In vitro models system
Material
Bifidobacterium animalis subsp. lactis and Lactobacillus casei were purchased from the DSMZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH, Braunschweig, Germany). Both strains were activated at 37°C using an anaerobic condition (Anaerocult A, Merck, Germany). Xanthan gum was obtained from the As Food Co., (Istanbul, Turkiye).
Bacterial Basal Media Preparation and Inoculation
To test the effect of xanthan gum on the growth of B. lactis and L. casei, the basal medium carbohydrate-free, Tryptone Peptone Yeast Extract (TPY) for B. lactis and a De Man Rogosa ve Sharpe (MRS) broth for L. casei were prepared (Table 1).
The both medium were sterilized at 121oC for 15 min. Stock solutions of xanthan gum were prepared in distilled water and 0.45 µm-filter-sterilized (Millipore-Stericup-GP). In order to evaluate the effect of xanthan gum on metabolic activity and viability of B. lactis, based on the TPYcomposition, five different basal media were prepared firstly TPY without carbohydrates (negative control), secondly TPY+ glucose (0.50% w/v; Merck, Germany) and TPY + inulin (0.50% w/v Orafti® HSI, Belgium) as positive controls, and lastly TPY + 0.25% and 0.50% w/v xanthan gum (test substrates). Similarly, to investigate the effect of xanthan gum on L. casei, five different basal media were prepared using by MRS broth which was composed of glucose (0.5% w/v), inulin (0.5% w/v), xanthan gum (0.25 and 0.50% w/v) and without carbohydrate (negative control). 2% (v/v) of overnight probiotic strains, were inoculated into the TPY and MRS broth supplemented with glucose, inulin (0.5%) and xanthan gum (0.25 and 0.50%) as the sole carbon source.
Measurement of pH and optical cell density (OD)
The pH was determined at 12-hour intervals for up to 48 hours during fermentation using a pH-meter (pH 315i / SET; WTW, Germany). The OD was recorded at 600 nm absorbance with a spectrophotometer (Shimadzu UV 1800, Kyoto, Japan) at the 0, 12, 24, 24, 36 and 48th hour of fermentation. The sterile TPY and MRS broth without bacteria were used as blanks for the OD measurements.
Calculation of prebiotic activity score
In this assay, TPY and MRS broth supplemented with 0.50 % of xanthan gum and glucose were prepared for B. lactis and L. casei, respectively. A tryptic soy broth (TSB) supplemented with 0.50 % of xanthan gum and glucose were prepared for Escherichia coli as the representative enteric species. Afterwards, 1% of the inoculum of each strains were added to a TPY, MRS and TSB medium with 0.50% xanthan gum and glucose. Test cultures were incubated at 37°C for 24 hours anaerobically. Samples were taken at the 0 and 24th hour of fermentation, and the cell counts were determined by using TPY agar for B. lactis, MRS agar for L. casei and tryptic soy agar (TSA) for E. coli. Prebiotic activity scores (PAS) were calculated according to Huebner et al. (2008). 
Analytical methods for postbiotics
The postbiotics such as lactic acid and SCFAs were evaluated in the media with glucose (0.50%), inulin (0.50%), xanthan gum (0.25 and 0.50%) and without carbohydrate (control) after 48-hour fermentation. Standars for postbiotics were obtained from Merck, Germany. High Performance Liquid Chromatography (HPLC, Shimadzu Prominence, Japan) equipped in a Diode Array Detector (DAD, SPD-M20A) was used for the determination of lactic acid amount. The HPLC equipment consisted of the Shimadzu CTO-10ASvp Column. Analysis of short-cahin fatty acids was carried out on the Headspace Gas Chromatography (HS-GCMS, Agilent 7697A Headspace, USA). Samples were filtered with a 0.45 μm syringe filter and injected. Temperatures were set 35°C and 150°C for column, 200oC for dedector and 180oC for injector. Relative standard deviation values for the responses of the retention times obtained from standard solution mixtures containing different levels of acid were calculated for the repeatability of the method. Quantity of lactic acid and SCFAs was calculated as g L-1 (Fiori et al. 2018).
Skim milk model system
Culture activation
Freeze dried Bifidobacterium lactis (DANISCO, Sassenage, France) and Lactobacillus casei (Chr-Hansen, Istanbul, Turkiye) were used in milk model system. Reconstituted sterile non-fat milk (12 g 100 mL-1) was autoclaved at 121oC for 15 min for the culture activation. The strains were inoculated in sterile reconstituted skimmed milk and incubated anaerobically at 37±1°C (Anaerobic System Anaerogen, Oxoid, Basingstoke, UK).
Production of skim milk model system with xanthan gum
Skim milk powder was reconstituted in pure water at 10.70% (w/v). Four groups of milk model systems were manufactured by using probiotic strains and xanthan gum; BL (milk with B. lactis), BLG (milk with B. lactis and 0.25% xanthan gum), LC (milk with L. casei) and LCG (milk with L. casei and 0.25% xanthan gum). Reconstituted milk was divided into two experimental batches; one batch of which was supplemented with xanthan gum to final concentrations of 0.25% w/v. Each batch was subjected to pasteurization at 90°C for 10 minutes and subsequently cooled to 37°C. The samples were inoculated with the probiotic strains. The inoculum levels of each strain were determined to give a final concentration of approximately 8–9 log10 CFU mL-1 in milk and then incubated at 37°C until the final pH of 4.7 was achieved. The samples were cooled and stored for 28 days, which were then analyzed for probiotic viability, technofunctional properties (titratable acidity, syneresis, color and textural attributes) at days 1, 7, 14, 21 and 28 of cold storage (4°C.)
Enumeration of probiotic viability in milk model system
B. lactis and Lb. casei counts were enumerated by using Man, Rogosa and Sharpe agar (MRS; Merck, Germany) (Tharmaraj and Shah 2003). The plates were incubated at 37°C for 72 h anaerobically (AnaeroGen Gas Packs, Oxoid, Basingstoke, UK). Plates containing 30–300 colonies were counted and converted into log10 colony forming units (cfu g-1). Viability proportion index (VPI) of microorganisms was calculated according to the following formula:

Evaluation of technofunctional properties in milk model system
Total titratable acidity (TTA) was determined by the titration method and expressed as a percentage of lactic acid equivalents (AOAC 2000). In order to evaluate syneresis, 25 g sample was weighed and then filtered through Whatman paper at 4°C for 2 h. The syneresis was calculated as the weight of collected whey divided by the initial sample weight (Ozcan et al. 2020). Color parameters (L*, a* and b*) of samples were measured by using a Minolta Chromameter (Konica Minolta Co., Ltd., Osaka, Japan). L* parameter ranges from 0 (black) to 100 (white), a* shows from red (+a*) to green (−a*), and b* varies from yellow (+b*) to blue (−b*). Textural characteristics of samples were determined using a texture analyzer TA-XT Plus (Stable Micro System Ltd, Model TA-XT plus, UK) with the back extrusion method fitted with a 5 kg load cell. A 40-mm-diameter cylinder probe was used and penetrated the samples to 75% of their original depth. The speed of the probe was arranged at 0.1 mm/s during the pretest compression and relaxation of the samples. Firmness, consistency, cohesiveness and viscosity indexes were calculated using the Texture Expert Exceed software (v 2.55) extracted from the resulting force time curves (Patrignani et al. 2007).
Statistical analysis
The results obtained from the present study were statistically evaluated using analysis of variance (ANOVA) and Fisher’s least significant difference (LSDs) method (p<0.05 and p<0.01, Minitab 17, USA). Significant differences statistically were indicated using different letters. The hierarchical cluster analysis was performed following an unweighted pair group method with an arithmetic average based on a dissimilarity matrix. In order to analyze the relationship among tehnofunctional properties of skim milk matrices, the Pearson Correlation Coefficient (r) was calculated on individual data using the Statistical software (IBM SPSS Statistics, USA).