Development of an automated system for pyruvic
acid analysis in onion breeding
Kil Sun Yoo
*, Leonard M. Pike
Vegetable Improvement Centre, Department of Horticultural Sciences, Texas A&M University, College Station, TX 77843, USA
Accepted 5 May 1999
Abstract
An automated system for pyruvic acid analysis was developed in order to screen a large number of onion bulbs in our breeding program. The system includes two high performance liquid chromatography (HPLC) pumps, an autosampler, a column heater, a spectrophotometric detector, and an integrator. With this new method, human error can be eliminated with great repeatability (CV0.9%), and we can run one sample per min. In a test using five cultivars, this automated
system showed highly significant correlations with the HPLC and the spectrophotometric method (r2> 0.99), and estimated about 7% and 17% more pyruvic acid over the HPLC and spectrophotometric methods, respectively. Interference by sugars (48 mg mlÿ1
) and quercetin compounds (68mg mlÿ1) was about 0.1mmoles mlÿ1 as pyruvic acid.# 1999 Elsevier Science
B.V. All rights reserved.
Keywords: Allium cepaL.; Flavor; Pungency; Breeding; Screening
1. Introduction
Onion tissues produce distinctive volatile sulfur compounds when the cells are damaged mechanically. S-alk(en)yl-L-cysteine-sulfoxides, the flavor precursor
compounds, are hydrolyzed by the action of alliinase to produce thiopropanal S-oxide (tear-causing factor), pyruvic acid, and ammonia (Whitaker, 1976). The
* Corresponding author. Tel.: +1-409-847-8848; fax: +1-409-862-4522 E-mail address:k-yoo@tamu.edu (K.S. Yoo)
thiopropanal S-oxide degrades into mono-, di-, or tri-sulfides through complex chemical reaction, such as via thiosulfinates. Onion pungency has been estimated by measuring one of the enzymatic reaction products by gas chromatography, high performance liquid chromatography (HPLC), or spectrophotometric methods. Because pyruvic acid measurement is simple and fast, it has been widely used to estimate onion pungency since the 1960's.
The original method by Schwimmer and Weston (1961) (SW) includes blending onions with water, filtering, centrifuging, and diluting before analysis. This method was slow and tedious when used in a breeding program requiring analysis of many samples. Several modifications are made to the original method for fast analysis. Randle and Bussard (1993) developed a fast juice extraction method by using a press instead of blending bulb tissues with water, and reported that 120 samples could be extracted per day. Yoo et al. (1995) (Yoo) extracted juice by blending onion tissues without adding water, then reacted the juice with reagents without two steps of dilution. We saved time by eliminating dilution steps and by reduced dish-washing work. Using frozen juice samples, two technicians could process up to 250 samples a day by this method.
To improve the current analysis system, we developed an automated system to avoid pipetting error and to reduce labor costs. This new method uses the same principle as the original Schwimmer and Weston (1961) method, but is automatic and should reduce human error.
2. Materials and methods
2.1. Apparatus
Two HPLC pumps (Model Acuflow II, Fisher Scientific, Pittsburgh, USA) were used to deliver dinitro phenyl hydrazine (DNPH) and NaOH solutions (Fig. 1). Each pump delivered 0.7 ml minÿ1 DNPH solution (25 mg DNPH and
25 ml phosphoric acid per liter) or 1.4 ml minÿ1
NaOH solution (20 g lÿ1
). Onion juice sample was filtered through a 0.45mm nylon filter (microfilter) and a 5ml sample was injected into the DNPH solution by an autosampler (Model 570,
Alltech, Deerfield, IL, USA) at an interval of 1 min. The mixture of onion juice and DNPH was heated in a 3 m long HPLC tubing coiled in a column heater set to 708C. The NaOH solution was mixed with the DNPH/onion juice at the end of the heating coil. The final mixture was passed through a column (4.6 mm150 mm) packed with glass beads (0.2 mm). Detection was made
by a UV/Vis detector set to 485 nm and peak area was recorded by an integrator or a Perkin-Elmer data collection system. Pyruvic acid standards between 0 and 12mmoles mlÿ1 were used to quantify unknown samples. The DNPH solution
was kept at 58C in a refrigerator and there was no degradation for seven days.
2.2. Reproducibility test
Twenty identical juice samples were analyzed by both the Yoo and the automated methods. Coefficients of variation (CV) were calculated from peak areas or absorbency.
2.3. Background pyruvic acid and interfering compounds in onion juice
Onion bulbs (cv. `TG 1015Y' and `Texas Early White' from our storage and an unknown onion from a local store) were cut in halves longitudinally. One half was wrapped with plastic film and cooked completely in a microwave oven (5 min). The tissue was blended in a home mixer for 2 min and juice was collected through a paper filter. The other fresh half was blended and juice collected. The juice sample was filtered again by the microfilter and injected into the automated system.
Compounds in onion juice that may interfere with pyruvic acid analysis were also tested. Sugar mixture (sucrose, glucose, and fructose mixture of 16 mg mlÿ1
each) and quercetin (68mg mlÿ1
) were injected and peak areas recorded. The effect of protein compounds was tested by injecting microfiltered juice samples with or without adding trichloro acetic acid in juice (3%).
2.4. Comparison with other methods using spiked samples
1 ml of 0±16mmoles mlÿ1 pyruvic acid was added to 1 ml of identical onion
juice and the total pyruvic acid content was measured by the SW, Yoo, automated, and HPLC methods. In the SW method, 49-fold diluted juice was used, while the other three methods used undiluted juice. The HPLC method used an Alltech IOA 1000 organic acid column (3007.8 mm) and solvent was 0.01 N sulfuric acid
at a rate of 0.5 ml minÿ1. A 20ml sample was injected after microfiltration.
2.4.1. Five cultivar samples
Longday onions with red, yellow, and white colors (cultivars unknown) bought from a local store, `TG 1015Y', and `Texas Early White' (both from cold storage) were used. Five bulbs of onions were used from each cultivar as replications. Each onion bulb was cut into two pieces longitudinally and half was cooked as in Section 2.3. The fresh and cooked halves were blended with the same weight of water and juice extracted. The juice sample was filtered through a microfilter and analyzed by the Yoo, automated, and HPLC methods. Correlation between the Yoo method and the other two was calculated.
2.5. Application to screening of a breeding line
Forty onion bulbs of a breeding line were blended without adding water, and juice samples were collected. These samples were frozen at ÿ208C, defrosted,
microfiltered, and analyzed by both Yoo and the automated methods. Correlation between pyruvic acid concentrations by these two methods was calculated.
3. Results and discussion
The new automated system, without a true HPLC column, produced very smooth and symmetrical peaks of a chromatograph, resembling a true chromatogram (Fig. 2). After much trial and error, flow rates of 0.7 and 1.4 ml minÿ1
for the DNPH and NaOH, respectively, were found to be the optimum condition for an analysis. A batch of 96 samples could be loaded to the autosampler and about 500 samples could be analyzed per day. Because this method needs no tubes or pipetting for mixing chemicals, we could save time and eliminate human error in the analysis. On the other hand, microfilters must be used to clean the juice, adding to the cost of analysis.
3.1. Reproducibility
The CV of the automated method was 0.9%, as compared with 1.7% of the Yoo method. This result demonstrated that this new automated method is superior to a conventional method in consistency because of automation and elimination of human error.
3.2. Background pyruvic acid and interfering compounds
Background pyruvic acid content in `TG 1015Y', `Texas Early White', and a long day onion were 0.2, 0.3, and 0.3mmoles mlÿ1
content were 3.3, 4.1, and 3.3mmoles mlÿ1
, respectively. These results indicate that the automated system can accurately measure low levels of pyruvic acid without much interference in onion juice. These background contents were similar to our previous results using undiluted onion juice sample (Yoo et al., 1995).
A total sugar mixture of 48 mg mlÿ1
and quercetin solution (68mg mlÿ1 ) resulted in peak areas equivalent to about 0.1mmoles mlÿ1pyruvic acid. Removal
of protein substances in the onion juice reduced its pyruvic acid content by about 0.1mmoles mlÿ1, which was regarded as negligible. Therefore, this automated
system may over-estimate pyruvic acid content by about 0.3mmoles mlÿ1 in onions with a high sugar or quercetin concentration. This problem was thought to be avoided by blending onion tissue with one or two volumes of water and thereby reducing the concentration of interfering compounds. Microfiltration is always required in sample preparation to avoid clogging the tubing of an autosampler and a UV/Vis detector.
3.3. Comparison with other methods using spiked samples
In the test of adding a known amount of pyruvic acid into onion juice, all methods resulted in very linear increases (Fig. 3). The Yoo, automated, and HPLC methods estimated very similar concentrations of pyruvic acid in juice. The SW method requiring two-step dilution to obtain 49-fold dilution, however, estimated about 0.7±1.0mmoles mlÿ1 less than the other three methods. The
sticky nature of onion juice was thought to cause sample loss during pipetting, as compared to a non-sticky standard solution, resulting in reduced concentrations against the standard. Measuring pyruvic acid within a range of 8mmoles mlÿ1
seemed to be more accurate than at the higher levels. This can be easily accomplished by using 2- or 3-fold diluted onion juice. In this result, the automated method proved to be a reliable way to accurately estimate pyruvic acid content in onion juice.
3.4. Five cultivar samples
Onion samples with wide genetic backgrounds could be accurately measured by the automated method. The automated and HPLC methods estimated about 17% and 7% over the Yoo method, respectively (Fig. 4(A) and (B)). Because the HPLC method is believed to measure pyruvic acid specifically, as compared to the non-specific Yoo method reacting with any carbonyl compounds, the actual over-estimation was thought to be about 10%. Background pyruvic acid in the cooked sample again were less than 0.8mmoles mlÿ1, as similarly mentioned in
levels were less than 0.5mmoles mlÿ1
and was considered negligible in a breeding program.
3.5. Application to screening of a breeding line
In the actual screening of a breeding line, the automated system showed highly significant correlation with the Yoo method (Fig. 5). The automated method, using undiluted juice, estimated about 15% higher pyruvic acid than the Yoo method, probably due to combined interference of sugars, quercetin, or other Fig. 4. Comparison of pyruvic acid analysis by the automated and HPLC methods with the Yoo et al. (1995) method using juice extracted from paired fresh and cooked halves of five onion cultivars (25 bulbs).
compounds. Since the interference proportionally increased with pyruvic acid concentration, some by-products of the alliinase reaction was thought to be the compound. This deviation could not be eliminated even if diluted juice were used, as in Sections 3.3 and 3.4.
3.6. Comparison with a commercial autoanalyzer
Because actual analysis has not been performed by an autoanalyzer (Scientific Instruments, Westco, CT), a brief subjective comparison between the two systems was made. The new automated system looks simpler in structure and is more flexible to change flow rate, as compared with the autoanalyzer which has a complex tubing system and fixed flow rate determined by diameter of a tubing. The peak shape in our method resembled an ordinary HPLC peak with a smooth baseline, and an ordinary integrator or a data collection system can be used to record peak areas. The autoanalyzer, however, produced step- or wave-like chromatograms, and peak height data can be measured manually or with a special computer program. Over all, our automated system was thought to be equal or more versatile and efficient than a commercial autoanalyzer.
3.7. Costs
Two HPLC pumps for $4000; an autosampler for $7000; a column heater for $1500; a UV/Vis detector for $8000; an integrator for $1500; and tubing and other materials for $500. Total cost is about $22 500.
4. Conclusion
A new automated system using two HPLC pumps, an autosampler, and a detector was demonstrated to be a consistent and accurate way to measure pyruvic acid content in onion juice. The advantages of this system were the elimination of pipetting error and the speed of analysis. Using this system, we could run about 500 samples per day, compared to 250 samples by our current method employing two people. This system has proven to be effective in a mass screening of onion bulbs in a breeding program and could be used in a commercial analysis laboratory where fast, efficient, highly uniform results are required.
References
Schwimmer, S., Weston, W.K., 1961. Enzymatic development of pyruvic acid in onion as a measure of pungency. J. Agric. Food Chem. 9, 301±303.
Whitaker, J., 1976. Development of flavor, odor, and pungency in onion and garlic. Adv. Food Res. 22, 73±133.
