UNDERSTANDING THE ANTIHYPERGLYCEMIC AGENTS OF

THIAZOLIDINEDIONES ANALOGUES USING QUANTITATIVE

STRUCTERE ACTIVITY RELATIONSHIP (QSAR) MODELS
ABSTRACT
Quantitative Structure Activity Relationship (QSAR) is a useful tool to establish quantitative
relationship between various physico-chemical properties and biological activities of compounds.
The present studies have been carried out on series of 5-(4'-alkoxy benzyl)-2,4-thiazolidinediones
derivatives, which are a class of synthetic antihyperglycemic (anti-diabetic) drugs exerting action
upon activation of the peroxisome proliferators-activated receptor-γ (PPAR-γ). The 68 compounds
with similar array of biological activities were selected for QSAR model and divided into 53
training set and 15 test set. On performing multiple linear regression (MLR), the training set
showed high correlation (r=0.91) of significance. The model study indicated that thermodynamic,
topological descriptors interplay important role with PPAR-γ receptor.
Key Words: QSAR, PPAR-γ inhibitor, TZDs, MLR, Antihyperglycemic. NIDDM.
INTRODUCTION
Diabetes is the root cause of several chronic and progressive diseases, which adversely
affect number of organs including the nervous and vascular systems. More than 90% of the patients
suffer from Non-Insulin Dependent Diabetes Mellitus (NIDDM) i.e. type 2 diabetes. Resistance
of peripheral tissue to the action of insulin is a characteristic feature of human obesity and NIDDM.
Thus drugs that reverse the onset of insulin resistance fulfill a major unmet medical need for the
treatment of NIDDM. A new series 5-(4'-alkoxy benzyl)-2,4-thiazolidinediones derivatives are
reported by Sohda as antihyperglycemic agent[1] and oral insulin sensitizing agents that improve
glucose utilization without stimulating insulin release.
The molecular mechanism of thiazolidinedione is activation of peroxisome
prolifertoractivated receptor-γ (PPAR-γ). The PPAR-γ nuclear receptors are a class of transcription
factor proteins that play an important role in the regulation of lipid and glucose metabolism in
vertebrates. They are linked to severe human diseases such as cardiovascular disease and
INDDM.[2-5] When PPAR-γ bound to a small molecule called an agonist is activated by undergoing
a conformational change,[6] it binds in the form of a heterodimer with an retinod X receptor (RXR)
to a specific binding element in the DNA.[7[ PPAR-γ is proposed to be a possible molecular target
for type of compounds.[8,9] The 5-(4'-alkoxy benzyl)-2,4-thiazolidinedones compounds have few
essential pharmacophore elements. These are an acidic group linked to a central flat ring and a
large lipophilic substructure. Recently several QSAR studies of agonist binding to human PPAR-
γ have been undertaken.[10-14] While the first of these dealt with Thiazolidinediones, others treated
one or a few series of tyrosine derivatives originating from Glaxo Well come research.[15-17]
EXPERIMENTAL METHODS
To obtain a reliable and robust Quantitative Structure Activity Relationship (QSAR) model, it
is desirable to build the model based on a large data set that covers reasonable chemical diversity
and biological activity spread. A prerequisite for QSAR study is a congeneric series of compounds.
The training and test sets were selected to cover all structural features and biological activities.
Selection of the training and test set molecules was done by considering the fact that test set
molecules represent a range of biological activity, similar to that of the training set. We have
divided the molecules such that all possible structural features are included in both the training and
test sets. A training set containing 53 molecules was used for generation of QSAR models, whereas
a test set of 15 molecules to test the predictive ability of generated models Table no.5. The QSAR
model and all the studies were carried out using QSAR lite1.0 software [21] for generation of
descriptors values and to performed equation.

BIOLOGICAL ACTIVITY
The compounds chosen for the present study were obtained from the literature. The structure and
biological activity values of the compounds forming the training set are shown in the Table 1-3
and test set Table no 4. The hypoglycemic activities of compounds were tested using genetically
obese and diabetic yellow KK mice. The response of animal was measured as the reduction in the
blood glucose level from the control animals. The biological activity was expressed as pED25,
which is –log of effective molar dose required to reduce blood glucose by 25%.[18, 19,20]
RESULT AND DISCUSSION
Step-wise Multiple Linear Regressions (MLR) was performed to obtain correlation between
physico-chemical descriptors and the biological activity. The validity of the regression was
confirmed by correlation coefficient, standard deviation, F-test value. The best models were
selected on the basis of statistical significance. Initially all the 53 molecules were taken in the
study.
MLR analysis was carried out to find out the factors responsible for biological activity. It is an
attempt to maximize the fit of data to the regression equation by minimizing the variance (r2) by
adjusting each of the available parameter up or down. This task was approached stepwise. That is
successive regression equation were derived in which parameters are added, removed or replaced
until r2 were optimized. The magnitudes of the coefficient derived in this manner indicate the
relative contribution of the associated parameter to the biological activity. QSAR models were
generated for the whole series. All parameters in the Table no. 6 were used for generation of QSAR
equations. The first step of model generation was to identify the potential contributing parameters.
No equation having single parameter showed significant correlation with the activity. But
parameters showing good correlation with biological activity were combined in the next step for
the generation of equation with parameters. All the equation had variation in the parameters. Only
the parameters which were mutually orthogonal were combined. Therefore, equations were
derived with different combinations of three parameters. The regression equation is given in Table
no 5. The QSAR models were generated by choosing proper combination of descriptors belonging
to different categories that had proper balance of statistical parameters. Out of three equations, the
best QSAR model was found to be Equation no. 3 having r2 =0.83, r= 0.91 and F value 30.55.
Thus it was concluded to be statistically significant model

CONCLUSION
QSAR study was carried out for reported series of thiazolidinedione of 68 compounds derivatives
having carboxylic ester appendage at nitrogen-3 and biphenyl and heteroaryl constituents at
carbon-5 and sulphur-2. From the regression equation, it was found that after running regression
and generating mathematical relationship, the best QSAR equation was found to be significant.
The best QSAR equations in rotable bonds indicates that less rigid molecules show good biological
activity. It is evident from results that, better inhibitory activity of molecules (45 and 36) can be
attributed to presence of 9 and 11 rotable bond, respectively. Molecules 6 and 7 with 7 and 6
rotable bonds, respectively, showed lowest activity. SlogP indicates that incorporation of
hydrophobic atoms increase biological activity which can be exemplified by molecule 36 having
higher SlogP value as compared to less active molecule 6. The positive correlation Smr indicates
that in addition to hydrophobic interactions, the electronic interactions also play an important role
in PPAR-γ inhibitory activity. For all 53 treated molecules k2 alpha had no significance, whereas
Connectivity Index, ChiV0 showed higher coefficient. PSA show positive co-relation with
activity, incorporation of polar substituent such as OH in case of active molecules 45, 51 and 52
increases their activity. In conclusion, the present study showed that thermodynamic, topological
and structural descriptors are important for inhibition of PPAR-γ and series of 5-(4'-alkoxy
benzyl)-2, 4-thiazolidinedones molecules can be used as effective alternatives tosome of related
antihyperglycemic agents. The work can be further extended to find better and potent
antihyperglycemic compounds within the series.