J Am Oil Chem Soc (2016) 93:587–594 DOI 10.1007/s11746-016-2795-8
ORIGINAL PAPER
Analysis of Vitamin K1 in Soybean Seed: Assessing Levels in a Lineage Representing Over 35 Years of Breeding Melissa M. Thompson1 · Ashley Niemuth1 · Jane Sabbatini1 · David Levin1 · Matthew L. Breeze2 · Xin Li2 · Timothy Perez2 · Mary Taylor2 · George G. Harrigan2
Received: 3 March 2015 / Revised: 5 January 2016 / Accepted: 18 January 2016 / Published online: 25 February 2016 © AOCS 2016
Abstract Recommendations by the Organization of Economic Cooperation and Development (OECD) for compositional assessments of new genetically modified (GM) soybean now include measurement of vitamin K1. Here we employ an HPLC method to provide information on variation in levels of vitamin K1 in seed from soybean representing over 35 years of breeding. Vitamin K1 analysis was conducted on seed from six conventional lines and three GM glyphosate-tolerant lines grown concurrently at two sites in the US. Vitamin K1 levels in seed were influenced by both location and germplasm. Levels were higher in the newer higher-yielding lines when compared with the older varieties. The effect of GM was negligible. Overall, our study showed, using a method suitable for compositional assessments under OECD guidelines, that (i) variation in levels of vitamin K1 in soybean is associated with a safe history of consumption and (ii) that modern breeding strategies can be employed for soybean varietal development without adversely impacting vitamin K1 levels. Keywords Vitamin K · Phylloquinone · Phytomenadione · Phytonadione · HPLC · Soybean · GM · Composition variation · Statistical analysis
Electronic supplementary material The online version of this article (doi:10.1007/s11746-016-2795-8) contains supplementary material, which is available to authorized users. * David Levin
[email protected] 1
Covance Laboratories Inc., 3301 Kinsman Boulevard, Madison, WI 53704, USA
2
Monsanto Company, 800 North Lindbergh Boulevard, St. Louis, MO 63167, USA
Introduction Decades of compositional studies on GM crops have consistently shown no meaningful differences between GM crops and their conventional counterparts [1–9] and have contributed to an assurance of GM safety. Composition studies conducted as part of comparative safety assessments follow internationally recognized guidelines published by the Organization of Economic Co-operation Development (OECD). A recently revised consensus document on compositional considerations for new varieties of soybean published by the OECD [10] included a new recommendation to evaluate vitamin K in addition to the nutrients and anti-nutrients that have typically been assessed in the past. Vitamin K comprises a group of fat-soluble vitamins and is known to be present in soybean [11, 12]. It is found in various leafy green vegetables and plant oils [13] and is recommended as part of a balanced diet due to its role as an essential nutrient [14–16]. While there is now extensive literature on naturally occurring variation in levels of soybean nutrient and other metabolites that have traditionally been measured in comparative assessments of new GM crops, there is less information on vitamin K1 (phylloquinone). The purpose of the work reported here was to conduct a case study on variation in soybean seed vitamin K1 levels. In the context of this goal, it should be noted that numerous breeding programs are committed to developing new soybean varieties with improved agronomic characteristics [17, 18] or nutritional profiles [19] and that extensive research on the genetic base and diversity of modern soybean [20] as well as on the influence of germplasm and environment on yield [21] and seed composition [1, 3, 22–24] are keys to enhanced trait development. For this study, a genetically characterized
13
588
soybean lineage representing ~35 years of breeding (launch years 1972–2008) and increasing yield potential was utilized. This allowed an evaluation of not only variation in vitamin K1 levels but whether any observed variation could be associated with genetic or agronomic features. The selected varieties included six different conventional (nonGM) and three different glyphosate-tolerant (GM) lines and all were grown concurrently at two replicated field sites in the US during the 2011 growing season. The analytical method was based on reversed-phased C30 HPLC and fluorescence detection. Reported values are for the biologically active vitamin K1.
J Am Oil Chem Soc (2016) 93:587–594 Table 1 Effect of varietal development on vitamin K1 levels Soybean varie- launch year Mean Min value Max value ties (mg/kg dwt) (mg/kg dwt) (mg/kg dwt) Williams 1972 A3127 1979 CX366 1986 CX375 1996 AG3701 (GM) 1997 A3469 1999 AG3705 (GM) 2006 A3555 2008
0.575 0.650 0.627 0.608 0.719 0.774 0.803 0.730
0.490 0.521 0.503 0.516 0.647 0.681 0.749 0.667
0.631 0.770 0.716 0.837 0.788 0.909 0.903 0.784
AG3803 (GM) 2008
0.818
0.711
0.931
Materials and Methods Biological Material A list of the selected soybean varieties with associated commercial launch dates is presented in Table 1. Details on the growth and harvest, and measured yield of the soybean lines at two different sites in the US (Jacksonville and Jerseyville Counties, IL) have been presented elsewhere [22, 25]. Seeds were harvested at maturity, homogenized by grinding with dry ice to a fine powder, and stored frozen at approximately −20 °C. Samples were lyophilized prior to analysis and stored frozen when not in use. Compositional (excluding vitamin K1) and metabolomics assessments have been presented in [22, 25]. Subsamples from these studies were utilized in the vitamin K1 assessments presented here. Genetic characterization of all varieties using the Illumina® Infinium platform is presented in Ref. [25]. Standards A stock standard solution of phytonadione (vitamin K1) (US Pharmacopeial Convention, Rockville, MD) was prepared in iso-octane at 1 mg/mL. Calibration standard solutions were prepared from the stock with a final concentration range of about 0.008–0.2 µg/mL, and a solvent composition of 10 % iso-octane in methanol. Chemicals All solvents used were HPLC grade. All other reagents were ACS grade.
evaluated but showed low recoveries (~50 %) based on fortified samples. Extraction was therefore pursued utilizing the procedure recommended by USP 35-NF 30 [27]. The vitamin K1 USP method uses DMSO and hexane as extraction solvents. The method was validated for soy using full fat soy flour as the test matrix. Average recovery when the matrix was fortified with a known amount of vitamin K1 reference material was 96.7 %. Precision on this matrix for 20 replicate measurements resulted in a relative standard deviation of 3.5 % with the average vitamin K1 content of 0.437 µg/g. For the soybean era samples of this study, test portions (1.0 g) of ground soybean seed were weighed into 50-mL plastic centrifuge tubes. To each test portion were added 10 mL of DMSO and 15 mL of hexane, and the contents were mixed on a vortexing apparatus for about 5 s. The samples were placed in a shaking water bath at 60 °C for 45 min. After centrifugation at 3000 rpm for 10 min, the upper hexane layer was transferred to a 100-mL volumetric flask. Four additional extractions were performed by adding 15 mL of hexane followed by shaking (sample tubes oriented horizontally) for 5 min using a platform orbital action shaker. Following centrifugation at 3000 rpm for 10 min, the hexane layer was drawn off and combined with the previous extract in the 100-mL volumetric flask. A 10-mL aliquot of the hexane extract was transferred to a 15-mL glass culture tube and evaporated to dryness under a stream of nitrogen using a heating block temperature of 30 °C. Sample residues were reconstituted by adding 1 mL of 30:70 DCM/MeOH to each culture tube and vortex mixing for about 5 s. Vitamin K1 analyses were conducted in a laboratory fitted with UV filtered lighting (GE Ecolux® covRguard® T8) to mitigate against photodegradation.
Extraction HPLC Analysis Extraction of vitamin K1 from food matrices typically involves direct extraction of organic solvents and several methods exist. Initially, AOAC 999.15-2003 [26] was
13
Analysis of vitamin K was carried out using reversed phase chromatography with post-column zinc reduction. The
J Am Oil Chem Soc (2016) 93:587–594
0.9
Vitamin K (mg/kg dwt)
Fig. 1 This figure above shows, through the use of separate boxplots to represent data from the two different growing locations [Jacksonville, IL (ILJA) and Jerseyville, IL (ILJE)], the influence of environment on vitamin K1 values in soybean. For each boxplot, the “whiskers” extend to the minimum and maximum values, the box indicates the 25th and 75th percentiles of the data, the middle line indicates the median, and the diamond indicates the mean. The ILJA site mean value was 0.742 mg/kg dwt and the ILJE site mean was 0.658 mg/ kg dwt. For both sites, the data is based on all nine different soybean varieties where, n = 6 biological replicates (except for AG3705; n = 5 at ILJA, and CX375; n = 5 at ILJE)
589
0.8
0.7
0.6
0.5 ILJA
Table 2 Comparisons of group means across sites b
ILJE
Site
a
p value
Comparison of group means
Difference (mg/kg dwt)
SE
Group A–Group B Group A–Group B conv. Group A–Group B GMO
0.156 0.138 0.167
0.0077 0.0099 0.0088
0.0000 0.0000 0.0000
Group B conv.–Group B GMO
0.029
0.0105
0.0074
a
Group A includes the four lines from 1996 and earlier and Group B includes the all five later lines. All lines in Group A are conventional whereas two lines in Group B are conventional (conv.) and three lines are GMO b
Standard error
analytical HPLC system was an Agilent 1200 Series (Santa Clara, CA) with a binary pump using only a single channel, autoinjector, thermostatically controlled column compartment and fluorescence detector. The separation was carried out isocratically with a YMC C30 column (4.6 × 250 mm, 5 µm) (Kyoto, Japan) with the column compartment temperature set at 26 °C. The mobile phase was prepared as follows: 200 mL dichloromethane, 1800 mL methanol, 2.76 g zinc chloride, 0.82 g anhydrous sodium acetate, and 0.54 mL glacial acetic acid were combined, mixed, and degassed. A zinc post-column reduction assembly was placed in line between the analytical column and the detector. The reductor was prepared by dry packing a Restek 4.6 × 30 mm assembly (Bellefonte, PA, cat #25131), with zinc dust (particle size <10 µm, ≥98 %, SigmaAldrich). The hydroquinone products of the vitamin K isomers were detected with an Agilent G1321A fluorescence
detector operated at excitation wavelength of 243 nm and emission wavelength of 430 nm. Injections of standards and sample extracts were made with a 50 µL volume. No correlation (r = −0.03) between phylloquinone and fat content (based on fat data presented in Ref. [22]) was observed confirming the validity of the method and its insensitivity to lipid content (data not shown). Statistical Analysis Equation (1) shown below was used to generate leastsquares means for all varieties and relative contrasts for Group A, Group B, and sub-groups within Group B as described later. Groups A and B refer to earlier (pre-1997) and newer (post-1997) varieties, respectively. This grouping was established ex post facto and was based on observations from earlier studies on the genetic similarity of the nine test varieties.
yijk = µ + Li + R(L)j(i) + Pk + LPik + eijk
(1)
yijk response of plot; μ overall mean across treatments; Li fixed location effect; R(L)j(i) random replicate effect within location; Pk fixed pedigree effect; LPik fixed location by pedigree effect; eijk random residual effect. Equation (1) was also used for assessing distribution of differences to assess the effect of location and pedigree on vitamin K1 levels. Equation (2) shown below, was applied to serve for estimation of variance components: yijklm = µ + Li + R(L)j(i) + Gl + M(G)m(l) + P(MG)k(ml) + eijklm
(2)
13
590
J Am Oil Chem Soc (2016) 93:587–594
Vitamin K (mg/kg dwt)
0.9
0.8
0.7
0.6
0.5 A
A
A
08 20 380 G3
8 00 -2 55 35
72 19 S-
06 20 570 G3
M
9 97 -1 27 31
A LI IL
9 99 -1 69 34 A 97 19 170 G3 A 96 19 537 CX 86 19 636 CX
A
W
Pedigree Fig. 2 Demonstration of how varietal development has influenced vitamin K1 levels in soybean. For each boxplot (associated with a different soybean variety) the “whiskers” extend to the minimum and maximum values, the box indicates the 25th and 75th percentiles of the data, the middle line indicates the median, and the diamond indicates the mean. For each soybean variety, values are based on data
from both growing sites (ILJA and ILJE) and on n = 12 biological replicates (except for AG3705 and CX375; n = 11). Prior to 1997, vitamin K1 mean values across the different soybean varieties were relatively consistent between 0.575 and 0.650 mg/kg dwt. Following 1997, vitamin K1 mean values increased to a range of 0.719– 0.818 mg/kg dwt
yijklm response of plot; μ overall mean across treatments; Li random location effect; R(L)j(i) random replicate effect within location; Gl random group effect; M(G)m(l) random GMO effect within group; P(MG)k(ml) random pedigree effect within GMO and group; eijklm random residual effect. All statistical computations were performed in SAS [SAS Software Release 9.4 (TS1M0)].
in yield potential. Vitamin K1 was unlikely to be a target of selection for these varieties so any changes may be incidental or unintended. The soybean varieties were also grown at two different, albeit geographically close, sites to provide information on possible environmental influences on variation in levels of vitamin K1. The first step in the evaluation showed that levels of vitamin K1 were influenced by both location and germplasm. Figure 1 shows that, overall, vitamin K1 levels were higher at the Jacksonville, IL (ILJA) site when compared to the Jerseyville, IL (ILJE) site. The ILJA site was associated with higher seed yields [22, 25] for all varieties when compared to the site ILJE, suggesting a possible positive association between vitamin K1 and yield. The influence of geographic location on vitamin K1 levels on leafy green vegetables has been reported [28]. For example, when cabbage, Swiss chard, leaf lettuce, spinach, and kale were grown at Boston, US, and Montreal, Canada, during the same growing season, vitamin K1 levels were observed to be higher in the Montreal samples clearly illustrating the influence of environment [28]. Values of vitamin K1 were also influenced by germplasm (Tables 1, 2; Fig. 2). As shown in Fig. 2, levels of vitamin K1 were higher overall in the newer higheryielding varieties (Group B; 1998 launch and later) when
Results and Discussion Vitamin K1 was evaluated in this study because of its inclusion in the 2012 revised consensus document on compositional considerations for new varieties of soybean published by the OECD [10] as well as the limited amount of data on natural variation in levels of vitamin K1 this crop. To better provide information on whether variation in vitamin K1 levels could be associated with genetic or agronomic features a lineage representing ~35 years of soybean breeding was assessed. In other words, the soybean varieties used in this study were selected to evaluate vitamin K1 levels in the context of decades of soybean breeding (association with a history of safe consumption), genetic diversity (influence of conventional breeding), and improvement
13
J Am Oil Chem Soc (2016) 93:587–594
591
BG
B
0.8
BC
BG
B
BG
Vitamin K (mg/kg dwt)
BC B BC
0.7 A
A
0.6 A
A
B
BC
BG
ILJA ILJA - ILJA - ILJA - ILJA - ILJE ILJE - ILJE - ILJE - ILJE A B BC BG A B BC BG
Group / Subgroup LSmean
Fig. 3 Least-square means for Groups A, Group B, conventional lines in Group B (BC) and GM lines in Group B (BG). The data reflect the relative lack of influence of GM on vitamin K levels when compared to location and conventional breeding. Values in the leftpanel are based on data from both Jacksonville, IL (ILJA) and Jerseyville, IL (ILJE) sites combined; values in the middle are based on data from ILJA, and values in the right panel are from ILJE. All
Lower
Fig. 4 Percentage of total variability attributed to each factor as estimated by variance components analysis (see “Materials and Methods” for details of statistical model). The group term captures variation that is attributable to differences between the older (Group A, 1996 and earlier) and newer (Group B, 1997 and later) varieties; the location term captures variation due to differences between the two growing locations (Jacksonville and Jerseyville). The pedigree (Group*GMO) term is associated with variation attributable to differences between varieties within a group; the GMO (Group) term is associated with differences between the GMO and conventional lines within Group B
Upper
biological replicates for the soybean samples were included. Group A includes the four varieties from 1996 or earlier (Williams, A3127, CX366, and CX375), Group B includes all five later lines (AG3701, A3469, AG3705, A3555, and AG3803), Group BC is a subset of Group B and includes only the conventional lines (A3469 and A3555), and Group BG is a subset of Group B that includes only the GM lines (AG3701, AG3705, and AG3803)
Percent 70.00% 62.46% 60.00%
50.00%
40.00%
30.00% 20.19%
20.00%
9.51%
10.00%
0.00%
7.61% 0.24%
Gr
oup
compared to the older lower-yielding lines (Group A; 1996 launch and earlier). This is consistent with the suggestion, based on the effect of growing location above, of a possible
LO
CA
TIO
Re N
PE
sid
DI
ual
GR
BL
0.00%
EE
OC
1(G
rou
p*G
MO )
GM
O(
K(
LO CA
TIO
Gr
oup
)
N)
Factor
positive association of vitamin K1 with yield. There is, to our knowledge, little information on the relationship between levels of vitamin K1 and soybean yield but these
13
592
Fig. 5 a Distribution of site differences in vitamin K1 means of soybean varieties within the same group (Group A; 1996 and earlier varieties) or subgroup [Group B_Conv (conventional lines in Group B; 1997 and later varieties) and Group B_GMO (GM lines in Group B)]. For each panel, and within the group or subgroup associated with that panel, a series of pairwise comparisons involving the Jacksonville, IL (ILJA) and Jerseyville, IL (ILJE) soybean varieties were conducted. As an illustration, panel (6) shows differences in vitamin K1 levels involving four pairwise comparisons of the ILJA and ILJE means for A3469 and A3555 means. The fact that the differences are above the zero reference line indicates that values of both varieties are higher at the same location (i.e., there is a location effect). This location effect extends to all groups and subgroups as shown in the remaining pan-
13
J Am Oil Chem Soc (2016) 93:587–594
els. b Distribution of differences in vitamin K1 individual site means between soybean varieties associated with different groups (Group A; 1996 and earlier varieties) or subgroups [Group B_Conv (conventional lines in Group B; 1997 and later varieties) and Group B_GMO (GM lines in Group B)]. For each panel, a series of pairwise comparison involving the Jacksonville, IL (ILJA) and Jerseyville, IL (ILJE) soybean varieties of each represented group were conducted. As an illustration, panel (3) shows the distribution of differences in vitamin K1 levels based on a total of 24 pairwise comparisons involving ILJA and ILJE means for the three varieties associated with Group B_GMO (AG3701, AG3705, AG3803) with the means for the two varieties associated with Group B_Conv (13,469, A3555)
J Am Oil Chem Soc (2016) 93:587–594
initial results imply that yield gains in soybean will not compromise levels of this vitamin. The distinction in levels of vitamin K1 between the older and newer lines, shown in Fig. 2, was of interest since an earlier genetic fingerprinting study [22, 25] had previously established that although these varieties were all genetically distinct from each other they could be classified into two major groups [25]. In other words, the results of the vitamin K1 analysis appeared to match the genetic distinction of the four earlier lower-yielding varieties (Group A) and the five later higher-yielding varieties (Group B). We were therefore interested in contrasting the data from these two groups. Because Group B contained the three glyphosate-tolerant (GMO) varieties as well as two conventional lines, a comparison of these two B subgroups was warranted although Fig. 2 appeared to indicate no differences between these two subgroups (i.e., between the GM and conventional non-GM lines represented in Group B). Statistical Eq. (1) was used to generate the data represented in Fig. 3 and Table 2. Summary statistics are presented in the Supporting Information (Tables S1–S4). These data demonstrate that conventional breeding contributes to variation in vitamin K1 levels, and this is reflected in differences between newer and older varieties. In addition to the differences between the older (Group A) and newer (Group B) varieties the data illustrated no differences between the two Group B subgroups (i.e., between the GM and conventional non-GM lines represented in Group B). In the variance component analysis of these data (Fig. 4), the variation due to GM was estimated to be negligible relative to the other sources of variation. In fact, the GM effect was estimated to be the smallest source of variation, although this effect is actually confounded by the fact that GM and non-GM comparators in this study are not near-isogenic, i.e., effects of conventional breeding could, at least in principle, still be a factor if differences had been observed between these sets of comparators. The data was further evaluated as to the distribution of differences (of least-squares means) between varieties represented in Groups A and B. Figure 5 shows that by comparing site mean differences between varieties a clear location effect could be discerned (Fig. 5a). The mean difference between sites for all groups evaluated was around 0.1 mg/kg. Figure 5 also shows that the distribution of mean differences between A and the two B subgroups is essentially similar with the B subgroups also both showing average higher values (Fig. 5b). In contrast, the distribution of mean differences between the two B subgroups centers closely around zero confirming the results of the contrasts conducted in Eq. (1) and the variance component analysis. This confirmed that, in this study, GM contributed less to variation in vitamin K levels than the other sources of variation.
593
Conclusion A method for the measurement of vitamin K1 in soybean to support revisions in the OECD composition guidelines for soybean is described. It was applied here in a test case on variation in vitamin K1 levels in soybean. The genetic diversity associated with our soybean samples implied an appropriately representative assessment of vitamin K1 levels associated with a history of safe consumption and that represented yield potential differences. It further indicated that GM is not an important source of variation in vitamin K1 levels. Thus, we have demonstrated with the current method that (i) variation in levels of vitamin K1 in soybean are associated with a safe history of consumption and (ii) that modern breeding strategies can be employed for soybean yield development without adversely impacting vitamin K1 levels.
References 1. Harrigan G et al (2010) Natural variation in crop composition and the impact of transgenesis. Nature Biotech 28:402–404 2. Herman RA, Price WD (2013) Unintended compositional changes in genetically modified (GM) crops: 20 years of research. J Agric Food Chem 61:11695–11701 3. Berman KH et al (2010) Compositional equivalence of insectprotected glyphosate-tolerant soybean MON 87701 × MON 89788 to conventional soybean extends across different world regions and multiple growing seasons. J Agric Food Chem 59:11643–11651 4. Zhou J et al (2011) Compositional variability in conventional and glyphosate-tolerant soybean varieties grown in different regions in Brazil. J Agric Food Chem 59:11652–11656 5. Zhou J et al (2011) Stability of the compositional equivalence of grain from insect-protected corn and seed from herbicidetolerant soybean over multiple seasons, locations and breeding germplasms. J Agric Food Chem 59:8822–8828 6. Harrigan GG, Harrison JM (2012) Assessing compositional variability through graphical analysis and Bayesian statistical approaches; Case studies on transgenic crops. Biotechnol Genet Eng Rev 28:15–32 7. Harrison JM, Breeze ML, Harrigan GG (2011) Introduction to Bayesian statistical approaches to compositional analyses of transgenic crops 1. Model validation and setting the stage. Reg Toxicol Pharmacol 60:381–388 8. Harrison JM et al (2013) Principal variance component analysis of crop composition data: a case study on herbicide-tolerant cotton. J Agric Food Chem 61:6412–6422 9. Herman et al (2009) Compositional assessment of transgenic crops: an idea whose time has passed. Trends Biotechnol 27:555–557 10. OECD (2012) Revised consensus document on compositional considerations for new varieties of soybean [Glycine max (L.) Merr.]: key food and feed nutrients, antinutrients, toxicants and allergens. OECD, Rome 11. Damon M et al (2005) Phylloquinone (vitamin K1) content of vegetables. J Food Comp Anal 18:751–758 12. SELF Nutrition Data. Nutrition facts, calories in food, labels, nutritional information and analysis. http://nutritiondata.self.
13
594 com/facts/legumes-and-legume-products/4375/2. Accessed 6 December 2013 13. USDA National Nutrient Database for Standard Refer ence, Release 25. https://www.ars.usda.gov/SP2UserFiles/ Place/12354500/Data/SR25/nutrlist/sr25w430.pdf. Accessed 6 December 2013 14. Fujita Y et al (2012) Association between vitamin K intake from fermented soybeans, natto, and bone mineral density in elderly Japanese men: the Fujiwara-kyo osteoporosis risk in men (FORMEN) study. Osteoporos Int 23:705–714 15. Stevenson M, Lloyd-Jones M, Papaioannou D (2009) Vitamin K to prevent fractures in older women: systematic review and economic evaluation. Health Technol Assess 45:1–134 16. National Research Council (2001) Dietary reference intakes for vitamin A, vitamin K, arsenic, boron, chromium, copper, iodine, iron, manganese, molybdenum, nickel, silicon, vanadium, and zinc. The National Academies Press, Washington 17. Mikel M et al (2012) Genetic diversity and agronomic improvement of North American soybean germplasm. Crop Sci 50:1219–1229 18. Thompson J, Nelson R (1998) Utilization of diverse germplasm for soybean yield improvement. Crop Sci 38:1362–1368 19. Clemente T, Cahoon E (2009) Soybean oil: genetic approaches for modification of functionality and total content. Plant Physiol 151:1030–1040 20. Lam H et al (2010) Resequencing of 31 wild and cultivated soybean genomes identifies patterns of genetic diversity and selection. Nature Gen 42:1053–1059
13
J Am Oil Chem Soc (2016) 93:587–594 21. Specht J, Hume D, Kumudini S (1999) Soybean yield potential— a genetic and physiological perspective. Crop Sci 39:1560–1570 22. Harrigan GG (2013) Investigation of biochemical diversity in a soybean lineage representing 35 years of breeding. J Ag Food Chem 61:10807–10815 23. Rotundo J, Westgate M (2009) Meta-analysis of environ mental effects on soybean seed composition. Field Crop Res 110:147–156 24. Medic J et al (2014) Current knowledge in soybean composition. J Am Oil Chem Soc 91:363–384 25. Kusano M et al (2015) Assessing metabolomic and chemical diversity of a soybean lineage representing 35 years of breeding. Metabolomics 11:261–270 26. AOAC International Official Method 999.15 (2003) Vitamin K in milk and infant formulas. Liquid chromatographic method. Association of Analytical Chemists International, Gaithersburg 27. United States Pharmacopeia and National Formulary (USP/NF) (2012) Oil and water-soluble vitamins with minerals tablets, method 1. USP 35-NF 30, p 1567 28. Ferland G, Sadowski JA (1992) Vitamin K1 (phylloquinone) content of green vegetables: effects of plant maturation and geographical growth location. J Ag Food Chem 40:1874–1877