Revels: an pr 2020. | Reis: ly Canpreteasie ABIES Wiey COMPREHENSIVE REVIEWS IN FOOD SCIENCE AND FOOD SAFETY Impact of wheat endogenous lipids on the quality of fresh bread: Key terms, concepts, and underlying mechanisms SaraMelis® | Jan A. Delcour KU Leuven, Laboratory of ood Chemistry and Biochemiszy and Leuven Food Science and Nutrition Research Centre (LFoRCe), Leuven, Belgium Abstract While wheat (Triticum aestivum L.) flour contains only low levels of lipids (2.0% t03.0%), they tremendously affect fresh bread quality. They are either starch (30% to.40%%) ornonstarch (60% to 70%) lipids. While the former areimportantin bread staling, they affect neither bread loaf volume nor crumb structureas they are only set free at the very end of the baking process and prior to that they are not avail- able because of their location inside starch granules. This review mainly focuses ‘on wheat nonstarch lipids and how they impact on bread quality. Traditional ‘approaches for investigating their role in bread making have been to use flours varying in bread making quality, to defat and reconstitute flour with (fractions of) the extracted lipids and/or to supplement flour with lipids from wheat or other sources. More recently, lipases have been successfully applied to investigate how ‘wheat lipids affect bread making. It is generally accepted that their impact on bread loaf volume and crumb structure largely if not entirely relates to that on bread dough gas cells stability. However, today there are still different views and hypotheses on the mechanistn(s) whereby they impact fresh bread quality. This review first defines and introduces the key terms, concepts, and theories related to lipids, lipases, and bread making, Next, the effects that wheat endogenous lipids and their enzymatically released hydrolysis products have on fresh bread. properties and the mechanisms whereby they exert these effects are reviewed. Correspondence ‘Sara Melis, Kaste)park Arenberg20(box 2463), B-3001 Leven, Belgium. mal: sara. meliserkuleaven.be Fundinginformation Agentschap voor Innovatie dar We schapen Technclgie, GraatAward Num- ber WEMI795 1 | INTRODUCTION ing bread making, and therefore, have great potential as targets for improving bread quality Bread isa staple food in many parts of the world. In Europe, ‘America, and Australia, the average per capita consump- tion in 2019 stood at 30.3, 20.0, and 18.7 kg, respectively 2 | (Statista, 2020). Bread not only provides energy (primarily through starch)but also contains protein, dietaryfiber,and = 2.1 | awide array of vitamins and minerals (Delcour & Hoseney, LIPIDS Definition and classification 2010). Eating bread thus helps consumers to reach their everyday needs of many nutrients, Consumers evidently expect bread to be of high sen- sory quality, Bread quality is not only affected by major ‘wheat flour constituents but also by minor ones such as lipids. Wheat flour lipids indeed play important roles dur- Lipids are an extremely diverse group of natural com- pounds for which no widely accepted definition exists. ‘They exhibit greater structural variety than other classes of biological molecules (such as proteins and polysac- charides) and are similar only in that they are largely hydrophobic and only sparingly soluble in water (Voet, © tia jeatae offend Techaslog@® | a7ame | REVIEWS @) o e) wk, xd LA, ° a as FIGURE 1 ud i Structures of tripalmitn (a), linoleic acid (b), a digalactosydiaeylglycerol (6), an N-acyl phosphatidylethanolamine (), anda Iysophosphatidylcholine (e), with RI-6 random esterified fetty acids. Source: htps://wwwlipidmeps.org/ Voet, & Pratt, 2016). Christie in 1987 put forward the following definition: “Lipids are fatty acids (FAs) and their derivatives, and substances related biosynthetically or functionally to these compounds.” This definition to date is still used by the American Oil Chemists’ Society (Christie, 2020; Christie & Han, 2010) andi adopted in this review. Lipids of plant and animal origin are mainly FAs linked by ester bonds to glycerol or to other alcohols (such as cholesterol) or by amide bonds to long-chain bases (sph- ingoids or bases thereof) or, occasionally, to other amines. In addition, some lipids contain moieties such as phospho- rous groups, organic bases, and/or carbohydrates in their structure (Christie & Han, 2010). Based on the distinct hydrophobic and hydrophilic elements that they contain, lipids have been divided into eight categories: FAs, gle erolipids, glycerophospholipids, sphingolipids, sterol lipids, prenol lipids, saccharolipids, and polyketides (Fahy et al., 2008). The majority of wheat lipids belong to the first three categories (Chung, Ohm, Ram, Park, & Howitt, 2009; Mor- rison, 1978). Each lipid category includes distinct classes and subclasses (Fahy et al., 2005). Many of these contain numerous lipid species that vary in the (esterified) FAs which they contain, For example, tripalmitin (Figure 1a) and trilinolein both are triacylglycerols (TAGs) and are thus glycerolipids. FAs are the main building blocks of lipids. They are carboxylic acids that as part of their structure contain an aliphatic chain that is either saturated (containing no dou- ble bonds) or unsaturated (containing one or more dou- ble bonds). Most natural FAs contain an even number of carbon atoms ranging between 10 and 22 (Christie & Han, 2010; Fahy et al, 2005; Gunstone, Harwood, & Difk- stra, 2007). The most common FA in wheat is linoleic acid (€18:2) (Chung et al, 2009; Morrison, 1978). Itis shown in its nonesterified form in Figure 1b. Nonesterified or free FAs (FFAs) are rather uncommon in nature and FAs are mostly esterified such as in glycerolipids. Glycerolipids contain all lipids with a glycerol back- bone. The most prevalent glycerolipids are acylglycerols.Here, the glycerol backbone is esterified to one, two, or three FAs resulting in monoacylglycerols (MAGS), diacyl- glycerols (DAGs), or TAGs, respectively (Christie & Han, 2010; Fahy et al., 2005; Gunstone et al, 2007). Next to the acylglycerols, also glyceroglycolipids belong to the glycerolipids. Here, at least one sugar residue is con- nected via a glycosidic bond with the sn-3 position of the glycerol backbone (Fahy et al., 2005). Most glycero- glycolipids in wheat are galactolipids. Indeed, the suger residues are almost exclusively galactose units (Christie & Han, 2010). This terminology is adopted in this review. Wheat galactolipids include monogalactosyldiacylglyc- erols (MGDGs) and digalactosyldiacylglycerols (DGDGs) (Figure 1c) (Morrison, 1978; Pareyt, Finnie, Putseys, & Deleour, 2011) Glycerophospholipids consist of a glycerol-3-phosphate backbone esterified to one or two FAS at the sn-l and/or sn-2 position(s). They are frequently and also here referred to as phospholipids. The phosphate group ean be linked to a polar group such as ethanolamine, choline, or serine (Christie & Han, 2010; Fahy et al., 2005). When only one estetified FA is present (usu- ally at the sn-1 position), the lipid is referred to as a lysophospholipid (Gunstone et al., 2007). Wheat phospho- lipids include N-acyl phosphatidylethanolamines (NAPEs) (Figure 1c), phosphatidylcholines (PCs), and lysophos- phatidylcholines (LPCs) (Figure le) (Morrison, 1978; Pareyt et al., 2011). Additionally, lipids are subdivided into “simpl “complex” lipids based on the number of products released ‘upon hydrolysis. Simple lipids consist of one (when they are FFAs) or two different types of structural moietiessuch as FFAS, acylglycerols, and sterols. Hydrolysis of the lat tergives two product types. Complex lipids contain more than two types of structural moieties such asis the case for phospholipids and galactolipids. They yield three or more product types upon hydrolysis (Chung et al, 2009; Fahy et al., 2005; Pareyt et al, 200). More commonly, lipids are classified a5 “nonpolar” and “polar.” However, the distinction between these two ‘groups is not always clear and can even be misleading (Christie & Han, 2010; Chung etal, 2008). FFAs, forexam- ple, are frequently considered to be nonpolar lipids, regard- less their slightly polar nature (Chung et al., 2009). FFAS and acylglycerols such as MAGs, DAGs, and TAGs are here considered to be nonpolar lipids in line with common practice in the cereal research field. Examples of polar lipids are phospholipids and galactolipids. They account for the majority of polar lipids in wheat (Chunget al.,2009; Morrison, 1978; Pateyt et al, 2011). They are amphiphilic and therefore have surface-active properties (Christie & Han, 2010), and fees | Lipids constitute 2.0% to 3.5% of wheat kernels and are unevenly distributed over the different structural parts (Chung et al., 200%; Morrison, 1978). About 35% to 45% ‘occur in the endosperm, 30% to 36% in the germ, 25% to 29% in the aleurone, and only a minor portion (<4%) in the pericarp (ITargin & Morrison, 1980; Morrison, 1978). ‘The majority of the lipids in germ, aleurone, and peri- carp tissues are nonpolar. They include TAGs, DAGS, MAGS, and FFAs (Figure 2a). TAGsare the most dominant types (Hargin & Morrison, 1980). They are typically stor- age lipids and present in plant seeds in separate oil bodies or spherosomes with sizes ranging between 0.2 and 2.5 4am (Figure 2b). Oil bodies are generally spherical but can have irregular shapes in mature seeds if they are compressed by other cell structures. They have a core of TAGs surrounded by a phospholipid monolayer. Oleosins are low molecu- lar weight (15 to 26 kDa) basic proteins that are embed- ded in and bulge from this monolayer (Huang, 1992; Tzen & Huang, 1992). In wheat oil bodies, almost two-thirds of the total phospholipids are PCs. The remainder are com- parable portions of phosphatidylethanolamines and phos- phatidylinositols (lelsema, Morré, Ruddat, & Turner, 1977). Wheat germ and aleurone tissues contain larger num bers of oil bodies than subaleurone and endosperm tissues (Hargin & Morrison, 1980). Lipids in the endosperm are subdivided into starch internal and nonstarch lipids. Some sources mention a third category, that is, starch surface lipids, which is made up by nonstarch lipids firmly adsorbed onto or inside starch granules during starch synthesis (Chung et al.,2009; Morrison, 1981, 1988). However, starch surface lipids espe- cially in recent publications and also here are considered to be part of the starch lipids rather than belonging to a separate category. Starch lipids are located inside the starch granules as their name implies (Chung et al., 2009; Morrison, 198!). They consist almost exclusively (89% to 94%) of phospho- lipids (Figure 2a) and, more in particular, of lysophos- pholipids. The latter contain very high levels of LPCs and lower levels of lysophosphatidylethanolamines and lysophosphatidylglycerols. Other predominantly present lipids are FFAs (Hargin & Morrison, 1980). Starch lipids in native starch granules (likely) form a complex with amylose, Such amylose-lipid complexes contain FAs of, the monoacyl lipids in the hydrophobic cavity of amy- lose helices (Morrison, 1981, 1988; Morrison, Law, & Snape, 1993), Nonstarch lipids in the endosperm comprise nonpolar lipids (33.2% to 47.4%) as well as comparable portions of galactolipids and phospholipids (Figure 2a) (Hargin 2.2 | Lipids in wheat kernelsow | ) ‘ALEURONE 400 100% iis armoss | Nonpolar lipids 72310890% Galactolpids 2219 98% Phospholipids raat 17a% | PERICARD eatouen Lids 19% Nonpolaripis 96.1% Galactolnis 71% Phospholipids 6% 1 cenw aswsox | Lids as7ms05% Nonpolaripis 7oatme4e% Galactolpids 0012 95% Phospholipis 196101724 mm FIGURE 2 WHEAT ENDOGENOUS LIPIDS IN BEEAD MAKING. 707 to oas% ENDOSPERM | Starch lipids Nonpoar| 08m 1.2% s9%059% 12067% Galactolipids Phospholipid 00.40 04.9% Non-starchlipids 0.9 101.1% Nonpolarlipids s.2047.4% | Golactolipids ids 21.91095.5% rostoonane | (@) Lipid disteibution in wheat. Values in bold represent the relative weight ofthe kemel tissues and are expressed as of ‘whole kernel dey matter. Values in italics represent lipid levels in the kemel tissues (on dry matter base). Other values represent portions of specific ips in the kerne tissues expressed as % of total lipids present. Al data originate from four cultivars except for lipid levels present in pericarp tissues for which the data are based on only one cultivar. Figure constructed with data from Hargin and Morrison (1980).(b) Model of 2 maize ol body. igure adapted from Ten and Huang (1992) & Morrison, 1980). ‘The nonpolar lipids are predomi- nantly TAGS originating from oil bodies located in sub- aleurone and endosperm cells (Hargin & Morrison, 1980) Polar lipids are mainly DGDGs, MGDGs, NAPEs, N- acyl lysophosphatidylethanolamines (NALPES), and PCs (Finnie, Jeannotte, & Faubion, 2009; Hargin & Morri- son, 1980). Most of these are remainders of lipid bilay- ers surrounding amyloplasts, that is, organelles synthe- sizing and storing starch during kernel development. ‘These amyloplast membranesare (partially) degraded dur- ing seed desiccation (Hargin & Morrison, 1980; Tan & Morrison, 1973). The original locations of NAPEs and NALPEs are unfortunately not known. These polar lipids are presumably distributed throughout the kernel (Pareyt etal,, 2010). Overall, the relative quantities of lipids in an entire ‘wheat kernel decrease in the order nonpolar lipids, phos- pholipids, and galactolipids. The nonpolar lipids are mainly TAGs with lower levels of other lipids such as PAGS, MAGs, and FFAs. They mainly occur in the germ and aleurone tissues. Phospholipids are mainly NAPEs, NALPEs, PCs, and LPCs and occur in the endosperm and especially in starch granules, where high levels ‘of lysophospholipids (mainly LPCs) are (presumably) present as amylose-lipid complexes. Galactolipids, that is, the smallest wheat kernel lipid fraction, are mainly DGDGs and MGDGs and occur almost exclusively in the ‘endosperm (Finnie et al., 2009; Hargin & Morrison, 1980; Morrison, 1978 1981). The level and composition of ana- lyzed wheat lipids not only depend on growing environ- ‘ment and genetic background of the wheat but also on extraction conditions (including applied solvents, temper- atures, and sample moisture contents) and quantification ‘methodology (Chung et al., 2009; Pareyt etal, 2011) 2.3 | Lipids in wheat flour Upon milling wheat into flour, kernels and kernel pieces are broken. Next, they are further reduced in size andini |» Total wheat flour lipids (2.0 12.5% of our) Non-starch lipids (60%) Starch lipids (40%) Free lipids (35%) |_| Bound lipids (25%) Polar =r 26%) | on) TAGs peDGs Lees DAGs Mapes LpGs MAGs NAPES Les FFAS Pes FIGURE 3 Classification of wheat lourlipids based on sequent J extraction of fre, bound, and starch lipids. Light gray and dark graybbars indicat lipids extracted with nonpolar (eg, hexane ether, o petroleum ether) and polar (eg., water-saturated butan-1-ol or isopropanol-water [90:10)) solvent, respectively: When the bars have 2 fll or a dotted line, lipids are extracted at room or elevated (90 to 100 °C) temperatures, respectively: Reproduced with permission from Pareyt tal (201!) Abbreviations: DAGs, diacylglyeerols; DGDGs, digalactosyldiacylglycerols; FFAs, frce fatty acids, LPCs, lysophosphatidylcholines; LPs, Iysophosphatidylethanolamines; LPGs, Iysophosphatidyiglycerals, MAGS, monoacylglycerols, MGDGs, monogalactosyliacylglycerols; [NAPEs, N-acylphosphatidylehanolamines; PCs, phosphatidylcholines; TAGS, triacyiglycerols sieved to separate the endosperm from the bran (ie. peri- ‘carp and aleurone) and the germ (Posner, 2009). Lipids in the resultant flour are not only those of the endosperm, but they also include a portion of those of the aleurone and germ. Approximately 50% of the ‘TAGs in flour orig- inate from the germ (Hargin, Morrison, & Fulcher, 1980; Morrison & Hargin, 1981). Hence, while the ratio of non- polar to polar nonstarch lipids in endosperm is about 1: it approaches 1:1 in flour (MacMurray & Morrison, 1970). It follows that flour lipid level and composition depend on ‘wheat milling practices such as those leading to differences in flour extraction rate (Chung etal, 200°). ‘Wheat flour typically comprises 2.0% to 3.0% li These ate 60% to 70% nonstarch and 30% to 40% starch, lipids (Figure 3). Nonstarch lipids consist of appro mately 60% free and 40% bound lipids. Free lipids can be extracted at room temperature with nonpolar solvents such a8 hexane, ether, or petroleum ether. After prior removal of free lipids, bound lipids can be extracted at room temperature with more polar solvents like water- saturated butan-l-ol (WSB). After removal ofall nonstarch lipids, starch lipids are typically extracted at 90 to 100 °C with polar solvents like WSB or isopropanol:water (90:10) (Chungetal, 2009; Morrison, 1978; Morrison, Mann, Soon, & Coventry, 1975; Pareyt etal, 2011) ‘The extractability of lipids in different solvents not only relates to the match with their polarities but also to inter- actions between lipids and other flour components. Anon- polar solvent such as hexane can undo hydrophobic inter actions and is able to extract free lipids that thus either do not or do interact hydrophobically with other flour compo- nents. More polar solvents such as WSB break ionic inter- actions and/or by hydrogen bonds that are binding forces ‘occurring between bound lipids and flour components. Alternatively, bound lipids may be physically entrapped in a matrix, The presence of water as a low molar volume component allows a solvent such as WSB to penetrate the ‘matrix better and extract such lipids (Carr, Daniels, & Fra- zier, 1992; Wootton, 1966). Nonstarch lipids in wheat flour are S1% to 63% nonpo- lar lipids, 22% to 27% galactolipids, and 13% to 23% phos- pholipids. The free lipid fraction consists of about 75% nonpolar and 25% polar lipids, while the bound lipid frac- tion accommodates almost exclusively polar galactolipids and phospholipids (Figure 3). Nonstarch wheat flour lipids are mostly TAGs (21% to 47%), DGDGs (13% to 17%), MGDGs (5% to 6%), NAPEs (4% to 5%), and PCs (4% to 6%) ‘Their major FAs are linoleic ($0% to 65%), palmitic (19% to 26%), and oleic (10% to 21%) acids (Blaszezak, Fornal, Amarowicz, & Pegg, 2003; Chung et al., 2009; MacMur- ray & Morrison, 1970; Morrison, 1978; Morrison etal, 19 Pomeranz, Chung, & Robinson, 1966). Starch lipids in wheat flour are mostly phospholipids (85% to 91%) of which about 85% are LPCs. Their FA com- position differs from that of nonstarch wheat flour lipids in that they contain less unsaturated (38% to 48% linoleic, and 8% to 12% oleic acids) and more saturated (39% to 45% palmitic acids) FAs (Blaszczak et al, 2003; Chung et al, 2009; MacMurray & Morrison, 1970; Morrison, 1978; Mor- rison etal, 1975; Pomeranz.et al., 1966).smo | evens FIGURE 4 (@) Rhigomucor michel lipase. The () Enzyme unt o 1 2 Saturation the closed conformation (black) prevents the substrate (ethyl phosphonate) from entering. The open lid conformation (gray) allows access tothe active site, Reproduced with permission from Casas-Godoy etal (2012) (0) [Bnayme activity asa function of substrate level for porcine pancreatic lipase (ctcls) and horse liver esterase triangles), Substrate level is from Ali etal. (2012) and based on da 3 | LIPASES 3.1 | Classification and definition Lipases afler peptidases and carbohydrases are the third largest group of commercial enzymes. They ate widely applied in the food, detergent, and fine chemical indus- tries (Casas-Godoy, Duquesne, Bordes, Sandoval, & Marty, 2012). Lipases are hydrolases (EC 3) acting on ester bonds (BC 31) (hitps://www:brenda-enzymes.org). The term “lipases” is ambiguous. Indeed, itis often used to specify ‘TAG acylhydrolases (EC 3.11.3) that catalyze both hydrol- ysis and synthesis of (long chain) TAGs, DAGs, and MAGs (Andualema & Gessesse, 2012; Casas-Godoy et al., 2012) However, the term is equally often used to cover the com- plete range of lipases without implying a specific selectiv- ity Gerits, Pareyt, Decamps, & Delcour, 2014). The latteris also the case in the present review. Within the EC subclass of the carboxylic-ester hydro- Iases (EC 3.11), lipases used to be distinguished from car boxylesterases (EC 3.1.1) based on having an “interfa cial activation” feature. When positioned at an interface between an immiscible organic phase (containing the sub- strate) and an aqueous phase (containing the enzyme), ‘most lipases are activated and swiftly exert their catalytic action. Interfacial activation originates from the presence of a hydrophobic oligopeptide surface loop acting as a lid that covers the active site of most lipases. Without an essed in multiples of saturation, The dashed line epresents the solubility limit ofthe substrate triacetin, Figure reproduced with permission from Sarda and Desnuelle (1958) interface, this lid blocks the entrance to the active site and renders the enzymes inactive. In the presence of a lipid/water interface, the active site is uncovered due to a change in the conformation of the enzyme. This increases, the hydrophobicity of the catalytic site and allows the substrate to enter (Figure 42) (Brady et al., 1990; Brzo- ‘owski et al., 1991; Casas-Godoy et al., 2012; van Tilbeurgh et al., 1993). Due to this unique feature, lipase acti ity drastically increases when the substrate concentration ‘exceeds the critical micelle concentration. Asa result, most lipases do not follow Michaelis-Menten kinetics and sig- ‘moidal curves are obtained when reaction rates are plot- ted against substrate concentration (Figure 4b). In con- trast, carboxylesterases hydrolyze substrates dissolved in ‘an aqueous phase (Chapus, Semeriva, Bovier-Lapierre, & Desnuelle, 1976; Gerits, Pareyt, Decamps, et al., 2014; Sarda & Desnuelle, 1958; Verger, 1997). ‘Since some lipases do not exhibit interfacial activation, it is better to distinguish them from carboxylesterases by the nature of their substrate rather than by the fea- ture of interfacial activation. In this context, lipases have bbeen described as enzymes that can release carboxyl esters from long-chain acylglycerols (210 carbon atoms) and carboxylesterases as enzymes that hydrolyze carboxyl esters in short-chain acylglycerols (<10 carbon atoms) (Casas-Godoy et al., 2012; Verger, 1997). Moreover, lipases catalyze a broad range of synthesis reactions (ie., esterifi- cation and transesterification) at low water activity (Casas- Godoy et al, 2012). However, such conditions are not@ (b) ° oO tay, ott lh Lo’ ‘r, rita “OH OH FIGURE 5 Sites in wiayllycerol (TAG) for hydrolysis (X) by TAG lipases (a, in phospholipid fr hydolsis by phospholipases with indication ofthe diffrent phospholipase types: A, (Ki, (9, CRY, and D (% 4 (6) andi golactlipid for hydrolsishygalactoipases (©). Ria represent random esterified fatty acids. Figure adapted from Geri, Pareyt, Decamps, etal. (2014) relevant in the context of bread making since the water activity of dough varies between 0.96 for mixed and 0.98 for proofed dough (Cauchajowska, Pomeranz, & Jeffers, 1989). It is furthermore useful to mention here that lipases such as phospholipases C and D do not exhibit carboxylic-ester (EC 3.11) but phosphoric-diester (EC 3.14) hydrolase activity (https://wwvw.brenda-enzymes.org). Given the above, for the purposes of this review, the following definition is put forward: “Lipases are lipid degrading enzymes catalyzing the hydrolysis of (phosphodijester bonds of glycero(phospho)lipids.” The reactions are catalyzed at the interface of @ bipha- sic system consisting of both water and an immisci ble organic phase (containing the hydrophobic glyc- ero(phospho)lipids) (Casas-Godoy et al., 2012) 3.2 | Selectivity Lipase selectivity or specificity relates to their preference to catalyze specific reactions. Three types of selectivity have been identified: type-selectivity, regio-selectivity, and enantio-selectvity. ‘Type-selectivity refers to the preference of lipases for a given substrate such as TAGs, DAGs, or MAGs or a particular FA structure. This selectivity is related to the shape of the binding site of the enzymes and the amino acids composing it. Additionally, lipases can show chemo- selectivity, that is, specificity for a functional group such as an ester or phosphodiester. The term regio-selectivity Js used to refer to lipases preferentially attacking a spe- cific (phosphodiester bond in a glycero(phospho)lipid. For ester bonds, a primary and secondary ester bond is defined and regio- selectivity can be for sn-l(3) or sn-2. Most microbial lipases such as those produced by Thermomyces lanuginosus release substituents from the sn-I(3) positions of TAGs. Only a few do so from the sn-2 position. Some lipases are non-regio-specific and act randomly on TAGs. Enantio-selectivity is defined as lipase preference for 4 specific enantiomer of a chiral molecule. Enantiomers often have different biological functions, forexample, ther- apeutic versus inactive or even toxic. Therefore, enantio- selectivity is important in the pharmaceutical industry (Casas-Godoy et al., 2012) Based on their selectivity, three important lipase cate- gories have been identified. ‘The first are TAG lipases (EC 3.1.1.3). They hydrolyze ester bonds in TAGs thereby releasing DAGs and FFAs. They can, however, also act on DAGs and MAGs and further hydrolyze these lipids into FFAs and glycerol (Figure 5a). Lipases specifically acting on MAGs have been separately classified asacylglycerol lipases (synonym: MAG lipases, EC 3.11.23). ‘The second category contains the phospholipases. ‘They are divided further into phospholipases A, (BC 3.1132), A, (EC 3.114), C (EC 31.43), and D (BC 3.1.4.4) depending on the bond which they hydrolyze (Figure 5b). Phospholipase A, and A; enzymes act on ester bonds at sm-l and sn-2 positions in phospholipids, respectively. Phospholipases C and D have phosphoric diester hydrolase activity. While phospholipases C cleave glycerophosphate groups, phospholipases D detach the polar head group but not the phosphoryl group. Phos- pholipases with specificity toward lysophopholipids are separately classified as_lysophospholipases (BC3115). ‘The third group is that of the galactolipases (BC 311.26). They hydrolyze ester bonds in galac- tolipids, as their name implies (Figure Se) (Gerits, Pareyt, Decamps, etal, 2014), Itis important to recognize that lipase selectivity is often not absolute. Many lipases are indeed promiscuous and act fon a range of substrates depending on the applied con- ditions and lipase level (Casas-Godoy et a., 2012; Gerits, Pareyt, Decamps, et al., 2014; Gerits, Pareyt, & Delcour, 2014).= | ETE 3.3. | Wheat endogenous lipases Cereal lipases such as those of wheat mostly act on TAGs. ‘The main substrate of wheat endogenous lipases is tri olein. It is hydrolyzed in a non-regio-specific way (Barros, Fleuri, & Macedo, 2010). Lipasesare endogenously present in wheat with most lipase activity present in its bran and germ tissues (De Almeida, Pareyt, Gerits, & Delcour, 2018; Kapranchikov, Zherebisov, & Popova, 2004; Rose, (Ogden, Dunn, & Pike, 2008). Hence, lipase activity in white (straight-grade) flour is orders of magnitude lower than that in whole meal. Indeed, for wheat from cv. Apache, De ‘Almeida et al. (2014) could not detect any lipase activity in its white flour, whereas its whole meal had a consid- erable lipase activity. While lipase activity in white wheat flout is thus usually low enough to avoid the occurrence of enzymatic acid rancidity during storage, wheat whole ‘meal is more susceptible to such deterioration (Doblado- Maldonado, Pike, Sweley, & Rose, 2012; Rose et al., 2008; Shearer & Warwick, 1983). The release of FFAs in flour has been linked to deterioration of bread making quality (oblado-Maldonado et al., 2012; Rose et al., 2008). 4 | AIR: AN IMPORTANT “INGREDIENT” IN BREAD MAKING Bread is consumed all over the world in a wide range of sizes, shapes, textures, and tastes (Deleour & Hoseney, 2010). Many different procedures are used for its manufac- ture, all with the goal to convert wheat flour into an aer- ated and palatable food (Cauvain & Young, 2007; Deleour & Hoseney, 2010) Straight-dough bread making is the most traditional bread making procedure (Cauvain & Young, 2007; Del cour & Hoseney, 2010). Itis commonly used in many parts of Europe and also the subject of study of this review. In straight-dough bread making, all ingredients (inelud- ing air) are mixed into developed dough that is then fer- mented, molded, proofed, and baked. During fermenta- tion, dough is usually punched or remixed at least once (elcour & Hoseney, 2010) When flour is processed into bread, loaf volume (LV) and crumb structure or grain are important quality param- eters. For typical straight-dough breads such as white pan bread, high LV anda fine, uniform cell structure (i. thin ‘walled, elongated, small- to medium-sized cells) are essen- tial (Cauvain & Young, 2007; Delcour & Hoseney, 2010). Achieving such high quality requires incorporation of air during mixing and stabilization of gas cells during the sub- sequent stages of the bread making process. Indeed, the gas cells give rise to volume and the typical cellular struc- ture of wheat bread—which gives ita soft spongy texture— when fermented dough is heat-set during baking (Cauvain & Young, 2007). The void fraction, or volume fraction, of air ranges between 4% and 15% in freshly mixed dough, Increases up to 78% during fermentation and proofing, and, ranges between 58% and 70% in baked bread (Jekle, Fuchs, & Becker, 2018; Mehta, Scanlon, Sapirstein, & Page, 200% Zihiga & Le-Bail, 2009). Straight-dough bread making typically starts by mix- ing all essential ingredients, that is, wheat flour, water, yeast, and salt, and possibly nonessential ones in appro- priate ratios (Cauvain & Young, 2007; Delcour & Hoseney, 2010). During mixing, all ingredients are uniformly dis- persed to form homogeneous dough in which—as a result, of mixing—air is incorporated (Cauvain & Young, 2007) 41 | Dough formation or development Wheat flour is the major bread recipe constituent. It con- sists mainly of starch (ca. 70% to 75%), water (ca. 14%), and protein (ca. 10% to 12%) (Goesaert et al., 2005). Together with water, it forms viscoelastic dough that retains gas and is responsible for the structure of bread (Delcour & Hoseney, 2010) ‘When water is added to flour and mixing is started rapid hydration of flour particles occurs. During mixing, the particles rub against each other and/or mixer surfaces. ‘This action removes hydrated surfaces and exposes fresh particle surfaces to the excess water present. As a result, the flour particles slowly disintegrate and are hydrated. As ‘more protein and starch is hydrated by free water, the resis- tance to extension of the developing dough increases (Del- ‘cour & Hoseney, 010). Dough development essentially resulls from flour par ticles being fully hydrated (Delcour & Hoseney, 2010). The term “developed dough” is poorly defined. Bakers use it to indicate that it has all necessary physicochemical proper: ties to deliver the required bread characteristics (Cauvain & Young, 2007). On amacroscopic level, dough development can be mon- itored with a mixograph or a Farinogeaph. Both measure and record the resistance of dough to mixing. In a typi- cal mixogram (Figure 6 top) (AACCI Approved Method 54-40.02), the height of the mixing curve initially gradu- ally increases to a peak as mixing proceeds. At this max- imum, all protein and starch is hydrated and dough is ‘optimally mixed. Such dough is either termed “optimally mixed dough” or “dough with minimum mobility.” Fur- ther mixing produces wet, sticky dough, a phenomenon referred to as “dough breakdown” (Delcour & Hoseney, 2010). The mixing curve obtained with a Farinograph also increases until reaching a maximum after which it grad ually decreases. From a typical farinogram, a number‘wring ENNOGENOUSLIPINS I NAEAD MAKING ates == 1.20 A o 1 2 3 4 5 6 7 8 MINUTES MIXING Arival Departure tolerance index (6 min after peak) consistency, BU 0 5 0 5 20 time, minutes FIGURE 6 Top: Mixogram of flour-water mixture, with a curve showing change in dough density (g/ems) during mixing. Reproduced with permission from Junge etal. 1981. Bottom: Typical frinogram showing frequently used parameters to four. Dough development time or peak time isthe time needed t be the cheological behavior of reach the maximum consistency. Arrival time is the time when the top of the ‘curve reaches 00 Brabender Unis (BU) and represents the rate of absorption. Departure imei te time when the top of the curve goes below the S00 BU line and marks the start of dough breakdown or overmixing. Stability is the time difference between departure and arrival time nd indicates four strength (which relates to gluten content and quality ofthe flout). Mixing tolerance inder's the difference in BU between the top ofthe curve a the peak and the top of the curve at min after reaching the peak. It estimates dough softening. Reproduced with permission from Pomeranz (1971) of parameters can be deduced to evaluate the absorp- network (Cauvain & Young, 2007). Indeed, dough mixing tion of flours, dough stability, and other dough charac- characteristics largely relate to the ability of gluten present. teristics (Figure 6, bottom) (AACCI Approved Method in flour to form dough (Delcour & Hoseney, 2010). Gluten 54-21,02), is the storage protein of wheat and good for 80% to 85% On a molecular level, dough development is essentially of total protein in wheat flour. It consists of gliadin and the formation of a three-dimensional viscoelastic gluten _glutenin in about equal levels (Day, Augustin, Batey, &om | AS Wrigley, 2006; Deleour & Hoseney, 2010; Delcour et al., 2012). The gliadin fraction comprises monomeric proteins that are extractable in aqueous ethanol. The glutenin frac- tion contains large polymers of glutenin subunits that are linked through disulfide bonds and are largely extractable in dilute acid or alkali, Reducing the disulfide bonds ren- ders glutenin subunits that like gliadin also are extractable in aqueous ethanol (Delcour & Hoseney, 2010). Upon hydration, gliadin becomes sticky and exhibits litte if no resistance to extension. Hydrated glutenin is rubbery, stiff, and breaks easily upon extension (Delcour & Hoseney. 2010; Wrigley, Békés, & Bushuk, 2006). The strength and elasticity of developed dough relate to the high molecular weight glutenins present (Delcour & Hoseney, 2010), In flour, glutenin at its lowest free energy is present as random coil structures (Figure 7a) Hydration and application of energy during mixing opens up these structures. As a result of extensional and shear forces, coiled glutenin is stretched and long chains align in the direction of shear (Figure 7b). Wheat glutenins with the smallest molecular weights probably open up first (Létang, Piau, & Verdier, 1999; MacRitchie, 1986, 1992) ‘When larger glutenin molecules are extended, they entan- gle. This increases viscosity and probably corresponds to the onset of dough development. The extension and entan- glement of the largest glutenins in flour probably corre- sponds to the peak mixing time (MacRitchie, 1986; Singh & MacRitchie, 2001). Viscoelastic dough thus results from entanglement of high molecular weight glatenin and its noncovalent (.e., hydrogen and hydrophobic) interactions with gliadin and the other glutenin fractions. The above result in dough continuity, a prerequisite for elastic behav- ior. Gliadins confer viscosity upon dough. Acting as plas- ticizers, they prevent too intensive interactions between glutenin chains. This contributes to the flow behavior of, dough. Once the gluten proteins are hydrated, further mix- ing results in progressive breakdown of disulfide bonds in gluten (Figure 7c) (Delcour & Hoseney, 2010; Létangeet al, 1998) Overall, hydration of gluten protein and application of energy during mixing result in a continuous three- dimensional viscoelastic gluten network that embeds the swollen starch granules (Cauvain & Young, 2007; Wrigley et al, 2006). Based on its water level (about 45%), dough can also be described as a bicontinuous system of a gluten gel and a free water phase. The main constituents solubi lized or dispersed in this bicontinuous system are (dam- aged) starch, albumins, globulins, and arabinoxylan (Elias- son & Larsson, 1993). As a result of dough development, the viscoelastic network can uniformly expand and retain gas during fermentation, proofing, and baking, which is reflected in a high bread LV and fine crumb structure (Cau- vain & Young, 2007) 42 | Airincorporation: A crucial event during mixing Air incorporation during dough mixing is of major impor- tance for final product quality characteristics such as bread LV and crumb structure, Whether a fine crumb structure is obtained partly depends on incorporation of air bub- bles in dough and subdivision thereof into small cells. ‘There is indeed a clear relation between air bubble struc- ture in dough during mixing and crumb cellular struc- ture in the baked bread (Baker & Mize, 1941; Cauvain & Young, 2007). During the initial stages of mixing, litle air is incorpo: rated, As dough becomes cohesive and begins to develop resistance to mixing, air is entrapped and dough density decreases (Figure 6, top). Some minimum viscoelasticity is required to envelope and as such retain included air bubbles (Cauvain & Young, 2007; Delcour & Hoseney, 2010; Junge, Hoseney, & Varriano-Marston, 1981). As mix- ing flour from cereals such as rice and maize with water does not lead to viscoelastic dough, it is not suited for producing breads with good LV and crumb structure (He & Hoseney, 1991). Included air bubbles are further subdi- vided into small gas cells by the shear exerted during mix- ing, The number of gas cells included during mixing is yp- ically 10° to 10/mms (Bloksma, 1990), Air incorporation during mixing is crucial as new gas bubbles cannot be generated in later stages of the bread making process. This follows from the Young-Laplace ‘equation for spherical gas bubbles: ap= 7, o where APis the pressure difference between the inside and outside of a gas bubble (ie., the Laplace pressure), y is the surface tension of the gas/liquid interface, and r is the radius ofthe gas bubble (Mills, Wilde, Sat, & Skeggs, 2003) ‘When r approaches zero and ys constant, this relationship shows that the pressure required to createa new gas bubble is infinitely high. Hence, gases produced during fermenta- tion, proofing, and baking cannot create new bubbles and ‘must diffuse into preexisting bubbles or, alternatively, to the atmosphere surrounding the dough. Air bubbles incor- porated during mixing thus serve as nuclei for gases to dil fuse into during fermentation and are at the basis of the cellular sponge-like structure of bread (Baker & Mize, 1941; Delcour & Hoseney, 2010). ‘Altogether, an insoluble though hydrated viscoelastic gluten network forms the continuous phase in mixed ‘wheat flour dough with starch and air bubbles present as discontinuous phases (Delcour & Hoseney, 2010). Devel- ‘oped dough can thus be considered to be a foam of discrete‘wear enpocenoustinns nyAREAD MAKEN aeirwg_| = a) Long glutenin chain Small glutenin subunits Non covalent crosslinks with intermolecular resulting from the breaking - Hydrogen bonds covalent disulfide bonds of disulfide bonds - Hydrophobic interactions FIGURE 7 Molecular interpretation of gloten development during dough mixing (2) At the beginning of dough mixing, (b) optimally mixed dough, and (c) overmixed dough. Reproduced with permission fom Létang etal (1299)sme | REVIEWS gas cells embedded in a continuous gluten-starch (GS) matrix. During baking, this foam structure is converted into the gas-continuous sponge structure of bread (Gan, Ellis, & Schofield, 1995. 4.3 | Gas cell stabilization: The dual-film hypothesis In wheat flour dough, the air bubbles incorporated dur- ing mixing biaxially expand during fermentation, proofing, and (the early stages of) baking (Cauvain & Young, 2007; Dobraszczyk, 1997). Yeast ferments sugar and thereby for the most part produces ethanol and carbon dioxide. As carbon dioxide is released in the aqueous phase, the pH of dough decreases from 6.0 (after mixing) to about 5.0 Once the aqueous phase is saturated, newly produced carbon dioxide diffuses into preexisting air bubbles as it cannot create new gas cells (sce Section 4.2) (Delcour & Hoseney, 2010). During baking, gas cells rapidly expand for a number of reasons. When temperature increases, yeast bbecomes more active (up toabout 55°C) and thus produces more carbon dioxide. Also as a result of the temperature Increase, carbon dioxide becomes less soluble and diffuses more into gas cells; water and ethanol evaporate; and all gases present expand. This leads to rapid volume gain in the oven that is referred to as the oven spring (Delcour & Hoseney, 2010), Not only the extent of gas production but also gas cell stability during fermentation, proofing, and baking determines bread LV and crumb structure. Indeed, dough expansion can only take place as long as gas is effectively retained (Cauvain & Young, 2007). It follows that LV is a measure of the degree to which gas cells expand prior to being opened (Gandikota & MacRitchie, 2005; He & Hoseney, 1991) Foam structures such as that of wheat flour dough are in general thermodynamically unstable (Ewers & Suther- land, 1952; Wilde, 2003). Three mechanisms cause instabil- ity of gas cells in foams: drainage, disproportionation, and coalescence (Mills eta, 2003; Ornebro, Nylander, & Elias- son, 2000). The later two are of importance in the context of bread making, Disproportionation or Ostwald ripening isthe growth of large gas cells at the expense of smaller ones in their neighborhood, thereby reducing interfacial area and pro- ducing coarser foam. It result from differences in Laplace pressure (Equation (1)) between gas cells of various sizes. ‘The pressure in small gas cells is higher than that in large gascellsso that net transfer of gas occurs from the smaller to the larger bubbles when they come close to each other. Smaller bubbles eventually disappear completely, which coarsens the foam structure in time (Garrett, 1993; Mills et al,, 2003; Voorhees, 1992; Wilde, 2003). In bread mak- ing, an increase in the number of large and a decrease in the number of small gas cells in dough is observed upon advancing from fermentation over proofing to baking (Li & Dobraszezyk, 2004). Evidently, the uniformity of the gas cells incorporated during dough mixing contributes to their stability since foam coarsening is promoted by (large) differences in gas cell sizes (Wilde, 2003). During mixing, air bubbles are not only included but also subdivided into small gas cells by the shearing action (MacRitchie, 1986). Dough punch- ing and molding also subjects the dough to high shear forces, This causes subdivision of existing gas cells, and increases in their numbers and improved size distribution (MacRitchie, 1976a). It contributes to a fine and uniform crumb structure in bread (Baker & Mize, 1941; Cauvain & Young, 2007). Coalescence is the fusion of two neighboring gas cells into one cell that then obviously has a lower surface-to- volume ratio (and hence a lower free energy). It implies ‘rupture of the cell walls that separate the two. Coalescence is only important during the later stages of fermentation and/or proofing and during the initial stage ofbaking since gas cells are completely embedded in the continuous GS ‘matrix after mixing, punching, or molding (Mills et al, 2003). ‘A long-accepted dogma in cereal chemistry has been that gas cell stability in bread making is exclusively estab- lished by the GS matrix surrounding expanding gas cells indough (Bloksma, 1990; Delcour & Hoseney, 2010). How- ‘ever, Gan, Angold, et al. (1990) with scanning electron microscopy illustrated that toward the end of proofing, discontinuities appear in the GS matrix surrounding gas cells (Figure 8, left). That dough continues to expand substantially and hence retains its gas-holding capacity ‘well after the discontinuities appear, provided evidence for a secondary stabilization mechanism. In their view, the integrity of expanding gas cells in dough is maintained bya liquid film around them that contains surface-active components. Initially, this liquid film is surrounded and strengthened by the GS matrix. The latter fails to fully ‘envelope the gas cells at later stages of expansion (Gan, Angold, etal, 1990; Gan, Ellis etal, 1995) (Figure 8, right) ‘The presence of liquid films surrounding gas bubbles in dough had already been suggested before (MacRitchie, 1076a) and was later supported by (indirect) experimen- tal evidence (Sahi, 1994; Sroan & MacRitchie, 2009; Sroan, Bean, & MacRitchie, 2009; Turbin-Orger et al. 2015a). The here described cooperative mechanism by which gas cells indough arestabilized by both the GS matrixas well as thin liquid films has been referred to as the dual-film hypoth- ‘esis (Gan, Angold, etal., 1990; Gan, Elis, et al., 1995; Stoan ‘& MacRitchie, 2009; Sroan et al, 2008).guid film Early stages of fermentation ‘Advanced stages of fermentation to ‘early stages of baking + (c) (© =Starch granule SH ~cluten-starch matrix © <6asceltned wih a aud fire End of oven spring or baking, FIGURE & Left: Scanning electron microscopy picture of $0 min proofed and freeze-dried dough showing discontinuities in the gluten- starch matrix surrounding gas cells, Reproduced with permission from Gan et al (1990). Right: Schematic representation af dough expansion ‘uring bread making as proposed by Gan etal. (1990) and Gan et a. (1995). Mixed dough contains discrete gas eells lined with liguid fllms embedded in a continuous gluten-starch matrix (a). This matrix falls to completely enclose the gas cells al advanced stages of fermentation, ‘eaving areas that contain only a thin liquid film (b). Baking increases the rate of gas cell expansion until the liquid films are no longer able to _generate new surface area, Acthis point in the process, the dough foam structure is converted into an open sponge structuce (C. In this view, loss of gas retention is not caused by rupture ofthe glut etal. 1995), 43.1 | Primary gluten-starch matrix ‘The GS matrix is a primary support of the expanding gas, cells in wheat flour dough. It should be extensible enough to expand when gas pressure increases but also sufficiently strong to avoid breaking (Sroan et al., 2009). The bal- ance between gliadin and glutenin is thus of major impor tance in bread making. Indeed, gliadin provides dough viscosity and cohesiveness and glutenin is responsible for dough elasticity (see Section 4.1) (Delcour & Hoseney, 2010; Wrigley et al., 2006). ‘The GS matrix surrounding gas cells expand biaxially during fermentation, proofing, and baking, This causes thinning of the cell wall matrix separating adjacent gas cells. When these thin regions of the cell walls continue expanding, at a given point, they can rupture, This is avoided as stress typically increases more than propor- tional in comparison to strain in these thin regions. As such, deformation or rupture is prevented and gas cells continue to expand along thicker parts of the enveloping GS matrix. This phenomenon, that is, local increases of stressin response to strain preventing coalescence, is called strain hardening (Dobraszezyk & Roberts, 1994; van Vliet, 2008), The tendency to strain harden is a good indicator starch matrix but by thal of the liquid films. Reproduced with permission fxom Gan, of wheat flour bread making performance (Dobraszczyk, 1997; Dobraszczyk & Roberts, 1994; Kokelaar, van Vliet, & Prins, 1996; Meerts, Cardinaels, Oosterlinck, Courtin, & Moldenaers, 2017) ‘Although dough rheology is primarily determined by its gluten network (Deleour et al, 2012; Wrigley et al, 2006), starch granules also contribute to its viscoelastic- ity (Amemiya & Menjivar, 192; Edwards, Dester, & Scan- Jon, 2002; Larsson & Eliasson, 1997; Uthayakumaran, New berry, PhanThien, & Tanner, 2002; Watanabe, Larsson, & Eliasson, 2002). Indeed, they are not only inert filles, ‘but also interact with the gluten network (Yang, Song, & Zheng, 2011). In this context, it has been demonstrated that dough elasticity is significantly reduced when par- tially replacing starch granules by glass beads of compara- ble size and shape (Edwards etal, 2002; Larsson & Elias- son, 1997). 43.2 | Secondary liquid films ‘As noted above, liquid films surrounding expanding gas cells in dough are, next to the GS matrix, atthe basis of a secondary stabilization mechanism (Gan, Angold, etal,ans_| ABIEWS Low molecular weight surfactants (e.g. lipids) 2D gel network of partly denatured, aggregated surface active proteins ‘Soluble arabinoxy/ stabilising interfacial layer? an) mata, Gluten-starch matrix FIGURES, from Mills eal (2002) 1990; Gan, Ellis, et al., 1995; Sroan & MacRitchie, 2009; Sroan et al., 2009). This is because they contain surface- active components (Mills et al., 2003) (Figure 9). Both surface-active proteins and lipids can stabilize gas cells, be it through different mechanisms. Surface-active proteins stabilize gas cell interfaces by forminga viscoelastic protein film (Figure 10a). When they adsorb at an interface, they (partly) denature and form a continuous network. The resultant elastic adsorbed layer stretches and deforms when gas cells deform. As long as the deformation does notexceed the elastic limit, the liquid film remains intact and forms a physical barrier between, adjacent gas cells thereby preventing coalescence (Mills et al,, 2003; Ornebro et al., 2000; Wilde, 2003). The pres- ence of arabinoxylan may enhance gas cell stability by increasing the viscosity of the liquid film and mediating interactions between adsorbed proteins (Mills etal., 2003). Protein adsorption at interfaces is practically irreversible and leads to foams with relatively small bubbles (Bos & van Vliet, 2003). Amphiphilic lipids stabilize gascell interfaces through the Gibbs-Marangoni mechanism (Figure 10b). ‘This mechanism relies on the high mobility of adsorbed lipids. Deformation of gas cells lined with lipids causes concen- tration gradients and local thinning of the liquid films sur- rounding gas cells. As a result, lipid molecules migrate to depleted areas to reduce the concentration gradients. ‘Their flow drags interlamellar fluid toward thinner regions in the liquid film and hence restores their original thick- ness (Bwers & Sutherland, 1952; Mills et al., 2003; Orne- ‘Schematic representation of aliquid film and the components involved in the stabilizationof gas cell interfaces. Figure adapted bro et al, 2000), Adsorption of low molecular weight sur- factants such as amphiphilic lipids at interfaces does not induce conformational changes. Itis usually reversible and produces foams with relatively large bubbles (Bos & van Vliet, 2001). Foam structures are efficiently stabilized by either the protein viscoelasticity or the Gibbs-Marangoni mecha- nism, However, mixtures of surface-active proteins and lipidsare highly unstable since proteins and lipids compete for the interface and impair each other's ability to stabilize it (Figure 10c) (Mills etal., 2003). 5_| WHEAT ENDOGENOUS LIPIDS IN BREAD MAKING: IMPACT ON QUALITY. OF FRESH BREAD As noted already ($2.3), lipids are important determinants of bread quality (Chung et al,, 2009; Pareyt et al., 2011) ‘Their roles in wheat flour bread making have been investi gated in different ways, thereby mostly focusing on bread LV and crumb structure. Traditionally applied approaches have been (i) tose flours varying in bread making quality fr lipid analyses as well as for bread making, Gi) to first defat and then reconstitute flours with the extracted lipids, and/or Gil) to add lipids of wheat (or other sources) to regular (nondefatted) flour.Strong intermolecular interactions —low mobility terete ©) eS. cae Reduced lipid mobility No protein-protein interactions FIGURE 10 Film stretching Fs se Film stretching EX w nee ww Protein deformation Film rupture due to slow diffusion and no protein deformation ‘Schematic representation of possible mechanisms for (destabilization of gas/liqld interfaces with (a) surface-active pro- teins, b) amphiphilic lipids, and (¢) mixtures ofthe two components. Cross-sections ofthe liquid flmsare shown with the letter “W" indicating the watery layer in between two interfaces. Figure adapted from Grnebro etal. (2000) More recently, the role of wheat endogenous lipids in bread making has been studied by taking advantage of lipase technologies. It is of note here that most studies in this context have only targeted flour nonstarch lipids. Starch lipids are due to their location inside starch granules not available in the phases of the bread making process preceding starch gela- tinization. Although they may be of importance for bread crumb firming, they do not or only marginally affect LV or crumb structure (MacRitchie, 1981; MacRitchie & Gras, 1973; Morrison, 1978). In what follows, the focus is also on the role of the flour nonstarch lipids and the term “lipids” refers to nonstarch lipids. Furthermore, within the scope of this manuscript, the term “LV” is interchangeable with “specific LV.” Since the above-mentioned approaches to investigate the roles of wheat endogenous lipids in bread ‘making hardly affect the mass of bread loaves, itis safe to assume that observed trends in LV are similar to those in specific LV. 5.1 | Flours varying in bread making quality Relationships between the identity and levels of wheat flour lipids and bread quality have been investigated by using flours differing in bread making performance for lipid analyses as well as for bread making (Chung et al, 2008). Chung, Pomeranz, and Finney (1982) extracted lipids with petroleum ether from straight-grade flour from 23 hard red winter wheats varying in bread making potential.vo L AEVEWS TABLE 1 ‘making parameters (Overview of (linear) correlation coefficients between wheat flour fee (hexane or petroleum ether extractable) lipid and bread Bread making parameter v Dough handling Crumb grain score Flour fee lipid parameter TL 0.55" 0.65 10081" NPL. 0.72" 0.7710 092" 0.40%" PL O89 087, 043100056 CAL ost" oar pepertt, one NPLIPL 0.68" 0.91%" 0.98 0,70" NPLIGAL ——-0.97" Pancaza et 1900), ‘ohm and Chung (2003), schungetal (982) sekes eta. (88), *Graytosch eta (8 sp <5" p01; p< 00, Abbreviations: DGDG, cigalactosycacyigiyceol level: GAL, galactlipi level: LV, loaf volume; NPL, nonpolar lip evel; PL, polar ip level; TL. tt ipit love Free lipids were fractionated into nonpolar and polar lipids and analyzed for their carbohydrate constituents (mainly galactose). Positive linear relations were found between LV and the levels of free polar lipids and free galactolipids in flour. In contrast, a negative linear relation was noted between LV and the ratio of free nonpolar to polar flour lipids (Table 1). Other researchers have also reported (significant) rela- tionships between flour lipid parameters and LV (Table 1). Inline with findings by Chung etal. (1982), the ratio of free nonpolar to polar lipids in flour has often been reported to bbe negatively related with bread LV (Békés, Zawistowska, Zillman, & Bushuk, 1986; Larsen, Baruch, & Humphrey ‘Taylor, 1989; Matsoukas & Morrison, 1991; Morrison, Law, Wylie, Coventry, & Seekings, 1989; Zawistowska, BEkés, & Bushuk, 1984), Békés et al. (1986) have shown that LV can bbe estimated with high accuracy from the ratio of free non- polar lipids to galactolipids in flour. Morrison et al. (1989), like Chung et al. (1982), noted a positive relation between LV and the level of free polar lipids, whereas Larsen et al (1989) found no such relation. Furthermore, free polar lipid levels in flour showed substantial positive contributions to dough handling (Graybosch, Peterson, Moore, Stearns, & Grant, 1993) Although negative relations between LV and free total as well as nonpolar lipid levels have been reported (Chung et al., 1982 Larsen et al., 1989; Ohm & Chung, 2002), Panozzo et al. (1990) found these lipid fractions to be pasi- tively correlated with LV (Table 1). This may have resulted from the positive relation between free nonpolar and polar lipid levels fr the sets of flour they used, Therefore, it most likely does not indicate that free nonpolar lipids positively contribute to LV. Ohm and Chung (2002) investigated whether variations. in the level of petroleum ether extractable galactolipids {ie., DGDGs and MGDGs) could be correlated with flour bread making properties. They used 12 North American hard winter wheats multiplied at six locations and noted significant negative correlations between LV and free non- polar lipid levels. Furthermore, bread crumb scores cor- related positively with the percentages of DGDG in free lipids among the 12 cultivars (Table 1). This suggested that increasing free DGDG levels would improve bread crumb structure While in all above-mentioned studies, relationships between free hexane or petroleum ether extractable) lipids and bread making quality of flours were investigated, McCormack, Panozzo, Békés, and MacRitchie (1991) exam- ined links between chloroform extractable lipids and LV by using flours from 14 wheat cultivars grown in three suc- cessive crop years and at six levels of nitrogen fertli Consistent correlations were found between LV and sev: ‘ral lipid related parameters. In particular, negative corre- lations were noted between LV and the ratio of nonpolar to total or polar flour lipids. They estimated that differences in flour lipid level and the ratio of nonpolar to polar lipids accounted for only up to 5% and 11% of the variation in, bread LV, respectively (McCormack, Panozzo, Békes, et al., 1991a). However, it is equally true that LV is very sensi- tive to changes in flour lipid level and composition, as will bbe demonstrated in Sections 5.2 and 5.3. It can therefore bbe concluded that these small contributions of lipids todifferencesin LV are due to substantially constant lipid lev- els and their compositions inthe applied flours. Overall, studies in which flours varying in bread making quality were used indicate detrimental roles for nonpolar and beneficial roles for polar flour lipids in bread mak- ing, With the exception of McCormack, Panozzo, BEkéS, et al. (1991, all researchers exclusively investigated rela- tions between free flour lipids and mainly bread LV. The conclusions re therefore mostly restricted tothe effects of fice lipids, which only account for approximately 60% of the total nonstarch lipids in flour (see Section 2.3). Varietal variations in free flour lipids appear to parlly account for differences in bread LV, with free galactolipids havingan outspoken beneficial function. However, since correlations do not necessarily reflect functional cause-and-effect rela- tionships, complementary information is required to con- firm the roles of different (iree) lipid fractions of flour in bread making. 5.2 | Fractionation-reconstitution experiments ‘More direct testing of relations between wheat endogenous lipids and bread LV has been done by first defatting and then reconstituting flours with (portions of the) extracted lipids and by supplementing nondefatted flours with lipids from wheat or other sources. Hoseney et al. (1959) compared the role of the free nonpolar and free and bound polar lipids from flour in bread making when reconstituted with an almost com- pletely defatted flour. They obtained such flour without irreversibly damaging its bread making quality through a complex procedure involving multiple extraction and frac- tionation steps. They found small levels of free or bound ppolar lipids to be detrimental to LV (Figure 11 top), unless they co-oceurred with free nonpolar or bound polar lipids in their native state. Adding high levels of free or bound polar lipids allowed restoring LV, but larger levels of bound than of free polar lipids were required (Figure, top). Free nonpolar lipids had no effect on bread LY. Other researchers have tried to avoid uncertainty intro- duced by complex defatting procedures such as the one applied by Hoseney et al. (1969) by using only extrac- tion solvents that do not alter flour functionality. Rela- tively nonpolar solvents such as petroleum ether were found to be suitable to research the role of lipids in bread making as they do not harm interactions between lipids and other flour constituents such as protein or carbohy- drates (Section 2.3) (Chungetal., 2009; MacRitchie & Gras, 1973). With such procedures, however, only the free lipids are extracted and not the other nonstarch lipids in flour (Section 2.3). few =» Attempts to extract all bound flour lipids without irre- versibly impacting on flour rheological properties, dough fermentation and bread quality have in essence been unsuccessful. Extraction conditions (in terms of solvent, temperature, method, ete.) required to raise the extraction yield of flour polar lipids increasingly damage wheat flour bread making performance (Chung, Pomeranz, Finney, & Shogren, 1977; Chung, Pomeranz, Finney, Hubbard, & Shogren, 1977; Schaffarezyk, Ostdal, Matheis, Jekle, & Kochler, 20162). A compromise between lipid extractabil- ity and flour bread making quality thus needs to be made. Chloroform has often been used as extraction solvent. It removes relatively high portion of lipid without affecting flour properties (MacRitchie & Gras, 1973; MeCormack, Panozzo, & MacRitchie, 1991; Stoan & MacRitchie, 2009), It extracts significantly higher levels of galactolipids and phospholipids than do nonpolar solvents such as hexane but significantly lower levels of galactolipids than do more polar solvents such as WSB (Bahrami et al., 2014) Plots of LV as a function of (chloroform extractable) flour lipid level show outspoken minima at lipid levels in between those of chloroform defatted and regular flours (Figure 1, bottom). Atthese minima, LVs are up to 35% and 20% smaller than those of breads prepared from chloro- form defatted and regular flours, respectively (MacRitchie & Gras, 1973; Stoan & MacRitchie, 2009). Bread LV is thus very sensitive to changes in flour lipid level. Not only the level, but also the flour lipid composi- tion tremendously affects bread LV. In fractionation and reconstitution studies, nonpolar and polar lour lipids were found to lower and increase LV, respectively (Daftary, Pomeranz, Shogren, & Finney, 1968; De Stefanis & Ponte, 1976; MacRitchie & Gras, 1973; McCormack, Panozz0, & MacRitchie, 1991; Ponte & De Stefanis, 1969; Sroan & MacRitchie, 2009). This is in line with outcomes of most studies in which flours varying in bread making quality were used (see Section 5.1. ‘When varying the relative amount of nonpolar lipids in flour from 0% to 100% while keeping the total amount of, (Chloroform extractable) lipids constant, LV decreases with up to 25% to 32% (McCormack, Panozzo, & MacRitchie, 1991), Furthermore, unlike incremental addition of total flour lipids, addition of increasing levels of nonpolar flour lipids to chloroform defatted flour causes a continuous decrease in LV (Figure 11, bottom) (MacRitchie & Gras, 1973; Sroan & MacRitchie, 2009). This seems to be in con- trast with the finding of Hoseney et al. (1969) that flour nonpolar lipids have no effect on bread LV. Important to note is that the flour nonpolar lipid fraction contains not only nonpolar lipids but also lipids such as FFAS (see Section 2.1) (Christie & Han, 2010; Chung et al., 2008). ‘These lipids of intermediate polarity mainly cause nega- tive effects. In contrast, the volume of bread prepared froma) 5 ar 5 3 x Bound 3 Saree 02 04 06 08 Polar flour lipids (%) 400 je Polar flour lipids 380 (a) em almitic aci £ Palmitc acid E 250 2 Total flour lipids z 3 = 200 Myristic acid ®) Gas-iquidinterface 150 Nonpolarflou lipids 1, aajexsatniedd Linoleic acid 10 | g surtce-ctive protein 0% 100% 200% 300% 400% % of chloroform extractable flour lipids ricuRE 11 ‘Top: Effect offre polar and bound polar lipid addition on loaf volume of bread baked from 10g almost completely defatted four. The recipe contained 3% shortening. Reproduced with permission from Hoseney etal. (968). Bottom: Loaf volume as a function of the level of diferent types of lipids (expressed as percentage of chloroform extractable flour lipids) added to 38g chloroform-defatted our In the absence of lipids gas/liguid interfaces are stabilized by a viscoelastic film of surface-active proteins (a). When small levels of (pola) flour lipids are added back tothe flour, ga/liquid interfaces in dough are unstable due tothe presence ofboth surface-ative lipids and proteins (b). High levels of (pola) lou lipids gradually displace surface-active proteins from the interface, thereby forming stable lipd-dominated gasiliquid interfaces. Figures adapted from Sroan and MacRitehie (2009) and Gan etal. (1995) defatted flour is little affected by nonpolar compounds such as TAGs (De Stefanis & Ponte, 1976; MacRitchie, 1977). Linoleic and myristic acids depress while palmitic acid has little impact on LV (Figure ll, bottom) (De Stefanis & Ponte, 1975; Sroan & MacRitchie, 2009). Presumably, the nonpolar lipid fraction isolated by Hoseney et al. (1959) contained litteif any lipids of intermediate polarity so that ithad no effect on bread LV when added to defatted flour, Incremental addition of polar flour lipids to chloro- form defatted flour, like that of total flour lipids, initially resullsin LV depression followed by LV increases at higher addition levels (Figure 11, bottom) (MacRitchie & Gras, 1973;Stoan & MacRitchie, 2009).Such outcomes are inline with the findings of Hoseney et al. (1969). They reported that added small levels of polar lipids are detrimental to, whereas high added levels of polar lipids restore LV(Figure 11, top). It i further of note that galactolipids are better at improving bread quality than phospholipids (Daf- tary etal, 1968; Helmerich & Koehler, 2005; Selmair, 2010; Selmair & Koehler, 2008, 2009). ‘The bread making performance of galactolipids has been examined by adding synthesized (ie., MGDGs and monogalactosylmonoacylelycerols [MGMG5}) or isolated (from lecithin, such as DGDGs) galactolipids to regular (nondefatted) lourand comparing their functional proper- ties in bread making with those of commercial surfactants (Selmair & Koehler, 2008, 2009). The major flour galac- tolipids, that is, DGDGs and MGDGs (Section 2.3), and MGMGs displayed excellent bread making performance when compared with classical surfactants. DGDGs and MGMGsby far had the best impacton bread making, while that of MGDGs was lower but still comparable to that of most surfactants. Interestingly, galactolipids with two galactose units have to contain two esterified FAs to exert effects similar to those exerted by galactotipids containing one galactose unit and one esterified FA. This underlines that the balance between hydrophilic and lipophilic mot ties in galactolipid structures is important for their impact in bread making (Selmair & Koehler, 2009). Earlier, Elias- son and Larsson (1993) already proposed that a high ratio of MGDGs to DGDGs in flour is detrimental to bread quality Differences in FA composition result in slight differences in activities of galactolipids in bread making (Selmair & Koehler, 2002). ‘The functionality of several phospholipids in bread making has been studied by isolating them from crude lecithin preparations and adding them to regular (non- defatted) flour. Of the major wheat flour phospholipids (Section 2.3), NAPEs have an overall negative effect when added at levels of 0.05% to 0.40%, In contrast, PCs have no or a moderately positive effect on bread LV depend- ing on their origin when added at levels of 0.02% to 0.10% (Helmerich & Koehler, 2005; Selmair, 2010). Of the minor ‘wheat phospholipids (Section 2.2), phosphatidylinositols increase bread LV when added at levels of 0.02% to 0.06%, However, they are less effective when added at levels of, 0.08% to 0.10%. Phosphatidylethanolamines, depending on their origin, have no or a negative effect on bread LV when added at 0.02% to 0.10%. As for galactolipids, it can be assumed that the FA composition of a given phospholipid has a substantial impact on its functionality (Helmerich & Koehler, 2008). ‘Altogether, fractionation-reconstitution approaches clearly have demonstrated that flour nonstarch lipids, although present in very small levels (§2.3), have substantial effects on bread making performance in terms of LV and crumb structure. More in particu- lar, their outcomes have confirmed the detrimental roles of nonpolar and beneficial roles of polar flour ies =» lipids in general and especially galactolipids in bread ‘making. Although fractionation-reconstitution experiments have provided a better understanding of the effects that wheat endogenous lipids have on fresh bread quality, they have some limitations. First, they are not free from impacts of extraction solvent(s) on flour constituents other than lipids. This introduces uncertainty about the observed effects. Second, they inevitably result in reloca- tion of flour lipids so that they do not occur at their native (endogenous) position when (re-Jadded to (defatted or regular) flour. Relocation of flour lipids presumably affects their functionality in bread making. Pauly et al. (2014b) found chloroform treatment (ie., addition and subsequent evaporation of chloroform) of isolated gluten and starch fractions to reduce extension until fracture and total force needed for fracture of GS doughs (as measured with a Kieffer extensibility rig) as well as LV of GS breads. This indicated that either altering the location of endogenous lipids or contact with chloroform affects dough and bread properties. Maximum load until fracture of GS doughs and crumb structure of GS breads were not affected by chloroform treatment (Pauly, Pareyt, Fierens, & Delcour, 20140). Furthermore, Pycarelle et al, (2019) showed that relocating flour lipids with hexane or hexanesisopropanol (322, v/v) negatively impacts the properties of sponge cake batter and those of the resultant cakes. 5.3 | Lipase technologies More recently, the role of wheat endogenous lipids (and their enzymatically released hydrolysis produets) in bread ‘making has been investigated by using lipases and relating (Changes in) the dough lipid composition to bread LV (Ger- its et al, 2014; Gerits, Pareyt, Masure, & Delcour, 2015a, 2015b; Melis, Meza Morales, & Delcour, 2019; Melis, Ver- bauwhede, Van de Vondel, Meza Morales, & Delcour, 2019; Schaffarczyk, Ostdal, & Koehler, 2014; Schaffarcayk, Ost- dal, Matheis, & Koehler, 2016b). Lipases hydrolyze (phos- phodijester bonds of glycero-(phospho)lipids (see Section 3.1) and therefore allow to change the molecular struc- ture of endogenous lipids in situ without affecting other constituents, Such approach overcomes the limitations of fractionation-reconstitution experiments as itis free from impacts ofextraction solvents and flour lipids occurat their native (endogenous) location when in contact with lipases. In bread making, the free and bound nonstarch lipids are substrates for lipases. They make up 60% to 70% of, the lipids in wheat flour. The other wheat flour lipids are essentially unavailable as they are starch lipids occur- ring inside the granules (De Maria, Vind, Oxenbell, Svend- sen, & Patkar, 2007). In general, lipases hydrolyze wheatow | A wear ENDOGENOUS LIPIDS TAGs ‘nonpolar ips MDGs 606s Goectolpids NAPES: Pes Phospholipids FIGURE 12, Lysoupins ———+ tas Potential effects of lipases (indicated by arrows) on the dough lipid population when applied in wheat flour bread making. ‘The desirable specificity of perfect bread making lipases proposed in ths review is indicated in green (hydrolysis) and red (no hydrolysis). Released glycerol backbones are not included Abbreviations: DAGS, diacylglycerols; DGDGs, digalactosyidiacylglycerols; DGMGSs, digalactosylmonoacylglyeerols; FFAS, fee fatty acids; LPCS, ysophosphatidyicholines; MAGS, monoacylglyeerols; MGDGs, monogalactosyldiacylglycerols; MGMGs, monogalactosylmonoacylgly- exols.N elycerois. endogenous lipids into the corresponding lysolipids, thereby releasing FFAs (Figure 12). Lysolipids may, in turn, be further hydrolyzed into their glycerol backbone and FFAs. The wheat endogenous lipids accessible to lipases in bread making are mainly TAGs (21% to 47%), DGDGs (13% to 17%), MGDGs (5% to 6%), NAPEs (4% to 5%), and PCs (4% to 6%) (see Section 2.3). Their cotrespond- ing lysolipids are DAGs, MAGs, digalactosylmonoacylglyc- erols (DGMGs), MGMGs, NALPEs, and LPCs (Figure 12). Apart from smaller quantities of other FAs, (esterified) FAs in nonstarch lipids are 50% to 65% linoleic, 19% to 26% palmitic, and 10% to 21% oleic acids (see Section 2.3. Assuming that lipases have no preference for a given FA. structure (in terms of chain length and/or degree of unsat- uration), the composition of the FFA population present in dough/bread as a result of lipase action is similar to that of esterified FAs in nonstarch lipids. Today, itis unclear whether this assumption holds true as this has not been reported for bread making lipases yet. Lipase use improves processing of dough and overall bread quality. It increases the stability and maximum resis- tance to extension of dough while decreasing its stick- iness (Colakoglu & Ozkaya, 2012). Furthermore, lipase action in bread making increases LV (Aravindan, Anbu- mathi, & Viruthagiri, 2007; De Maria et al., 2007; Frauen- lob, Scharl, D'Amico, & Schoenlechner, 2018; Gerits etal, 2014; Gerits, Pareyt, Masure, et al., 2015b; Melis, Meza ALPES, N-acylIysoplhosphatldylethanolamines; NAPEs, N-ayl phosphatidylethanolamines; PCs, phosphsatidylcholines; TAGS, tiacy- Morales, et al., 2019; Melis, Verbauwhede, et al., 2019; “Moayedallaie, Mirzaei, & Paterson, 2010) improves crumb softness and structure (Aravindan et al., 2007; Poulson, Soe, Rasmussen, Madrid, & Zargahi, 2010) and delays or ‘even reduces the extent to which starch retrogradation ‘occurs during storage (Colakoglu & Ozkaya, 2012; Frauen- lob et al, 2018; Gerits, Pareyt, Masure, et al, 2015; Sis- ‘woyo, Tanaka, & Morita, 1999).In this context, lipases have logically been proposed to be (partial) alternatives for sur factants in bread making as they in situ produce molecules with a hydrophilichead and a hydrophobic tail resembling surfactants (Aravindan et al., 2007; Colakoglu & Ozkaya, 2012; De Maria et al,, 2007; Moayedallaie et al., 2010). In general, their effects have been related to an improved surface activity of the lipid population. Indeed, lipases produce more polar lipids by freeing one or more FA(s) from nonpolar and/or polar lipids (Figure 12) (Colakoglu & Gzkaya, 2012; Gerits et al, 2014; Moayedallaie et al, 2010; Schaffarcayk et al, 2014). However, not every lipase improves bread quality and some even reduce LV (De Maria et al., 2007; Gerits etal, 2014; Schaffarczyk et al., 2014). Indeed, the selectivity of given lipase clearly affects its functionality in bread mak- ing. De Maria et al. (2007) proposed that a perfect bread ‘making lipase has optimal activity toward the different sub- strates available in flour (i.e., acylglycerols, galactolipids, and phospholipids) and as such results in gas cell stability