One hundred S. thermophilus strains were isolated from 80 naturally fermented dairy products, including koumiss, kurut, fermented cow milk, and fermented goat milk, from six provinces in China. These strains were identified by 16 S rDNA sequencing and stored in Wecare Probiotics Co., Ltd.

The isolated S. thermophilus strains were cultured in M17 broth (Oxoid, Basingstoke, United Kingdom) supplemented with 1% lactose at 37°C for 16–18 h in an anerobic environment. Their optical density (OD) was tested after 5 times diluted with distilled water at 600 nm. The bacterial cells were collected by centrifugation at 7,000 × g for 5 min, and the collected cells were washed once with sterile saline and resuspended in saline for further use. Monocultures were prepared by inoculating at a concentration of 1 × 10⁶ cfu/mL S. thermophilus strains individually in fresh milk pasteurized at 95°C for 5 min (Brightdairy, Shanghai, China) or UHT milk. The co-cultures of different S. thermophilus strains were prepared by inoculating at a ratio of 1:1 in fresh milk pasteurized at 95°C for 5 min at a total concentration of 1 × 10⁶ CFU/mL. For milk fermentation the temperature was maintained at 42°C.

Acid production during fermentation was expressed as changes in pH and increases in the titratable acidity (TA) of the final product. After fermentation, the fermented yogurt samples were titrated with 0.1 mol/L NaOH (Merck, Darmstadt, Germany) under continuous magnetic stirring until pH 8.30 was reached. The amount of 0.1 mol/L NaOH (mL) required to titrate 100 g of yogurt was referred to as the TA. The pH changes of the samples were determined from the start of the fermentation every 5 min using a Cinac pH meter system (Alliance Instruments, AMS Company, France).

A fermentation medium containing different carbon sources (initial concentration of 40 g/L), yeast extract (10 g/L), peptone (10 g/L), sodium acetate (2 g/L), sodium glutamate (2 g/L), K₂HPO₄ (2 g/L), Tween 80 (1 g/L), MgSO₄ (0.25 g/L), and MnSO₄ (0.05 g/L) was prepared. All of the ingredients used to prepare the medium were purchased from Sinopharm (Beijing, China). The initial pH was adjusted to 6.8 ± 0.2, and the medium was sterilized at 121°C for 20 min.

pH-controlled fermentation by each strain was carried out in a 3-L jar fermentor (Korea Fermentor Co., Inchon, Korea) containing 2 L medium with an inoculum size of 2% (v/v). All experiments were performed in triplicate.

Cells were harvested under aseptic conditions at the beginning of the stationary phase of fermentation by centrifugation at 10,000 × g and 4°C for 15 min (Eppendorf Company, Hamburg, Germany). The growth medium was decanted, and the harvested cells were re-suspended in the cryoprotective medium containing glycerol, sodium glutamate, and skim milk, as the protective agents, suspended in distilled water to obtain cell suspensions containing approximately 1.0 × 10¹⁰ CFU/mL. The samples were then freeze-dried (FreeZone 6L, Labconco, America).

When the TA of the yogurt samples reached 70°T, the samples were stirred to stop fermentation and stored in a refrigerator (4°C) for 24 h before testing. The viscosity of all the yogurt samples was then tested using a rotary viscometer (proRheo, R180, Germany) at a rotational speed of 64/s and an interval time of 10 s.

A sensory analysis was conducted to evaluate the sensory properties of the fermented milk products. Twenty trained panelists were invited and asked to score the consistency, stringiness, smooth, sweetness, and aroma of the samples using a 9-point intensity scale (1 = disliked extremely; 9 = liked extremely). The panelists were guided to taste the fermented samples and drink water between the samples to cleanse the palate. Scores were calculated as means of twenty trained panelists.

The rheological tests of the yogurt samples were performed using an ARES-G2 rheometer (TA Instruments, United States) with parallel titanium plate geometry (50 mm diameter, 1 mm gap). The exposed surfaces of the samples were covered with a thin layer of dimethylsiloxane to avoid dehydration. A thixotropic loop test was conducted by increasing the shear rate from 0.1/s to 100/s and then decreasing the shear rate at the same rate, and do this again. The storage modulus (G′) and loss modulus (G″) of the samples were measured by oscillation tests. A stress sweep test was performed to determine the range of linear viscoelasticity. Thereafter, frequency sweeps ranging from 0.1 to 100 rad/s were implemented within the linear viscoelastic region (γ = 0.5%) to obtain the relationships with G′ and G’’. All of the rheological measurements were performed in duplicate.

The acid- and EPS-producing abilities of 100 S. thermophilus strains were studied after 5–6 h of fermentation. The role of S. thermophilus in starter cultures is related primarily to its fast growth and acid-producing capability (Vendramin et al., 2017). Their lactic acid production ability was positively correlated with their growth ability (Fig. 1AB). According to the acidification rates of different strains in pasteurized fresh milk and UHT milk, all of the strains were divided into six major categories as shown in Figure 1C. Class 1 contained 27 strains with the best acid-producing capability. These strains could grow independently in pasteurized fresh milk, and their acidification rates were stable in different types of milk. Strains in Classes 2 and 3 could rapidly produce acid, but their acidification rates were depended on the milk sources. Strains from Classes 4–6 could not be used alone, but rather only as assistant strains to improve the flavor or texture of the fermented products. Zhao et al. (2021) evaluated the acidification capability of 85 strains of S. thermophilus based on the TA and the acid production rate. The TA achieved with and the acid production rate of strains carrying the PrtS gene were significantly higher than those of strains not carrying this gene, and these strains were clustered together in the phylogenetic tree.

Of all the tested strains, 34% could not produce enough EPS to improve the texture of the fermented milk as shown in Figure 1D. The texture of their fermented samples was rough, and the graininess and lumpiness of large protein aggregates was usually present in their fermented samples. The other 64% strains generated enough EPS to produce smooth and thick fermented samples in which no or very few small grains (1–2 mm) were observed. These results are in agreement with those reported by Nguyen et al., (2018) and Hassan et al., (2003), who observed reduced graininess in yogurt fermented with a high EPS-producing culture compared with low-EPS producing cultures (Hassan et al., 2003; Nguyen et al., 2018).

Before S. thermophilus strains can be used for milk fermentation in the industry, they need to be cultured in a suitable medium under optimal conditions to increase their growth to yield high cell density. After such culturing followed by centrifugation, bacterial suspensions containing large amounts of cells are used to produce directed vat-set (DVS) starter cultures with which to ferment milk.

Several types of strains are used to produce different DVS starter cultures (de Vin et al., 2005; Settachaimongkon et al., 2014; Karadeniz et al., 2021), so the composition of the culture medium needs to be changed according to the different strains used to prepare the DVS starter culture. In our study, we found that the carbon source was the primary limiting factor for preparing a universal culture medium for different S. thermophilus strains.

Carbon source utilization capacity of the strains tested in our study is shown in Figure 2. When using glucose as the carbon source, 36 strains could not grow, with the OD₆₀₀ being <0.1 after 6 h of fermentation by these strains at 37°C. Similarly, five strains could not grow when using sucrose as the carbon source, only seven strains showed a weak growth when using fructose as the carbon source, and none of the strains showed growth when galactose was used as the carbon source.

Principal component analysis (PCA) was performed using prcomp function for carbon source utilization in different strains, and the effect of carbon source on principal components was investigated (Figure 2A). Cluster analysis was performed using kmeans function to classify the utilization of different carbon sources by strains based on PCA analysis result (Figure 2B). All the strains can be divided into 3 clusters. The first cluster has 42 strains and they can grow well in glucose, sucrose and lactose. The second cluster has 44 strains and they grow weak in glucose. The third cluster has 14 strains they all grow weak both in glucose and sucrose. This revealed that lactose was the best carbon source for all S. thermophilic strains.

S. thermophilus is a homofermentative bacterium that ferments lactose via the Embden–Meyerhof pathway (EMP) to lactate. When S. thermophilus is cultured in milk, lactose is transported into the bacterial cell in association with the expulsion of galactose via an antiport system. Lactose is then hydrolyzed by bacterial β-galactosidase, but only glucose is metabolized further via the EMP to lactate. Overall, 1 mol of lactose is fermented to 2 mol of lactate plus 1 mol of galactose. Most of the S. thermophilus strains are galactose-negative, that is, they are unable to efficiently metabolize galactose. This metabolic defect is due, in part, to the production of insufficient levels of galactokinase, a key enzyme of the Leloir pathway encoded by galK (Vaillancourt et al., 2002; de Vin et al., 2005). In the fermentation course using lactose as the carbon source is not economical not only because lactose costs approximately three times more than glucose but also the galactose produced from lactose metabolism remains unused. As reported by Filip de Vin et al., 2005 only 8 of the 49 S. thermophilic strains they tested could completely utilize lactose, while the other 41 strains could process little or none of the secreted galactose.

Thus, a composite carbon source (0.5% lactose and 3.5% glucose) was used, instead of pure lactose, to prepare the universal culture medium in our study. We found that 0.5% lactose helped to begin fermentation by all glucose-non-fermenting strains, after which 3.5% glucose helped maintain the rapid growth of all strains. After optimizing the carbon source, all 100 strains of S. thermophilic were successfully cultured under the same fermentation conditions. This suggests that our optimized culture medium is conducive to large-scale fermentation using different S. thermophilic strains.

Forty strains were used for pH-controlled fermentation using 0.5% lactose and 3.5% glucose as the carbon source. After 5–7 h of fermentation, the OD₆₀₀ values of their fermented products all exceeded 10, indicating that high-density cultivation was achieved. The growth curves of five glucose-non-fermenting strains and five glucose-fermenting strains are shown in Figures 3A,B, respectively. After 2 h of fermentation, all of the 40 strains entered the logarithmic growth phase. Lactose promoted the rapid initiation of fermentation by the glucose-non-fermenting strains, after which these strains continued utilizing glucose similar to the glucose-fermenting strains.

Glucose fed-batch fermentation was also studied to increase the growth of S. thermophilus. The growth curve of strain ST20 under pH-controlled fermentation with glucose supplement was shown in Figure 4. When glucose was supplied during the course of fermentation, the OD₆₀₀ continued increasing until an OD₆₀₀ of 15.8 was reached after 12 h of fermentation, while the lactic acid content continued increasing until all of the supplemented glucose was consumed. The lactic acid in the system reached 60 g/L after 20 h of fermentation.

The cell cultures were collected by centrifugation of fermented samples obtained at 6, and 12 h and mixed with the same weight of cryoprotective solution separately. Each of these prepared bacterial suspensions was then freeze-dried and stored at −70°C. We found that the high levels of lactic acid in the fermentation medium affected the freeze-drying process of the products and even led to the failure of freeze-drying because the low Tg (−60°C) of lactic acid decreased the melting point of the suspension. Particularly, we found that the 9-h and 12-h suspensions had collapsed powder cakes stuck to the drying trays after freeze-drying. For culture broth samples containing more than 35 g/L of lactic acid, we recommend that the concentrated cell suspensions obtained after centrifugation be rinsed with buffer to remove lactic acid before freeze-drying.

The main roles of the starter culture used for yogurt production are acidification through the conversion of lactose to lactic acid, creation of viscous texture by denaturation of proteins and production of exopolysaccharides, and development of the typical yogurt flavor. After termination of fermentation, the yogurt is stored until consumed. During this period, the LAB may cause further acidification of the product, particularly when stored at temperatures above 4°C. This phenomenon is called post-acidification and is generally considered undesirable, as it may contribute to defects such as syneresis, reduction in viable LAB counts, accumulation of lactic acid in the product, and development of an undesirable flavor (Han et al., 2012; Wang et al., 2013; Mani-López et al., 2014; Akgun et al., 2018). Thus, starter cultures with a mixture of strains capable of rapid acidification, low post-acidification, and creating a viscous texture are favorable. Han et al. showed that the use of Cu²⁺ at the concentration of 1.25 mg/kg (CuSO₄ 3.15 mg/kg) could reduce post-acidification (Han et al., 2012), but the concentration of this supplemented CuSO₄ was too high compared with the DVS starter culture usually added at 30 mg/kg.

Class 1–3 strains (Figure 1C) had the ability to produce enough acid for fermentation. The post-acidification capacity of these strains was investigated by testing their fermented products stored at 10°C for 28 days after the TA reached 65–70°T. The strains ST40, ST33, ST42, and ST75 showed the highest post-acidification capability. Notably, although among all classes, Class 1 strains showed the highest post-acidification ability (Figure 5A), we found that the post-acidification ability was related to the strains’ acid fermentation rates that were used to classify the groups.

Next, we used combinations of different strains to improve the yogurt fermentation rate and texture and control post-acidification. Two strains ST53 and ST49 belonged to different classes that showed low post-acidification were selected to study their growth characteristics and interactions when used together with other strains for fermentation.

ST49 was found to be a high-EPS-producing strain but with low proteolytic capacity. When ST49 was co-cultured with other 13 EPS-producing strains at a ratio of 1:1, all of the pair-wise strain combinations except with ST20 increased the fermentation rate (Figure 5B). These results suggest that different S. thermophilus strains may promote or inhibit one another’s growth.

The acid production rate of ST53 was high, but there was no accumulation of EPS, so the texture of the product was rough when used ST53 alone for fermentation. Co-cultured with other 13 EPS-producing strains at a ratio of 1:1, we found that ST53 combinations with ST20, ST50, ST56, and ST49 produced smooth-textured yogurt after 6 h of fermentation, whereas its combinations with ST44, ST29, ST05, ST52, and ST57 significantly decreased the fermentation rate (Figure 5B).

Our mixed fermentation experiments showed that ST53, ST49, and ST50 promoted one another’s growth (Figure 6), and that post-acidification by the ST53 + ST50 and ST53 + ST49 co-cultures was lower than that by the ST49 + ST50 co-culture; thus, we further studied the characteristics of co-culture strategies using these strain combinations. The viscosity, post-acidification, and sensory evaluation results of the yogurt samples produced using various co-cultures were shown in Figure 7. Compound E yielded the highest viscosity, the second lowest post-acidification, and the best flavor. Thus, a DVS starter culture comprised of ST53 + ST49 + ST50 at a ratio of 1:2:2 was selected to explore its post-acidification capability and resulting yogurt quality.

The rheological properties of the fermented yogurt produced by Compound E was studied. The apparent viscosity decreased with the increase in the shearing rate, but the sample had good ability to withstand mechanical handling (Fig. 8AB). When the shearing rate increased or decreased from 0.1/s to 100/s, the viscosity of the stirred yogurt samples remained nearly the same from the second flow ramp. This indicated that the yogurt produced by Compounds E had the ability to withstand mechanical handling during dairy processing and transport. The thixotropic properties of fresh yogurt have been well established in previous studies (Sakin-Yilmazer et al., 2014; Xu et al., 2018). These could be quantified by the yogurt’s ability to regain its initial structure when allowed to rest for a certain period after disturbing.

The linear viscoelastic region γ < 1% (Figure 8C), γ = 0.5% was selected for the oscillation test. From 0.1 to 100 rad/s, the G′ of yogurt sample remained larger than G″, indicating that the structure of EPS and protein networks had higher elasticity than viscosity (Figure 8D). A higher elasticity of yogurt is known to reduce its stringiness.