104.

Lira T. O., Colombo R., Ribeiro J. S., Ferreira M. M. C., Yariwake J. H., "Chemometric Treatment of Two Varieties of Transgenic Sugarcan Leaves by HPLC Fingerprinting". São Pedro, SP, Brazil, 04-07/11/2007: 1st Brazilian Conference on Natural Products (BCNP: Recent Advances on Natural Products, covering all the latest and outstanding developments in the area) & XXVII Annual Meeting on Micromolecular Evolution, Systematics and Ecology (XXVII RESEM: Reflections on the Current Status of Chemosystematics), Abstracts (2007) BOSC-1. Oral Session C: Manufacturing and Quality Control of Herbal Drugs and Essential Oils.



Chemometric Treatment  of  Two Varieties  of  Transgenic Sugarcan Leaves  by  HPLC
Fingerprinting

Tatiana O. Liraa,  Renata Colomboa,  Juliano S. Ribeirob,  Márcia M. C. Ferreirab,  Janete  H.
Yariwakea

aInstituto de Química de São Carlos, Universidade de São Paulo, 780, 13560-970 São
Carlos, SP, Brazil  bInstituto de Química, Universidade Estadual de Campinas (UNICAMP),
13083-970 Campinas, SP, Brasil

          The aim of this study was  to  evaluate  the differences in the  HPLC  chromatographic
profile  between   two  varieties   of   transgenic  sugarcane   ("Browman-Birk"  and  "Kunitz",
furnished  by  Prof. Marcio de Castro Silva-ESALQ/USP) and,  consequently,  to  identify the
secondary  metabolites  which  are  responsible  for  these  differences.  For this purpose,  two
pattern  recognition  methods,   partial   least   squares   discriminant  analysis  (PLS-DA)  and
principal  component  analysis   (PCA)   were    performed   [1-2].   High-performance   liquid
chromatography   (HPLC)   method  with  photo-diode  array   (DAD)   detection,   previously
described [3],  was used  in  order to obtain the HPLC fingerprints  of  flavonoids of sugarcane
leaves  for  chemometric  analysis.  Preprocessing  (baseline  correction  and  autoscaling)  and
pretreatments  (correlation  optimized  warping - COW)  of  fingerprints  chromatograms  were
required   to  reduce  drifts  in  the  retention  times   as   well  as  alignment  correction  in  the
chromatographic data, due to its peak shape and area [4]. PLS-DA (RMSECV = 0.0662, RCV
=  0.991  using  two  latent  variables)   successfully  classified   the  leaves  of  two  sugarcane
varieties.   PC1  versus   PC2  scores   plot  (describing  84.2%  of  total  variance)  effectively
distinguished th Kunitz and Bowman-Birk sample groups.  Through  PLS-DA  it  was  possible
to decrease the number of variables from 2701 to 70 and thus improve the PCA results.  From
the  loadings  (PC1  and  PC2),  it  could  be  concluded   that  mainly  7  peaks  were  able  to
discriminate  the  two sets  of  transgenic sugarcane leaves  (tR1 = 6.74,  tR2 = 11.8,  tR3 = 17.8,
tR4 = 19.9,  tR5 = 23.3, tR6 = 27.2,  tR7 = 45.1 min).  Some of the compounds corresponding  to
those  peaks  were elucidated by comparison with previous LC-MS data  [6-7].  These  results
demonstrated  that our approach is capable to identify the metabolites which are significant  for
the discrimination of these two transgenic sugarcane leaves.

Acknowledgements: CAPES, CNPq, FAPESP
 
 

Refs.
1. S. Wold et al., Chemometrics and Intelligent Laboratory Systems, 2, 37 (1987)
2. L. Xie et al., Analytica Chimica Acta, 584, 379 (2007)
3. R. Colombro et al., Journal of Chromatography A, 1103, 118 (2006)
4. N. P. V. Nielsen et al., Journal of Chromatography A, 805, 17 (1998)
5. R. Colombo et al., Phytochemical Analysis, 17, 337 (2006)
6. R. Colombo et al., Journal of Chromatography A, 1082, 51 (2005)