Yoghurt-Like Drink Enriched with Lactobacilli and Bifidobacteria ^†

Abstract

To develop an enriched fermented milk, an experimental design was used in the first step by combining Lactiplantibacillus plantarum (isolated from oenological matrix) and Bifidobacterium animalis subsp. lactis strains, at different inoculum levels (4, 5 and 6 log CFU/mL) and temperatures (30, 35 and 40 °C) to study the effects on the acidification. Then, for the realization of the yoghurt-like drink, 5 log CFU/mL of both strains was inoculated in milk + 10% honey at 35 °C for 24 h; the fermented milk was stored at 4 °C for 50 days. The microbial concentrations were always >8 log CFU/mL.

1. Introduction

The creation of a product requires a careful and scrupulous study of all the variables and an evaluation of how these can influence the final product. In the literature, many papers focus on the production of drinks with a probiotic or prebiotic added [1,2,3]. This type of product responds to the constant demand of consumers who are increasingly attentive to the nutritional and health characteristics of foods. Yoghurt is one of the oldest foods, and today, it has become a trendy product, in terms of nourishment and versatility, considered a healthy protein food and a valid alternative to milk [4]. Like other fermented drinks, yoghurt brings several benefits to our body; it has anti-inflammatory and purifying properties, the acidity of the yoghurt favours the growth of an intestinal bacterial biota capable of counteracting the putrefactive phenomena that occur in the intestine. Many functions are performed by the synergistic action of Streptococcus thermophilus and Lactobacillus delbrueckii subsp. Bulgaricus [4]; however, different bacteria (e.g., probiotics) can be added.

The aim of this work was the formulation of a yoghurt-like drink with Lactobacilli and Bifidobacteria, through the use of microorganisms isolated from oenological matrices of the Apulian territory; an experimental design, the Centroid, was used to optimize the product.

2. Materials and Methods

Microorganisms and Milk: Bifidobacterium animalis subsp. lactis DSM 10140 (Deutsche Sammlung von Mikroorganismem und Zellkulturen’s collection, Braunschweig, Germany) and one strain of Lactiplantibacillus plantarum coded as 33 isolated from Apulian (Italy) oenological matrix, were used in this research. The strains were stored at −20 °C in MRS broth (Oxoid, Milan, Italy) and MRS Broth + 0.5% cysteine, for Lactiplantibacillus and Bifidobacterium, respectively, with 33% (V/V) of sterile glycerol (J.T. Baker, Milan, Italy). Before each assay, the bacteria were grown twice in the same medium at 30 (Lpb. plantarum) and 37 °C (B. animalis subsp. lactis) for 24 h; then, the cultures were centrifuged at 4000× g for 10 min at 4 °C, the supernatant was discarded, and the pellet re-suspended in distilled water. Apulian fresh whole pasteurized homogenized cow’s milk (3.2 g/L protein; 4.9 g/L carbohydrates; 3.6 g/L fats) purchased from a local dealer was used.

Experimental design—Centroid: The optimization of the production of the yoghurt-like drink was performed according to a centroid design [5,6]. In this research, the independent variables were the inoculum level of Lpb. plantarum 33 (L) and of B. animalis subsp. lactis (B), and temperature (T) (

Table 1. Combination, coded levels and values of Centroid.

Combinations	Coded Levels			Values
	B *	L **	T ***	B	L	T
A	1	0	0	6	4	30
B	0	1	0	4	6	30
C	0	0	1	4	4	40
D	0.5	0.5	0	5	5	30
E	0.5	0	0.5	5	4	35
F	0	0.5	0.5	4	5	35

* B = Bifidobacterium animalis subsp. lactis, CFU/mL; ** L = Lactiplantibacillus plantarum 33, CFU/mL; *** T = temperature, °C.

).

According to the design, Lpb. plantarum and B. animalis subsp. lactis were inoculated to 4, 5 and 6 log CFU/mL in 100 mL of pasteurized milk supplemented with honey at 10% (V/V); then, the samples were incubated at 30, 35 or 40 °C for 7 days. For each combination, microbiological sampling was carried out and the acidification was periodically monitored through pH measuring using a pH-meter (Crison, Barcelona, Spain). Each analysis was performed in duplicate.

Modeling: Microbial growth data (log CFU/mL), were plotted on an Excel worksheet and processed as mean value ± standard deviation.

Concerning pH, data were modeled as acidification (ΔpH), i.e., pH decrease referred to the beginning of the experiment; then, it was used as a dependent variable for a primary modeling through the “lag-exponential model” of van Gerwen and Zwietering [7] and Baty and Delignette-Muller [8] to obtain α, the time before the beginning of the acidification kinetic (h); d_max is the maximal acidification rate (1/h), and ΔpH_max is the maximum level of acidification.

Successively, ΔpH_max e d_max were used as input values for a multiple regression approach; temperature and inoculum levels of both strains were used as independent variables. The analysis was performed through the software Statistica for Windows (StatSoft, Tulsa, OK, United States). The model was built by using the option “quadratic,” for the evaluation of the individual (“B. animalis subsp. lactis inoculum level”, “Lpb. plantarum 33 inoculum level” and “temperature”) and interactive effects (“B. animalis subsp. lactis inoculum level * Lpb. plantarum 33 inoculum level; “B. animalis subsp. lactis inoculum level * temperature,”; and “temperature ∗ Lpb. plantarum 33 inoculum level”). Finally, the effect of each independent variable on the fitting parameters of the acidification kinetic (ΔpH_max and d_max) was evaluated through the individual desirability functions, as reported by Speranza et al. [3].

Yoghurt-like realization: an amount of 5 log CFU/mL of Lpb. plantarum 33 and B. animalis subsp. lactis was inoculated in a 10 mL sample composed of 9 mL of milk and 1 mL honey and incubated at 35 °C for 24 h; then, the samples were stored at 4 °C. Microbiological analyses and pH measurements were periodically assessed. For microbiological analyses, the following media were used: MRS Agar acidified to pH 5 for Lpb. plantarum 33, and MRS Agar added with a solution of 4 antibiotics (Paromomycin sulphate 0.01 g, Neomycin sulphate 0.005 g, Lithium 0.15 g, Nalidixic acid 0.75 g) for B. animalis subsp. lactis. The experiments were repeated twice on two independent samples [9,10].

3. Results

The microbial strains used, B. animalis subsp. lactis DSM 10140 and Lpb. plantarum 33 were preliminary analysed to evaluate their acidifying capability and subjected to adaptation trials to improve their performances [11]; they were then inoculated in milk + honey at 10% (data not yet published). Successively, the product optimization was realised through an experimental design (Centroid).

3.1. Centroid

Table 2. Acidification of Lpb. plantarum 33 and B. animalis subsp. lactis: fitting parameters of the “lag-exponential equation” (mean ± standard error).

Combinations
	ΔpHmax	dmax	α	R
A	2.38 ± 0.01	0.08 ± 0.00	3.53 ± 0.61	0.999
B	2.24 ± 0.05	0.09 ± 0.00	5.19 ± 2.10	0.994
C	2.20 ± 0.07	0.19 ± 0.02	5.34 ± 2.00	0.980
D	2.34 ± 0.03	0.10 ± 0.00	6.59 ± 1.15	0.996
E	2.59 ± 0.06	0.11 ± 0.01	-	0.995
F	2.52 ± 0.07	0.11 ± 0.00	-	0.994

ΔpH_max = maximum acidification (decrease in pH at the end of the assay); d_max = maximal acidification rate (ΔpH/day); α = the time before the beginning of the acidification kinetic (d); R = regression coefficient.

shows the combinations of the experimental design and the performances of both strains referred to ΔpH_max, d_max and α. ΔpH_max varied from a minimum value of 2.20 (combination C) to a maximum of 2.59 (combination E); the acidification rate (d_max) ranged from 0.08 to 0.19 g⁻¹. The parameter α was statistically significant only for combinations A, B, C and D, ranging from 3.53 ± 0.61 (combination A) to 6.59 ± 1.15 days. The high standard error for B, C and D combinations suggests that acidification could occur even more rapidly.

Successively, ΔpH_max and d_max were used as dependent variables for a multiple regression approach. The first output was the table of the standardized effects, showing the statistical weight and the significance of each individual (inoculum of both microorganisms and temperature) and interactive factors.

Table 3. Standardized effects of the inoculum of B. animalis subsp. lactis DSM 10140 [B], Lpb. plantarum [L] and temperature [T] on the maximum acidification (ΔpH_max) and on the acidification rate (d_max).

EFFECTS	ΔpHmax	dmax
[B]	73.814	12.746
[L]	69.330	13.429
[T]	68.251	28.331
[B] × [L]	- (A)	-
[B] × [T]	7.385	−2.999
[L] × [T]	7.614	−3.275
R2 (B)	0.901	0.930

^(A) Not significant; ^(B) determination coefficient adjusted for the multiple regression.

shows the standardized effects of the three variables on ΔpH_max. As individual term, B. animalis showed the greatest effect (73.814), followed by Lpb. plantarum 33 (69.330) and temperature (68.251), while the interaction “Lpb. plantarum-temperature” was the must significant, followed by “B. animalis subsp. lactis-temperature”. Concerning d_max, the maximum effect was attributable to the temperature followed by the level of inoculum of Lpb. plantarum and B. animalis strains. The “Lpb. plantarum–temperature” and “B. animalis subsp. lactis –temperature” interactions were also significant.

Other outputs of centroid are the ternary plots; Figure 1a,b show the triangular surface for the effects of B. animalis subsp. lactis (B), Lpb. plantarum (L) and temperature (T) on the maximum acidification (ΔpH_max) (Figure 1a) and on the acidification rate (d_max) (Figure 1b). The model predicted the maximum acidification (ΔpH_max = 2.5) when either Lpb. plantarum or B. animalis and temperature were at the coded levels 0.5 (that is the microorganisms at 5 log CFU/mL and temperature at 35 °C) (Figure 1a). Concerning d_max (Figure 1b) the model predicted the highest values (d_max = 0.18) at the coded level 1 of the temperature (40 °C).

Triangular surfaces provide a quantitative assessment of interactive effects, but not of each individual term. This type of evaluation is possible through the desirability profile.

The desirability profile of ΔpH_max (Figure 2a) shows that both for B. animalis subsp. lactis and Lpb. plantarum, the increase in the inoculum level negatively affected acidification with a decreased performance of the strains at coded levels 0.75 or 1 (corresponding to 5.50 and 6 log CFU/mL). Concerning temperature, ΔpH_max was minimal at 30 °C and also decreased at coded levels 0.75 and 1 (corresponding to 37.5 and 40 °C).

The desirability profile of the acidification rate (d_max) (Figure 2b) shows that both B. animalis subsp. lactis and Lpb. plantarum 33 reduced the acidification rate at the inoculum coded level 1 (corresponding to 6 log CFU/mL). With regard to temperature, a positive correlation was observed: as temperature increased (for values coded as 0.75 and 1 = 37.5 and 40 °C, respectively), d_max increased.

As shown by the desirability profiles (Figure 2a,b), the temperature acted differently on the maximum acidification and on the rate: in relation to the ΔpH_max, the increase in temperature led to a reduction in acidification; instead, in the case of d_max, the increase in temperature was positively correlated with acidification rate. This phenomenon known as parameters uncoupling [12] highlights that increasing the temperature increases the acidification rate; however, temperature cannot reach the maximum values because a reduction in acidification capability could occur.

3.2. Yoghurt-Like Drink

The main result of optimization of Centroid design was the best combination for obtaining the final product; as a result of modelling, the conditions were identified by the software as follows: 5 log CFU/mL of Lpb. plantarum 33 and of B. animalis subsp. lactis 10,140, and temperature at 35 °C for 24 h. The concentration of both strains reached > 8 log CFU/mL. Then, the samples were stored at 4 °C for 50 days and both strains were able to survive and never resulted below the criticality threshold (7 log CFU/mL) (

Table 4. Cellular concentration (log CFU/mL) Lpb. plantarum 33 and B. animalis subsp. lactis DSM 10140 inoculated into the yoghurt-like drink and stored at a temperature of 4 °C. Lowercase letters indicate statistically significant differences along the column (effect of time for each strain); capital letters refer to the differences between the 2 strains at the same time of analysis (one-way ANOVA and Tukey’s test, p < 0.05). Data are expressed as mean values ± standard deviation.

Time (Days)	33	10,140
0	8.78 ± 0.00 a,A	8.40 ± 0.00 a,B
1	8.74 ± 0.01 a,A	8.90 ± 0.02 b,A
4	8.90 ± 0.03 a,A	8.85 ± 0.00 b,A
6	8.93 ± 0.02 a,A	8.72 ± 0.03 b,A
19	8.88 ± 0.03 a,A	8.78 ± 0.02 b,A
26	8.85 ± 0.00 a,A	8.88 ± 0.03 b,A
33	8.52 ± 0.02 b,A	8.60 ± 0.00 a,A
40	8.74 ± 0.01 a,A	8.88 ± 0.02 b,A
50	8.74 ± 0.04 a,A	8.88 ± 0.03 b,A

). Furthermore, the pH did not undergo significant changes during storage (initial pH = 4.84; final pH = 4.22). Finally, the texture, odor and color of the yoghurt-like drink were optimal.