International

Shaheen Begum et al. : 3D QSAR Journaland

Studies Benzoxazoles of Pharmaceutical Sciences
Oxazolo-(4, 5-b)pyridines and Nanotechnology
as Anti-fungal agents 557
Volume 2 Issue 2 July September 2009

Research Paper
3D QSAR Studies on Benzoxazoles and Oxazolo-(4, 5-b)pyridines as
Anti-fungal agents

Shaheen Begum1, Satya Parameshwar K2,*, Ravindra G K3, Achaiah G4,*

1
Institute of Pharmaceutical Technology, Sri Padmavati Mahila Visvavidyalayam, Tirupathi, Andhra Pradesh.
2
L.B. College, Warangal.
3
National institute of Pharmaceutical Education and Research, Hyderabad.
4
University College of Pharmaceutical Sciences, Kakatiya University, Warangal.

ABSTRACT: Benzoxazoles and Oxazolo-[4,5-b]pyridines have been reported as potent anti-fungal agents. 3D QSAR
tools including CoMFA and CoMSIA have been known to be a promising approaches is to correlate structures and activity
which further enable the medicinal chemists to design more potent molecules thus curtailing the cost and time in drug
research. CoMFA and CoMSIA studies have been carried out on 31 molecules of benzoxazole and oxazolopyridines in
order to determine the structural properties required for effective antifungal activity. 26 compounds were evaluated for
establishing QSAR model, which was then validated by predicting the activities of five test set molecules. All the molecules
were aligned by SYBYL database alignment which led to a best model with q2 value of 0.835, r2=0.976 and r2pred=0.773.
This model was further employed to derive CoMSIA models, a best model with steric, electrostatic, hydrophobic and
hydrogen bond acceptor indices exhibited q2 = 0.812, r2=0.971 and r2pred=0.81. The models thus obtained from this study
can be useful for the design and development of new potential anti-fungal agents.

Introduction activity. It is imperative to improve predictability of QSAR

model of test compounds by considering the structural and
Azoles have been one of the extensively studied scaffolds physicochemical features. Three dimensional quantitative
in medicinal chemistry as anti-microbial agents in structureactivity relationship (3D-QSAR) techniques
particular as anti-fungal agents [Cavelleri et al., 1978, including comparative molecular field analysis (CoMFA)
Cuckler et al., 1966, Elamine et al., 1981]. In the last two and comparative molecular similarity indices analysis
decades there has been shift in the development of anti- (CoMSIA) could facilitate in designing unsynthesized
fungal drugs. Inspite of expansion in anti-fungal drug agents for therapeutic purpose [Lepper et al., 2004,
discovery there remains a wide gap in the range and scope Ravindra et al., 2007, Sperandio da Silva et al., 2004].
of current anti-fungal chemotherapy (Graybill et al., 1996). Both the techniques predict the activities based on the
Benzoxazoles have been extensively studied for their existing set of molecules, CoMFA develops the 3D QSAR
antibacterial and anti-fungal activity [Oren et al., 1999, model based on the relationship between the
Temiz et al., 1998], anticancer activity [Kumar D et al., steric/electrostatic properties and biological activities,
2002], and also as new non-nucleoside topoisomerase I while CoMSIA develops the models using steric,
poisons [Kim et al., 1996] and HIV-1 reverse transcriptase electrostatic, hydrophobic and hydrogen bonding
properties. The addition of other fields in CoMSIA resulted
inhibitors [Perrin et al., 1996].
in augmented significance and predictive power of model
Quantitative structureactivity relationships and also interpretation of the resultant correlation of the
(QSARs) are computational statistical methods which model. CoMSIA is known to be less sensitive to molecular
reveal explainable difference in the observed biological alignments/ conformations than CoMFA [Bandyopadhyaya
et al., 2005].
* For correspondence: Satya Parameshwar K, Achaiah G, Inumerable reports are available containing 2D QSAR
E-mail: spkonda@yahoo.co.in ; achaiah_g@yahoo.co.in on various benzoxazoles as anti-fungal agents and very
few 3D QSAR on the same scaffolds are available. In this
connection, it is aimed is to analyze 3D structural

557
558 International Journal of Pharmaceutical Sciences and Nanotechnology Volume 2 Issue 2 July - September 2009

requirements of benzoxazoles and Oxazolo-(4,5- and CoMSIA models [Oren et al., 2004, Yalcin et al.,
b)pyridines as anti-fungal agents and also to understand the 2000]. The structures and their anti-fungal activities have
structural basis for their affinity for the activity and to been mentioned in Table 1. The anti-fungal activity of the
design still better potent molecules. data set has been converted into pIC50 using the formula
pIC50 = -log IC50 and which was in the range of 3.8-4.4.
Materials and methods
The inhibitors were randomly sorted into training set and
A series, containing 31 benzoxazole and oxazolopyridines test set containing 26 and 5 molecules respectively.
have been obtained from the literature to establish CoMFA

Table1 Structure and in vitro anti-fungal activity of training set compounds

R1 X N
R
O

Comp. No X R R1 pIC50
1 CH -H -H 3.892
2 CH -C(CH3)3 -H 4.001
3 CH -NH2 -H 3.924
4* CH -NHCH3 -H 3.952
5 CH -CH2CH3 -Cl 4.013
6 CH -NHCOCH3 -Cl 4.059
7 CH -NHCH3 -Cl 4.015
8 CH -Cl -Cl 4.024
9 CH -NO2 -Cl 4.04
10 CH -H -Cl 4.282
11 CH -CH3 -NO2 4.308
12 CH -C(CH3)3 -NO2 4.375
13* CH -NH2 -NO2 4.31
14 CH -Cl -NO2 4.342
15 CH -Br -NO2 4.406
16 CH -CH2CH3 -NH2 3.976
17 CH -F -NH2 3.96
18 CH -N(CH3)2 -NH2 4.005
19* CH -CH3 -CH3 3.95
20 CH -CH2CH3 -CH3 3.977
21 CH -CH2CH3 -CH3 3.98
22 CH -F -CH3 3.958
23 CH -NHCOCH3 -CH3 4.027
24* CH -NHCH3 -CH3 3.979
25 CH -N(CH3)2 -CH3 4.004
26 N -CH3 -H 4.225
27 N -CH2CH3 -H 4.253
28 N -OCH3 -H 4.257
29 N -OCH2CH3 -H 4.283
30* N -NH2 -H 4.227
31 N -NO2 -H 4.285
 Shaheen Begum et al. : 3D QSAR Studies Benzoxazoles and Oxazolo-(4, 5-b)pyridines as Anti-fungal agents 559

All the molecular modeling studies were performed Alignment

using Sybyl 6.8 version [Sybyl 6.8, Tripos Inc.] on silicon
The important aspect of 3DQSAR is the alignment of the
graphics O2 workstation with IRIX 6.5 operating system.
inhibitors that defines the presumed pharmacophore of the
All the structures were constructed using sketch module of
series under the study. As no structural information is
SYBYL using standard library models. The modeled
available about ligand-target complex, the lowest energy
structure was minimized using Tripos force field with
conformation obtained after MOPAC of most active
distance dependent dielectric function and applying
molecule has been employed as templet of alignment using
Gasteiger-Hckel partial atomic charges. These structures
SYBYL database alignment (Fig.1).
were further optimized by semi-empirical MOPAC
calculations using AM1 method. These structures were
considered for generating CoMFA and CoMSIA
calculations.

Fig.1 Database alignment of anti-fungal agents.

Methodology of CoMFA and CoMSIA having charge +1, hydrophobicity +1, and hydrogen bond
donor and acceptor property of +1 was assigned to
A 3D cubic lattice was defined in CoMFA automatically calculate similarity indices. Atom based parameters were
by extending at least 4 beyond all the aligned molecules employed to compute hydrophobic descriptors
in all the three axes (X, Y, Z directions) with a grid [Viswanadhan et al., 1989] and rule based method was
spacing of 2.0 . Lennard-Jones and Columbic potentials used to get the hydrogen bond donor and acceptor fields.
were used to calculate the steric and electrostatic
interaction fields respectively at each lattice intersection. Partial least square analysis
Tripos force field with distant dependent dielectric
constant e = e0 Rij with e0 =1.0 was considered for the Partial least square analysis (PLS) is the best statistical tool
to derive the correlation between biological activity and
calculation of interaction energies considering an sp3
descriptors calculated by CoMFA and CoMSIA
carbon atom with a radius of 1.52 bearing +1 charge as a
[Ciubotariu et al., 1993]. Leave one out (LOO) method
probe atom. Steric and electrostatic cutoffs were applied to
was carried out as a rudimentary step to ascertain the
truncate the contribution up to 30 kcal/mol [Cramer et al., predictivity of the model and also to determine the
1988]. optimum number of components with minimum standard
error of prediction (SEP). The optimum number of
Klebe et al introduced CoMSIA analysis which
components thus obtained from the previous LOO method
calculates five similarity descriptors namely steric, was further applied to derive the final non-cross validated
electrostatic, hydrophobic, hydrogen bond donor and correlation r2. Group cross validation and bootstrapping
acceptor [Klebe et al., 1994, Klebe et al., 1999)]. Gaussian were performed to get the confidence intervals (mean and
type function was employed to gauge similarity indices at standard deviation) and also to assess the robustness of the
all the grid points, a default attenuation factor () of 0.3 model. In the group cross validation, which was conducted
was employed as a smoothening function. A probe atom for 50 iterations in which the training set molecules were
560 International Journal of Pharmaceutical Sciences and Nanotechnology Volume 2 Issue 2 July - September 2009

randomly divided into groups and subjected to the cross validated r2 value of 0.976 with six components, and
correlation. The mean of 50 cross validation runs was standard error of estimation (SEE) of 0.106 as listed in
considered as correlation coefficient, rcv2. The Table 2. The contributions of steric and electrostatic were
bootstrapping (rbs2) analysis for 100 runs was performed 60.5% and 39.5%, respectively. Group cross validation
with optimum number of components, which provides an was carried out for 50 iterations to prove for the
estimate of the variability of the parameters in a final confidence of the model that resulted in a mean group
model [Bunce D et al., 1988]. cross-validated r2 value of 0.81. Bootstrapping analysis of
100 runs was led to an r2bs of 0.943 with deviation of 0.03
Result and discussions (Table 3) and the CoMSIA predicted activities of both
training and test sets are presented in Fig 2.
CoMFA model resulted from SYBYL database alignment
(Fig.1) has yielded a crossvalidated q2 value of 0.835, non-

Table 2 CoMFA and CoMSIA statistics.

CoMFA CoMSIA
Parameters
I II
2
r 0.976 0.971

q2 0.835 0.812

SEE 0.029 0.023

SEP 0.075 0.032

Number of components 6 6

F value 129.92 105.45

% Contribution

Steric 60.5 38.7

Electrostatic 39.5 14.3

Hydrophobic -- 34.4

H acceptor -- 12.6

r2pred 0.773 0.81

Table 3 Bootstrapping results of CoMFA and CoMSIA.

r2bsa r2cvb

CoMFAc CoMSIAd CoMFAc CoMSIAd

Mean 0.945 0.98 0.81 0.785

SD 0.03 0.019 0.059 0.07

a
From 100 run boot strapping
b
From mean of 50 run group cross validation
c
CoMFA with atom fit based alignment
d
CoMSIA model generated by the combination of steric, hydrophobic and hydrogen bond
acceptor fields
 Shaheen Begum et al. : 3D QSAR Studies Benzoxazoles and Oxazolo-(4, 5-b)pyridines as Anti-fungal agents 561

CoMSIA results
CoMFA model obtained as explained in the previous
section, has been considered to derive CoMSIA model and
five different models with different CoMSIA field
combination were generated. One best model containing
steric, electrostatic, hydrophobic and hydrogen bond
acceptor fields demonstrated q2 and r2 value of 0.812 and
0.971, respectively with acceptable F-value of 105.52 from
six components. The contributions of all fields were found
to be 32.4 % Steric, 12.3 % Electrostatic, 28.2 %
Hydrophobic, 16.5 % Hydrogen bond acceptor and 10.6 %
Hydrogen bond donor. The aforementioned model was
further subjected to group cross validation for 50 runs and
correlation of 0.785 was obtained and the boot strapping
for 100 runs was found to be 0.98 with 0.19 SD as
summarized in Table 3. The predicted activity of both test
set and training set molecules by CoMFA and CoMSIA
models are listed in Table 5 and the graphs of experimental
Fig. 2 Predicted and observed activities of training activity against the CoMSIA predicted activities of both
and test sets using CoMFA. training and test sets are presented in Fig 3.

Table 5 actual and predicted (CoMFA and CoMSIA) activities of test set molecules.

Comp. No Actual pIC50 Predicted Residuals Predicted pIC50 Residuals

pIC50 CoMFA CoMFA CoMSIA CoMSIA
4 3.952 4.027 -0.077 3.988 -0.038
13 4.31 4.252 0.058 4.315 -0.005
19 3.95 3.962 -0.012 3.952 -0.002
24 3.979 4.004 -0.024 3.965 0.015
30 4.227 4.157 0.073 4.165 0.065

Fig. 3 Predicted and observed activities of training and test sets using CoMSI.
562 International Journal of Pharmaceutical Sciences and Nanotechnology Volume 2 Issue 2 July - September 2009

was found to be the most influential for the antifungal

CoMFA and CoMSIA contour map analysis
activity. Further insight into the contour plots at this
The steric and electrostatic contour plots of CoMFA position led to the conclusion that not only steric property
revealed that three green colored plots at the 5th position of but also electronic effect guides the activity.
benzoxazole, 3 or 4 of 2-phenyl infer the sterically Comparatively, electron releasing groups demonstrated
favorable groups required for anti-fungal activity (Fig. 4). reduced activity, whereas groups with electron
A small red contour plot was also noticed at 5th position withdrawing nature such as chlorine and nitro groups
along with green. From the SAR it is found that showed better activity as noticed in molecule 10 and 15
unsubstituted molecule 1 exhibited least anti-fungal respectively. A big blue contour map which covered the
activity against Candida albicans. Substitution at 5th benzoxazole was found which indicates that the electron
position by chlorine, nitro, amino and methyl groups has withdrawing groups would create lesser electron cloud
led to molecules with greater potency, however nitro group over the aromatic ring by withdrawing the electrons.

Fig. 4 CoMFA steric and electrostatic contour maps. Green contour favors steric bulk, whereas yellow color
disfavors steric bulk. Blue contour indicates electropositive charge and red contour electronegative charge.

Two more green colored plots have been spotted at the 4 substitution at 5th position might not be essential for the
of 2-phenyl ring which suggests that substitution with activity however bulky groups alone at the 4 position on
bulky groups at this position could influence positively for 2-phenyl ring is vital for the activity.
the antifungal activity. Activity of compounds was found
Four yellow isopleths could be sighted one at the 5th
increasing from fluorine to bromine as in molecules 22, 14
position, one on the 2-phenyl ring and two across the 4
and 15. Compound 12 has also exhibited potent activity
having t-butyl group at the same place. Moderate increase position of 2-phenyl ring infer about the requirement of
in the potency of the compounds was also observed in the hydrophobicity (Fig. 6). From the previous explanation,
molecules bearing substituted amino groups. substitution at the 5th position might not be key
requirement for potent activity, hence yellow isopleth at
CoMSIA steric and electrostatic contour plots are quite
this position can not be explained. However two yellow
similar however one big yellow plot was found at the 5th
contours across the 4-position indicates that the sustituents
position of heterocyclic ring (Fig 5). A careful observation
of the data set considered for the model development that increase the lipophilicity contribute positively for the
revealed that a few molecules (26-29, 31) which are activity. Compounds like 11 and 12 bearing methyl and t-
unsubstituted found to be potent This indicates that butyl demonstrated potent anti-fungal activity.
 Shaheen Begum et al. : 3D QSAR Studies Benzoxazoles and Oxazolo-(4, 5-b)pyridines as Anti-fungal agents 563

Fig. 5 CoMSIA steric and electrostatic fields are represented by green, yellow, blue and red colour regions.

Fig. 6 CoMSIA hydrophobic contour map.

Validation of CoMFA and CoMSIA models greater agreement with the observed activities with
marginal error range with predictive correlation coefficient
The most critical and important task of 3D-QSAR is model of r2pred of 0.771 and 0.81 for CoMFA and CoMSIA
validation in which the predictive ability of the model models respectively. The predicted activities of both test
established by reproducing the activity of test set set and training set compounds by both CoMFA and
compounds which are not included in training set. The best CoMSIA verses actual activities has been shown in Fig.
CoMFA and CoMSIA models thus resulted were subjected These results indicate that the CoMFA and CoMSIA
to external validation by predicting the activity of test set models could be consistently employed in the design for
compounds. All the test set compounds were constructed developing inhibitors against fungal infections caused by
and aligned similarly. Table 4 contains the predicted candida albicans.
activity of test set compounds which are found to be in
564 International Journal of Pharmaceutical Sciences and Nanotechnology Volume 2 Issue 2 July - September 2009

Table 4 actual and predicted activities of training set molecules.

Comp. Actual pIC50 Predicted Residuals Predicted pIC50 Residuals

No pIC50 CoMFA CoMFA CoMSIA CoMSIA

1 3.892 3.977 -0.087 3.882 0.010

2 4.001 4.016 -0.016 3.970 0.030
3 3.924 3.912 0.008 3.934 -0.010
5 4.013 3.996 0.014 4.049 -0.036
6 4.059 4.043 0.017 4.071 -0.012
7 4.015 4.018 -0.008 4.020 -0.005
8 4.024 3.980 0.040 4.013 0.011
9 4.04 4.016 0.024 4.081 -0.040
10 4.282 4.317 -0.037 4.263 0.019
11 4.308 4.334 -0.024 4.349 -0.041
12 4.375 4.356 0.014 4.377 -0.002
14 4.342 4.341 -0.001 4.346 -0.004
15 4.406 4.345 0.065 4.356 0.050
16 3.976 3.977 0.003 3.960 0.19
17 3.96 3.957 0.003 3.919 0.040
18 4.005 3.983 0.027 3.978 0.027
20 3.977 3.983 -0.003 3.972 0.005
21 3.98 3.984 -0.004 4.001 -0.021
22 3.958 3.990 -0.030 3.992 -0.034
23 4.027 4.026 0.004 4.039 -0.012
25 4.004 3.983 0.017 4.030 -0.027
26 4.225 4.240 -0.010 4.202 0.024
27 4.253 4.263 -0.013 4.238 0.016
28 4.257 4.263 -0.003 4.266 -0.008
29 4.283 4.286 -0.006 4.295 -0.012
31 4.285 4.284 -0.004 4.165 0.065

Conclusion References
CoMFA and CoMSIA methods could derive best models Bandyopadhyaya AK, Johnsamuel J, Al-Madhoun A, Eriksson S,
employing twenty six anti-fungal agents. These models Tjarks W. Comparative molecular field analysis and
were successfully validated by predicting the anti-fungal comparative molecular similarity indices analysis of human
activity of five test set molecules. In comparison, the thymidine kinase 1 substrates. Bioorg. Med. Chem. 13: 1681
1689 (2005)
predictivity of the CoMFA model was found to be superior
to that of the CoMSIA model. However, a combined use of Cavelleri B, Volpe G, Arioli V, Pizzocheri F, Dienna A.
both the CoMFA and the CoMSIA methods would further Synthesis and biological activity of new 2-Nitroimidazole
enhance the predictivity of the derived models. derivatives J. Med. Chem. 21: 781-784 (1978).
 Shaheen Begum et al. : 3D QSAR Studies Benzoxazoles and Oxazolo-(4, 5-b)pyridines as Anti-fungal agents 565

Ciubotariu RD, Deretey E, Opera TI, Sulea T, Simon Z, Kurunczi Oren I, Temiz O, Yalcin I, Sener EA, Altanlar N. Synthesis and
L, Chiriac A, Quant. Struct.-Act. Relat. 12: 367-372 (1993) antimicrobial activity of some novel 2,5- and/or 6-substituted
Cramer D, Patterson E, Bunce D. Comparative molecular field benzoxazole and benzimidazole derivatives.Eur. J. Pharm.
analysis (CoMFA). 1. Effect of shape on binding of steroids Sci. 7: 153-160 (1999).
to carrier proteins. J Am Chem Soc. 110: 5959-5967 (1988)
Oren I, Yalcin I, Sener E, Ucarturk N. Synthesis and structure
Cramer RD, Bunce D, Patterson E. Quant. Struct. Act. Relat 7:18 activity relationships of new antimicrobial active
(1988)
multisubstituted benzazole derivatives. European J. Med.
Cuckler AC, Mezey KC. The therapeutic efficacy of Chem. 39: 291-298 (2004)
thiabendazole for elminthic infections in man. Arzneim.
Forsch. 16: 411-428 (1966). Perrin L, Rakik A, Yearly S, Baumberger C, Kinloch-deLoies S,
Pechiere M, Hirschel B. Combined therapy with zidovudine
Elamine IE, Uppal ZM, Al-Bard AA. Antibacterial and
and L-697,661 in primary HIV infection. AIDS 10: 1233-
Antifungal Activities of Benzimidazole and Benzoxazole
1237 (1996)
Derivatives. Antimicrobial Agents and Chemotherapy. 981,
p.29-32 Ravindra GK, Srivani P, Achaiah G, Sastry GN. Strategies to
Graybill JR. The future of antifungal therapy. Clin. Infect. Dis., design pyrazolyl urea derivatives for p38 kinase inhibition: a
22: 166-168 (1996) molecular modelling study. J Comput Aided Mol Des.
21:155166 (2007)
Kim JS, Sun Q, Gatto B, Liu A, Liu LF, LaVoie E. J. Structure-
activity relationships of benzimidazoles and related Sperandio da Silva GM, Sant Anna CMR, Barreiro EJ. A novel
heterocycles as topoisomerase I poisons. Bioorg. Med. Chem. 3D-QSAR comparative molecular field analysis (CoMFA)
4: 621-630 (1996)
model of imidazole and quinazolinone functionalized p38
Klebe G, Abraham U, Mietzner T. Molecular similarity indices in MAP kinase inhibitors. Bioorg. Med. Chem. 12: 3159-3166
a comparative analysis (CoMSIA) of drug molecules to (2004)
correlate and predict their biological activity. J Med Chem.
37: 4130-4146 (1994). Sybyl 6.8, Tripos Inc., 1699 S. Hanley Rd., St Leuis, MO 63144
USA
Klebe G, Abraham U. Comparative molecular similarity index
analysis (CoMSIA) to study hydrogen-bonding properties and Temiz O, Oren I, Sener E, Yalcin I, Ucartrk N. Synthesis and
to score combinatorial libraries. J Comp. Aided Mol. Des. 13: microbiological activity of some novel 5- or 6-methyl-2-(2,4-
1-10 (1999)
disubstituted phenyl) benzoxazole derivatives. Il Farmaco.
Kumar D, Jacob MR, Reynolds M, Kerwin SM. Synthesis and 53: 337-341 (1998).
evaluation of anticancer benzoxazoles and benzimidazoles
related to UK-1. Bioorg. Med. Chem. 10: 3997-4004 (2002) Viswanadhan VN, Ghose AK, Revankar GR, Robins RK. J
Chem. Inf. Comput. Sci. 29:163 (1989)
Lepper ER, Ng SSW, Guetschow M, Weiss M, Hauschildt S,
Hecker TK, Luzzio FA, Eger K, Figg WD. Comparative Yalcin I, Oren I, Teniz O, Sener EA. QSARs of some novel
molecular field analysis and comparative molecular similarity isosteric heterocyclics with antifungal activity. Acta
indices analysis of thalidomide analogues as angiogenesis Biochemic. Polon. 47: 481-486 (2000)
inhibitors. J. Med. Chem. 47: 2219-2227 (2004)