Lactobacillus paracasei ZFM54 alters the metabolomic profiles of yogurt and the co-fermented yogurt improves the gut microecology of human adults

Gut microbiota imbalance could lead to various diseases, making it important to optimize the structure of flora in adults. Lactobacillus paracasei ZFM54 is a bacteriocin and folic acid producing Lactobacillus strain. Herein ZFM54 was used as the potentialy probiotic bacterium to ferment milk together with a yogurt starter. We optimized the fermentation conditions and the obtained yogurts were then subjected to volatile and non-volatile metabolome analysis, showing that ZFM54 cannot only improve the acidity, water holding capacity and live lactic acid bacteria counts, but also improve many volatile acid contents and increase some beneficial non-volatile metabolites such as N-ethyl glycine and L-Lysine, endowing the yogurt with more flavor and better function. The regulatory effects of the co-fermented yogurt on intestinal microecology of volunteers were investigated by 16S rRNA sequencing and short-chain fatty acids (SCFAs) analysis after a continuous consuming the yogurt of 2-week, showing better effect to increase the relative abundance of beneficial bacteria such as Ruminococcus and Alistipes , decrease harmful bacteria ( Escherichia - Shigella and Enterobacter ), and enhance the production of SCFAs (acetate, propionate and butyric acid) than the control yogurt. In conclusion, L. paracasei ZFM54 can significantly improve the health benefits of yogurt, laying the foundation for its commercial application in improving gut microbiota.


INTRODUCTION
The human gut contains a wide variety of commensal microbiota referred to as the gut microbiota, playing an important role in host physiological functions, including drug metabolism, energy metabolism, nutrient absorption, and immune regulation (Walsh et al., 2014).Therefore, the gut microbiota is considered to be an important organ acquired by the human body, called as the "ignored organ" (Gomaa, 2020).The abundance and diversity of gut microbiota and bacterial metabolites such as short-chain fatty acids (SCFAs) affect a series of physiological processes essential to host health (Morrison and Preston, 2016).Although "healthy" intestinal flora is regarded as a stable community(O'Toole and Claesson, 2010), it can be altered due to various internal and external factors (such as age (Mariat et al., 2009), diet (Claesson et al., 2012), antibiotics (Robinson and Young, 2010), smoking (Biedermann et al., 2013), psychological pressure (Jiang et al., 2015), geographical position (Tyakht et al., 2013), etc.) during the whole human life, resulting in the imbalance of intestinal microecology that may harm the human health (Li et al., 2017).Although numerous drugs for gut microbiota have been emerging, diet remains the most rapid and important factor in remodeling the structure and function of gut microbiota (Thursby and Juge, 2017).
In the last decades, researchers have found that several strains of Lactobacillus cannot only form a film to protect the mucosa of the gastrointestinal tract (Pessione, 2012), but also produce a variety of active substances (Wang et al., 2021b), which can optimize the structure of gut microbiota.At present, several strains of Lactobacillus have been used to make various oral dietary supplements and drugs (Mani-López et al., 2014) marketed as a potential way to prevent a variety of gastrointestinal diseases.Among them, yogurt has been confirmed to offer a positive impact on human health (Savaiano and Hutkins, 2021).Astrup et al.(Astrup, 2014) showed that yogurt consumption is associated with a reduced risk of cardiovascular disease.Buendia et al. (Buendia et al., 2018) conducted a follow-up study of 3 groups of participants for 20-30 years and found that higher yogurt intake was associated with a lower risk of hypertension.Panahi et al.(Panahi and Tremblay, 2016) reported that yogurt is involved in the control of weight and energy homeostasis and plays a role in reducing the risk of type 2 diabetes.In fact, regular consumption of yogurt cannot only balance trace elements (Cifelli et al., 2020), fight against aging (Teneva-Angelova et al., 2018), moisturize the skin and hair, brighten the eyes and replenish calcium (Keast et al., 2015), but also boost immunity (Daliri and Lee, 2015), regulate brain activity (Tillisch et al., 2013), and relieve irritable bowel syndrome (Mofid et al., 2020).
At present, the lactic acid bacteria starter used in market are mostly foreign commercial strains, and the development and commercial application of strains with independent intellectual property rights in China are limited.Recently, several reports have been revealed that some potentialy probiotic can be used to co-ferment with starter cultures to produce yogurt, such as Lactobacillus plantarum P-8 (Dan et al., 2018), Lactobacillus casei Zhang and Bifidobacterium animalis (Wang et al., 2021a), and Bifidobacterium ssp.lactis Probio-M8 (Wang et al., 2021a), which could further improve the flavor and nutritional value of yogurt.Previous studies in our laboratory showed that Lactiplantibacillus paracasei ZFM54 is a lactic acid bacterium (Qureshi et al., 2020) that can produce bacteriocin (Zhu et al., 2021) and folate acid, and inhibit a variety of pathogens including Helicobacter pylori (Zhou et al., 2022).
To further understand its potentialy probiotic function and metabolic properties, herein we investigated the effect of L. paracasei ZFM54 on yogurt ordinary qualities, volatile and non-volatile metaboloms and further evaluated the effects of the yogurt on human fecal microbiota ecology and SCFAs production after consuming by the volunteers.This study not only promotes the research and development of lactic acid bacteria starter in China, but also lays the foundation for the commercial application of L. paracasei ZFM54 in improving the gut microbiota.

Strain culture
L. paracasei ZFM54 is screened from infant feces and stored in China Model Culture and Collection Center (CCTCC; Wuhan, China; Collection No. M 2, 016, 667).The L. paracasei ZFM54 storage tube was propagated twice in de Man-Rogosa-Sharpe (MRS) broth before use.Twenty μL of bacterial suspension of ZFM54 was inoculated into a 1-L MRS liquid culture medium.The broth culture was centrifugated at 8,000 rpm for 30 min at 4°C.The precipitate was washed twice with PBS and resuspended to a bacterium concentration of 1 × 10 7 cfu/mL for subsequent yogurt fermentation.
Sensory assessment.Twenty volunteers (20 healthy volunteers, half male and half female, normal taste and smell,no history of intestinal diseases) were recruited to form a sensory evaluation group.The operation was as follows: taking an appropriate amount of samples into a 50 mL beaker, the volunteers were first asked to observe the color, appearance and mobility of various samples, then smell the samples, and finally taste the samples after gargling with warm boiled water.Referring to the industrial standard of Chinese dairy industry (RHB 601-2005), we developed the sensory scoring criteria (Table S1), with the evaluation indexes including sweet and sour degree, color, taste, odor and tissue state.Based on the criteria, volunteers gave a comprehensive score of the yogurts, and the average value served as the score of each yogurt.
Quantifying pH and acidity.Each sample was read 3 times repeatedly using a pH meter, and the average value was taken and recorded.Acid-base titration method was used to determine the titrated acidity as described in the national Standard GB5009.239-2016,Determination of Food Acidity in National Food Safety Standards of China.

Yogurt fermentation
12.4% skim milk powder and 6.5% sucrose were placed in an appropriate amount of distilled water at 50°C, and stirred well to fully dissolve.The solution was homogenized twice under pressure of 15 MPa and 35 MPa, respectively.The homogenized milk matrix was pasteurized (95°C, 5 min) and then quickly cooled in cold water to 4°C.The fermentation was carried out by inoculating L. paracasei ZFM54 bacterial suspension of 1 × 10 7 cfu/mL and 0.00109% commercial starter VEGE 033 (Danisco, America, ≈5 × 10 6 cfu/ mL), marked as yogurt LS.Commercial starter VEGE 033 was inoculated with the same amount of inoculum separately for fermentation, serving as control and marked as yogurt S. The 2 yogurt samples were placed in a yogurt machine (Rikon Electrical Co., Ltd., Beijing, China), and the fermentation temperature and time were set to 42°C and 8 h, respectively.After the fermentation, it was removed to 4°C and refrigerated.
Determination of water holding capacity.Ten g sample (M) was accurately weighed into the centrifuge tube (M1), centrifuged for 5 min at a speed of 7,100 g.After centrifugation, the supernatant was removed, and the weight of the sample and the centrifuge tube (M2) was recorded, and the water holding capacity (W) was calculated according to the following formula: W = (M2-M1)/M × 100%.
Determination of nutrient contents.Specific determination methods for protein, fat and some nonfat milk solids refer to National Food Safety Standard Pasteurized Milk (GB 19645-2010).
Determination of microbial indexes.The viable counts of lactic acid bacteria, Escherichia coli, Staphylococcus aureus, Salmonella, mold and yeast in yogurt LS or S were determined according to GB 4789.35-2016, GB 4789.9.3-2016, GB 4789.10-2016, GB 4789.9.4-2016and GB 4789.15-2016, respectively.Briefly, The viable counts of lactic acid bacteria, Escherichia coli, Staphylococcus aureus, Salmonella, mold and yeast were determined using Membrane filtration method.This method involves filtering the yogurt sample onto a specific membrane, then placing the membrane into different culture dishes containing nutrient liquid required for lactic acid bacteria, Escherichia coli, Staphylococcus aureus, Salmonella, mold and yeast growth, respectively.After culturing under suitable conditions for a certain time, the number of colony-forming units (cfu) is counted.

Determination of volatile metabolites
The volatile metabolites of yogurt LS and S were analyzed based on SPME-GC-MS.Solid phase Extraction (SPE) process: yogurt samples were placed in a 20 mL headspace bottle, sealed using a cap with a silicone rubber septum.The headspace bottle was kept in 50°C water for 30 min, and then the extraction head was inserted into the headspace bottle at a distance of 1 cm from the liquid level.The extraction head was adsorbed at 50°C for 30 min, removed and inserted into the GC injection port, and analyzed at 250°C for 2 min before GC mass spectrometry analysis.
Data processing: spectrograms obtained by GC-MS were searched in NIST14 database and compared with standards for substance qualitative analysis, and the relative peak area referred to the content of volatile substances.
Determination of non-volatile metabolites An untargeted UPLC-Q-TOF-MS platform was used to analyze the non-volatile metabolites of yogurt LS and S.
Sample pretreatment: thaw the yogurts on ice (all subsequent operations are carried out on ice); the samples were then vortexed and mixed for 10 s, and take 50 μL of each sample into a centrifuge tube.After adding 300 μL acetonitrile methanol internal standard extract, the samples were vortexed for 3 min and centrifuged at 4°C, 10,000 rpm for 8 min.After centrifugation, 180 μL of the supernatant was transferred to the lining tube of the corresponding injection bottle for machine analysis.
The chromatographic conditions: HSS C18 column (1.8 μm, 2.1 mm × 100 mm, Waters); column temperature: 40°C; flow rate: 0.40 mL/min; injection volume: 2 μL; mobile phase A: ultra-pure water (0.1% formic acid); mobile phase B: acetonitrile (0.1% formic acid).The gradient conditions of mobile phase on the chromatographic column were shown in Table S2.The positive and negative ion scanning mode was used for quality spectrum signal acquisition, and the MS conditions were shown in Table S3 Data analysis: after correcting and filtering the original data, the metabolites were searched in the laboratory's self-built database and integrated public databases for qualitative analysis; R program was used to conduct Student's t-test and fold change analysis for univariate statistical analysis, principal component analysis (PCA) and orthogonal partial least squares discriminant (OPLS-DA) analysis for multivariate statistical analysis.

Experimental design of yogurt intervention
According to the principle of "informed consent and voluntary participation," 20 healthy volunteers, half male and half female, were recruited.Inclusion criteria: normal body mass index, 29.9 ≥ BMI ≥18.5; no history of intestinal diseases; no use of antibiotics or antifungal drugs in the last 3 mo before the test; consent to consume the test products during the whole period; consciously abstain from spicy and oily foods and other probiotic drinks and foods.The volunteers were randomly divided into 2 groups to consume yogurt LS or yogurt S at a dose of 400 mL/day.The trial period was 21 d in total, divided into a 7-d voiding period and a Chen et al.: Lactobacillus paracasei ZFM54… 14-d drinking period, and stools were collected from all volunteers on d 7 and 21.The study was conducted as approval by the Ethics Committee of Zhejiang Gongshang University (Approval number: 20212203-01).Among the 20 volunteers, 2 of each group did not consciously abstain from spicy and oily foods and other probiotic drinks and food, thus finally 8 samples of each group were used.

Fecal sample collection and pretreatment
Fecal samples were collected individually following an SOP of self-collection (http: / / www .microbiome-standards .org),and a sampling kit with stabilizing solution was obtained from Beijing Genomics Institute (BGI).Samples were frozen in liquid nitrogen and stored at −80°C for further sequencing at LC-Bio (Hangzhou, China) or detection of SCFAs.

Determination of SCFAs
The contents of SCFAs were determined using a gas chromatograph-mass spectrometer (GC-MS QP2010-Ultra, Shimadzu, Japan) as reported previously (Liu et al., 2022).

DNA Extraction and 16S rRNA Sequencing
DNA from different samples was extracted using the QIAamp DNA Stool Mini Kit (cat 51604) as per manufacturer's instructions.The quality of DNA extraction was detected by agarose gel electrophoresis.The 5′ ends of the primers were tagged with specific barcodes per sample and sequencing universal primers.PCR amplification was then performed.The PCR products were purified by AMPure XT beads (Beckman Coulter Genomics, Danvers, MA, USA) and quantified by Qubit (Invitrogen, USA).The amplicon pools were prepared for sequencing and the size and quantity of the amplicon library were assessed on Agilent 2100 Bioanalyzer (Agilent, USA) and with the Library Quantification Kit for Illumina (Kapa Biosciences, Woburn, MA, USA), respectively.The libraries were sequenced on NovaSeq PE250 platform provided by LC-Bio.
Paired-end reads was assigned to samples based on their unique barcode and truncated by cutting off the barcode and primer sequence, and merged using FLASH.NovaSeq 6000 sequencer, and quality control and chimeric filtering were performed to obtain high quality ASVs (Amplicon Sequence Variants).Then the concept of ASVs was used to construct the class OTU (Operational Taxonomic Units) table to obtain the final ASV feature table as well as the feature sequences.Alpha diversity and β diversity were calculated by nor-malized to the same sequences randomly.Alpha diversity and β diversity were calculated with QIIME2, and the graphs were drew by R package (v3.5.2).Blast was used for sequence alignment, and the feature sequences were annotated with SILVA database.Other difference analysis and diagrams were implemented using the OmicStudio cloud platform (https: / / www .omicstudio.cn/index).

Statistical analysis.
All results are presented as mean ± standard deviation (SD).Graphs of in vitro indexes, and SCFAs were drawn using GraphPad Prism 9.0, and the differences between the 2 groups were compared by Student's t-test.The differences in intestinal microbiota were analyzed using the Wilcoxon rank-sum test.A P-value <0.05 was considered statistically significant.

Influence of single factor on the co-fermented yogurts
First, L. paracasei ZFM54 was compounded with L. delbrueckii ssp.bulgaricus and S. thermophilus (1:1000) yogurt starter at 4 different inoculation ratios (1:2, 1:1, 2:1, and 3:1) to ferment milk.The other conditions were set as follow: inoculum volume, 1%; sucrose content, 6.5%, fermentation temperature, 42°C; fermentation time, 8 h.After fermentation, the optimal strain ratio was determined according to the sensory evaluation and the acidity of the yogurt (Figure 1A and 1B).When the strain ratio was 2:1, the yogurt had delicate tissue, suitable taste and the highest sensory score.With the increase of the relative proportion of L. paracasei ZFM54, the acidity of yogurt increased and then decreased.The high proportion of L. paracasei ZFM54 might accelerate the acid production, resulting in the high acidity to inhibit the growth of cocci.Thus when the strain ratio is 2:1, the fermentation strains in yogurt are compatible with each other and have the best symbiotic effect.
Second, a series of inoculum quantities (0.5%, 1%, 1.5%, 2%, 3%) were set, and the other conditions were as follow: the inoculum ratio, 1:1; the sucrose content, 6.5%; fermentation temperature, 42°C; fermentation time, 8 h.The results showed that with the increase of inoculation amount, the comprehensive sensory score of yogurt gradually decreased (Figure 1C).The inoculation quantity had little effect on the color and smell of yogurt, but had great influence on the organizational state and taste.When the inoculation quantity was 1.5%, the obtained yogurt had the best quality, with an acidity greater than 70 °T.The acidity kept stable when the inoculation amount increased continuously (Figure 1D).Therefore, 1.5% of the inoculation amount would be the best.
Third, we set various fermentation time (4, 6, 8, 10, and 12 h) and found that with the extension of fermentation time, the acidity value of yogurt increased gradually, and the total sensory score increased first and then decreased (Figure 1E and 1F).When the fer- mentation time was 4 h, the texture of yogurt was soft and the taste was poor.When the fermentation time was 8 h, the total score was the highest.Moreover, the gel structure of yogurt was damaged after a long time fermentation, especially 12 h, with cracks appearing on the surface.Thus, the optimal fermentation time was set as 8 h.
Following, we evaluated the effect of fermentation temperature.As shown in Figure 1G and 1H, 5 temperature gradient (37°C, 40°C, 42°C, 43°C and 45°C) was set and the best fermentation temperature was determined to be 42°C, as with the increase of temperature, the curdling effect and taste of yogurt became worse and worse.When the temperature was 42°C, the yogurt had inviting flavor, excellent tissue state and the highest total score.When fermentation temperature was 37°C or 40°C, low acidity yogurt was milky, not thick enough, and was too sweet.
Finally, we screened the proper sucrose content.The added amounts of sucrose were set as 2.5%, 4.5%, 6.5% and 8.5% and found that the sucrose content had little effect on acidity, but great effect on taste (Figure 1I  and 1J).When the sucrose content was 6.5%, the total score of yogurt was the highest.

Optimization of yogurt fermentation process using response surface test
Based on the results of single factor experiments, 3 independent variables including starter inoculated amount (A), fermentation time (B) and fermentation temperature (C) were selected, and the sensory score (Y) was taken as the response value.Box-Behnken was used to design the 3-factor and 3-level experiment.As shown in Table S4 and Table S5, the quadratic multiple regression model was extremely significant (P < 0.01).The missing fitting term P = 0.9442 > 0.05 indicated that the error was not significant compared with the pure error, suggesting that the model had a good degree of fitting.The coefficient of determination R 2 = 0.9983, indicated that 99.83% of the data changes could be explained by this model.P < 0.01 for A, C, AC, BC, B 2 and C 2 , indicated that these factors had a great influence on the quality of yogurt.According to the F value of each factor, the order of influence of each factor in the primary term was as follow: inoculation amount (A) > fermentation temperature (C) > fermentation time (B).
Design-Expert 8.0.6 software was used to make response surfaces and contour lines of each factor with sensory scores as response values.When the response surface graph is oval or saddle shape, it means that the interaction between the 2 factors is significant.As shown in Figure 2, the ovality of contour plots AB, AC and BC was very large, indicating that the interaction effects of AB, AC and BC were extremely significant (P < 0.01), which was consistent with the results of ANOVA (Table S5).Among them, the 2-dimensional contour map of BC showed an oval shape, the stability point was close to the center point, and the 3D response surface graph presented an arch, indicating that fermentation time (B) and fermentation temperature (C) have the most significant interaction influence.
Conclusively, the optimal fermentation conditions of L. paracasei ZFM54 compound yogurt obtained were as follows: strain ratio, 2:1; inoculation amount, 1.5%; fermentation time, 8 h; fermentation temperature, 42°C; sucrose content, 6.5%.The verification fermentation experiment was carried out under the optimal condition.After 6 parallel experiments, the actual sensory score was 90, close to the predicted value, proving that the process parameters of fermented yogurt obtained by response surface analysis method were reliable and of guiding significance.

Analysis of physicochemical properties of the yogurts
The physical and chemical properties, and microbial levels of yogurts can comprehensively reflect the overall quality of the products.Under the optimal fermentation conditions, yogurt LS and yogurt S are produced.Referring to GB19302-2010 "National Standard for Food Safety," the physical and chemical indexes, microbial indexes and viable numbers of lactic acid bacteria of the yogurts (LS and S) are determined (Table 1).
Both the physical and chemical indexes of the 2 yogurts met the national standards.Compared with commercial starter culture, adding L. paracasei ZFM54 had little effect on the contents of fat, non-milk fat solids and protein in yogurt, but increased the acidity and water retention of yogurt.This may attribute to the symbiotic effect when L. paracasei ZFM54 coexists with the strains in commercial starter cultures, accelerating the acid production, promoting the hydration of lactic acid and protein, and changing the structure of casein.In addition, yogurt LS contained 6.0 × 10 9 cfu/mL lactic acid bacteria, which was 6 times that of yogurt S. The more viable lactic acid bacteria, the more colonized in the gastrointestinal tract, giving more regulatory effect to improve gut microbiota.Thus, L. paracasei ZFM54 has a certain contribution to the formation of unique sensory quality of yogurt and the improvement of probiotic levels.

Comparison of volatile metabolomic profiles of yogurt S and LS.
A total of 20 volatile components were detected by GC-MS, including 8 ketones, 5 acids, 4 alcohols, 2 alkenes and 1 ester.As shown in Figure 3 and Table 2, the relative contents of 2, 3-butanedione, 3-hydroxy-butanone, 2-heptanone, caproic acid and octanoic acid in these 2 kinds of yogurts are relatively high, and should be the main flavor components in the yogurt.Compared with yogurt S, the contents of 2, 3-butanedione, butyric acid, caproic acid, octanoic acid and n-decanoic acid in sample LS were significantly increased, and D-limonene flavor substance specifically existed in sample LS. 2, 3-butanedione is a fatty acid formed from methyl ketone oxide, giving sour cream or nutty flavor.Butyric acid, caproic acid, octanoic acid and n-decanoic acid can regulate the acidity of yogurt and emerge a unique taste.Butyric acid can give cheese flavor, and caproic acid can show coconut oil flavor, etc. (Magne et al., 2020).Numerous fatty acids such as caproic acid, octanoic acid, and capric acid help alleviate nutritional malabsorption syndrome, small bowel dysfunction, and hereditary pancreatic diseases.D-limonene imparts fresh orange and lemon-like aroma to yogurt.Thus, ZFM54 cannot only assist commercial starter cultures to ferment, but also improve the volatile acid contents in yogurt and offer more flavor.

Comparison of nonvolatile metabolomic profiles of yogurt S and LS
Many non-volatile bioactive metabolites, such as certain vitamins, essential aminoacids, bioactive peptides, other organic acids or fatty acids, can give yogurt good sensory quality and more health benefits (Settachaimongkon et al., 2014).The differences of non-volatile metabolites between yogurt LS and S samples were investigated using an untargeted UPLC-Q-TOF-MS platform.
1,328 and 861 nonvolatile components were identified in the positive and negative ion mode, respectively.The PCA plot provided a clear difference between S and LS (Figure 4A and 4B).Volcano plots were created to compare the distribution of 56 significantly different nonvolatile metabolites in yogurt LS compared with Chen et al.: Lactobacillus paracasei ZFM54… yogurt S (Figure 4C and 4D), of which 49 were identified in the positive ion mode and 9 were in the negative ion mode with 2 shared by both.Of these metabolites, the levels of 19 metabolites were higher and the levels of 37 were lower in LS than in S. To identify pathways of significantly changed metabolic features induced by ZFM54, we employed the mummichog algorithm included in MetaboAnalyst for pathway enrichment assay, which considers both the number of detected metabolomic features in individual pathways and their alterations between conditions.The enriched pathways included metabolism of Tyrosine, 2-Oxocarboxylic acid, Cyanoamino acid, and Phenylalanine, biosynthesis of Tropane, piperidine and pyridine alkaloid, plant secondary metabolites, folate and aminoacyl-tRNA, and absorption of Protein and Mineral (Figure 4E).

Effects of yogurt on the diversity index
The results of yogurt intervention on each index of α diversity of the gut microbiota are shown in Figure S1A.The Chao1 index and Observed species index of the gut microbial community of volunteers consuming yogurt S were not significantly changed, but when L. paracasei ZFM54 was added to yogurt, the Chao1 index and Observed species index were significantly increased.However, neither of the 2 yogurts intervention had a significant effect on the Shannon and Simpson indices.These results indicated that addition ZFM54 to yogurt significantly increased gut microbiota richness, but had no significant effect on gut microbiota diversity.In the 2-dimensional plots of both bray_curtis PCoA and jaccard PCoA, the structural composition of the fecal microflora of volunteers before and after drinking either yogurt was similar with little variability (Figure S1B and S1C), indicating that addition L. paracasei ZFM54 in yogurt cause no significant changes in the main composition structure of the gut microbiota.

Effect of yogurt on species composition of gut microbiota
The relative abundance of various bacteria in different groups were shown in Figure 6A and 6B.At the phylum level (Figure 6A), all samples mainly included Firmicutes, Actinobacteria, Proteobacteria, and Bacteroidota, accounting for more than 98% of the total bacteria, followed by Verrucomicrobiota, Desulfobacterota and others.Among the stool samples obtained from volunteers before yogurt S consumption, Firmicutes (65.75%) accounted for the highest percentage, followed by Actinobacteria (26.25%),Bacteroidota (4.80%) and Proteobacteria (2.90%).Firmicutes (79.01%) and Bacteroidota (8.17%) increased, and Actinobacteria (8.98%) decreased after yogurt S consumption.The samples of volunteers who drank yogurt LS mainly contained the same above 4 phyla.The relative abundance of Firmicutes (from 56.32% to 79.59%) and Actinobacteria (from 10.88% to 7.65%) obtained similar results in the LS group.It's worth noting that the relative abundance of Proteobacteria (from 21.45% to 6.35%) decreased extremely after drinking LS, while consuming yogurt S induced a slight increase in the Proteobacteria abundance (From 2.90% to 3.62%).In addition, the relative abundance of Bacteroidota (from 9.99% to 6.07%) also decreased in the Y_LS group, which was increased in the Y_S group.
At the genus level (Figure 6B), the gut microbiota of volunteers with a relative high abundance (>4%) mainly include Faecalibacterium, Bifidobacterium, Escherichia-Shigella, Megamonas, Bacteroides, Dialister, Agathobacter, Ruminococcus, and Klebsiella.After consuming yogurt S, the relative abundance of Faecalibacterium, Bacteroides, Megamonas, and Ruminococcus in the intestinal flora increased, while the relative abundance of Bifidobacterium, Escherichia-Shigella, Dialister, and Agathobacter decreased.Similar variation trend was obtained when ZFM54 co-fermented yogurt was consumed.Noteworthy, compared with yogurt S group, drinking yogurt LS exacerbated the decline of Escherichia-Shigella and alleviated the decline of Bifidobacterium (Figure 6B).Especially, presentation of data at individual genus level revealed even more consistent differences after consumption yogurt LS or yogurt S (Figure 6C).These results indicate that addition of ZFM54 during yogurt fermentation could enhance the regulation of human gut microbiota to gain a better homeostasis.

Significant differences in gut microbiota
To further figure out the differences between yogurt LS and yogurt S in regulating gut microbiota, we con- ducted LEfSe analysis to identify the relevant flora with significant differences, and the results were shown in Figure 7.There were 38 (15 upregulated, 23 downregulated) significantly different species in the fecal microflora after yogurt S consumption, of which 12 at the level of "genus" including Faecalibacterium, Lachnospira, Leptotrichia, Eubacterium_siraeum_group, Neisseria, Methylobacterium_Methylorubrum, Lactiplantibacillus, Gemmobacter, Companilactobacillus, Mitochondria_unclassified, Raoultella and Staphylococcus (Figure 7A and 7B).Among them, Faecalibacterium, Lachnospira, Leptotrichia, Eubacterium_siraeum_group, and Neisseria increased.As for ZFM54 co-fermented yogurt, the number of significantly different species was 32 (17 upregulated, 15 downregulated), including 8 at the level of "genera" (Figure 7C and 7D).The increased genera included Lentilactobacillus, Lacticaseibacillus, Phocea, Phascolarctobacterium, Pediococcus, Ruminococcaceae_ unclassified, Alistipes, and Butyricimonas.The relative abundance of beneficial bacteria Lacticaseibacillus and Lentilactobacillus increased significantly, while that of harmful bacteria Bacillus decreased significantly after yogurt LS consumption.In addition, the relative abundance of genera associated with the production of SCFAs, such as Alistipes (Parker et al., 2020) and Lachnospira (Li et al., 2020), showed an upward trend, which might help to increase the levels of SCFAs in the intestinal contents following yogurt intervention.The results of the LEfSe analysis indicated that addition of L. paracasei ZFM54 to yogurt was able to further increase the content of beneficial bacteria and decrease the content of harmful bacteria.

Effects of yoghurt on SCFAs
We measured the contents of SCFAs (acetic acid, propionic acid, butyric acid, and isobutyric acid) in the feces of volunteers before and after drinking yogurt LS or S using GC.The levels of acetic and propionic acids decreased and the levels of butyric and isobutyric acids increased after yogurt S intervention (Figure 8).However, the concentrations of acetic acid, propionic acid and butyric acid were significantly higher after consuming yogurt LS than consuming yogurt S (Figure 8).In conclusion, addition of L. paracasei ZFM54 to yogurt may provide better substrate source for SCFAs producing bacteria, thus promoting the increase of acetic acid, propionic acid and butyric acid content, beneficial to human health.

DISCUSSION
The potential probiotic of L. paracasei ZFM54 to coordinate intestinal flora and inhibit Helicobacter pyloriinduced gastritis has been confirmed previously (Zhu et al., 2021, Zhou et al., 2022).Herein we evaluate the effect of L. paracasei ZFM54 on improving the flavor and quality of yogurt and how the co-fermented yogurt regulate the human intestinal flora.We obtained the optimal fermentation conditions of L. paracasei ZFM54 compound yogurt: strain ratio, 2:1; inoculation amount, 1.5%; fermentation time, 8 h; fermentation temperature, 42°C; sucrose content, 6.5%.Through SPME-GC-MS and UPLC-Q-TOF-MS, we confirmed that L. paracasei ZFM54 could significantly influence volatile and nonvolatile metabolomic profiles of yogurt to improve many volatile acid contents and alter lots of nonvolatile metabolites, resulting in more flavor.Finally, we compared the effects of L. paracasei ZFM54 co-fermented yogurt and commercial fermented yogurt on the gut microbiota of the volunteers, as well as on the contents of SCFAs in the feces of the volunteers, showing that ZFM54 optimized the flora structure and promoted the production of acetic acid, propionic acid and butyric acid.This work provides theoretical basis and experimental evidence for the commercial application of L. paracasei ZFM54 in improving the quality of yogurt and coordinating intestinal microecology.
Through sensory evaluation, we optimized the fermentation conditions for L. paracasei ZFM54 to co-ferment with commercial starter VEGE 033 and obtained high-quality yogurt with slight sweetness and sour, as well as excellent physicochemical properties.The viable probiotic cell count is a core index of fermented milk, driving the milk industry to improve the number of viable bacteria in its yogurt products (Tamang et al., 2016).Herein the yogurt containing ZFM54 has more live Lactobacillus, about 6 times as much as control yogurt S, suggesting that ZFM54 could promote the proliferation and survival of Lactobacillus during the fermentation under the optimized conditions.
By comparing the volatile and nonvolatile metabolites, different metabolomic profiles were observed between yogurt LS and S.    As for the volatile metabolites, 2,3-Butanedione, acetoin, 2-butanone, and 2,3-pentanedione contribute most to the typical flavor of yogurt, similar to the previous study (Yuan et al., 2022).L. paracasei ZFM54 could significantly increase the level of 2,3-butanedione and 2,3-pentanedione, which might be due to accumulation of the precursors of the diketones, 2-acetolactate and 2-acetohydroxybutyrate during fermentation (Ott et al., 1999).2,3-butanedione and 2,3-pentanedione could contribute to yogurt aroma with buttery smells and offer a creamy and nutty flavor (Ott et al., 1999, Papaioannou et al., 2021).Moreover, adding ZFM54 could extremely  increase the content of D-limonene, which gives yogurt a fresh orange and lemon-like aroma (He et al., 2021).
As for the non-volatile metabolites, adding ZFM54 induced 56 differentially altered components.Among the differential metabolites include 13 amino acids and their metabolites, of which 4 (N-lactoyl-phenylalanine, N-ethylglycine, L-lysine and Leu-Asn) were significantly upregulated by ZFM54 involvement.A recent report revealed that exercise stimulates the production of Nlactoyl-phenylalanine to suppress feeding and obesity (Li et al., 2022).Another report also showed that systemic treatment with N-ethylglycine results in an efficient amelioration of hyperalgesia and allodynia without affecting acute pain and with no side effects (Werdehausen et al., 2015).L-lysine needs to be intaked with adequate diets to keep human healty (Hayamizu et al., 2020).As for the organic acids, 3 metabolites including (2Z, 4E, 7E)-2-hydroxy-6-oxonona-trienedioic acid, atrolactic acid, and phenyllactate were significantly regulated.Atrolactic acid, also known as 2-phenyllactic acid commonly found in fermented food, is an intermediate in the conversion of phenylalanine (Valerio et al., 2004).Phenyllactate has been revealed to exert excellent antibacterial activity (Liu et al., 2021).These results indicated that adding ZFM54 during fermentation could significantly increase healthy beneficial function metabolites to improve the nutritive value of yogurt.
What's more, after the volunteers drank yogurt for 14 d continuously, the α diversity index analysis showed that both Chao1 index and Observed Species index of yogurt LS intervention group were observably higher, indicating higher species richness, while yogurt S intervention group had no such effect.The results of PCoA analysis showed that the 2 intervention methods of yogurt S and yogurt LS had little difference in the overall structure of the volunteers' fecal microbial community, in accordance with the concept that short-term dietary interventions do not overcome the dominant inter-individual variation in the healthy intestinal microbiome (Wu et al., 2011).However, short-term yogurt intervention can induce significant changes in the proportion of certain intestinal bacteria of volunteers.
At the Phyla level,drinking yogurt LS could increase the proportion of Firmicutes and decrease Proteobacteria in volunteers.Chronic liver disease and hepatocellular carcinoma were associated with the significant decrease of Firmicutes and the significant increase of Proteobacteria in the intestinal flora system of patients (Chen et al., 2021), suggesting a better beneficial effect of yogurt LS.At the genus level, compared with yogurt S, significant difference analysis showed that drinking yogurt LS specifically increased the relative abundance of Lentilactobacillus, Lacticaseibacillus, Phocea, Phascolarctobacterium, Pediococcus, Ruminococcaceae_unclassified, Alistipes, and Butyricimonas, and specifically decreased Escherichia-Shigella, Acetobacter, Chloroplast_unclassified and Bacillus.It has been reported that the abundance of Lachnospiraceae and Ruminococcaceae decreased significantly in Crohn patients (Humbel et al., 2020).Bifidobacterium have been incorporated as active ingredients in many functional foods due to their healthpromoting properties(O'Callaghan and van Sinderen, 2016).Lentilactobacillus from milk kefir were found to lower serum total cholesterol, low-density lipoprotein (LDL) and triglyceride levels in Sprague Dawley rats fed with a high cholesterol diet (Zheng et al., 2013).The upregulation of Lacticaseibacillus might be the result of adding L. paracasei ZFM54.Lacticaseibacillus has been revealed to inhibit C. difficile, a bacterium which causes diarrhea and colon inflammation in human (Lessa et al., 2015).Pediococcus could help to ameliorate constipation (Huang et al., 2020) and the progression of non-alcoholic fatty liver disease (NAFLD) (Yu et al., 2021).Alistipes is also associated with NAFLD and that patients with NAFLD have a reduced abundance of Alistipes compared with healthy controls (Rau et al., 2018).Phascolarctobacterium is abundantly colonized in human gastrointestinal tract and can produce SC-FAs such as acetate and propionate (Wu et al., 2017).Butyricimonas is a probiotics that can produce SCFAs to reduce inflammation (Xie et al., 2020).Both Phocea and Ruminococcaceae_unclassified are beneficial and produce butyric acid (Sakamoto et al., 2018).Several diseases have been shown to be associated with increased levels of Escherichia-Shigella, such as coronary artery disease (Zhu et al., 2018), autism spectrum disorder (Dan et al., 2020), acute gastroenteritis (Castaño-Rodríguez et al., 2018), as a potential pathogenic agent in human.High abundance of Escherichia-Shigella in the intestinal flora may impair warfarin anticoagulation in patients undergoing cardiac valve replacement (Wang et al., 2020).In addition, most well-studied members of Bacillus are known for their pathogenic potential such as B. anthracis, B. cereus, and B. thuringiensis(Ehling-Schulz et al., 2019).These results indicate that addition of ZFM54 during yogurt fermentation could enhance the regulation of human intestinal flora to gain a better homeostasis.
SCFAs play an important role in intestinal physiology and our results showed that the alteration of SCFAs levels were coincided with the intestinal flora.Yogurt LS consumption significantly enhanced the production of SCFAs (acetic acid, propionic acid, butyric acid) while yogurt L could not.Notably, Alistipes is an acetic acid (Oliphant and Allen-Vercoe, 2019) and propionic acid (Polansky et al., 2015) producer and patients with advanced fibrosis had reduced fecal acetic acid and propionic acid levels.Faecalibacterium and Megamonas, as the most upregulated genera by yogurt LS consumption, expressed acetyl-CoA acetyltransferase and at least one of the 2 additional enzymes necessary for the production of butyric acid, namely 3-hydroxyl-CoA dehydrogenase or alkyl-CoA hydrase (Vital et al., 2014).In addition, Megamonas also produced propionic acid through methylmalonyl-CoA mutase, epimerase, and decarboxylase (Polansky et al., 2015).Moreover, Bifidobacterium and Ruminococcus are the producers of acetic acid (Oliphant and Allen-Vercoe, 2019), Eubacterium is the producer of acetic and butyric acid (Oliphant and Allen-Vercoe, 2019), Bacteroide is the producer of propionic acid (Adamberg et al., 2014), and Subdoligranulum is the producer of butyric acid (Vital et al., 2014).

CONCLUSION
In this study, we evaluate the effect of L. paracasei ZFM54 on improving the flavor and quality of yogurt and how the co-fermented yogurt regulate the human intestinal flora.We obtained the optimal fermentation conditions of L. paracasei ZFM54 compound yogurt: strain ratio, 2:1; inoculation amount, 1.5%; fermentation time, 8 h; fermentation temperature, 42°C; sucrose content, 6.5%.Through SPME-GC-MS and UPLC-Q-TOF-MS, we confirmed that L. paracasei ZFM54 could significantly influence volatile and nonvolatile metabolomic profiles of yogurt to improve many volatile acid contents and alter lots of nonvolatile metabolites, resulting in better flavor and quality.Finally, we compared the effects of yogurt LS and S on the gut microbiota of the volunteers, as well as on the contents of SCFAs in the feces of the volunteers, showing that ZFM54 optimized the flora structure and promoted the production of acetic acid, propionic acid and butyric acid.This work provides theoretical basis and experimental evidence for the commercial application of L. paracasei ZFM54 in improving the quality of yogurt and coordinating intestinal microecology.
Chen et al.: Lactobacillus paracasei ZFM54… Figure 1.Influence of single factor on sensory quality of co-fermented yogurt.(A-B), effect of strain ratio on sensory quality (A) and acidity (B) of yogurt.(C-D), effect of inoculation amount on sensory quality (C) and acidity (D) of yogurt.(E-F), effect of fermentation time on sensory quality (E) and acidity (F) of yogurt.(G-H), effect of fermentation temperature on sensory quality (G) and acidity (H) of yogurt.(I-J), effect of sucrose content on sensory quality (I) and acidity (J) of yogurt.
Figure 1 (Continued).Influence of single factor on sensory quality of co-fermented yogurt.(A-B), effect of strain ratio on sensory quality (A) and acidity (B) of yogurt.(C-D), effect of inoculation amount on sensory quality (C) and acidity (D) of yogurt.(E-F), effect of fermentation time on sensory quality (E) and acidity (F) of yogurt.(G-H), effect of fermentation temperature on sensory quality (G) and acidity (H) of yogurt.(I-J), effect of sucrose content on sensory quality (I) and acidity (J) of yogurt.
Figure 2. Optimization of yogurt fermentation process using response surface test.(A), the interaction between fermentation time and inoculation amount.(B), the interaction between fermentation temperature and inoculation amount.(C), the interaction between fermentation temperature and fermentation time.

Figure 3 .
Figure 3.Total ion flow diagram of volatile components in 2 different yogurt samples.(A), total ion flow diagram of volatile components in commercial starter fermented yogurt S. (B), total ion flow diagram of volatile components in L. paracasei ZFM54 co-fermented yogurt LS.

Figure 4 .
Figure 4. Distribution of yogurt samples based on nonvolatile metabolomic profiles.(A-B), distribution of yogurt LS and S samples in the first 2 principal component analyses under positive ion mode (A) or negative ion mode (B).(C-D), a volcano plot in the comparison of nonvolatile metabolomic profiles between yogurt LS and S samples under positive ion mode (C) or negative ion mode (D).(E), the enriched pathways involved by significantly regulated metabolites when comparing yogurt LS with S. In (C-D), significantly up-and downregulated metabolites (|fold change| ≥ 2, x-axis; log10 P-value ≤ 0.05, y-axis).

Figure 5 .
Figure 5.The heat-map of differentially regulated nonvolatile metabolites of yogurt LS and S samples under positive ion mode (upper) and negative ion mode (down).

Figure 6 .
Figure 6.Bacterial community distribution among various groups at the phylum and genus levels.(A) Distribution of abundant phyla in different groups.(B) Distribution of abundant genera in different groups.N_LS was the non-intervention group of yogurt LS, N_S was the non-intervention group of yogurt S, Y_LS was the intervention group of yogurt LS, Y_S was the intervention group of yogurt S.

Figure 7 .
Figure 7. Difference in bacterial taxa among various groups using LEfSe (LDA Effect Size) analysis.(A-B), the Cladogram (A) and Distribution bar graph (B) of volunteer intestinal flora before and after yogurt S intervention.(C-D), the Cladogram (A) and Distribution bar graph (B) of volunteer intestinal flora before and after yogurt LS intervention.

Figure 8 .
Figure 8. Effects of yogurt LS and S on the contents of major SCFAs in the feces of volunteers.(A-D), concentrations (mg/L) of acetic acid (A), propionic acid (B), butyric acid (C), and isobutyric acid (D) in volunteer feces before or after consuming different yogurts.N_LS was the non-intervention group of yogurt LS, N_S was the non-intervention group of yogurt S, Y_LS was the intervention group of yogurt LS, Y_S was the intervention group of yogurt S. Data are presented as mean ± SD, n = 8 per group.*P < 0.05, **P < 0.01, ns, showing no statistically significant differences.

Table 2 .
The ingredient list of volatile flavor substances in yogurts