Volatile Organic Compound Based Markers for the Aroma Trait of Rice Grain

A study was conducted to determine the volatile organic compounds (VOCs) associated with rice grain aroma in 37 commonly grown lines within Uganda, as well as elites. The aim of the study was to identify potential volatile biochemical markers, if any, for the rice grain aroma trait. Certified rice seeds were obtained from the Uganda National Crops Resources Research Institute germplasm collection. The seeds were sown into experimental plots, under field conditions and the mature paddy harvested. Polished rice grains were heated to 80 C and the liberated VOCs subjected to untargeted metabolite analysis using gas chromatography-time-of-flight mass spectrometry. In total, nine functional groups were present; hydrocarbons, alcohols, ketones, aldehydes, N-containing compounds, S-containing compounds, esters, oxygen heterocycles and carboxylic acids. More specifically, 148 VOCs were identified across the 37 rice lines, of which 48 (32.4%) including 2-acetyl-1-pyrroline (2-AP) appeared to elucidate the difference between non-aromatic and aromatic rice. Furthermore, 41 (27.7%) VOCs were found to be significantly correlated with 2-AP abundance, the principle rice aroma compound. Amongst the 41 VOCs, only ten compounds were found to contribute highly towards variation in 2-AP abundance, indicative of their possible modulation roles in regard to rice aroma. Within the ten influential volatiles, three aroma active compounds; toluene, 1-hexanol, 2-ethyl and heptane, 2,2,4,6,6-pentamethylwere established as the most reliable biochemical surrogates to the rice aroma trait. Thus, the aforementioned compounds may be used in rice breeding programme for enhancing development of the grain aroma trait.


Introduction
Rice (Oryza sativa L.) is among the most important staple food crops for approximately two billion persons worldwide and a substantial source of income for several rural households (Sharma et al., 2018). Globally, several rice species exist including O. sativa thought to have originated in Asia (Kovach et al., 2009) and Oryza glaberrima of West African origin (Linares, 2002). The origin of a rice species is known to influence the physio-chemical attributes of the grain such as aroma and taste, primarily due to genetic inheritance (Rai et al., 2015). Considering aroma, the responsible gene betaine aldehyde dehydrogenase 2 (badh2) is believed to have different alleles (Bindusree et al., 2017), with specific allele configurations dependent upon species origin (Pachauri et al., 2010). Consequently, substantial variations in rice grain aroma has been observed in relation to differences in genetic backgrounds (Rai et al., 2015). Source: Kanaabi et al. (2018).

Preparation of the Polished Rice Grains
Paddy rice lines were processed according to the method by He et al. (2018) with slight modifications. Rice lines (300 g each) were submerged in 1 L distilled water and the floated kernels immediately discarded. The settled rice kernels were briefly washed and then sun dried for 48 h, final moisture content approximately 14%. The dry paddy rice were milled and polished using a bench top milling machine (Satake, Tokyo, Japan). Fifty (50) grams of polished rice grains were vacuum packed into air-tight polyethylene bags and delivered to the School of Agriculture and Food Sciences laboratory, University of Queensland, Australia for biochemical analyses.

Identification and Quantification of Volatile Organic Compounds
Polished rice grains were ground using a cryogrinder (Qiagen, Hilden, Germany) to particle size < 25 µm, and the flours weighed (1 g) into autosampler tubes. The tightly sealed tubes, in triplicates, were stored at -80 °C until required. Pre-analysis, the frozen flours were left overnight at room temperature to equilibrate (Daygon et al., 2017). Thawed flours were then randomised and analysed in batches of 50. Blank samples were run before analysis of the rice flours to equilibrate the machine and quality control (QC) standards were placed at every 10 th queue position. Rice flours were assayed following the method by Daygon et al. (2017). In summary, rice flours were heated to 80 °C with agitation for 10 min on a CombiPal Autosampler (Agilent, CA, USA) to release the VOCs. Liberated VOCs within the tube headspace (1.5 ml) were collected using 2.5 ml headspace syringes at 80 °C and injected in splitless mode (Pegasus 4D GC×GC-TOF-MS Leco; St. Joseph, MI, USA). Temperature of the gas chromatographer (GC) inlet and transfer lines were maintained at 250 °C. Separation was performed first on a primary column (Agilent DB-624UI midpolar, 30 m × 250 µm × 1.4 µm; Agilent, CA, USA) and then on a secondary column, Stabilwax (polar, 0.9 m × 250 µm × 0.50 µm; Restek, Bellefon, USA). The primary column was initially set to 45 °C for 1 min and then ramped at a rate of 10 °C/min to 235 °C. The secondary column and the modulator were set at 15 °C and 25 °C higher than the primary column respectively, during the entire run. The modulation period was set at 2.5 s, with 0.4 s hot pulse time and 0.85 s cool time between stages. The carrier gas (helium) was maintained at a constant flow rate of 1 ml/min. Data were acquired using a TOF-MS after a 200 s delay with an acquisition rate of 200 spectra/s. The MS scanned analytes within the mass range of 35 to 500 m/z and the ion source was held at 240 °C.
Data pre-processing, alignment and noise correction were done using ChromaTof v4.50. Signal to noise ratio was set at 25. The absence of instrument drift and batch effects was verified using the QC standards and technical replicates. Identification of VOCs was achieved through comparison of retention time and electron ionization (EI) fragmentation patterns of rice flours to an in-house mass spectral library created by running authentic analytical standards (Daygon et al., 2017). The relative amounts of VOCs were calculated by measuring the area under the curve of the VOC peak.

Data Analysis
Overall mean abundance of VOCs, representative of previously established non-aromatic and aromatic rice lines (Ocan et al., 2019), were subjected to the Student's T-test at 5% significance level using Genstat software 18 th Ed.
(VSN International, Hemel Hempstead, UK). Correlation analysis between and within VOCs significantly (p < 0.05) associated to 2-AP abundance was conducted. Consequentially, VOCs significantly correlated with 2-AP abundance were subjected to simple regression analysis in a bid to determine the key volatiles, if any, as potential biochemical markers for the rice aroma trait.

Overall Mean Abundance of the Key VOCs within Non-aromatic and Aromatic Rice Lines
A comparison between the overall mean abundance of the 148 VOCs across non-aromatic and aromatic lines revealed 48 VOCs that appeared to differentiate the 37 rice lines into two aroma categories (Table 3).   Considering the 48 VOCs that revealed significantly (p < 0.05) different quantities between non-aromatic and aromatic rice categories; hydrocarbons constituted the majority at 50.0% (24 VOCs), followed by alcohols 12.5% (6 VOCs), ketones 10.4% (5 VOCs), aldehydes 8.3% (4 VOCs), O-heterocycles 8.3% (4 VOCs), Ncontaining compounds 4.2% (2 VOCs), esters 4.2% (2 VOCs) and S-containing compounds 2.1% (1 VOC) (Table 3). With the exception of the carboxylic acid group, the total number of functional groups present within both the non-aromatic and aromatic rice categories remained unchanged.
However, a substantial reduction in the number of VOCs established as possible discriminators between the rice aroma categories (Table 3) and the total number of VOCs that constitute rice aroma (Table 2) was observed.

Relationships, between and within, 2-Acetyl-1-pyrroline and Associated Volatile Organic Compounds
A correlation analysis of the 148 VOCs (Table 2) revealed 12 (8.1%) compounds to be highly associated to 2-AP, while 13 (8.7%) compounds were moderately associated and 16 (10.8%) compounds weakly associated (Tables  4a, 4b and 4c). Among the highly associated compounds to 2-AP, five functional groups were represented (Table  4a), while for the moderately associated compounds three functional groups were present (Table 4b). On the other hand, several VOCs were determined as weakly associated to 2-AP (Table 4c), thus, of limited importance in the present study.
Furthermore, with the exception of methyl acetoacetate and undecane, 2,6-dimethyl-, all compounds found to be highly associated to 2-AP were also highly or moderately associated amongst themselves (Table 4a). For moderately associated compounds to 2-AP, it was noted that all compounds were either highly or moderately associated amongst each other with the exception of: cyclopentane, methyl; cyclohexane; 1-penten-3-one, 2-methyl and hexane, 2,3-dimethyl (Table 4b).

Identification and Classification of the Volatile Organic Compounds
Earlier studies have revealed wide variations in the number and chemical nature of VOCs associated with rice grain aroma. For example; 47 volatiles in Lee et al. (2019), 50 volatiles in Shanthine et al. (2019) and 140 compounds in Sansenya et al. (2018). Thus, the number of 148 VOCs in the present study is more closely aligned to 140 VOCs as reported by Sansenya et al. (2018). The wide variation (47 148 VOCs) in-between the numbers of volatile organic compounds reported under the different studies appears most likely due to differences in the technologies employed, nature and number of cultivars utilized. Generally, studies involving more advance techniques such as GC×GC-TOF-MS and numerically more rice lines, appeared to report higher numbers of VOCs (Lee et al., 2019;Fatemi et al., 2014;Shanthine et al., 2019;Sansenya et al., 2018). In regards to functional groups, earlier studies revealed slight variations in the nature and number constituting rice aroma.
In the present study, nine functional groups were established (Table 2). This number is in agreement with earlier works by Fatemi et al. (2014) and Lee et al. (2019). These authors, reported the rice aroma metabolites to comprise; ketones, aldehydes, alcohols, carboxylic acids, esters, hydrocarbons, oxygen heterocycles, N-containing compounds, benzene and benzene derivatives. Considering the nature of the functional groups involved in rice aroma, the three studies were similar with the exception of S-containing compounds.

Volatile Organic Compounds Able to Differentiate Non-aromatic and Aromatic Rice Categories
The presence of several VOCs capable of differentiating between non-aromatic and aromatic rice lines is probably due to the genetic diversity of rice lines within Uganda (Table 3). Similar to earlier report by Hinge et al. (2019), most volatiles in both the rice aroma categories were similar, but with different propositions. Therefore, for the 15 VOCs with aroma description, aromatic rice lines on average had more abundance of the compounds. Intriguingly, these included three volatiles with undesirable aroma tints namely; tetradecane (gasoline-like), 1-pentanol (plastic, fusel oil-like) and heptanal (rancid). Among these compounds, the contribution of tetradecane is most likely negligible due to the high odour threshold limitation of most hydrocarbons (Hu et al., 2020). In contrast to Setyaningsih et al. (2019) who found; pentanal, hexanal, 2-pentyl-furan, 2,4-nonadienal, pyridine, 1-octen-3-ol and (E)-2-octenal as volatile markers capable of differentiating between non-aromatic and aromatic rice, the present study found no similarity. Given the similarities in methodology, the absence of commonality in the results was most likely due to genetic differences in the rice lines. This highlights the fact that identified volatile biochemical markers may only be applicable in differentiating between specific rice lines and within similar environs. The present study also revealed that not all key volatile biochemical compounds possess aroma descriptions. This could possibly be construed to imply that not all volatile biochemical markers are aroma active compounds. This would be in contrast to the study by Setyaningsih et al. (2019) in which all the seven volatile biochemical markers were aroma active compounds. Hence, the present study deemed it necessary to further refine the list of possible volatile biochemical markers by subjecting the VOCs to correlation and regression analyses. Hinge et al. (2016) reported that 2-AP abundance was not correlated to other aroma active compounds in aromatic rice. The authors postulated that the expression pattern of 2-AP was probably unique and specifically delinked from other volatiles of aroma importance. However, recently Daygon et al. (2017) documented a strong correlation between 2-AP abundance and amine heterocycles, namely; 6-methyl, 5-oxo-2,3,4,5-tetrahydropyridine (6M5OTP), 2-acetylpyrrole, pyrrole and 1-pyrroline. In the present study, a total of 25 VOCs were found to be highly or moderately associated to 2-AP abundance (Tables 4a and 4b). These 25 compounds comprised of six functional groups; hydrocarbons, alcohols, ketones, N-containing compounds, esters and O-heterocycles. In contrast, the functional groups for the present study did not contain amine heterocycles as earlier reported by Daygon et al. (2017). This is probably due to genetic differences in the constitution of the rice lines studied. Specifically, the present study employed genotypes developed from O. barthi, O. longistaminata, O.glaberrima and O. sativa backgrounds (Kitara et al., 2015;Lamo et al., 2017;Kanaabi et al., 2018) in contrast to the O. sativa intra-specific crosses used by Daygon et al. (2017).

Relationships, between and within, 2-Acetyl-1-Pyrroline and Associated Volatile Organic Compounds
In the present study, hydrocarbons and alcohols constituted the highest percentage of VOCs correlated to 2-AP, cumulatively 84%. In contrast, ketones, N-containing compounds, esters and O-heterocycles constituted only 16%. This is consistent with the importance of alcohols and hydrocarbons in rice aroma as earlier reported by Stefano and Agronomia (2011) and Lee et al. (2019). The absence of carboxylic acids, S-containing compounds and aldehydes (Table 4a and 4b) may be construed to imply that these compounds are of limited contribution towards 2-AP synthesis. However, this would be in contradiction to the finding by earlier authors concerning the importance of S-containing compounds and aldehydes in rice aroma (Sansenya et al., 2018). Thus, it is important to note that the absence of correlation between a given volatile compound and 2-AP may not indicate limitations in its sensory contribution towards rice aroma. Interestingly, with the exception of; methyl acetoacetate, heptane, 2,2,4,6,6-pentamethyl-and ethanol, 2-butoxy, all compounds highly associated to 2-AP were equally highly associated to each other (Table 4a). This implies that methyl acetoacetate, heptane, 2,2,4,6,6-pentamethyl-and ethanol, 2-butoxy may be important through associated pathways such as those influencing antagonistic effects beneficial to the perception of rice grain aroma (Champagne et al., 2008;Chambers et al., 2013).
In conclusion, volatile organic compounds associated with rice grain aroma are quantitatively enormous and of diverse functional group classification. However, it is apparent that not all VOCs associated to rice aroma aid in the segregation of the rice grains into non-aromatic and aromatic types as perceived by consumers. Further, amongst the VOCs associated with the aroma trait in rice grains, not all compounds contribute equally. A few VOCs are responsible for both the discrimination of rice into non-aromatic and aromatic groups in addition to aiding desired sensory perception. These three compounds, namely: toluene, 1-hexanol, 2-ethyl and heptane, 2,2,4,6,6-pentamethyl-are therefore potential volatile biomarkers for the rice aroma trait. Thus, the aforementioned compounds may be used in rice breeding programmes for enhancing development of the aroma trait.