Intensification of the global meat industry has stimulated innovation in the animal feed sector. Advances in feed technology strive towards intensification through a reduction in feed conversion ratio (FCR), greater energy and nutrient utilization, improved animal welfare and environmental sustainability, reduction in endogenous grain anti-nutrients, and optimization of costs (Makkar and Ankers, 2014). Using poultry production as a specific example, the yield of chicken meat (with Australia as a model market) has exponentially increased from the 1970s until today where it has begun to level off (Supplementary Figure S1). This increase in production is mirrored by a similar decrease and leveling off of FCR (increased efficiency). This gain of efficiency is due to several global innovations, including directed poultry and feed grain breeding, implementation of new feed additives, and use of higher quality grains and supplements in dietary formulations (Mottet and Tempio, 2017). Supplementation of monogastric feed with exogenous enzymes has become routine to support measures of performance, as well as to mitigate the effects of anti-nutrients, most commonly phytate, non-digestible starches and proteins, and polyphenolic compounds (Cowieson et al., 2006). Polyphenols are well-established anti-nutrients and antifeedants, particularly to monogastrics, and routinely cause reduced feed intake and weight gain, increased FCR (reduced efficiency), and enzyme inhibition (Bravo, 1998; Cadogan and Finn, 2010; Pasquali et al., 2016; Alu’datt et al., 2018).

While phenolic and polyphenolic compounds are found in all feed grains, sorghum, Sorghum bicolor (L.), is well-established as having markedly high levels of these secondary metabolites, including condensed tannins (Glennie et al., 1981; Awika and Rooney, 2004). Sorghum is the fifth-most important cereal crop grown in the world and has many diverse applications, including alcoholic beverages, biofuel, human food products, and animal feed. Approximately, 59 million tons of sorghum was produced in 2018 with half of the production in Africa and a third in the Americas (Food and Agricultural Organization of the United Nations, 2020). As sorghum has high levels of phenolic and polyphenolic compounds, it has developed a split nature as these metabolites have proven positive effects in human diets. The polyphenols found in sorghum are well-established antioxidants that can reduce oxidative stress and the diseases that arise from imbalances in reactive oxygen species (Awika and Rooney, 2004; Stefoska-Needham et al., 2015).

The higher concentrations of polyphenols, up to 10% of the grain’s mass, have played a key role in sorghum being stigmatized as having lower nutritional quality when incorporated into monogastric animal feed (Jambunathan and Mertz, 1973; Armstrong et al., 1974; Bravo, 1998; Selle et al., 2017). Previous studies on sorghum have identified a diverse range of phenolics, from small ferulic and caffeic acids to large condensed tannins with high degrees of polymerization (Stafford, 1965; Gupta and Haslam, 1978; Kang et al., 2016). High-tannin sorghum varieties are not commonly used in monogastric animal feed as deleterious nutritional effects have been observed in animals fed these particular grains (Nyachoti et al., 1996). These negative effects have encouraged sorghum breeders to preferentially select low-tannin varieties for use in monogastric animal feed. Sorghum grains low in tannin content, such as most red and white varieties in use today, have also been reported as having higher levels of digestible protein (Youssef, 1998). White sorghum grain, most commonly the Liberty variety in Australia, has been found to better support weight gain, FCR, and growth performance in pigs and chickens than its red colored counterparts. This may be due to the absence of large polyphenols, such as condensed tannins (Cadogan and Finn, 2010; Liu et al., 2010).

Currently, there is discussion about whether modern varieties, important to the animal feed industry, contain relevant/detectable levels of condensed tannins (Perez-Maldonado and Rodrigues, 2009; Liu et al., 2015). This debate seeks to move the conversation from condensed tannins to smaller phenolic and polyphenolic compounds which may contribute subtle differences in varietal performance, even in sorghum grains designated “tannin-free.” In their study of six “tannin-free” sorghum diets, Truong et al. (2016) found no difference in broiler chicken performance with regard to FCR and weight gain but did find differences in nutrient utilization between white and red grains. Negative correlations were found between phenolic acids, flavonols, kafirin protein (sorghum’s major storage protein), and measures of digestibility. Even the beneficial impacts of certain feed additive enzymes have been reported to be muted when formulated into sorghum diets. This has been observed primarily with phytase, with regard to the enzyme’s extra-phosphoric effects on protein and amino acid digestibility, (Liu et al., 2014; Truong et al., 2014; Selle et al., 2017). As the grains used in these studies were “tannin-free,” compounds other than traditional condensed tannins may have caused the anti-nutritional effects observed.

The majority of sorghum phenolic analyses have used liquid chromatography-mass spectrometry (LC-MS) with identifications achieved through use of standards, retention time comparison, and MSⁿ fragmentation (Kang et al., 2016; Rao et al., 2018; Tugizimana et al., 2019; Jiang et al., 2020; Zhou et al., 2020). The use and comparison of less intensive methodological approaches, including direct ionization and infrared spectroscopy, has been little studied in sorghum, especially with regard to characterizing metabolic variation between grain varieties important to the animal feed industry. Currently, there exists no comparative framework for the assessment of orthogonal methods of analysis for polyphenolic extracts, particularly crude extracts, from feed-relevant sorghum grains. In this paper, we present an alternative analytical framework for characterizing phenolic anti-nutrients in crude polyphenol extracts from three Australian sorghum varieties (MR-Buster, Cracka, and Liberty). Using a series of analytical techniques from simple spectroscopy to more complicated mass spectrometric methods, untargeted and targeted metabolomics methodologies were applied to the data to determine both bulk and subtle differences in metabolite profiles.

While modern monogastric animal feed has been formulated for optimum nutrient utilization and digestive efficiency, performance gains still remain, especially in feeds composed of sorghum grain. With poultry production as a model, increases in efficiency with sorghum as a feed grain, as measured by FCR, have begun to level out (Supplementary Figure S1). While efforts to reduce anti-nutrient content in sorghum, most notably tannin and polyphenol concentrations, have been successful, gaps in efficiency and efficacy of feed additives remain, possibly due to unintended consequences in feed quality sorghum breeding. An understudied alternative analytical approach was thus used to identify anti-nutrients that may be causing varied performance in sorghum feed and to determine the suitability of different analytical techniques for assessing metabolic variation among sorghum grain extracts.

In this study, all three sorghum extracts, especially the Liberty variety, were found to have high ion counts for mass bins putatively identified as fatty acids, including oleic acid, linoleic acid, vernolic acid, ricinoleic acid, and trihydroxyoctadecenoic acid, which could result in nutritional variation in monogastric diets composed primarily of sorghum. With oleic acid as a starting point, vernolic acid is formed through an epoxidation reaction, ricinoleic acid through hydroxylation, linoleic acid through desaturation, and trihydroxyoctadecenoic acid through the hydroxylation of linoleic acid (Mazur et al., 1999; Cao and Zhang, 2013). The identifications made through mass spectrometry echo the strong, sharp FT-IR peaks between 3,000 and 2,900 cm⁻¹ which correlate with C – H bonding found extensively in fatty acids (Shapaval et al., 2014). MR-Buster has been previously found to contain linoleic and oleic acids as the most dominant fatty acids, making up 80% of unsaturated fatty acid content (Mehmood et al., 2008). White varieties are similarly dominated by oleic and linoleic acids but at slightly higher proportions (Afify et al., 2012). Broiler chickens fed diets high in oleic acid were found to have a higher FCR, as well as reduced muscle and carcass weights (Toomer et al., 2003). Similar long chain fatty acids have also been shown to inhibit enzyme activity which could result in muted responses of exogenous feed enzymes (Kido et al., 1984). The high levels of fatty acids detected in the current study indicate the potential of implementing an exogenous lipase or emulsifier, as high levels of these compounds could be detrimental to growth and performance parameters.

Sorghum breeding efforts may have triggered metabolic alterations by favoring fatty acid synthesis over polyphenols, particularly condensed tannins. This hypothesis is supported by current work on sorghum grain and its management. Xie et al. (2019) studied sorghum varieties and preference by birds for feeding. They found that varieties avoided by birds had higher anthocyanin and tannin precursors (flavan-3-ols) than those that they preferred to eat. This correlated with the presence or absence of the Tannin1 gene, previously found in sorghum to be involved in the regulation of polyphenols and tannins (Wu et al., 2012). In addition to determining the difference in polyphenols, the bird-preferred sorghum was found to have increased volatiles associated with fatty acids, as well as higher concentrations of fatty acids, including linolenic acid. Xie et al. (2019) concluded that the modulation of the Tannin1 gene affects SbGL2 which is involved in transcription of fatty acids.

In addition to the fatty acids and common polyphenols detected, the red sorghum varieties were found to have peaks with high ion masses and ion count, m/z 689, 851, 1107, and 1269, not clearly identifiable as traditional sorghum tannins. Similar peak masses have been previously identified in sorghum extracts as either pyrano-compounds or glucosylated heteropolyflavans. In sorghum leaf sheath, Khalil et al. (2010) describes the structural determination of a novel pyrano-3-deoxyanthocyanidin, pyrano-apigenindin [m/z 371.091 (+)]. Red and black sorghums were found to have unique flavanone structures, including pyrano-naringenin-catechin (m/z 689), pyrano-naringenin-catechin-glucoside (m/z 851), pyrano-eriodictyol-catechin-glucoside (m/z 867), and pyrano-naringenin-pyrano-eriodictyol-catechin (m/z 1107) (Yang, 2013; Rao et al., 2018). Glucosylated heteropolyflavans, described by Gujer et al. (1986) and Krueger et al. (2003), are composed of unique polymerizations of eriodictyol/naringenin and luteolinidin/apigeninidin with varying degrees of glycosylation. The masses of these compounds were predicted with the equation 288 + 272a + 256b + 162c + cation with 288 representing the mass of an eriodictyol base unit, 272 referring to a proluteolinidin unit, 256 to a proapigeninidin unit, 162 to additional sugar units, and the letters referring to possible degrees of polymerization. The MS² spectra support these identifications as the primary fragment masses detected correspond to losses of 256, 272, and 162. Purification of these unknowns is needed along with structural evidence that could be gained using NMR techniques.

In the current study, spectra obtained from ESI (+, −) were clearer than those produced using MALDI (+) with a greater number of clearly identifiable peaks, most likely due to the lack of need for a matrix compound with ESI. ESI (+, −) analyses were also successful in detecting routinely identified polyphenol compounds in sorghum, including apigenin, naringenin, and phenylpropanoid glycerides (Kang et al., 2016). Sorghum grain extracts have been sparingly analyzed using MALDI (Krueger et al., 2003; Qi et al., 2018; Jiang et al., 2020; Reeves et al., 2020). Unsupervised analysis of the ESI and MALDI spectra allowed for the clear separation between red and white varieties and in some cases among all three extracts. Identifications allowed for pathways of interest to be highlighted, as well as comparisons of percent relative abundances of selected masses. These metabolomic methods are most likely critical when first releasing a new variety of grain to the market. Recently, Zhou et al. (2020) compared three sorghum varieties (black, red, and white) to study metabolic variation based on color. PCA revealed a separation of grains based on grain color, and compound identification found that darker grains contained more flavonoids than lighter colored grains, similar to what was determined in the current work. Typically, the color of the grain gives some indication of the polyphenols present with darker and highly colored grains containing higher concentrations and often larger, more complex polyphenols (Rhodes et al., 2014).

FT-IR has been previously shown to be successful in both distinguishing bulk differences among general metabolite profiles (Johnson et al., 2003) as well as nuancing more subtle variations, e.g., tannin extracts separated based on tannin chemistry (Grasel et al., 2016a). In the current study, FT-IR analysis revealed there were structural similarities, with regard to polyphenols and tannins, among the sorghum polyphenol extracts. The subtle differences in sorghum extract peak structures and maxima suggest the presence of competing compounds indicative of a complex plant extract. As reviewed by Ricci et al. (2015), the spectra matched the general profile of extracts containing phenolic, polyphenolic, and tannin compounds, including characteristic O – H hydroxyl groups, aromatic C – H bonds, aromatic C = C bonds, and C – O groups. A similar approach in evaluating the presence/absence of specific metabolite structures was taken by Cameron et al. (2006) in their investigation into alteration to lignin and suberin content in grasses, legumes, and forbs subject to attack by a root hemiparasitic plant.

Comparable values for the O – H maxima have previously been reported in sorghum flour (Manuhara et al., 2017). The strong bands around 3,000 cm⁻¹, indicating aliphatic C – H structures, have been detected in similar extract types, including sorghum, and possibly indicate the presence of sugars and/or fatty acids (She et al., 2010; Shapaval et al., 2014; Manuhara et al., 2017). In their natural state, polyphenols are most likely to be conjugated to sugars (Bravo, 1998). Polyphenol and tannin extracts typically contain peaks corresponding to C = O groups (1,700 s cm⁻¹), especially those containing hydrolyzable tannins (Grasel et al., 2016b; Reeves et al., 2020). This region can also indicate the presence of amide functional groups, most commonly found in proteins. Duodu et al. (2001) studied protein structure in highly digestible sorghum and maize mutants and identified bands between 1,670 and 1,620 cm⁻¹ as amide I and from 1,550 to 1,500 cm⁻¹ as amide II. The subtle differences between red and white sorghum grains in these regions may indicate important differences in protein content and structure with possible nutritional implications. Selle et al. (2020) studied amino acids and kafirin protein in several sorghum varieties, including a Buster variety and Liberty. With regard to crude protein and kafirin content, Liberty had 80.9 and 41.4 g/kg, while Buster reported 99.2 and 44.6 g/kg, respectively (Selle et al., 2020). The higher proportion of kafirin protein found in Liberty may be causing the spectral differences observed in the current study.

The region from 1,630 to 1,400 cm⁻¹ is strongly diagnostic for the presence of polyphenols as it indicates aromatic C = C bonding found in the aromatic rings of phenolic compounds. The peaks identified between 1,400 and 1,000 cm⁻¹ are also characteristic of C – OH as well as C – O – C bonding (Ricci et al., 2015). The differences in the sorghum extract spectra in these regions indicate that the white variety (Liberty) likely has reduced polyphenol content compared to the red varieties (MR-Buster, Cracka), a common finding in sorghum polyphenol studies, which can correlate with nutritional variations observed in feeding (Truong et al., 2016). FT-IR is a useful, simple method of analysis that could be used in on-line agricultural settings, as opposed to full scale laboratories. Although this study analyzed more laboratory intensive freeze-dried extracts, FT-IR can easily be applied to simple liquid extracts taken on-site.

This study of an alternate analytical framework for polyphenol characterization highlighted the need for complementary methods to fully understand the complexity of sorghum polyphenol extracts. FT-IR spectroscopy provided general chemical profiles which highlighted functional groups and classes of compounds specific to polyphenol and tannin structural chemistry. Multivariate analysis of the FT-IR spectra demonstrated that the technique was robust enough to separate different extract types and to explain greater variance in the data than any MS method. Both ESI analyses produced similar plots to that of FT-IR, albeit with slightly better grouping of the sorghum extracts. ESI also provided a clearer metabolite profile than MALDI. These results indicated that, with regard to untargeted analysis, FT-IR and ESI provide essentially the same end-product allowing for similar conclusions to be drawn on bulk differences in the spectra. Based on these results, compatibility, and pricing, FT-IR may be the most effective tool for determining the applicability of certain grains to feed formulations, particularly with regard to polyphenol content. This application could be especially important in varietal selection for grain breeding and feed applications. Markers for chosen nutritional parameters, such as protein structure and anti-nutrient content, could be selected for and used as a screening tool prior to more intensive analytical methodologies should they be needed. However, mass spectrometric studies of metabolites present in these grains should be used to guide the interpretation of FT-IR spectra in the field to further highlight subtle differences in the grains that may result in monogastric feed performance variation.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

All authors contributed to the conception and design of the study. HH collected the data. HH, RF, and DC analyzed the data. HH wrote the manuscript. All authors contributed to manuscript revision, reading, and approval of the submitted version.