Food Chemistry 172 (2015) 613–621

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Native and enzymatically modified wheat (Triticum aestivum L.) endogenous lipids in bread making: A focus on gas cell stabilization mechanisms Lien R. Gerits ⇑, Bram Pareyt, Hanne G. Masure, Jan A. Delcour Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), KU Leuven, Kasteelpark Arenberg 20 Box 2486, B-3001 Leuven, Belgium

a r t i c l e

i n f o

Article history: Received 6 May 2014 Received in revised form 21 August 2014 Accepted 13 September 2014 Available online 20 September 2014 Keywords: Wheat flour lipids Lipases Dough liquor Dough extensibility

a b s t r a c t Lipopan F and Lecitase Ultra lipases were used in straight dough bread making to study how wheat lipids affect bread loaf volume (LV) and crumb structure setting. Lipase effects on LV were dose and dough piece weight dependent. The bread quality improving mechanisms exerted by endogenous lipids were studied in terms of gluten network strengthening, which indirectly stabilizes gas cells, and in terms of direct interfacial gas cell stabilization. Unlike diacetyl tartaric esters of mono- and diacylglycerols (DATEM, used as control), lipase use did not impact dough extensibility. The effect on dough extensibility was therefore related to its lipid composition at the start of mixing. Both lipases and DATEM strongly increase the levels of polar lipids in dough liquor and their availability for and potential accumulation at gas cell interfaces. Lipases form lysolipids that emulsify other lipids. We speculate that DATEM competes with (endogenous) polar lipids for interacting with gluten proteins. Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction Endogenous wheat lipids, although a minor fraction of flour, are important in bread making (Chung, Ohm, Ram, Park, & Howitt, 2009; Pareyt, Finnie, Putseys, & Delcour, 2011). Selective modification of the lipid composition drastically affects bread loaf volume (LV) (Gerits, Pareyt, & Delcour, 2014; MacRitchie & Gras, 1973). High bread LV has been related to efficient gas holding capacity (Delcour & Hoseney, 2010; van Vliet, Janssen, Bloksma, & Walstra, 1992). The latter depends on dough aeration and subsequent gas cell stabilization. These complex phenomena are of major importance in bread making. The gluten network formed during mixing determines how much air is incorporated at that stage. It embeds the gas cells and acts as a primary stabilization mechanism. Gas cells present in dough at the end of mixing act as nuclei which are filled with carbon dioxide produced as a result of fermentation since yeast cannot create new cells. Punching subAbbreviations: DATEM, diacetyl tartaric esters of mono- and diacylglycerols; DGDG, digalactosyldiacylglycerols; DGMG, digalactosylmonoacylglycerols; DL, dough liquor; EP, enzyme protein; FFA, free fatty acids; LV, loaf volume; NAPE, Nacyl phosphatidylethanolamine; NALPE, N-acyl lysophosphatidylethanolamine; PC, phosphatidylcholine; TAG, triacylglycerols. ⇑ Corresponding author. Tel.: +32 16 32 1634; fax: +32 16 32 19 97. E-mail address: [email protected] (L.R. Gerits). http://dx.doi.org/10.1016/j.foodchem.2014.09.064 0308-8146/Ó 2014 Elsevier Ltd. All rights reserved.

divides the expanded cells into more but smaller cells (Delcour & Hoseney, 2010). Remarkably, whereas discontinuities appear in the gluten films surrounding the gas cells already during early stages of fermentation, dough does not lose its gas holding capacity at that time. Based on circumstantial evidence, it has been hypothesised that a thin liquid film which contains surface active proteins and (polar) lipids then surrounds and stabilizes the gas cells (Gan et al., 1990). The liquid film is part of the dough liquor (DL) phase, i.e. the liquid phase of dough obtained by ultracentrifugation. DL has been proposed to act as the medium for incorporation and subsequent growth of gas cells in dough (MacRitchie, 1976; Primo-Martin, Hamer, & de Jongh, 2006; Salt et al., 2006). The above described (cooperative) mechanism for stabilizing gas cells in dough by the gluten network and the thin liquid film has been referred to as the dual film hypothesis (Gan et al., 1990; Gan, Ellis, & Schofield, 1995; Sroan, Bean, & MacRitchie, 2009; Sroan & MacRitchie, 2009; Turbin-Orger et al., 2012). Thus, lipids can (in)directly stabilize gas cells, and, as such, contribute to bread LV. Their indirect impact is related to their interaction with gluten proteins which promotes gluten aggregation. Their direct impact arises from their ability to position themselves at the gas cell interface, thereby preventing coalescence of neighbouring gas cells. Finally, like some surfactants (Eliasson, 1983), lipids may postpone starch

614

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

gelatinization and therefore impact bread LV by prolonging the oven rise (Ahmad et al., 2014). By including lipases in a dough recipe and analysing the lipid composition after fermentation, it was recently observed that an overall shift in the lipid population towards higher levels of polar lipids (i.e. by conversion of galactolipids, and to a lesser extent, phospholipids to their lysolipid form) positively impacts bread LV (Gerits et al., 2014). Moreover, the hypothesis that the type of mesomorphic phase formed when lipids are in contact with water is crucial for proper gas cell stabilization and that lipases catalysing the transition from lipid classes promoting the hexagonal II mesophase to lipid classes promoting the lamellar and/or hexagonal I mesophases positively impact bread LV was put forward (Gerits et al., 2014). Against this background, this study further elaborates on the specific mechanisms whereby different lipid classes endogenously present in wheat flour improve bread LV. As in a previous study, the present study uses two lipases (Lecitase Ultra and Lipopan F) in a straight dough bread procedure to in situ selectively alter only the endogenous lipid fraction. We also used a well-known surfactant [i.e. diacetyl tartaric esters of mono- and diacylglycerols (DATEM)] in bread making. Indeed, lipases are known to generate surfactants in dough (Gerits, Pareyt, Decamps, & Delcour, 2014) (and references cited therein) and we found it useful to compare the effects of DATEM with those exerted by lipases. DATEM also positively impacts bread LV (Gerits et al., 2014; Koehler & Grosch, 1999). As in our previous paper, lipid compositions of fermented dough, but now also its DL and baked bread crumb were determined. Furthermore, the Kieffer dough and gluten extensibility rig was used to study whether in situ modification of the lipid population affects gluten network strength. Differences in dough height and the moment at which they became apparent for control dough and its lipase or DATEM containing counterparts were determined. Starch gelatinization in the different dough samples and whether and to what extent it is affected by differences in lipid composition were analysed with differential scanning calorimetry (DSC). Finally, final bread crumb structure was analysed with 2D image analysis.

2. Materials and methods 2.1. Materials Grains from soft wheat cultivar Claire were from Limagrain (Rilland, The Netherlands) and conditioned to 16.0% moisture before milling with a Bühler (Uzwil, Switzerland) MLU-202 laboratory mill. Milling yield of straight grade flour was 69.4%, and its moisture and protein contents respectively 12.6% and 9.3% [on dry matter basis]. The latter were determined with AACC-I Approved Method 44-19.01 (AACC-I, 1999) and an adaptation of the AOAC Official Method (AOAC, 1995) to an automated Dumas protein analysis system (EAS Vario Max CN, Elt, Gouda, The Netherlands) with 5.7 as nitrogen to protein conversion factor. Lipopan F and Lecitase Ultra were kindly donated by Novozymes (Bagsvaerd, Denmark). Lipopan F, a Fusarium oxysporum enzyme preparation, is used in bread making as a source of both lipase and phospholipase activities. Lecitase Ultra is a phospholipase used in degumming of edible oils. It is the result of combining homologous genes encoding Thermomyces lanuginosus lipase and F. oxysporum phospholipase (De Maria, Vind, Oxenboll, Svendsen, & Patkar, 2007). Lecitase Ultra and Lipopan F had a lipase activity based on p-nitrophenyl palmitate of 0.15 units (U) and 56.50 U, respectively, with one U being the number of lmol of p-nitrophenol released per minute and per mg enzyme under the conditions of the assay (Gerits et al., 2014). DATEM was from Puratos (Groot-Bijgaarden,

Belgium). Lipid standards were as in Gerits, Pareyt, and Delcour (2013) and Gerits et al. (2014). All solvents used were from VWR (Haasrode, Belgium) unless specified otherwise and of at least analytical grade. 2.2. Methods 2.2.1. Bread making Bread was made in triplicate on a 10 g and 100 g of flour scale according to the straight dough bread making procedures of Shogren and Finney (1984) and Finney (1984), respectively, but without adding shortening. Flour (10.0 g or 100.0 g, 14.0% moisture basis), water, sugar (6.0% on flour basis), compressed yeast (5.3% on flour basis) and salt (1.5% on flour basis) were mixed in the appropriate pin mixers (National Manufacturing, Lincoln, NE, USA). The amount of water added and the optimal mixing time were determined by Mixograph analysis according to AACC-I Approved Method 54-40.02 (AACC-I, 1999) and were respectively 45% (on flour basis) and 150 s. In what follows, bread making procedures for 10.0 g or 100.0 g of flour will be referred to as 10 g or 100 g bread making, respectively. Lipases were added in levels ranging from 0.25 to 5.0 mg enzyme protein (EP)/kg flour, DATEM in levels between 0.5% and 1.5% (on flour basis). Baked bread samples were cooled (2 h) and their LVs determined with a Volscan Profiler (Stable Micro Systems, Godalming, Surrey, UK) for 100 g bread making and as in Vanhamel, Van den Ende, Darius, and Delcour (1991) for 10 g bread making. Standard deviations on triplicates did not exceed 3.0%. The temperature in the centre of the baking dough/ bread for both 10 g and 100 g bread making was monitored in duplicate using a Datapaq temperature logger (MultiPaq 21) with type T thermocouples (Datapaq, Cambridge, UK). A stainless steel thermal barrier (Datapaq) was used to protect the MultiPaq logger from the heat in the oven. Changes in dough height for 100 g bread making during late fermentation and baking were monitored in duplicate with time-lapse photography with a Canon (Machelen, Belgium) Powershot S50 digital camera. For this specific case, baking pans with smaller dimensions (50  100  45 mm3) were used to monitor dough height already during fermentation. Partially baked bread (100 g scale) was withdrawn during baking when the maximum dough/bread height was reached and immediately frozen in liquid nitrogen to stop the heat transfer and ongoing transformations. Thereafter, its outer and inner layers were separated based on visible differences between baked (i.e. crumb) and unbaked (i.e. dough like) aspect and further used for DSC analyses (Section 2.3.3). 2.2.2. Rheofermentometer analyses Control dough (100.0 g flour basis, prepared as in Section 2.2.1), or with added Lecitase Ultra or Lipopan F (0.5 mg EP/kg flour), or DATEM (0.5% on flour basis), was placed in the Rheofermentometer (Chopin, Villeneuve-La-Garenne, France) basket and covered with a 254 g piston. Temperature was 30 °C during the entire run (360 min). Parameters derived from the resulting dough development graph were maximum dough height (mm) and dough tolerance (min). The latter is the difference between the moments (before and after reaching the maximum dough height) at which a dough height of 90% of the maximum dough height was reached. A greater such difference indicates a higher dough tolerance. The time of gas cell opening (min) was derived from the Rheofermentometer gas release graph as the moment of carbon dioxide release from the dough. Measurements were performed in triplicate. 2.2.3. Uniaxial dough extensibility analysess Uniaxial dough extensibility was measured with the Kieffer dough and gluten extensibility rig mounted on an Instron (Norwood, MA, USA) 3342 Single Column Testing System with a 50 N

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

load cell. Dough pieces prepared from 10.0 g of flour as described above (Section 2.2.1) but without yeast, sugar and salt, and either with or without 0.5 mg EP Lecitase Ultra or Lipopan F/kg flour or 0.5%, 1.0% and 1.5% DATEM (on flour basis), were placed on a paraffin coated Teflon mold and formed with a lubricated Teflon top plate to obtain uniform dough strands (approximately 1 g, 52 mm length, 3.2 mm width). The Teflon mold and dough were allowed to rest (32 °C, 95% relative humidity) for either 30 or 120 min (only upon addition of Lecitase Ultra and Lipopan F). In a second set up, dough pieces with and without added Lecitase Ultra and Lipopan F were stored for 90 min (same conditions), remixed for 30 s, and placed in the Teflon mold for a final resting time of 30 min (same conditions) before analysis. Triplicate dough pieces each yielded 6–8 dough strands. Parameters derived from the load–displacement curves included maximum extensibility (mm), maximum force during extension (N) as a measure for the resistance to extension, and total work needed for breakage (mJ), i.e. the area under the load–displacement curve (Kieffer, Wieser, Henderson, & Graveland, 1998). 2.2.4. Dough liquor isolation Fermented dough samples (prepared from 10.0 g flour as outlined above) were placed in ultracentrifuge tubes (38 mL thickwalled polycarbonate tubes, Beckman Coulter, Brea, CA, USA) and centrifuged (165,000g, 60 min, 20 °C, L7 Ultracentrifuge, Beckman Coulter). The tubes were then placed on ice to minimise enzymatic activity. The obtained supernatants further referred to as DL, were collected in amber coloured vials, weighed and freeze-dried. When using fermented dough samples, a foam was formed on top of DL instead of a lipid pellicle. This foam, with the lipids suspended in it, could much more easily be collected than the lipid pellicle obtained as a result of ultracentrifugation of freshly mixed dough. Hence, it was easier to include the lipids of interest in the DL. DL samples were prepared in triplicate. 2.2.5. Image analysis Bread crumb structure was analysed on bread loaves prepared from 100.0 g of flour. Three of such bread loaves were sliced at the centre with an electric slicer (Affettatrice Slicer 30 N, Galesecca, Italy), yielding 3 slices (1.0 cm thickness) per bread. Thus, for each recipe, 9 bread slices were analysed. The slices were placed one by one on a flatbed scanner (HP Scanjet 3800, Hewlett–Packard, Beijing, China). An area of 40  40 mm at the centre of each slice was selected and evaluated. Image resolution was 300 dpi and images were stored as .bmp files (473  473 pixels). The Otsu thresholding algorithm (Otsu, 1979) was used as an Image J (Bethesda, MD, USA) plug-in for image segmentation (conversion to binary images), as commonly done for optimal thresholding (Gonzales-Barron & Butler, 2006). Crumb cell detection on the binary images was performed with the image processing toolbox of Matlab 8.0 (Mathworks, Natick, MA, USA). Crumb grain features determined were mean cell area (mm2) and number of cells/cm2. 2.2.6. Differential scanning calorimetry Differential scanning calorimetry analyses of fermented dough and partially baked bread samples (4.0–7.0 mg) were performed [at least in triplicate, in excess water (1:3 dry matter/water ratio)] with a Q1000 TA Instruments (New Castle, DE, USA) differential scanning calorimeter as in Bosmans et al. (2012). Onset and conclusion temperatures (in °C) and enthalpies (in J/g dry matter sample) of control and lipase containing fermented dough samples (prepared from 10.0 g flour), as well as the outer and the centre layer of partially baked 100 g bread (withdrawn from the oven at the end of oven rise) were determined with TA Instruments Universal Analysis software.

615

2.2.7. Lipid analyses Lipid extraction, purification, separation and detection of fermented dough and bread crumb were as in Gerits et al. (2013) and Gerits et al. (2014). DL samples were extracted according to Bligh and Dyer (1959). Briefly, chloroform, methanol and water were added to freeze dried DL samples in a ratio 1.0/1.0/0.9. After centrifugation (540g, 20 min, 21 °C), the upper phase containing non-lipid material (e.g. proteins) was removed whereas the lower chloroform phase with the lipid material was collected. This was repeated three times. The chloroform phases were combined and evaporated, dissolved in 0.5 ml isooctane, and finally injected (2.0 ll) for HPLC analysis. 2.2.8. Statistical analyses Statistical analyses were performed with the Statistical Analysis System software 9.3 (SAS Institute, Cary, NC, USA). For several variables, it was verified whether mean values, based on at least three individual measurements, significantly differed (significance level a = 0.05, ANOVA analysis). Pearson’s correlation coefficients (p < 0.05) for mean values were calculated.

3. Results 3.1. Specific bread volume Fig. 1 shows bread LVs for 10 and 100 g bread making, which had a specific bread LV of 4.18 ± 0.04 vs. 3.94 ± 0.10 ml/g, respectively for control bread, i.e. without added lipases or DATEM. Low levels of lipases first increased bread LV, whereas higher levels again decreased it. The impact of lipase on specific bread LV differed for both (10 g vs. 100 g) bread making scales. Optimal lipase levels for 100 g and 10 g bread making were respectively 0.5 and 1.0 mg EP/kg flour for both lipases. At the optimal dosage level, Lipopan F addition yielded the highest bread LV increase (up to 17.4% and 30.8% LV for 100 and 10 g bread making, respectively). However, with higher doses of Lipopan F, bread LV for a 100 g scale more rapidly decreased than with equal doses of Lecitase Ultra. Already at 1.0 mg EP Lipopan F/kg flour, specific bread LV was lower than that of control bread. Remarkably, the increase in specific bread LV was higher for 10 g than for 100 g bread making irrespective of the type and concentration of the lipases added. Bread LV showed no such dose dependent effect upon addition of DATEM, which at all dosage levels (0.5%, 1.0%, and 1.5% on flour basis) equally increased specific bread LV (14.6 ± 2.6% for 100 g bread making and 21.7 ± 4.3% for 10 g bread making, respectively). 3.2. Changes in dough and bread properties during fermentation and baking Fig. 2 shows dough height during fermentation in the Rheofermentometer for control dough and dough containing added lipase or DATEM. Maximum dough height when including Lecitase Ultra (45.1 ± 1.7 mm), Lipopan F (45.6 ± 2.0 mm) or DATEM (45.5 ± 1.6 mm) in the dough recipe was slightly, although not significantly (a = 0.05) higher than for control dough (42.6 ± 2.0 mm). Also, inclusion of lipases or DATEM both significantly postponed the moment at which gas cell opening occurred during baking. Indeed, cell opening for the control occurred after 83.5 ± 4.8 min, whereas that for the Lecitase Ultra dough occurred after 94.3 ± 2.4 min, that for the Lipopan F containing dough after 94.0 ± 2.0 min, and that for the DATEM supplemented dough after 89.3 ± 5.9 min. (Fig. 2). Furthermore, lipase use lowered dough tolerance (Lecitase Ultra: 58.9 ± 8.7 min, Lipopan F: 48.5 ± 5.7 min) as

616

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

Fig. 1. Increase in specific bread loaf volume (LV, %) upon addition of 0.25, 0.50, 1.00 and 5.00 mg enzyme protein (EP) Lecitase Ultra or Lipopan F control produced with no addition of lipase in its recipe.

/kg flour compared to a

Fig. 2. Dough height during fermentation as measured with the Rheofermentometer for dough samples with or without ( ) or (0.5 mg EP/kg flour) added Lecitase Ultra ( ) and Lipopan F (- - - -), and DATEM (0.5% on flour basis, ). The moment of gas cell opening is indicated by the vertical lines (upper right figure); 1, control dough; 2, dough containing DATEM; and 3, dough containing Lipopan F and Lecitase Ultra (which showed similar moments of gas cell opening).

did DATEM (61.8 ± 12.9 min). Indeed, control dough tolerance was 76.5 ± 13.7 min. Lipase or DATEM use in the straight dough bread making procedure did not impact height of dough from 100.0 g flour such as monitored by time-lapse photography during the last 36 min of fermentation (results not shown), i.e. after dough moulding and panning. Table 1 lists maximum dough height and the time at which it was reached during baking and Fig. 3 shows the evolution of dough height (in 100 g bread making) and temperature in the centre of the dough/bread (100 g vs. 10 g bread making) during baking. Optimum lipase levels (in terms of bread LV, i.e. 0.5 mg EP/kg flour) increased maximum dough height to a similar extent (ca. 5.7%) for both lipases (Table 1). DATEM increased maximum dough height during baking by 11.8% and 6.7% when used at 0.5% and 1.5% addition levels, respectively. In contrast, no such differences in the time at which the maximum dough height was reached during baking were observed. When lipases were overdosed (5.0 mg EP/kg flour), both the maximum height reached during baking and the length of the oven rise (43% and 34% for Lecitase Ultra and Lipopan F, respectively) decreased (Table 1).

3.3. Dough extensibility Kieffer extensibility analyses showed no significant effects on dough extensibility of the use of lipases (0.5 mg EP/kg flour) and this neither for 30 min nor for 120 min dough rest (results not shown). However, dough rested for 120 min with intermediate mixing (after 90 min and further resting for 30 min in the Teflon mould) needed a slightly but significantly higher amount of total work for breaking the dough strands upon use of (0.5 mg EP/kg flour) Lecitase Ultra or Lipopan F, (Table 2). In contrast, DATEM already significantly increased the resistance to extension and work input after 30 min dough rest but did not impact dough extensibility (Table 2). 3.4. Dough liquor analysis The ratio of the DL dry matter and DL total mass did not significantly differ between the different samples and averaged 18.6% (w/w). Additionally, inclusion of Lecitase Ultra and Lipopan F lipases (0.5 mg EP/kg flour) in the dough recipe (10.0 g flour) did

617

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

Table 1 Maximum height of baking dough (cm) for 100 g bread making and corresponding time (s) at which it is reached for dough with or without added lipases [0.5 or 5.0 mg enzyme protein (EP) Lecitase Ultra and Lipopan F per kg flour] or DATEM [0.5 or 1.5% on flour basis]. Averages of duplicate measurements are given with exception of the control, which was measured in triplicate.

Fig. 3. Dough height ( ) and temperature–time profile ( ) of a 100 g baking control dough and temperature–time profile of a baking 10 g (- - - -) control dough. Temperature–time profiles represent those of the centre of the bread loaf. Table 2 Dough extensibility (mm), resistance to extension (N) and total work (mJ) for control dough and dough containing DATEM (0.5%, 1.0% and 1.5% on flour basis) after 30 min of resting and for dough with and without added Lecitase Ultra or Lipopan F (0.5 mg EP/kg flour) after 90 min rest, remixing and 30 min final rest. Extensibility (mm)

Resistance to extension (N)

Total work (mJ)

Dough rest

Control DATEM 0.5% DATEM 1.0% DATEM 1.5%

53.38 53.64 56.96 52.96

0.23 0.27 0.31 0.31

9.25 (±1.21 a) 10.14 (±0.40 ab) 13.12 (±2.27 b) 12.88 (±0.94 b)

Dough rest + remixing

Control Lecitase Ultra Lipopan F

92.03 (±9.84 a) 97.30 (±11.7 a) 96.71 (±7.37 a)

(±0.56 (±3.77 (±2.70 (±0.28

not significantly impact the amounts of DL dry matter (443.1 ± 10.1 mg). However, analysis of lipid extracts of DL obtained from dough containing Lecitase Ultra and Lipopan F revealed that DL contained 1.81 (±0.22)% (w/w) and 3.09 (±0.17)% (w/w) lipids (on dry matter basis), respectively, whereas DL from control dough contained only 0.96 (±0.09)% (w/w) lipid (on dry matter basis). In contrast, DATEM (0.5% dosage level) decreased the amount of DL dry matter to 349.4 (±7.7) mg, whereas the level of lipids (including DATEM) in DL increased to 3.66 (±0.41)% (w/w) (on dry matter basis). Fig. 4 shows the corresponding HPLC profiles of the lipid extracts. Clearly, in line with the above gravimetric determinations, lipid profiles of DL from dough containing lipases or DATEM contained more lipid than that of DL from control dough (Fig. 4A).

a) a) a) a)

(±0.03 (±0.02 (±0.03 (±0.03

a) ab) b) b)

0.073 (±0.007 a) 0.071 (±0.007 a) 0.076 (±0.007 a)

5.65 (±0.51 a) 5.99 (±0.69 ab) 6.50 (±0.87 b)

Whereas triacylglycerol (TAG) levels remained fairly constant, levels of all other lipid classes increased upon addition of Lecitase Ultra or Lipopan F. Lipid classes showing the highest increase were free fatty acids (FFA), digalactosylmonoacylglycerols (DGMG) and N-acyl lysophosphotidylethanolamine (NALPE). The elution profile of DATEM itself (Fig. 4C) showed large peaks around the elution times of TAG, N-acyl phosphatidylethanolamine (NAPE) and smaller peaks around the elution times of FFA and monogalactosylmonoacylglycerols. The HPLC elution profile of the lipid extract of DL from dough with added DATEM (Fig. 4B) showed an increase in all lipid classes in dough, with highest increases in TAG, monogalactosyldiacylglycerols (MGDG), digalactosyldiacylglycerols (DGDG), NALPE and phosphatidylcholine (PC).

618

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

A

B

C

Fig. 4. Elution profiles of lipid extracts obtained from DL from dough (A) with (0.5 mg EP/kg flour) ( Lecitase Ultra; Lipopan F) or without (- - - -), (B) with ( ) or without (- - - -) (0.5% on flour basis) DATEM. Figure (C) shows the elution profile of ‘pure’ commercial DATEM. TAG, triacylglycerols; Chol, cholesterol (internal standard); FFA, free fatty acids; MAG, monoacylglycerols; MGDG, monogalactosyldiacylglycerols; MGMG, monogalactosylmonoacylglycerols; DGDG, digalactosyldiacylglycerols; NAPE, Nacyl phosphatidylethanolamine; DGMG, digalactosylmonoacylglycerols; NALPE, N-acyl lysophosphatidylethanolamine; PC, phosphatidylcholine; LPC, lysophosphatidylcholine.

3.5. Differential scanning calorimetry DSC analyses revealed similar onset and conclusion temperatures and enthalpies of starch gelatinization of fermented dough samples (10.0 g flour) prepared with or without lipase and DATEM addition (results not shown). DSC measurements of outer and inner layers of partially baked dough (100.0 g flour) at the end of the oven rise (205 s) revealed that starch gelatinization had only occurred in the outer layer (up to 82.6% lower gelatinization enthalpy than for the inner layer). 3.6. Image analyses of bread crumb Mean cell area (0.21 mm2 for control bread loaves) significantly decreased to about 0.17 mm2 upon addition of Lipopan F (all dosages), Lecitase Ultra (5.0 mg EP/kg flour) or DATEM (1.5%). In contrast, the number of cells/cm2 (216 cells/cm2 for control bread loaves) significantly increased upon use of Lipopan F (ca. 242

cells/cm2 for all dosages) or Lecitase Ultra (246 cells/cm2 for recipes with 5.0 mg EP/kg flour). 4. Discussion 4.1. Effect of lipases on bread LV in different scale bread making procedures Lipases impact bread LV in a dose dependent way (Fig. 1). Also in line with earlier observations (Koehler & Grosch, 1999), inclusion in the recipe of 0.5% DATEM increased bread LV, but no further impact was observed when higher DATEM levels were added. In contrast, too high lipase dosage levels again decreased LV. This was earlier (Gerits et al., 2014) related to (too) extensive lipid hydrolysis. Remarkably, the impact of lipase on bread LV was not only dose dependent, but the lipase to bread LV dose–response relation differed for the 10 g and 100 g bread making procedures (Fig. 1). First, increases in bread specific LVs obtained in 10 g bread

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

making upon use of lipases were relatively higher than those with similar lipase levels in 100 g bread making, irrespective of the lipase dosage. This was all the more remarkable because given the fact that specific bread LV of control bread was already significantly higher for 10 g than for 100 g scale bread making (4.18 ± 0.04 vs. 3.94 ± 0.10 ml/g, respectively). Second, optimal dosages of both lipases for 10 g bread making (1.0 mg EP/ kg flour) were higher (i.e. double) than for 100 g bread making (0.5 mg EP/kg flour). This demonstrates that lipase dosages should be carefully considered and adapted to the respective scale of bread making. Third, the small scale 10 g bread making procedure yielded bread LVs that were less susceptible to overdosing lipase. Indeed, although lipase dosage levels exceeding the optimal level decreased LV, in 10 g bread making, LVs were significantly higher than that of control bread, irrespective of the applied lipase levels tested. This was in line with our previous observations (Gerits et al., 2014). In contrast, for 100 g bread making, lipase dosage levels exceeding the optimal level decreased bread LV to such extent that the obtained breads were of lower volume than control bread. The effects observed upon addition of (different amounts of) DATEM, on the other hand, did not depend on the scale of bread making (10 g vs. 100 g). A strong correlation (R2 = 0.99) was observed between LVs of 10 g and 100 g bread making when using DATEM. This is in line with earlier observations by Shogren and Finney (1984) and Koehler (2001). Koehler (2001) demonstrated that the effects of DATEM in a 10 g bread making procedure were representative for those observed in a 300 g protocol. However, as indicated above, no such similar and/or representative observations were made in the present study for 10 g and 100 g bread making when using different levels of lipases [as deduced from the low Pearson’s correlation coefficient (R2 = 0.41)]. Shogren and Finney (1984) noted high correlations between the volumes obtained with 10 g and 100 g bread making in baking tests with flour samples with different protein levels, or upon addition of (a combination of) ascorbic acid and potassium bromate. This made the authors conclude that the 10 g bread making procedure is a good alternative for the 100 g procedure. Because bread dough mixing procedures on both scales were similar if not identical, the observed differences between both bread making procedures logically originate from the baking phase. During baking, heat is transferred by convection from the heating medium, radiation from the oven walls to the surface of the baking dough, and conduction inside the sample to its geometric centre. Logically, since convection and radiation were similar in both (small and large scale) procedures, the latter mechanism (i.e. conduction inside the sample) presumably for some reason caused the observed differences in bread specific LV. When we monitored the temperature–time profile of the centre of the baking dough (Fig. 3), for 100 g bread making, we noted that the centre dough temperature was isothermal for approximately 360 s before rapidly increasing. After ca. 870 s, the temperature reached a maximum (ca. 97°C). Dough centre temperature in 10 g bread making showed a much shorter isothermal (ca. 90 s) ‘lag’ phase. Also, dough centre temperature reached 100°C after ca. 330 s (Fig. 3), i.e. when temperature was still only 30°C in 100 g bread making. In contrast to 100 g bread making, the centre temperature exceeded 100°C when the bread was removed from the oven. It is further noteworthy that for 100 g control bread making, dough rise during baking ends at the moment (205 s, Table 1) at which dough centre temperature is still only 30°C (Fig. 3). It seems reasonable to assume that this also holds for other (lipase- or DATEM-containing) recipes tested. According to Mills, Wilde, Salt, and Skeggs (2003), the end of oven rise is caused by starch gelatinization at ca. 60°C, which ruptures the gas cells. This then leads to a porous (i.e. with interconnected gas cells) sponge struc-

619

ture which sets the bread loaf crumb structure (Mills et al., 2003). In the present case, DSC measurements of control product withdrawn after 205 s of baking revealed that (pronounced) starch gelatinization took place in the outer layer (ca. 2 cm), but not in the inner layer where starch was, hence, still semi-crystalline at that moment. This indicates that starch gelatinization in the outer layer, and the concomitant (e.g. pasting) transitions rigidifying the bread matrix, end the oven rise. Hence, the mechanisms exerted by endogenous lipids and how they are affected by lipase addition (which will be subject of the discussion below) are mainly important in dough before the structure rigidifies due to starch gelatinization and/or pasting. This may well explain why the effects observed upon addition of lipases differ between different scales of bread making, in spite of similar if not equal lipid compositions of the baked bread at different scales (results not shown). In 10 g bread making, where temperature increases faster during baking (Fig. 3), the starch gelatinizes earlier in time which leads to shorter oven rise due to earlier structure setting than in 100 g bread making. In the latter case, the mechanisms of gas cell stabilization exerted by endogenous lipids and how they are affected by lipase seem, at least for a longer time during baking, important. Our results show that when using lipases, mechanisms other than those improving gluten functionality, such as obtained with bromate, are in play. In what follows, we further elaborate on such alternative mechanisms exerted by lipase action and the concomitant altered endogenous lipid composition in bread. 4.2. Lipid functionality in bread making as related to bread LV and crumb structure The heights of the proofed control and lipase or DATEM containing panned dough samples from 100.0 g of flour immediately before baking were very comparable (results not shown). Thus, the differences in bread volume after baking resulted from differences in dough oven rise during baking. As described in the Introduction, discontinuities appear in the gluten network surrounding the gas cells already during early stages of fermentation (Gan et al., 1990). As we were wondering whether the (polar) lipids impact the thin liquid film that surrounds the gas cells, as suggested by Rouillé, Bonny, Della Valle, Devaux, and Renou (2005), rather than strengthen the gluten network, we further explored this by studying dough fermentation with the Rheofermentometer, dough extensibility with the Kieffer dough and gluten extensibility rig, and DL lipid composition. Rheofermentometer analyses revealed that the increased dough rise and resulting larger maximum height of dough samples containing added lipases or DATEM was related to postponed gas cell opening and, therefore, increased gas cell stability (Fig. 2). After partial gas cell opening, as deduced from the point of CO2 release, dough height continuously increased until it reached a maximum. At this point, the further expansion of some gas cells was compensated by loss of gas from other such cells. Later, dough height even decreased as a result of an increased loss of gas which was enhanced by the pressure exerted by the weight used in the experimental setup. Because of decreased dough tolerance as a result of addition of lipases or DATEM, the height benefit obtained during fermentation for dough containing lipases and DATEM was of short term. Indeed, after such dough had reached its maximum height, it proved to be unstable as it quickly lost height to a degree which approached the height of control dough (Fig. 2). It is of note that, in regular bread making (with no weight put on samples), both lipases tested and DATEM all increased bread LV. Kieffer extensibility analyses showed that use of DATEM significantly increased dough resistance to extension (Table 2) as also reported earlier (Koehler, 2001). DATEM use also increased the total work needed for breaking the dough strands. According to

620

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

Nash et al. (2006), total work input during Kieffer extensibility analysis strongly correlates with bread LV. Lipases at their optimal levels in terms of improving bread LV did not affect dough extensibility. This was probably because the lipid composition of fresh dough produced from a lipase containing recipe resembles that of control dough (i.e. without added lipases). Indeed, lipase use revealed only detectable changes in lipid composition after at least 10 min of fermentation (Supplementary Fig. 1), i.e. long after the gluten network has been developed. In contrast, when adding DATEM at the start of mixing, like the endogenous lipids, it can immediately interact with and/or become trapped in the gluten network. By doing so, DATEM acts as dough strengthener (Ahmad et al., 2014) and thus indirectly stabilizes the gas cells. We hypothesise that dough rheology is probably only impacted by those lipid components readily available at the start of mixing and, thus, not by the ones released by lipases during fermentation. When we remixed control and lipase containing dough samples that had rested for 90 min and in which thus, in the latter case, the lysolipids and FFA formed by lipase action were able to interact with the gluten proteins during the second mixing step, the total work input needed to break dough strands of dough increased in the order control, Lecitase Ultra containing dough and Lipopan F containing dough (Table 2). This observation would seem to be in line with the above proposed hypothesis. Probably, the impact of lipases and especially that of hydrolysed lipids has to be found in the direct or secondary stabilization mechanism of the liquid film(s) surrounding the gas cells. We next analysed DL lipid levels and compositions (Fig. 4). Lipase inclusion in a dough recipe increased the levels of polar lipids in DL (Fig. 4A). This was not only because of increased levels of ‘lysolipids’ but unexpectedly also of all other polar lipids. The latter probably originates from formation of hexagonal I mesophases by the lysolipids which emulsify other lipids (Gerits et al., 2014). Higher levels of polar lipids in DL indicate a higher availability for and potential accumulation of these components at the gas cell interface of the gas cells. Both lipases mainly increased the levels of FFA, DGMG and NALPE. As described earlier (Gerits et al., 2014), NALPE forms the lamellar mesophase which forms condensed monolayers at the gas cell interface. Such monolayers are favourable in terms of gas cell stabilisation. FFA and DGMG, on the other hand, promote formation of the hexagonal I mesophase, which probably emulsifies the deleterious (non-polar) lipids. We earlier found only small and negligible differences in the lipid compositions of fermented dough when prepared with use of Lecitase Ultra or Lipopan F (0.5 mg EP/kg flour) despite clear differences in their lipase activity and in bread LV responses (Gerits et al., 2014). However, the present data show clear and significant differences in the amount of polar lipids in DL from dough made with these lipases. This reveals a different distribution of the lysolipids formed by the lipases in dough and, hence, a difference in action exerted by the enzymes. This can explain the differences in specific bread LV observed when including them in bread making recipes. Whilst lipase addition did not impact the amount of DL dry matter, DATEM addition decreased it. Possibly, this resulted from the observed increased gluten network strength (cfr. supra) and, hence, from increased interactions between gluten proteins. As for dough prepared with lipase addition, DL of dough with added DATEM (Fig. 4B) increased the levels of all lipid classes, even of those that did not appear in the HPLC elution profile of the DATEM used (Fig. 4C). It would seem that DATEM preferentially interact with gluten proteins, thereby limiting interactions and/or entrapment of other (endogenous) lipids with the gluten proteins. That the largest increases were noticed for components eluting at elution times similar to those of TAG and NAPE is not surprising, as these components were originally present in the DATEM sample (Fig. 4C). Their higher levels in DL from DATEM containing dough

than in that from control dough can therefore be related to DATEM itself. The increased levels of other lipid classes (mainly DGDG but also NALPE and PC) in DL indicate that, in the presence of DATEM, they compete with the latter for interaction with gluten proteins. Indeed, the DATEM component which, if present in a sample, would elute right before NAPE (14 min), was no longer present in DL from dough which contained DATEM. This observation indicated that part of the DATEM was not recovered in DL and presumably associated with the gluten protein fraction. The presence of DATEM and concomitant increase in the amounts of DGDG, NALPE and PC in DL can increase dough gas cell stability because DATEM (Krog, 1981) but also DGDG, NALPE and PC (Selmair, 2010) promote the lamellar mesophase which favours gas cell stability due to formation of condensed monolayers at the interface. The present results suggest the existence of the thin liquid film in order to stabilize the gas cells and demonstrated the importance of analysing DL (polar) lipids composition in relation to bread LV. They also demonstrated that optimal levels of lipases in terms of bread making (0.5 mg EP/kg flour) or DATEM (0.5% and 1.5%, on flour basis) did not extend the oven rise in time during baking (Table 1). This was in line with our observations that neither lipases (indirectly) nor DATEM (directly) affect starch gelatinization. This logically demonstrates that the increased height at the end of the dough rise was due to an increased dough expansion rate for dough containing optimal lipase or DATEM levels than for control dough. At the same time, we noted that use of lipases or DATEM led to later gas cell opening during baking. In this context, it is of note that the Rheofermentometer data also pointed to increased gas cell stability as a result of the use of DATEM or lipases. More research will be need to determine whether such data are predictive for increased gas cell stability during baking. However, for 100 g bread making the highest levels (5.0 mg EP Lecitase Ultra and Lipopan F per kg flour) of lipases tested induced earlier end of oven rise (118 and 137 s respectively, vs. 205 s for the control) (Table 1). As mentioned above, thorough breakdown of the lipids, especially of those promoting the formation of a lamellar mesophase, decreases gas cell stability or even completely disrupts gas cells. Hence, in this case, gas cell stabilization seemed to have failed before the starch gelatinized and rigidified the structure. It may well be that the cause of gas cell disruption may be their growth and, hence, the growing interfacial area to cover, in combination with the mass of the baking dough exerting pressure on the gas cells. The above brings us to the inconsistency in LV responses to lipase addition in 10 g and 100 g scale bread making of which one of the reasons may well be the different masses of baking dough exerting pressure on the gas cells. Even more importantly, it would seem that the resultant difference in time–temperature profile in baking (see Fig. 3) has implications for the time during which the liquid film surrounding the gas cells needs to be stabilized. In 100 g scale bread making, the film evidently needs to be stable for a longer time than in 10 g bread making. As mentioned before, the centre temperature of a baking dough from 100 g of flour is still only 30°C when oven rise is finished (Fig. 3). Since the outer layers limit expansion of the inner layers, the overall expansion capacity of the gas cells in the centre of the dough seems to be only partially exploited. In addition, it seems that the phenomena responsible for increasing bread LV that are based on improving gas cell stability also improve crumb structure during further heating of the inner bread layers. Indeed, Lipopan F and Lecitase Ultra decreased mean gas cell area and, logically, increased the number of cells/cm2, yielding a finer and more uniform crumb structure. After ca. 650 s of baking, the centre of the dough reached 60°C (Fig. 3), which allowed the starch to gelatinize and rigidified the crumb structure. This completed the transition from semi-solid foam to solid sponge and brought an

L.R. Gerits et al. / Food Chemistry 172 (2015) 613–621

end to gas cell stabilizing mechanisms and the role of lipids in terms of increasing bread LV and crumb structure. 5. Conclusion The effect of lipases on bread LV depends on the applied dosage and the (scale of the) bread making procedure. Optimal concentration for 100 g bread making (0.5 mg EP/kg flour) of the two lipases tested (Lipopan F and Lecitase Ultra) was half of that for 10 g bread making (1.0 mg EP/kg flour). Whilst DATEM did, lipases did not impact dough rheology because the hydrolysed lipids were released only to a significant degree during fermentation. A prerequisite for lipid(-like) components impacting dough rheology seems that they need to be present at the start of mixing. DL recovered from dough prepared with lipases contained higher lipid levels than that of control dough. In fact, the level of all polar lipid classes in DL increased upon lipase addition. Lipopan F induced a larger increase in the portion of polar lipids in DL. This explained its higher impact on bread LV than that of Lecitase Ultra. Inclusion of DATEM in a dough recipe also increased the levels of the different lipid classes in isolated DL. This was probably related to competition of (endogenous) lipid classes with DATEM for interacting with gluten proteins. Differences in bread LV during baking arose from differences in oven rise. Too high lipase dosages induced earlier oven rise termination due to gas cell coalescence. Optimal lipase dosages did not alter oven rise duration. The data demonstrated that oven rise strongly depends on starch gelatinization in the outer baking dough layers. This transformed their structure from a liquid foam to a rigid sponge and prevented further expansion of inner dough. Nevertheless, the phenomena responsible for proper gas cell stabilization such as the composition of the thin liquid film not only impacted bread LV, but also contributed to uniform and fine crumb structure. Acknowledgements The authors are grateful to H. Van den Broeck (at this laboratory) for technical assistance and to Novozymes (Bagsvaerd, Denmark) for providing the enzymes. This work is part of the Methusalem programme ‘Food for the future’ (2007–2014). B. Pareyt acknowledges the Research Foundation – Flanders (FWO – Vlaanderen, Brussels, Belgium) for a position as postdoctoral researcher. J.A. Delcour is W.K. Kellogg Chair in Cereal Science and Nutrition at the KU Leuven. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foodchem. 2014.09.064. References AACC-I. Approved Methods of Analysis, 11th Ed. Methods 44-19.01 and 54-40.02. Mixograph Method. Approved November 3. (1999). AACC International, St. Paul, MN, USA. http://dx.doi.org/10.1094/AACCIntMethod-54-40.02; http:// dx.doi.org/10.1094/AACCIntMethod-44-19.01.6. Ahmad, A., Arshad, N., Ahmed, Z., Bhatti, M. S., Zahoor, T., Anjum, N., et al. (2014). Perspective of surface active agents in baking industry: an overview. Critical Reviews in Food Science and Nutrition, 54, 208–224. AOAC. (1995). Official Methods of Analysis, 16th 402 ed. Association of Official Analytical Chemists. Bligh, E. G., & Dyer, W. J. (1959). A rapid method of total lipid extraction and purification. Canadian Journal of Biochemistry and Physiology, 37, 911–917. Bosmans, G. M., Lagrain, B., Deleu, L. J., Fierens, E., Hills, B. P., & Delcour, J. A. (2012). Assignments of proton populations in dough and bread using NMR relaxometry

621

of starch, gluten and flour model systems. Journal of Agriculture and Food Chemistry, 60, 5461–5470. Chung, O. K., Ohm, J. B., Ram, M. S., Park, S. H., & Howitt, C. A. (2009). Wheat Lipids. In K. Khan & P. R. Shewry (Eds.), Wheat: Chemistry and Technology (fourth ed., pp. 363–399). St. Paul: AACC International. De Maria, L., Vind, J., Oxenboll, K. M., Svendsen, A., & Patkar, S. (2007). Phospholipases and their industrial applications. Applied Microbiology and Biotechnology, 74, 290–300. Delcour, J. A., & Hoseney, R. C. (2010). Principles of Cereal Science and Technology (3 ed.). St. Paul, MN, US: American Association of Cereal Chemists. Eliasson, A. C. (1983). Differential scanning calorimetry studies on wheat starchgluten mixtures. 2. Effect of gluten and sodium stearoyl lactylate on starch crystallization during ageing of wheat starch gels. Journal of Cereal Science, 1, 207–213. Finney, K. F. (1984). An optimized, straight dough, bread-making method after 44 years. Cereal Chemistry, 61, 20–27. Gan, Z., Angold, R. E., Williams, M. R., Ellis, P. R., Vaughan, J. G., & Galliard, T. (1990). The microstructure and gas retention of bread dough. Journal of Cereal Science, 12, 15–24. Gan, Z., Ellis, P. R., & Schofield, J. D. (1995). Gas cell stabilization and gas retention in wheat bread dough. Journal of Cereal Science, 21, 215–230. Gerits, L. R., Pareyt, B., Decamps, K., & Delcour, J. A. (2014). Lipases and their functionality in the production of cereal-based food systems. Comprehensive Reviews in Food Science and Food Safety, 13, 978–989. Gerits, L. R., Pareyt, B., & Delcour, J. A. (2013). Single run HPLC separation coupled to evaporative light scattering detection unravels wheat flour endogenous lipid redistribution during bread dough making. LWT – Food Science and Technology, 53, 426–433. Gerits, L. R., Pareyt, B., & Delcour, J. A. (2014). A lipase based approach for studying the role of wheat lipids in bread making. Food Chemistry, 156, 190–196. Gonzales-Barron, U., & Butler, F. (2006). A comparison of seven thresholding techniques with the k-means clustering algorithm for measurement of breadcrumb features by digital image analysis. Journal of Food Engineering, 74, 268–278. Kieffer, R., Wieser, H., Henderson, M. H., & Graveland, A. (1998). Correlations of the breadmaking performance of wheat flour with rheological measurements on a micro-scale. Journal of Cereal Science, 27, 53–60. Koehler, P. (2001). Study of the effect of DATEM. 3: synthesis and characterization of DATEM components. LWT – Food Science and Technology, 34, 359–366. Koehler, P., & Grosch, W. (1999). Study of the effect of DATEM. 1. Influence of fatty acid chain length on rheology and baking. Journal of Agricultural and Food Chemistry, 47, 1863–1869. Krog, N. (1981). Theoretical aspects of surfactants in relation to their use in breadmaking. Cereal Chemistry, 58, 158–164. MacRitchie, F. (1976). The liquid phase of dough and its role in baking. Cereal Chemistry, 53, 318–326. MacRitchie, F., & Gras, P. W. (1973). The role of flour lipids in baking. Cereal Chemistry, 50, 292–302. Mills, E. N. C., Wilde, P. J., Salt, L. J., & Skeggs, P. (2003). Bubble formation and stabilization in bread dough. Trans IchemE, 81, 189–193. Nash, D., Lanning, S. P., Fox, P., Martin, J. M., Blake, N. K., Souza, E., et al. (2006). Relationship of dough extensibility to dough strength in a spring wheat cross. Cereal Chemistry, 83, 255–258. Otsu, N. (1979). A threshold selection method from gray-level histograms. IEEE Transactions on Systems Man and Cybernetics, 9, 62–66. Pareyt, B., Finnie, S. M., Putseys, J. A., & Delcour, J. A. (2011). Lipids in bread making: sources, interactions, and impact on bread quality. Journal of Cereal Science, 54, 266–279. Primo-Martin, C., Hamer, R. J., & de Jongh, H. H. J. (2006). Surface layer properties of dough liquor components: Are the key parameters in gas retention in bread dough. Food Biophysics, 1, 83–93. Rouillé, J., Bonny, J. M., Della Valle, G., Devaux, M. F., & Renou, J. P. (2005). Effect of flour minor components on bubble growth in bread dough during proofing assessed by magnetic resonance imaging. Journal of Agricultural and Food Chemistry, 53, 3986–3994. Salt, L. J., Wilde, P. J., Georget, D., Wellner, N., Skeggs, P. K., & Mills, E. N. C. (2006). Composition and surface properties of dough liquor. Journal of Cereal Science, 43, 284–292. Selmair, P.L. (2010). Structure–Function Relationship of Glycolipids in Breadmaking, PhD dissertation. Technische Universität München, München (Germany). Shogren, M. D., & Finney, K. F. (1984). Bread-making test for 10-grams of flour. Cereal Chemistry, 61, 418–423. Sroan, B. S., Bean, S. R., & MacRitchie, F. (2009). Mechanism of gas cell stabilization in bread making. I. The primary gluten–starch matrix. Journal of Cereal Science, 49, 32–40. Sroan, B. S., & MacRitchie, F. (2009). Mechanism of gas cell stabilization in breadmaking. II. The secondary liquid lamellae. Journal of Cereal Science, 49, 41–46. Turbin-Orger, A., Boller, E., Chaunier, L., Chiron, H., Della Valle, G., & Réguerre, A. L. (2012). Kinetics of bubble growth in wheat flour dough during proofing studied by computed X-ray micro-tomography. Journal of Cereal Science, 56, 676–683. van Vliet, T., Janssen, A. M., Bloksma, A. H., & Walstra, P. (1992). Strainhardening of dough as a requirement for gas retention. Journal of Texture Studies, 23, 439–460. Vanhamel, S., Van den Ende, L., Darius, P. L., & Delcour, J. A. (1991). A volumeter for breads prepared from 10 g of flour. Cereal Chemistry, 68, 170–172.