates the best binding. Based on the ability to inhibit protein biosynthesis, it is on the same level as erythromycin, and it is significantly better than desmethyl-erythromycin. Molecular dynamic modeling of complexes of the derivatives with ribosomes was conducted to explain the observed effects.

Full text of DOI:10.1134/S0006297915110127

ISSN 0006ꢀ2979, Biochemistry (Moscow), 2015, Vol. 80, No. 11, pp. 1500ꢀ1507. © Pleiades Publishing, Ltd., 2015.
Modeling Interactions of Erythromycin Derivatives
with Ribosomes
A. V. Shishkina1, T. M. Makarova2, A. G. Tereshchenkov2,
G. I. Makarov2, G. A. Korshunova1*, and A. A. Bogdanov1
1Belozersky Institute of PhysicoꢀChemical Biology, Lomonosov Moscow State University,
Moscow, Russia; Eꢀmail: korsh@genebee.msu.ru
2Lomonosov Moscow State University, Faculty of Chemistry, 119991 Moscow, Russia; fax: +7 (495) 939ꢀ3181
Received July 6, 2015
Revision received July 24, 2015
Abstract—Using a method of static simulation, a series of erythromycin A analogs was designed with aldehyde functions
introduced instead of one of the methyl substituents in the 3′ꢀNꢀposition of the antibiotic that was potentially capable of
forming a covalent bond with an amino group of one of the nucleotide residues of the 23S rRNA in the ribosomal exit tunꢀ
nel. Similar interaction is observed for antibiotics of the tylosin series, which bind tightly to the large ribosomal subunit and
demonstrate high antibacterial activity. Binding of novel erythromycin derivatives with the bacterial ribosome was investiꢀ
gated with the method of fluorescence polarization. It was found that the erythromycin analog containing a 1ꢀmethylꢀ3ꢀ
oxopropyl group in the 3′ꢀNꢀposition demonstrates the best binding. Based on the ability to inhibit protein biosynthesis, it
is on the same level as erythromycin, and it is significantly better than desmethylꢀerythromycin. Molecular dynamic modꢀ
eling of complexes of the derivatives with ribosomes was conducted to explain the observed effects.
DOI: 10.1134/S0006297915110127
Key words: ribosomal tunnel, macrolides, erythromycin, modeling
Macrolides are a family of natural or semisynthetic al Gramꢀnegative bacteria. The molecular target for
antibiotics with characteristic structures comprising a 14ꢀ macrolides is the 50S subunit of the bacterial ribosome.
16ꢀmembered macrocyclic lactone ring substituted with They are bound in the first third of the ribosomal tunnel
one or more sugar residues. Erythromycin and its derivaꢀ (RT) below the peptidyl transferase center (PTC). The
tives – roxithromycin, clarithromycin, and azithromycin – exact location and conformation of macrolides in the RT
are the most widely used in clinical practice [1]. This can were established by Xꢀray diffraction analysis (XRD) of
be explained among other things by the fact that the 14ꢀ ribosome in complex with antibiotics. The macrolides are
membered macrolides do not suppress functions of mitoꢀ bound to the RT elements via hydrophobic interactions of
chondrial and cytoplasmic ribosomes of eukaryotes and, their soꢀcalled “hydrophobic site” formed by unidirecꢀ
hence, are not toxic for humans [2].
The total synthesis of macrolides is a challenging via hydrogen bonds with bases in the 23S rRNA [3].
task due to their large number of asymmetric centers. Tylosin and related macrolides containing a 16ꢀ
tional orientation of hydrophobic substituents as well as
Erythromycin A, for example, contains 10 such centers membered ring with an 2ꢀoxoethyl group in the C6ꢀposiꢀ
only in the lactone ring, and this is why new antibiotics of tion comprise a unique family of ribosomal antibiotics.
this family are obtained via chemical modification of the When they bind to a ribosome, the aldehyde group of the
natural ones.
antibiotic reversibly reacts with the N6ꢀatom of the A2062
The macrolides belong to broadꢀspectrum antibiꢀ nucleotide of 23S rRNA, forming a hemiaminal (Fig. 1).
otics exhibiting activity against Gramꢀpositive and severꢀ
It was shown that transformation of the aldehyde
group in tylosin into an alcohol or methyl group resulted
in significant loss of antimicrobial activity [4]. We also
verified recently that the reduction of the tylosin aldehyde
group decreases its ability to compete with erythromycin
for the ribosomeꢀbinding site [5]. With the goal to
increase the antibiotic affinity to the ribosome, we set
Abbreviations: BODIPY, 4,4ꢀdifluoroꢀ4ꢀboraꢀ5,7ꢀdimethylꢀ
3a,4aꢀdiazaꢀsꢀindaceneꢀ3ꢀpentanoic acid; DIPEA, diisoꢀ
propylethylamine; Ery, erythromycin; LCꢀMS, liquid chroꢀ
matography/massꢀspectrometry; PTC, peptidyl transferase
center; RT, ribosomal tunnel; XRD, Xꢀray diffraction analysis.
* To whom correspondence should be addressed.
1500

INTERACTION OF MACROLIDES WITH RIBOSOMES
1501
Fig. 1. Schematic representation of hemiaminal formation between the aldehyde group of tylosin and the amino group of A2062 in 23S rRNA.
ourselves the task of modifying erythromycin based on (2.78 mmol) of BrCH2CH(OC2H5)2 and 243 μl
preliminary modeling in such a way that it becomes capaꢀ (1.39 mmol) of diisopropylethylamine (DIPEA) were
ble of forming in RT a hemiaminal with the 23S rRNA added. The reaction mixture was kept at 50°C for 72 h and
bases similarly to tylosin.
then poured into 20 ml of water. The product was extractꢀ
ed with ethyl acetate, washed with 10 ml of 5% NaHCO3
aqueous solution and then with 10 ml of saturated NaCl
aqueous solution, and dried with anhydrous MgSO4. The
solvent was removed under vacuum; the reaction product
MATERIALS AND METHODS
Erythromycin A from Fluka (Germany) as well as was precipitated by addition of hexane and purified using
reagents and solvents from Merck (Germany) and Reanal column chromatography in system (2). The yield of comꢀ
(Hungary) were used in this work. Silica gel 60 F 254 pound III was 25 mg (22%). TLC: Rf (1) 0.60, Rf (2) 0.50.
plates (Merck) were used for TLC. Silica gel 60 (0.040ꢀ LCꢀMS: found 837.1 [M + H+], calculated for
0.063 mm) from Merck was used for column chromatogꢀ C42H77NO15 – 836.0; tR = 1.63 min.
raphy. Chromatography was performed using the followꢀ
3′ꢀNꢀ(2ꢀoxoethyl)ꢀ3′ꢀNꢀdesmethylꢀerythromycin A
ing solvent systems: (1) AcOEt–MeOH–Et3N, 12 : 4 : 1; (IV). A sample of 25 mg (0.035 mmol) of compound
(2) AcOEt–MeOH–Et3N, 30 : 6 : 1; (3) CHCl3–MeOH, III was dissolved in 300 μl of CH2Cl2, and 15 mg
9 : 1. Molecular mass and chromatographic purity grade (0.042 mmol) of Bi(NO3)3⋅5H2O was added. The mixture
were confirmed with LCꢀMS using an Acquity UPLC was stirred well at room temperature for 6 h and then
System chromatograph (Waters, USA) and a TQD tandem poured into 10 ml of water; the product was extracted
quadrupole MS detector (Waters) in the positive ion mode with ethyl acetate, and the extract was washed with 5 ml
with gradient 0ꢀ95% CH3CN in 0.1% TFA for 3 min.
of 5% NaHCO3 solution and then with 5 ml of saturated
3′ꢀNꢀdesmethylꢀerythromycin A (II). The synthesis NaCl solution and dried over anhydrous MgSO4.
was conducted according to the protocol described in [6] Following solvent removal under vacuum, the yield of IV
from 5.00 g (6.81 mmol) of erythromycin A (I), 4.68 g was 18 mg (78%). TLC: Rf (1) 0.83, Rf (2) 0.76. LCꢀMS:
(34.4 mmol) of CH3CO2Na·3H2O, and 1.75 g (6.81 mmol) found 762.9 [M + H+], calculated for C38H67NO14 – 761;
of I2. The yield was 1.7 g (34%). TLC: Rf (1) 0.35. LCꢀMS: desosamine fragment in the LCꢀMS spectrum: found
found 721.0 [M + H+], calculated for C36H65NO13 – 719.9; 186.3, calculated for C9H16NO3 – 186.2; tR = 1.32 min.
′
′
desosamine fragment in the LCꢀMS spectrum: found
144.2, calculated for C7H14NO2 – 144.1; tR = 1.21 min.
3 ꢀNꢀ(1ꢀmethylꢀ3ꢀoxopropyl)ꢀ3 ꢀNꢀdesmethylꢀ
erythromycin A (V). A sample of 100 mg (0.139 mmol) of
′
′
3 ꢀNꢀ(2,2ꢀdiethoxy)ethylꢀ3 ꢀNꢀdesmethylꢀerythroꢀ compound II was dissolved in 300 μl (3.06 mmol) of croꢀ
mycin A (III). A sample of 100 mg (0.139 mmol) of comꢀ tonaldehyde and the solution was kept for 96 h at 50°C.
pound II was dissolved in 600 μl of CH3CN, and 418 μl The reaction mixture was evaporated to dryness. The
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015

1502
SHISHKINA et al.
product was isolated by column chromatography in sysꢀ tional to the intensity of luminescence recorded during
tem (3). The yield of compound V was 25 mg (23%). the cleavage of luciferin. The value of residual level of
TLC: Rf (3) 0.65. LCꢀMS: found 791.1 [M + H+], calcuꢀ translation was calculated as the ratio of luminescence
lated for C40H71NO14 – 789.9; desosamine fragment in intensity in the system with antibiotic to the level of lumiꢀ
the LCꢀMS spectrum: found 214.3, calculated for nescence intensity in the experiment without translation
C11H20NO3 – 214.2; tR = 1.32 min.
inhibitors.
Molecular dynamic simulation. A starting system was
Investigation of antibiotic binding to ribosomes with
fluorescence polarization method. 70S bacterial ribosomes designed based on the crystal structure of E. coli 70S riboꢀ
of E. coli, strain BW25113, isolated according to the stanꢀ some complex with erythromycin obtained using XRD
dard technique [7] and sensitive to macrolideꢀtype antibiꢀ method with 3.1 Å resolution (PDB code: 4V7U) [11].
otics were used in this work. The ribosome concentration The investigated derivatives were positioned in the RT
was determined from optical density at 260 nm (OD260
=
instead of the initial erythromycin. For further calculaꢀ
1 at C(Rs) = 24 pmol/ml). BIND buffer (20 mM tions, all ribosome residues were selected that have at
HEPESꢀKOH, pH 7.5, 50 mM NH4Cl, 10 mM MgCl2, least one atom in the cubic region with 7ꢀnm edge length,
0.05% Tween 20) was used for preparation of solutions. which included the entire RT and PTC, so that the cenꢀ
The dissociation constants of the investigated compounds ter of this region was in the tunnel, and the tunnel itself
were determined by competitive replacement of fluoresꢀ was located along the imaginary applicate axis.
cently labeled erythromycin (BODIPYꢀEry) in its comꢀ
All molecular dynamic calculations and analysis of
plex with ribosomes. Escherichia coli ribosomes were obtained trajectories of atom movements were conducted
incubated at 37°C for 15 min and then diluted in the with the GROMACS software package [12, 13] (versions
BIND buffer. BODIPYꢀEry (4 nM) obtained according 4.5.4 and 4.6.5) and force field parm99sb. All simulation
to the described technique [8] was incubated with riboꢀ experiments were accomplished at 300 K under control of
somes (17 nM) for 30 min at room temperature. Then a velocityꢀscaling thermostat with an additional stochasꢀ
solution of competing ligand (compounds I, II, IV, or V) tic term [14] and a Berendsen barostat [15]. Electrostatic
at concentrations from 1 nM to 10 μM was added to the interactions were treated via Ewald particle mesh [16].
formed complex in a 384ꢀwell plate so the final volume The modeled system was centered in a cubic cell with
was 50 μl. The resulting mixture was incubated for 1.5 h 10ꢀnm edge filled with water molecules TIP4P [17] so that
to reach equilibrium, and then fluorescence polarization the edges of the system were coated with the 0.40ꢀ0.65 nm
was recorded with a VICTOR™ X5 Multilabel Plate solvent layer. Negative charge of the system was compenꢀ
Reader (Perkin Elmer, USA). The excitation wavelength sated with magnesium and sodium ions. Magnesium(II)
was λex = 485 nm, and the recorded emission wavelength ions were placed so they form “magnesium bridges”
was λem = 530 nm. Gꢀfactor was corrected using a fluoꢀ between neighboring phosphate groups, and Na+ ions
rescence standard. The recording time for each well was were located close to negatively charged groups to comꢀ
1 s. The measured fluorescence polarization values were pensate the residual negative charge [18]. Residues with at
converted into anisotropy (mA). The data were approxiꢀ least one atom located within 0.1 nm from the face of
mated with a function corresponding to the solution of a modeled site of the ribosome were positionꢀrestricted,
cubic equation describing competitive binding of two ligꢀ while all other atoms could move freely. The integration
ands on a single site [9] using the GraphPad Prizm 6 comꢀ time step in all calculations was 2 fs, and coordinates were
puter program. The dissociation constant of the BODꢀ recorded in a trajectory files every 60 ps. Graphic repreꢀ
IPYꢀEry complex with ribosome was measured previousꢀ sentation of the data as well as static modeling of comꢀ
ly and was 16.4 nM. Dissociation constants of the comꢀ plexes of erythromycin A derivatives with E. coli ribosome
plexes of the derivatives with ribosomes were calculated were conducted with the PyMOL 1.5.1 program.
based on four replicate measurements. The confidence
interval (95%) was calculated for the presented dissociaꢀ
tion constant values.
RESULTS AND DISCUSSION
Translation inhibition in vitro. Escherichia coli S30
extract and Renilla luciferase mRNA were used for in vitro
Design and synthesis of erythromycin derivatives. In
translation. The reaction was conducted according to a the first stage of this study, we conducted a search for posꢀ
described protocol [10]. The mRNA concentration was sible sites for erythromycin A modification with the goal
20 ng/μl, antibiotic concentration – 1 μM. The measureꢀ to obtain derivatives that are capable of forming a covaꢀ
ments were conducted 1 h after the start of translation. lent bond with the amino group of one of the bases in the
For this purpose, a commercial Renilla Luciferase Assay binding site of macrolides with the RT. For this purpose,
System (Promega, USA) was added to the product of the XRD data of the erythromycin complex with E. coli
translation, and chemiluminescence intensity was ribosome (PDB code: 4V7U) were analyzed, and static
recorded with a multiꢀreader 20/20n (Turner Biosystems, modeling of new derivatives was performed using this
USA). The amount of synthesized luciferase is proporꢀ structure. The positions 6ꢀO and 3′ꢀN in erythromycin A
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015

INTERACTION OF MACROLIDES WITH RIBOSOMES
1503
The 3′ꢀNꢀdesmethylꢀerythromycin A (II) was
obtained from the initial erythromycin A (I) according
the known technique via reaction with iodine under modꢀ
erate heating in the presence of a base. Alkylation of comꢀ
pound II was conducted with 20ꢀfold excess of (2,2ꢀ
diethoxy)ethyl bromide at 50°C. The resulting diethylacꢀ
etal (III) was hydrolyzed under mild conditions by bisꢀ
muth (III) compounds [20]. HPLCꢀpure compound IV
was obtained (scheme).
Desosamine
3′ꢀNꢀ(1ꢀmethylꢀ3ꢀoxopropyl)ꢀ3′ꢀNꢀdesmethylꢀeryꢀ
thromycin A (V) was obtained via the Michael addition
reaction of crotonaldehyde with the secondary amino
group of 3′ꢀNꢀdesmethylꢀerythromycin A (II) (scheme).
The derivatives were purified to the HPLCꢀindividꢀ
ual state and their structures were affirmed with the LCꢀ
MS method. Localization of the substituent in the hydroꢀ
carbon moiety was confirmed from the LCꢀMS spectrum
based on modified desosamine fragment characteristic for
erythromycin derivatives.
Fig. 2. Suggested conformations of 2ꢀoxoethyl and 3ꢀoxopropyl
groups of 3′ꢀNꢀdesmethylꢀerythromycin A (only the desosamine
residue is shown) in the RT.
Measurement of dissociation constants of antibiotic–
ribosome complexes. Fluorescence polarization was used
for investigation of the affinity of the compounds to the
macrolide binding site in the ribosome. Fluorescently
labeled erythromycin (BODIPYꢀEry) [16] was displaced
from its complex with the ribosome by the investigated
compounds. The fraction of the displaced fluorescent
compound was determined from the level of the fluoresꢀ
cence polarization, and based on the known dissociation
constant for BODIPYꢀEry the effective dissociation conꢀ
stants (hereafter referred to as dissociation constant, Kd)
of the complexes of the erythromycin derivatives with the
ribosome were calculated. For this purpose, a series of
experiments was conducted in which the concentration of
ribosomes (17 nM) and BODIPYꢀEry (4 nM) were conꢀ
stant, and the concentration of the investigated comꢀ
pounds was varied from 1 nM to 10 μM (Fig. 3). Based on
these data, Kd of the complexes of derivatives IV and V as
well as of initial compounds I and II with ribosome were
calculated (table).
were suggested as the most suitable from the point of view
of closeness with amino groups of the bases in 23S rRNA:
the distance of the closest N6ꢀamino group of adenine to
the 6ꢀO atom is 5.2 Å (A2062), to the 3′ꢀN atom – 4.6 Å
(A2059). Based on the structure of the complex, the aldeꢀ
hydeꢀcontaining substituents to these atoms were identiꢀ
fied that could form a hemiaminal in the reaction with the
adenine amino group without changing the position of
the antibiotic.
In accordance with the modeling data, a 3ꢀoxoꢀ
propyl or even a 4ꢀoxobutyl fragment should be introꢀ
duced into the 6ꢀOꢀposition of erythromycin. However,
the bulky hydrophobic substituents could affect the posiꢀ
tion of erythromycin in the RT; moreover, they are not
able to provide sufficient rigidity of the nucleotide bindꢀ
ing with the antibiotic. It is also known that introduction
of the shorter 2ꢀoxoethyl substituent into this position
results in decreased antibacterial activity of the obtained
compound [19]. Considering this, we concentrated on the
synthesis of aldehyde derivatives of erythromycin at the
3′ꢀNꢀposition.
The table shows that the removal of one methyl
group from the erythromycin molecule (I) decreases
According to static modeling, replacement of one of
the 3′ꢀNꢀmethyl groups for the 2ꢀoxoethyl group allows
bringing the carbonyl atom close to the nitrogen atom of
A2059 to a distance of 1.9 Å (Fig. 2). Considering the regꢀ
ularly available flexibility of the antibiotic structure in the
RT and of the 23S rRNA nucleotides, a slight shift of the
desosamine residue or of the A2059 base could result in
minimization of distance limitations and formation of a
covalent bond. Introduction of the 3ꢀoxopropyl substituent
in this position also allows these atoms to come closer, so
the distance becomes sufficient for their interaction.
Hence, the erythromycin A derivatives containing an
aldehyde function at a distance of one or two methylene
groups from the nitrogen atom of desosamine were selectꢀ
ed as target compounds.
Dissociation constants of complexes of derivatives of
erythromycin A with ribosome
Compound
Kd, nM
I
3.5 0.9
II
IV
V
44
8
280 60
8.1 1.2
Note: Arithmetic means 95% confidence interval are presented.
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015

1504
SHISHKINA et al.
d
a
b
c
Synthesis of aldehydeꢀcontaining erythromycin A derivatives. a) I2, CH3COONa, MeOH/H2O; b) BrCH2CH(OEt)2, DIPEA, 50°C; c)
Bi(NO3)3·5H2O, CH2Cl2; d) CH3CH=CHCHO, 50°C
Scheme
affinity of the antibiotic to the ribosome by approximateꢀ introduction into compound II of the 1ꢀmethylꢀ3ꢀoxoꢀ
ly one order of magnitude (compound II). Introduction propyl group (compound V) increases stability of its comꢀ
of the 2ꢀoxoethyl substituent results in an even larger plex with the ribosome sixꢀfold, which becomes compaꢀ
decrease in affinity (compound IV). At the same time, rable with that of the initial erythromycin.
Hence, the affinity of the antibiotic to the ribosome
is rather sensitive to any modification at the 3′ꢀNꢀgroup,
removal of existing or introduction of a new substituent.
150
Most likely, this can be explained by the effect of this part
of the structure on the formation of a hydrogen bond
125
between the 2′ꢀOH of the antibiotic and N1 of A2058 of
100
the ribosome, which plays a key role in the interaction of
3
2
1
the antibiotic with the RT, because the distance between
3′ꢀN and 2′ꢀO atoms is only 2.9 Å. The importance of this
bond is corroborated by the fact that the antibacterial
activity of 2′ꢀ(deoxy)ꢀerythromycin is lower than that of
erythromycin by two orders of magnitude [21].
4
75
50
1
10
100
1000 10 000
Investigation of ability of antibiotics to inhibit translaꢀ
tion in vitro. Inhibitory activity was measured in a cellꢀ
free translation system of Renilla luciferase mRNA. The
differences in the inhibitory activity of the compounds
were most pronounced at antibiotic concentration of
1 μM (Fig. 4).
Qualitative correlation was observed between the
affinity to ribosome and ability to inhibit protein biosynꢀ
thesis: the decrease in biological activity in the series
I–V–II–IV corresponded to the same series in the reducꢀ
Antibiotic concentration, nM
Fig. 3. Displacement curves of BODIPYꢀerythromycin from its
complex with 70S E. coli ribosomes by investigated compounds with
constant concentrations of BODIPYꢀerythromycin (4 nM) and
ribosomes (17 nM). Curves: 1) erythromycin A (I); 2) 3′ꢀNꢀ(1ꢀ
methylꢀ3ꢀoxopropyl)ꢀ3′ꢀNꢀdesmethylꢀerythromycin A (V); 3) 3′ꢀ
Nꢀdesmethylꢀerythromycin A (II); 4) 3′ꢀNꢀ(2ꢀoxoethyl)ꢀ3′ꢀNꢀ
desmethylꢀerythromycin A (IV). Each point represents the arithꢀ
metic mean with standard deviation calculated from four replicates.
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015

INTERACTION OF MACROLIDES WITH RIBOSOMES
1505
in the trajectory (frequency around 7%), but the same
hydroxyl of the antibiotic forms a new stable hydrogen
bond with O6 of G2057 (frequency 76%). Moreover, due
to the large length of the hydrocarbon chain, the aldehyde
group of compound V is incorporated into the coordinaꢀ
tion sphere of one of the Mg2+ ions, where phosphate
groups G2056, G2057, and G2505 are also located (Fig.
5).
The energy of aldehyde coordination by magnesium,
~170 kJ/mol, is comparable with the energy of all other
noncovalent interactions of this compound with the riboꢀ
some (about 280 kJ/mol). The sum of these energies
(450 kJ/mol) is significantly higher than the energy of
noncovalent interaction of compound IV not coordinated
by the magnesium ions to the ribosome (350ꢀ370 kJ/mol)
(Fig. 6).
70
60
50
62
40
30
20
10
0
20
6
3
I
II
IV
V
Fig. 4. Relative values of translation suppression by compounds I, II,
IV, and V at their concentration of 1 μM and reaction time 60 min.
The results of molecular dynamic simulation explain
the reason for the drop in ribosome affinity to compound
tion in stability of the complexes of these compounds with IV: as suggested, the critically important bond of the 2′ꢀ
the ribosome; moreover, compound V was found to OH of the macrolide with N1 of A2058 was disrupted
demonstrate inhibiting ability at the level of erythromyꢀ though not due to the interferences with 3′ꢀNꢀsubꢀ
cin.
stituent, but due to the shift of the desosamine residue
Molecular dynamic modeling. To explain differences from the position optimal for the bond formation.
in the binding of the 3′ꢀNꢀmodified erythromycin A
The interactions not observed in the trajectory of the
derivatives containing an aldehyde group, molecular initial erythromycin were much stronger in the case of
dynamic modeling of compounds IV and V as well as of compound V (new hydrogen bond as well as coordination
the initial erythromycin (I) in a ribosome fragment conꢀ bond with Mg2+) than in the case of compound IV (one
taining the RT was conducted. Molecular dynamics canꢀ hydrogen bond). High lability of bonding of this comꢀ
not describe rearrangements of covalent bonds in the sysꢀ pound to the ribosome can be explained by the competiꢀ
tem, but it can describe dynamics of reaction centers tion between approximately equal hydrogen bonds of
prior to bond formation and stability of closely positioned compound IV with the RT that is in agreement with the
states. Moreover, molecular dynamics can predict hydroꢀ experimental results. Based on the data of molecular
gen bond formation based on the geometric position of dynamic simulation, the covalent bond between the aldeꢀ
potential donor and acceptor; hence, this method can hyde group of the resulting compound and the 23S rRNA
help to confirm or reject the hypothesis on the effect of
3′ꢀN substituents on hydrogen bond formation between
the 2′ꢀOH group of an antibiotic and the A2058
nucleotide. For example, the aldehyde group of the 3′ꢀNꢀ
Desosamine
(2ꢀoxoethyl)ꢀ3′ꢀNꢀdesmethylꢀerythromycin (IV) forms a
stable hydrogen bond with the amino group of A2059 with
84% frequency during the entire calculation process. This
seemingly favors high probability of covalent bond formaꢀ
tion between the mentioned groups; however, formation
of this hydrogen bond removes the desosamine fragment
from A2058, which results in the loss of interaction
between the 2′ꢀOH group and this nucleotide that is a key
element for erythromycin binding and leads to destabiꢀ
lization of the whole structure. According to the dynamic
modeling data, the frequency of the bond between the 2′ꢀ
OH group of the desosamine fragment and N1 of A2058
drops from 76% for erythromycin to 22% for compound
IV. It can be suggested that the new bond of the aldehyde
substituent with A2059 competes with the former conꢀ
Fig. 5. Magnesium ion coordinating aldehyde group of 3′ꢀNꢀ(1ꢀ
methylꢀ3ꢀoxopropyl)ꢀ3′ꢀNꢀdesmethylꢀerythromycin
phosphate groups of G2056, G2057, and G2505 residues. Only
the desosamine fragment of erythromycin is shown.
tacts of the desosamine fragment with A2058.
In the case of compound V, the hydrogen bond
between 2′ꢀOH and N1 of A2058 also is practically absent
A and
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015

1506
SHISHKINA et al.
a
b
–150
–150
desMeEry–CH2CHO
desMeEry–СН(СН3)CH2CHO
–200
–250
–200
–250
–300
–350
–400
–300
–350
–400
–350
–500
–550
–600
–350
–500
–550
–600
0
50
100
150
200
250
300
0
50
100
150 200
250
300
350
Time, ns
Fig. 6. Energy of noncovalent interactions of antibiotics with the ribosome according to the results of molecular dynamic simulation: a) for
compound IV; b) for compound V. The energy of noncovalent interactions is shown in gray; the solid black line indicates energy values
smoothed using the moving average with 8ꢀns time window, and the black dashed line – the moving standard deviation.
nucleotide is not formed: the carbonyl carbon during the
entire trajectory is in the plane of the base, while the
hemiaminal is formed as a result of attack of the elecꢀ
trophilic center by the unshared pair of electrons of the
nitrogen atom perpendicularly to the aromatic ring plane.
In other words, the observed relative position of reaction
groups is shifted by 90° relative to the required configuraꢀ
tion.
It is likely that the binding model including combiꢀ
nation of coordination and hydrogen interactions is realꢀ
ized for compound V, which is not exceeding in efficienꢀ
cy the one characteristic for erythromycin and other
macrolides, but is comparable with it. Hence, this model
can be used as a basis for design of new compounds capaꢀ
ble of binding to ribosomes of bacterial strains resistant to
erythromycin.
REFERENCES
1. Sunazuka, T., Omura, S., Iwasaki, S., and Mura, S. (2002)
Macrolide Antibiotics: Chemistry, Biology, and Practice, 2nd
Edn., Elsevier Science, p. 100.
2. Bottger, E. C., Springer, B., Prammananan, T., Kidan, Y.,
and Sander, P. (2001) Structural basis for selectivity and
toxicity of ribosomal antibiotics, EMBO Rep., 2, 318ꢀ323.
3. Hansen, J., Ippolito, J. A., Ban, N., Nissen, P., Moore, P.
B., and Steitz, T. A. (2002) The structures of four macrolide
antibiotics bound to the large ribosomal subunit, Mol. Cell,
10, 117ꢀ128.
4. Kirst, H. A., Toth, J. E., Debono, M., Willard, K. E.,
Truedell, B. A., Ott, J. L., Counter, F. T., FeltyꢀDuckworth,
A. M., and Pekarek, R. S. (1988) Synthesis and evaluation
of tylosinꢀrelated macrolides modified at the aldehyde
function: a new series of orally effective antibiotics, J. Med.
Chem., 31, 1631ꢀ1641.
5. Starosta, A. L., Karpenko, V. V., Shishkina, A. V.,
Micolajka, A., Sumbatyan, N. V., Schlunzen, F.,
Korshunova, G. A., Bogdanov, A. A., and Wilson, D. N.
(2010) Interplay between the ribosomal tunnel, nascent
chain, and macrolides influences drug inhibition, Chem.
Biol., 17, 504ꢀ514.
6. Freiberg, L. A. (1973) Process for the demethylation of 3ꢀ
aminomacrolides, US Patent 3725385.
7. Noll, M., Hapke, B., and Noll, H. (1973) Structural
dynamics of bacterial ribosomes, II. Preparation and charꢀ
acterization of ribosomes and subunits in the translation of
natural messenger RNA, J. Mol. Biol., 80, 519ꢀ529.
The authors are grateful to Dr. I. A. Osterman
(Faculty of Chemistry, Lomonosov Moscow State
University) for help in conducting in vitro testing.
This research is supported by the Russian Science
Foundation (project No. 14ꢀ50ꢀ00029, molecular modelꢀ
ing) and the Russian Foundation for Basic Research
(projects 13ꢀ04ꢀ00986ꢀa, synthesis, and 13ꢀ04ꢀ40211ꢀH,
ribosome interacting studies). The authors acknowledge
support from M. V. Lomonosov Moscow State University
Program of Development.
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015

INTERACTION OF MACROLIDES WITH RIBOSOMES
1507
8. Yan, K., Hunt, E., Berge, J., May, E., Copeland, R. A., and
Gontarek, R. R. (2005) Fluorescence polarization method to
characterize macrolide–ribosome interactions, Antimicrob.
Agents Chemother., 49, 3367ꢀ3372.
9. Wang, Z. X. (1995) An exact mathematical expression for
describing competitive binding of two different ligands to a
protein molecule, FEBS Lett., 360, 111ꢀ114.
10. Svetlov, M. S., Kommer, A., Kolb, V. A., and Spirin, A. S.
(2006) Effective cotranslational folding of firefly luciferase withꢀ
out chaperones of the Hsp70 family, Protein Sci., 15, 242ꢀ247.
11. Jack, A., Dunkle, J. A., Xiong, L., Mankin, A. S., and
Cate, J. H. D. (2010) Structures of the Escherichia coli riboꢀ
some with antibiotics bound near the peptidyl transferase
center explain spectra of drug action, Proc. Natl. Acad. Sci.
USA, 107, 17152ꢀ17157.
12. Van der Spoel, D., Lindahl, E., Hess, B., Groenhof, G.,
Mark, A. E., and Berendsen, H. J. C. (2005) GROMACS:
fast, flexible, free, J. Comput. Chem., 26, 1701ꢀ1718.
13. Hess, B., Kutzner, C., Van der Spoel, D., and Lindahl, E.
(2008) GROMACS 4: algorithms for highly efficient, loadꢀ
balanced, and scalable molecular simulation, J. Chem.
Theory Comput., 4, 435ꢀ447.
ics with coupling to an external bath, J. Chem. Phys., 81,
3684ꢀ3690.
16. Darden, T., York, D., and Pedersen, L. (1993) Particle
mesh Ewald: an N log(N) method for Ewald sums in large
systems, J. Chem. Phys., 98, 10089ꢀ10092.
17. Jorgensen, W. L., Chandrasekhar, J., and Madura, J. D.
(1983) Comparison of simple potential functions for simuꢀ
lating liquid water, J. Chem. Phys., 79, 926ꢀ935.
18. Athavale, S. S., Petrov, A. S., Hsiao, C., Watkins, D.,
Prickett, C. D., Gossett, J. J., Lie, L., Bowman, J. C.,
O’Neill, E., Bernier, C. R., Hud, N. V., Wartell, R. M.,
Harvey, S. C., and Williams, L. D. (2012) RNAꢀfoldꢀ
ing and catalysis mediated by iron(II), PLoS One, 7,
e38024.
19. Clark, R. F., Ma, Z., Wang, S., Griesgraber, G., Tufano,
M., Yong, H., Li, L., Zhang, X., Nilius, A. M., Chu, D. T.
W., and Or, Y. S. (2000) Synthesis and antibacterial activity
of novel 6ꢀOꢀsubstituted erythromycin A derivatives,
Bioorg. Med. Chem. Lett., 10, 815ꢀ819.
20. Eash, K. J., Pulia, M. S., Wieland, L. C., and Mohan, R. S.
(2000) A simple chemoselective method for the deprotecꢀ
tion of acetals and ketals using bismuth nitrate pentahyꢀ
drate, J. Org. Chem., 65, 8399ꢀ8401.
21. LeTourneau, N., Vimal, P., Klepacki, D., Mankin, A., and
Melman, A. (2012) Synthesis and antibacterial activity of
desosamineꢀmodified macrolide derivatives, Bioorg. Med.
Chem. Lett., 22, 4575ꢀ4578.
14. Bussi, G., Donadio, D., and Parrinello, M. (2007)
Canonical sampling through velocity rescaling, J. Chem.
Phys., 126, doi: 10.1063/1.2408420.
15. Berendsen, H. J. C., Postma, J. P. M., Van Gunsteren, W.
F., DiNola, A., and Haak, J. R. (1984) Molecular dynamꢀ
BIOCHEMISTRY (Moscow) Vol. 80 No. 11 2015