the conformational and tautomeric equilibrium of 5a,6-anhydrotetracycline in aqueous solution at ph...

The conformational and tautomeric equilibrium

of 5a,6-anhydrotetracycline in aqueous solution at pH 7

Helio F. Dos Santosa,*, Clebio S. Nascimento Jr.a, Paulo Belletatob,Wagner B. De Almeidac

aNEQC: Nucleo de Estudos em Quımica Computacional, Departamento de Quımica, ICE, Universidade Federal de Juiz de Fora (UFJF),

Campus Universitario Martelos, 36036-330, Juiz de Fora, MG, BrazilbDepartamento de Fısica, ICE, Universidade Federal de Juiz de Fora (UFJF), Campus Universitario Martelos, 36036-330,

Juiz de Fora, MG, BrazilcLQC-MM: Laboratorio de Quımica Computacional e Modelagem Molecular, Departamento de Quımica, ICEx, Universidade Federal de

Minas Gerais (UFMG) Campus Universitario Pampulha, 31270-901, Belo Horizonte, MG, Brazil

Received 12 November 2002; revised 3 February 2003; accepted 3 February 2003

Abstract

The conformational and tautomeric equilibrium in tetracyclines has been considered to be important for the action mode of

this class of antibiotics. In the present study the structures and thermodynamic properties for the distinct isomers of the 5a,6-

anhydrotetracycline (AHTC) derivative were calculated in gas phase and aqueous solutions for the LH2 ionized species, which

is predominant at pH 7. Ab initio HF and MP2 levels of theory were used in gas phase and SCRF, IPCM and PCM continuum

models in aqueous solution. The solvent effect was also analyzed through the Monte Carlo simulation. In the gas phase the

folded (FLD) form was found to be predominant at the HF and MP2 levels. At the HF level, the inclusion of polarization

functions on the heteroatoms (N and O) was found to be important to the stability order. The HF/3-21G and HF/6-31G results

showed the extended conformer (EXT) as the most favorable in gas phase. This is the same result obtained at the semiempirical

AM1 level. In aqueous solution, the equilibrium is shifted to the twisted (TWS) form when the PCM and Monte Carlo solvation

energies are considered. In gas phase and water solution the tautomer ionized at O11 was found to be preferred. In general the

results obtained show that the conformational and tautomeric equilibrium are solvent-dependent, with the stability order in

water solution determined by the overall effect of the solvent, being not localized on a specific molecular moiety.

q 2003 Elsevier Science B.V. All rights reserved.

Keywords: Tetracycline; Conformational-analysis; Continum-models; Monte-Carlo

1. Introduction

Tetracyclines (TCs) are broad-spectrum antibiotics

that act blocking the protein synthesis [1,2]. Two main

factors have been related to the biological potency: (i)

the conformational flexibility [3–11] and (ii) the

affinity of tetracyclines by metals ions present in the

organism [12–21]. The former involves the equili-

brium between three distinct conformers named

extended (EXT), twisted (TWS) and folded (FLD)

[22], being the relative population dependent on

0166-1280/03/$ - see front matter q 2003 Elsevier Science B.V. All rights reserved.

doi:10.1016/S0166-1280(03)00126-X

Journal of Molecular Structure (Theochem) 626 (2003) 305–319

www.elsevier.com/locate/theochem

* Corresponding author. Fax: þ55-32-3229-3314.

E-mail address: [email protected] (H.F. Dos Santos).

http://www.elsevier.com/locate/theochem

the solvent and pH [22]. Stezowski and co-workers

showed in a series of papers [3–10] that some TCs

derivatives change the conformation from TWS to

FLD when the solvent changes from water to organic,

respectively [5,7]. The role played by the solvent on

the conformation might be indirectly related to the

partition coefficient, which has been quantitatively

correlated with the biological response of TCs against

resistant cell [23].

On the other hand, the interaction of TC with metal

ions is involved in many different biological processes

[12–21]. Experimental studies [16] showed that the

fraction of the drug not bound to the proteins is found

in the complexed forms with Ca(II) and Mg(II). The

complexes with Fe(II), Zn(II) and Al(III) are also

present in the biological medium, being responsible

for a decrease in the bioavailability of the drug and

some observed side effects [15,19]. More recently it

has been proved the importance of the TCs–Mg(II)

complex for the resistance mechanism involving

proteins associated to the cellular membrane

[24–26]. These studies showed relevant structural

aspects related to the interaction modes of TCs with

cellular proteins that can be useful to develop new

derivatives with improved biological properties.

In order to contribute to the understanding of the

biological properties of TCs, we have used quantum

mechanical theory to investigate the physical chem-

istry processes related to the action mode of this class

of antibiotics [22,27–31]. In our previous studies [22,

27–31] the solvent effect has been included using

polarizable continuum models at semiempirical level

of theory. In the present work we analyze the

conformational and tautomeric equilibrium for the

LH2 ionized species of the 5a,6-anhydrotetracycline

(AHTC) derivative (Fig. 1). That ionized form is

present in aqueous solution at pH 7 [17] and is usually

involved in the coordination processes with distinct

metal ions [17–21]. Gas phase and aqueous solution

ab initio calculations are reported and it is also

described an adequate procedure to explicitly include

the solvent within the Monte Carlo simulation.

2. Theoretical methodology

2.1. Quantum mechanical

The geometries of the distinct conformations

(extended, twisted and folded) and tautomeric forms

involving the O10–O11 moiety of the LH2 species

(hereafter named LH2(O10) and LH2(O11) for the

molecules ionized at O10 and O11, respectively) were

fully optimized at the ab initio Hartree–Fock (HF)

level of theory. The following basis sets were used: 3-

21G, 6-31G, 6-31G(d) and GEN (3-21G for C and H

and 6-31G(d) for N and O). The stationary points on

the Potential Energy Surface (PES) were character-

ized as minima through the harmonic frequencies

analysis at the HF/3-21G and HF/GEN levels. The

electronic correlation effects on the equilibrium

structures were included using the Møller–Plesset

second order perturbation theory (MP2) in single

point energy calculations on the HF/GEN fully

optimized geometries (MP2/GEN//HF/GEN).

Within the quantum mechanical formalism the

solvent effect (water) was considered using conti-

nuum models: Self-Consistent Reaction Field (SCRF)

[32], Isodensity Polarizable Continuum Model

(IPCM) [33] and Polarizable Continuum Model

(PCM) [34]. The solvation energy was calculated

using the HF/GEN gas phase optimized geometries. In

the SCRF and IPCM approaches only the

electrostatic contribution for the solvation free energy

ðDGsolv; elÞ was considered (Eq. (1)) and for the PCM

method the cavitation ðDGsolv;cavÞ and repulsion–

dispersion ðDGsolv;rdÞ terms were also included.

DGsolv ¼ DGsolv;el þ DGsolv;cav þ DGsolv;rd ð1Þ

The calculations in solution were carried out in water

ð1 ¼ 78:39Þ being the set of parameters specified for

each model on the text. The ab initio calculations were

performed using the GAUSSIAN98 suit program [35].Fig. 1. Structure and numbering scheme of the 5a,6-anhydrote-

tracycline (AHTC).

H.F. Dos Santos et al. / Journal of Molecular Structure (Theochem) 626 (2003) 305–319306

2.2. Statistical mechanics

Although we do not intend to extend in the

simulation methodology, some comments about the

simulation procedures will be made. Any further

details can be obtained with the authors by request.

The Monte Carlo (MC) simulations were done

using the Metropolis algorithm as implemented in the

DIADORIM.TPT [36] program in the isobaric–

isothermal ensemble. A cubic box ðL ¼ 38:97 �AÞ

consisting of one solute (AHTC molecule) and around

1970 Tip4p [37] water molecules were built to

adequately hydrate the solute molecule. Cubic

periodic conditions were imposed to eliminate

boundary effects and two different spherical cut-off

were used for solvent–solvent interactions ðrc ¼ 8:5�AÞ and solute–solvent interactions ðrc0 ¼ 19:0 �AÞ:

The latter was chosen so that the estimated first solute

solvation shell securely does not interact with the first

solvation shell of the solute nearest neighbor images

in order to simulate an infinitely dilution.

Energies for the solvent–solvent and solute–

solvent interactions were calculated using the

Lennard-Jones plus Coulomb site–site potential

according to Eq. (2).

EðrijÞ ¼ 41ij

sij

rij

!12

2sij

rij

!6" #þ

qiqj

rij

ð2Þ

The Lennard-Jones parameters for the water mol-

ecules were taken from the Tip4p model [37] and

for the AHTC (LH2 species) we used the OPLS

force field parameters taken from BOSS [38]

program without optimization. The Lennard-Jones

parameters for cross interactions were calculated

with the rules: sij ¼ ðsisjÞ1=2 and 1ij ¼ ð1i1jÞ

1=2:

Atomic CHELPG charges and the solute molecular

geometries were taken from ab initio (HF/GEN)

calculations.

After a very large equilibration phase (11 £ 106

MC steps) with T ¼ 25 8C and p ¼ 1 atm we started

averaging configurations for the final results. Each

complete averaging configuration phase (1 chunk)

entailed 20 £ 106 MC steps for the thermodynamic

properties and a total of 4 chunks were made using

different initial configurations in order to verify the

reproducibility of the results. Final means and

standard deviations were obtained from this set of

chunks, as required by the MC procedure. A

preferential sampling [39] was applied in the average

phase in order to improve the water moves by three

times in a spherical shell of approximately 10 A from

the center of mass of the solute molecule.

To study the liquid structure around the solute

molecule a site–site radial pair distribution function

gijðrÞ were determined using the well known

relation:

gðr þ 1=2drÞ ¼nsimlð½r; r þ dr�Þ

nrandomð½r; r þ dr�Þð3Þ

where nsimlð½r; r þ dr�Þ is the average number of

particles whose distances from a given particle lie

within the range ½r; r þ dr� in the simulation and

nrandomð½r; r þ dr�Þ denotes the same property calcu-

lated from a completely random system at the same

density r :

nrandomð½r; r þ dr�Þ ¼ ð4p=3Þr½ðr þ drÞ3 2 r3� ð4Þ

The solute–solvent site–site pair distribution func-

tion gives the probability to find a specific solvent

site at a distance r from a specific solute site,

relative to a non-structured one. The presence of

hydrogen bonds between solvent and solute mol-

ecule can be inferred from such a function.

Table 1 shows some solution properties where we

can see that the typical thermodynamic solvent

properties could be well reproduced in the simu-

lations. Solvent heat of vaporization and density does

not diverge more than 2% from a pure liquid and the

solvent correlation functions (not shown) correctly

reproduce the bulk water Tip4p contours.

Table 1

Solution properties obtained from the Monte Carlo simulations for

the distinct forms of the LH–(O10) tautomer of the AHTC molecule

Conformations Density (g/cm3) DHvap (kcal/mol)

Extended (EXT) 1.008 ^ 0.001 10.69 ^ 0.01

Twisted (TWS) 1.012 ^ 0.001 10.68 ^ 0.01

Folded (FLD) 1.009 ^ 0.001 10.68 ^ 0.01

Experimental (pure water) 0.997 10.51

H.F. Dos Santos et al. / Journal of Molecular Structure (Theochem) 626 (2003) 305–319 307

3. Results and discussion

3.1. Structures

The HF/GEN fully optimized geometries of the

extended (EXT), twisted (TWS) and folded (FLD)

forms of the LH2(O11) are shown in Fig. 2. Only the

tautomer ionized at O11 is shown. The corresponding

structures for the LH2(O10) species are similar to the

LH2(O11) ones, except for the O10–O11 moiety (see

Fig. 1) where the tautomeric process occurs. Some

relevant structural parameters, calculated at the HF/

GEN level, are reported in Table 2 and compared with

the X ray experimental data for the fully protonated

AHTC molecule (LH3þ) [40].

The conformation named EXT, TWS and FLD can

be characterized by using some of the dihedral angles

reported in Table 2. The EXT(TWS) ! FLD inter-

conversion can be followed through the values of v1;

v2 and v3: For the EXT and TWS forms these

dihedral angles are on the range of 50–608 and for the

FLD conformer the values are close to 2608. The

v1 –v3 torsional angles define the spatial relative

position between the ring A and the rings system BCD

on TCs molecules (see Figs. 1 and 2). The

interconversion process EXT(TWS) ! FLD involves

a pseudo-rotation over the C4a–C12a bond. The

transition state (TS) structure optimized at the HF/

GEN level for the TWS ! FLD process involving the

LH2(O11) species is shown in Fig. 3a. In this

structure, v1 was found to be 22.08.

The EXT and TWS conformers are distinct relative

to the position of the dimethylamonium (DMA) group

at C4. This stereochemistry characteristic can be

quantified by using the dihedral angles v5 and v6: For

the EXT form these are close to 240 and 21708,

respectively and for the TWS conformer v5 , 21508

and v6 , 808: Furthermore, a structural parameter

that is also important to distinguish the EXT and TWS

forms is the intramolecular hydrogen bond involving

the N4–H group. In the EXT isomer the hydrogen

bond is N4–H· · ·O12a and in the TWS form it is N4–

H· · ·O3. At the HF/GEN level, the N4· · ·O12a (EXT)

and N4· · ·O4 (TWS) distances were found to be 2.621

and 2.513 A [LH2(O11)] and 2.627 and 2.513 A

[LH2(O10)]. By comparing the experimental data

with those obtained for the different forms, it can be

seen that the observed structure in the solid state for

Fig. 2. HF/GEN optimized structures for the distinct conformations

of the LH–(O11) tautomer of the AHTC molecule: (a) extended

(EXT); (b) folded (FLD) and (c) twisted (TWS).


the LH3þ species is classified as EXT (see Table 2).

The hydrogen bond distance N4· · ·O12a is 2.783 A in

accordance with the calculated value for the EXT

form. The HF/GEN optimized geometry for the

transition state involved with the EXT ! TWS

process is shown in Fig. 3b. For this structure v5

and v6 were found to be 289.4 and 143.68,

respectively.

The overall effect of the basis-set on the structure

was analyzed using the Root Mean Square (RMS)

deviation of the atomic positions. The optimized

geometries from our highest level of theory (HF/6-

31Gp) were taken as reference, being overlapped with

the optimized structures from HF/3-21G, HF/6-31G

and HF/GEN levels. The results are (in A for the EXT,

TWS and FLD forms, respectively): 0.157, 0.274,

0.129 (HF/3-21G|HF/6-31Gp), 0.082, 0.080, 0.110

(HF/6-31G|HF/6-31Gp) and 0.077, 0.058, 0.118

(HF/GEN|HF/6-31Gp). These values were obtained

for the LH2(O11) species. The analysis for the

LH2(O10) tautomer was also performed with the

results being essentially the same in a quantitative

sense. The greatest deviation from the HF/6-31Gp

structure was found at HF/3-21G showing that the

enlargement of the basis-set at least up to 6-31G might

be important to calculate the structure of AHTC

molecule. However, by comparing the RMS for

(HF/6-31G|HF/6-31Gp) and (HF/GEN|HF/6-31Gp) it

can be seen that the improvement of the basis-set on

the carbon and hydrogen atoms from 3-21G

(HF/GEN) to 6-31G (HF/6-31G) has no effect on

the geometries. The role played by the basis-set on the

structures is centered on the oxygen and nitrogen

atoms. The inclusion of polarization functions on

Table 2

Main dihedral angles (vi in degree) calculated for the LH–(O11) ionized form of the AHTC molecule (HF/GEN). The values in parenthesis are

for the LH–(O10) tautomer

Dihedral anglesa Extended Twisted Folded Expt.b

v1 :[O12a,C12a,C4a,H4a] 57.1 (56.8) 58.6 (58.3) 259.6 (260.4) 67.1

v2 :[C4,C4a,C12a,C1] 50.2 (49.9) 53.5 (53.6) 259.4 (259.7) 59.0

v3 :[C12,C12a,C4a,C5] 59.7 (59.2) 59.3 (58.8) 258.8 (259.3) 58.5

v4 :[C1,C2,C2am,O2am] 149.6 (149.5) 2168.6 (2169.1) 2165.2 (2165.1) 2176.9

v5 :[C3,C4,N4,Cm1] 244.4 (244.6) 2150.5 (2150.5) 2138.8 (2139.3) 245.0

v6 :[C3,C4,N4,Cm2] 2170.2 (2170.4) 79.0 (79.0) 88.3 (87.9) 2170.8

a The numbering scheme is represented in Fig. 1.b From Ref. [40].

Fig. 3. HF/GEN optimized structures for the transition states on the processes TWS ! FLD (a) and EXT ! TWS (b). The main dihedral angles

are (in degree): (a) v1 ¼ 22:0; v2 ¼ 22:7; v3 ¼ 18:0; v4 ¼ 2164:6; v5 ¼ 2151:7; v6 ¼ 77:2 and (b) v1 ¼ 66:2; v2 ¼ 61:7; v3 ¼ 64:8;

v4 ¼ 2165:4; v5 ¼ 289:4; v6 ¼ 143:6:


the heteroatoms also leads to an improvement on the

geometries, if we consider the HF/6-31Gp geometries

trustful.

Summarizing the structural analysis reported in

this section, it can be concluded that the less

expensive HF/GEN calculation might be considered

good enough to describe the structures of the distinct

conformers of the AHTC molecule. Although there is

no experimental data available for the LH2 ionized

form, the comparison of the calculated dihedral angles

important for the conformation definition with those

observed for the fully protonated form showed a very

good agreement (Table 2). A similar mixed basis-set

(GEN) was recently used to study the structure and

tautomeric processes in the azo-dye Sudan III [41]. In

general this methodology has been proved to be useful

to study large molecular systems at the ab initio level

of theory.

3.2. Thermodynamic analysis in gas phase

The thermodynamic quantities for the confor-

mational and tautomeric equilibrium involving the

LH2 species were calculated in gas phase and water

solution at different levels of theory. In Table 3, the

data for the conformational equilibrium in gas phase

are reported. At the HF/3-21G and HF/6-31G levels,

the most stable form was found to be the EXT, being

present on a relative concentration greater than 90%.

The inclusion of the thermal correction in the total

energy shifts the equilibrium toward the FLD and

TWS forms, but the EXT conformer is still the

predominant one. At the HF/3-21G and HF/6-31G

levels there is no significant change on the equilibrium

position due to the tautomeric process [LH2(O11) !

LH2(O10)]. For both tautomers the stability order is

EXT . FLD . TWS. It is important to refer to our

previous paper regarding the conformational analysis

of AHTC using the semiempirical method AM1 [22].

In that work the values found for DGg were (in kcal/

mol): 0.000 (EXT), 3.032 (FLD) and 7.535 (TWS)

[LH2(O11)], showing the EXT conformation as the

most abundant form in gas phase. For the LH2(O10)

tautomer the relative energies were essentially the

same [22]. So the HF/3-21G, HF/6-31G and AM1

levels predicted the same global minimum and

stability trend on the gas phase PES for the LH2

ionized form of the AHTC molecule.

Table 3

Relative energies (DEg) and Gibbs free energies (DGg; T ¼ 298:15 K and p ¼ 1:0 atm) calculated for the distinct conformers of the LH–(O10)

and LH–(O11) species in the gas phase at different levels of theory. The values are in kcal/mol and the Gibbs populations (%) are given in

parenthesis

Level of theory Extended Folded Twisted

DEg DGg DEg DGg DEg DGg

Conformers of LH–(O10)

HF/3-21Ga 0.000 0.000 (94) 2.358 1.685 (6) 5.523 4.841 (0)

HF/6-31Gb 0.000 0.000 (97) 2.695 2.022 (3) 4.670 3.988 (0)

HF/6-31Gpb 0.000 0.000 (8) 20.809 21.482 (91) 1.813 1.131 (1)

HF/GENc 0.000 0.000 (11) 20.986 21.265 (89) 4.022 3.406 (0)

MP2/GEN//HF/GENd 0.000 0.000 (11) 20.972 21.251 (89) 6.505 5.889 (0)

Conformers of LH2(O11)

HF/3-21Ga 0.000 0.000 (98) 3.374 2.401 (2) 5.919 5.125 (0)

HF/6-31Gb 0.000 0.000 (99) 3.515 2.538 (1) 5.079 4.285 (0)

HF/6-31Gpb 0.000 0.000 (19) 0.120 20.853 (79) 2.230 1.435 (2)

HF/GENc 0.000 0.000 (36) 0.032 20.334 (64) 4.465 3.829 (0)

MP2/GEN//HF/GENd 0.000 0.000 (42) 0.166 20.200 (58) 6.928 6.293 (0)

a The vibrational frequencies were scaled with the factor 0.9085 (see Ref. [43]).b Thermal correction for the Gibbs free energy (see Ref. [44]) from HF/3-21G calculation.c The vibrational frequencies were scaled with the factor 0.8929 (see Ref. [43]). The basis-set called GEN was definied as: 3-21G for C and H

and 6-31Gp for N and O.d The thermal correction for the Gibbs free energy from HF/GEN calculation.


The inclusion of polarization functions on all

heavy atoms (HF/6-31Gp) and on the N and O atoms

(HF/GEN) shifts the equilibrium toward the FLD

conformer. This effect is less pronounced for the

LH2(O11) isomer (see Table 3). At HF/6-31Gp the

FLD Gibbs populations were 91% [LH2(O10)] and

79% [LH2(O11)], being close to the results from HF/

GEN: 89% [LH2(O10)] and 64% [LH2(O11)]. The

main difference between the results from HF/6-31Gp

and HF/GEN is regarding the relative energy

calculated for the TWS form. At HF/6-31Gp it was

,1 kcal/mol and at HF/GEN it was ,3.5 kcal/mol,

showing that the TWS conformer is more favorable at

HF/6-31Gp than at HF/GEN level.

The electron correlation effect was accessed

through MP2 single point energy calculation

(MP2/GEN//HF/GEN). The results reported in

Table 3 show that the inclusion of the correlation

effect is less important than the basis set improve-

ment, being the equilibrium position kept unchanged

at MP2 level relative to HF/GEN. Once more the

greatest effect was observed for the TWS form.

However, the overall conformational distribution is

the same at MP2 and HF levels of theory (see values

in brackets in Table 3). So in gas phase considering

our best calculations it can be concluded that the FLD

form is predominant with the EXT conformer present

in the equilibrium on a relative concentration ranging

from 8–42% depending on the tautomer considered

and the level of theory used. In general the EXT

population is greater for the LH2(O11) isomer

(19 – 42%) than for the LH2(O10) tautomer

(8–11%). For the LH2(O11) species the energy

barriers for the interconversion processes EXT !

TWS and TWS ! FLD were calculated in gas phase

only at HF/GEN level. The values found were

respectively 7.374 and 11.367 kcal/mol.

In Table 4 the relative energy and relative

Gibbs free energy are reported for the LH2(O10) !

LH2(O11) tautomeric process considering the three

conformations and distinct levels of theory. As can be

seen the equilibrium is strongly shifted toward the

LH2(O11) tautomer, being the results non-sensitive to

the conformation and the level of theory considered.

On a systematic way the thermal correction slightly

shifts the equilibrium toward the LH2(O10) form.

In Section 3.3, the conformational and tautomeric

equilibrium are analyzed in water solution using

quantum mechanical continuum models and Monte

Carlo simulation. At the ab initio level, only the HF/

GEN calculations were performed once it was shown

for the gas phase that this level of theory is good

enough to predict the equilibrium position at a lower

computational cost.

3.3. Thermodynamic analysis in water solution

In the present work we used the standard Self

Consistent Reaction Field model (SCRF–Onsager)

and Polarizable Continuum Model (PCM) with the

solute cavity defined as an isodensity (IPCM) and

using the United Atoms approach (UA–PCM).

Table 4

Relative energies ðDEgÞ and Gibbs free energies ðDGgÞ calculated in gas phase considering the tautomeric process LH–(O10) ! LH–(O11) at

different levels of theory for the three conformers. The values are in kcal/mol

Level of theory Extended Folded Twisted

DEg DGg DEg DGg DEg DGg

Tautomeric process LH–(O10) ! LH–(O11)

HF/3-21Ga 24.689 23.805 23.673 23.089 24.292 23.520

HF/6-31Gb 24.871 23.987 24.054 23.471 24.462 23.690

HF/6-31Gpb 23.843 22.959 22.914 22.330 23.426 22.654

HF/GENc 24.473 24.379 23.455 23.448 24.030 23.956

MP2/GEN//HF/GENd 24.518 24.424 23.380 23.372 24.095 24.020

a The vibrational frequencies were scaled with the factor 0.9085 (see Ref. [43]).b Thermal correction for the Gibbs free energy (see Ref. [44]) from HF/3-21G calculation.c The vibrational frequencies were scaled with the factor 0.8929 (see Ref. [43]). The basis-set called GEN was definied as: 3-21G for C and H

and 6-31Gp for N and O.d The thermal correction for the Gibbs free energy from HF/GEN calculation.


The latter models are more accurate in the sense that

the solute cavity presents a real molecular shape.

According to Barone and co-workers [34], the UA–

PCM gives better results than the IPCM method for

solvation free energy in water.

The solvation free energies calculated from

distinct methodologies are reported in Table 5. For

the SCRF calculation, the spherical cavity radium

was set to 5.5 A. This is an average value for the

different forms. For each conformation the radium

was calculated using the VOLUME keyword in the

GAUSSIAN program [35]. In the IPCM method, the

isodensity used was 0.0005. This value was found to

be optimal in our quite recent work on the

conformational analysis of the herbicide DCMU

[42]. In the SCRF and IPCM approaches, only the

electrostatic contribution to the solvation free energy

was included (see Eq. (1)). In the UA–PCM

method, the electrostatic and non-electrostatic con-

tributions were taken into account. In the last line of

Table 5 the solute–solvent interaction energy

obtained from the MC simulation is reported.

These values were obtained considering four

simulations for each conformer following the

protocol described in Section 2.

By analyzing the values in Table 5 it is difficult to

draw conclusions about the preferred form of each

tautomer in aqueous solution. However, when the total

Gibbs free energy in solution ðDGaqÞ is considered, a

trend is observed according to the methodology used.

The values in Table 6 were calculated from the equation

below:

DGaq ¼ DGg þ ðDGsolvB 2 DGsolv

A Þ ð5Þ

Table 5

Solvation energies (DEsolv from Monte Carlo method) and solvation Gibbs free energies (DGsolv from continuum models) calculated for the

distinct conformers of the LH–(O10) and LH–(O11) tautomeric species. The values are in units of kcal/mol

Solvation model LH–(O11) LH–(O10)

Extended Folded Twisted Extended Folded Twisted

SCRF–HF/GENa,b 210.59 213.85 213.34 215.26 214.25 216.25

IPCM–HF/GENc 278.86 279.21 267.77 277.21 274.79 271.16

PCM–HF/GENd 271.01 271.07 280.99 271.86 271.41 281.20

Monte Carloe 2220 2222 2232 2225 2223 2231

a The solute cavity radius is 5.5 A.b The basis set called GEN was definied as: 3-21G for C and H and 6-31Gp for N and O.c The isodensity value was set to 0.0005.d The United-Atom Hartree–Fock (UAHF) method was used to construct the solute cavity.e The HF/GEN optimized geometries were used in the Monte Carlo simulation.

Table 6

Relative Gibbs free energies (DGaq in kcal/mol) calculated in aqueous solution (T ¼ 298:15 K and p ¼ 1:0 atm) for the distinct conformers of

the LH–(O10) and LH–(O11) species using different solvation models. The Gibbs populations (%) in water solution are given in parenthesis

LH–(O11) LH–(O10)

Extended Folded Twisted Extended Folded Twisted

SCRF–HF/GENa,b 0.00 (0) 23.59 (100) 1.08 (0) 0.00 (39) 20.26 (60) 2.42 (1)

IPCM–HF/GENc 0.00 (24) 20.68 (76) 14.92 (0) 0.00 (88) 1.16 (12) 9.46 (0)

PCM–HF/GENd 0.00 (0) 20.39 (0) 26.15 (100) 0.00 (0) 20.82 (0) 25.93 (100)

Monte Carloe 0.00 (0) 22.33 (0) 28.17 (100) 0.00 (1) 0.74 (0) 22.59 (99)

a The solute cavity radius is 5.5 A.b The basis set called GEN was definied as: 3-21G for C and H and 6-31Gp for N and O.c The isodensity value was set to 0.0005.d The United-Atom Hartree–Fock (UAHF) method was used to construct the solute cavity.e The HF/GEN optimized geometries were used in the Monte Carlo simulation.


where DGg are given in Table 3 (HF/GEN) and DGsolvB

correspond to the solvation energies calculated for the

conformers FLD and TWS (values in Table 5). The

DGsolvA refers to the solvation free energy calculated for

the EXT conformer (see Table 5), which was taken as

reference.

For the LH2(O11) tautomer, the FLD form was

found to be the most abundant conformer in solution

Fig. 4. Pair correlation functions ðgrÞ calculated for the O3 and O12a atomic sites of the LH2(O11) tautomer in distinct conformations.


using the SCRF and IPCM continuum models. In

these methodologies, the solvent effect shifts the

equilibrium toward the FLD form. For the LH2(O10)

species the effect is contrary to those observed for the

LH2(O11) tautomer, with the equilibrium shifted

toward the EXT form due to the solvent effect

according to the SCRF and IPCM approaches.

The UA–PCM results are unique in the sense that

the TWS form is the preferred one for both tautomers.

As has been pointed out by Machado and co-workers

[18], the AHTC molecule should be found in the TWS

form in aqueous solution at pH 7, giving support to the

UA–PCM conformational analysis. This same quali-

tative trend was observed from the MC simulation, as

shown in the last line of Table 6.

The conclusion drawn from these results is that the

TWS form should be preferred in aqueous solution,

although it was found to be the least stable in the gas

phase. This conclusion is in agreement with the

experimental finding [18]. An attempt to understand

the molecular reason for the greater solvation energy

calculated for the TWS form is given in Section 3.4,

where the solution structure is analyzed using the

configuration from the MC simulation. From the

values in Tables 4 and 5 it can also be seen that

the LH2(O11) tautomer is still the most stable species

in aqueous solution. Considering the UA–PCM

method, the calculated solvation energies is essen-

tially the same for the LH2(O11) and LH2(O10) being

a little bit more negative for the ionized species at O10

site. This is also the case for the MC results, except for

the TWS conformer that was found to be 1 kcal/mol

less solvated in the LH2(O10) tautomer (see Table 5).

3.4. Monte Carlo simulation: structure of solution

In the last part of this article we analyze the solution

structure from the results of the MC simulation. The

radial distribution functions (RDFs, gr) were calculated

considering distinct sites on the molecule. The EXT,

FLD and TWS conformers and the LH2(O10) and

LH2(O11) tautomers were analyzed.

For the atomic sites on the A ring the RDFs for

LH2(O10) and LH2(O11) are essentially the same, so

only those calculated for the most stable tautomer

[LH2(O11)] are discussed. In Fig. 4 the pair correlation

functions gr(O3–Ow/Hw) and gr(O12a–Ow/Hw) are

depicted. The gr profile observed for the enolic

oxygen (O3) shows that this site is involved in

hydrogen bonds with the solvent molecules, with well

defined peaks centered at 1.9 [gr(O3–Hw)] and 2.8 A

[gr(O3–Ow)]. The integration up to 2.55 [gr(O3–Hw)]

and 3.55 A [gr(O3–Ow)] gives 2–3 water molecules

Fig. 5. Supermolecules obtained from the MC simulations of EXT

(a), FLD (b) and TWS (c) conformers of AHTC molecule in water

solution. The last configuration was used for each isomer and only

water molecules hydrogen bonded to the O1, O3, O12 and O12a sites

are shown.


on average hydrogen bonded to the O3 site (Fig. 5).

The solvation structure on this site is not very

sensitive to the conformation. In the RDF calculated

for the FLD form, the peaks are smaller and wider,

indicating a less structured solvation shell around this

site. This is probably due to the fact that in this

conformation the A ring lies over the hydrophobic

ABC ring system (see Fig. 5b). The gr calculated for

the O12a site presented peaks centered at 1.9 and 2.8 A

for the FLD form and at 2.0 and 3.0 A for the EXT and

TWS conformers. The integration up 3.35 A

[gr(O12a–Ow)] gives 1–2 water molecules for the

FLD form and 0.5–1 water molecule for the EXT and

TWS forms (see Fig. 5). The solvent structure around

the O12a site is well defined in the FLD conformer. In

this form the O12a site is not steric blocked by the

hydrophobic methyl groups on N4. The hydrogen

bond is also better defined on the TWS form than in

the EXT one. The reason is that in the latter the O12a

site is involved an intramolecular hydrogen bond with

N4.

The O3 and O12a atoms are those directly

involved in the EXT ! TWS conformational equi-

librium. Considering the analysis of the RDFs

previously described, the solute–solvent interaction

in these sites could not explain the fact that the

TWS is about 10 kcal/mol more solvated than the

EXT and FLD ones. This solvation energy

difference is enough to shift the equilibrium toward

the TWS in aqueous solution. The position of the

amide group is slightly affected due to the

structural change, however the solvent structure

around this site is essentially the same for EXT and

TWS conformers as can be seen from Fig. 6.

The RDFs for O1 and O12 sites are shown in Fig. 7.

No significant differences between the distinct forms

were observed, being observed 2 water molecules

hydrogen bonded to each site on average (Fig. 5).

Once more the peaks calculated for the FLD form

show a less structured solvation shell around the O1

site, being explained by the same reason discussed for

the O3 moiety (Fig. 5b).

The O10 and O11 atomic sites are involved in

the LH2(O11) ! LH2(O10) tautomeric process. In the

LH2(O11) tautomer the O11 site is ionized and the

LH2(O10) the O10 site is deprotonated. As expected,

the solvent structure around these sites is not affected

with the conformational change that involves mainly

the groups on the A ring. Therefore, the following

discussion will be done considering only the TWS

form, which is the most stable form in aqueous

solution. In Fig. 8 the RDFs calculated for the O10 and

O11 sites are represented for both tautomers. First

analyzing the O11 site (Fig. 8a and b), it can be seen

well defined hydrogen bond peaks centered at 1.9 and

2.8 A when the molecule is ionized at O11. The

integration up to 2.55 [gr(O11–Hw)] and 3.25 A

[gr(O11–Ow)] gives 2 water molecules on average.

This same profile was not observed in the RDF for the

LH2(O10) species (Fig. 8b). An interesting result was

observed when the O10 site was analyzed (Fig. 8c and

d). As expected the solvent structure is well defined

in the LH2(O10), where the O10 site is ionized,

however the peaks are also characteristic of hydrogen

bond on the LH2(O11) tautomer. The supermolecules

Fig. 6. Pair correlation functions ðgrÞ calculated for the Oam atomic site of the LH2(O11) tautomer in the EXT and TWS conformations.


involving the O10 and O11 sites for the LH2(O11) and

LH2(O10) tautomers are shown in Fig. 9. Despite the

larger number of hydrogen bond observed for the

LH2(O10) tautomer, the solvation energy calculated

from the MC simulation was found to be the same as

those for the LH2(O11) (see Table 5).

Finally, the conclusion from this last analysis is

that despite the differences on the solvation processes

Fig. 7. Pair correlation functions ðgrÞ calculated for the O12 and O1 atomic sites of the LH–(O11) tautomer in distinct conformations.


Fig. 8. Pair correlation functions ðgrÞ calculated for the O11 and O10 atomic sites of the LH2(O11) and LH2(O10) tautomers in the TWS

conformation.

Fig. 9. Supermolecules obtained from the MC simulations of LH2(O11) (a) and LH2(O10) (b) of AHTC molecule in the TWS conformer. The

last configuration was used for each tautomer and only water molecules hydrogen bonded to the O11 O10 sites are shown.


for the conformers and tautomers of AHTC molecule

in pH 7, the stability order in water solution is

determined by the overall effect of the solvent, being

not localized on a specific moiety of the molecule.

4. Conclusions

The conformational and tautomeric equilibrium for

the 5a,6-anhydrotetracycline (AHTC) at pH 7 was

analyzed in gas phase and water solution using

quantum mechanical ab initio methods. The tautomer

ionized at O11 [LH2(O11)] was found to be favorable

in gas phase and aqueous solution, being independent

on the level of theory applied. On the other hand, the

conformational equilibrium is much more sensitive to

the solvent effect and also to the level of theory

applied. In gas phase, the results from HF/3-21G and

HF/6-31G showed that the extended (EXT) form is

the global minimum on the PES. This is the same

result as found at semiempirical AM1 and PM3

calculations. The inclusion of d polarization functions

on N and O only, shifts the equilibrium toward the

folded (FLD) conformer that was found to be present

on a ratio raging from 64% [HF/GEN; LH2(O11)] to

89% [HF/GEN; LH2(O10)]. At the HF/6-31Gp level

(with inclusion of polarization functions on all heavy

atoms) the conformational distribution was essentially

the same (relative population of the FLD form): 79%

[HF/6-31Gp; LH2(O11)] and 91% [HF/6-31Gp;

LH2(O10)], showing that the effect of the improve-

ment of the basis-set of the carbon atoms on the

conformational equilibrium is not significant. The

electronic correlation effect was found to be less

important than the increase of the basis set. At the

MP2/GEN//HF/GEN level the FLD form is present at

equilibrium on a ratio of 58% [HF/GEN; LH2(O11)]

and 89% [HF/GEN; LH2(O10)]. So, considering our

higher levels of theory the stability order in gas phase

was FLD . EXT . TWS.

In water solution, the SCRF and IPCM continuum

models showed an equilibrium between the EXT and

FLD forms and the PCM showed the TWS conformer

as the most stable. This last finding is in accordance

with the Monte Carlo simulation which is supported

by the experimental proposal. The analysis of the

solution structure from the RDFs calculated for

distinct sites on the molecule showed that the higher

solvation energy of the TWS form should be treated as

an overall solvent effect, not attributed to a specific

site of the molecule. So, in aqueous solution the

stability order was found to be TWS . FLD . EXT.

Acknowledgements

This research was supported by the Brazilian

agencies Conselho Nacional de Desenvolvimento

Cientıfico e Tecnologico (CNPq) and Fundacao de

Amparo a Pesquisa do Estado de Minas Gerais

(FAPEMIG). C.S.N, Jr. thanks the Fundacao Coorde-

nacao de Aperfeicoamento de Pessoal de Nıvel

Superior (CAPES) for the research grants. The

authors thank the CENAPAD-MG/CO-NAR-UFJF

for providing computational facilities.

References

[1] A.I. Laskin, J.A. Last, Antibiot. Chemother. 17 (1971) 1.

[2] D. Voet, J.G. Voet, Biochemistry, second ed., Wiley, New

York, 1995.

[3] J.J. Stezowski, J. Am. Chem. Soc. 98 (1976) 6012.

[4] K.H. Jogun, J.J. Stezowski, J. Am. Chem. Soc. 98 (1976)

6018.

[5] R. Prewo, J.J. Stezowski, J. Am. Chem. Soc. 99 (1977) 1117.

[6] J.J. Stezowski, J. Am. Chem. Soc. 99 (1977) 1122.

[7] L.J. Hughes, J.J. Stezowski, R.E. Hughes, J. Am. Chem. Soc.

101 (1979) 7655.



[10] R. Prewo, J.J. Stezowski, R. Kirchlechner, J. Am. Chem. Soc.

102 (1980) 7021.

[11] H. Lanig, M. Gottschalk, S. Schneider, T. Clark, J. Mol.

Model. 5 (1999) 46.

[12] W.A. Baker Jr., P.M. Brown, J. Am. Chem. Soc. 88 (1966)

1314.

[13] G.W. Everett Jr., J. Gulbis, J. Shaw, J. Am. Chem. Soc. 104

(1982) 445.

[14] L. Lambs, B. Decock-Le Reverend, H. Kozlowski, G.

Berthon, Inorg. Chem. 27 (1988) 3001.

[15] L. Lambs, G. Berthon, Inorg. Chim. Acta 151 (1988) 33.

[16] M. Jezowska-Bojczuk, L. Lambs, H. Kozlowski, G. Berthon,

Inorg. Chem. 32 (1993) 428.

[17] J.M. De Siqueira, S. Carvalho, E.B. Paniago, L. Tosi, H.

Beraldo, J. Pharm. Sci. 83 (1994) 291.

[18] F.C. Machado, C. Demicheli, A. Garnier-Suillerot, H.

Beraldo, J. Inorg. Biochem. 60 (1995) 163.

[19] S.V. De Mello-Matos, H. Beraldo, J. Braz. Chem. Soc. 6

(1995) 405.


[20] J.M. Wessel, W.E. Ford, W. Szymczak, S. Schneider, J. Phys.

Chem. B 102 (1998) 9323.

[21] A.A.M. Aly, A. Strasser, A. Vogler, Inorg. Chem. Commun. 5

(2002) 411.

[22] H.F. Dos Santos, W.B. De Almeida, M.C. Zerner, J. Pharm.

Sci. 87 (1998) 190.

[23] G.H. Miller, H.L. Smith, W.L. Rock, S. Hedberg, J. Pharm.

Sci. 66 (1977) 88.

[24] M. Takahashi, L. Altschmied, W. Hillen, J. Mol. Biol. 187

(1986) 341.

[25] J. Degenkolb, M. Takahashi, G.A. Ellestad, W. Hillen,

Antimicrob. Agents Chemother. 35 (1991) 1591.

[26] C. Kisker, W. Hinrichs, K. Tovar, W. Hillen, W. Saenger,

J. Mol. Biol. 247 (1995) 260.

[27] W.B. De Almeida, L.R.A. Costa, H.F. Dos Santos, M.C.

Zerner, J. Chem. Soc., Perkin Trans. 2 (1997) 1335.

[28] W.B. De Almeida, H.F. Dos Santos, W.R. Rocha, M.C.

Zerner, J. Chem. Soc., Dalton Trans. 15 (1998) 2531.

[29] W.B. De Almeida, H.F. Dos Santos, M.C. Zerner, J. Pharm.

Sci. 87 (1998) 1101.

[30] H.F. Dos Santos, M.C. Zerner, W.B. De Almeida, J. Chem.

Soc., Perkins Trans. 2 (1998) 2519.

[31] H.F. Dos Santos, E.S. Xavier, M.C. Zerner, W.B. De Almeida,

J. Mol. Struct. 527 (2000) 193.

[32] L. Onsager, J. Am. Chem. Soc. 58 (1936) 1486.

[33] J.B. Foresman, T.A. Keith, K.B. Wiberg, J. Snoonian, M.J.

Frisch, J. Phys. Chem. 100 (1996) 16098.

[34] (a) M. Cossi, V. Barone, R. Cammi, J. Tomasi, Chem. Phys.

Lett. 255 (1996) 327.

(b) V. Barone, M. Cossi, J. Tomasi, J. Chem. Phys. 107 (1997)

3210..

[35] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A.

Robb, J.R. Cheeseman, V.G. Zakrzewski, J.A. Montgomery,

Jr., R.E. Stratmann, J.C. Burant, S. Dapprich, J. M. Millam,

A.D. Daniels, K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi,

V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C.

Adamo, S. Clifford, J. Ochterski, G.A. Petersson, P.Y. Ayala,

Q. Cui, K. Morokuma, D.K. Malick, A.D. Rabuck,

K. Raghavachari, J.B. Foresman, J. Cioslowski, J.V. Ortiz,

B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi,

R. Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham,

C.Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe,

P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, J.L. Andres,

M. Head-Gordon, E.S. Replogle, J.A. Pople, GAUSSIAN98,

Revision A.6., Gaussian, Inc.: Pittsburgh PA, 1998.

[36] DIADORIM.TPT: L.C.G. Freitas, Fortran Code, 1990.

[37] M.W. Mahoney, W.L. Jorgensen, J. Chem. Phys. 112 (2000)

8910.

[38] BOSS Version 3.5: W.L. Jorgensen, Biorganic and Organic

Simulation System, 1995.

[39] M.P. Allen, D.J. Tildesley, Computer Simulation of Liquids,

Oxford University Press, Oxford, 1987.

[40] G.J. Palenik, M. Mathew, R. Restivo, J. Am. Chem. Soc. 100

(1978) 4458.

[41] H.F. Dos Santos, L.F.C. De Oliveira, S.O. Dantas, P.S. Santos,

W.B. De Almeida, Int. J. Quantum Chem. 80 (2000) 1076.

[42] H.F. Dos Santos, P.J. O’Malley, W.B. De Almeida, Theor.

Chem. Acc. 99 (1998) 301.

[43] J.B. Foresman, A. Frisch, Exploring Chemistry with Elec-

tronic Structure Methods, second ed., Gaussian Inc, USA,

1996.

[44] H.F. Dos Santos, W.R. Rocha, W.B. De Almeida, Chem. Phys.

280 (2002) 31.


the conformational and tautomeric equilibrium of 5a,6-anhydrotetracycline in aqueous solution at ph...

Documents