Selection of a galactose-positive mutant strain of Streptococcus thermophilus and its optimized production as a high-vitality starter culture

Streptococcus thermophilus is a common starter in yo-gurt production and plays an important role in the dairy industry. In this study, a galactose-positive (Gal + ) mutant strain, IMAU20246Y, was produced using the chemical mutagen N-methyl-Nʹ-nitro-N-nitrosoguanidine (NTG) from wild-type S. thermophilus IMAU20246, which is known to have good fermentation characteristics. The sugar content of milk fermented by either the mutant or the wild type was determined using HPLC; metabolism of lactose and galactose was significantly increased in the mutant strain. In addition, we used response surface methodology to optimize components of the basic M17 medium for survival ratio of the mutant strain. Under these optimal conditions, the viable counts of mutant S. thermophilus IMAU20246Y reached 4.15 × 10 8 cfu/mL and, following freeze-drying in the medium, retained cell viability of up to 67.42%. These results are conducive to production of a high-vitality starter culture and development of “low sugar, high sweetness” dairy products.


INTRODUCTION
Streptococcus thermophilus is a thermophilic grampositive lactic acid bacterium (LAB) that is an important yogurt starter culture and is commonly mixed with Lactobacillus delbrueckii ssp.bulgaricus for yogurt production (Sieuwerts et al., 2010).Most strains of S. thermophilus, and most strains of all described L. delbrueckii ssp.bulgaricus, are unable to metabolize galactose, and this sugar is secreted into milk (Hols et al., 2005), resulting in dairy products containing large amounts of galactose as well as incompletely metabolized lactose.Excess amounts of galactose in dairy products may have adverse effects on product quality and human health, especially in patients with galactosemia (Novelli and Reichardt, 2000).Bley et al. (1985) reported that when non-galactose S. thermophilus was used as a starter culture to produce Cheddar cheese, a large amount of galactose accumulated, which aggravated the degree of cheese browning.Stang et al. (2005) reported that consumption of milk fat and galactose increased the likelihood of testicular cancer developing in adolescence.
Sugar metabolism in fermented milk mainly refers to LAB metabolizing lactose to produce lactic acid.Lactose is a disaccharide in milk, consisting of 1 molecule of glucose and 1 molecule of galactose condensed by a β-1,4 glycosidic linkage (Qi and Tester, 2020).Devos and Vanghan (1994) reported that LAB take up lactose in 2 ways.One way is via phosphoenolpyruvate-dependent phosphotransferase systems (PEP-PTS) followed by further metabolism by the tagatose-6-phosphate and Embden-Meyerhof-Parnas (EMP) pathways; this is the mechanism used by Lactobacillus rhamnosus and Lactococcus lactis but not S. thermophilus, which lacks this type of system (Neves et al., 2010;Wu and Shah, 2017).The second mechanism for lactose uptake by LAB is by intracellular transport via lactose permease, which is metabolized by β-galactosidase to produce glucose and galactose, and subsequently, 2 monosaccharides are metabolized by the EMP and Leloir pathways (de Vos and Vaughan, 1994).Both S. thermophilus and L. delbrueckii ssp.bulgaricus take up lactose into their cells using lactose permease and secrete 1 molecule of galactose for every molecule of lactose (Poolman, 2002).
After entering the Leloir pathway, galactose is converted into UDP-galactose and galactose-1-phosphatphosphorylated by galactokinase (GalK); using gal-1-phosphate uridylyltransferase (GalT) it is converted into glucose-1-phosphate and UDP-galactose.Galactose-1-phosphate is converted into glucose-6-phosphate under the catalysis of phosphoglucomutase enzyme (PGM).Thomas and Crow (1984) and Xiong et al. (2019) reported that galactose metabolism in S. thermophilus was limited due to insufficient production of key enzymes in the Leloir pathway and lack of a galactose transport system.
Given galactose metabolism deficiencies in S. thermophilus, and the effects of free galactose in milk on product quality and human health, genetic improvement of S. thermophilus is a priority (Konar and Datta, 2022).Mutation breeding provides an opportunity for selection of mutants that can metabolize galactose (Oladosu et al., 2016).In particular, N-methyl-Nʹ-nitro-N-nitrosoguanidine (NTG) is considered useful in mutation breeding and has been used as a common mutagen to induce mutants that meet the needs of industrial production; NTG induces a relatively wide spectrum of mutations by alkylating purines and pyrimidines (Adelberg et al., 1965;Ohnishi et al., 2008).Thomas et al. (2014) successfully applied NTG mutagens to screen urease-defective S. thermophilus mutants for high lactic acid-yield mutants.Miyamoto et al. (1983) used NTGtreated Lactobacillus casei ssp.alactosus and found that mutants produced more diacetyl than parental strains, which has great potential in dairy applications.
High cell-density cultivation and low-temperature vacuum drying are key technologies for production of high-vitality starter cultures (Bauer et al., 2012).These techniques make it possible to produce starter culture powders on a large scale.Freeze-drying makes it easier to store and transport biological samples and reduces the risk of phage infection, making it suitable for production of concentrated starter cultures (Kanmani et al., 2011).Hou et al. (2016) used high cell-density cultivation and low-temperature vacuum drying to produce a highvitality starter culture of Lactiplantibacillus plantarum NDC 75017.However, during freeze-drying, the freezing process led to formation of ice crystals and dehydration caused changes in solute concentration, which caused irreversible damage and decreased the survival rate of freeze-dried LAB (Vinderola et al., 2012).Therefore, improving the survival rate of microorganisms during freeze-drying has become a focus of research in recent years.In industrial production, survival rate during freeze-drying is generally improved by optimization of strain culture conditions, freeze-drying protection types, and freeze-drying process parameters (Peiren et al., 2015).Schubiger (2015) significantly improved cell density and increased survival of microorganisms after freeze-drying by optimizing the complex ratio of glucose, yeast extract, and plant peptone in the medium.The response surface methodology (RSM) and Box-Behnken design (BBD) are effective ways to optimize complex processes, which means they are widely used to improve product yield and reduce development time costs (Ren et al., 2008).These statistical techniques have been widely used in many fields, including optimization of enzyme catalysis conditions, food processing conditions, and medium conditions (Wang and Lu, 2004;Gültekin-Özgüven et al., 2016).Li et al. (2019) verified the effectiveness of parameter optimization based on BBD-RSM by establishing a quadratic equation model to determine optimal culture conditions for Escherichia coli.
The aim of the present study was to use NTG to obtain a stable galactose-positive mutant of a strain of S. thermophilus known to have good fermentation properties (IMAU20246); the metabolic pathway of the mutant would enable preferential utilization of galactose and excretion of glucose to increase yogurt sweetness without adding exogenous sugars.Therefore RSM was used to optimize the basic medium in which the mutant was preserved, and vacuum freeze-drying was used to prepare a high-vitality bacteria powder, which provided a starter culture for industrial production.

Isolates and Reagents
Streptococcus thermophilus IMAU20246 was originally isolated from traditionally fermented milk collected in Kent Province, Mongolia, and was used throughout this study (Ren et al., 2015).Galactose medium M17 (1%) was prepared using galactose as the only carbon source at 1% (wt/vol) instead of lactose as in M17.Lactose, galactose, glucose, and lactic acid standard products used were of chromatographic grade, purity ≥98%.Phosphatebuffered saline was made up of 0.8% NaCl, 0.02% KCl, 0.02% KH 2 PO 4 , 0.115% Na 2 HPO 4 ; pH 7.4.

Treatment With NTG and Isolation of Mutants
The procedure used to treat bacteria with NTG (Shanghai Yien Chemical Technology, China) was a modification of that reported by Benateya et al. (1991).Briefly, frozen S. thermophilus IMAU20246 was activated by at least 3 routine cultivations in M17 broth (QuingDao HopeBiol, China) for 24 h at 42°C.Subsequently, the volume of the medium was increased to 500 mL of M17 broth for 24 h and then centrifuged for 5 min at 5,000 × g (4°C).The cell pellet was washed twice in PBS and resuspended in the same buffer solution so that the final cell concentration corresponded to 0.1 g/mL (dry weight).Then NTG was added to the cell suspension at final concentrations of 0, 100, 200, 300, 400, and 500 μg/ mL, and the contents were carefully mixed.An NTG-free sample served as a control.The NTG was applied to the oscillating cell suspension for 1 h and kept in darkness Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE at 42°C.Cell suspensions were centrifuged for 20 min (6,000 × g, 4°C), and the cells were washed twice with PBS.The washed cells were suspended in 5 mL of M17 broth and incubated for 2 h at 42°C.Mortality rate was calculated according to the formula in Equation [1].The cell suspension showing a mortality rate of 50% (Adelberg et al., 1965) compared with the control was used to select for mutants with growth activity on galactose.Single mutant colonies were isolated on M17 agar containing 1% (wt/vol) galactose as the sole carbon source at 42°C for 48 h: [1]

Enzymatic Activity Assay
Wild-type S. thermophilus IMAU20246 and its mutants were cultured in M17 broth at 42°C for 24 h, centrifuged at 15,000 × g (4°C, 10 min), and then washed twice with PBS.After centrifugation, the supernatant was retained and placed on ice for testing.β-Galactosidase, glucokinase, pyruvate kinase, and galactokinase activities were determined using activity assay kits (Beijing Solarbio Technology Co., China) and the instructions provided in the kits for specific experimental methods.

Growth Curve
Streptococcus thermophilus IMAU20246 and its mutant were inoculated in M17 broth with lactose and galactose as single carbon sources at 2% and cultured at 42°C for 24 h.Optical density (OD) values were measured at 600 nm (OD 600 ) for 24 h to draw the growth curves.

Metabolic Profiling
Fermented milk was prepared using the method described previously by Dan et al. (2022).Streptococcus thermophilus IMAU20246 and its mutant were inoculated into whole milk (New Zealand Heng Nature Co. Ltd.) at 5 × 10 6 cfu/mL and fermented at 42°C.Samples were taken every 2 h during fermentation and stored at −20°C.Fermentation terminated when the pH reached 4.5 (fermentation end point), and samples were then stored at 4°C for 0, 1, 3, 7, 14 d.The quantities of lactose, galactose, glucose, and lactic acid (purchased from Shanghai Yuanye Biotechnology, China) in fermented milk were determined via HPLC (Agilent 1260 series, Agilent Technologies).
Determination of Lactose, Galactose, and Glucose Contents in Fermented Milk.Accurately weighed samples (1 g) of well-mixed fermented milk were each placed in a 25-mL flask; 20 mL of ultrapure water was added and shaken for 30 s; the volume was fixed; sonication was performed for 30 min; centrifugation was performed (8,000 × g, 10 min at 4°C); and the supernatant was filtered through a 0.45-μm microporous membrane.Standard solutions were configured with the concentrations of 5, 2.5, 1.25, 0.625, and 0.3125 mg/mL of lactose, galactose, and glucose.The chromatographic separation column for HPLC was an Aglient amino column (5 µm, 250 mm × 4.6 mm) for 35°C with a differential refractive index detector (1260 RID, Agilent Technologies).The mobile phase was a mixture of acetonitrile and ultrapure water (70:30) at a flow rate of 1 mL/min and an injection volume of 10 μL.
Determination of Lactic Acid Content of Fermented Milk.Fermented milk samples (0.5 g) were placed in 10-mL volumetric flasks to which 8 mL of H 2 SO 4 (0.1 mol/L) was added, and the volume was fixed after shaking for 30 s.A supernatant was obtained after centrifugation (8,000 × g, 10 min at 4°C) and filtration through 0.45-μm microporous membranes.The HPLC detector was an UV detector with a Zorbax SB-Aq column (5 μm, 4.6 mm × 150 mm; Agilent Technologies) at 35°C, and the detection wavelength was 210 nm.The mobile phase consisted of methanol and 0.01 mol/L phosphate buffer (pH = 2.0); the volume ratio was 3:97, and the eluent flow was 0.5 mL/min.The sample size was 10 μL.

Genetic Stability
Stability of the mutant S. thermophilus IMAU20246Y was determined using the previously described method of Sørensen et al. (2016) with some modifications.Streptococcus thermophilus IMAU20246Y was inoculated into M17 broth (2% [wt/vol]) incubated for 24 h at 42°C, which was the first-generation strain.After that, every 24 h a 2% transfer was made for 10 continuous subcultures at 42°C.The enzyme activities of β-galactosidase, galactokinase, glucokinase, galactokinase, and pyruvate kinase were determined.

Single-Factor Optimization
Carbon and Nitrogen Sources.Lactose in the M17 basal medium (Supplemental Table S1, see Notes) was replaced with equal amounts of either galactose, fructose, maltose, or glucose as the sole carbon source.To this 2% of bacterial inoculum was added to the medium containing each carbon source, incubated at 42°C for 24 h, and then the OD 600 measured as the response value indicative of bacterial cell density.Each experiment had 3 replicates (n = 3).
Five nitrogen sources (soy peptone, bacteriological peptone, casein, yeast powder, beef paste) from M17 basal medium were each provided as a single nitrogen source; 2% of the activated inoculum was inoculated into the different nitrogen sources' medium and incubated at 42°C for 24 h, and the OD 600 was measured as the response value indicative of bacterial cell density.Multiple nitrogen sources identified from this initial screening were then evaluated in different ratios but with a constant total amount; as before, they were each inoculated with the activated culture.
Based on the results of carbon and nitrogen source evaluation, the total amounts and proportions of carbon and nitrogen sources in the medium were manipulated to determine the optimal total amount and proportion of carbon and nitrogen.
Trace Elements.The M17 medium was supplemented with either ZnSO 4 and inoculated with the culture at 2%, incubated at 42°C for 24 h, and the OD 600 measured (as the response value indicative of bacterial cell density) and compared with medium without trace elements as the control group.
Growth Factors.We evaluated growth in basal M17 medium supplemented with either arginine, proline, leucine, valine, methionine, glutamic, histidine, lysine, aspartic, alanine, glycine, glutamine, serine, cysteine; vitamin B 1 , vitamin B 2 , vitamin B 6 , and vitamin C; or l-ascorbate sodium, as in the previous single-factor test, and compared it against growth in medium without these additional factors as the control group.As before, bacterial cultures were inoculated at 2% and incubated for 24 h before the OD 600 was measured as the response value indicative of bacterial cell density.pH.To investigate the effect of pH on growth of the mutant, the pH of the M17 medium was manipulated to 6, 6.5, 7, 7.5, or 8, respectively, for each trial.Inoculum was added at a concentration of 2% and incubated for 24 h at 42°C.
Inoculum Concentration.The effect of inoculation concentration on growth of the mutant was studied by inoculating different volumes of inoculum into 5 mL of medium (representing 1%, 2%, 3%, 4%, and 5% of total medium volume).Growth was estimated by evaluation of final cell concentration after incubation at 42°C for 24 h.

Evaluation Using the Plackett-Burman Design
According to the results of single-factor experiments, 6 independent variables (X n ) for galactose content (X 1 ), nitrogen source content (X 2 ), trace element content (X 3 ), growth factor content (X 4 ), inoculum volume (X 5 ), and pH (X 6 ) were investigated via Plackett-Burman (PB) design.Two levels (concentrations) of each variable, high and low, were used and were designated as +1 and −1, respectively (Supplemental Table S2, see Notes).The value of the OD 600 taken after growth was used as the response value indicative of bacterial cell density.

BBD Experiment
Based on medium components that had a significant effect on bacterial density in the PB design experiment, a Box-Behnken model was designed and applied to the independent variables (Supplemental Table S3, see Notes).The OD 600 taken after growth was used as the response value indicative of cell density.An RSM experimental design was generated in Design Expert 7.0 (StatEase) software to develop a quadratic model suitable for fitting the experimental data and plotting the response in 3 dimensions.

Preparation of a Bacterial Powder
The method we used followed Yao et al. (2022) with some improvements.Activated cultures were inoculated into optimized medium at a rate of 3% and incubated for 14 to 16 h at 42°C and then centrifuged (4,000 × g, 5 min) at 4°C; bacterial precipitates were collected and mixed in a 25% concentration of freeze-drying protectant (skimmed milk 60 g/kg, seaweed sugar 60 g/kg, yeast Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE powder 40 g/kg).The bacterial precipitates with freezedrying protectant were pre-frozen at −80°C for 22 h, and then placed into a freeze dryer (Xinzhi Biotechnology Co. Ltd., Shanghai, China) for freeze-drying (cold trap temperature, −53°C; vacuum, 6.93 Pa; duration, 24 h).
A gradient dilution plate colony counting method was used to determine the survival rate of each strain before and after freeze-drying.The lyophilization survival rate was calculated using the following formula: where A 1 represents the number of viable bacteria in the sample after lyophilization (cfu/mL) and A 2 represents the number of viable bacteria in the sample (cfu/mL) before lyophilization.

Statistical Analysis
All experiments were performed in triplicate (n = 3), and the mean data ± SD were reported.Data were analyzed in Excel (Microsoft Corp., Redmond, WA), SPSS 19.0 (IBM Corp., Chicago, IL) and Origin 7.5 (Origin-Lab, Northampton, MA).One-way ANOVA was used to determine differences between mean values with a significance level of P < 0.05.

Survival of Strains After NTG Treatment and Selection of Gal + Strains
Mortality of S. thermophilus IMAU20246 cell suspensions was evaluated in different concentrations of NTG after shaking for 60 min in darkness at 42°C (Table 1).Mortality increased with increasing NTG concentration; when the NTG concentration was 300 μg/mL, viable cell counts decreased from 3.06 × 10 9 cfu/mL (control) to 1.50 × 10 9 cfu/mL, representing a mortality rate close to 50%.Fifteen potential galactose-positive (Gal + ) mutant strains were initially isolated from the cells treated with 300 μg/mL NTG.These potential mutants were tentatively named IMAU20246-1 to IMAU20246-15.

Enzymatic Activity Assay
In the process of milk fermentation, the activity of β-galactosidase directly affects metabolism of lactose (Liu et al., 2019).Although glucokinase catalyzes the initial metabolism of glucose, pyruvate kinase enables phosphoenolpyruvate and ADP to produce pyruvate and ATP (Zhang et al., 2021).Considering the role of the 3 enzymes, the mutant that had higher β-galactosidase activity, but weaker glucokinase and pyruvate kinase activities than the wild-type was selected as the experimental strain for subsequent experiments.Therefore, to screen galactose-metabolizing S. thermophilus mutants, we measured the activities of pyruvate kinase, β-galactosidase, and glucokinase in the wild-type and its mutants (Figure 1).The mutant IMAU20246-8 had a high β-galactosidase activity (0.09 nmol/h per 10 4 cells) and galactokinase activity (11.42 IU/L), whereas the wild-type IMAU20246 was weak.In addition, glucokinase (2.6 mU/10 4 cells) and pyruvate kinase (2.1 mU/10 4 cells) had weak activities in the mutant, but they were significantly higher in the wild type.Therefore, S. thermophilus mutant IMAU20246-8 was selected as the experimental strain for subsequent experiments and named S. thermophilus IMAU20246Y.

Growth of S. thermophilus IMAU20246 and Its Mutant
The growth of S. thermophilus IMAU20246 and its mutant on M17 broth containing lactose and galactose was determined by OD 600 (Figure 2).Lactose was the best energy source for IMAU20246, whereas for IMAU20246Y it was galactose.This may be because S. thermophilus IMAU20246Y is a Gal + phenotypic mutant and therefore has a certain preference for galactose during growth.In addition, when cultured in galactose medium, IMAU20246Y showed a lag period of about 4 h compared with IMAU20246 in lactose medium.The final cell density obtained with the mutant grown on galactose (0.91) was much higher than that measured on lactose (0.49).

Amino Acid Sequence Alignment
The activities of β-galactosidase, glucokinase, galactokinase, and pyruvate kinase in the mutant S. thermophilus IMAU20246Y were significantly different to the wild type, and it was believed that the binding of NTG and DNA had caused mutations in the amino acid sequence encoding these enzymes.Therefore, PCR amplification and sequencing of the amino acids that encode β-galactosidase (LacZ), glucokinase (glck), galactokinase (galK), and pyruvate kinase (pyk) in S. thermophilus IMAU20246 and its mutant were undertaken.Bands of PCR amplification products were single and clear, indicating good amplification (Figure 3).Sequences of the mutant strain showed that the LacZ and glck amino acid sequences had different degrees of genetic code deletion and substitu-tion phenomena (Figure 4).Similarities in lacZ and glck sequences between S. thermophilus IMAU20246 and its mutant were 40.76% and 20.26%, respectively.

Lactose, Galactose, Glucose, and Lactic Acid Contents of Fermented Milk
To further verify the ability of the mutant to metabolize galactose, the quantities of lactose, galactose, glucose, and lactic acid in the medium during milk fermentation by S. thermophilus IMAU20246 and its mutant S. thermophilus IMAU20246Y, and following storage, were determined via HPLC (Table 2).At 0 h the lactose contents of milk fermented by the wild type and the mutant were 4.43 and 4.36 g/100 g, respectively, and gradually metabolized as fermentation progressed.After 14 d of storage, the lactose contents of the wild-type and the mutant media were significantly different to each other (P < 0.05) and had reduced to their lowest values of 2.95 and 2.57 g/100 g, respectively.During fermentation and storage of milk, S. thermophilus IMAU20246 and its mutant IMAU20246Y metabolized 33% and 41% of lactose, respectively, and the lactose content of the fermented milk of the mutant was always lower than that of S. thermophilus IMAU20246, indicating that the ability of the mutant to metabolize lactose was significantly higher than that of the wild type (P < 0.05).
During fermentation (0, 3, and 6 h), the galactose contents of milk fermented by S. thermophilus IMAU20246 and its mutant S. thermophilus IMAU20246Y were not significantly different (P > 0.05).In contrast, the ga-lactose content of milk fermented by S. thermophilus IMAU20246Y was significantly lower than that of milk fermented by the wild type during storage (P < 0.05).At 14 d of storage, milk fermented by the wild type and the mutant had 0.6 and 0.41 g/100 g galactose content, respectively.The ability of S. thermophilus IMAU20246Y to metabolize galactose was significantly greater than the wild type (P < 0.05).
Trace amounts of glucose (0.09 g/100 g) were detected in milk fermented by S. thermophilus IMAU20246 at 0 d of storage, with no glucose detected at any other time points, which may be due to the rapid metabolism of glucose through EMP during milk fermentation, such that the low levels achieved failed to reach the detection limit (<0.02 g/100 g).In contrast, glucose in milk fermented by the mutant S. thermophilus IMAU20246Y began to accumulate at 3 h and reached a peak of 0.57 g/100 g at 6 h.It was slightly metabolized during storage, but levels were not significantly different (P > 0.05).These results suggest that the metabolic pathway of S. thermophilus may have changed to some extent after chemical mutagenesis.
Lactic acid has an antibacterial effect in fermented milk and is also a natural flavor substance (Teshome, 2015).Lactic acid was detected in milk fermented by the wild type and the mutant at 3 h, specifically 1.15 and 0.37 g/100 g, respectively.Lactic acid content gradually increased as lactose in fermented milk was gradually metabolized.After 14 d of storage, the lactic acid content of milk fermented by the mutant was 7.76 g/100 g.This result indicated that the acid production capacity of the mutant was significantly greater than that of the wild type (P < 0.05), which was consistent with the results of lactose metabolism in milk fermented by the wild type and its mutant.

Stability
Desirable characteristics of strains obtained via mutagenesis are prone to degradation and loss during passage and culture (Filannino et al., 2016).During fermentation, key influences on sugar metabolism in LAB are critical enzymes in the metabolic pathway (Iskandar et al., 2019).Therefore, we measured the activity of key enzymes in the metabolic pathway, β-galactosidase, glucokinase, and pyruvate kinase, over 10 continuous generations to evaluate genetic stability of the mutant (Table 3).The β-galactosidase activity of S. thermophilus IMAU20246Y remained in the range of 0.080 to 0.088 nmol/h per 10 4 cells for 10 successive generations; galactose activity was between 8.32 and 8.46 IU/L; and the activities of glucokinase and pyruvate kinase ranged from 2.30 to 2.60 and 1.92 to 2.26 mU/10 4 cell.No significant difference in enzyme activities occurred during passage over 10 generations (P > 0.05).These results indicate that β-galactosidase, glucokinase, and pyruvate kinase activities of the mutant, S. thermophilus IMAU20246Y, remained stable after 10 successive generations; that is, they showed good genetic stability.

Screening Individual Culture Medium Components: Single-Factor Optimization
Effect of Carbon Source on Growth.Carbon is not only an essential element for microbial structure; it also provides energy for growth and other cellular processes (Wang et al., 2019).We evaluated the effects of different carbon sources on growth of the mutant (Figure 5a).Results showed that the growth rate of the mutant was highest in the galactose medium, significantly higher than in media with other carbon sources (P < 0.05).Galactose was selected as the carbon source for subsequent evaluations.
A suitable concentration of carbon can meet the higher nutritional requirements of a strain (Osma et al., 2011).Therefore, different galactose concentration gradients were set to evaluate effects on growth of the mutant (Figure 5b).Greatest growth, as indicated by the highest ΔOD 600 (0.69) for mutant S. thermophilus IMAU20246Y, was obtained at a galactose concentration of 1%, which was significantly greater than at other concentrations   (P < 0.05); growth was negatively affected when this concentration was exceeded.
To verify these experimental results, we enumerated the viable cells of the mutant when the concentrations of galactose in the medium were 0.5%, 1%, and 1.5% (Supplemental Table S4, see Notes).The highest viable cell count (3.52 × 10 7 cfu/mL) was obtained when the concentration of galactose in the medium was 1%, which was consistent with the results of mutant bacterial density.
Effect of Nitrogen Source on Growth.Nitrogen sources are essential nutrients for microbial synthesis of amino acids, DNA, RNA, ATP, and other biomolecules (Durán-Sequeda et al., 2021).The effects of 5 nitrogen sources individually on mutant growth were evaluated (Figure 5c).When soy peptone and yeast powder were the single nitrogen sources, the ΔOD 600 of mutant IMAU20246Y was 0.69; when casein peptone was the single nitrogen source, the density of the mutant was the lowest (0.1).
Growth was compared when soy peptone and yeast powder were provided together as nitrogen sources in ratios of 1:1, 1:2, 1:3, 2:1, and 3:1 (Figure 5d).When the ratio of soy peptone and yeast powder was 1:1, the ΔOD 600 , as indicative of cell density, was 0.85, and the density of the mutant was significantly higher than for the other experimental groups (P < 0.05).To verify these results, cell suspensions from the 5 experimental groups were gradient diluted and the number of cells counted (Supplemental Table S5, see Notes); results confirmed that the highest number of viable cells (5.8 × 10 7 cfu/mL) was achieved with a ratio of soybean peptone to yeast powder of 1:1.
Effect of Total Amount and Proportion of Carbon to Nitrogen on Growth.Based on the above experimental results, the total amount and proportion of carbon to nitrogen were evaluated and optimized (Figure 5e).The density of the mutant strain increased gradually with increasing total carbon and nitrogen, reaching its highest value at 4%; the largest bacterial density was achieved when the ratio of carbon to nitrogen was 1:3.To further evaluate the optimal total amount of carbon and nitrogen, viable cell counts were performed (Supplemental Table S6, see Notes).When the total amount of carbon and nitrogen in the medium was 4% and the ratio of carbon to nitrogen was 1:3, the viable cell count of S. thermophilus IMAU20246Y reached 8.23 × 10 7 cfu/mL, which was significantly greater than in all other treatments (P < 0.05).

Effect of Trace Elements on Growth.
Trace elements from inorganic salts play an important role in the growth and metabolism of microorganisms (Mutanda et al., 2023).Among the elements evaluated, FeSO 4 •7H 2 O inhibited growth of S. thermophilus IMAU20246Y, and the bacterial cell density was always lower than that of the  verify the accuracy of the results (Supplemental Table S7, see Notes).When concentrations of MnSO 4 •5H 2 O and MgSO 4 •7H 2 O were 0.20 and 1.20 g/L, the viable cell counts of the strain reached 9.01 × 10 7 and 8.73 × 10 7 cfu/ mL, respectively, and were significantly higher than that of the control group (P < 0.05).Therefore, these 2 trace elements were selected as components of the growth medium of S. thermophilus IMAU20246Y in the subsequent experiment.
Effect of Growth Factors on Growth.Most vitamins appeared to inhibit bacterial growth, and the degree of growth inhibition increased gradually with increasing concentration (Figure 5h and 5i).Among these, vitamin B 6 promoted growth of S. thermophilus IMAU20246Y; the maximum density reached 1.12 at a concentration of 0.5 g/L.Viable cell counts were enumerated at the optimum concentrations of the above growth factors (Supplemental Table S8, see Notes).Results showed that when histidine at a concentration of 0.5 g/L was added to the medium, the number of viable bacteria increased significantly (P < 0.05) to 1.15 × 10 8 cfu/mL.Therefore, 0.5 g/L histidine was the selected growth factor in the medium of S. thermophilus IMAU20246Y.

Effect of PH and Inoculum Volume on Growth.
Using the above optimized medium components, the optimal initial pH for the growth of S. thermophilus IMAU20246Y was determined.As the initial pH of the medium increased, the ΔOD 600 , measured as indicative of growth, reached a maximum when the pH reached 7.5 (P < 0.05; Figure 5j).Subsequently, viable cell counts were made on samples (Supplemental Table S9, see Notes).At a pH of 7.0, S. thermophilus IMAU20246Y had the best growth, and the viable cell count reached 2.04 × 10 8 cfu/mL.Therefore, pH 7.0 was selected as the initial pH of the medium.
In addition, we continued to optimize inoculation concentration of the strain (Figure 5k).Results showed that the ΔOD 600 was highest when the inoculation rate of S. thermophilus IMAU20246Y was 2% (wt/vol).To verify results, viable cell counts were conducted on samples (Supplemental Table S10, see Notes).When the inoculum rate was 3%, the maximum number of viable bacteria reached 3.02 × 10 8 cfu/mL, which was significantly greater than that of the other 2 groups (P < 0.05), so the optimal inoculum rate of mutant S. thermophilus IMAU20246Y selected was 3%.

PB Design to Identify Key Medium Factors Affecting Strain Growth
After single-factor selection of medium components, PB was used to determine the most influential factors affecting growth of the mutant strain.Six factors, namely (X 1 ) galactose content, (X 2 ) nitrogen source content, (X 3 ) trace element content, (X 4 ) histidine content, (X 5 ) inoculum volume, and (X 6 ) initial pH, were integrated into a PB matrix that included a set of 12 experimental runs (Supplemental Table S11, see Notes).The amount of growth achieved was greatest in run 9. Analysis of variance showed that the model was significant (P = 0.0036); the model F-value was 16.72, also indicating that the model was significant (Table 4).The roles of galactose content and nitrogen source content were extremely significant (P < 0.01); pH was significant (0.01 < P < 0.05), but trace element content, histidine content, and inoculum rate were not significant (P > 0.05).Therefore, 3 factors, galactose content, nitrogen source content, and pH, were selected for the subsequent BBD test.

Box-Behnken Design
Based on the results of PB, the 3 most important factors with a significant effect on mutant strain growth were further optimized by RSM using BBD.Taking the result of the PB test with galactose content (X 1 ), nitrogen source content (X 2 ), and pH (X 3 ) as the center point, a 3-factor 3-level BBD was adopted (Supplemental Table S12, see Notes) and used to generate a BBD matrix with 17 runs.The highest bacterial density (1.482 ± 0.03) was obtained at run 13.Regression fitting was performed on the data to obtain the regression equation of the secondorder model: where X 1 , X 2 , and X 3 are coded values for galactose content, nitrogen source content, and pH, respectively.We found no significant difference in the lack of fit test (P 1 X 1 -X 6 = independent variables tested in single-factor experiments: X 1 = galactose content, X 2 = nitrogen source content, X 3 = trace element content, X 4 = histidine content, X 5 = inoculum volume, and X 6 = pH.R 2 = 0.9525; adjusted R 2 = 0.8956.**P < 0.01, extremely significant difference; *0.01 < P < 0.05, significant difference; P > 0.05, not significant. > 0.05), indicating that the quadratic model had a good degree of fit to the data (Table 5).The R2 was 0.9888, indicating practically perfect agreement between the predicted and experimental values, which can be used to predict growth and proliferation of S. thermophilus IMAU20246Y.
The model primary terms X 1 and X 2 had a highly significant effect on bacterial density (P < 0.0001); the effect of primary term X 3 on bacterial density reached a significant level (P < 0.05).The X 1 X 2 and X 2 X 3 interactions on bacterial density were not significant (P > 0.05), whereas effects of X 1 X 3 and the secondary terms X 1 2 , X 2 2 , and X 3 2 on the bacterial density interaction were significant (P < 0.05).The relationship between the factors was complex, showing a surface relationship; relationships between the 3 factors on bacterial density were even more complex.In summary, the order of significance of the influence of these 3 factors on the amount of strain growth was as follows: galactose content (X 1 ) > nitrogen source content (X 2 ) > pH (X 3 ).

Effects of Varying Medium Factors
We obtained 2-dimensional contours and 3-dimensional stereograms through BBD (Figure 6).Galactose quantity and nitrogen source significantly affected the cell density of mutant S. thermophilus IMAU20246Y.Density of the mutant increased with increasing total content but decreased significantly when the total content increased beyond a certain level.In addition, the plot of the interaction between carbon and nitrogen sources had obvious curvature, which proved that carbon and nitrogen sources had a significant interaction effect on the density of S. thermophilus IMAU20246Y.The low curvature of the plot of the interaction between nitrogen source and pH indicated that this interaction was not significant.The optimal combination was obtained by RSM-BBD as follows: the contents of galactose, nitrogen source, and pH in the medium were 12.45 g/L, 38.88 g/L, and pH 7.04, respectively, and the bacterial density (OD) predicted by the model was 1.488.To verify the reliability of the results, S. thermophilus IMAU20246Y was cultured under the conditions obtained, and the measured bacterial density (OD) was 1.588, which was not significantly different from the predicted value.

Survival Ratio of Viable Cells
The vacuum freeze-drying method for the preparation of LAB powder was based on the principle that LAB slow down cell metabolism in response to lack of water after drying, which promotes dormancy and maintains its original strain characteristics (Ma et al., 2018).The mutant strain S. thermophilus IMAU20246Y was cultured under the optimum culture conditions determined according to the effects of varying medium factors.A protective agent was added to the bacteria and a bacterial powder prepared after vacuum freeze-drying.The survival ratio after freeze-drying was measured by the dilution plate pouring method.Before and after S. thermophilus IMAU20246Y was freeze-dried, the viable cell counts were 1.78 × 10 9 and 1.20 × 10 9 cfu/mL, respectively, and the survival ratio of viable cells was 67.42%.In contrast, after freezedrying the S. thermophilus IMAU20246 viable cell count was 1.87 × 10 7 cfu/mL, which was significantly lower than for the mutant.

DISCUSSION
During evolution of S. thermophilus, many genes related to carbon source transport and metabolism degrade into pseudogenes as an adaptation to the single carbon source provided by milk (McDonnell et al., 2016).Therefore, the available carbon sources for S. thermophilus production are limited mainly to sucrose, lactose, galac-  tose, and glucose.However, most S. thermophilus strains currently used in the dairy industry can only metabolize glucose and cannot directly metabolize galactose, resulting in large extracellular accumulations of free galactose (Poolman, 2002;Sørensen et al., 2016).Galactose that accumulates in dairy products affects the quality of the product; for example, excessive galactose content leads to cheese browning during heating and the production of carbon dioxide via heterolactic fermentation, which results in cheese cracks, fractures, and other organizational defects (Michel and Martley, 2001).Therefore, improving the utilization efficiency of galactose and developing yogurt with low galactose content has market application value (Sørensen et al., 2016).In this study, we used NTG to develop a galactose-metabolizing mutant strain of S. thermophilus (IMAU20246Y), which was able to metabolize 31.67% of galactose compared with the wildtype strain.
Streptococcus thermophilus can obtain lactose operons (LacSZ) by horizontal gene transfer (Settachaimongkon et al., 2014).These operons enable lactose to be transported into the cell in a non-phosphorylated form, as by lactose permease (LacS), and then hydrolyzed into glucose and galactose by β-galactosidase (LacZ), which is essential for the hydrolysis of lactose.Ibrahim and O'Sullivan (2000) treated S. thermophilus with NTG and ethyl methylsulfone and found that β-galactosidase activity of the resulting mutant increased significantly, indicating that strains with higher β-galactosidase activity could be selected for using NTG mutagenesis.In this study, β-galactosidase activity was weak in the wild-type strain (Figure 1) but higher in the mutant, indicating that the mutant was better than the wild-type strain at hydrolyzing lactose.Glucokinase and pyruvate kinase are rate-limiting enzymes during glycolysis; decreasing enzyme activity inhibits metabolism of glycolysis and enables glucose accumulation in fermented milk.Lower glucokinase pyruvate kinase activity was detected in the mutant S. thermophilus IMAU20246Y, indicating rates of glycolysis were lower than in the wild type.
Metabolic modification does not necessarily require the use of recombinant DNA techniques; mutagenesis breeding methods can be used to control the direction of strain mutagenesis artificially, thereby significantly increasing the mutation rate (Lee et al., 2014).Adelberg et al. (1965) found that NTG was the most effective chemical mutagen and effective in LAB species.Burrow et al. (1970) used NTG to develop mutagenic strains of Streptococcus diacetylactis that produced more acetoin and diacetyl than wild-type strains.Benateya et al. (1991) developed a galactose-positive mutant of S. thermophilus after treatment with NTG that not only retained the original biological characteristics of the wildtype strain but also metabolized 70% more galactose than the parental strain.The use of NTG to change genetic control of the metabolic pathways of strains is effective in production of strains more suited to industrial production (Konar and Datta, 2022).In this study, genetic code deletion and substitution phenomena were observed in both β-galactosidase (LacZ) amino acid sequences (Figure 4).This resulted in selected mutants with lactose and galactose metabolism abilities superior to those of wild-type strains (Table 2).Based on NTG mutation of S. thermophilus IMAU20246, we were able to significantly (P < 0.05) change the metabolites secreted into the growth medium by the resulting mutant.Although the galactose content of the mutant strain tended to accumulate during fermentation, S. thermophilus was always in a viable state when inoculated into milk and could still be metabolized despite being stored at low temperature.Our results are similar to those of Benateya et al. (1991).During fermentation of milk, the mutant, IMAU20246Y, was able to consume more lactose, metabolize some galactose, and secrete the remaining galactose and most of the glucose back into the milk.This may benefit yogurts with low lactose content and enhance natural sweetness.
To better compare the sweetness of fermented milk by the S. thermophilus IMAU20246 and its mutant IMAU20246Y, sensory evaluation was performed at 0 d of storage.The result showed that the yogurt produced with IMAU20246Y had a slightly sweet flavor and good taste (result not shown).Each individual has a different threshold of sweetness perception.The most acute will be around 0.2 g/100 g sugar by volume, but most people begin to perceive sweetness at around 1 g/100 g (Mao et al., 2019).The recognition threshold for glucose ranged from 2.7 to 148.3 mM (0.4 g to 26 g; Low et al., 2017).In this study, the glucose content of fermented milk with the wild type was 0.09 g/100 g at 0 d of storage, whereas the mutant was significantly higher (P < 0.05), reaching 0.35 g/100 g (Table 2), which was close to the glucose recognition threshold, consistent with the result of the sensory evaluation.Similar results were also found in Sørensen et al. (2016).During fermentation, glucose and galactose are produced by lactose; glucose is 4 times sweeter than lactose, and galactose is twice as sweet as lactose (Dan et al., 2023).We surmise that the use of mutant IMAU20246Y would reduce the amount of added sugar in actual production, while still providing consumers with the desired taste experience, which would also reduce the caloric content effects of foods, as people are now more interested in low-calorie foods (Daher et al., 2022).Although the content of this study was extensive, we only compared the sensory evaluation of single-strain fermented milk and did not evaluate the sweetness of wild-type S. thermophilus and its mutant compounded with L. delbrueckii ssp.bulgaricus for fermented dairy production, respectively, and did not compare whether different amounts of sucrose could maintain the same perceived sweetness, which is one of the limitations of this study.In future experiments, we will discuss the effects of strain combination and sucrose addition on product sweetness.
Streptococcus thermophilus is an important industrial starter culture and can be used alone or in combination with other microorganisms for production of dairy products such as cheese and yogurt (Han et al., 2016).However, during production of starter culture powders, viable cell counts are often significantly reduced during lyophilization, which affects industrial production (Marcial-Coba et al., 2018).Therefore, it is of great importance to improve freeze-drying survival of microorganisms.Changing the culture conditions of microorganisms is an effective way to improve survival following lyophilization.The conventional M17 medium can satisfy the normal culture of S. thermophilus, but it is necessary to culture S. thermophilus with stronger physiological activity if it is to better resist damage due to vacuum freeze-drying.
As an effective statistical technique for optimizing multiple variables and predicting the best performance from the fewest number of experimental runs (Gouda et al., 2001), RSM has been applied to optimize the composition of media for biological production (Nikerel et al., 2006).Lu et al. (2017) used RSM to optimize the cryoprotective medium of S. thermophilus STX2; after freeze-drying in the optimal medium, cell survival increased to 93.58%.In this study, we used a singlefactor experiment to identify individually the best carbon source, nitrogen source, growth factor, trace element, pH, and inoculum quantity (by volume) of the medium (Figure 5).Then, a PB experimental design was used to determine which 3 factors had the greatest influence on mutant growth; these were found to be galactose content, nitrogen source content, and pH (Table 4).Finally, the medium composition of S. thermophilus IMAU20246Y was optimized in a BBD experiment to establish the best medium composition to achieve maximum growth (Supplemental Table S12, see Notes; Figure 6).Under these conditions, the viable cell count of the mutant reached 4.15 × 10 8 cfu/mL, which was 11.79 times higher than could be achieved before optimization.The survival rate of S. thermophilus IMAU20246Y was 67.42% after vacuum freeze-drying with a selected protective agent.Through an RSM optimization experiment, we obtained the best medium components for mutant IMAU20246Y, improved the vitality of mutant significantly (P < 0.05), and prepared a high-vitality bacterial powder, which can be applied to the preparation of fermented milk, cheese, and other products.
At present, we have successfully developed a strain of S. thermophilus using NTG that can metabolize galac-tose and optimized its culture conditions using the RSM method to prepare a high-vitality bacteria powder.However, the whole metabolic regulation of the mutant strain is still unclear, and future work will focus on exploring the metabolic mechanisms of the mutant using a combination of proteomics and transcriptomics.

CONCLUSIONS
This study used NTG treatment of S. thermophilus IMAU20246 to produce 15 mutants that metabolized galactose; from these, the mutant strain IMAU20246Y was selected.β-Galactosidase, glucokinase, and pyruvate kinase were significant in this mutant compared with the wild-type strain.Furthermore, the genes encoding β-galactosidase and glucokinase were changed and the original metabolic pathway of the strain was altered following exposure to NTG.The mutant was able to metabolize more lactose and galactose and secrete more glucose extracellularly.We determined the optimal culture conditions for the mutant by optimizing the basic medium.Under these conditions, microorganisms were cultured and freeze-dried to prepare a bacterial powder that significantly improved cell survival.This study not only provided data supporting use of beneficial mutagenesis of LAB to develop an improved starter culture but also provided new and useful microbial resources for industrial applications.

NOTES
This research was supported by the National Natural Science Foundation of China (Beijing; No. 32072235), the Natural Science Foundation of Inner Mongolia (Hohhot; No. 2022MS03013), Fundamental Research Funds of Inner Mongolia Agricultural University (Grant No. BR221203), and the earmarked fund for CARS .Author contributions: HH, writing original draft and editing, data curation, visualization; QM and JN, investigation, methodology, data curation; NW, investigation, data curation; TD, data curation, review and editing, project administration.Supplemental material for this article is available at https: / / doi .org/ 10 .17632/w37b6t8gtf .1.Because no human or animal subjects were used, this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.The authors have not stated any conflicts of interest.

Nonstandard abbreviations used
Figure 1.β-Galactosidase (a), glucokinase and pyruvate kinase (b), and galactokinase (c) activities of Streptococcus thermophilus IMAU20246 and its mutants.Lowercase letters above bars represent the significance of differences between strains in the activity of the same enzyme; those with the same letter were not significantly different to each other.Error bars indicate SD.
Figure 2. Growth of Streptococcus thermophilus IMAU20246 and mutant IMAU20246Y on M17 broth containing lactose and galactose.OD 600 = optical density at 600 nm.
indicate significant differences in sugars or acids of different strains at the same time (P < 0.05); those with the same letter were not significantly different to each other.A-H Uppercase superscript letters indicate significant differences in sugars or acids of the same strain at different times (P < 0.05); those with the same letter were not significantly different to each other.1 Dashes indicate no detection (content <0.02 g/100 g).Data are presented as mean ± SD.

Figure 5 .
Figure 5. Effects of different (a) carbon sources, (b) galactose content, (c) nitrogen source, (d) nitrogen source compound ratios, (e) carbon: nitrogen ratios, (f, g) trace elements, (h, i) growth factors, (j) initial pH of medium, and (k) inoculation volume on the density of mutant Streptococcus thermophilus IMAU20246Y.OD 600 = optical density at 600 nm.Lowercase letters on the bars indicate the significance of differences between strains under each medium condition evaluated; bars with the same letter were not significantly different to each other.Error bars indicate SD.

Figure 6 .
Figure 6.Contour plots and response surface plots described by the second-order model for the interaction between (a) galactose content and nitrogen source content, (b) galactose content and pH, and (c) nitrogen source content and pH.OD 600 = optical density at 600 nm.

Table 1 .
Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE Mortality rate of Streptococcus thermophilus IMAU20246 after treatment with different concentrations of NTG after incubation for 60 min at a-f Lowercase superscript letters represent significant differences in the mortality ratio of strains with different concentrations of NTG (P < 0.05); those with the same letter were not significantly different to each other.

Table 3 .
Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE Changes in genetic stability of glucokinase, pyruvate kinase, β-galactosidase, and galactokinase activity of the mutant strain Streptococcus thermophilus IMAU20246Y after passage through 10 consecutive generations 1 EnzymeLowercase letters indicate significant differences in the same enzyme activity at different times (P < 0.05); those with the same letter were not significantly different to each other.
1Data are presented as means ± SD.Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE

Table 4 .
Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE ANOVA analysis for the Plackett-Burman design 1

Table 5 .
Hu et al.: S. THERMOPHILUS MUTANT STRAIN AS STARTER CULTURE ANOVA analysis for the Box-Behnken design 1