Natural products have long demonstrated their important role as a rich source of new targets for drug discovery. The huge structural diversity of natural compounds together with unique mechanism of action has prompted promising applications in the fields of medicine, pharmacy and biology [1].

Among challenges encountered in product outcome is standardization of plant inputs for the pharmaceuticals industry. Factors such as soil, climate conditions, genotypes, harvesting, and post-harvest conditions have shown to have an important affect both on chemical and biological activity. Therefore, growth control of cultivars and their processing have a straight outcome on health benefits and economic issues. According to Ibáñez-Marcelo and Alarcon [2]. Phenotypes arise from the combination of genotype together with the environment where the species is grown.

Previously our research group reported the antoniceptive properties of compounds isolated from Pterodon pubescens Benth. fruit (Fabaceae) that are commercially available in Brazil. The plant’s crude alcoholic extract is used in folk medicine in anti-inflammatory, analgesic, and anti-rheumatic preparations. Phytochemical studies of Pterodon genus revealed the presence of alkaloids, isoflavones and diterpenes. Furan diterpenes were identified and isolated from Pterodon species. Authors [3-5] reported that furanoditerpenes possessing vouacapan skeleton are involved with the anti-inflammatory, antinociceptive and antiproliferative properties of Pterodon pubescens fruit crude extract. Among compounds reported involved with the antinociceptive and anticancer activity were geranylgeraniol, 6α- acetoxy,7α- hydroxyl-vouacapan, 6α,7α-di-hydroxy-vouacapan-17α-oate methyl esther, isomers 6α-hydroxy-7α-acetoxy-vouacapan-17α-oate methyl esther and 6α-acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther (Figure 1).

Figure 1: Chemical structures of compounds: A) geranylgeraniol (m/z 288);, B) 6α- acetoxy,7α- hydroxyl-vouacapan (m/z 360); C) 6α,7α-di-hydroxy-vouacapan-17α-oate methyl esther (m/z 362); D) isomers 6α-hydroxy-7α-acetoxy-vouacapan-17α-oate methyl esther and E) 6α-acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther (m/z 404).

The chemical variability encountered in samples from different origins throughout the country prompted the study of the secondary metabolite 6α-acetoxy-7α-hydroxy-vouacapan variability among samples encountered in Minas Gerais and São Paulo states of Brazil, to determine if the compound was consequence of the production as an artifact throughout storage.

Chemical analysis

IR spectra were recorded on a JASCO-FT/IR-410 spectrometer. ¹H, ¹³C NMR and 2D experiments were conducted on a Varian Inova-500 spectrometer (11 T). Chemical shifts were recorded in CDCl₃ solutions and quoted relative to TMS (δH 0.0, 1H NMR) and CHCl₃ (δC 77.0, ¹³C NMR). High-resolution electron impact ionization mass spectrometry (HREIMS) was recorded on a VG-AutoSpec high resolution mass spectrometer (70 eV) using direct probe. GC-MS was carried out using a HP-6890/5970 system equipped with a J&W Scientific DB-5 fused capillary column (25 m x 0.2 mm x 0.33 μ). Temperature program was 40 (5°C min^-1) - 300ºC (10 min). Injector 250°C, detector 300°C. Helium was used as carrier gas (0.7 bar, 1 mL min^-1). The MS were taken at 70 eV. Scanning speed was 0.84 scans s^-1 from 40 to 550. Sample volume was 1 μL. Split: 1:40. Column chromatography (CC): silica gel (0.063 x 0.200 mm, Merck^®). TLC (thin layer chromatography): precoated plates (775554 Merck^®), UV detection and anisaldehyde solution.

Phytochemistry: Plant Material

Pterodon pubescens (Pp) fruits were previously collected in São Carlos, São Paulo in March 2012 and identified by Prof. Dr. Jorge Yoshio Tamashiro from IB-Unicamp (Department of Botany), who identified the plant species. The voucher specimen was deposited at the University of Campinas (Unicamp) Herbarium, under number UEC 179139. Fruits from Ponto Chique, Minas Gerais were collected in July 2012 and identified by comparison with an authentic voucher deposited at Unicamp and further deposited under number UEC 179140. The genetic patrimony number issued by CGEN-IBAMA was process 010495/2012-6.

Extract Production Compounds Isolation

For time zero sampling freeze-dried fruits (10 g) were ground (POLI model LS-06) and extracted either with dichloromethane three times during 2 h periods, with 5:1 solvent/plant ratio, at room temperature with constant stirring (RTS) or in Soxhlet system with dichloromethane. The extract was dried over anhydrous Na₂SO₄ to remove residual water, filtered and concentrated under vacuum (BUCHI model RE 215), providing the crude extract.

Further monthly sampling used fruits collected (1000 g),that were maintained at 40oC throughout the 330 days of analysis with exception of samples taken at time zero. A total of thirteen samples were analyzed from São Carlos and Minas Gerais origins respectively at times 0, 30, 60, 90, 120, 150, 180, 210, 240, 270, 300 and 330 days. Every month the whole batch was homogenized and samples in triplet were collected for analysis.

The extracts were monitored monthly for geranylgeraniol, 6α-acetoxy-7α- hydroxyl-vouacapan, 6α,7α-di-hydroxy-vouacapan- 17α-oate methyl esther, 6α-acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther and 6α-hydroxy-7α-acetoxy-vouacapan-17α-oate methyl esther content and total growth inhibition against prostate cancer cell lines (PC-03).

The isolated compounds were obtained from the crude oil (18.2 g) that was purified on pre-column chromatography using silica gel (Merck 7734) (5 × 60 cm). The obtained fractions (FR) were: hexane (FR1) (0-450 mL); hexane/ethyl acetate (95:5) (FR2) (500-900 mL); hexane/ ethyl acetate 1 (80:20) (FR3-4) (1000-1350 mL); hexane/ethyl acetate (60:40) (1400-1800 mL) (FR5-6); hexane/ethyl acetate (40:60) (1900- 2300 mL) (FR7); rest flushed with methanol. The resulting fractions were monitored by thin layer chromatography (TLC) in comparison to previously purified standards, exposed with anysaldehyde reagent (50 mL acetic acid, 0.5 mL sulfuric acid and 0.5 mL anysaldehyde) followed by heating at 110°C.

Cell proliferation

Human tumor cell line PC-03 (prostate) was kindly provided by National Cancer Institute (NCI). Stock culture was grown in medium containing 5 mL RPMI 1640 (GIBCO BRL) supplemented with 5% fetal bovine serum. Gentamicine (50 μg mL^-1) was added to experimental cultures. Cells in 96 well plates (100 μL cells well^-1) were exposed to sample concentrations in DMSO/RPMI (0.25, 2.5, 25 and 250 μg mL^-1) at 37ºC, 5% of CO₂ in air for 48 h. Final DMSO concentration did not affect cell viability. Afterwards cells were fixed with 50% trichloroacetic acid and cell proliferation determined by spectrophotometric quantification (540 nm) of cellular protein content using sulforhodamine B stain assay. Using the concentration-response curve for each cell line, TGI (concentration that produces total growth inhibition) was determined through non-linear regression analysis using software Origin 7.5 (OriginLab Corporation) [6].

Principal Component Analysis (PCA)

The data correlations obtained from São Paulo (n=21) and Minas Gerais (n=22) samples were plotted in the light of PCA (Principal Component Analysis). The different parameters (variables) were analyzed by a multivariate approach considering TGI values compared to geranylgeraniol, isomers 6α-hydroxy-7α-acetoxy-vouacapan-17α- oate methyl esther and 6α-acetoxy-7α-hydroxy-vouacapan-17α- oate methyl esther, 6α-acetoxy-7α-hydroxy-vouacapan, 6α,7α-dihydroxy- vouacapan-17α-oate methyl esther content, total vouacapan content (considering the sum of the vouacapans monitored) and geranylgeraniol over 6α-acetoxy-7α-hydroxy-vouacapan ratio content throughout twelve months maintained at 40oC. These variables in 43 samples were submitted to principal component analysis (PCA) with autoscaling pretreatment, resulting initially in a 43 x 7 data matrix. For analysis based on TGI potency a 21 x 7 (São Paulo) and 22 x 7 (Minas Gerais) data matrices were proposed. The Unscrambler 8.0 (CAMO, Norway) chemometrics package was employed for PCA calculations.

The overall extraction yield was approximately the same ranging from 2.9-3.2% (m/m), without affecting overall 6α-acetoxy-7α- hydroxy-vouacapan content among different extraction process of the same sample. The extracts of Minas Gerais samples were monitored monthly for geranylgeraniol and vouacapans content and total growth inhibition against prostate cancer cell lines (PC-03) (Table 1).

Table 1: Minas Gerais sample Extracts monitored monthly for geranylgeraniol, 6α-acetoxy-7α-hydroxy-vouacapan, 6α,7α-di-hydroxy-vouacapan-17α-oate methyl esther, 6α- acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther, 6α-hydroxy-7α-acetoxy-vouacapan-17α-oate methyl esther content and total growth inhibition against prostate cancer cell lines (PC-03).

Fraction 7 (10 g), containing vouacapan compounds was successively chromatographed by CC on silica-gel (70-230 mesh) (5 × 60 cm) and eluted with hexane/dichloromethane (7:3) (900-1800 mL) yielded 6α-acetoxy-7α-hydroxy-vouacapan (333 mg, 3% yield), Rf 0.75; hexane/dichloromethane (6:4) (1900-2600 mL) yielded 6α-hydroxy- 7α-acetoxy-vouacapan-17α-oate methyl esther and 6α-acetoxy-7α- hydroxy-vouacapan-17α-oate methyl esther (990 mg, 9.9% yield), Rf 0.56; hexane/dichloromethane (2:8) (2650-3100 mL) yielded geranylgeraniol (963 mg, 9.63% yield), Rf 0.29; (3350-4100 mL) yielded 6α,7α-di-hydroxy-vouacapan-17α-oato methyl esther (1.2 g, 12% yield) Rf 0.17.

The different parameters analyzed by a multivariate approach considering TGI values compared to geranylgeraniol and vouacapans content, and geranylgeraniol over 6α-acetoxy-7α-hydroxy-vouacapan ratio content throughout twelve months maintained at 40°C. The original data set was re-normalized by an auto scaling transformation since the values for the various parameters were expressed in different units.

Considering all the samples (n=43) a complete separation based on the geographic location (São Paulo and Minas Gerais) was possible. The score plots separate groups by similarity and the loading plot indicate the variables that influence the separation. In accordance with scores plot (Figure 1S and Graphic a) a good separation could be observed of groups. The variables that contributed most significantly to the separation were 6α-acetoxy-7α-hydroxy-vouacapan, 6α,7α-di-hydroxy-vouacapan-17α- oate methyl esther and the sum of isomers 6α-acetoxy-7α-hydroxyvouacapan- 17α-oate methyl esther and 6α-hydroxy-7α-acetoxyvouacapan- 17α-oate methyl esther content (Figure 1S and Graphic b).

Previous work reported by Cabral et al. [7] highlighted the chemical ESI(+)–MS fingerprinting profiles of samples from different geographical regions of Brazil, showing essentially the same constitutes with small variations in abundances. According to Cabral et al. [7] the results showed that PCA extracted three major principal components. The plot of PC1 versus PC2 versus PC3 accounted for ca. 99% of data variance (PC1=62.87%, PC2=32.62%, PC3=3.46%) and showed that all oils from the different states in Brazil were clearly grouped. Samples from Sergipe and Mato Grosso states had a higher absolute 6α-acetoxy- 7α-hydroxy-vouacapan content compared to samples from São Paulo, Minas Gerais and Bahia states. Previously our group [3] demonstrated that activity guided fractionation of Pterodon pubescens, purchased at commercial market, methylene chloride-soluble fraction afforded novel 6α-acetoxy-7α-hydroxy-vouacapan and 6α,7α-di-acetoxy-vouacapan, 7α-di-acetoxy-vouacapan and 6α,7α-di-hydroxy-vouacapan-17α-oate methyl esther. The isolated compounds were evaluated for in vitro cytotoxic activities against human normal cells and tumor cell lines UACC-62 (melanoma), MCF-7 (breast), NCI-H460 (lung, non-small cells), OVCAR-03 (ovarian), PC-03 (prostate), HT-29 (colon), 786- 0 (renal), K562 (leukemia) and NCI-ADR/RES (ovarian expressing phenotype multiple drugs resistance). These compounds showed selectivity in a concentration-dependent way against human PC- 03. The main component of that sample extract was 6α-acetoxy-7α- hydroxy-vouacapan. Further work by the same authors [3] with samples freshly collected at São Paulo state to continue with evaluation of the antinociceptive activity of the crude extract contribution demonstrated differently to contain mainly geranylgeraniol and 6α,7α-di-hydroxy-vouacapan- 17α-oate methyl estestherer with only traces of 6α-acetoxy- 7α-hydroxy-vouacapan. Thereafter Servat et al. [8] demonstrated the antinociceptive properties of samples obtained at Minas Gerais state containing mainly isomers 6α-hydroxy-7α-acetoxy-vouacapan- 17α-oate methyl esther and 6α-acetoxy-7α-hydroxy-vouacapan- 17α-oate methyl esther. The variability observed among the samples prompted the investigation to determine if the differences observed were consequence of chemical modifications of the original secondary metabolites encountered in the fruits due to samples post-harvest conditions or if this was consequence of different genotypes among the species. In order to standardize an extract for the production of an herbal medicine those issues need to be well understood. Therefore this study monitored monthly during one year the geranylgeraniol, 6α-acetoxy- 7α-hydroxy-vouacapan, 6α-hydroxy-7α-acetoxy-vouacapan-17α-oate methyl esther and 6α-acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther content of samples from São Carlos, São Paulo and Ponto Chique, Minais Gerais freshly picked and further observed monthly to verify if there was an increase of 6α-acetoxy-7α-hydroxy-vouacapan content overtime when maintained under 40°C temperature.

A statistical evaluation of the extraction procedures for Minas Gerais samples was performed by applying ANOVA and correlation analysis (Table 1), and we observed that there was no significant difference between the Soxhlet and RTS extraction methods (at 95% confidence level) with high linear correlation for both extraction methods.

In vitro cytotoxicity screening against human prostate cancer cell line displayed higher selectivity and potent anticancer activity with TGI 11.43 μg mL⁻¹ when higher 6α-acetoxy-7α-hydroxy-vouacapan over total voaucapan ratio (3.14) was achieved. Nevertheless, 6α-acetoxy- 7α-hydroxy-vouacapan maintained approximately the same content throughout the year among the samples in opposition to overall voaucapan content (Figure 2).

Figure 2: Ratio of 6α-acetoxy-7α-hydroxy-vouacapan (m/z 360) in relationship to total vouacapan content compared to total growth inhibition on prostrate cell line (PC-03) monitored throughout twelve months of dichloromethane extracts produced from Pterodon pubescens fruit of: a) Samples from Minas Gerais extracted by Soxhlet system and b) Samples from Minas Gerais extracted at room temperature with stirring. Arrows indicate the respective scale.

Samples from São Carlos at time zero had 26% geranylgeraniol content whereas Minas Gerais samples only achieved the highest content of 1.3%. Throughout the year, the samples maintained at 40oC demonstrated geranylgeraniol concentration to decrease with a straight relationship with the overall increase of total vouacapan content. Among the chemometrics tools, the score plot of PCA analysis from Minas Gerais samples, which showed different group patterns of samples evaluated (Figure 3a), seem to demonstrate tendencies that give a comprehensive view of the chemical composition linked to the in vitro antiproliferative activity differences observed.

Figure 3: a) Score plot of samples from Minas Gerais (n=22) according to total growth inhibition (TGI) against prostate (PC-03) cell lines, considering low (L), moderate (M) and high (H) TGI potency. b) Loadings plot according to total growth inhibition against prostate (PC-03) cell lines grouped according to geranylgeraniol (A), 6α-acetoxy-7α-hydroxy-vouacapan (B), 6α,7α-di-hydroxy-vouacapan-17α-oate methyl esther (C), sum of isomers 6α-acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther and 6α-hydroxy-7α-acetoxyvouacapan-17α-oate methyl esther content (D), TGI (E), total voucapan content (F) and geranylgeraniol/isomers concentration ratio (G). Total variance, PC1+PC2=78%.

The correlations of data obtained from Minas Gerais samples were plotted in the light of PCA. The different parameters were analyzed by a multivariate approach considering TGI values compared to geranylgeraniol, isomers 6α-hydroxy-7α-acetoxy-vouacapan-17α-oate methyl esther and 6α-acetoxy-7α-hydroxy-vouacapan-17α-oate methyl esther, 6α-acetoxy- 7α- hydroxy-vouacapan and 6α,7α-di-hydroxy-vouacapan-17α-oate methyl esther content throughout twelve months maintained at 40oC. As reported in the score plot (Figure 3a) the PC1 includes most of the information (up to 51% of the total variance) due to 6α-acetoxy-7α- hydroxy-vouacapan ratio over total vouacapan content. The high loading values of geranylgeraniol, isomers 6α-hydroxy-7α-acetoxy-vouacapan- 17α-oate methyl esther and 6α-acetoxy-7α-hydroxy-vouacapan-17α- oate methyl esther, 6α-acetoxy-7α- hydroxyl-vouacapan and 6α,7α-dihydroxy- vouacapan-17α-oate methyl esther variables confirm, according to multivariate analysis, the major role of TGI potency where PC2 explains 27% of the total variance. The score plot reported in Figure 3a shows the inhibition of prostate cell line (PC-03) growth in the space of the two new variables PC1 and PC2. Moving along PC1 from left to right in the graph, we find different patterns of grouping with low anticancer potency (highest TGI values) close to the moderate anticancer potency samples separated from samples with high 6α-acetoxy-7α- hydroxy-vouacapan ratio over total vouacapan content with high anticancer potency (lowest TGI values) grouped in the first quadrant. For the sake of analyzing the loading plots TGI values above 52.74 μg mL⁻¹ were considered weak potency; between 52.74 μg mL⁻¹ and 19.28 μg mL⁻¹ considered medium potency and below 19.28 μg mL⁻¹ considered high potency.

Ultimately, the loadings plot reported in Figure 3b shows that variable with major influence in separation group with high anticancer potency (lowest TGI values) was 6α-acetoxy-7α-hydroxy-vouacapan, whereas the other variables were responsible for grouping samples with low and moderate anticancer potency. Samples from São Paulo (n=21) showed no obvious groups separation (Figure 2S).

This study highlights the complexity of factors involved in the production of secondary metabolites in plants. To enable the development of an herbal medicine, one of the greatest challenges is plant input standardization in order to meet efficacy, safety and reproducibility final product’s requirements as recommended by Brazilian Sanitary Regulatory Agency (ANVISA).

The aim of this work was to interpret the analytical data for the understanding of Pterodon pubescens Benth. Fruit features necessary to enhance in vitro antiproliferative proprieties viewing plant input standardization. The combination of antiproliferative tests, comparative analyses, and chemometric evaluation provided information to characterize the species properties involved with important in vitro antiproliferative events of PC-03 cancer strains.

Data presented herein suggest that 6α-hydroxy-7α-acetoxyvouacapan plays an important role over PC-03 cancer cell line growth inhibition. This furanditerpene over the time span studied was not affected by temperatures applied during extraction production process or storage temperature.

The significant 6α-hydroxy-7α-acetoxy-vouacapan concentrations detected among samples from different Brazilian regions could be attributed to different genotypes other than post-harvest and atmosphere conditions.

The application of PCA allowed the evaluation of similarities and differences between samples, being observed the formation of three main groups from Minas Gerais samples based on total growth inhibition (TGI) against prostate (PC-03) cell lines.

Further studies with microsatellite markers, one of the most informative and versatile DNA-based markers used in plant genetic research, are recommended to understand the variability among this species permitting to establish the most convenient phenotype for standardization of plant input.

The authors declare no competing financial interest.

Supplementary data relative additional PCA analysis.

The authors acknowledge Fapesp 2012/19661-4, CNPq and CAPES for financial support granted.