True ileal amino acid digestibility

To determine true ileal amino acid digestibility, a human or animal is fed a test diet and ileal digesta are collected. The amino acid content of both diets and digesta are then determined using conventional amino acid analysis and by difference digestibility is calculated. Amino acid analysis techniques have remained largely unchanged since first developed in the 1950's and involve the hydrolysis of protein in 6 M HCl at 110°C for approximately 24 h in an oxygen free environment. These hydrolysis conditions are severe and some amino acids are chemically altered during hydrolysis and cannot be quantitatively determined. For such amino acids, other hydrolytic strategies must be used. Asparagine and glutamine are determined as their acid derivatives aspartic acid and glutamic acid. Tryptophan is determined using base hydrolysis which improves tryptophan yields in comparison with hydrochloric acid hydrolysis but even then complete recovery of tryptophan is seldom achieved. For cysteine and methionine, quantitative conversion to cysteic acid and methionine sulphone using performic acid oxidation must be carried out prior to hydrochloric acid hydrolysis to improve yields, however losses of cysteine are still observed. Serine and threonine undergo hydrolytic losses such that 10–15 % of these amino acids are lost after 24 h acid hydrolysis.

Lysine

Lysine is a dietary essential (indispensible) amino acid for animals and humans. Lysine also possesses a reactive side chain amino group that can react with other compounds present in a food to form nutritionally unavailable derivatives (Maillard products)⁽Reference Hurrell and Carpenter¹⁾. This reaction can occur at any temperature but is greatly accelerated at elevated temperatures such as those used during heat processing or cooking. Rutherfurd and Moughan⁽Reference Rutherfurd and Moughan⁶⁾ showed that the decrease in the reactive lysine content of heated field peas was more than three times greater after heating at 165°C compared to heating for the same length of time at 150°C. Other factors can influence the rate of Maillard product formation, including pH, water activity and reactant concentration⁽Reference Franzen, Singh and Okos⁷⁾. Some of these lysine derivatives formed during processing or long-term storage are acid labile and can revert back to lysine during hydrochloric acid hydrolysis leading to an overestimate of lysine content⁽Reference Hurrell and Carpenter¹⁾. When conventional amino acid analysis techniques are employed, the reverted lysine and undamaged lysine cannot be distinguished and are determined as “total lysine” and consequently total lysine estimates may be misleading⁽Reference Rutherfurd and Moughan²⁾. Furthermore, digestibility coefficients for total lysine, which are derived from the total lysine in the diet and digesta, in samples that contain significant amounts of acid labile lysine derivatives are confounded and should not be used.

A number of assays have been developed to determine reactive lysine (the lysine that possesses an unreacted side chain amino group), the most common of which include the fluorodinitrobenzene (FDNB) method⁽Reference Carpenter^8, Reference Booth⁹⁾, guanidination method⁽Reference Mauron and Bujard¹⁰⁾, trinitrobenzenesulphonic acid (TNBS) method⁽Reference Kakade and Leiner¹¹⁾ and furosine method⁽Reference Desrosiers, Savoie and Bergeron¹²⁾. However, reactive lysine is not completely absorbed from the digestive tract⁽Reference Moughan, Gall and Rutherfurd¹³⁾ and none of the latter methods account for the incomplete digestion and absorption from the gastrointestinal tract that is common among heat-treated foods⁽Reference Moughan and Rutherfurd¹⁴⁾. Moughan and Rutherfurd⁽Reference Moughan and Rutherfurd¹⁴⁾ developed a method for determining the true ileal digestibility of reactive lysine whereby the reactive lysine content of both diets and digesta were determined using the guanidination reaction (reaction of O-methylisourea with the ɛ-amino group of lysine to form acid stable homoarginine) and subsequent analysis of homoarginine. The guanidination reaction with O-methylisourea (OMIU) is specific for the ɛ-amino group of lysine and OMIU does not react with the α-amino group of amino acids⁽Reference Moughan and Rutherfurd^14–Reference Zhang, Li and Qiao²¹⁾ with the exception of glycine⁽Reference Wang, Xu and McGrath²²⁾. Consequently, guanidination is suitable for the analysis of reactive lysine in digesta where free lysine may be present. True ileal digestible reactive lysine can then be estimated by multiplying the true ileal reactive lysine digestibility by the reactive lysine content of the food or feedstuff. From a theoretical standpoint, digestible reactive lysine is equivalent to available lysine. Moreover, Rutherfurd et al. ⁽Reference Rutherfurd, Moughan and Morel²³⁾ conducted a study to investigate the accuracy of the assay to predict the available lysine content of a heated skim milk powder. The study, based on body lysine retention in the growing pig, employed a test diet and two control diets containing either heated skim milk powder or enzymatically hydrolysed casein (EHC) and free amino acids respectively as the sole nitrogen source. All diets were identical except for the nitrogen source and lysine in the EHC/free amino acid-based diet was assumed to be completely digested and absorbed and lysine was shown experimentally to be the first limiting amino acid in all diets. The control diets were formulated to contain lysine at the same level as the heated skim milk powder diet based on either true ileal total lysine digestibility (traditional amino acid analysis), or true ileal reactive lysine digestibility (guanidination). Rutherfurd et al. ⁽Reference Rutherfurd, Moughan and Morel²³⁾ reported that the whole body lysine deposition of the pigs fed the heated skim milk powder-based diet was similar to that in the pigs fed the control diet formulated based on reactive lysine digestibility but was significantly and markedly different to that for the pigs fed the control diet formulated based on total lysine digestibility (Table 1). This study demonstrated the inaccuracy of total lysine digestibility and the accuracy of reactive lysine digestibility as a measure of lysine availability and showed that the inaccuracy of the true ileal digestible total lysine measure in processed feedstuffs was due to inadequacies in the traditional amino acid analytical approach rather than being due to an inherent flaw in the ileal digestibility approach.

Table 1 Least-squares means (n=8) of whole body lysine deposition (g d⁻¹) in pigs fed a heated skim milk powder based diet and one of two EHC1 control diets

Reproduced with permission from Rutherfurd et al. ⁽Reference Rutherfurd, Moughan and Morel²³⁾; copyright 1997 American Chemical Society.

¹ Enzymatically hydrolysed casein.

² EHC Diet A was formulated to contain lysine equal to the digestible lysine content of the heated skim milk powder determined using the conventional ileal digestibility assay (reactive lysine in heated skim milk powder x true digestibility of total lysine (determined using conventional methods) for the heated skim milk powder).

³ EHC Diet B was formulated to contain lysine equal to the digestible lysine content of the heated skim milk powder determined using the new ileal reactive lysine digestibility assay (reactive lysine in heated skim milk powder x true digestibility of reactive lysine (determined using the new method) for the heated skim milk powder).

^a Means with difference superscripts were significantly (P < 0·001) different.

The difference between digestible total lysine and digestible reactive (available) lysine for processed foods and feedstuffs tested in our laboratory ranges from nil for minimally processed foods such as UHT milk or a protein supplement to 143 % for severely heat-treated feedstuffs such as moist cat food (Table 2). Consequently, ileal digestible lysine predicts available lysine accurately for some processed foods and feedstuffs but for many others the overestimation is considerable. As a result, where the accurate prediction of available lysine is required it is advisable that the digestible reactive lysine rather than digestible total lysine be determined particularly since to do so requires only an additional chemical analysis step rather than a separate bioassay.

Table 2 Digestible total and reactive (available) lysine contents (g  kg⁻¹) for a range of protein sources

Data reported by Rutherfurd et al. ⁽Reference Rutherfurd, Markwick-Rutherfurd and Moughan^84–Reference Rutherfurd, Moughan and van Osch⁸⁶⁾ and Rutherfurd and Moughan⁽Reference Rutherfurd and Moughan⁸⁷⁾.

¹ Based on the total lysine content of the food or feedstuff and the true ileal total lysine digestibility.

² Based on the reactive lysine content (guanidination method) of the food or feedstuff and the true ileal reactive lysine digestibility (guanidination method).

Methionine and cysteine

The sulphur amino acids methionine and cysteine can be oxidised in foods that undergo processing or prolonged storage to methionine sulphoxide and methionine sulphone (for methionine) and cysteic acid (for cysteine)⁽Reference Anderson, Ashley and Jones²⁴⁾. In addition cysteine can also react with glucose to produce Maillard products⁽Reference Tai and Ho²⁵⁾ and with lysine to form lysinoalanine⁽Reference Friedman, Levin and Noma²⁶⁾. Methionine sulphone and cysteic acid have been shown to be nutritionally unavailable in rats⁽Reference Anderson, Ashley and Jones^24, Reference Friedman and Gumbmann^27–Reference Sjøgerg and Bostrøm²⁹⁾, chickens⁽Reference Kuzmicky, Kohler and Walker²⁸⁾ and turkeys⁽Reference Parsons and Potter³⁰⁾, although cysteic acid can be used by some species to synthesise taurine⁽Reference Anderson, Ashley and Jones^24, Reference Edgar, Hickman and Marsden³¹⁾. Methionine sulphoxide on the other hand can be partially utilised in rats⁽Reference Anderson, Ashley and Jones^24, Reference Sjøgerg and Bostrøm^29, Reference Baker^32–Reference Njaa³⁵⁾, chickens⁽Reference Haaland, Arnesen and Njaa³⁶⁾, mice⁽Reference Friedman and Gumbmann^27, Reference Baker³²⁾ and turkeys⁽Reference Parsons and Potter³⁰⁾ by the action of methionine sulphoxide reductase which converts methionine sulphoxide to methionine in vivo ⁽Reference Moskovitz^37–Reference Nakamura, Higuchi and Iwami³⁸⁾. The extent of dietary methionine sulphoxide utilisation is also dependent on the dietary concentrations of methionine, cysteine and methionine sulphoxide⁽Reference Anderson, Ashley and Jones^24, Reference Haaland, Arnesen and Njaa^36, Reference Gjøen and Njaa³⁹⁾ as dietary methionine sulphoxide will have a sparing effect on methionine so long as the amount of conversion of methionine sulphoxide to methionine required does not exceed the capacity of the methionine sulphoxide reductase enzyme.

True ileal sulphur amino acid digestibility takes into account incomplete digestion and absorption but may not give accurate values for availability for protein sources that have undergone oxidation, since the sulphur amino acids (methionine and cysteine) are traditionally determined following performic acid oxidation to quantitatively convert methionine to methionine sulphone and cysteine to cysteic acid which are then determined using amino acid analysis. Consequently, if significant amounts of methionine sulphoxide or methionine sulphone are naturally present in the food the methionine will be overestimated. The same is true for cysteic acid and cysteine. Recently, Rutherfurd and Moughan⁽Reference Rutherfurd and Moughan⁴⁰⁾ proposed that available methionine in oxidised foods could be predicted as shown below:

$$\begin{eqnarray} Available\,methionine = Met_{food}\times Met\,digestibility + MetO_{food}\times MetO\,digestibility\times MetO\,utilisation \end{eqnarray}$$ $$

Where Met_food is the actual methionine content of the food, Met digestibility is the true ileal digestibility of methionine, MetO_food is the methionine sulphoxide content of the food, MetO digestibility is the true ileal methionine sulphoxide digestibility and MetO utilisation is the utilisation of dietary methionine sulphoxide based on the methionine, methionine sulphoxide and cysteine contents of the food. An accurate estimation of available methionine therefore requires the accurate determination of methionine and methionine sulphoxide in both diets and digesta of a test animal or human as well as an understanding of the quantitative relationship between the extent of methionine sulphoxide utilisation and the dietary concentrations of methionine, methionine sulphoxide and cysteine. To that end the analysis of methionine, cysteine and methionine sulphoxide will be discussed.

While performic acid oxidation followed by hydrochloric acid hydrolysis is the most common method used for determining sulphur amino acids in foods, in samples where all the methionine molecules are unoxidised, hydrochloric acid hydrolysis alone can be used to release methionine if rigorous removal of oxygen is undertaken prior to hydrochloric acid hydrolysis⁽Reference Finley⁴¹⁾ since the loss of methionine during hydrochloric acid hydrolysis under oxygen free conditions is negligible⁽Reference Rutherfurd, Schneuwly and Moughan⁴²⁾. Methionine sulphoxide can also be reduced to methionine during hydrochloric acid hydrolysis⁽Reference Rutherfurd and Moughan^40, Reference Puchala, Piór and von Keyserlingk^43–Reference Todd, Marable and Kehrberg⁴⁵⁾ which is in contrast to the commonly held belief that methionine is oxidised during acid hydrolysis. Rutherfurd and Moughan⁽Reference Rutherfurd and Moughan⁴⁰⁾ suggested that this may be due to the relative flux of oxidation of methionine and reduction of methionine sulphoxide as a result of the extent to which oxygen is removed from the samples being hydrolysed. Consequently, hydrochloric acid would be an unsuitable hydrolysing agent for foods that contain significant amounts of methionine sulphoxide.

Several hydrolytic methods have been reported for determining methionine sulphoxide and these include hydrolysis with methanesulphonic acid⁽Reference Puchala, Piór and von Keyserlingk^43–Reference Sochaski, Jenkins and Lyons^44, Reference Chiou and Wang^46–Reference Simpson, Neuberger and Liu⁴⁷⁾ or p-toluenesulphonic acid⁽Reference Hayashi and Suzuki⁴⁸⁾, and alkaline hydrolysis⁽Reference Sjøgerg and Bostrøm^29, Reference Nakamura, Higuchi and Iwami^38–Reference Gjøen and Njaa^39, Reference Todd, Marable and Kehrberg^45, Reference Cuq, Besancon and Chartier⁴⁹⁾. A method has also been reported based on barium hydroxide hydrolysis where methionine and methionine sulphoxide can be determined in the one analysis procedure⁽Reference Njaa⁵⁰⁾. Rutherfurd and Moughan⁽Reference Rutherfurd and Moughan⁴⁰⁾ reported that hydrolysis in methanesulphonic acid resulted in a substantial loss (50 %) of methionine sulphoxide in an oxidised fishmeal after 24 h methanesulphonic acid hydrolysis which contrasts with the findings of some workers who have reported near complete recovery of methionine sulphoxide using methanesulphonic acid hydrolysis in purified proteins⁽Reference Chiou and Wang⁴⁶⁾ and rumen samples spiked with pure methionine sulphoxide⁽Reference Puchala, Piór and von Keyserlingk⁴³⁾. Given the conflicting reports for the analysis of methionine sulphoxide, it would appear that more work needs to be conducted on methods for the accurate analysis of methionine sulphoxide. Furthermore, and to determine ileal methionine sulphoxide digestibility, any promising method would also need to be optimised for digesta samples.

Finally, while the discussion above describes the inadequacies of current analytical methods for determining methionine in proteins that contain methionine sulphoxide or methionine sulphone, it must also be recognised that a significant amount of methionine is used metabolically for purposes other than for protein synthesis, for example, as a methyl donor. Consequently, estimates of utilisable methionine based on body methionine retention such as the slope ratio assay will inevitably underestimate the available methionine content of food proteins as they do not account for the methionine utilised for purposes other than protein synthesis.

Threonine

During heating, threonine can react with sugars to form pyrazines, pyridines and pyrroles⁽Reference Ho and Chen⁵¹⁾ with hundreds of individual compounds being identified⁽Reference Baltes and Bochmann^52–Reference Baltes and Bochmann⁵⁵⁾. These structures, however, are unlikely to interfere with the analysis of threonine and such reactions are unlikely to be quantitatively significant under normal conditions of food processing.

The presence of phosphorylated amino acids may lead to an overestimate of the availability of those phosphorylated amino acids, particularly threonine. Phosphothreonine cannot be used directly for the synthesis of body proteins. O- and N-phospho-amino acids occur naturally in many dietary proteins but in addition, controlled phosphorylation is commonly used as a manufacturing technique to alter the functional properties of food proteins⁽Reference Shih⁵⁶⁾. Furthermore, uncontrolled phosphorylation may occur when foods are processed at higher temperatures.

Phosphorylated serine, threonine and tyrosine are base labile and partially acid labile⁽Reference Matheis, Penner and Feeney^57–Reference Anderson and Kelley⁵⁸⁾. Under the hydrolytic conditions employed during amino acid analysis (6 M HCl, 24 h, 110°C) recoveries of synthetic phosphothreonine and phosphoserine have been reported to be 36 % and 2 % respectively with 46 % and 81 % of the original phospho-amino acids being dephosphorylated to threonine and serine respectively⁽Reference Bylund and Huang⁵⁹⁾. Similar results were observed for both phosphothreonine and phosphoserine when hydrolysed from phosphothreonine and phosphoserine containing peptides⁽Reference Bylund and Huang⁵⁹⁾. It is likely that for processed protein sources conventional hydrochloric acid hydrolysis overestimates the amounts of threonine and serine if the phosphorylated forms of these amino acids are present and this may have implication for the determination of available threonine and serine if phosphothreonine and phosphoserine cannot be utilised. In addition, as is the case for unphosphorylated threonine and serine, not all the phosphothreonine and phosphoserine can be recovered during hydrochloric acid hydrolysis.

Threonine is a quantitatively important amino acid in the production of intestinal proteins, particularly mucins. As much as 60-100 % of the absorbed threonine is extracted and utilised by the portal drained viscera (PDV) (the intestine, pancreas, spleen and stomach) of pigs⁽Reference Stoll, Henry and Reeds^60–Reference van der Schoor, Reeds and Stoll⁶¹⁾. Furthermore, systemic amino acids would appear to be poorly utilised by the intestine since Dudley et al. ⁽Reference Dudley, Wykes and Dudley⁶²⁾ showed that mucosal protein synthesis was substantially lower in piglets fed parenterally compared to those fed enterally. Thus the bulk of dietary threonine is used for PDV protein synthesis and the bulk of the threonine used for PDV protein synthesis is likely to be of dietary origin.

Alkaline phosphatase catalyses the dephosphorylation of a wide range of molecules and is present in the enterocytes of the small intestine, predominantly in the duodenum⁽Reference Hietanen^63–Reference Nakano, Kawamoto and Takano⁶⁵⁾. However, during a meal containing large amounts of phosphothreonine the activity of intestinal alkaline phosphatase may be insufficient to quantitatively dephosphorylate the absorbed phosphothreonine before it is transported from the brush border to the portal vein. Consequently, the latter phosphothreonine molecules would be largely lost to PDV protein synthesis. The absorbed phosphothreonine may be dephosphorylated in the body but given that most of the threonine requirement is for intestinal protein synthesis the net result may be poor utilisation of threonine in processed foods that contain large quantities of phosphothreonine.

The impact of phosphorylation on the in vitro digestibility of phosphorylated casein⁽Reference Matheis, Penner and Feeney⁵⁷⁾, and soya bean proteins⁽Reference Hirotsuka, Taniguchi and Narita^66–Reference Sung, Chen and Liu⁶⁷⁾ using a selection of proteases has been reported to be minor. In contrast, Giec et al. ⁽Reference Giec, Stasifiska and Skupin⁶⁸⁾ reported a 10 % decrease in the in vitro digestibility of a yeast homogenate after phosphorylation. To the authors' knowledge no in vivo studies have been conducted investigating the impact of phosphorylation on protein digestibility. Matheis et al. ⁽Reference Matheis, Penner and Feeney⁵⁷⁾ used a tetrahymena bioassay to show that a broth containing phosphorylated casein reduced the growth of T. thermophili by 20 % in comparison to a broth containing proteose peptone (animal protein hydrolysate) suggesting that with T. thermophili, at least, there was a lower utilisation of phosphorylated amino acids compared to their unphosphorylated counterparts.

To summarise, significant amounts of phosphothreonine may be present in some processed food proteins. The presence of phosphothreonine probably does not significantly affect protein digestibility but may impact on threonine utilisation if the phosphorylated threonine cannot be dephosphorylated in the brush border at a rate sufficient to support PDV protein synthesis.

Tryptophan

Tryptophan can also be modified during processing, usually through oxidation, to a range of products depending on the oxidising agents present⁽Reference Simat and Steinhart⁶⁹⁾. Weck et al. ⁽Reference Weck, Nielsen and Finot⁷⁰⁾ have shown that the tryptophan oxidation products generated after hydrogen peroxide treatment are not nutritionally available. Tryptophan can be quantitatively recovered during the hydrochloric acid hydrolysis of pure proteins⁽Reference Gruen and Nicholls⁷¹⁾, but in foods where significant amounts of carbohydrates are present tryptophan recoveries tend to be low⁽Reference Finley⁴¹⁾. Consequently, base hydrolysis is commonly used for the determination of tryptophan in foods but even then recoveries are incomplete⁽Reference Finley⁴¹⁾.

When analysing samples for tryptophan content, it is common practice to include an internal standard, usually 5-methyl tryptophan, and to correct the determined tryptophan values based on the recovery of the internal standard. This method relies on a similar loss rate for both tryptophan and the internal standard. If the loss rate of the internal standard during hydrolysis is greater than that of tryptophan, then tryptophan content will be overestimated. A disparity in loss rates between tryptophan and the internal standard may explain why ileal digestible tryptophan overestimates available tryptophan in some proteins. However, it is the experience in our laboratory that the loss rates of tryptophan and 5-methyl tryptophan are similar and if they do differ 5-methyl tryptophan degrades more slowly than tryptophan which would lead to an underestimate of tryptophan rather than an overestimate. Overall, more work is required for the accurate analysis of tryptophan in foods and digesta.

Least-squares nonlinear regression

While many amino acids are stable under the strong acidic conditions used to hydrolyse proteins for amino acid analysis, some amino acids, for example cysteine, methionine, serine, threonine and tryptophan, are destroyed either partially or totally during hydrochloric acid hydrolysis. Typically hydrolytic losses of 10–15 % for serine and 5–10 % for threonine are not uncommon after 24 h hydrolysis⁽Reference Darragh, Garrick and Moughan^72–Reference Ozols⁷³⁾. In contrast, branched chain amino acids are difficult to hydrolyse and require more than 24 h hydrolysis for complete liberation. Essentially, the hydrolysis interval of 24 h is a compromise that permits the quantitative hydrolysis and recovery of most but not all amino acids. Hydrolysis for 72–96 h has been used to determine isoleucine and valine content⁽Reference Rayner^74–Reference Rowan, Moughan and Wilson⁷⁵⁾, while using multiple hydrolysis intervals and extrapolating (linear regression) back to 0 h hydrolysis has been used to estimate serine and threonine content⁽Reference Rowan, Moughan and Wilson⁷⁵⁾. The latter approach assumes that the release of an amino acid from the protein and its subsequent degradation occurs sequentially but it is much more likely that for any given amino acid, liberated residues begin to degrade while other residues are yet to be hydrolysed. Least-squares nonlinear regression, which takes into account the simultaneous hydrolysis and destruction of amino acids has been used to estimate the amino acid contents of foods and other protein sources⁽Reference Rutherfurd, Schneuwly and Moughan^42, Reference Darragh, Garrick and Moughan^72, Reference Darragh and Moughan^76–Reference Rutherfurd⁸⁰⁾. The method involves determining the amino acid yield for samples that have been hydrolysed over multiple intervals and using least-squares nonlinear regression to estimate the amino acid content based on the following model:

$$\begin{eqnarray} B ( t ) = \frac { A _{ o }\, h \,( e ^{ - lt } - e ^{ - ht })}{ h - l } + B _{ o }\,( e ^{ - lt }) \end{eqnarray}$$ $$

Where A _o is the protein bound amino acid content, B(t) is the amino acid concentration at time t, B_o is the free amino acid concentration prior to hydrolysis, h is the hydrolysis rate (the rate at which amino acids are released from the protein), l is the loss rate (the rate at which amino acids are degraded).

In Table 3 the threonine and serine contents predicted using either least-squares nonlinear regression or determined after conventional 24 h hydrochloric acid hydrolysis are given for several protein sources. The threonine and serine content determined using 24 h hydrochloric acid hydrolysis underestimated that predicted using least-squares nonlinear regression by 3–13 % for threonine and 2–14 % for serine. The main drawback of the least-squares nonlinear regression is the time and cost to run the analyses, given that multiple (between 10 and 25) hydrolysis intervals are required. Consequently, this method is not suitable for routine amino acid analysis of foods. Where it does have a place, however, is when highly accurate amino acid compositional data are required.

Table 3 Threonine and serine contents (g kg⁻¹) of selected protein sources predicted using either least-squares nonlinear regression or determined after conventional 24 h hydrochloric acid hydrolysis

Data reported by Rutherfurd et al. ⁽Reference Rutherfurd, Moughan and Lowry⁷⁹⁾, Rutherfurd⁽Reference Rutherfurd⁸⁰⁾ and Darragh and Moughan⁽Reference Darragh and Moughan⁷⁶⁾ and Darragh et al. ⁽Reference Darragh, Garrick and Moughan⁷²⁾.

¹ The actual threonine and serine content of lysozyme based on amino acid sequence is 7 and 10 residues/mol protein⁽Reference Canfield⁸⁸⁾.

Racemisation

Racemisation is another effect of processing, particularly at high pH, leading to the formation of D-amino acids. Amino acids will racemise at differing rates depending on the electron-withdrawing ability of the amino acids side chain⁽Reference Liardon and Ledermann⁸¹⁾ with amino acids like cysteine, serine and threonine being the most affected, and the neutral amino acid valine, isoleucine and leucine being the least affected⁽Reference Liardon and Ledermann⁸¹⁾. The proteins within which the amino acids are bound also impact on amino acid racemisation rates⁽Reference Liardon and Ledermann⁸¹⁾. Negligible racemisation occurs in heat-treated proteins in the absence of alkaline treatment after heating at moderate temperatures (120°C for 1 h), and low levels ( < 3 %) of racemisation occurs for most amino acids after heating at high temperatures (230°C for 20 min)⁽Reference Liardon and Hurrell⁸²⁾. D-amino acids are generally not nutritionally available but conventional amino acid analysis determines the D- and L-amino acids collectively thereby overestimating the available amino acid content in alkaline-treated proteins. For proteins that contain significant amounts of the D-amino acid enantiomers, analytical methods that can distinguish between D- and L-amino acids are required. Such methods are discussed in detail by Rutherfurd and Sarwar-Gilani⁽Reference Rutherfurd and Sarwar-Gilani⁸³⁾.