Interlaboratory comparison of simmondsin analysis
Thomas P. Abbott
a,*, Gerda Flo
b, Lothar Frank
c, Ronald A. Holser
a,
Paul Kolodziejczyk
d, David York
e, Terry Nelsen
aaNew Crops Research,National Center for Agricultural Utilization Research,Agricultural Research Ser6ice,USDA**,
1815N. Uni6ersity St.,Peria,IL61604,USA
bInterdisciplinary Research Center,Katholieke Uni6ersiteit Leu6en Campus Kortrijk,B-8500,Kortrijk,Belgium cLark Enterprises,Inc.,Webster,MA01570,USA
dPOS Pilot Plant Corp.,Saskatoon,SK,Canada S7N2R4
ePennington Biomedical Research Center,Louisiana State Uni6ersity,Baton Rouge,LA70808-4124,USA Accepted 12 May 2000
Abstract
Eleven samples containing various amounts of simmondsin (S), simmondsin ferulate (SF), demethyl simmondsins (DMS) and didemethyl simmondsins (DDMS) were analyzed by five different laboratories. The samples were made from chromatographically pure simmondsin, animal feed formulations containing jojoba meal, defatted jojoba meal, water extracts of jojoba meal and combinations of these ingredients. Where mixes were made, all materials were ground together in a coffee mill and sieved. Four laboratories analyzed for simmondsin and related components and one laboratory analyzed for only S. The means of the S, SF, DMS and DDMS percentages in the samples were determined to be: high performance liquid chromatography (HPLC) purified simmondsin, 94.1% S, 0 SF, 6.30% DMS, 0.52% DDMS; recrystallized simmondsin, 99.6% S, 0 SF, 1.01% DMS, 0 DDMS; water extract of jojoba meal 1, 29.2% S, 2.62% SF, 3.45% DMS, 9.47% DDMS; water extract of jojoba meal 2, 20.6% S, 2.00% SF, 2.81% DMS, 8.66% DDMS; formulated pet food with simmondsin, 0.59% S, 0 SF, 0 DMS, 0.01% DDMS; defatted jojoba presscake 7.05% S, 1.55% SF, 1.34% DMS, 3.48% DDMS. Using a Rank-Sum test, no laboratory demonstrated a consistently higher or lower bias compared to other laboratories for simmondsin analysis. Simmondsin analysis had less variability (C.V.=44) than other component analyses. Reproducibility for a blind duplicate sample of defatted jojoba presscake demonstrated that four of the five laboratories were consistent in simmondsin analysis. Water extracts of jojoba meal were shown to be highly variable in simmondsin content. Published by Elsevier Science B.V.
Keywords:Simmondsin; Analysis; Interlaboratory; Capillary electrophoresis
www.elsevier.com/locate/indcrop
Names are necessary to report factually on available data; however, the USDA neither guarantees nor warrants the standard of the product, and the use of the name by USDA implies no approval of the product to the exclusion of others that may also be suitable.
* Corresponding author. Tel.: +1-309-6816533; fax: +1-309-6816524.
E-mail address:[email protected] (T.P. Abbott).
1. Introduction
Jojoba, Simmondsia chinensis (Link) Schneider
is native to the desert Southwestern US and Mex-ico. It is currently being grown on 8000 – 10 000 acres in the US and elsewhere worldwide. Jojoba has a unique wax ester oil which is 50 – 60% of its seed weight. Jojoba oil has good markets in cos-metics and lubricants. The rest of the seed is not used as much as the oil although it contains about 25% crude protein after the oil is removed. The defatted meal contains sugars and 11 – 15% of a unique group of natural products, all structurally related to simmondsin (S). Cokelaere has shown that S is an effective hunger satiation agent and reduces food intake in mice, rats and chickens (Cokelaere et al., 1995). Jojoba meal has been used for its S content to regulate food intake of animals, and the water washed meal is an effective cattle feed supplement (Vontungelin et al., 1997). The isolation of S was first described by Elliger et al. (1973). Several methods, usually high perfor-mance liquid chromatography (HPLC), for quan-titative analysis of S and related substances in jojoba have been published (Verbiscar et al., 1980; Van Boven et al., 1996; Abbott et al., 1999). One problem is that of the four major components S, simmondsin ferulate (SF), demethyl simmondsins (DMS) and didemethyl simmondsins (DDMS) the latter three have either positional or geometric isomers (Van Boven et al., 1996, 2000). The de-gree by which these isomers are separated by the
various procedures and which isomeric peaks should be measured led to differences in analysis results from different laboratories. In some cases, the extraction solvents were different and some were shown to extract more or less completely depending on the sample type and extraction method (Van Boven et al., 1996). This study was undertaken to compare S, SF, DMS and DDMS analyses between laboratories to provide insight to each one about differences between laborato-ries, interlaboratory biases and the consistency of results within each laboratory.
2. Materials and methods
Samples 1 – 11 and an approximate range for S content were provided to each laboratory in the collaborative study (Table 1). The ranges were
given as \50% samples 1, 2; 10 – 50% samples 3,
4, 7, 10; 0 – 10% samples 5, 6, 8, 9, 11. Pilot scale HPLC purified S, sample 1, and recrystallized S, sample 2, were prepared as described in Abbott et al. (1999). The water extracts of defatted jojoba meal, samples 3 and 4, were prepared as described in Erhan et al. (1997) with defatted jojoba meal provided by Purcell Jojoba International using two different meals at two different times. Animal feed, sample 5, was a proprietary formulation for experimental studies and is no longer available. Defatted jojoba presscake was also provided by Purcell Jojoba International. Mixtures of samples in Table 1 were made by grinding the components in a coffee mill for 1 min. HPLC was used by four of the five laboratories. Two HPLC procedures are described in Holser and Abbott (1999), and Van Boven et al. (1996). In these cases either methanol or water was used to extract the compo-nents of interest. Percent S was determined by comparison to standards made from isolated S and SF.
A third was performed on Waters NOVA PAK
C18, 4mm, 4.6×250 mm column, using a Waters
590 pump, a UV detector at 220 and 325 nm to
separate a 10ml injection of sample solution. The
sample solution was prepared by extracting 10 mg of solid sample with 4 ml of absolute ethanol in amber vials and diluting samples 1, 2, and 10 1:4 Table 1
Composition of samples tested in the study
Sample Components
Pilot scale HPLC pure simmondsin fraction 1
Recrystallized pure simmondsin 2
3 Water extract of jojoba meal 1 4 Water extract of jojoba meal 2 5 Animal feed containing jojoba meal
Defatted jojoba presscake 6
80% sample 5, 16% sample 2, 4% simmondsin 7
ferulate
8 90.9% sample 5, 9.1% sample 2 95.2 sample 5, 4.8% sample 2 9
Table 2
Simmondsin content of 11 samples — results from five laboratories
Laboratory no.
aValues significantly outside the 95% confidence limits of the means.
with ethanol. The solvent gradient was 0 – 50% solvent B (35% acetonitrile in solvent A) in 1 min, 50 – 100% B in 20 min, returning to
sol-vent A (5 mM Na2HPO4 in water plus
triethy-lamine, 1 ml/l titrated to pH 6.8 with acetic
acid). Retention times (min) were DDMS 8.5, DMS 9.3, S 10.3, SF 27.2. S and SF supplied by T. Abbott was used for calibration and standard curves gave correlation constants greater than 0.9993.
One laboratory, POS Pilot Plant, used a capil-lary electrophoresis method described elsewhere in this issue by Kolodziejczyk et al.
Means and confidence intervals at the 95% confidence level were calculated and values outside those intervals identified. A Rank-Sum
test was run atPB0.05 by adding the rank of the
result of each component in each sample as the highest (1), next highest (2), etc. value and adding the ranking of each laboratory for each compo-nent in all 11 samples and comparing the rank sums to values of limits of probable similar re-sults. The analyses of sample 6 and sample 11 were compared for internal laboratory reproduci-bility because it is the same sample and not a mixture.
3. Results and discussion
The results reported from five laboratories for S content is shown in Table 2. The values were determined by comparison to standards made from isolated S with a purity of 99%. Values significantly different from the means occurred more often in laboratory 5. If this were a result of heterogeneity of mixtures, they would have
oc-Table 3
Sample composition — average of five laboratories
SFa
Fig. 1. Reproducibility of percent simmondsin (S) in two identical samples (6 and 11) in four laboratories.
3) is about 94% pure with the chief identifiable contaminant being DMS. Recrystallized S from HPLC purified S (sample 2) is about 99.5% pure, if one assume values over 100% are part of a random distribution of errors, with DMS being the chief impurity. However, the recrystallizing solvent may be entrained in the crystals and not seen in the HPLC analysis. One has recently learned that denatured ethanol (5% methanol, 5% isopropanol) was used in the pilot-scale produc-tion and recrystallizaproduc-tion. In addiproduc-tion, repeated recrystallization of sample 2 leaves a slight amount of a brown residue. Identification of this minor impurity is proceeding because even minor impurities can have significant effects. S, prepared in the same way as sample 2 should be thoroughly ground and dried under vacuum or recrystallized from absolute ethanol before use in feeding stud-ies. The water extracts of jojoba meal (samples 3 and 4) are clearly enriched in S, DMS, DDMS and SF compared to jojoba meal (sample 6). There have been reports of breakdown of S in aqueous solutions but usually in the presence of the meal (Abbott et al., 1989). The differences between samples 3 and 4 may be attributed to the different meals extracted, the thoroughness of the extraction or degradation dependent on the length of time the extract was in contact with the meal. The average values for samples 6 and 11 (the same material) were quite close. The S measure
was less variable (average C.V.=47) than SF
(average C.V.=83, DDMS (average C.V.=89)
or DMS (average C.V.=93).
To test reproducibility within each laboratory, plots of the analyses for S, SF or DDMS in sample 6 versus those for sample 11 were made. Because the samples were from the same starting material, results not falling on a line with a slope of 1.0 would indicate lack of reproducibility. Fig. 1 indicates that laboratory 5 had poor reproduci-bility for S, but the other laboratories reproduced their results for the two samples well for all components (Figs. 1 – 3). Table 4 shows the Rank-Sum results and limit values of probable bias. For SF, laboratories 1 and 4 were consistently higher in their analyses. For DDMS, laboratory 3 was consistently higher and for DMS, laboratory 4 was consistently higher. For S, no one laboratory was consistently higher or lower.
Fig. 2. Reproducibility of percent simmondsin ferulate (SF) in two identical samples (6 and 11) in five laboratories.
Fig. 3. Reproducibility of percent didemethyl simmondsin (DDMS) in two identical samples (6 and 11) in four laborato-ries.
Table 4
Rank-sum test atPB0.05
SF
LaboratoryS DDMS DMS
17 20 19
1 44
23 22
34 28
2
32
3 27 15
31
4 15 18 14
23 5
Limits B21,\45 B18,\37 B18,\37 B15,\29
and related jojoba constituents. Ind. Crops Prod. 10 (1), 65 – 72.
Cokelaere, M., Flo, G., Decuypere, E., Vermaut, S., Daenens, P., Van Boven, M., 1995. Evidence for a satiating effect of defatted jojoba meal. Ind. Crops Prod. 4, 91 – 96. Elliger, C.A., Waiss, A.C. Jr, Lundin, R.E., 1973.
Sim-mondsin, an unusual 2-cyanomethylenecyclohexyl glu-coside fromSimmondsia californica. J. Chem. Soc. Perkin Trans. 1, 2209 – 2212.
Erhan, S., Abbott, T., Nabetani, H., Purcell, H., 1997. Sim-mondsin concentrate from defatted jojoba meal. Ind. Crops Prod. 6, 147 – 154.
Holser, R.A., Abbott, T.P., 1999. Extraction of simmondsins from defatted jojoba meal using aqueous ethanol. Ind. Crops Prod. 10, 41 – 46.
Van Boven, M., Daenens, P., Tygat, J., Cokeleare, M., 1996. Determination of simmondsins and simmondsin ferulates in jojoba meal and feed by high-performance liquid chro-matography. J. Agric. Food Chem. 44, 2239 – 2243. Van Boven, M., Busson, R., Cokelaere, M., Decuypere, E.,
Daenens, P., 2000. 4-Demethyl simmondsin from Sim
-mondsia chinensis. Ind. Crops Prod. 12, 203 – 208. Verbiscar, A.J., Baningan, T., Weber, W., Reid, B., Frei, J.,
Nelson, N.E., Raffauf, R.F., Kosersky, D., 1980. Detoxifi-cation of jojoba meal. J. Agric. Food Chem. 28, 571 – 578. Vontungelin, A., Phillips, W.A., Abbott, T.P., 1997. Dry matter intake and body weight changes in lambs fed differ-ent amounts of water washed de-oiled jojoba meal. J. Anim. Sci. 75 (Suppl. 1), 26.
The results of this interlaboratory study will be used to improve reproducibility of S, SF, DMS and DDMS analyses.
References
Abbott, T.P., Peterson, R.E., Nakamura, L.K., Nelsen, T.C., Bagby, M.O., 1989. Monitoring jojoba toxins by fourier transform infrared spectroscopy and HPLC. In: Baldwin, A.R. (Ed.), Proceedings of the Seventh International Con-ference on Jojoba and Its Uses. AOCS, Champaign, IL, pp. 440 – 450.
Abbott, T.P., Holser, R.A., Plattner, B.J., Plattner, R.D., Purcell, H.C., 1999. Pilot-scale isolation of simmondsin
