Skip to main content
ARS Home » Research » Publications at this Location » Publication #210796


item Delwiche, Stephen - Steve

Submitted to: Journal of the American Oil Chemists' Society
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/18/2007
Publication Date: 10/18/2007
Citation: Delwiche, S.R., Pordesimo, L.O., Panthee, D., Pantalone, V.R. 2007. Assessing glycinin (11s) and beta-conglycinin (7s) fractions of soybean storage protein by near-infrared spectroscopy. Journal of the American Oil Chemists' Society. 84(12):1107-1115.

Interpretive Summary: Soybean breeders have mentioned the complexity, slowness, and danger from the use of hazardous chemicals in HPLC and gas chromatography procedures as reasons for their desire to have near infrared procedures in soybean breeding programs. It would help breeders tremendously if there was a rapid method of assessing desirable quality traits for screening purposes at the early stages of plant breeding. However, at early stages in plant breeding samples are very limited; oftentimes only a handful of seeds is available. Currently, the fastest method for determining total soybean crude protein and oil is near-infrared (NIR) spectroscopy, but conventional instrumentation typically requires hundreds of grams per sample of whole beans. Alternatively, instrumentation can scan ground material, though this preliminary step is time consuming and destructive. Obtaining near infrared spectra from intact, ideally individual, seeds would be advantageous because of sample preservation and simple sample preparation. Plant breeders could use a nondestructive, single-seed spectroscopic method to screen genotypes without sacrificing any of the limited seed stock. A study was conducted to evaluate two sample presentation methods for NIR spectroscopy of soybean: transmission through single beans standard reflectance of ground samples. Not only protein content was examined, but also the chief storage protein components, glycinin (11S) and beta-conglycinin, were also examined, as well as the sub-fractions that comprise these components. Glycinin is especially of interest because it contains elevated levels of sulfur-containing amino acids that are traditionally limited in soybean, such that breeding efforts are currently underway to enhance this trait. The results demonstrated that although overall protein content is measurable using either method, NIR measurement of the protein components and their sub-fractions is at best applicable in only preliminary screening stages.

Technical Abstract: Soybean breeding programs underway today are addressing the goal of improving the protein profile to benefit the human diet as well as that of livestock. Glycinin, a globulin storage protein of the meal and designated as the 11S size fraction by ultracentrifugation, is desirable because of its relative abundance of sulfur-containing amino acids, such as methionine and cysteine. Soybean breeding programs for enhancing glycinin stand to become more efficient provided a method such as near-infrared (NIR) technology can be used to rapidly, and ideally, nondestructively screen for this condition. The current study examined the feasibility of NIR measurement of glycinin and the other prevalent protein fraction, beta-conglycinin (7S size fraction), as well as the electrophoretically separable sub fractions that comprise these two components. From a population of 101 F6-derived recombinant inbred lines in a field replicated trial, initially developed from a cross of two breeding lines (one noted for high protein, the other for high yield), single whole soybeans were scanned in transmittance (800 to 1798 nm usable wavelength region, 24 beans / sample x 197 samples total). Additional scanning of the ground form of these samples was performed in reflectance (1100 to 2500 nm). Partial least squares (PLS) calibrations were developed, using the 24-bean average log(1/T) spectrum for each sample, as well as the average spectrum from duplicate packs of log(1/R) spectra of the meal. The results indicate that NIR prediction of 11S and 7S, as well as the sub fractions thereof, is not of sufficient accuracy for use in soybean breeding programs. The probable reasons for this inability include an inherent lack of spectral specificity of the protein fractions and a non-constant proportion of soluble-to-total protein.