In Situ Effects of 4,6-α- and 4,3-α-Glucanotransferases During Sourdough Fermentation: Assessing Microbial Community Dynamics, Bread Glycemic Index, Staling, and Texture

Abstract

Although some lactic acid bacteria species harbor genes encoding α-glucanotransferases (4,6-α-GTase and 4,3-α-GTase), the technological and functional implications of these enzymes within sourdough ecosystems remain poorly understood. This study investigated the effects associated with α-GTase-positive and α-GTase-negative strains of Limosilactobacillus reuteri and L. fermentum on sourdough microbiota, bread quality, and starch digestibility. Starter culture inoculation steered the assembly of the sourdough microbiota, resulting in stable acidification and altered microbial community structures. High-throughput sequencing analysis showed that starter culture addition generally increased taxonomic richness, while Shannon and Simpson diversity indices varied according to strain genotype and inoculation level. Sourdoughs fermented with α-GTase-positive strains exhibited higher viscosity while maintaining shear-thinning behavior, which was putatively linked to in situ enzymatic synthesis. Additionally, inoculation with these strains improved bread specific volume, suggesting a possible structural modification of the starch matrix or the overlapping effects of fermentation metabolites. During storage, the 4,3-α-GTase-positive L. fermentum PFC282 strain delayed the increase in bread hardness, reflecting a possible mitigation of amylopectin retrogradation and indicating a potential anti-staling effect. In addition, the 4,6-α-GTase-positive L. reuteri PFC338 strain produced breads with lower estimated glycemic index values, suggesting a lower susceptibility to enzymatic starch hydrolysis likely due to the hypothesized alterations in structural linkages. Overall, the findings highlight the potential of α-GTase-positive sourdough starter cultures to modulate the technological and nutritional properties of bread.

1. Introduction

Sourdough fermentation is a traditional biotechnological process driven by the symbiotic activity of lactic acid bacteria (LAB) and yeasts, which improves nutrient bioavailability and shelf life of cereal-based products [1]. Although starch, the predominant component of the complex sourdough matrix, provides essential technological functionalities in bakery products, its inherent structural limitations, such as low solubility in dough systems and high susceptibility to retrogradation, can often compromise end-product quality [2]. Driven primarily by the recrystallization of amylopectin chains, starch retrogradation is a fundamental cause of bread staling, leading to substantial economic losses and food waste in the baking industry [3]. Furthermore, while sourdough fermentation can lower the estimated glycemic index (eGI) of bread, the high-glycemic nature of starch-rich bakery formulations remains a persistent challenge [4]. Consequently, targeted structural modification of starch is imperative to impart desirable physicochemical and functional attributes, thereby optimizing its industrial compatibility [5,6].

Structural drawbacks of native starch can be mitigated by utilizing chemical, physical, or biotechnological (specifically enzymatic) modifications, either individually or in combination [7,8,9]. Among these, enzymatic approaches have attracted considerable interest, driven by the growing consumer demand for clean-label products and the need to minimize chemical waste or hazardous by-products [10]. In this context, enzymes of the glycoside hydrolase family 70 (GH70) have gained increasing interest as natural tools for tailoring the structural and functional properties of starch. Traditionally known for glucansucrases that synthesize α-glucans from sucrose, this family also encompasses specialized α-glucanotransferases (α-GTases) [11]. Produced by specific LAB strains, these enzymes (4,6-α-GTase and 4,3-α-GTase) possess the unique capability to modify starch architecture by converting linear α-1,4-glucosidic linkages into branched α-1,6 or α-1,3 linkages, respectively [12]. This enzymatic rearrangement introduces novel branching points that exert steric hindrance against the realignment of amylopectin chains, thereby suppressing retrogradation enthalpy and delaying staling [13]. Furthermore, the resulting complex branched matrices and malto-oligosaccharides exhibit enhanced resistance toward human digestive enzymes, consequently lowering the glycemic index [14,15].

In our previous study [16], genes encoding 4,6-α-GTase and 4,3-α-GTase from Limosilactobacillus reuteri E81 (reclassified as L. reuteri PFC338) and L. fermentum PFC282, respectively, were successfully expressed in Lactococcus lactis MG1363 to investigate their glycosidic bond-forming capabilities and catalytic efficiencies on starch substrates. In a subsequent study exploring the direct application of these purified enzymes as additives in bakery products [17], they were found to alter starch granule morphology, delay staling without compromising sensory quality, and reduce the eGI values of bread by 18.01% (4,6-α-GTase) and 13.61% (4,3-α-GTase). However, the industrial-scale utilization of purified or partially purified enzymes is often limited by substantial production costs and the requirement for food additive declarations. Consequently, shifting from direct enzyme supplementation to an in situ production strategy within the sourdough ecosystem offers a highly sustainable, cost-effective, and clean-label alternative.

Although certain LAB strains are documented as α-GTase producers in the literature [12], to the best of our knowledge, no study has yet investigated the in situ production of these specific enzymes within a sourdough matrix. To address this gap, the present study aimed to utilize L. reuteri PFC338 and L. fermentum PFC282, natural producers of 4,6-α-GTase and 4,3-α-GTase, respectively, as sourdough starter cultures, and to comprehensively evaluate their impacts on the microbiota, eGI, and staling properties of both sourdough and the resulting bread. This novel approach seeks to provide insights into the potential role of in situ enzymatic modification on the sourdough matrix and its subsequent macromolecular outcomes.

2. Materials and Methods

2.1. Screening and Cultivation of Starter Strains Based on α-Glucanotransferase Genes

The LAB strains utilized in this study were selected from the Pamukkale University Food Engineering Culture Collection (PUFECC) based on their potential to produce α-GTase. The presence of genes encoding 4,6-α-GTase (gtfB1) and 4,3-α-GTase (gtfB2), which were characterized in our previous study [16], was confirmed in candidate strains by colony PCR. Species-specific primer pairs used for the PCR assays were as follows: GTFB1-M-ECORI-F1 (5′-ACA AGC CAG ATG GTA AGA C-3′) and GTFB2-M-KPNI-R1 (5′-CGG TTA ACA ATA CTA-3′) for L. reuteri strains; GTFB1-F (5′-ACA AGC CCA RGA TGG TAA-3′) and GTFB2-R (5′-TCG CCA TTA GST AAT TT-3′) for L. fermentum strains. Amplification reactions were performed in a total volume of 50 µL containing 10 µM of each primer, 10 mM dNTP mix (Merck, Darmstadt, Germany), and OneTaq DNA Polymerase (New England Biolabs, Ipswich, MA, USA), following a protocol adapted from the manufacturer’s instructions. The thermal cycling profile consisted of an initial denaturation at 94 °C for 1.5 min, followed by 30 cycles of denaturation at 95 °C for 30 s, annealing at the specific optimized melting temperature (Tm −5 °C) of each primer pair for 1 min, and extension at 68 °C for 60 s/kb, with a final extension at 68 °C for 5 min. PCR products were resolved by 1% (w/v) agarose gel electrophoresis (Merck, Germany). Strains yielding specific bands of the expected molecular size were designated as α-GTase-positive, while those without bands were classified as α-GTase-negative. Consequently, 4,6-α-GTase-positive L. reuteri PFC338, α-GTase-negative L. reuteri PSC73, 4,3-α-GTase-positive L. fermentum PFC282, and α-GTase-negative L. fermentum PFC268 were selected for further use. The $\alpha$-GTase-negative strains were utilized as strain-specific baseline controls due to their comparable sourdough acidification profiles and microbial growth kinetics relative to the $\alpha$-GTase-positive strains, thereby aiming to minimize fermentative variations. Stock cultures were maintained at −80 °C in MRS broth (Merck, Germany) supplemented with 20% (v/v) glycerol (Riedel-de Haën, Seelze, Germany).

2.2. Preparation of Sourdoughs

To prepare the sourdough samples, the α-GTase-positive and α-GTase-negative starter cultures were inoculated into MRS broth and incubated until the late-exponential phase at 30 °C for L. fermentum strains and at 37 °C for L. reuteri strains. Bacterial cells were harvested by centrifugation (Hitachi Centrifuge, Tokyo, Japan) at 9000 rpm for 30 min, washed twice with sterile PBS buffer, and adjusted to an OD_600nm of 1.0. The cell suspensions were then serially diluted in sterile PBS to reach the target initial cell counts. Sourdoughs were prepared with a dough yield of 200 (w/w) using commercial additive-free white wheat flour (Carrefoursa, Istanbul, Türkiye) and potable water. According to the manufacturer’s specifications, the flour contained 10.4% protein, 71.9% carbohydrate (2.9% sugars), 3.5% dietary fiber, and 1.1% fat, with a maximum ash content of 0.7% (dry basis). In terms of functional properties, it exhibited a water absorption capacity of 56.23% as determined by Farinograph analysis. Initially, a master dough was prepared and divided into equal portions. According to the experimental design (

Table 1. Experimental design applied for sourdough production carried out at 30 °C.

Starter Culture	Inoculation Rate (CFU/g)	Sample Codes
L. reuteri PFC338 (4,6-α-GTase-positive)	102	SD338L
L. reuteri PFC338 (4,6-α-GTase-positive)	106	SD338H
L. reuteri PSC73 (4,6-α-GTase-negative)	102	SD73L
L. reuteri PSC73 (4,6-α-GTase-negative)	106	SD73H
L. fermentum PFC282 (4,3-α-GTase-positive)	102	SD282L
L. fermentum PFC282 (4,3-α-GTase-positive)	106	SD282H
L. fermentum PFC268 (4,3-α-GTase-negative)	102	SD268L
L. fermentum PFC268 (4,3-α-GTase-negative)	106	SD268H
Control *	-	SDC

* No starter culture was added to the control sample.

), each dough portion was inoculated with 4,6-α-GTase-positive, 4,6-α-GTase-negative, 4,3-α-GTase-positive, or 4,3-α-GTase-negative strains at two different initial inoculation levels (10² and 10⁶ CFU/g). Sourdough prepared without any starter culture inoculation served as the spontaneously fermented control group. Fermentation was carried out at 30 °C until a target pH of 4.5 ± 0.1 was achieved.

Upon reaching the target pH, a 10% (w/w) aliquot of the fermented sourdough was propagated into a fresh flour–water mixture, and this backslopping process was repeated for three consecutive cycles. At the end of each backslopping stage, fermented sourdough samples were collected aseptically and stored in sterile falcon tubes at −20 °C and protected from light and moisture, specifically for subsequent microbiological counting and DNA extraction.

2.3. Total Titratable Acidity

The total titratable acidity (TTA) values of the sourdough samples were determined at each backslopping cycle (BC). To measure TTA, a 10 g dough sample was homogenized in 100 mL of distilled water. The suspension was then titrated with 0.1 N NaOH (Merck, Germany) using a 1% phenolphthalein indicator (Merck, Germany) under continuous stirring until a faint pink color persisted for 30 s. The total acidity of the samples was calculated from the volume of titrant consumed and expressed as a percentage of lactic acid equivalent (%).

2.4. Microbiological Analysis

At each BC, the fermented sourdough samples were homogenized with sterile peptone physiological saline at a 1:10 (w/v) ratio using a stomacher blender (Seward Medical, London, UK), and subsequent ten-fold serial dilutions were prepared. LAB were enumerated on MRS agar supplemented with 0.01% cycloheximide and GM17 agar (containing 0.5% glucose, Condalab, Madrid, Spain) following incubation at 30 °C for 72 h. Total aerobic mesophilic bacteria (TAMB) counts were determined on Plate Count Agar (PCA; Merck, Germany) after incubation at 30 °C for 72 h. Yeasts were enumerated on Dichloran Rose Bengal Chlortetracycline (DRBC) agar (Merck, Germany) after an incubation period of 3–5 days at 25 °C. All microbiological counts were expressed as log CFU/g of sourdough.

2.5. Culture-Independent PCR-DGGE Profiling of LAB in Sourdoughs

To profile the LAB diversity within the sourdough samples, a culture-independent denaturing gradient gel electrophoresis (DGGE) approach was employed, modified from the protocols described by Van der Meulen et al. [18] and Oguntoyinbo and Dodd [19]. For microbial genomic DNA extraction, sourdough samples were homogenized with peptone water at a 1:10 (w/v) ratio. To separate large dough particles from the microbial fraction, the homogenate was initially centrifuged at 1000× g for 5 min. The resulting supernatant was then subjected to a second centrifugation at 5000× g for 15 min to harvest the bacterial cells. The recovered cell pellets were washed twice with PBS buffer, and genomic DNA was isolated using the Bacterial & Yeast Genomic DNA Purification Kit (EurX, Gdańsk, Poland). The V3 region of the bacterial 16S rDNA was amplified using the GC-clamped forward primer P338F and the reverse primer P518R. PCR amplifications were conducted in a total reaction volume of 40 µL, containing 8 µL of 5X FIREPol^® Master Mix (Solis BioDyne, Tartu, Estonia), 1 µM of each primer, and 2 µL of template DNA. The thermal cycling profile consisted of an initial denaturation at 95 °C for 12 min, followed by 30 cycles of denaturation at 95 °C for 20 s, annealing at 55 °C for 45 s, and extension at 72 °C for 1 min, with a final extension at 72 °C for 5 min. DGGE analysis was performed on a vertical electrophoresis system (Thermo Fisher Scientific, Waltham, MA, USA) using polyacrylamide gels formulated with a 30% to 60% denaturant gradient (where 100% denaturant corresponds to 7 M urea and 40% formamide). Electrophoresis was carried out in 1X TAE buffer at a constant temperature of 60 °C, initially at 50 V for 15 min, followed by a separation phase at 150 V for 4 h. Finally, the gels were stained with SYBR Gold (Invitrogen, Carlsbad, CA, USA) and visualized under UV illumination.

2.6. High-Throughput 16S rRNA Amplicon Sequencing of Sourdough Microbiota

At the end of BC₃ (pH 4.5 ± 0.1), total genomic DNA was extracted from three independent biological sourdough replicates and evaluated according to standard quality criteria regarding concentration and purity. To accommodate sequencing optimization and represent the baseline microbial profile of the master fermentations while managing technical variations, equal amounts of purified DNA from the biological replicates were pooled into a single representative composite sample per group prior to sequencing. To elucidate bacterial diversity, high-throughput sequencing of the V3 region of the 16S rRNA gene was performed. The V3 region was amplified using specific primers and purified. During library preparation, Illumina dual indices and sequencing adapters were appended using the Nextera XT Index Kit (Illumina, San Diego, CA, USA), followed by a secondary purification step. The concentration of the constructed libraries was quantified via real-time PCR (qPCR), and all samples were normalized by dilution to a final concentration of 4 nM. Normalized libraries were subsequently pooled. Amplicon sequencing was performed on the Illumina platform using a 2 × 250 bp paired-end read length based on the sequencing-by-synthesis principle. The resulting fluorescence data were converted into raw FASTQ files. All bioinformatics analyses were executed using the QIIME 2 (v2018.11) platform [20]. As part of the data quality control pipeline, reads with a Phred score below 20, along with primers and barcodes, were trimmed and filtered. After removing chimeric sequences, Amplicon Sequence Variants (ASVs) were resolved using the DADA2 plugin within QIIME 2. Taxonomic classification of the filtered sequences was performed against the SILVA database. Sequences matching plant organelles (chloroplasts and mitochondria) originating from the wheat flour matrix were filtered out prior to downstream analyses. Alpha and beta diversity analyses were carried out utilizing the QIIME 2 core-diversity pipelines to determine intra- and inter-sample diversity. Rarefaction curves were constructed to verify whether the sequencing depth sufficiently represented the microbial diversity within the sourdough matrix. The curves generally appeared to approach a plateau at the selected standardized sequencing depth threshold of 2000 reads per sample, suggesting that the sub-sampling depth was adequate to capture the core microbial community structure (Figure S1). Due to the pooling of biological replicates into a single representative composite sample per group, alpha and beta diversity values are reported descriptively to present the baseline microbial profile of each formulation without inferential group comparisons.

2.7. Breadmaking from Sourdoughs

Bread production was carried out in three independent batches for each formulation using sourdoughs prepared according to the experimental design (Table 1) that reached the target pH (4.5 ± 0.1) at the end of BC₃. The bread dough formulation consisted of commercial additive-free white wheat flour (Carrefoursa, Türkiye) on a 14% moisture basis, 1.5% table salt, and 2% commercial fresh yeast (Pakmaya, Istanbul, Türkiye) calculated on a flour basis. These ingredients were initially blended at a low speed (speed setting 2) for 1 min using a KitchenAid mixer (KitchenAid, Greenville, OH, USA), after which matured sourdough was incorporated into the matrix at a level of 15% (w/w). The specific water addition was determined by reducing the water absorption capacity obtained from Farinograph analysis by 2%. Upon adding the calculated amount of water based on flour weight, the dough was kneaded for 5 min. The resulting dough was subjected to bulk fermentation at room temperature (25 °C ± 1.0) for 30 min at a relative humidity above 80%. Following the bulk fermentation, the dough from each independent batch was divided into 160 g pieces, allowed to rest for 10 min, shaped, and placed into baking molds. The final proofing stage was conducted at 30 °C for 45 min, and the loaves were subsequently baked at 220 °C for 17 min. The produced breads were stored in polyethylene bags within a climate cabinet (Nüve-TK-120, Ankara, Türkiye) maintained at 25 °C and 65% relative humidity. For the accelerated staling and thermal properties analyses, a separate set of bread samples was kept under controlled storage at 4 °C.

2.8. Viscosity and Rheological Properties of Sourdoughs

The apparent viscosity of sourdough samples (at day 0 and mature preferment stage) was measured using a rotational viscometer (Brookfield Model RVDV-II, Middleboro, MA, USA) equipped with an SC4-14 spindle at 25 °C. The measurements were performed over a progressive rotational speed range (1 to 100 rpm) to obtain the shear stress and shear rate data. To describe the steady-shear flow behavior of the sourdoughs, the experimental data were fitted to the Power Law (Ostwald–de Waele) model:

τ = K (γ)ⁿ

where τ is the shear stress (Pa), K is the consistency index (Pa.sⁿ), γ is the shear rate (s⁻¹), and n is the flow behavior index.

2.9. Specific Volume of Sourdough Breads

The specific volumes of the breads were determined using a rapeseed displacement method modified from Kazakos et al. [21]. The values were calculated by dividing the loaf volume by its weight.

2.10. Textural Properties of Sourdough Breads

Texture profile analysis (TPA) of the breads was performed during storage (days 0, 3, 5, and 7) using a texture analyzer (Brookfield CT3-4500, Middleboro, MA, USA) equipped with a TA3/1000 cylindrical probe, following a modified method of Sun et al. [22]. Slices of 18 mm thickness were cut from the center of the loaves. The samples were subjected to a two-cycle compression test at 50% deformation with a trigger force of 0.07 N and a test speed of 1 mm/s. Hardness, chewiness, cohesiveness, gumminess, and springiness values were calculated from the resulting force–time curves.

2.11. Thermal Properties of Sourdough Breads

The thermal properties of retrograded amylopectin in the bread samples were determined using a differential scanning calorimeter (DSC131 EVO, Setaram, Caluire-et-Cuire, France) following a modified method of Hadaegh et al. [23]. Approximately 3.2 mg of bread crumb sample (stored at 4 °C at days 0, 3, 5, and 7) was weighed directly into aluminum pans, hermetically sealed, and equilibrated prior to analysis. The measurements were carried out under an argon atmosphere from 25 °C to 135 °C at a constant heating rate of 5 °C/min.

2.12. Estimated Glycemic Index of Sourdough Breads

The estimated glycemic index (eGI) of the bread samples was determined via an in vitro starch hydrolysis procedure based on the methods described by Englyst et al. [24] and Yaman et al. [25]. Briefly, 1 g of homogenized bread sample was mixed with 5 mL of distilled water and 10 mL of a pepsin (Sigma-Aldrich, St. Louis, MO, USA)–guar gum (Merck, Germany) solution, and incubated at 37 °C and 175 rpm for 30 min. Following the gastric phase, the pH was adjusted to 5.0–5.25 using a 0.5 M sodium acetate solution (Merck, Germany). An enzyme mixture containing pancreatin (Sigma-Aldrich, USA) and amyloglucosidase (13.4 U/mL; Megazyme, Bray, Ireland) was then added, and the total volume was completed to 50 mL. During the 180 min intestinal incubation, 0.5 mL aliquots were collected at specific time intervals (20, 30, 60, 90, 120, and 180 min) and immediately immersed in a boiling water bath for 5 min to terminate enzyme activity. After centrifugation (4000× g, 5 min), the glucose content in the supernatant was measured spectrophotometrically (UV-1800, Shimadzu, Tokyo, Japan) at 510 nm using a glucose determination kit (GOPOD-format, Megazyme, Ireland). To account for differences in moisture among the samples, all calculations were corrected based on the dry matter content of the breads (on an equal dry weight basis). Commercial white wheat bread crumb (crust-free) was utilized as the reference bread. The area under the hydrolysis curve (AUC) for both the samples and the reference bread was calculated using the trapezoidal rule. The hydrolysis index (HI) was calculated using the following formula:

$$\mathrm{HI}=\left({\displaystyle \frac{\mathrm{AUC} \mathrm{of} \mathrm{the} \mathrm{sample}}{\mathrm{AUC} \mathrm{of} \mathrm{the} \mathrm{reference} \mathrm{white} \mathrm{bread}}}\right)\times 100$$

Consequently, the HI value of the reference white bread was defined and set as 100.00. The eGI values were obtained using the equation proposed by Goñi et al. [26] and Yaman et al. [25]:

eGI = 39.71 + (0.549 × HI)

These estimated values, representing predicted glycemic responses based on in vitro starch hydrolysis, were classified into three categories according to the adapted FAO/WHO international guidelines: low-glycemic (<55), medium-glycemic (55–69), and high-glycemic (≥70) categories [27].

2.13. Statistical Analysis

Sourdough fermentation and bread baking processes were executed in three independent biological batches. Analytical measurements were performed in technical triplicates for each individual batch. The technical replicates within each biological batch were averaged, and the resulting mean values from the three independent batches were utilized as the replicates (n = 3). The data were expressed as mean ± standard deviation (SD). Prior to ANOVA, the assumptions of normality and homogeneity of variance were verified using the Shapiro–Wilk test and Levene’s test, respectively. To evaluate the differences among the samples, a two-way analysis of variance (ANOVA) was performed for storage-dependent parameters, while one-way ANOVA was applied for fermentation monitoring parameters and the remaining physicochemical and functional properties. For post hoc multiple comparisons, the Bonferroni correction was applied following the two-way ANOVA, whereas Tukey’s test was employed after the one-way ANOVA to determine significant differences between the groups. All statistical analyses were conducted using IBM SPSS Statistics software (Version 26, Armonk, NY, USA), and the level of statistical significance was set at p < 0.05. High-throughput 16S rRNA amplicon sequencing data were excluded from these inferential statistical analyses due to sample pooling and were reported descriptively.

3. Results

3.1. Genotypic Verification of α-GTase Genes in Starter Strains

PCR analysis targeting the 4,6-$\alpha$-GTase (gtfB1) and 4,3-$\alpha$-GTase (gtfB2) genes within the candidate genomes confirmed the presence of these genetic determinants, clearly showing that specific amplification fragments of approximately 1000 bp successfully accumulated for L. reuteri PFC338 (4,6-α-GTase-positive) and L. fermentum PFC282 (4,3-α-GTase-positive). Conversely, no amplification product was present in the genomes of L. reuteri PSC73 (4,6-α-GTase-negative) and L. fermentum PFC268 (4,3-α-GTase-negative). These results genotypically verified the potential of the positive strains to encode for $\alpha$-GTase activity, establishing them as suitable models for evaluating in situ technological effects.

3.2. Monitoring of Sourdough Fermentation

Table 2. Total titratable acidity (TTA) and microbiological profiles of sourdoughs across successive backslopping cycles.

Sample	Backslopping Cycle	TTA	LAB * (on MRS)	LAB * (on GM17)	TAMB *	Yeast and Mold *
SDC	T0	0.60 ± 0.02 a	8.47 ± 0.01 cd	8.73 ± 0.04 bc	8.58 ± 0.11 b	3.67 ± 0.20 b
	BC1	0.62 ± 0.02 abc	8.80 ± 0.03 abc	8.60 ± 0.01 b	9.00 ± 0.10 a	3.25 ± 0.08 ab
	BC2	0.48 ± 0.01 de	8.80 ± 0.03 ab	8.60 ± 0.01 cd	9.23 ± 0.23 a	3.29 ± 0.01 c
	BC3	0.48 ± 0.02 b	8.83 ± 0.06 bc	9.04 ± 0.05 a	8.75 ± 0.15 abc	3.32 ± 0.04 c
SD338L	T0	0.59 ± 0.03 a	8.45 ± 0.15 de	9.15 ± 0.15 a	8.45 ± 0.15 bc	3.16 ± 0.09 c
	BC1	0.49 ± 0.03 d	8.60 ± 0.30 bcd	8.73 ± 0.04 b	9.00 ± 0.10 a	3.29 ± 0.02 ab
	BC2	0.49 ± 0.00 de	9.23 ± 0.23 a	9.15 ± 0.15 ab	9.15 ± 0.15 a	3.09 ± 0.05 d
	BC3	0.48 ± 0.06 b	9.15 ± 0.15 ab	9.15 ± 0.15 a	9.15 ± 0.15 a	ND
SD338H	T0	0.68 ± 0.09 a	8.00 ± 0.10 f	8.10 ± 0.10 d	8.00 ± 0.00 d	3.46 ± 0.00 bc
	BC1	0.72 ± 0.01 a	8.45 ± 0.15 cd	8.60 ± 0.00 b	8.60 ± 0.10 ab	ND
	BC2	0.72 ± 0.03 a	9.23 ± 0.23 a	9.23 ± 0.23 a	9.00 ± 0.10 a	ND
	BC3	0.71 ± 0.06 a	9.30 ± 0.10 a	9.30 ± 0.10 a	8.60 ± 0.30 bc	ND
SD73L	T0	0.70 ± 0.09 a	9.15 ± 0.15 a	8.83 ± 0.23 bc	9.30 ± 0.10 a	3.19 ± 0.05 c
	BC1	0.60 ± 0.01 abcd	9.00 ± 0.10 ab	9.00 ± 0.10 a	8.38 ± 0.38 b	2.83 ± 0.06 c
	BC2	0.53 ± 0.02 bcd	9.15 ± 0.15 a	9.15 ± 0.15 ab	8.83 ± 0.06 a	3.11 ± 0.00 d
	BC3	0.69 ± 0.13 a	8.83 ± 0.06 bc	9.00 ± 0.10 a	8.80 ± 0.20 abc	ND
SD73H	T0	0.59 ± 0.02 a	8.30 ± 0.00 def	8.30 ± 0.00 d	8.15 ± 0.15 cd	3.62 ± 0.26 b
	BC1	0.57 ± 0.10 bcd	9.15 ± 0.15 a	9.00 ± 0.00 a	9.00 ± 0.00 a	ND
	BC2	0.49 ± 0.02 de	8.77 ± 0.30 ab	8.72 ± 0.12 c	9.15 ± 0.15 a	ND
	BC3	0.53 ± 0.05 ab	8.83 ± 0.06 bc	9.00 ± 0.10 a	8.77 ± 0.17 abc	ND
SD282L	T0	0.59 ± 0.03 a	8.15 ± 0.15 ef	8.64 ± 0.01 c	9.00 ± 0.10 a	4.54 ± 0.02 a
	BC1	0.54 ± 0.05 bcd	8.29 ± 0.01 cd	9.00 ± 0.10 a	8.15 ± 0.15 b	3.32 ± 0.05 a
	BC2	0.59 ± 0.04 b	9.25 ± 0.05 a	9.00 ± 0.10 abc	9.00 ± 0.10 a	3.07 ± 0.00 d
	BC3	0.63 ± 0.08 ab	8.64 ± 0.04 c	9.00 ± 0.10 a	9.00 ± 0.20 ab	ND
SD282H	T0	0.40 ± 0.03 b	8.76 ± 0.07 bc	8.90 ± 0.10 abc	9.25 ± 0.05 a	ND
	BC1	0.64 ± 0.02 ab	8.23 ± 0.23 d	8.29 ± 0.01 c	8.29 ± 0.01 b	ND
	BC2	0.57 ± 0.01 bc	8.38 ± 0.08 b	8.80 ± 0.03 bc	8.23 ± 0.23 b	4.04 ± 0.02 a
	BC3	0.56 ± 0.04 ab	9.00 ± 0.10 ab	8.53 ± 0.23 b	9.10 ± 0.10 a	4.01 ± 0.05 a
SD268L	T0	0.58 ± 0.04 a	9.00 ± 0.10 ab	9.20 ± 0.05 a	8.69 ± 0.00 b	4.48 ± 0.00 a
	BC1	0.51 ± 0.02 cd	8.57 ± 0.27 bcd	9.00 ± 0.10 a	8.86 ± 0.09 a	3.18 ± 0.04 b
	BC2	0.50 ± 0.04 cde	9.15 ± 0.15 a	9.15 ± 0.15 ab	8.92 ± 0.08 a	3.13 ± 0.06 d
	BC3	0.61 ± 0.05 ab	9.20 ± 0.20 a	9.30 ± 0.10 a	8.79 ± 0.10 abc	ND
SD268 H	T0	0.66 ± 0.01 a	9.00 ± 0.10 ab	9.00 ± 0.10 ab	9.15 ± 0.15 a	ND
	BC1	0.52 ± 0.04 bcd	8.58 ± 0.11 bcd	8.15 ± 0.15 c	8.23 ± 0.23 b	ND
	BC2	0.43 ± 0.02 e	8.62 ± 0.15 b	8.23 ± 0.23 d	8.23 ± 0.23 b	3.50 ± 0.04 b
	BC3	0.54 ± 0.04 ab	9.15 ± 0.15 ab	8.54 ± 0.24 b	8.38 ± 0.08	3.84 ± 0.05 b

* Microbial populations are expressed in log CFU/g. Results represent mean values (n = 3) ± SD. Different lowercase superscript letters within the same column indicate statistically significant (p < 0.05) differences between sourdough groups within the same backslopping cycle (BC). T₀ represents the unfermented fresh mix immediately after inoculation and prior to fermentation. TTA values are expressed as percentage (%) of lactic acid equivalents. LAB: Lactic acid bacteria; TAMB: Total aerobic mesophilic bacteria; SD_C: Spontaneous sourdough control; SD₃₃₈: L. reuteri PFC338 (4,6-α-GTase-positive); SD₇₃: L. reuteri PSC73 (4,6-α-GTase-negative); SD₂₈₂: L. fermentum PFC282 (4,3-α-GTase-positive); SD₂₆₈: L. fermentum PFC268 (4,3-α-GTase-negative). Subscripts “L” and “H” represent low (10² CFU/g) and high (10⁶ CFU/g) initial inoculation levels, respectively. ND: Not detected (below the detection limit of 1.00 log CFU/g).

summarizes the TTA values of the sourdoughs fermented at 30 °C upon reaching the target pH of 4.5.

TTA values revealed that starter culture formulations, inoculation dosages (10² and 10⁶ CFU/g), and consecutive BCs exerted a statistically significant effect (p < 0.05) on acid accumulation (Table 2). Prior to fermentation (T₀), the dough samples generally exhibited similar (p > 0.05) TTA levels (0.58–0.70% expressed as lactic acid), whereas only SD_282H (0.40%) showed a significantly lower (p < 0.05) initial acidity. Throughout all BCs (BC₁, BC₂, and BC₃), SD_338H stood out with the highest TTA values (0.71–0.72%).

Starter culture formulations and inoculation dosages significantly influenced (p < 0.05) all microbial counts (LAB on MRS and GM17, TAMB, and yeast–mold) across all BCs (Table 2). Prior to fermentation (T₀), LAB counts ranged between 8.00 and 9.15 log CFU/g on MRS and 8.10 and 9.20 log CFU/g on GM17. By BC₃, the LAB counts maintained stability (8.64–9.30 log CFU/g on MRS; 8.53–9.30 log CFU/g on GM17), where samples SD_338H and SD_268L secured the highest cell loads. Similarly, TAMB remained stable throughout consecutive passages, shifting from an initial 8.00–9.30 log CFU/g (T₀) to 8.38–9.15 log CFU/g at BC₃. In contrast, yeast–mold counts declined sharply; while sample SD_282L had the highest initial load (4.54 log CFU/g) and two samples (SD_268H and SD_282H) showed no initial growth at T₀, the yeast–mold counts dropped below the detection limit in six out of nine samples by the final cycle (BC₃).

3.3. DGGE Characterization of the Microbial Profile

DGGE gel analysis revealed shifts in LAB profiles of the sourdough samples fermented with 4,6- and 4,3-α-GTase variants at different inoculation dosages at the end of BC₃ (Figure 1). Although each well received an equal concentration (20 ng) of PCR product (DNA), the resulting specific band profiles and intensities displayed variations depending on the starter cultures and inoculation dosages.

DGGE band profiles of sourdoughs fermented with L. reuteri strains showed intensity variations (Figure 1A). Band “a”, which remained weak in the SD_C, SD_338L, and SD_338H samples, displayed higher relative intensity in SD_73L and SD_73H. Meanwhile, band “b”, prominent in SD_C, weakened in both low-dose samples (SD_73L and SD_338L); however, it intensified in SD_338H but was less intense in SD_73H. In region “f”, the starter inoculation led to a reduction in the intensity of the thick band cluster present in SD_C. For the common bands “c” and “d”, the α-GTase-negative strains (SD_73L and SD_73H) had wider and darker profiles than their positive counterparts (SD_338L and SD_338H), reaching maximum width and peak intensity in SD_73H (Figure 1A). For sourdoughs fermented with L. fermentum strains (Figure 1B), band “b” showed higher intensity in SD_268L and SD_268H compared to both SD_C and the SD₂₈₂ counterparts. Similar to Figure 1A, L. fermentum strains inoculation decreased the intensity of the band structure in region “f” across all lanes relative to SD_C. Conversely, bands “c” and “d”, which were strong in L. reuteri strain profile, were nearly undetectable or fainter in all L. fermentum samples, reflecting potential species-specific variations. Additionally, band “e”, prominent especially in high-dose L. reuteri samples (SD_338H and SD_73H), remained faint in all L. fermentum treatments.

3.4. 16S rRNA Gene Sequencing Data and Quality Metrics

High-throughput 16S rRNA gene sequencing yielded a total of 1,252,423 high-quality reads to resolve the microbial composition and diversity of the sourdough samples. The filtered read count per sample varied between 99,409 and 179,553, with an average of 139,158 reads. The proportion of high-quality reads retained after the bioinformatic filtering pipeline suggested satisfactory library quality and provided sufficient sequencing depth for downstream taxonomic characterization.

3.5. Bacterial Community Stability and Alpha Diversity Analysis

Table 3. Alpha diversity metrics of composite pooled sourdough samples.

Sample	Chao1	Shannon	Simpson	Good’s Coverage
SDC	42.62 ± 5.85	3.24 ± 0.05	0.1814 ± 0.0101	0.9961 ± 0.0009
SD338L	75.64 ± 22.86	3.24 ± 0.03	0.1205 ± 0.0088	0.9912 ± 0.0024
SD338H	58.65 ± 10.09	3.24 ± 0.04	0.1077 ± 0.0067	0.9940 ± 0.0017
SD73L	62.79 ± 8.93	3.18 ± 0.05	0.1159 ± 0.0047	0.9926 ± 0.0014
SD73H	113.28 ± 25.78	3.09 ± 0.05	0.0656 ± 0.0039	0.9848 ± 0.0033
SD282L	54.57 ± 10.32	3.81 ± 0.02	0.2575 ± 0.0255	0.9945 ± 0.0016
SD282H	113.90 ± 35.14	3.40 ± 0.05	0.1386 ± 0.0080	0.9868 ± 0.0015
SD268L	52.58 ± 11.01	3.17 ± 0.03	0.1538 ± 0.0094	0.9944 ± 0.0016
SD268H	90.73 ± 16.44	2.08 ± 0.04	0.0584 ± 0.0041	0.9877 ± 0.0016

Data represents the mean values ± SD calculated across 10 computational rarefaction iterations performed at a standardized sequencing depth of 2000 reads per sample. SD_C: Spontaneous sourdough control; SD₃₃₈: L. reuteri PFC338 (4,6-α-GTase-positive); SD₇₃: L. reuteri PSC73 (4,6-α-GTase-negative); SD₂₈₂: L. fermentum PFC282 (4,3-α-GTase-positive); SD₂₆₈: L. fermentum PFC268 (4,3-α-GTase-negative). Subscripts “L” and “H” represent low (10² CFU/g) and high (10⁶ CFU/g) initial inoculation levels, respectively.

and Figure 2 show the bacterial richness and diversity indices (Chao1, Shannon, and Simpson) of the composite pooled sourdough samples. Good’s coverage values were at or above 98.48% for all samples, suggesting that the standardized rarefaction subset of 2000 reads per sample was adequate to represent the primary taxonomic profiles. Aligned with these metrics, the rarefaction curves approached a plateau, indicating that the sequencing depth captured the bacterial diversity within the groups.

The bacterial richness (Chao1) of the sourdough samples varied among the starter culture formulations and inoculation dosages. The spontaneous control group (SD_C) exhibited the lowest taxonomic richness with a value of 42.62, while starter culture incorporation generally increased the taxonomic richness of the sourdoughs (Figure 2A). In line with these variations, the Shannon and Simpson indices, which reflect the diversity and dominance profiles, showed shifts based on strain variants. SD_282L exhibited the highest Shannon index (3.81) along with a Simpson dominance value of 0.2575, suggesting potential changes in taxonomic distribution within the ecosystem. Conversely, SD_268H displayed the lowest diversity metrics, with a Shannon index of 2.08 and a Simpson value of 0.0584 (Figure 2B). Specifically, the enzyme-active strain at a low dosage (SD_282L) contributed to a different profile, establishing a high Shannon index despite a higher microbial dominance profile, particularly when comparing the 4,3-$\alpha$-GTase variants, whereas the 4,3-$\alpha$-GTase-negative strain at a high inoculation level (SD_268H) influenced the overall uniformity and dominant profiles in the sourdough ecosystem.

3.6. Bacterial Community Composition and Taxonomic Distribution

Genus-level taxonomic profiles, obtained after filtering plant organelle sequences (chloroplast and mitochondria) and converting the data to relative abundances, suggested that LAB heavily dominated the microbiota structure. Across all analyzed samples, the genera Lactobacillus, Weissella, Leuconostoc, Lactococcus, and Pediococcus constituted the core components of the bacterial community (Figure 3).

Compared to SD_C, starter culture inoculation induced distinct shifts in the bacterial composition. Notably, in the SD_268H sample, inoculated with a high dosage of the 4,3-α-GTase-negative L. fermentum PFC268 strain, the genus Lactobacillus constituted most of the population with a high relative abundance of 94.90%, leading to a lower relative abundance of other taxa. Conversely, the SD_338L sample, which featured the 4,6-α-GTase-positive L. reuteri PFC338 strain, maintained a co-existence of Weissella (40.90%), Lactococcus (22.98%), and Lactobacillus (18.83%) at high relative abundances. Similarly, in the SD_282L sample inoculated with the 4,3-α-GTase-positive L. fermentum PFC282 strain, Lactobacillus (39.83%) and Leuconostoc (23.02%) emerged as the dominant taxa, while the “others” group reached 23.85%, revealing a notable variation in taxonomic distribution (Figure 3). As a general trend, increasing the initial inoculation dosage (from 10² to 10⁶ CFU/g) enhanced the dominance of the respective genera and reduced community heterogeneity. Hierarchical clustering and heatmap analysis further supported these dosage- and strain-associated similarity patterns among the samples (Figure 4).

The hierarchical clustering tree indicated that the high-dose samples SD_73H and SD_338H, which had similar compositions, clustered within the same primary group, while the taxonomically uniform SD_268H sample formed a distinct branch. A Venn diagram further delineated the shared and unique genera among the groups (Figure 5).

The Venn diagram revealed that 10 genera were shared among all analyzed sourdough groups (Figure 5). There were five specific genera present exclusively in both enzyme-positive groups (4,3- and 4,6-α-GTase) but completely absent from the spontaneous control and α-GTase-negative groups. Furthermore, the 4,3-α-GTase-positive group alone harbored 47 unique genera.

3.7. Beta Diversity and Community Structure

PCoA analysis based on distance matrices, performed to evaluate the bacterial community similarities among the samples, suggested that the starter culture type and inoculation level distinctly influenced the sourdough ecosystem (Figure 6).

The first two principal components in Figure 6 explained 70.38% of the total variation (PC1: 46.43%; PC2: 23.95%), delineating the taxonomic variation patterns among the groups. While SD_C occupied a central position on the PCoA plot, starter culture inoculation shifted the bacterial profiles in distinct directions. Notably, samples SD_268H (light yellow) and SD_268L (light green) separated from both SD_C and all other groups, positioning themselves on the far-left outer margins of the plot. Conversely, the samples inoculated with L. reuteri strains (PFC338 and PSC73) exhibited a taxonomic convergence. The high-dose samples SD_338H (olive green) and SD_73H (dark green) clustered closely together in the upper right quadrant, while the low-dose samples SD_338L (light blue) and SD_73L (dark blue) were tightly grouped in the lower right quadrant. In contrast, the samples inoculated with L. fermentum PFC282 strain (pink and brown) were widely separated.

3.8. Viscosity and Rheological Properties of Sourdoughs

The Power Law model evaluated the rheological flow behavior and viscosity parameters of the matured sourdough preferments prepared at an inoculation level of 10⁶ CFU/g at the end of the third backslopping cycle (BC₃) for use as breadmaking starters (

Table 4. Rheological parameters of sourdough preferments.

Sample	Consistency Coefficient (Pa.sn)	Flow Behavior Index (n)	R2
SDC	3.45 ± 1.08 b	0.78 ± 0.11	0.98
SD338H	20.48 ± 6.54 a	0.65 ± 0.04	0.99
SD73H	19.44 ± 6.19 a	0.69 ± 0.05	0.99
SD282H	18.40 ± 4.22 a	0.70 ± 0.04	0.98
SD268H	13.33 ± 2.92 ab	0.76 ± 0.04	0.99

Results were expressed as mean values (n = 3) ± SD. Different lowercase superscript letters within the same column indicate statistically significant (p < 0.05) differences between sourdough samples. SD_C: Spontaneous sourdough control; SD_338H: L. reuteri PFC338 (10⁶ CFU/g); SD_73H: L. reuteri PSC73 (10⁶ CFU/g); SD_282H: L. fermentum PFC282 (10⁶ CFU/g); SD_268H: L. fermentum PFC268 (10⁶ CFU/g).

).

According to the rheological modeling data, the flow behavior index values of all samples, including SD_C, were below 1. There was no statistically significant (p > 0.05) difference among the flow behavior index values of the samples. Conversely, starter culture inoculation generally induced a statistically significant (p < 0.05) variations in the consistency coefficient values. While the consistency coefficient of SD_C was 3.45 Pa.sⁿ, all sourdoughs fermented with starter cultures, except for SD_268H, exhibited significantly higher (p < 0.05) consistency coefficient values compared to SD_C.

3.9. Specific Volume and Loaf Characteristics of Sourdough Breads

Table 5. Technological and physical properties of sourdough breads.

Bread Sample	Loaf Weight (g)	Loaf Volume (mL)	Specific Volume (mL/g)	Baking Loss (%)
SDC-Bread	137.30 ± 1.35	458.00 ± 12.00 c	3.34 ± 0.05 d	14.19 ± 0.84
SD338H-Bread	138.65 ± 0.48	500.00 ± 5.77 b	3.61 ± 0.04 b	13.34 ± 0.30
SD73H-Bread	139.52 ± 1.22	455.00 ± 5.00 c	3.19 ± 0.01 e	12.80 ± 0.76
SD282H-Bread	139.70 ± 0.40	542.50 ± 12.50 a	3.88 ± 0.08 a	12.69 ± 0.25
SD268H-Bread	137.30 ± 2.60	476.00 ± 6.00 c	3.47 ± 0.02 c	14.19 ± 1.63

Results were expressed as mean values (n = 3) ± SD. Different lowercase superscript letters within the same column indicate statistically significant (p < 0.05) differences between sourdough samples. SD_C-Bread: Spontaneous sourdough bread control; SD_338H-Bread: L. reuteri PFC338 (10⁶ CFU/g); SD_73H-Bread: L. reuteri PSC73 (10⁶ CFU/g); SD_282H-Bread: L. fermentum PFC282 (10⁶ CFU/g); SD_268H-Bread: L. fermentum PFC268 (10⁶ CFU/g).

summarizes the specific volume and loaf characteristics of the breads produced from the sourdough samples.

There was no statistically significant difference (p > 0.05) among the loaf weights (137.30–139.70 g) and baking loss rates (12.69–14.19%) of the sourdough bread samples. Conversely, starter culture inoculation exerted a significant effect (p < 0.05) on specific volume values. The SD_282H-Bread sample, prepared with the 4,3-α-GTase-positive L. fermentum PFC282 strain, recorded the highest loaf volume (542.50 mL) and specific volume (3.88 mL/g) values (p < 0.05). Similarly, the SD_338H-Bread sample produced with the 4,6-α-GTase-positive L. reuteri PFC338 strain led to a significant increase in loaf specific volume (3.61 mL/g) compared to SD_C-Bread (3.34 mL/g) (p < 0.05). However, the specific volumes of the SD_268H-Bread (3.47 mL/g) and SD_73H-Bread (3.19 mL/g) samples, produced with α-GTase-negative strains, were statistically significantly lower (p < 0.05) than those of the breads produced with α-GTase-positive strains.

3.10. Textural Properties of Sourdough Breads During Storage

Table 6. Textural profile analysis of sourdough breads during seven days of storage.

Textural Parameter	Bread Sample	Day 0	Day 3	Day 5	Day 7
Hardness (N)	SDC-Bread	3.72 ± 0.29 Da	9.53 ± 0.69 Ca	11.89 ± 0.71 Bb	23.42 ± 1.64 Aa
	SD338H-Bread	3.70 ± 0.10 Ca	8.59 ± 0.88 Bab	14.73 ± 0.78 Aa	15.37 ± 1.00 Ab
	SD73H-Bread	3.29 ± 0.47 Da	7.18 ± 0.26 Cbc	16.43 ± 0.98 Ba	24.24 ± 1.71 Aa
	SD282H-Bread	2.63 ± 0.14 Ca	7.33 ± 0.21 Bbc	9.03 ± 0.29 Bc	11.53 ± 0.85 Ac
	SD268H-Bread	2.11 ± 0.10 Da	6.36 ± 0.28 Cc	9.01 ± 0.37 Bc	15.53 ± 0.99 Ab
Chewiness (N)	SDC-Bread	22.04 ± 1.39 Cab	39.63 ± 2.12 Ba	50.57 ± 2.45 Aa	53.35 ± 2.35 Aa
	SD338H-Bread	22.57 ± 0.62 Ca	37.87 ± 3.15 Ba	52.88 ± 2.40 Aa	52.05 ± 2.12 Aa
	SD73H-Bread	23.24 ± 2.82 Ca	39.07 ± 2.76 Ba	51.61 ± 1.77 Aa	51.70 ± 4.28 Aa
	SD282H-Bread	16.87 ± 0.80 Bbc	37.75 ± 0.70 Aa	34.03 ± 1.43 Ab	36.44 ± 3.26 Ab
	SD268H-Bread	14.17 ± 0.41 Cc	30.31 ± 1.04 Bb	37.29 ± 1.42 Ab	41.60 ± 1.58 Ab
Cohesiveness	SDC-Bread	0.74 ± 0.01 Ab	0.54 ± 0.01 Bb	0.47 ± 0.01 Ca	0.38 ± 0.01 Db
	SD338H-Bread	0.78 ± 0.00 Aa	0.57 ± 0.01 Bb	0.50 ± 0.01 Ca	0.43 ± 0.01 Da
	SD73H-Bread	0.79 ± 0.01 Aa	0.55 ± 0.02 Bb	0.50 ± 0.02 Ca	0.40 ± 0.01 Dab
	SD282H-Bread	0.79 ± 0.01 Aa	0.57 ± 0.01 Bb	0.48 ± 0.02 Ca	0.41 ± 0.01 Dab
	SD268H-Bread	0.80 ± 0.01 Aa	0.62 ± 0.02 Ba	0.49 ± 0.01 Ca	0.43 ± 0.02 Da
Gumminess (N)	SDC-Bread	2.78 ± 0.18 Ca	5.35 ± 0.48 Ba	6.49 ± 0.44 Aa	7.36 ± 0.35 Aab
	SD338H-Bread	2.86 ± 0.07 Ca	5.24 ± 0.45 Ba	7.00 ± 0.41 Aa	7.37 ± 0.44 Aab
	SD73H-Bread	2.55 ± 0.32 Cab	4.90 ± 0.40 Bab	6.43 ± 0.38 Aa	6.63 ± 0.66 Ab
	SD282H-Bread	2.07 ± 0.09 Bab	4.07 ± 0.08 Ab	4.45 ± 0.24 Ab	4.76 ± 0.41 Ac
	SD268H-Bread	1.68 ± 0.08 Cb	4.08 ± 0.24 Bb	4.50 ± 0.20 Bb	7.58 ± 0.80 Aa
Springiness	SDC-Bread	7.95 ± 0.08 Aab	7.71 ± 0.07 Bb	7.70 ± 0.12 Bc	7.39 ± 0.10 Cc
	SD338H-Bread	7.87 ± 0.06 Ab	7.73 ± 0.08 Ab	7.69 ± 0.07 ABc	7.54 ± 0.09 Bbc
	SD73H-Bread	8.13 ± 0.07 Aa	7.94 ± 0.13 Ba	7.74 ± 0.13 Cbc	7.72 ± 0.10 Cb
	SD282H-Bread	8.14 ± 0.06 Aa	8.05 ± 0.05 ABa	7.93 ± 0.04 Bab	7.65 ± 0.06 Cb
	SD268H-Bread	8.12 ± 0.07 Aa	8.01 ± 0.04 Aa	8.01 ± 0.07 Aa	7.98 ± 0.05 Aa
ANOVA (p values)	Hardness (N)	Chewiness (N)	Cohesiveness	Gumminess (N)	Springiness
Sample (S)	p < 0.001	p < 0.001	p < 0.001	p < 0.001	p < 0.001
Day (D)	p < 0.001	p < 0.001	p < 0.001	p < 0.001	p < 0.001
Interaction (S × D)	p < 0.001	p < 0.001	p = 0.001	p < 0.001	p = 0.002

Results were expressed as mean values (n = 3) ± SD. Different lowercase superscript letters within the same column indicate statistically significant (p < 0.05) differences between bread samples for the same storage day. Different uppercase superscript letters within the same row indicate statistically significant (p < 0.05) differences between storage days (0, 3, 5, and 7 days) for the same sample. SD_C-Bread: Spontaneous sourdough bread control; SD_338H-Bread: L. reuteri PFC338 (10⁶ CFU/g); SD_73H-Bread: L. reuteri PSC73 (10⁶ CFU/g); SD_282H-Bread: L. fermentum PFC282 (10⁶ CFU/g); SD_268H-Bread: L. fermentum PFC268 (10⁶ CFU/g).

summarizes the textural properties of the breads produced from the sourdough samples evaluated over a seven-day storage period.

All bread samples showed statistically significant (p < 0.05) increases in hardness values on the final day of storage (day 7) compared to the beginning of storage. On day 0, no statistically significant differences (p > 0.05) were observed in the hardness values among all bread samples, which ranged from 2.11 N to 3.72 N. By the final day of storage, while SD_C-Bread and SD_73H-Bread reached the highest hardness levels with values of 23.42 N and 24.24 N, respectively, the hardness of SD_282H-Bread (11.53 N) remained significantly lower (p < 0.05) than all other samples.

Overall, the chewiness and gumminess values of the breads increased as the storage period extended; however, this increase varied among the samples and exhibited minor fluctuations. For the SD_282H-Bread sample, chewiness values stabilized after day 3, remaining statistically similar (p > 0.05) among days 3, 5, and 7. Notably, on the final day of storage, SD_282H-Bread distinguished itself by exhibiting a statistically significantly (p < 0.05) lower gumminess value compared to the other samples. Specifically, while SD_C-Bread reached a gumminess of 7.36 N and a chewiness of 53.35 N, the gumminess and chewiness of SD_282H-Bread were 4.76 N and 36.44 N, respectively (p < 0.05).

The cohesiveness decreased significantly (p < 0.05) in all samples as the storage period extended. Among the fresh breads, all breads produced using starter cultures possessed significantly higher (p < 0.05) cohesiveness values (0.78–0.80) compared to SD_C-Bread (0.74). At the end of storage, the cohesiveness value of SD_C-Bread dropped to 0.38, whereas the cohesiveness values of the SD_338H-Bread and SD_268H-Bread samples were 0.43 (p < 0.05). The springiness value of all samples, except for SD_268H-Bread, decreased significantly (p < 0.05) on the final day compared to day 0 of storage. Conversely, the springiness value of SD_268H-Bread remained similar (p > 0.05) throughout the storage period. Among all samples, SD_C-Bread and SD_338H-Bread exhibited the lowest springiness values on day 7.

3.11. Thermal Properties of Sourdough Breads During Storage

Figure 7 summarizes the thermal properties of the breads over a seven-day storage period.

Throughout the storage period, the endothermic peak onset temperatures (T₀) for the melting of amylopectin crystals in all bread samples remained at approximately 40 °C. Both storage time and sample variations demonstrated a significant effect (p < 0.001) on the enthalpy (ΔH) values of the breads. Prolonging the storage time increased the retrogradation enthalpy values of all breads, and the difference between the first day (day 0) and the final day (day 7) of storage was significant (p < 0.05). At the beginning of storage, SD_268H-Bread (176.46 J/g) and SD_C-Bread (147.36 J/g) exhibited the highest initial enthalpy values (p > 0.05) (Figure 7b), whereas the other samples had significantly (p < 0.05) lower enthalpy values than SD_268H-Bread. On day 3 of storage, the SD_338H-Bread (170.88 J/g) and SD_282H-Bread (175.90 J/g) samples produced with α-GTase-positive strains showed the lowest enthalpies (p < 0.05) among all samples. Conversely, on the same day, SD_268H-Bread (285.01 J/g), had the highest enthalpy (p < 0.05). On days 5 and 7, although significantly lower (p < 0.05) than the other samples, the enthalpy values of the SD_338H-Bread, SD_282H-Bread, and SD_C-Bread samples remained similar (p > 0.05) to one another. At the end of storage, the sample with the lowest amylopectin retrogradation enthalpy peak was SD_338H-Bread (236.10 J/g) (Figure 7a), while the samples with the highest retrogradation enthalpy values were SD_268H-Bread (311.63 J/g) (Figure 7b) and SD_73H-Bread (294.54 J/g), respectively. The enthalpy values of all samples remained similar (p > 0.05) between days 5 and 7.

3.12. Estimated Glycemic Index of Sourdough Breads

Table 7. In vitro starch hydrolysis index, estimated glycemic index, and glycemic category of sourdough breads.

Bread Sample	Hydrolysis Index (HI)	Estimated Glycemic Index (eGI)	Glycemic Category
White Bread	100.00 ± 5.95 a	94.61 ± 3.27 a	High
SD338H-Bread	52.99 ± 0.63 c	68.80 ± 0.35 c	Medium
SD73H-Bread	83.31 ± 2.65 b	85.45 ± 1.46 b	High
SD282H-Bread	85.89 ± 1.98 b	86.86 ± 1.09 b	High
SD268H-Bread	91.41 ± 0.33 b	89.90 ± 0.19 b	High

Results were expressed as mean values (n = 3) ± SD. Different lowercase superscript letters within the same column indicate statistically significant (p < 0.05) differences between bread samples. SD_338H-Bread: L. reuteri PFC338 (10⁶ CFU/g); SD_73H-Bread: L. reuteri PSC73 (10⁶ CFU/g); SD_282H-Bread: L. fermentum PFC282 (10⁶ CFU/g); SD_268H-Bread: L. fermentum PFC268 (10⁶ CFU/g). Glycemic categories were classified based on Venn and Green [27] guidelines: low (<55), medium (55–69), and high (≥70).

presents the hydrolysis index (HI), estimated glycemic index (eGI), and glycemic category results of the bread samples.

Sourdough fermentation with different starter cultures demonstrated a significant reducing effect (p < 0.05) on the HI and eGI values of the breads compared to the reference white bread, which was classified in the “high-glycemic” category. The SD_338H-Bread sample, produced with the 4,6-α-GTase-positive L. reuteri PFC338 strain, scored an HI of 52.99 and an eGI of 68.80; these values were significantly lower (p < 0.05) than those of all other samples, making it the only sample in the “medium-glycemic” category. Conversely, the remaining sourdough formulations (SD_73H-Bread, SD_282H-Bread and SD_268H-Bread) still remained within the “high-glycemic” category, exhibiting statistically similar (p > 0.05) eGI values to one another.

4. Discussion

The selection of paired α-GTase-positive and -negative models within the same species (L. reuteri and L. fermentum) provided a framework for tracking how the starch matrix undergoes restructuring during fermentation. Both L. reuteri and L. fermentum, frequently isolated from sourdough ecosystems [28], possess documented capabilities to synthesize α-GTase enzymes [12,29].

The TTA results (0.48–0.72%) obtained during fermentation align with the lower limits reported by Syrokou et al. [30], who documented TTA values of 0.50–1.59% in wheat sourdoughs. The high TTA levels (0.71–0.72%) maintained by sample SD_338H throughout all BCs suggest that the L. reuteri PFC338 strain rapidly adapted to the dough environment, contributing to a stable establishment in the matrix. The stability of the LAB and TAMB loads between 8 and 9.30 log CFU/g across successive BCs supports the overall adaptability of the microflora under these fermentation conditions. This microbial stability and acidity development likely created a biopreservative barrier against other competitive microorganisms in the ecosystem. Indeed, the decline in yeast and mold counts as the cycles progressed could be attributed to the antifungal activity of organic acids or potential compounds like reuterin synthesized particularly by L. reuteri strains [31]. The high and stable metabolic activity of the LAB population suggests the successful establishment of the biotic infrastructure required for in situ α-GTase functionality in the dough rheology and starch matrix during the subsequent stages.

The DGGE profiles of the samples at BC₃ indicated an impact of starter culture inoculation on the bacterial profiles. The addition of LAB starters led to a reduction in the intensity of the complex microbial band clusters present in SD_C. This outcome suggests that the starter cultures adapted to the ecosystem, influencing the overall bacterial band profile. Indeed, the literature [32] indicates that target LAB species dominate at the end of sourdough fermentations initiated with a high inoculation dosage, although the DGGE band profile may still include other strains of the same species naturally present in the flour microflora. However, the presence of bands in distinct positions and variations in band intensities reflect potential shifts within the dough matrix.

The lower Chao1 value observed in SD_C can be attributed to the selective proliferation of specific dominant taxa from the native flour microflora that adapted to the fermentation conditions. Conversely, the overall increase in taxonomic richness following starter culture addition suggests that the starter culture inoculation enriched the sourdough matrix and expanded the existing microflora diversity. The variations observed in the Shannon and Simpson indices across different strains highlight the potential role of starter culture genotypic traits and inoculation dosages in modulating microbial profiles. Particularly when comparing the 4,3-α-GTase variants, the enzyme-active L. fermentum PFC282 strain at a low dosage (SD_282L) promoted a more balanced and stratified community structure, yielding the highest Shannon index along with a distinct Simpson dominance profile within the dataset. In contrast, using the 4,3-α-GTase-negative L. fermentum PFC268 strain at a high inoculation level (SD_268H) markedly reduced the evenness and diversity in the ecosystem compared to other samples. This 4,3-α-GTase-negative strain, inoculated at a high population (10⁶ CFU/g), free from the potential metabolic burden of in situ enzyme synthesis, and acid-tolerant, likely dominated the limited carbon sources in the dough environment, limiting the relative abundance of alternative taxa and leading to a community structure with low evenness [28,33]. The genus-level taxonomic distribution of the samples supports this trend. While the genus Lactobacillus was predominantly represented in SD_268H, the samples inoculated with low dosages of α-GTase-positive strains (SD_338L and SD_282L) maintained a more balanced and heterogeneous distribution of Lactobacillus, Weissella, Lactococcus, and Leuconostoc. This outcome can be tentatively attributed to the in situ α-GTase activity at low initial populations, which potentially generates alternative prebiotic carbon sources, such as modified glucans or oligomers (e.g., isomalto-oligosaccharides), within the dough matrix, thereby supporting bacterial homeostasis [33]. However, the increased Lactobacillus dominance and reduced heterogeneity in samples inoculated with high dosages of the same strains (SD_338H and SD_282H) were likely driven by the intense metabolic competition imposed by the initial cell load. The Venn diagram results identified 10 shared genera forming the “core bacterial community” of the system, while the absence of unique taxa in the SD_C indicates that the starter cultures modified the bacterial profile of the dough. The presence of 47 unique genera in the samples produced with the 4,3-α-GTase-positive strain suggests that the diversity of in situ synthesized structural carbohydrates expanded through specific synergistic and metabolic interactions established between the starter culture and the flour microflora. Furthermore, the distribution of samples on the PCoA coordinate plot indicates that the bacterial community structure was shaped not only by the starter culture species but also by the strain genotype and inoculation dosage. The positioning of L. fermentum PFC268-fermented samples (SD_268L and SD_268H) on the far-left outer margin of the coordinate axis, separated from the other samples, further supports the high adaptability and subsequent impact of this starter culture on the bacterial profile. Conversely, the distinct taxonomic convergence displayed by the doughs inoculated with both L. reuteri strains (PFC338 and PSC73), clustering according to inoculation dosages rather than enzymatic variations, suggests that the compositional stabilization capacity of L. reuteri in the flour matrix relates to its species-specific metabolic traits rather than the presence of α-GTase [28].

All samples exhibited a shear-thinning flow behavior, which is characteristic of flour-based matrices [34,35]. The statistically similar (p > 0.05) flow behavior index values among the samples indicate that starter culture variations and enzymatic activities did not alter the fundamental flow mechanism of the dough. Conversely, starter culture inoculation generally induced statistically significant (p < 0.05) variations in consistency coefficient values, with the SD_268H sample showing values statistically comparable (p > 0.05) to the SD_C. The higher consistency coefficients in most of the starter-inoculated preferments compared to SD_C can be attributed to the exopolysaccharides (EPS) potentially synthesized by these strains during fermentation [36].

The improved specific volume values recorded for the breads produced with the 4,3-$\alpha$-GTase-positive L. fermentum PFC282 (SD_282H-Bread) and 4,6-$\alpha$-GTase-positive L. reuteri PFC338 (SD_338H-Bread) strains are consistent with the core hypothesis of this study. Notably, this volumetric performance exceeds the basal specific volume values (2.13 mL/g) reported by Verdonck et al. [37] for breads produced using sourdough. Furthermore, our findings align with a previous study [38] reporting a 2.3–11.4% increase in bread specific volume compared to the control upon adding the 4,6-α-GTase enzyme to the dough environment. The higher specific volumes of the α-GTase-positive samples compared to both SD_C and the breads produced with α-GTase-negative strains (Table 5) can be attributed to the in situ synthesis of modified α-glucans and oligosaccharides via enzymatic transglucosylation. Similar to the network-building and water-binding behavior of microbial hydrocolloids (such as dextrans) in cereal systems, these polysaccharide structures likely exerted a hydrocolloid-like effect within the dough matrix, interacting favorably with the gluten network to enhance dough stability and gas retention capacity [39]. In addition to this enzymatic mechanism, alternative and contributory biochemical factors, such as moderate acidification affecting the gluten network, microbial proteolysis influencing dough extensibility, and in situ EPS production [40], may have also affected bread specific volumes by modifying dough rheology [37]. Additionally, sugars like glucose and maltose released as a result of the enzymatic reaction may have provided an extra carbon source, thereby stimulating greater CO₂ production [41].

The general increasing trend observed in the hardness, gumminess, and chewiness values of the breads during the seven-day storage period (Table 6), along with the decrease in cohesiveness and springiness values, can be attributed to the retrogradation of starch molecules gelatinized during baking and the migration of water from the crumb to the crust [3]. On day 0, no significant differences (p > 0.05) in hardness were observed across the freshly baked samples. However, when these textural profiles are evaluated together with thermal dynamics (Figure 7) it is shown that on the final day of storage, the SD_282H-Bread sample exhibited significantly lower (p < 0.05) hardness and gumminess compared to all other samples, which aligns with the literature data reporting that α-GTase enzyme activity reduces crumb hardness, gumminess, and chewiness at the end of storage [38]. This anti-staling effect can be associated with the ability of the L. fermentum PFC282 strain to alter the branching architecture of flour starch through its in situ transferase activity [38] and to synthesize structural EPSs [42]. The general increase observed in the enthalpy values during bread storage demonstrates at the molecular level that starch chains recrystallize over time and enter the staling process [43]. Specifically, during the early stages of storage (day 3), breads produced with α-GTase-positive strains (SD_338H-Bread and SD_282H-Bread) exhibited significantly lower (p < 0.05) retrogradation enthalpies compared to breads produced with $\alpha$-GTase-negative strains (SD_73H-Bread and SD_268H-Bread) and the SD_C-Bread, suggesting a delayed crystallization rate. This thermal stability observed during the first 3 days of storage can be potentially explained by the ability of the α-GTase enzyme to disrupt the linear symmetry of amylopectin by increasing the number of short branches and the ratio of α-(1 → 6) linkages in the starch chains; this modification geometrically hinders the formation of regular crystal lattices and slows down the crystallization rate [38]. However, toward the end of the storage period (on days 5 and 7), the enthalpy values of the SD_338H-Bread and SD_282H-Bread samples were similar to that of SD_C-Bread, implying that α-GTase enzyme activity slows down crystallization kinetics in the early storage phase rather than completely halting amylopectin retrogradation.

The decrease observed in the HI and eGI values of the breads compared to white bread can be attributed to the organic acids formed during sourdough fermentation, which limit starch gelatinization and promote gluten–starch interactions that reduce starch digestibility [44]. The observation that the SD_338H-Bread sample, produced with L. reuteri PFC338, exhibited the lowest eGI value among all samples, becoming the only sample classified in the “medium-glycemic” category, can be linked to the ability of the 4,6-α-GTase enzyme to cleave α-(1 → 4) glycosidic bonds in starch molecules and transfer glucose residues to α-(1 → 6) positions. This enzymatic modification potentially increases the branching density of starch and establishes novel linkage structures unrecognized by digestive enzymes. Consequently, it reduces rapidly digestible starch fractions while promoting the formation of slowly digestible and resistant starches, thereby lowering the eGI by reducing the hydrolysis rate of amylolytic enzymes and delaying glucose release [38,45]. Conversely, the other α-GTase-positive sample, L. fermentum PFC282 (SD_282H-Bread), remained in the “high-glycemic” group similarly to the α-GTase-negative strains (SD_73H-Bread and SD_268H-Bread) suggesting that enzyme specificity exerts molecular impacts on starch architecture. These findings highlight the potential of utilizing the in situ 4,6-α-GTase activity of L. reuteri PFC338 to develop functional bakery products with a reduced eGI.

Despite these observed physical and technological changes, a key limitation of this study is that the in situ expression or specific enzymatic activity of the 4,6-$\alpha$- and 4,3-$\alpha$-GTases during sourdough fermentation was not directly measured using enzyme assays or RT-qPCR of gtfB genes. The suggested structural modifications of starch and the subsequent changes were deduced from genetic confirmation and the observed differences between the $\alpha$-GTase-positive and -negative strains. Therefore, future studies focusing on proteomic confirmation or starch linkage analysis are needed to clarify the expression kinetics of these enzymes within the sourdough matrix. Furthermore, since the glycemic properties in this study were evaluated solely through in vitro starch hydrolysis and human postprandial glycemic responses were not measured, future in vivo clinical trials are required to validate these potential metabolic benefits.

5. Conclusions

This study demonstrates that α-GTase-positive starter cultures were associated with improved bread quality and lower estimated starch digestibility in sourdough ecosystems. The selected L. reuteri and L. fermentum starter cultures successfully adapted to the matrix, modifying the native flour bacterial profile to establish stable acidity. Furthermore, high-throughput sequencing revealed that the inoculation level and strain genotype shaped the bacterial assembly, where lower-dose α-GTase-positive starters preserved a balanced taxonomic evenness while high-dose inoculation led to pronounced Lactobacillus prevalence. The use of these strains corresponded to enhanced apparent viscosity without altering characteristically pseudoplastic flow behavior, along with significantly improved bread specific volume. During storage, the application of the 4,3-α-GTase-positive L. fermentum PFC282 strain was linked to retarded amylopectin retrogradation, thereby significantly reducing hardness and gumminess, alongside lowering chewiness, to exert a distinct anti-staling effect. Crucially, the 4,6-α-GTase-positive L. reuteri PFC338 strain showed an association with lower enzymatic hydrolysis, which may be related to potential modifications in structural linkages, thereby emerging as a prominent strain profile capable of shifting bread from the “high-glycemic” to the “medium-glycemic” category. In conclusion, the application of α-GTase-positive starter cultures holds significant potential for the development of both long shelf life and reduced eGI clean-label products in the bakery industry, while further mechanistic investigations are warranted to fully elucidate in situ enzymatic pathways.