11 thermal processing and nutritional quality

The last reactions belong essentially to four categories: • breaking and/or recombination of intramolecular or intermolecular disulfidebridges; • reactions of the basic and acidic side c

dif-as potatoes, although they are considered the bdif-ase of a balanced diet in view ofthe most up-to-date dietary recommendations, are never consumed raw.

With the exception of milk, fruit juices, and some other foods, in which a freshand natural appearance is required, thermal treatments have also relevant hedonistic consequences, as they confer the desired sensory and texture features

to foods Bread and baked products, or chocolate, coffee, and malt are well known products that are consumed world-wide; here thermal treatments producethe characteristic aroma, taste, and colour (Arnoldi, 2001) Such sensory charac-teristics have positive psychological effects that facilitate digestion and thereforecontribute to an individual’s well-being

During thermal treatment many reactions take place at a molecular level:

• Denaturation of proteins, with the important consequence of the deactivation

of enzymes that destabilise foods or decrease their digestibility, such aslipases, lipoxygenases, hydrolases, and trypsin inhibitors

• Lipid autoxidation

• Transformations of minor compounds, for example vitamins

• Reactions involving free or protein-bound amino acids

The last reactions belong essentially to four categories:

• breaking and/or recombination of intramolecular or intermolecular disulfidebridges;

• reactions of the basic and acidic side chains of amino acids to give tides (for example Lys + Asp);

isopep-• reactions involving the side chains of amino acids and reducing sugars in avery complex process generally named as ‘Maillard reaction’ (MR);

• reactions involving the side chains of amino acids through leaving group elimination to give reactive dehydro intermediates, which can produce cross-linked amino acids

The Maillard reaction is described in this chapter and some information given onthose reactions involving the side chains of amino acids The Maillard reaction,

or non-enzymatic browning, is one of the most important processes involving onone hand amino acids, peptides and proteins, and on the other reducing sugars(Ledl and Schleicher, 1990; Friedman, 1996) The MR is a complex mixture ofcompetitive organic reactions, such as tautomerisations, eliminations, aldol con-densations, retroaldol fragmentations, oxidations and reductions Their interpre-tation and control is difficult because they occur simultaneously and give rise tomany reactive intermediates

Soon after the discovery of the MR it became clear that it influences the nutritive value of foods The loss in nutritional quality and, potentially, in safety is attributed to the destruction of essential amino acids, interaction withmetal ions, decrease in digestibility, inhibition of enzymes, deactivation of vitamins and formation of anti-nutritional or toxic compounds However, while the reaction has its negative effects, the positive effects are considerablygreater

11.2 The Maillard reaction

About 90 years ago Maillard (1912) observed a rapid browning and CO2opment while reacting amino acids and sugars: he had discovered a new reactionthat became known as the ‘Maillard reaction’ or non-enzymatic browning Nine-teen years later Amadori (1931) detected the formation of rearranged stable products from aldoses and amino acids that became known as the Amadorirearrangement products (ARPs) The development of industrial food processing,especially after World War II, gave a large impulse to research in this field and

devel-after some years Hodge (1953) was able to propose an overall picture of the tions of non-enzymatic browning in a review that, after almost 50 years, remainsone of the most cited in food chemistry.

reac-The mechanism of non-enzymatic browning is generally studied in simplemodel systems in order to control all the parameters and the results are extrapo-lated to foods quite efficiently

The reactants include reducing sugars Pentoses, such as ribose, arabinose orxylose are very effective in non-enzymatic browning, hexoses, such as glucose

or fructose, are less reactive, and reducing disaccharides, such as maltose orlactose, react rather slowly Sucrose as well as bound sugars (for example glycoproteins, glycolipids, and flavonoids) may give reducing sugars throughhydrolysis, induced by heating or very often by yeast fermentation, as in cocoabean preparation before roasting or dough leavening

The other reactants are proteins or free amino acids; these may already bepresent in the raw material or they may be produced by fermentation In somecases (e.g cheese) biogenic amines can react as amino compounds Smallamounts of ammonia may be produced from amino acids during the Maillardreaction or large amounts added for the preparation of a particular kind of caramelcolouring

A very simplified general picture of the MR may be found in Fig 11.1 lowing the classical interpretation by Hodge (1953), the initial step is the con-densation of the carbonyl group of an aldose with an amino group to give an

Fol-unstable glycosylamine 1 which undergoes a reversible rearrangement to the ARP (Amadori, 1931), i.e a 1-amino-1-deoxy-2-ketose 2 (Fig 11.2) Fructose reacts

in a similar way to give the corresponding rearranged product,

Early stage

First interactions between sugars and amino groups, rearrangements

Fissions, cyclisations, dehydrations, condensations, oligomerisations

2-aldose 3 (Fig 11.3, Heyns, 1962) The formation of these compounds, that have

been separated from model systems as well as from foods, takes place easily even

at room temperature and is very well documented also in physiological tions Here long-lived body proteins and enzymes can be modified by reducingsugars such as glucose through the formation of ARPs (a process known as gly-cation) with subsequent impairment of many physiological functions This takesplace especially in diabetic patients and during aging (Baynes, 2000; Furth, 1997;James and Crabbe 1998; Singh et al, 2001; Sullivan, 1996) A detailed descrip-tion of the synthetic procedures, physico-chemical characterisation, propertiesand reactivity of the ARPs may be found in an excellent review by Yaylayan andHuyggues-Despointes (1994)

condi-Where the water content is low and pH values are in the range 3–6, ARPs areconsidered the main precursors of reactive intermediates in model systems

H

H HO

H

HO OH

OH

HC OH H H HO OH H OH H

CH2OH NR

HC OH H HO OH H OH H

CH2OH NHR

H2C O H HO OH H OH H

CH2OH NHR

O H

HO

H HO

H

H OH

OH

O OH

H

H HO

H

OH OH H

Fig 11.2 Mechanism of the Amadori rearrangement.

Below pH 3 and above pH 8 or at temperatures above 130°C (caramelisation),sugars will degrade in the absence of amines (Ledl and Schleicher, 1990) Ringopening followed by 1, 2 or 2, 3-enolisation are crucial steps in ARP transfor-mation and are followed by dehydration and fragmentation with the formation ofmany very reactive dicarbonyl fragments This complex of reactions is con-sidered the intermediate stage of the MR.

Maillard observed also the production of CO2, which is explained by a processnamed the Strecker degradation (Fig 11.4) The mechanism involves the reac-tion of an amino acid with an a-dicarbonyl compound to produce an azovinylo-gous b-ketoacid 4, that undergoes decarboxylation In this way amino acids are

converted to aldehydes containing one less carbon atom per molecule These arevery reactive and often have very peculiar sensory properties The aldehydes thatderive from cysteine and methionine degrade further to give hydrogen sulfide, 2-methylthio-propanal, and methanethiol: that means that the Strecker degradation

is responsible for the incorporation of sulfur in some Maillard reaction products(MRPs) Another important consequence of the Strecker reaction is the incorpo-

ration of nitrogen in very reactive fragments deriving from sugars, such as 5.

H

HHO

H

HNHR

OH

2-amino-2-desoxy-D-glucose 3

Heyns rearranged product

Fig 11.3 Heyns products.

C C O

O

H2N

HCCOOHR

C C O

N

HCCOOHR

C C OH

N CH R

C C OH

NH2

O CH R

NH3 CHC OH O

+

+

4

Strecker aldehyde

5

Fig 11.4 Mechanism of the Strecker degradation of amino acids.

However, in the last two decades other mechanisms have been proposed Forexample, starting from the experimental observation of free radical formation atthe start of the MR, Hayashi and Namiki (1981; 1986) proposed a reducing sugardegradation pathway that produces glycolaldehyde alkylimines without passingthrough the formation of ARPs (Fig 11.5).

Very recently, on the basis of extensive experiments with 13C- and 2H-labelledsugars, a detailed reaction scheme was proposed by Tressel et al (1995 and1998a): the formation of various C6-, C5-, and C4-pyrroles and furans from bothintact and fragmented hexoses and amines could be unambiguously attributed to

distinct reaction pathways via the intermediates A–C without involving the

CH-NH-RCHOHCH=OCHOHR'

+ RNH2

- H2O

reverse aldol reaction 5

Fig 11.5 Possible pathway for the formation of glycolaldehyde alkylimines proposed by

Namiki and Hayashi (1986).

CH2OH

NR

O H OH OH

CH2OH

NR

O H

OH

CH2OH

NR

O H O OH

CH2OH NHR

OH

CH2OH

H O OH OH

CH2OH

O H

CH 2 OH

NR

OH H

O

NR

O H O

CH2OH

NHR

OH H O O NHR

cyclisation cyclisation

beta-dicarbonyl route

3,4-dideoxy aldoketose route

3,4-dideoxy aldoketose route

3,4-dideoxy aldoketose route

bonyl route

Fig 11.6 Transformation of hexoses and pentoses to C 5 - and C 4 -pyrroles and -furans.

(Reproduced with permission from Tressel et al, 1998a)

Amadori rearrangement (Fig 11.6) These pyrroles and furans polymerise veryeasily to highly coloured compounds that may be involved in the formation ofmelanoidins.

By means of experiments showing that sugars and most amino acids alsoundergo independent degradation (Yaylayan and Keyhani, 1996), a new concep-tual approach to the MR has been proposed recently by Yaylayan (1997) He sug-gested that in order to understand the MR better, it is more useful to define asugar fragmentation pool {S}, an amino acid fragmentation pool {A}, and aninteraction fragmentation pool {D}, deriving from the Amadori and Heyns com-pounds (Table 11.1) Together they constitute a primary fragmentation pool ofbuilding blocks that react to give a secondary pool of interaction intermediatesand eventually a very complex final pool of stable end-products

However, most foods contain also lipids that can degrade by autoxidation(Grosch, 1987) giving reactive intermediates, mainly saturated or unsaturatedaldehydes or ketones and also glyoxal and methylglyoxal (in common with the

Table 11.1 Composition of the primary fragmentation pools

Type of pool Constituents

fragmentation pool Carboxylic acids

Aldehydes Amino acid specific side chain fragments: H 2 S (Cys), CH 3 SH (Met), styrene (Phe)

Sugar fragmentation C1 fragments: formaldehyde, formic acid

pool {S} C2 fragments: glyoxal, glycoladehyde, acetic acid

C3 fragments: glyceraldehyde, methylglyoxal, hydroxyacetone, dihydroxyacetone, etc.

C4 fragments: tetroses, 2, 3-butanedione, 1-hydroxy-2-butanone, 2-hydroxybutanal, etc.

C5 fragments: pentoses, pentuloses, deoxy derivatives, furanones, furans

C6 fragments: pyranones, furans, glucosones, deoxyglucosones Amadori and Heyns C3-ARP/HRP derivatives: glyceraldehyde-ARP, amino

fragmentation pool acid-propanone, amino acid-propanal, etc.

{D} C4-ARP/HRP derivatives: amino acid-tetradiuloses, amino

acid-butanones C5-ARP/HRP derivatives: amino acid-pentadiuloses C6-ARP/HRP derivatives: amino acid-hexadiuloses, pyrylium betaines

Lipid fragmentation Propanal, pentanal, hexanal, octanal, nonanal

pool {L} 2-Oxoaldehydes (C6–9)

C2 fragments: glyoxal C3 fragments: CHOCH 2 CHO, methylglyoxal Formic acid, acids

Maillard reaction) and malondialdehyde (Table 11.1) These belong to a fourthpool, the lipid fragmentation pool {L} (D’Agostina et al, 1998) and in this waythe scheme proposed by Yaylayan (1997) was revised to include it (Fig 11.7).Clear interconnections between the MR and lipid autoxidation have been exten-sively studied in the case of food aromas, where many end-products deriving fromlipids and amino acids or sugars are very well documented (Whitfield, 1992), butcertainly they may be relevant also for other sensory aspects, such as colour ortaste, or for nutrition, although these research areas have been almost completelyneglected until the present.

Depending on food composition and heating intensity applied, thousands ofdifferent end products may be formed in the advanced stage of the MR: they areclassified here according to their functions in foods (Fig 11.8) Very volatile com-pounds, such as pyrazines, pyridines, furans, thiophenes, thiazoles, thiazolines,and dithiazines are of interest, when considering aroma; some low molecularweight compounds relate to taste (Frank et al, 2001; Ottiger et al, 2001), othersbehave as antioxidants and a few are mutagenic Polymers (melanoidins) that insugar/amino acid model systems and some foods such as coffee, roasted malt, orchocolate are the major MRPs, and determine the colour of the food

This review will discuss only the mechanism of formation of MRPs that havesome nutritional significance or may be used as molecular markers for quantify-ing the MR in foods A very detailed description of the pathways leading to mostMaillard reaction products may be found in an excellent review by Ledl andSchleicher (1990)

11.3 Nutritional consequences and molecular markers of the Maillard reaction in food

As the MR involves some of the most important food nutrients, its nutritionalconsequences must be carefully considered Researcher attention has previouslybeen focused mainly on milk and milk products, where thermal treatments arenecessary for obtaining microbial stabilisation and the preservation of high nutritional quality Vegetable products, which become edible only after thermaltreatments, have been relatively neglected so far

The degradation of sugars per se is never considered a problem because it is

only rarely they are lacking in diet However, free or protein-bound essentialamino acids may be damaged irreversibly; the amount of free amino acids in food

is always very low and they are important as constituents of proteins This meansthat the most relevant nutritional effect of the Maillard reaction is non-enzymaticglycosylation of proteins which involves mostly lysine, whose bioavailabilitymay be drastically impaired This should be distinguished very clearly from enzy-matic glycosylation, a normal step in the biosynthesis of glycoproteins, in whicholigosaccharides are bound to serine or asparagine through a glycosidic bond.The first glycation products are then converted to the Amadori product, fruc-tosyllysine, that eventually can cross-link with other amino groups intramolecu-larly or intermolecularly The resulting polymeric aggregates are called advancedglycation end products (AGEs)

Lysine availability is an important nutritional parameter especially in foodsfor particular classes of consumers, such as infant formulas (Ferrer et al, 2000) Statistically significant losses of available lysine (about 20%) with respect to raw milk have been reported as a consequence of the thermal treatment applied

in the preparation of these foods

Because the reactions of lysine are so relevant in nutrition, over a period oftime different MRPs have been proposed as markers of protein glycosylation

Fig 11.8 Functional classification of Maillard reaction products.

Fructosyllysine is unstable in the acid conditions of protein hydrolysis, ing about 30% furosine, pyridosine (a minor cyclisation product) and about 50%lysine (Fig 11.9) Furosine was first detected in foods by Erbersdobler and Zucker(1966) and can be easily analysed by HPLC: thus furosine quantification is con-sidered a good estimate of nutritionally unavailable lysine Milk proteins, owing

produc-to their nutritional relevance, have been considered with particular attention.Owing to the presence of lactose, the Amadori compound in this case is lactulo-syllysine and furosine is again a useful marker of lysine unavailability For thisreason several authors have used the furosine method for determining theprogress of the Maillard reaction in different foods (Chiang, 1983; Hartkopf andErbersdobler, 1993 and 1994, Henle et al, 1995; Resmini et al, 1990) However,today very powerful analytical techniques are disclosing new possibilities, permitting, for example, the direct determination of fructosyllysine (Vinale

et al, 1999) by the use of a stable isotope dilution assay performed in liquid chromatography – mass spectrometry (LC–MS) This method overcomes theproblems of hydrolytic instability of the analyte and the incompleteness

of the enzymatic digestion technique

Other possible markers of lysine transformation are N-e-carboxymethyllysine(CML) and 5-hydroxymethylfurfural (HMF) (Fig 11.10) CML was detected forthe first time in milk by Büser and Erbersdobler (1986) and an oxidative mech-anism was proposed for its formation (Ahmed et al, 1986) The formation of HMF

in foods has been explained in two ways: via the Amadori products through lisation (in the presence of amino groups) and via lactose isomerisation and degra-dation, known as the Lobry de Bruyn-Alberda van Ekenstein transformation(Ames, 1992) Because of this, it has recently been proposed to measure sepa-rately the HMF formed only by the acidic degradation of Amadori products and

eno-CH2

OHHO

OHH

OHH

H

C NH2HOOC

O

H2

C NH

(CH2)4O

COOH

NH2

N

H3C(CH2)4

O

NH2COOH

directly related to the MR, called bound HMF, and total HMF that also derivesfrom the degradation of other precursors (Morales et al, 1997) This method isconsidered more reliable than the previous spectrophotometric one of Keeney andBassette (1959).

Many authors have observed that the values of furosine, carboxymethyllysine,and HMF are very well correlated (Corzo et al, 1994; O’Brien, 1995) Hewedy

et al (1991), however, comparing several damage indicators for the classification

of UHT milk have shown that carboxymethyllysine is suitable only for ing very severe damage, because it is formed only in very small amounts, whereasfurosine and HMF have a more general applicability e-Pyrrolelysine (known also

monitor-as pyrraline, Fig 11.10) is another substance that hmonitor-as been proposed to memonitor-asurethe MR in foods It was observed for the first time in the reaction between glucoseand lysine (Nakayama et al, 1980) and is particularly useful in dry foods because

it is very stable: for example Resmini and Pellegrino (1994) have proposed amethodology for measuring protein-bound pyrraline in dried pasta The forma-tion of this MRP parallels very well the formation of furosine Another usefulsubstance for assessing protein damage is lysinoalanine (see section 11.7)

11.4 Melanoidins

The modifications of amino acids described in the preceding section take placealso during either mild treatments or short duration treatments at high tempera-

NHOH2C CHO

CH2COOH

N

NNHHN

tures, where the changing of food appearance is hardly perceptible However,many processes usually dealing with the processing of vegetables, such as breadbaking, roasting of coffee and nuts, and kiln drying of malt as well as roasting

of meat require severe thermal treatments In these cases the Maillard reaction isresponsible for colour formation, a very critical parameter in determining foodquality The most important contribution to colour comes from polymers, known

as melanoidins that in some foods are the major MRPs Their structure is stillelusive but many methodological difficulties have slowed down the progress ofknowledge in this field Some attempts have been made to isolate melanoidinsfrom such foods as soy sauce (Lee et al, 1987), dark beer (Kuntcheva andObretenov, 1996), malt or roasted barley (Milic et al, 1975; Obretenov et al,1991), coffee (Maier and Buttle, 1973; Steinhart and Packert, 1993; Nunes andCoimbra, 2001), but their very complex and probably non-repetitive structure haslimited structural characterisation

Studies on model systems have clearly indicated that reducing carbohydratesand compounds possessing a free amino group, such as amino acids, either infree or protein-bound form, are the basic material for their formation (Ledl andScleicher, 1990) They are reported to have molecular weight up to 100 000 Da,and to possess structural features, elemental analysis and degrees of unsaturationdepending on the reaction conditions

As already indicated at the beginning of this review, the Maillard reaction isoften studied through model systems; however, in the case of melanoidins thisapproach shows limitations Many authors have used model systems to isolateand characterise low molecular weight coloured compounds (Rizzi, 1997; Arnoldi

et al, 1997; Ravagli et al, 1999; Tressl et al, 1998a and 1998b; Wondrack et al,1997) Currently the most active in this field are Hofmann (Hofmann, 1998a, b,

c, d) and co-workers (Hofmann et al, 1999) However, it has been clearly strated that completely different compounds are produced by reacting glucosewith single amino acids or b-casein (Hofmann, 1998c) With amino acids themajority of coloured compounds have molecular weights below 1000 amu,whereas the reaction between glucose and casein gives rise to products with muchhigher molecular weights (Hofmann, 1998a) Moreover, only very rarely has itbeen demonstrated that the chromophores isolated from amino acids/sugarsmodel systems may be incorporated in melanoidins, as in the case of the freeradical chromophore named CROSSPY (Fig 11.11), which was successfullyidentified in several processed foods, such as coffee and bread crusts by EPR(Hofmann et al, 1999)

demon-However, the formation of melanoidins in real foods does not involve simplyproteins and sugars: biopolymers (proteins in animal proteins, both proteins andpolysaccharides in vegetables) probably act as a non-coloured skeleton bearing

a variety of chromophoric substructures Thus the melanoidins separated fromcoffee brews contain about 33% polysaccharides, 9% proteins, and 33% polyphe-nols (Nunes and Coimbra, 2001) The polyphenol substructures derive fromchlorogenic acids (Heinrich and Baltes, 1987) that disappear during coffee roasting Physico-chemical properties of melanoidins other than colour may be

important in foods For example, in espresso coffee brew, they may have foaming

properties (Petracco, 1999) that stabilise the foamy layer on top of the beverage,

which is well known by the Italian term crema.

Beside their obvious importance in determining the brown colour of roastedfoods, in recent years great interest has developed around melanoidins for theirpossible role in physiology and in food stabilisation: in fact some authors (Anese

et al, 1999; Nicoli et al, 1997) have demonstrated that they possess antioxidativeactivity Antioxidants are able to delay or prevent oxidation processes, typicallyinvolving lipids, which greatly affect the shelf-life of foods In coffee, the antiox-idant activity (attributed to the development of Maillard reaction products)increases with roasting up to the medium-dark roasted stage, then decreases withfurther roasting (Nicoli et al, 1997), possibly indicating a partial decomposition

in the antioxidative compounds

11.5 Transformations not involving sugars: cross-linked

amino acids

Another kind of transformation of the side chains of protein-bound amino acids induced by thermal treatments of foods, particularly in basic conditions,

is the formation of cross-linked amino acids A dehydroalanine residue may be

formed through elimination of a leaving group from protein-bound serine, phosphorylserine, O-glycosylserine, or cystine, and may undergo Michael

O-addition by another nucleophilic amino acid residue (Fig 11.12) For examplethe e-amino group of lysine may react to give a secondary amine, which is nor-mally indicated with the trivial name of lysinoalanine (LAL) (Maga, 1984) Anal-ogous reactions may involve ornithine to give ornithinoalanine (OAL), cysteine

to give lanthionine (LAN), and histidine to produce histidinoalanine (HAL)

NNprotein

protein

Fig 11.11 Structure of the radical cation CROSSPY.

(Finley and Friedman, 1977) Both nitrogens of histidine may react, giving rise

to the regioisomers Np-HAL and Nt-HAL (Fig 11.13) (Henle et al, 1993).The formation of cross-linked amino acids does not involve the participation

of reducing sugars and is particularly extensive when proteins are submitted toaqueous alkali treatments Such treatments include those used in the preparation

of soy protein concentrates and in the recovery of proteins from cereal grains,milling by-products, and oilseeds, such as cottonseeds, peanuts, safflower seedsand flaxseeds, and in the separation of sodium caseinate Other alkali proceduresare commonly used for destroying microorganisms, preparing peeled fruits, andinducing fiber-forming properties in textured soybean proteins, used, for example,

in the preparation of meat substitutes A recent review lists the processes and foods that have been studied for the formation of cross-linked amino acids(Friedman, 1999) The main feature of these compounds is that they are stableduring acidic protein hydrolysis and are relatively easy to analyse when other

H2C C

HProteinNHO

ProteinZ

H2C C ProteinHN

OProtein

N-tau-HAL

CH

H2N COOH

Fig 11.13 Main cross-linker amino acids.