Elaboration of a spontaneous gluten-free sourdough with a mixture of amaranth, buckwheat, and quinoa flours analyzing microbial load, acidity, and pH.

Pseudocereals are gluten-free, nutrient-dense raw materials that are being considered for the production of gluten-free products, especially bread. This study proposes a gluten-free sourdough formula based on equal amounts of amaranth, buckwheat, and quinoa with a dough yield of 250, and an elaboration method to obtain ripe sourdough. Sourdough was characterized in terms of microbiology, pH, and total titratable acidity. The established protocol made it possible to obtain a spontaneous ripe sourdough with lactic acid bacteria populations of 9.60 ± 0.02 log CFU/g and total yeasts and non-Saccharomyces yeast populations (lysine positive) of 7.91 ± 0.15 and 7.52 ± 0.10 log CFU/g, respectively. Great pH stability and total titratable acidity were maintained in the ripe sourdough phase, with values of 4.04 ± 0.02 and 18.39 ± 0.56 ml NaOH 0.1 M/10 g, respectively, at the time of the next refreshment. The use of this sourdough could be an interesting alternative for the production of not only gluten-free bread but also other gluten-free products.


INTRODUCTION
Finally, sourdough fermentation is a natural tool to extend bread shelf life because it can prevent microbial spoilage and retard bread staling. In fact, sourdough associated LAB produce many antimicrobial compounds such as organic acids, ethanol, CO2 and fungicins with activities against common bread spoilage organisms (Hassan et al., 2016). Moreover, bread staling delay is partly achieved thorough not only reduction of starch retrogradation as a result of amilolitic activity from sourdough LAB, but also expolysaccharide production by heterofermentative LAB (Galle et al., 2012). Therefore, incorporation of sourdough in gluten-free bread-making can contribute to avoiding the use of expensive chemical preservatives (Novotni et al., 2012) and will satisfy the request of consumers who demand a clean label. Elaboration processes and nutrient-dense ingredients (e.g. pseudocereals) are currently being tested to improve sensory and nutritional characteristics of gluten-free breads and enhance preservation and durability (Collar et al., 2015;Witczak et al., 2016).
Some researchers use different gluten-free flours, in addition to rice and corn, for sourdough making, cereal such as sorghum or millet (Galle et al., 2012;Akinola and Osundahunsi, 2017); pseudocereals such as amaranth, buckwheat, teff or quinoa (Sterr et al., 2009;Rühmkorf et al., 2012;Wolter et al., 2014;Rizzello et al., 2016); some legumes (Curiel et al., 2015); or others such as cassava or chestnut (Vogelmann et al., 2009;Aguilar et al., 2016); even some of them in germination form (Ogunsakin et al., 2015;Montemurro et al., 2019). All these flours provided nutrients that help the growth of the microorganisms. It has been reported that adding sourdough and some of these gluten-free flours has effect on bread quality (Campo et al., 2016;Rinaldi et al., 2017). Several protocols include the addition of LAB at the beginning of the sourdough (Marti et al., 2015) in order to ensure rapid dominance; some of them, inoculate strains from previous sourdough in order to take advantage of the characteristics of autochthonous strains (Picozzi et al., 2015). But there are procedures in which the fermentation occurs spontaneously, favoring the growth of autochthonous microorganisms (Gordún et al., 2015). Technological factors such as the frequency of the refreshments and the percentages of sourdough used are different depending on the procedures. Refreshments are proposed every 5 or 6 hours, and up to 24 hours, adding very different percentages of sourdough, ranging from 10, 20 and up to 40% (Aguilar et al., 2016). In spite of the current knowledge of gluten-free spontaneous sourdough, more efforts are needed to establish basic criteria of sourdough elaboration protocol and the behavior of the raw ingredients used. This study aims to formulate a gluten-free sourdough based on pseudocereals and proposes an elaboration method for it. The effectiveness of the formulation and elaboration method was assessed by an experiment in which the resulting sourdough was characterized in terms of pH, total titratable acidity (TTA) and microbial population (total yeasts, non-Saccharomyces yeasts and LAB).

Ingredients
Three types of commercial organic pseudocereals were used for elaboration of spontaneous glutenfree sourdough. Amaranth seeds (Amaranthus spp.) were provided by Bioprasad (Spain) and grounded using a Perten 3100 laboratory grinder with a 0.8 m sieve; stone-ground whole buckwheat (Fagopyrum esculentum) flour was provided by Rincón del Segura (Spain); and quinoa flour (Chenopodium quinoa) by Salutef (Spain). All products were certified as gluten-free (<20 mg gluten/kg, Commission Regulation (EC) nº41/2009). Items from the same manufacturing batch were used and the pseudocereal flours were stored at room temperature, sheltered from light.

Formulation and procedure of gluten-free sourdough making with pseudocereal flours
Gluten-free sourdough was formulated using equal amounts of the three pseudocereal flours: amaranth, buckwheat and quinoa, and with a constant dough yield of 250 (DY = [flour weight + water weight] x 100/ flour weight) ( Figure 1). The sourdough making procedure consisted of an initial phase of immature ferment that included five stages of dough fermentation at 30 ºC (± 0.2 ºC) and four periods of blockage at 5 ºC (± 0.2 ºC) prior to daily refreshment or back-slopping of the mature or ripe sourdough. This procedure was previously tested to fix a number of parameters such as a dough yield of 250 in order to obtain a fluid dough consistency. Previously fermented dough was used as a starter for the next dough, and the duration of the fermentation step at 30 °C required to reach a pH of 4.0-4.2 was tested. Finally, it was verified that typical mother dough microbiota (LAB and yeast) was obtained (results not shown).
The protocol of the proposed formulation and procedure is described in Figure 1. At the same time and under identical conditions, three sourdoughs were made in order to verify the designed protocol. To verify it, three sourdoughs were made simultaneously under identical conditions. The pseudocereal flours used at each step were mixed with water (30 ºC) and kneaded using a kneader (Kenwood, model KM 336) at 200 rpm for 2 min. Once the initial doughs were obtained, the first stage of fermentation started and lasted 24 h at 30 ºC until reaching a pH of 4.0-4.2. At that moment, an aliquot of each fermented dough (also called immature sourdough o pre-ferment) was mixed again with the pseudocereal flours and water. Fermentation times to obtain immature ferment at 30 ºC were initially 16 h and then 6 h. Blocking times at 5 ºC were flexible, ranging from 18 h to 66 h. After these steps of fermentation-blocking-renovation, daily refreshment of the already ripe sourdough was performed for 4 days. The percentage of ripe sourdough used as inoculum in the refreshments was 30% of the total dough weight, and fermentation time at 30 ºC and microbial activity blocking time at 5 ºC were 6 h and 18 h, respectively.

Measurements of flour and mature sourdough
Acidity. During the making of the three sourdoughs, pH and TTA were measured on dough samples taken before and after each nutrient renewal and at the end of the fermentation stage. pH was also determined after each blocking stage prior to refreshment. For all determinations, two independent measurements were taken on each sample and means were calculated.
The pH was measured directly in the dough with a glass electrode of a pH meter (Crison Instruments S.A., Spain) and in the aqueous preparation of 10 g dough samples blended with 90 ml of distilled water used to determine TTA. These measurements were called "Direct pH" and "Diluted pH", respectively.
TTA was determined by suspension of the doughs (10 g of dough diluted in 90 ml of distilled water) by an acid-base titration with 0.1 mol/l NaOH to pH 8.5 (at least 20 sec) under shaking.
TTA was expressed as the 0.1 mol/l NaOH volume used.

Microbial dynamics of yeast and LAB. Total yeasts, lysine positive yeasts (also called non-
Saccharomyces yeasts) and LAB present in the three pseudocereal flours and dough samples (preserved at 4 ºC) before renewal were analyzed at different steps of the sourdough making process. To this purpose, 10 g of flour or dough was homogenized with 90 ml of saline solution ringer ¼ (Sharlab, Spain) under shaking for 2 min. Decimal dilutions were made using the same solution, and microbiological seeding was performed with selective culture media on agar plates.
Total yeast count was carried out on WL nutrient agar supplemented by 0.5 g/l chloramphenicol (Scharlab, Spain), non-Saccharomyces yeasts on lysine agar (Scharlab, Spain) and LAB on Man-Rogosa-Sharpe agar (MRS agar) (Scharlab, Spain) supplemented by 15% grape juice, 15% tomato juice, 20 g/l maltose and 75 mg/l cyclohexamide (pH: 5.5). Yeasts were incubated for 5 to 10 days at 27 ºC, and LAB were incubated for 4 to 5 days at an atmosphere of reduced oxygen at less than 10% using a candle jar. Countable plates, between 15 and 150 colonies on WL nutrient or lysine agar, and between 30 and 300 colonies on MRS agar, were used to quantify the microbial population.

RESULTS
showed that quinoa and, to a lesser extent, buckwheat have a good buffering capacity when used separately. Some ingredients present in the flour, such as proteins, phytate or ash, also have a good buffering capacity. It has been shown that TTA is correlated with the phytate concentration (Hammes et al., 2005) and that the action of ash as a buffering agent is due to the higher concentration of minerals such as iron, sodium, potassium, magnesium and phosphorus (Salovaara and Valjakka, 1987).

LAB and Yeasts
LAB and total yeast counts were observed to be lower in amaranth flour than in quinoa and buckwheat flours (< 2 Log CFU / g of LAB and yeasts). Quinoa and buckwheat flours had initial LAB counts of 3.36 and 3.12 Log CFU / g, respectively. Regarding total yeasts, populations of < 2 Log CFU/ g in quinoa flour and 2 Log CFU/g in buckwheat flour were obtained. Counts for both microbial groups were quite similar to those found in wheat flour (Van Kerrebroeck et al., 2016).
Gram-positive (e.g. Bacillus sp) and Gram-negative (e.g. Pseudomonas sp and Enterobacteriaceae) bacteria populations had also been found but disappeared as sourdough fermentation progressed and pH decreased (Minervini et al., 2014). LAB, total yeast and non-Saccharomyces yeast counts are shown in Table 1. The first microbiological control of the three replicas of the sourdoughs carried out after 24 h of fermentation at 30 ºC revealed an average value and standard deviation of LAB counts of 9.54 ± 0.14 Log CFU /g. The viable population of LAB was high at the beginning and also throughout the entire process, in which constant counts were maintained both in the immature sourdough (24 h, 88 h and 160 h) and ripe sourdough (208 h and 256 h) phases. Average values and standard deviations of LAB counts in the immature and ripe sourdough phases were 9.55 ± 0.16 Log CFU / g and 9.60 ± 0.03 Log CFU / g, respectively. These figures were similar to those obtained for gluten-free sourdoughs by other authors (Sterr et al., 2009), even using LAB starter (Rühmkorf et al., 2012Rizzello et al., 2016). The LAB colonies observed on MRS agar plates had a uniform morphological appearance: they were white, circular and bright (Figure 3).
Unlike the LAB count, the total yeast count grown on WL nutrient agar increased throughout sourdough making (Table 1) The average count and standard deviation of the non-Saccharomyces yeast group, grown on lysine agar, was 4.62 ± 0.17 Log CFU / g at 24 h of fermentation. This group increased significantly to reach values around 7.5 Log CFU / g, remaining constant in the immature sourdough phase (160 h) and the ripe sourdough phase (208 h and 256 h). Gordún et al., (2015) reported that the high non-Saccharomyces yeast group value at the beginning of sourdough preparation was related to the addition of different non-essential ingredients. In the present study, the evolution of this group was inverse. This yeast growth could be due to the nutrients in the flours used.
WL nutrient agar and lysine agar made it possible to obtain total yeast counts and lysine positive yeasts (or non-Saccharomyces yeasts), respectively. The difference between both groups allows the concentration of the lysine negative yeast group, which includes among other genera Saccharomyces, to be known. This information is very useful because the different yeast groups are related to the dough texture observed during the making of the sourdough (Figure 4, Table 1). At 24 h of fermentation at 30 °C, the dough had a spongy appearance and a greater volume. In this control, the lysine negative yeast group represented practically 7% of the total yeast count, decreasing to 4.5% in a subsequent control. In the next stages of the immature sourdough phase, the dough barely sponged because very few CO2 bubbles formed. Swelling was recovered in the ripe sourdough phase and the dough increased in volume with the first refreshments. However, in the last two refreshments, after 6 h of fermentation at 30 ºC, the dough could not maintain the volume reached, which decreased. The difficulty of retaining gas in sourdough made with pseudocereals due to lack of gluten has been described (Marti et al., 2015). In the ripe sourdough phase, the lysine negative yeast count increased, becoming the dominant group in the last refreshment, where it represented 80.6% of the total yeast count. The increase of the lysine negative yeast group can be considered responsible for the increase of the total yeast population.
Two morphology types of yeast colonies were found (Figure 3). A first type of colony was observed on lysine agar (belonging therefore to the non-Saccharomyces group) and on WL nutrient agar. This type of colony was white, filamentous and with an umbonate center elevation. The second type of colony was only observed on WL nutrient agar (belonging therefore to the lysine negative yeast group), was circular and smooth, and had a creamy coloration and an umbonate center elevation too. These two colonies showed pinpoint or negligible growth on lysine agar.
Lysine agar medium uses L-lysine to provide organisms with a source of nitrogen, and was first used to distinguish wild yeasts in the brewing industry (Walters and Thiselton, 1953). Currently, it is used to control contamination in the manufacture of baker's yeast. Its application in sourdough allows lysine positive yeasts to be tracked (Gordún et al., 2015(Gordún et al., , 2017. In the present study, a type of non-Saccharomyces filamentous yeast was developed with stable high growth rates during sourdough making. Both lysine positive and negative yeasts participated in sourdough fermentation. Moreover, some lysine positive genera may find potential use in the baking industry because of their interesting aroma profiles (Aslankoohi et al., 2016).

Protocol for making sourdough
The pH, TTA, LAB and yeast population results ( Figure 2, Table 1) demonstrate that the mixture of equal proportions of amaranth, buckwheat and quinoa is a viable formulation for gluten-free sourdough. Additionally, the use of these ingredients provides bakery products with a greater diversity of nutrients, i.e. those found in these flours (USDA Food Composition Databases) and those generated during sourdough fermentation (Di Cagno et al., 2008;Arendt et al., 2011;Hager et al., 2012).
The proposed protocol (Figure 1) fulfills all the necessary conditions. The dough yield of 250 in the two phases (immature and ripe) of sourdough making provided a fluid consistency and may have favored the development of LAB and yeast populations. In the immature sourdough phase, the 5 stages of refreshment, fermentation time at 30 ºC until reaching pH 4 and activity blocking at 5 ºC allowed TTA to be consolidated and favored the development of LAB and yeast populations capable of overcoming cooling stages up to 66 h. In the ripe sourdough phase, maturity was evaluated by performing daily refreshments for 4 days, and constant acidification values (pH, TTA) were obtained. It has been shown that to keep TTA values constant it is important, among other things, to block fermentation without delay when a specific pH value is reached. At each step, the time required to reach the established pH value (4.0) is conditioned, among other factors, by the percentage of previous sourdough added. The increase of 30 to 40% of previous sourdough in one of the steps of immature sourdough, allowed the shortening of the necessary time, going from 16 to 6 hours. Good LAB:total yeast ratios (100 to 208 h and 10 to 256 h) and clear leavening capacity were verified. However, in this phase leavening capacity may have been influenced by the fact that yeast counts did not stabilize and gradually increased along refreshments. This could indicate that sourdough can take longer to reach maturity. Regarding gluten sourdoughs, it is widely accepted that between 5 and 7 days of sourdough propagation may be necessary for it to achieve maturity (Ercolini et al., 2013). On the other hand, gluten-free sourdoughs prepared with different pseudocereals have been reported to achieve maturity within 3 and 7 days (Sterr et al., 2009; Table 1. Means of the microbiological counts of lactic acid bacteria (LAB), total yeasts and non-Saccharomyces yeasts (Lysine positive yeast), and Lysine negative yeast (difference between total yeast and non-Saccharomyces yeast) expressed as Log CFU/g of the three samples of doughs, with the corresponding standard deviations (SD).