4
___________________________________________________________________________________________________________________
Structural  chemometrics  applied   to   small   bioactive  molecules   complexed   with
their  protein  receptors

Ferreira M. M. C. and Kiralj R.

Universidade Estadual de Campinas - Campinas SP

Resumo [Abstract]

The Protein Data Bank (PDB)  is a structural database which  currently contains  over  26000  macromolecular  structures,
mainly obtained  by  diffraction  methods  (over 80 percent).  Drug-protein  complexes  represent  a  valuable source of  in-
formation for further  experimental,  theoretical/computational  or  comparative studies.  Molecular design,  supported  with
advanced computer techniques,  chemometric methods  and  quantitative structure  activity/property  studies  represent  an
area which needs  3D  structures  of  macromolecules and their complexes.  It is a  common practice  that results from pro-
tein crystallography serve as  direct inputs for these methods.  It does  not  happen  rarely  that  published  macromolecular
structures are not completely characterized in  the sense of structural science.  Frequently,  structural  data  are  not studied
for their interconsistency and intercorrelations,  and  are not compared with other  similar  structures.  The purpose  of  this
work  is  to  present some examples  that  fill  this  "gap"  by  developing a new view on employing and interpreting  experi-
mental  3D  structures  deposited in the PDB. This is done through application of chemometric methods  such  as Principal
Component Analysis  and  Hierachical  Cluster  Analysis   on various structural data,  called  here  structural chemometrics.
The chemometric techniques tend  to recognize and visualize regular patterns among structural data and connect them  with
biological,  physico-chemical  and  other macro- or microproperties of substrates and/or receptors.  One example  is  HIV-1
protease complexed with inhibitor  L-700,417.  Steric and electronic protein-drug interactions were studies  based on simple
atom counts and excluding/including drug groups [1].  The  groups exhibited  various  interaction  profiles and clustering  in
the pattern recognition methods.  In the second example,  similar behaviour was observed for progesterone hydrogen atoms
inside the active hole of progesterone receptor protein [2]. Central, free and occuped hydrogens were detected, what agrees
with different biological activity obtained when the hydrogens  are substituted by small groups.  In the third example,  auxin
binding protein 1  is  complexed with its  1-naphthalenic acid  [3].  Simple  counts  and  distance parameters for  hydrogens
of the acid showed distinct clusters in chemometric analysis,  being  related  to  various pocket sizes  and consequent  auxin
activities of indole-3-acetic  acids.  Such findings represent valuable  information for design  of  new  bioactive  compounds
and insigh into the macromolecular structure-function relationships.

Agradecimentos [Acknowledgement]

FAPESP

Referências [References]

1. R. Kiralj, M. M. C. Ferreira, J. Mol. Graph. Mod. 21, 499 (2003)
2. R. Kiralj, Y. Takahata, M. M. C. Ferreira, QSAR Comb. Sci. 22 430 (2003)
3.M. M. C. Ferreira, R. Kiralj, Croat. Chem. Acta, submitted
 
 
 
 

44