Comprehensive overview of the quality of plant‐ And animal‐sourced proteins based on the digestible indispensable amino acid score

Abstract Indispensable amino acid (IAA) composition and standardized ileal digestibility (SID) of five animal‐ and 12 plant‐based proteins were used to calculate their respective Digestible Indispensable Amino Score (DIAAS) according to the three age categories defined by the Food and Agriculture Organization (FAO). Mean IAA content and mean SID obtained from each protein dataset were subsequently used to simulate optimal nutritional quality of protein mixtures. Datasets revealed considerable variation in DIAAS within the same protein source and among different protein sources. Among the selected protein sources, and based on the 0.5‐ to 3‐year‐old reference pattern, pork meat, casein, egg, and potato proteins are classified as excellent quality proteins with an average DIAAS above 100. Whey and soy proteins are classified as high‐quality protein with an average DIAAS ≥75. Gelatin, rapeseed, lupin, canola, corn, hemp, fava bean, oat, pea, and rice proteins are classified in the no quality claim category (DIAAS <75). Potato, soy, and pea proteins can complement a broad range of plant proteins, leading to higher DIAAS when supplied in the form of protein mixtures and at specific ratios. Such complementarity highlights the potential to achieve an optimal nutritional efficiency with plant proteins alone.


| INTRODUC TI ON
New protein sources have emerged in recent years to support the transition toward more sustainable food production dedicated to human nutrition. Animal-sourced proteins, for example, from milk and meat, have shown to have a substantial impact on greenhouse gases and are known to contribute to depletion of natural sources (Aiking, 2014;Godfray et al., 2018). These proteins are included to a large extent in Western diets, and their consumption is gradually increasing in developing countries (Godfray et al., 2018). While it has been recommended to shift global protein consumption toward plant-based proteins (Boland et al., 2013;Pyett, Vet, Trindade, Zanten, & Fresco, 2019), attention should be paid to the nutritional quality of new and alternative protein sources.
Dietary protein quality is primarily characterized by their indispensable amino acid (IAA) content. IAAs cannot be synthetized by the human body and must be obtained from the diet. The current standard to evaluate the nutritional quality of proteins is the Protein Digestibility Corrected Amino Acid Score (PDCAAS).
Due to several drawbacks of this method (Sarwar, 1997), the Food and Agriculture Organization (FAO) recommends another procedure called Digestible Indispensable Amino Acid Score (DIAAS) (FAO, 2013). DIAAS addresses the limitations of the PDCAAS method by considering the ileal digestibility of individual amino acids (AAs), with the growing pig as preferred model over the rat, and by avoiding truncation of the score obtained. Furthermore, reactive digestible lysine rather than total digestible lysine should be considered to determine the DIAAS of processed and cooked foods. The FAO DIAAS report also recommends classification of proteins using quality categories based on the DIAAS value: <75 (no quality claim); 75-99 (high-quality protein); and ≥100 (excellent quality protein). The FAO has updated the age-related AA reference scoring patterns and recommends that the general population be categorized according to three distinct age-related reference patterns: 0-6 months (infant), 0.5-3 years (children), and > 3 years (rest of the population).
Recent studies compared various protein sources based on their AA composition (Gorissen et al., 2018;Sá, Moreno, & Carciofi, 2020;van Vliet, Burd, & van Loon, 2015). However, these data do not inform to which extent a specific protein can be digested to meet the human body requirements. The most extensive studies to date provide DIAAS values of eight protein sources (Cervantes-Pahm, Mathai, Liu, & Stein, 2017). To broaden the comparison of DIAAS to more protein sources, IAA profiles and IAA standardized ileal digestibility (SID) data were used to calculate the DIAAS values of 17 protein sources according to the three reference scoring patterns.
Standardized ileal digestibility is preferred over apparent ileal digestibility (AID) since the latter does not correct for the AA endogenous loss inherent to the body function (Stein, Sève, Fuller, Moughan, & de Lange, 2007). Endogenous AAs are secreted into the small intestine, and an estimated 20%-35% is not reabsorbed by the body before reaching the distal ileum (de Lange, Sauer, Mosenthin, & Souffrant, 1989;Souffrant et al., 1993). Applying AID as digestibility coefficient therefore leads to underestimation of the actual AA digested. We also combine protein sources to optimize their nutrition profile, for which protein digestibility coefficients must be additive. SID therefore provides a more accurate and appropriate measurement of the AA ileal digestibility of single protein sources or protein mixtures (Stein, Pedersen, Wirt, & Bohlke, 2005). To improve comparison between results, studies involving rodents, gilts, sows, piglets, weanling pigs, and finishing pigs were excluded and instead only data obtained with growing pigs were used. In addition, only protein material containing at least 10% crude protein, and falling in categories such as seeds, meals, flours, concentrates, and isolates, was included in the investigation. Material in the forms of hulls, brans, or peels was excluded, being less suitable sources of protein for human consumption. The obtained datasets were subsequently used to simulate protein mixtures and highlight the complementarity of proteins at specific ratios to obtain higher values than that of the individual proteins.

| Data collection from the literature
Five animal (whey, casein, egg, gelatin, and pork meat) and 12 plant protein sources (soy, pea, lupin, fava bean, rapeseed, canola, hemp, wheat, potato, oat, rice, and corn) were selected, based on their occurrence in the Western diet or by the growing interest in the vegetarian, vegan, or flexitarian markets. Growing pig intervention studies providing complete IAA composition, crude protein content (CP), and IAA standardized ileal digestibility (SID) were selected.

| DIAAS of a single protein source
The most limiting digestible indispensable amino acid content (DIAA) defines the DIAAS value of a protein. DIAA ratios were determined according to the three reference pattern scores defined by FAO: infant (0-6 months), children (0.5-3 years), and children older than 3 years, adolescents, and adults (FAO, 2013).
For a given IAA "y," DIAA ratio is calculated as follows: where IAA y is expressed as mg/g CP.
The lowest DIAA ratio leads to the DIAAS value of a protein: Mean DIAAS values were obtained from the lowest mean DIAA for each protein source. Graphs were generated with Microsoft Excel Office 365 and Minitab software version 18.1.

| DIAAS of protein mixtures
Digestible indispensable amino acid content of protein mixtures were calculated as detailed by FAO and according to the reference pattern score of 0.5-to 3-year-old population group (FAO, 2013).
Average values of single proteins for IAAy and SIDy were used to estimate the DIAAys of protein mixtures. The latter is written as a linear combination with respect to the mixing ratios Ri. For a mixture of protein 1 (P 1 ), protein 2 (P 2 ) up to protein n (P n ), this results in: where: (1) DIAAy ratio = IAAy × SIDy Reference pattern score IAAy (2) DIAAS = 100 × lowest DIAA ratio among IAAs (3) DIAAy = c y,1 × R 1 + c y,2 × R 2 + ⋯ + c y,n × R n (4) c y,i = IAAyi × SIDyi Reference pattern score IAAy and Note that Equation (4) is independent of material crude protein content. Mi is the amount of pure protein from protein source i. By using pure protein, DIAAy remains independent of the protein content of individual material used in the mixture.
For a mixture of two protein sources, Equation (3) can be rewritten by applying R 2 = 1 -R 1 : The optimal mixture provides the maximum DIAAS among all possible ratios. The corresponding R 1 value was determined with the Solver tool available in Microsoft Excel, using the constraint 0 ≤ R 1 ≤ 1. The objective was set to find the highest DIAAS ≤ 100.  And using R 3 = 1 -R 1 -R 2 , this results in: The values of R 1 , R 2 , and R 3 for the optimal mixture were also calculated with the Solver tool available in Microsoft Excel. Equation (8) was optimized by modifying both R 1 and R 2 . Values of R 1 and R 2 were constrained to 0 ≤ R 1 ≤ 1 and 0 ≤ R 2 ≤ 1 with the objective to find the highest DIAAS ≤ 100.
Similar to Equation (2), the DIAAS score is obtained from the lowest DIAA value of the protein mixture:

| DIAAS variation within protein datasets
Studies providing complete IAA composition, CP content, and IAA SID were selected (Table 1). The amount of data available from the literature varies according to the protein source. Digestibility of corn, soy, canola, and wheat proteins has been extensively studied in growing pigs, evidenced by the relatively large number of datasets obtained through numerous references (Table 1 and Table S1).
Contrarily, complete data on ileal digestibility of amino acids from hemp and casein are limited to 1 and 2 datasets, respectively.
Each retrieved dataset was used to obtain the DIAAS value according to the three scoring patterns. As displayed in Figure 1, DIAAS values are scattered across the different protein quality categories for many protein sources. Based on the 0.5-to 3-year-old scoring pattern, soy and wheat proteins possess DIAAS values in all three quality categories (<75, no quality protein claim; 75-99, high-quality protein claim; and ≥100, excellent quality protein claim), while scores of wheat, rapeseed, lupin, pea, potato egg, and canola proteins fall in two of the quality categories. This is further highlighted with the broad absolute DIAAS variations observed within most protein datasets. Based on the 0.5-to 3-year-old scoring pattern, DIAAS values of wheat, rapeseed, and corn proteins display the greatest disparity with respective DIAAS ranges of 53, 43, and 44. DIAAS variation tends to decrease for the infant group (0-0.5 years old), with range values of 32, 20, and 34 for wheat, rapeseed, and corn proteins, respectively. Such decreases in variation can be explained by higher IAA requirement expressed in the scoring pattern and the shift in limiting amino acid as detailed in Table S2.
Given the varying protein content among protein sources, the absence of measurements of antinutritional factors (ANFs), and the limited description of the process used to obtain the various protein sources, a selection cannot be based on a specific determinant. For this reason, the average DIAAS value obtained from each protein dataset was selected.
For this age group, the limiting amino acid of protein obtained from cereal grains-such as corn, wheat, hemp, rice, canola, oat, rapeseed-is lysine (Lys), while the leguminous sources of protein (fava bean, pea, lupin, soy) are limited by the sulfur-containing amino acids methionine and cysteine (Met + Cys). Potato protein, which interestingly shows a high DIAAS value, is derived from a tuberous plant and therefore does not belong to any of these categories (Table 2).
(7) DIAAy = c y,1 × R 1 + c y,2 × R 2 + c y,3 × R 3 (8) DIAAy = c y,1 − c y,3 × R 1 + c y,2 − c y,3 × R 2 + c y,3 (9) DIAAS (P 1 + P 2 +…+P n ) = 100 × lowest DIAA (P 1 + P 2 +…+P n ) F I G U R E 1 Variation in DIAAS obtained from SID and IAA data available from pig intervention studies. DIAAS calculated for each scoring pattern as defined by FAO (2013): infant (0-0.5 years), children (0.5-3 years), and children older than 3 years, adolescents, and adults TA B L E 2 Digestible indispensable amino acid scores of various protein sources according to the 0.5-to 3-year-old reference pattern score Dataset does not include tryptophan measurements. This is however the only publication reporting the ileal digestibility of hemp in pig studies. These data were therefore included for comparison.
Due to different protein requirements between age groups, the limiting amino acid of the protein sources varies per age group.
A protein source reaching a DIAAS of 100 or above indicates that none of its amino acids is limiting and this sole protein source should be able to meet physiological requirements. Among the proteins selected in this study, potato protein, egg protein, casein, and pork meat reach this level based on the 0.5-to 3-year-old reference pattern. For a protein source with a DIAAS lower than 100, different strategies are possible to ensure an adequate protein intake based on such protein. A first option is to increase the protein intake of the limiting protein until the physiological requirement is reached. Higher protein intake has indeed shown to increase amino acid uptake in plasma (Gorissen et al., 2016). For example, based on the selected datasets, less than three portions of corn protein (2.8*36 = 101) would theoretically be needed to meet these requirements while 1.45 portions of pea protein (1.45*70 = 101) would be required. A second option is to combine protein sources and ensure complementarity of their amino acids to reach a higher DIAAS.

| Toward a higher DIAAS with protein combinations
The digestibility score of each indispensable amino acid highlights the potential of complementarity between protein sources (Table 2).
Cereal-based proteins, scoring low in Lys but high in Met + Cys, can to some extent complement leguminous proteins, scoring high in Lys but low in Met + Cys. Potato protein, which scores higher than 100 for each DIAA value, can also serve to increase DIAAS of many proteins. The degree of complementarity depends on the ratio of the combined protein sources, as illustrated in Figure 3.
Multiple simulations were used to calculate the maximum DIAAS of mixtures according to Equations (6), (8), and (9), enabling mixtures such as oat/lupin, fava bean/corn, and pea/wheat to reach the high-quality protein range (Table 3). Not all leguminous/cereal protein combinations will lead to higher scores: If Lys is the second or third limiting IAA in leguminous protein, the DIAAS increase will be minimal. This is the case of lupin protein scoring 75 for Lys (Table 2).
Among these mixtures, potato protein shows the ability to increase the DIAAS of most plant protein sources to 100. In combination with casein, egg protein, or pork meat, most plant proteins can reach a DIAAS of 100 until a certain ratio is reached (Table S4).

| Factors influencing DIAAS variability
Protein IAA content and SID coefficients were used to determine DIAAS values for 17 protein sources. Broad absolute DIAAS variations can be observed within most protein datasets. This can be attributed to the type of material fed to the growing pigs.
Similarly, the content of antinutritional factors (ANFs), for example, glycoalkaloid, glucosinolates, protease inhibitors, phytate, and saponins, differs among genotypes (Oomah et al., 2011;Sharma, Kaur, Goyal, & Gill, 2014). Along with processing conditions, ANFs are well-known to influence plant protein digestibility (Sá, Moreno, & Carciofi, 2019;Sarwar Gilani, Wu Xiao, & Cockell, 2012). For instance, Luo et al. (Luo & Xie, 2013) reported that phytate and trypsin inhibitor contents increase with dehulling of fava beans, but was most effectively reduced by first soaking the beans before dehulling followed by an autoclaving step. Pastuszewska et al. (Pastuszewska, Tuśnio, Taciak, & Mazurczyk, 2009) also highlighted the variations F I G U R E 2 Average DIAAS of various protein sources according to the three reference pattern scores: infant (0-0.5 years), children (0.5-3 years), and children older than 3 years, adolescents, and adults. Error bars represent standard deviation in the activity of solanidine glycoalkaloids and trypsin inhibitors in potato protein concentrate among process conditions from various potato starch production sites. Furthermore, fermentation or enzymatic treatments of soybean have also shown their potential to reduce the content of ANFs leading to an increased AA digestibility in young pigs (Zhang et al., 2013). In the studies selected for this analysis, the diverse processes used to obtain the protein materials-such as specific heat treatment, solvent extraction conditions, or enzymatic reaction-are not consistently described; the resulting activity of ANFs is rarely investigated. Moreover, many of the datasets used in the current study are based on raw feed ingredients.
Such material often displays a higher ANF content and ANF activity than dietary protein used in food products. Feedstuffs in their raw state do not reflect the processed and cooked forms of food used for human consumption. Processing and especially cooking can considerably affect amino acid composition and digestibility of proteins, resulting in a different DIAAS (Bailey, Mathai, Berg, & Stein, 2019;Friedman, Gumbmann, & Masters, 1984). While heat treatment can inactivate some of the ANFs, it can also induce molecular alterations making the protein more resistant to the action of digestive proteases or on the contrary more accessible (Carbonaro, Cappelloni, Nicoli, Lucarini, & Carnovale, 1997;Duodu, Taylor, Belton, & Hamaker, 2003;Liu, Zheng, & Chen, 2019). This is also illustrated in the present study in which pea protein obtains an average DIAAS of 70 while after extrusion its DIAAS can increase to 82 or 86 depending upon the extrusion temperature (Table S2).
Besides digestibility, amino acid composition of cooked protein food can differ greatly from that of its raw state due to leaching of soluble protein fractions into the boiling liquid and through the formation of amino acid derivatives (Carbonaro et al., 1997;Friedman et al., 1984;Nierle, 1985;Struthers, 1981). With an altered amino acid composition, the DIAAS value of the cooked protein, and possibly its limiting amino acid, may differ strongly. Animal-based proteins are also subjected to protein quality variation as a result of processing. Whey protein is generally considered an excellent protein with a DIAAS superior to 100, but surprisingly, multiple references led to DIAAS values ranging from 78 to 88 for whey-based products, implying that purity and the recovery process can be significant. Although findings based on hulls, brans, and peels were excluded, remarkable difference can be observed in the purity of the protein materials studied. Flours contain more fibers than protein isolates or concentrates, thereby also influencing the protein digestibility (Mosenthin, Sauer, & Ahrens, 1994;Myrie, Bertolo, Sauer, & Ball, 2008). It is suggested that protein and nonstarch polysaccharides increase production of pancreatic juices and bile, and stimulate secretion of gut mucin protein composing the mucosal layer, thus contributing to higher ileal AA endogenous losses (Low, 1989;Morel, Melai, Eady, & Coles, 2005). This indicates that applying endogenous loss obtained from a N-free diet control group to the intervention group could cause an underestimation of the amino acid SID of the studied protein material. True ileal digestibility (TID), also referred to as real ileal digestibility, accounts for diet-specific endogenous AA loss and is therefore more accurate. However, data based on TID remain scarce due to methodological complications involving such measurement.
Variations observed within each DIAAS dataset can also originate from the conditions in which the intervention diets were provided. Firstly, clear dissimilarities can be noted in starting age/ body weight of the growing pigs, varying from 17.8 ± 1.7 kg (Zhang et al., 2015) to 76.2 ± 5.6 kg (Pedersen, Boersma, & Stein, 2007).
Several publications (Cunningham, Friend, & Nicholson, 1962;Susenbeth, Dickel, Diekenhorst, & Höhler, 1999) reported a lower CP digestibility in early growing pigs compared with both growing and finishing-growing pigs. Likewise, younger gestating sows showed a lower apparent crude protein digestibility than older gestating sows (Jacyno et al., 2016). Based on similar results, Hennig and colleagues (Hennig, Bock, Wünsche, & Kreienbring, 1979) suggested that endogenous loss could be of influence. This hypothesis was confirmed by Nitrayova et al. (Nitrayová, Brestenský, Patráš, & Heger, 2013), who reported an increased AA TID concomitantly with a lower AID in pigs of 20.6 kg compared to pigs with average weight of 64.7 kg. Young animals thus tend to produce higher endogenous AA loss than older pigs. Since SID is obtained by correcting AID with endogenous AA losses, SID coefficients will consequently increase in younger animals. This suggests that neonatal pig and piglets could be more suitable models to calculate the DIAAS values of the infant group than growing pigs (Buddington, Ja, Puchal-Gardiner, & Sangild, 2001;Moughan, Birtles, Cranwell, Smith, & Pedraza, 1992).
Secondly, the daily amount of protein feed provided to the animals varies among studies. For instance, Cervantes-Pahm  (Rayadurg & Stein, 2003) reported a linear decrease in ileal endogenous losses of CP and most AAs as the diet increased from 1 to 3 times the maintenance requirement of growing barrows. Such reduced endogenous ileal AA losses will lead to a greater SID of AAs (Moter & Stein, 2004), eventually impacting the DIAAS value. This further highlights the importance of providing similar animal management conditions to obtain objective DIAAS values. Moreover, three major animal feed databases (CVB, 2016;NRC, 2012;Sauvant, Perez, & Tran, 2004) were included in our datasets, but distinctions between growing pigs, sows, finishing pigs, or piglets are not always detailed, and this also likely contributes to the variation observed in the collected data.
Overall, the differences in the setup of intervention studies highlight the need for harmonized methods when determining growing pig SID. Harmonization might reduce the variation observed in the current investigation and would generate more reliable data for plant proteins. These adjustments should be accompanied by a more detailed characterization of intervention material to ensure reliable DIAAS comparison between protein sources intended to human consumption. Ideally, each food protein source should be classified into subcategories detailing their content in fibers, content of ANFs, and specific processing conditions. sequence may influence the rate of hydrolysis by digestive proteases. Gastric pepsin is known for its affinity with hydrophobic sites (Fruton, 1970) and its propensity to cleave after phenylalanine and leucine residues (Powers, Harley, & Myers, 1977), while pancreatic trypsin favors basic amino acids (arginine and lysine) (Evnin, Vásquez, & Craik, 1990). Moreover, the protein secondary conformations seem to hinder the accessibility of digestive proteases to cleaving sites. High content of beta-sheet has been reported in soy protein (>40%) (Herrero, Jiménez-Colmenero, & Carmona, 2009), peas (>40%) (Beck, Knoerzer, & Arcot, 2017), rice (44,9%) (Wang, Wang, Wang, & Chen, 2016), wheat gluten (50%) , and oat (36,8%-74%) (Liu et al., 2009;Zhao et al., 2017), while such structural arrangement tend to account for less than 30% in milk-based protein (Carbonaro, Maselli, & Nucara, 2012;Curley, Kumosinski, Unruh, & Farrell, 1998) and meat proteins (Herrero, Carmona, Lopez-Lopez, & Jimenez-Colmenero, 2008). This relatively high proportion of beta-sheet conformations in leguminous and cereal-based proteins results in reduced protein in vitro digestibility (Carbonaro et al., 2012;Yang et al., 2016). Furthermore, multiple compounds present in plant proteins can interact with digestive enzyme by forming stable complexes and impairing the enzymatic functionality, thereby inhibiting protein digestion (Sarwar Gilani et al., 2012). As detailed earlier, it can be prevented or reduced by suitable processing in the form of heat-induced denaturation, enzymatic reaction, or fermentation (Sá et al., 2019;Sarwar Gilani et al., 2012).

| Distinctive DIAAS between protein sources
The vast majority of data used in this investigation are from animal nutrition studies, that is, protein preparations intended for animal feed. Protein products intended for human nutrition may differ in their processing and overall composition. New data on protein material that is specifically intended for food consumption are therefore needed to confidently apply their DIAAS value to human consumption.

| Seeking nutritional efficiency
The cause of disparities in DIAAS values appears multifactorial.
Depending on the processing conditions, purity of the material, or feeding conditions, a protein DIAAS will vary, making the choice of a representative product per protein source challenging. To further compare the DIAAS of protein sources, we therefore took an objective approach by considering the mean DIAAS values for the 0.5-to 3-year-old population group. Although DIAAS expresses the quality of protein based on its most limiting IAA score, it also highlights the opportunity to combine the strength of different protein sources.
The increased DIAAS values obtained from mixtures show the potential to achieve a protein nutritional efficiency with sustainable protein sources. Nutritional efficiency lies in meeting physiological requirements with minimal intake of high-quality protein, as opposed to higher protein intake of low-quality protein. This last scenario is not necessarily favorable due to a potentially high satiety effect resulting in inadequate protein intake and suboptimal sustainability impact of such a diet. Potato protein has shown to be the most promising source of plant protein to achieve nutritional efficiency: With high scores obtained for every IAA, it can boost most mixtures to the excellent protein quality category (DIAAS ≥100). Similarly, soy protein and pea proteins provide good potential to boost many cereal proteins toward the high-quality protein category (DIAAS = 75-99).
While simulating protein mixtures, crude protein content was not taken into consideration due to the considerable variation among and between proteins. Equations (6) and (8)  ing IAAs, is also one of the most reactive AAs (Hurrell, Carpenter, Sinclair, Otterburn, & Asquith, 1976). During thermal processing, chemical reactions between reducing sugars and lysine lead to formation of Maillard reaction products. Such reactivity reduces the availability of lysine (Nyakayiru et al., 2020), hence the FAO recom-

| CON CLUS IONS
In conclusion, this investigation shows that protein quality based on DIAAS differs not only between animal and plant proteins but also between plant protein sources. Based on the 0.5-to 3-year-old scoring pattern, potato and most animal-derived proteins tend to reach the excellent protein quality category (DIAAS ≥100). Soy and whey proteins fall into the high-quality protein range (DIAAS = 75-99).
Gelatin, corn, wheat, hemp, fava bean, oat, pea, canola, rapeseed, lupin, and rice proteins lie in the no quality claim category (DIAAS <75). Such scores differ greatly for the infant population, for which only casein is of good quality in the current selection. The differences observed in the protein quality lead to opportunities to enhance their nutritional efficiency in the form of protein mixtures. Any protein material can reach a higher DIAAS value when combined with an adequate complementary source. In the current selection, potato, soy, and pea proteins seem the most promising plant-based complementary sources to reach high or excellent quality mixtures to support a plant-based lifestyle. Clear dissimilarities were observed within each protein dataset suggesting that: (a) harmonized methods to evaluate protein ileal digestibility; (b) proper characterization of (cooked) protein material; and (b) digestible reactive lysine data are needed to obtain a coherent DIAAS database.

ACK N OWLED G M ENTS
The authors thank Robin Spelbrink for his support and constructive comments in completion of this manuscript. Our acknowledgments are also addressed to Michael Polhuis for proofreading this manuscript.

CO N FLI C T S O F I NTE R E S T
L.H., P.N., B.P., and M.C.L. are employees of Avebe U.A.