Free amino acids in African indigenous vegetables: Analysis with improved hydrophilic interaction ultra-high performance liquid chromatography tandem mass spectrometry and interactive machine learning

Highlights

• An improved HILIC-UHPLC-MS/MS method was developed for free amino acids (AAs).

• Removal of mineral buffer from mobile phase was key for high sensitivity.

• Addition of HCl in sample solvent was crucial for linear response of basic AAs.

• Free AAs were quantified in 544 African indigenous vegetable (AIV) samples.

• Machine learning methods were applied for AIVs prediction based on AA profile.

• Online open-source dashboard was developed to automate AIVs prediction.

Abstract

A hydrophilic interaction (HILIC) ultra-high performance liquid chromatography (UHPLC) with triple quadrupole tandem mass spectrometry (MS/MS) method was developed and validated for the quantification of 21 free amino acids (AAs). Compared to published reports, our method renders collectively improved sensitivity with lower limit of quantification (LLOQ) at 0.5~42.19 ng/mL with 0.3 μL injection volume (or equivalently 0.15~12.6 pg injected on column), robust linear range from LLOQ up to 3521~5720 ng/mL (or 1056 ~ 1716 pg on column) and a high throughput with total time of 6 min per sample, as well as easier experimental setup, less maintenance and higher adaptation flexibility. Ammonium formate in the mobile phase, though commonly used in HILIC, was found unnecessary in our experimental setup, and its removal from mobile phase was key for significant improvement in sensitivity (4~74 times higher than with 5 mM ammonium formate). Addition of 10 (or up to100 mM) hydrochloric acid (HCl) in the sample diluent was crucial to keep response linearity for basic amino acids of histidine, lysine and arginine. Different HCl concentration (10~100 mM) in sample diluent also excreted an effect on detection sensitivity, and it is of importance to keep the final prepared sample and calibrators in the same HCl level. Leucine and isoleucine were distinguished using different transitions. Validated at seven concentration levels, accuracy was bound within 75~125%, matrix effect generally within 90~110%, and precision error mostly below 2.5%. Using this newly developed method, the free amino acids were then quantified in a total of 544 African indigenous vegetables (AIVs) samples from African nightshades (AN), Ethiopian mustards (EM), amaranths (AM) and spider plants (SP), comprising a total of 8 identified species and 43 accessions, cultivated and harvested in USA, Kenya and Tanzania over several years, 2013~2018. The AN, EM, AM and SP were distinguished based on free AAs profile using machine learning methods (ML) including principle component analysis, discriminant analysis, naïve Bayes, elastic net-regularized logistic regression, random forest and support vector machine, with prediction accuracy achieved at ca. 83~97% on the test set (train/test ratio at 7/3). An interactive ML platform was constructed using R Shiny at https://boyuan.shinyapps.io/AIV_Classifier/ for modeling train-test simulation and category prediction of unknown AIV sample(s). This new method presents a robust and rapid approach to quantifying free amino acids in plants for use in evaluating plants, biofortification, botanical authentication, safety, adulteration and with applications to nutrition, health and food product development.

Introduction

African indigenous vegetables (AIVs) are an important food source in sub-Saharan Africa, and can contribute to enhancing food security and dietary diversity as well as create income generating opportunities for local people [1,2]. Many AIVs have been shown to be nutrient dense [2], [3], [4], [5], [6], and the total protein content typically ranges from ca. 20 ~ 50% or up to 70% of dry mass [7,8]. Despite literature abundance on total protein of AIVs, there is little information on the actual amino acid (AA) composition. In addition, knowledge of the content of free amino acids in the AIVs has also been limited. Free amino acids are important indicators of the vegetables’ biological response to the environment [9], provide dietary nutrition and play vital role in taste and flavor, and could be of interest for classification purposes [10,11].

Analysis methods for AA have been subjected to long-history of and continuous innovation and improvement. Of the numerous techniques developed, the first milestone was the 1950s-invention of ion-exchange chromatography with post-column derivatization using ninhydrin reagent with detection at 570 nm and 440 nm [12]. This technique was later developed into fully automated AA analyzers since 1960s [13], and remains as a frequently applied and classic method to date. This technique, however, suffers most from its hours-long elongated run time per sample and is being gradually replaced by many other methods [14,15]. One alternative technique is pre-column derivatization (PreCD) with reverse phase (RP) chromatography. The PreCD reagents, such as o-phthalaldehyde (OPA), fluorenylmethyl chloroformate (FMOCCl), phenyl isothiocyanate (PITC), dimethylaminonaphthalene-5-sulphonyl chloride (dansyl-Cl) [11,16,17], and 6-aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC) [18], [19], [20], etc., typically tag the polar AAs with a hydrophobic chromophore, and provides improved retention and separation on the RP column and feasible ultraviolet-visible light and/or fluorescence detection. In some applications using mass spectrometry (MS) analysis, the labeling reagents including their isotopic counterparts, such as iTRAQ™ (isobaric Tags for Relative and Absolute Quantitation) [21,22] and its later enhanced version aTRAQ™, etc. [23], [24], [25], react with amino acids such that the derivatives provide a characteristic fragmentation in MS to generate the product ion corresponding to the labeling reagent. This technique is a cost-efficient solution to provide isotopically-labeled internal standards for each AAs while using a single reagent. All such derivatization methods while providing many advantages also present various difficulties, such as the instability and/or inconsistency of derived products, laborious sample preparation, analysis artifacts and/or system contamination [18].

Different from the former two types of technique based on AAs derivatization, a third technique involves direct analysis of underivatized AAs on the RP column using ion-pairing reagents, such as various perfluorinated carboxylic acids followed mostly with MS detection. While this derivatization-free technique proves to be a more convenient and also powerful tool, the use of ion-paring reagents in the mobile phase, however, could induce noisy background, system peaks and contamination, ion suppression, and long equilibration time especially when the buffer is not balanced [14,15,18,26].

Another technique for underivatized AA analysis that is gaining increasing popularity in recent years is the hydrophilic interaction (HILIC) chromatography with MS detection. HILIC applies polar stationary phase, such as bare silica or polar bonded phase, and high percent of organic mobile phase for separation of polar and charged compounds. The separation mechanism involves analytes’ partition between the bulky organic phase and the thin aqueous layer immobilized along the surface of the stationary phase, and many other effects such as ionic interaction and dipole-dipole interaction [27,28]. While this technique has been extensively used in amino acid analysis [29], [30], [31], [32], [33], [34], [35], [36], and presents further analytical progress compared to many older more classical techniques, it still faces challenges including undesirable sensitivity, compromised chromatographic performance, limited linear range and long run-time, etc. A brief review of these reported methods is presented in Table 1.

To overcome the current analytical constraints reported in the literature, we developed an improved HILIC-MS/MS method for AAs analysis with significantly enhanced sensitivity, improved linear range and higher throughput. Using this newly developed method, the free AAs in a total of 544 AIVs samples were determined. Based on acquired free AAs profile, the AIVs categories were successfully predicted using machine learning methods, and an R-Shiny based online interactive application was constructed for interactive modeling simulation and classification prediction of unknown samples.