Effects of the addition of a meal deriving from toasted durum wheat kernels on dough properties and spaghetti cooking behavior
Efectos de la adición de una harina derivada de granos tostados de trigo duro en las propiedades de masa y comportamiento para el cocido de espaguetis

Abstract

The aim was to study the effects of toasting of durum wheat kernels on the properties of dough and spaghetti produced with semolina and increasing amounts of the meal deriving from toasted kernels. The replacement of semolina with the toasted meal caused a decrease in the dough strength and an increase of its tenacity already at a substitution level of 100 g kg⁻¹. Pasta made with 100% semolina or with a mixture of semolina and 200 g kg⁻¹ of toasted durum wheat meal appeared as the best, being the less compressible and showing the highest values of cohesiveness. Water sorption curves were successfully described through the Peleg's equation. The application of response surface regression followed by stepwise regression allowed to write one equation for each of the textural properties measured. Gluten index was the best predictor of compressibility, cohesiveness, and gumminess whereas elasticity was better predicted on the basis of both gluten content and gluten index. Observed and predicted values were highly correlated with R ² ranging from 0.707 (compressibility) to 0.867 (gumminess).

El objetivo fue investigar los efectos de tostado de granos de trigo duro en las propiedades de masa y espaguetis producidos con semolina y cantidades cada vez mayores de la harina derivada de granos tostados. La sustitución de semolina por la harina tostada originó una disminución de la fuerza de la masa y un incremento de su tenacidad ya a un nivel de sustitución de 100 g kg⁻¹. La pasta hecha con 100% semolina o con una mezcla de semolina y 200 g kg⁻¹ de harina tostada de trigo duro (TDWM) se mostró como la mejor, siendo la menos compresible y mostrando los valores más altos de cohesión. Las curvas de absorción de agua se describieron exitosamente a través de la ecuación de Peleg. La aplicación de la regresión de superficie de respuesta seguida de regresión progresiva permitió escribir una ecuación para cada una de las propiedades texturales medidas. El índice de gluten fue el mejor predictor de compresibilidad, cohesión y gomosidad, mientras la elasticidad fue predicha mejor en base a ambos contenido de gluten e índice de gluten. Valores observados y predichos fueron altamente correlacionados con R² entre 0,707 (compresibilidad) y 0,867 (gomosidad).

Keywords:

• cooking behavior
• dough
• spaghetti
• toasted durum wheat meal

Palabras clave:

• comportamiento de cocinado
• masa
• espaguetis
• harina tostada de trigo duro

Introduction

Pasta is a generic term for a popular food obtained by kneading, extrusion/lamination, cutting, and eventually drying of a mixture of flour and water. Semolina, obtained by the milling of durum wheat kernels, is the favorite meal for pasta production due to quality and amount of its starch and gluten (Baiano et al., 2008; Sissons, 2008), a cross-link of protein network that develops during the pasta-making process when the two native protein groups (glutenin and gliadin) come into contact with water.

Drying is the operation that mainly affects pasta quality, acting on both protein and starch fractions. Drying increases the protein molecular size since the heating of glutenin above 55 °C or gliadins above 70 °C induces disulfide/sulfydryl exchange reactions (Payne, Jackson, & Holt, 1984; Schofield, Bottomley, Timms, & Booth, 1983). Temperatures around 65 °C influence particularly the S–S structure of high-molecular weight (HMW) albumins (and perhaps their linkage to glutenin oligomers) whereas higher temperatures affect low-molecular weight (LMW) albumins and gliadins (Lavelli, Guerrieri, & Cerletti, 1996).

Processing of some traditional cereal by-products includes a kernel roasting before milling. Examples are represented by “Kavut”, a Turkish product made of whole wheat meal and barley flour (Karaoglu & Kotancilar, 2006) and by an Indian product made of wheat and other roasted flours mixed with water and alternatively sugar, jaggery, or salt (Vasundhara & Parihar, 1979). When thermal treatments are applied on kernels, the changes in structure and availability of starch and proteins could be responsible for a number of serious concerns affecting pasta quality. Heat treatments cause the formation of new non-starch linkages, mainly β-1→4 and 1→6 anomers (Laurentín, Cárdenas, Ruales, Pérez, & Tovar, 2003) that, by reducing the enzymic access of several adjacent α-glucosidic bonds, determine a decrease of starch digestibility and the formation of the so called “resistant starch”. Thermal treatments are also responsible for protein denaturation, causing conformational changes in gliadins and low-molecular weight albumins and globulins (Guerrieri, Alberti, Lavelli, & Cerletti, 1996; Schofield et al., 1983), formation of aggregates in glutenins and other changes of the secondary and tertiary structures (Michon, Wang, Ferrasson, & Gueguen, 1999; Tilley et al., 2001) through sulfhydryl (SH)–disulfide interchange reactions, strong cross-linking bonds and the exposure of hydrophobic groups on the surface making the proteins insoluble and preventing their successive hydration and dough formation (Riganakos & Kontominas, 1994).

Literature concerning the effects of heat treatment of wheat kernels on dough and pasta or bread is lacking. Riganakos and Kontominas (1994) constructed the water sorption isotherms of heat-treated wheat flour and their data showed a reduced water uptake. Miyazaki and Morita (2005) observed that elasticity of the dough added with high moisture-temperature treated meal (HMT-M) decreased as compared with the control and that HMT-M hardly swelled and gelatinized in the dough. Thorvaldsson, Stading, Nilsson, Kidman, and Langton (1999) studied the influence of heating rate on rheology and structure of heat-treated pasta dough and found that faster heating resulted in higher Young's modulus values and that energy required to cause a fracture in the fast-heated samples was more than for the slowly heated samples. In a study on the effects of heat-moisture treated maize starch on dough and bread properties. According to Xue, Fukuoca, and Sakai (2010), gelatinization degree increased slowly during the microwave intermittent heating of wheat flour dough. Pancakes baked with heat-treated wheat flour increased in springiness and decreased in gumminess (Seguchi, 2006).

An interesting application of toasted whole cereal grains is represented by a pasta product made of toasted wheat meal and recognized as traditional by the Italian Laws (Baiano et al., 2008). In the past, it was produced using a mixture of variable amounts of semolina (and/or soft wheat flour), and a whole meal deriving from durum wheat kernels of ears that escaped harvesting and undergone the stubble burning. As a consequence of burning, a meal from toasted wheat kernels shows a typical brown color and a roasted, aromatic flavor attributable to volatiles compounds generated by oxidation and the Maillard reaction. At present, this flour is obtained by toasting kernels previously harvested and threshed (Baiano et al., 2007).

Starch indigestibility caused by toasting could be a desirable characteristics of this type of pasta (Cho, Prosky, & Dreher, 1999; Rodwell & Schlenker, 2002). In fact, it has been shown that resistant starch has physiological functions similar to those of dietary fiber (Eerlingen & Delcour, 1995) and nutritional applications in the treatment of glycogen storage diseases and diabetes mellitus (Thomas & Atwell, 1999). Baiano et al. (2008) applied chemical, mechanical, sensory, and image analyses to two different types of toasted pasta, one made with a mixture of semolina (80%) and a meal deriving from toasted kernels (20%) and the other made of semolina (40%), a meal deriving from toasted kernels (40%), soft wheat flour (20%), and eggs (the latter in place of water). It was highlighted that the presence of toasted meal induced many structural and functional changes with respect to pasta made of 100% semolina but also that, among the formulations tested, the best one was the mixture of semolina and whole meal deriving from toasted durum wheat. Lamacchia, Baiano, Lamparelli, La Notte, and Di Luccia (2010), studied the changes in durum wheat kernel and pasta proteins induced by toasting and drying processes and found that the replacement of semolina with 5% and 10% of toasted durum wheat flour did not significantly change the unextractable polymeric proteins (UPP) when compared with spaghetti made with 100% durum semolina. On the other hand, the replacements of semolina with 15–30% toasted durum wheat meal (TDWM) showed significant increase in UPP.

In the light of these results, it seemed advisable: to increase the insight on the effects of toasting of durum wheat kernels on physical and chemical characteristics of meal; investigate the changes induced in dough properties and spaghetti cooking behavior by addition of increasing percentages of this meal; and find the better formulations among those usually applied in production of this traditional pasta. The possibility to use the characteristics of meals and doughs as predictors of the mechanical properties of pasta was also evaluated.

Materials and methods

Raw materials and pasta-making

Semolina and TDWM from the same commercial blend of Italian durum wheat grains (Simeto, Ciccio, Arcangelo) were supplied by Molino Daddario Antonia (Cerignola, Foggia, Italy). TDWM was a wholemeal obtained by the grinding durum wheat grains previously submitted to toasting (grains were moistened and arranged in 1 cm-layers on tin pans placed into direct contact with wood-fires, at 250 °C for 35–45 s, under continuous stirring), cleaning, and dampening. In order to apply the statistical analysis to the experimental data, the toasting procedure was repeated three times at least and the obtained TDWM batches were kept separately and used to prepare different doughs. Semolina and TDWM granules had the same size, in order to avoid differences in their hydration level that would have caused a color heterogeneity on the surface of pasta for the presence of granules not sufficiently hydrated and thus remained in their crystal form.

Spaghetti was produced in a 2 kg-pilot plant made of a mixing tank, an extruder and a dryer (NAMAD, Rome Italy). Starting from the meal moisture, tap water was added to the meals in the mixing tank in order to obtain a dough water content of 440–450 g kg⁻¹. The meal blends, obtained by replacing 0, 50, 100, 150, 200, and 300 g kg⁻¹ of semolina with the meal deriving from toasted durum wheat, were named semolina, 5% TDWM, 10% TDWM, 15% TDWM, 20% TDWM, and 30% TDWM-semolina, respectively. The product coming from the mixing tank was driven into the extruder barrel by a compression screw. The following extrusion conditions were applied: temperature 50 ± 5 °C; kneading time 15 min; pressure 7,092,750 ± 1,013,250 Pa (as a function of the specific formulation); vacuum degree 93333 Pa. A Teflon die-plate was used. Spaghetti was dried at 50 °C for 16 h. This temperature was applied from the beginning to the end of the drying cycle and then pasta was equilibrated to room conditions. The relative humidity of the hot air was about 50%. The diameter of the obtained spaghetti was about 1.70 ± 0.03 mm. Each meal blend was produced in an amount sufficient to repeat the pasta-making procedure three times, at least.

Meals and doughs were analyzed in order to highlight possible relationships among their characteristics and those of the final products. The analyses on raw materials, doughs, and spaghetti were performed as described in the following sections.

Analysis of semolina, toasted durum wheat meal and their blends

Moisture was measured according to the method 44-15A of the American Association of Cereal Chemist (2003). Water activity was measured through an Aqualab, Model Series 3 TE (Decagon, WA, USA). Total dietary fiber (TDF) was quantified according to the method of Prosky, Asp, Schweizer, De Vries, and Furda (1988) using a commercial kit supplied by Megazyme (Bray, Ireland). The evaluation of the particle size distribution was made by sieving 100 g samples on a set of sieves arranged from the largest (500 μm) to the smallest (90 μm) opening. The set of sieves was placed on the Ro-Tap sieve shaker (Tyler Industrial Products, OH). The duration of sieving was 10 min. After sieving, the mass retained on each sieve was weighed. The ash determination was performed according to the method 08-01 of the American Association of Cereal Chemist (2003).

Gluten index and gluten content were determined according to the methods 158 of the International Association for Cereal Science and Technology (1995) and 38-12A of the American Association of Cereal Chemist (2003), respectively.

Concerning colorimetric measurements, b* value (yellow/blue balance) was measured through a tri-stimulus colorimeter (Chromameter-400, Minolta, Osaka, Japan). The colorimeter was calibrated on a standard white tile (L* = 93.5, a* = 1.0, b* = 0.8) before each series of measurements. Meals were uniformly ground before analysis, in order to prevent the different particle sizes from affecting the color measurement.

Analysis of the doughs

The dough alveographic indices (P, resistance to extension or tenacity, expressed as Pa; L, extensibility, expressed as mm; W, deformation energy or flour strength, expressed as Joule) were determined according to the method 54-30 of the American Association of Cereal Chemist (1993). From each of the doughs, five curves were obtained.

Analysis of the spaghetti samples

Pasta color (L*, a*, b*) of multiple layers of parallel spaghetti strands (Wood, 2009) were measured using a tri-stimulus colorimeter.

The moisture content and water activity of ground spaghetti were performed as described for meals.

The water sorption kinetics were determined using 40.5 ± 0.2 mm long samples. Culture tubes containing about 9 mL of distilled water were equilibrated at 100 ± 0.5 °C in a thermostatic bath. Afterwards, the spaghetti strands were immersed into the tubes, one for each tube. At given times (each 5 s for the first 30 s; each 30 s until 10 min) the samples were removed from the tube, rapidly blotted, and weighed (Sartorius mod. BP 211 D, Goettingen, Germany). The optimal cooking time, corresponding to the time at which the core disappeared as a consequence of progressive hydration and gelatinization of starch granules, was measured by squeezing pasta between two glass plates (Del Nobile & Massera, 2000).

Total organic matter (TOM) lost into the cooking water was determined by cooking spaghetti in distilled water equilibrated at 100 ± 0.5 °C in a thermostatic bath. The ratio water/pasta 20:1 (w/w) was chosen in order to avoid that water represented a limiting factor for pasta hydration. At given times, cooking water was withdrawn, dried at 105 °C and weighed, in order to evaluate the presence of dried matter passed from pasta. The cooking losses were expressed as g kg⁻¹ of the raw spaghetti.

For the evaluation of amylose leached into the cooking water, spaghetti was prepared as for the analysis of the total cooking matter. Successively, 1 mL of cooking water (previously centrifuged at 15,000g for 10 min., 20 °C) was put in 20-mL flask, added with 1 mL of a iodine solution (40 g KI and 10 g I₂ L⁻¹ of distilled water) and made up to volume with distilled water. Absorbance at 600 nm was measured with a Beckman DU 640 spectrophotometer (Beckman Instruments, Inc., Irvine, CA) using 1 mL of iodine solution diluted to 20 mL as a blank.

Spaghetti samples were also submitted to the texture profile analysis for the evaluation of their mechanical properties. Spaghetti strands at the optimal cooking time were submitted to a double compression cycle by means of an Instron Universal Testing Machine (model 5567; Canton, MA, USA, maximum load 999 N). Tests were carried out at a cross-head speed of 25 mm s⁻¹. Two peaks were considered, the first at a sample deformation of 30% and the other one at a deformation of 90% that determined the sample breakdown. From the resulting force-time curve, the following mechanical parameters were determined: compressibility (an indirect measure of firmness that is the maximum load detected during the first compression cycle), cohesiveness (ratio between the work performed during the second compression cycle and the work performed during the first compression cycle, that is a measure of the strength of the internal bonds), elasticity (return of the compressed strand to the initial shape during the time within the end of the first cycle and the start of the second cycle), and gumminess (hardness x cohesiveness, energy requested to break up a semi-solid food until it is ready for swallowing) (Bourne, 1978; Martinez, Salmerón, Guillén, & Casas, 2004; Szczeniak, Brandt, & Friedman, 1963; Szczeniak, 1963).

Statistical analysis

Analyses of moisture, water activity, and ash were repeated at least three times. Colorimetric measurements were performed 10 times for each sample. Five water sorption kinetic curves were constructed for each sample. Analyses concerning TOM and amylose leached into the cooking water were repeated five times for each sample. For each sample, 10 spaghetti strands were submitted to the mechanical analysis.

Means and standard deviations were determined. Correlation analysis, ANOVA analysis, and the Holm test at a confidence level of 95% (P < 0.05) were performed by means of the Kaleidagraph Statistical Software (ver. 3.6.2; Synergy Software, Reading, PA, USA).

In order to determine the best single predictors of the spaghetti mechanical properties, simple correlations analyses were performed between gluten content, gluten index, or alveographic indices and textural properties and the relative determination coefficients (R ²) and standard errors (SEs) were reported. Statistical methods were also employed to check if the pasta textural properties could be predicted by the use of an algorithm incorporating gluten content, gluten index, and the dough rheological parameters. Models were calculated by means of response surface regression at first to test accuracy of prediction, and by means of stepwise regression to find the best subgroup of tested variables with the highest multiple determination coefficients (R ² adjusted). Both statistics were calculated on P < 0.05 using the software Winstat ver. 5.1 (Statsoft, Tulsa, USA).

Results and discussion

Meal and dough properties

Due to the drying effects of the toasting operation, semolina and TDWM showed significantly different water content (83.4 ± 1.0 and 67.0 ± 2.0 g kg⁻¹, respectively) and water activity (0.44 ± 0.00 and 0.41 ± 0.00, respectively). Also, the TDF content detected in semolina was equal to 28.8 ± 0.4 g kg⁻¹ on a dry basis whereas it was significantly higher in the whole meal extracted by toasted wheat kernel (98.3 ± 0.8 g kg⁻¹ on a dry basis).

The particle size distribution of semolina and TDWM was rather fine being distributed as follows: 26.5 ± 0.3% < 100 μm; 100 μm  ≤ 22.3 ± 0.9% < 160 μm; 160 μm  ≤ 15.1 ± 0.4% < 200 μm; 200 μm ≤ 36.0 ± 1.8% < 355 μm; 0.04 ± 0.01%  ≤ 355 μm. A fine granulometry is one of the factors that favors the formation of Maillard products in pasta as a consequence of the presence of low- to medium-molecular weight carbohydrates (Landi, 1995). Nevertheless, in the case of a meal derived from toasted kernels, the products of Maillard reaction had been already generated and the presence of particle of low to medium size is necessary to counterbalance, through the hydration of damaged starch, the low-water binding capacity of denatured proteins (Guerrieri & Cerletti, 1996).

The other characteristics of semolina, TDWM and their mixtures are reported in Supplementary Table 1. The presence of increasing amounts of TDWM in the mixtures accounted for significant increases of the ash content whereas the amount of gluten formed decreased as the percentage of substitution increased. The decreasing of gluten formation depended on several factors, including interference of bran on the dough structure (Baiano, Romaniello, Lamacchia, & La Notte, 2009) and the irreversible changes caused by toasting in the secondary and tertiary structure of proteins. The water unextractable solids interfere with gluten formation in a direct way, through interaction with gluten particles and also in an indirect way, by competing for water and thus changing conditions for gluten development (Wang et al., 2003). Heat treatments induce the exposure of hydrophobic groups that, by preventing protein hydration, reduced the amount of gluten formed (Riganakos & Kontominas, 1994). The gluten index values of TDWM and its mixtures with semolina were higher than those measured on doughs made of semolina only. In fact, the denaturation caused by heat treatment comes through the formation of disulfide linkages and cross-linking bonds (dityrosine formation) that strengthen the gluten network by increasing the molecular size of aggregates and thus the percentage of unextractable proteins (Lamacchia et al., 2010; Wagner & Anon, 1990; Weegels, de Groot, Verhoek, & Hamer, 1994). The yellow index (b*) of the mixtures dramatically decreased as the percentage of TDWM increased due to the presence of brown colored bran, non-volatile colored compounds of intermediate molecular mass, and brown substances of high molecular mass formed during Maillard reaction (Carpenter & Booth, 1973) consisting of unsaturated, nitrogenous polymers, and copolymers (Ames, 1992).

The dough alveographic indices are reported in Supplementary Table 1. The UNI 10709 Italian Technical Rule (1998) established three classes of quality for durum wheat on the basis of their technological characteristics. For example, durum wheat is included into the first, second, or third class if its alveographic W value is ≥250, 180, and 100, respectively. According to this index, the semolina used in this research belongs to the third class and, thus, it was not very suitable for the production of long pasta such as spaghetti. The replacement of semolina with the meal deriving from toasted durum wheat kernels did not cause further detrimental changes of the pasta quality only up to an amount of 200 g kg⁻¹. In order to be extruded, a dough should have a certain equilibrium between tenacity (P) and elasticity (L) and should oppose a certain resistance to extension (P>L) (UNI 10709, 1998). Nevertheless, a P/L ratio higher than 2.5 could be an index of an excessively hard dough that can easily break up giving rise to a pasta extremely brittle and subjected to breakage already at the exit of the die. On the basis of this index, mixtures containing more than 50 g kg⁻¹ of TDWM would seem less suitable for the production of long pasta. It was impossible to determine the alveographic indices on dough made with 500 and 1000 g kg⁻¹ of TDWM since they did not give rise to a workable dough.

During years of research, many empirical equations have been produced in order to predict pasta cooking quality. Increasing reliance on physical, chemical, and technological variables to provide estimates of product quality has been occurring in research laboratories (Alary & Kobrehel, 2006; Cole, 2007; D'Egidio, Mariani, Nardi, Novaro, & Cubadda, 1990). Pasta resistance to cooking stems mainly from a series of factors such as a high protein content (or a high content in nitrogenous substances), high gluten strength and elasticity, and a pasta drying cycle at temperatures above 80 °C so to prevent starch release from the structure. Researchers operating in pasta industry have developed a mathematical formula (reliability equal to 95%) based on multiple correlation of parameters routinely measured in company laboratories such as gluten content and alveographic indices and on drying temperature (Landi, 1995). According to such equation, the pasta cooking quality (also defined pasta value, PV, and expressed as a dimensionless number) was calculated as follows:

  where

  PV > 0;

K is an dimensionless factor depending on drying temperature, whose value decreases as the drying temperature increases;

  GluC is the dry gluten content (% d.b.);

W (J) and P/L (Pa mm⁻¹) are the alveographic indices.

 The pasta value has been calculated for each sample. According to the equation, PV was positively correlated with gluten content and dough strength. The term “P/L” is present two times in the empirical equation, one time with the minus sign (−2*(P/L)²) and another one with the plus sign (+8.5*P/L). A P/L equal to 4.25 would represent the limit value: at this point, the effect of P/L would be null; below this value, the effect is positive; above this value, the effect is negative. It is also conceptually correct: an excessively high P/L value is an index of a dough that can be subjected to breakage. Thus, concerning the present data, pasta cooking quality was positively affected by P/L value up to an amount of TDWM of 100 g kg⁻¹ and negatively affected when the percentage of replaced semolina was higher. In the present study, the drying temperature was the same for all the samples and thus the K value can be taken as a constant. The calculated PV are reported in Supplementary Table 1. According to this comprehensive index, the replacement of semolina with TDWM can be applied without serious detrimental changes of pasta quality up to an amount of 50 g kg⁻¹ (that is four times lower than that determined on the basis of the W value only). Furthermore, since K value dramatically decreases as drying temperature decreases (for example, in drying cycles at 80–85 °C this factor is 42 ± 2 and for a 85–90 °C drying cycle K is 47 ± 2), the PV of spaghetti made with 300 g kg⁻¹ of TDWM would result negative and, thus, not classifiable.

In order to check the existence of significant correlations between the percentage of TDWM and the physical–chemical characteristics of the meal mixtures and the corresponding doughs, regression analysis was performed. Simple linear regressions adequately described the relationships between gluten content, b* value, W, or P/L and the amount of TDWM present in the formulations. According to regression equations, gluten content, b* and W values were negatively (R ² = −0.895, −0.832, and −0.854, respectively) and significantly (P < 0.01) correlated to the percentage of TDWM whereas P/L ratio was positively and significantly correlated (R ² = 0.922, P < 0.01) to the amount of toasted meal, confirming that pasta-making quality decreased as the amount of TDWM increased.

Physical and chemical characteristics of spaghetti

According to P/L values and PV, formulations containing more than 50 g kg⁻¹ of TDWM would be unsuitable to be used in pasta production whereas, on the basis of the alveographic W, formulations containing up to 200 g kg⁻¹ of TDWM could be used in pasta-making. Since this traditional pasta can be made with amount of TDWM higher than 200 g kg⁻¹, it seemed advisable to produce pasta from formulations containing higher proportions, in order to check their quality and the possibility to find indices usable as predictors of their cooking behavior. Spaghetti produced from dough containing 400 and 500 g kg⁻¹ of TDWM was discarded due to their unpleasant taste and flavor and their brittleness. Physico-chemical characteristics and optimal cooking time of the other pasta samples are reported in Supplementary Table 2. The replacement of semolina with increasing percentages of TDWM did not significantly affect pasta moisture and induced only a slight decrease of the water activity compared to the pasta made of 100% semolina. Concerning colorimetric data, the amount of TDWM contained into the formulations was inversely correlated to brightness and yellow index according to second-order polynomial regression models (R ² > 0.99, P < 0.01) and to the red index according to a linear regression model (R ² > 0.99, P < 0.01). The optimal cooking time (indirect measure of starch hydration and gelatinization rate) did not change to a large extent with the addition of increasing amount of TDWM. This behavior could be explained considering a series of factors. First of all, though the proteins denatured by toasting hydrate slowly than the native ones, thermal treatments generally cause the breakdown of starch molecules (van den Einde, Akkermans, van der Goot, & Boom, 2004), and the consequent increase of the water absorption capacity counterbalances the effect of toasting on proteins. Furthermore, the addition of increasing percentages of TDWM increased the presence of bran particles that physically interfered with the gluten network (Manthey & Shorno, 2002) and made the starch granules more susceptible to hydration. Finally, it is worth noting that drying increases protein polymerization, and consequently, induces the strengthening of the gluten network (Lamacchia et al., 2007), thus improving the cooking behavior of dried pasta and counterbalancing the negative effects induced by semolina replacement.

Spaghetti cooking behavior

The water absorption curves of spaghetti during cooking and overcooking are shown in Supplementary Figure 1. As expected, ΔW/W (ratio between the increase in weight and the weight of dry spaghetti) increased with cooking time but without significant differences (P < 0.05) related to the amount of TDWM present in the formulations. In order to simplify the model of water absorption by food materials, the following two-parameter, non-exponential, empirical equation was proposed by Peleg (1988); Abu-Ghannam and McKenna (1997); Jideani and Mpotokwana (2009):

where

M_t  = moisture content at a known time on a dry basis (% d.b.);

M ₀ = initial moisture content on a dry basis (% d.b.);

t = time (min);

K ₁ = Peleg's rate constant (min/% d.b.);

K ₂ = Peleg's capacity constant (1/% d.b.).

The Peleg's equation was therefore applied to describe the water absorption curves of spaghetti during cooking and overcooking and the best fits to the experimental data are shown in Supplementary Figure 1. Supplementary Table 3 reports the Peleg's equation parameters and also the results of the linear regression models fitted to the data. The goodness of fit was expressed by the values of the coefficients of determination (R ²) that were higher than 0.925 with P < 0.01 and by the SE values that were very low. Hence, Peleg's equation was suitable for describing the water absorption characteristics of spaghetti made with semolina and increasing amounts of TDWM.

As expected, losses of organic matter and amylose into the cooking water significantly increased with cooking time (Supplementary Table 4). Although the statistical analysis highlighted several significant differences among samples, no relationships were detected between percentage of TDWM added and amount of cooking losses, probably due to the leveling effect of the drying cycle. TOM and amylose content of the cooking water were linearly correlated to each other. Their correlation coefficients ranged from 0.871 (P < 0.05, pasta made of 100% semolina) to 0.961 (P < 0.01, spaghetti made of 950 g kg⁻¹ semolina and 50 g kg⁻¹ TDWM). According to Matsuo, Malcolmson, Edwards, and Dexter (1992), strong linear relationships between TOM and amylose indicate that cooking loss can be estimated reliably from the absorption values of the amylose–iodine complex.

Determination of the textural characteristics of pasta after cooking is of great importance from the product acceptability by consumers (Sozer, Dalgıç, & Kaya, 2007). Good quality pasta is defined as having high degree of firmness and elasticity (Antognelli, 1980; Pomeranz, 1987). Cohesiveness can be as a good indicator of how the sample holds together upon cooking (Sozer et al., 2007). In order to both simulate human mastication and evaluate the textural properties of spaghetti at their optimal cooking time, a double compression cycle was applied and the results are reported in Supplementary Table 5. Spaghetti produced with 100% semolina showed the lowest values of compressibility (i.e. the highest firmness), the highest values of cohesiveness, and intermediate values of elasticity and gumminess. The mechanical performance of sample made of TDWM and semolina must be considered separately from those of pasta made of 100% semolina since they were strictly related to the changes induced in the product structure by increasing amounts of TDWM. From 50 to 200 g kg⁻¹ of TDWM, compressibility decreased whereas cohesiveness, elasticity, and gumminess showed opposite trends. Pasta made with 200 g kg⁻¹ of TDWM showed textural properties similar to those of spaghetti made of 100% semolina, with the exception of elasticity, that was higher in the first. Spaghetti made with 300 g kg⁻¹ of TDWM had the highest values of compressibility and elasticity and intermediate value of cohesiveness and gumminess. According to the definition of “good quality pasta”, the best samples were those made with 100% semolina and those produced with 200 g kg⁻¹ of TDWM. The good mechanical performance of the 100% semolina–spaghetti were related to the formation of a larger gluten network, able to absorb an amount of water higher than that bound by denatured proteins and to stronger hold back the starch granules. Concerning pasta made with 200 g kg⁻¹ of TDWM, its good textural properties depended on both the increase in protein molecular weight and the large amount of unextractable polymeric proteins deriving from polimerization of albumins and globulins, as well as glutenins and gliadins (Lamacchia et al., 2010). Also, it could be supposed that important contributions to the textural characteristics of these samples were given by the resistant starch formed during the kernel heat treatment (Sozer et al., 2007). The effect of these variables was dose-dependent since further increases in the TDWM content caused worsening of the textural properties.

The possibility of predicting spaghetti texture on the basis of gluten content, gluten index, or alveographic indices was examined since it could be very useful if a texture analyzer was not readily available. First of all, simple correlations between spaghetti mechanical properties (dependent variables) and meals and dough characteristics (independent variables) were tested. Statistically significant correlations were obtained only between gluten index and mechanical parameters such as compressibility (R ² = 0.762, P < 0.05) and cohesiveness (R ² = −0.796, P < 0.02), according to exponential regression models and between gluten index and gumminess (R ² = −0.867, P < 0.01), according to a linear regression model. The negative correlation between gluten index and cohesiveness could be explained by the research performed by Dhanasettakorn, Grün, Lin, and Ellersieck (2009). They found that pasta cohesiveness might involve the α-helical structures and hydrogen bonding formation in the gluten network whereas the gluten quality is directly and strongly related to the proportion of insoluble glutenin (Sapirstein, David, Preston, & Dexter, 2007). Successively, the ability of the gluten index as single predictor of spaghetti mechanical properties was checked. The goodness of the prediction models was evaluated on the basis of the corresponding R ² and SE values: R ² = 0.6602 and SE = 0.020 for compressibility; R ² = 0.8534 and SE > 1 for cohesiveness; R ² = 0.8471 and SE = 0.329 for gumminess. According to these results, gluten index was more suitable to predict spaghetti cohesiveness than the other mechanical properties.

For prediction of pasta textural properties, the application of response surface regression followed by stepwise regression allowed to write the following four final equations:

with a rate of explained variability R ² _adjusted = 0.6337.

with a rate of explained variability R ² _adjusted = 0.742.

with a rate of explained variability R ² _adjusted = 0.802.

with a rate of explained variability R ² _adjusted = 0.833.

According to the high R ² _adjusted value, the textural properties of pasta were greatly affected by the gluten index, and in the case of elasticity, also by the gluten content. These results confirmed the important role of gluten and its components in influencing durum wheat dough properties and spaghetti cooking quality (Sissons, Soh, & Turner, 2007) especially under low temperature drying conditions. The alveographic parameters did not appear among the predictors of the pasta textural properties. Observed and predicted values were highly correlated (R²  = 0.707, 0.794, 0.842, and 0.867 for compressibility, cohesiveness, elasticity, and gumminess, respectively) and the SE values were low for compressibility and elasticity (0.303 and 0.047, respectively) and slightly higher for cohesiveness and gumminess (9.2851 and 3.411, respectively).

Conclusions

Milling of the toasted whole kernels gave rise to a meal having lower moisture and gluten and higher TDF and ash contents than semolina. Toasting of wheat kernels caused a worsening of the dough alveographic indices already at a substitution level of 100 g kg⁻¹. On the basis of PV, the replacement of semolina with TDWM can be applied without detrimental changes of the pasta quality up to an amount of 50 g kg⁻¹. According to the experimental results, the addition of TDWM did not affect in a negative way the hydration kinetics of spaghetti. This was due to a number of causes including the stabilizing effect of drying on gluten network that counterbalance the compositional effects and the effect of toasting on starch chains. Peleg's equation was suitable for describing the water absorption characteristics of spaghetti made with semolina and increasing amounts of TDWM. Concerning cooking losses, differences among samples have been detected, though they were not correlated to the percentage of toasted meal present in the formulation.

In terms of textural properties, pasta made of 100% semolina together with those produced from a mixture of semolina and 200 g kg⁻¹ of TDWM appeared as the best, being the less compressible and showing the highest values of cohesiveness, and intermediate values of elasticity, and gumminess.

Gluten index and gluten content can be used as predictors of pasta textural properties and this ability could be used to modify dough formulations depending on the characteristics desired in the final products. In particular, gluten index was proved to be the best single predictor of spaghetti cohesiveness. Also, the application of response surface regression followed by stepwise regression allowed to establish that compressibility was positively affected by the gluten index, whereas cohesiveness and gumminess were negatively affected by this index. Elasticity was negatively affected by both gluten content and gluten index.

Supplementary material

Supplementary Figure 1. Increase in weight of spaghetti made of semolina and TDWM during cooking and overcooking. Application of the Peleg's equation to describe the trends. Symbols and curves representing the best fit of the Peleg's equation to the experimental data are the following: 5% TDWM-semolina (λ,——), 10% TDWM-semolina (σ,— —), 15% TDWM-semolina (ν, – – –), 20% TDWM-semolina (+,^{|||||||||||||||}), and 30% TDWM-semolina (×,^{|||||||||||||||}).

Figura adicional 1. Incremento del peso de espaguetis hechos con semolina y harina de trigo duro tostado durante cocción y exceso de cocción. Aplicación de la ecuación de Pelg para describir las pautas. Símbolos y curvas que representan el mejor ajuste de la ecuación de Peleg a los datos experimentales son los siguientes: 5% TDWM-semolina (λ,——), 10% TDWM-semolina (σ,— —), 15% TDWM-semolina (ν, – – –), 20% TDWM-semolina (+,^{|||||||||||||||}) y 30% TDWM-semolina (×,^{|||||||||||||||}).

**equal letters indicate not significant differences (P < 0.05).

Supplementary Table 1. Physico-chemical characteristics (ash and gluten content, gluten index) of meals, dough alveographic indices (W and P/L), and pasta value (PV–K) calculated according to the Equation (1).
Tabla adicional 1. Características físico-químicas (ceniza y contenido de gluten, índice de gluten) de harinas, índices alveográficos de masa (W y P/L) y valores de pasta (PV–K) calculados según la ecuación (1).

Meal mixtures	Ash (g kg^−1 dry weight)	Gluten (g kg^−1 dry weight)	Gluten Index	b*	W (J)	P/L (Pa mm^−1)	PV–K
Semolina	8.3 ± 0.2a	100 ± 2d	80 ± 2a	21.6 ± 0.30h	146 ± 2f	2.02 ± 0.00a	34.8 ± 1.2a
5% TDWM-semolina	8.6 ± 0.2a	94 ± 2c	89 ± 3c	16.5 ± 0.22f	143 ± 2f	2.29 ± 0.10b	33.5 ± 1.0b
10% TDWM-semolina	9.0 ± 0.1b	93 ± 1c	87 ± 2b,c	14.8 ± 0.13f	124 ± 4d	3.59 ± 0.15c	28.3 ± 1.5c
15% TDWM-semolina	9.3 ± 0.2b	92 ± 3c	85 ± 2b,c	13.2 ± 0.12e	133 ± 1e	4.88 ± 0.20d	17.6 ± 1.7d
20% TDWM-semolina	9.6 ± 0.2b	92 ± 2c	85 ± 2b,c	12.5 ± 0.05d	114 ± 1c	5.71 ± 0.35e	6.3 ± 1.6e
30% TDWM-semolina	10.2 ± 0.2c	88 ± 1c	84 ± 1b	11.5 ± 0.06c	72 ± 2b	9.54 ± 0.10f
40% TDWM-semolina	10.9 ± 0.3d	81 ± 2b	83 ± 2b	10.7 ± 0.20b	55 ± 3a	11.45 ± 0.10g
50% TDWM-semolina	11.9 ± 0.5e	77 ± 0a	89 ± 3c	10.8 ± 0.20b	n.d.	n.d.
TDWM	15.6 ± 0.3f	n.d.	97 ± 3d	9.8 ± 0.37a	n.d.	n.d.
Note: n.d., not determinable.
Within each column, different letters indicate significant differences (P < 0.05).
Nota: n.d., no determinado.
En cada columna, letras diferentes indican diferencias significativas (P < 0.05).

Supplementary Table 2. Physico-chemical characteristics and optimal cooking time (OCT) of spaghetti made of semolina and TDWM.
Tabla adicional 2. Características físico-químicas y tiempo óptimo de cocido (OCT) de espaguetis hechos con semolina y harina de trigo duro tostado.

Spaghetti samples	Moisture (g kg^−1)	aw	L*	a*	b*	OCT (s)
Semolina	64.0±2.9a	0.51±0.01c	61.32±0.44e	2.3.±0.43b,c	36.93±0.31e	390±20a,b
5% TDWM-semolina	65.5±4.0a	0.47±0.00b	44.86±1.15d	4.47±0.22e	19.74±1.12d	390±20a,b
10% TDWM-semolina	63.7±1.6a	0.46±0.00a	36.50±1.83c	3.56±0.23d	9.72±0.83c	390±20a,b
15% TDWM-semolina	70.2±5.3a	0.47±0.02b	31.78±2.16b	2.41±0.11c	5.29±0.37b	420±10b
20% TDWM-semolina	61.0±8.9a	0.47±0.00b	30.89±0.70a,b	1.83±0.06b	4.90±0.09b	360±30a
30% TDWM-semolina	66.0±3.2a	0.47±0.02b	29.11±0.77a	1.07±0.06a	3.36±0.22a	390±20a,b
Note: Within the same column, different letters indicate significant differences (P < 0.05).
Nota: En cada columna, letras diferentes indican diferencias significativas (P < 0.05).

Supplementary Table 3. Peleg's Equation (2) parameters (K ₁ and K ₂) of the water sorption curves of cooked and overcooked spaghetti made of semolina and TDWM.
Tabla adicional 3. Parámetros (K ₁ y K ₂) de la ecuación de Peleg (2) de las curvas de absorción de agua de espaguetis cocidos y cocidos en exceso hechos con semolina y harina de trigo duro tostado.

Spaghetti samples	K 1 (min/% d.b.)	K 2 (1/% d.b.)	R ^2	SE
Semolina	372.135±49.835	0.382±0.095	0.975	0.052
5% TDWM-semolina	479.430±77.04	0.210±0.143	0.962	0.064
10% TDWM-semolina	240.584±32.511	0.618±0.092	0.955	0.062
15% TDWM-semolina	150.615±12.873	0.8682±0.043	0.979	0.038
20% TDWM-semolina	204.211±22.397	0.6409±0.067	0.967	0.049
30% TDWM-semolina	220.445±35.489	0.626±0.1025	0.925	0.088
Note: Determination coefficients (R ^2) and standard errors (SEs) of the linear regression models fitted to the data.
Coeficientes de determinación (R ^2) y errores estándar (SE) de los modelos de regresión lineal ajustados a los datos.

Supplementary Table 4. Losses of TOM (expressed as g kg⁻¹) and amylose (expressed as absorbance at 600 nm) into the cooking water during cooking and overcooking of spaghetti made with semolina and increasing amounts of TDWM.
Tabla adicional 4. Pérdidas de materia orgánica total (TOM, expresado como g kg⁻¹) y amilosis (expresada como absorbancia a 600nm) en el agua de hervido durante cocido y exceso de cocido de espaguetis hechos con semolina y cantidades en aumento de harina de trigo duro tostado (TDWM).

	TOM (g kg−^1)						Absorbance at 600 nm
	Cooking time (s)						Cooking time (s)
Samples	120	210	300	480	540	600	120	210	300	480	540	600
Semolina	7.52±0.91a,A	23.92±8.97b,A,B	35.92±8.64b,c, B,C	37.02±8.30b,c, A,B	42.35±3.41c, B	45.52±7.90c, A	0.205±0.001a,C	0.215±0.003b,C	0.219±0.001c,A	0.250±0.001d,B	0.251±0.001d,A	0.256±0.003e,A
5% TDWM-semolina	29.29±1.0a,D	35.72±3.95b,D	41.67±0.41c,C	46.06±1.28 d, C	53.74±0.66e,C	56.09±4.30f,C	0.241±0.001a,D	0.248±0.001b,D	0.272±0.001c,D	0.283±0.003d,D	0.284±0.003d,C	0.297±0.001e,C
10% TDWM-semolina	20.31±5.13a,B,C	35.29±2.11b,D	38.74±0.44c,B	43.99±2.89 d, B	48.37±5.51d,e,B,C	49.15±3.32e,B	0.181±0.002a,A	0.188±0.005b,A	0.261±0.001c,C	0.263±0.003c,C	0.267±0.004c,d,B	0.273±0.002d,B
15% TDWM-semolina	24.47±6.18a,C,D	33.29±0.44b,C	40.66±2.58c,C	48.83±8.29 d, C	60.94±1.88e,D	71.92±3.78f,D	0.242±0.003a,D	0.267±0.001b,E	0.271±0.002c,D	0.291±0.007d,E	0.297±0.001e,D	0.373±0.005f,D
20% TDWM-semolina	14.81±3.65a,B	21.13±4.05b,A	22.20±0.28b,A	31.84±3.20 c, A	38.54±0.91d,A	46.67±5.40e,A	0.201±0.005a,B	0.205±0.005b,B	0.231±0.002c,B	0.242±0.003d,A	0.252±0.002e,A	0.257±0.004f,A
30% TDWM-semolina	25.41±3.13a,C	30.60±0.23b,B	56.56±3.01c,D	60.21±1.05 d, D	65.57±1.39e,E	69.97±0.41f,D	0.242±0.006a,D	0.267±0.002b,E	0.296±0.005c,E	0.314±0.007d,F	0.341±0.004e,E	0.375±0.001f,D
Note: Within the same line, different small letters indicate significant differences (P < 0.05). Within the same column, different capital letters indicate significant differences (P < 0.05).
Nota: En cada fila, letras minúsculas diferentes indican diferencias significativas (P < 0.05). En cada columna, letras minúsculas diferentes indican diferencias significativas (P < 0.05).

Supplementary Table 5. Textural properties of spaghetti made with semolina and increasing amounts of TDWM at their optimal cooking time.
Tabla adicional 5. Propiedades texturales de espaguetis hechos con semolina y con cantidades en aumento de harina de trigo duro tostado en su tiempo óptimo de cocido.

Spaghetti samples	Compressibility (N)	Cohesiveness	Elasticity (mm)	Gumminess (N)
Semolina	0.73±0.50a	60.39±33.62c	1.37±0.05b	42.58±7.89c
5% TDWM-semolina	2.21±0.22d	8.41±1.45a	1.25±0.03a	18.36±1.62a
10% TDWM-semolina	1.37±0.35b,c	19.58±7.91b,c	1.30±0.06a,b	24.83±4.99a,b
15% TDWM-semolina	1.33±0.39b,c	21.46±9.60b,c	1.48±0.04c	27.03±4.03b
20% TDWM-semolina	1.07±0.22a,b	33.62±7.96c	1.47±0.04c	34.49±2.83b,c
30% TDWM-semolina	1.58±0.02c	17.94±1.84b	1.51±0.02c	28.21±2.43b
Note: Within each column, different letters indicate significant differences (P < 0.05).
Nota: En cada columna, letras diferentes indican diferencias significativas (P < 0.05).