Supercritical Fluid Extraction and Chromatographic Analysis ( HRGC-FID and HRGC-MS ) of Lupinus spp . Alkaloids

Os extratos de alcalóides de Lupinus spp., obtidos por métodos convencionais (maceração/ sonicação – extração em fase sólida; maceração/sonicação –extração líquido-líquido) e por SFE (extração com fluido supercrítico) usando CO2 e CO2 modificado (CO2/MeOH, CO2/EtOH, CO2/iPrOH e CO2/H2O) foram analisados por CGAR-DIC (cromatografia gasosa de alta resolução com detector de ionização de chama) e CGAR-EM (cromatografia gasosa de alta resolução acoplada à espectrometria de massas). As análises quantitativas por CGAR-DIC foram feitas pelo método do padrão interno, para a quantificação de lupanina, multiflorina e um alcalóide derivado da esparteína. CGAR-EM permitiu a identificação dos constituintes químicos (alcalóides e outras substâncias) destes extratos.


Introduction
Lupinus spp.has been investigated in several countries as a potential alimentary source due to its relatively high protein and oil content (35 -40% and 8 -12%, respectively), high productivity and low cost.Although lupine protein levels are similar or larger than that of the soya bean, the main problem is the quinolizidine alkaloids, which are known to provide a bitter taste and toxicity to the seeds.An usual procedure for the elimination of these alkaloids is washing the seeds with flowing water; despite simple, this method requires large volumes of water for the commercial scale Lupinus processing [1][2][3] .SFE (supercritical fluid extraction) is a valuable method both for industrial scale food processing and also for analytical scale studies, as an alternative extraction method to reduce the use of liquid solvents (mainly organic solvents).However, analytical scale SFE of polar compounds is still underexplored, mainly due to the low diffusibility and low polarity of supercritical CO 2 .Some moderately polar natural products such as alkaloids and flavonoids have been extracted by SFE 4,5 .
In this work, the extraction of Lupinus spp.alkaloids by SFE is compared with conventional methods (maceration/sonication -solid phase extraction, and maceration/sonication -liquid-liquid extraction).The extracts were analysed by HRGC-FID (high resolution gas chromatography -flame ionisation detector), for the quantification of lupanine, multiflorine and a spartein-like alkaloid.HRGC-MS (high resolution gas chromatography coupled with mass spectrometry) analysis allowed the identification of the chemical constituents (alkaloids and other compounds) from the extracts obtained by different methods from Lupinus samples obtained from markets in São Paulo, SP.

Materials
Analytical reagent grade ethanol, ethanol, isopropanol, dichloromethane and acetone were purchased from Merck, Darmstadt, Germany.The carbon dioxide and nitrogen gas were supplied by White Martins, Rio de Janeiro, Brazil.

Sample extraction: conventional method (maceration/sonication)
Extraction was performed as schematically represented in Figure 1.Each sample was extracted in triplicate, and the extracts were analysed by HRGC-FID and HRGC-MS.The yield of each extraction was determined after drying (room temperature, under nitrogen flux, followed by drying in a vacuum oven at room temperature) until constant weight.

Sample extraction: SFE
Supercritical fluid extractions were performed on an analytical scale "home-made" system, previously described 6 .Powdered Lupinus samples (0.5 g each) were extracted for approximately 20 minutes in a dynamic mode, using as extraction fluids pure CO 2 or CO 2 modified with different solvents, as indicated in Table 1.
All the extracts were collected in analytical grade CH 2 Cl 2 (Merck, Darmstadt, Germany) contained in a test tube, in an ice bath.The solvent was removed at room temperature, under nitrogen flux, followed by drying in a vacuum oven at room temperature until constant weight.Whenever possible, extractions were made in triplicate (see remarks in Tables 2 and 3).Residues were dissolved in analytical grade methanol (Merck, Darmstadt, Germany) prior to HRGC analysis.

Clean up
Clean up was done by percolating the extract (obtained by conventional extraction or SFE), solubilized in 5.0 mL analytical grade methanol, through a glass Pasteur-type pipette containing 0.5 g silica gel 60, 70 -230 mesh (Merck, Darmstadt, Germany) and 0.5 g active charcoal (Reagen, Rio de Janeiro, Brazil).
Extracts obtained by SFE using aqueous mixtures have been submitted to SPE (solid phase extraction), using Sep-Pak C-18 cartridges (Waters), preconditioned with 8.0 mL methanol followed with 8.0 mL H 2 O.The extract was Sequential mode extractions were done firstly with pure CO 2 , followed by extraction of the same Lupinus sample using CO 2 modified with 10% EtOH.Temperature and pressure conditions of each step were as reported in Table 1.percolated through the cartridge and eluted with methanol, acetone, ethyl acetate and chloroform, successively (8.0 mL each).These fractions were collected and combined for total yield determination and chromatographic analysis.Some samples were not submitted to the clean up step and were directly analyzed by HRGC; they are indicated in Table 3.

Chromatographic analysis
HRGC-MS analyses were performed using a HP 5970 mass selective detector (Hewlett -Packard, USA), (EI, 70 eV), coupled to a HP 5890 GC.The column used was a 95% methyl, 5% phenylpolysiloxane, LM-5 (50 m x 0.25 mm x 0.65 mm) supplied by L & M (São Carlos, Brazil).Samples were injected using the split mode (1:30), with injector temperature and HRGC-MS interface temperature both at 300°C.The column temperature was programmed to rise from 170 °C (3.5 min), at 6 °C min -1 , to 300 °C (held during 20 min).Helium was used as carrier gas, at the average linear velocity of 35 cm sec -1 ; MS data were processed using a CPU HP 7946 / HP 9000-300.Tentative identifications were made by comparison of the obtained spectra with literature data 7,8 .The HRGC-FID analyses were performed on a HP 5890 GC, using the same column and the same temperature program as used for the HRGC-MS analysis.Detector (FID) temperature was 320 °C, split (1:30), injector temperature was 280 °C, data were obtained on HP 3396 A integrator.Hydrogen was used as the carrier gas, at an average linear velocity of 40 cm sec -1 .All quantitative analyses were made by the internal standard method, using caffeine as an analytical standard.The Lupinus extracts were diluted to 1.0 mL in methanol, and 0.3 mL of a caffeine standard solution (1 mg mL -1 ) was added to the sample, which was analyzed by HRGC-FID.For each alkaloid quantified a corresponding calibration curve was prepared (injections in triplicate for each concentration), and linearity for internal standard quantification was checked within the range of 0.05 -0.50 mg caffeine mL -1 .

Extraction methods
The yields for each extraction procedure were determined using both Brazilian commercial and Chilean samples (Table 2).The conventional extraction gave a similar yield, both by Extrelut preconcentration and liquid-liquid extraction (Figure 1).Most of the SFE procedures gave better yields than the conventional method.Some of the SFE experiments showed a large standard deviation, due to some specific problems of some fluid mixtures.For example, water extracted polar compounds, which plugged the system due to the production of foam.Another problem was the trapping of CO 2 extracts: losses of extracts due to plugging (ice formation in the extremity of restrictor) are inherent to extract collection in solvents (the process herein adopted).The time required for SFE was 20 minutes for each extraction while for conventional methods the total time was several hours for the complete procedure.

Chemical composition of the extracts
HRGC-MS analysis of the extracts obtained from a commercial Lupinus sample allowed tentative identification of three alkaloids, by comparison with literature data 7 : lupanine (1), multiflorine (2) and a spartein derivative (3).Many other compounds were identified as alkaloids, but a more detailed tentative identification was not possible.The main feature of these unidentified alkaloids was a peak at m/z = 58, which is found in several lupane-type alkaloids 7 .Some extracts also contained other compounds, mainly fatty acids and long chain hydrocarbons, which were identified by their MS profiles 9 .The CO 2 /H 2 O mixture required slightly stronger conditions, since the critical constants of H 2 O are significantly higher than those of organic solvents 10 .SFE using CO 2 modified with 5% H 2 O was the most selective condition for alkaloid extraction; however, the critical conditions of this mixture (Pc = 89.1 atm, Tc = 69.1 o C; calculated according to the literature 11 ), require a relatively high temperature for the usual working conditions with natural products.SFE using CO 2 modified with 10% methanol showed the best yield with a   lower temperature (Pc = 73.7 atm, Tc = 60.0 o C). Figure 2 (peaks key on Table 3) shows the TIC-HRGC-MS profile of these two extracts.

Quantitative analysis
Quantitative analyses were made for the main alkaloids.Lupanine was found in all of the Lupinus extracts.The regres-sion equations for the analytical curves were y= 0.0359 + 0.6647 x (r=0.999) for lupanine (1), y= -0.0236 + 0.1126 x (r= 0.999) for multiflorine (2) and y= 0.06203 + 0.02893 x (r= 0.968 ) for the spartein derivative (3).The average percentage standard error for the peak areas for replicate injections was less than 5%, showing good reproducibility.The content of each alkaloid in all Lupinus extracts is shown in Table 3.
The sum of the alkaloid content in the extract obtained by SFE using CO 2 modified with 10% ethanol and not submitted to clean-up before HRGC-FID was the greater of all extraction methods.Unfortunately EtOH usually shows problems in reproducibility as a SFE modifier, since commercial ethanol has a significant (for SFE) variation in water content.
Utilization of isopropanol as a modifier showed no significant improvement in extraction process (yield or selectivity), so this solvent should be considered only as a third option in the choice of modifiers, due to its cost and the difficulty of removing residual solvent from extracts.

Conclusions
The present results indicate that SFE can be used as an alternative to conventional methods for extraction of alkaloids from Lupinus.It is faster and had greater total yields for the extracts, and methanol was shown to be the best modifier for Lupinus extraction.Lupanin was the most abundant alkaloid in all the extracts (SFE and conventional method) from the samples studied.The utilization of water as a modifier may be a useful tool for the extraction of polar alkaloids, but only for qualitative analysis for the alkaloids herein analyzed.
Lupinus mutabilis, L. albus and L. sp.(Leguminosae-Papilonaceae) were collected in Chile by one of the authors (D. von B.).Commercial lupine samples were bought from local markets in São Paulo, SP, Brazil.The samples were dried (ca.40°C), powdered and ground (70-100 mesh).

Figure 1 .
Figure 1.Schematic drawing of the conventional extraction procedure and clean-up for Lupinus samples.

Table 1 .
Conditions of the fluid mixtures used in SFE experiments.

Table 2 .
Yield for total extracts from Lupinus samples obtained by conventional and SFE extraction methods (expressed in mg of alkaloid /g plant material).

Table 3 .
Content of alkaloids in Lupinus samples (expressed in mg of alkaloid /g plant material).

Table 4 .
Compounds and respective MS data (EI, 70 eV) found in the extracts of a commercial Lupinus sample.