Impacts of milk fat globules on the physicochemical properties of stirred yogurt

Abstract

Milk proteins play an essential role in yogurt manufacture, by adsorbing onto the fat globule interface. This study aims to clarify the impacts of milk fat globules on the physicochemical properties of stirred yogurt, by modifying the particle size distribution of fat globules in reconstituted milk. The results show that high-pressure homogenization increased the association between milk proteins and milk fat globules. The zeta potential and protein surface hydrophobicity of reconstituted milk samples increased. Stirred yogurt prepared with reconstituted milk homogenized at high pressure (≥ 20 MPa) exhibited a fine mesoscopic structure with higher water holding capacity and apparent viscosity. Based on transmission electron microscopy and atomic force microscopy, the milk fat globules tended to form effective connections with casein micelles, giving the yogurt a finer network structure and higher viscosity. These findings reveal that controlling milk fat globule size can effectively improve stirred yogurt quality.

Introduction

Yogurt is a popular fermented dairy product made by adding specific strains of bacteria, such as Streptococcus thermophilus and Lactobacillus bulgaricus, to milk (Herve-Jimenez et al., 2009). The bacteria ferment the lactose in the milk, creating lactic acid, which thickens the milk and gives yogurt its characteristic tangy flavor (Hati et al., 2013). Furthermore, the bacteria in yogurt produce beneficial compounds such as prebiotics, which can aid digestion and boost the immune system (Kopp-Hoolihan, 2001). Due to these health benefits and its taste, yogurt is widely consumed worldwide. Consumption of yogurt has been linked to a range of health benefits, including a reduced risk of cardiovascular disease (CVD), type 2 diabetes (T2D), and less weight gain over time (Savaiano and Hutkins, 2021). Furthermore, because yogurt is a nutrient-dense food with low energy density, it is thought to be associated with healthy dietary patterns and lifestyles (Tremblay and Panahi, 2017). It is important to remember that whole-fat yogurt contains 3–5 % fat, like other whole-fat dairy products, which can be considered high for people with certain medical conditions, such as those with cardiovascular disease (Kratz et al., 2013). Non-communicable diseases (NCDs) account for 71 % of all global deaths annually ⁱ. Consequently, many consumers choose functional foods, such as low-fat yogurt, to improve their health. However, it has been reported that reducing fat content in dairy products leads to poor food texture and low acceptability. Specifically, syneresis, viscosity changes, and taste deterioration are common defects in yogurt with reduced fat content (Lee and Lucey, 2004; Routray and Mishra, 2011; Laguna et al., 2017).

Yogurt is typically produced by fermentation using lactic acid starter cultures to reach the isoelectric point of casein micelles at a pH of 3.8–4.6 (Nakamura et al., 2006). Commercial yogurt can generally be classified into two types based on its manufacturing process: 1) set yogurt, which is incubated and cooled in its package, and 2) stirred yogurt, which is incubated in tanks and stirred before packaging (Haque et al., 2001). Various strategies are utilized to address the defects in food texture, such as syneresis, including fortifying the milk base with additives and stabilizers (i.e., skim milk powder, gelatin) as well as modifying processing conditions (i.e., homogenization, heat treatment); these are effective to a limited extent (Arab et al., 2022). In summary, it is difficult to produce reduced-fat yogurt that is as good as whole-fat yogurt in terms of sensory and physicochemical properties.

Homogenization (typically at 15–20 MPa) is a process that is often performed before fermentation to improve the firmness and consistency of yogurt and reduce whey separation (Sodini et al., 2004). Physical processes such as homogenization can reduce fat globule size and promote the adsorption/incorporation of casein micelles into the fat globules; this improves yogurt texture by increasing the extent of interaction between fat globules and proteins (Tribst et al., 2020a; Tribst et al., 2020b). However, by destroying the milk fat globule membrane (MFGM) under high-shear conditions, homogenization changes the interface structure, generating a new fat globule interface (Walstra, 1983). Milk proteins in the serum phase (mostly casein) then coat the newly exposed fat globule interface, reducing its interface-free energy. Obeid et al. found that the adhesion between casein-coated fat globules and casein micelles increased after homogenization using an atomic force microscope, which results in increased milk acid gel stiffness (Obeid et al., 2019). Adsorption of milk proteins, especially casein, onto the fat globule interface plays an important role in yogurt manufacture, linking fat globules to the physicochemical properties of yogurt.

This study aims to clarify the impacts of milk fat globules on the physicochemical properties of stirred yogurt. In this study, we hypothesize that milk fat globules influence the development of the yogurt network structure through milk proteins adsorbing fat globules, thereby affecting the physicochemical properties of stirred yogurt. Concretely, milk fat was dispersed in reconstituted milk prepared by skim milk powder and sucrose. The size of fat globules was controlled by the homogenization pressure, to investigate the impact of fat globule size and shape. To evaluate the effects of fat globules on reconstituted milk, particle size distribution, zeta potential, and protein surface hydrophobicity of reconstituted milk were measured. Syneresis, rheological properties, and mesoscopic structure were investigated to evaluate the physicochemical properties of yogurt.

Materials and Methods

Materials For yogurt production, skim milk powder (Meiji skim milk powder, Meiji Co., Ltd., Tokyo, Japan), lactic acid bacteria culture (Freeze-dried non-ropy-producing yogurt culture, Streptococcus thermophilus, and Lactobacillus bulgaricus, YC-X11 YoFlex, Chr. Hansen A/S, Hørsholm, Denmark) and sucrose (FUJIFILM Wako Pure Chemical Corp., Osaka, Japan) were used. A beaker with about 50 g of unsalted butter (Meiji butter, Meiji Co., Ltd.) was placed in a water bath (NTT-1200, Tokyo Rikakikai Co., LTD., Tokyo, Japan) and held at 60 °C for 20 min. The melted milk fat on the upper layer was collected for further processing. Fluorescein isothiocyanate (FITC; Merck KGaA, Darmstadt, Germany) and Nile red (FUJIFILM Wako Pure Chemical Corp., Osaka, Japan) were used as fluorescent staining dyes. Lead citrate, lead nitrate, lead acetate, and sodium citrate were used to constitute the lead stain solution for transmission electron microscopy, according to Sato (1968). Commercial milk (Noujou Gyunyu, Yuda Milk Corp., Iwate, Japan) purchased at the supermarket was used as a reference in the particle size distribution measurement.

Preparation of reconstituted milk The preparation of reconstituted milk and yogurt was according to Mutahi and Miura (Mutahi and Miura, 2020).Skim milk powder (12 % w/w) and sucrose (6 % w/w) were mixed with deionized water to prepare reconstituted skim milk. Subsequently, reconstituted skim milk was transferred to a water bath set at 60 °C, stirred by a magnetic stirrer, and added to 3 % (w/w) liquid milk fat. Pre-homogenization to disperse the milk fat was conducted at 1,500 rpm using a high-speed mixer (T.K. AUTO HOMO MIXER, PRIMIX Corp., Awaji, Japan) at 70 °C for 15 min. This was followed by an eight-pass homogenization process at 70 °C using a two-stage valve homogenizer (APV LAB1000, SPX FLOW, Inc., Tokyo, Japan). Five pressure levels were used for homogenization to obtain reconstituted milk samples with different fat globule size distributions (Table 1). After homogenization, the reconstituted milk was held at 90 °C for 5 min for sterilization using a circulating water bath (HAAKE DC10, Thermo Fisher Scientific, Inc., Waltham, USA), then cooled immediately by an ice bath. The prepared reconstituted milk was stored at 5 °C for further tests.

Table 1. Physicochemical properties of reconstituted milk.

Homogenization		Mean volume	Zeta potential	F max	Kd
pressure [MPa]		diameter	[mV]	[-]	[µM]	PSH
1st stage	2nd stage	[µm]
5	<1	1.420 ± 0.026a	−25.91 ± 0.74a	235.49 ± 20.83a,b	53.81 ± 15.74a	4.59 ± 1.12a
10	1	0.642 ± 0.027b	−28.58 ± 0.97b	247.80 ± 23.26b	47.80 ± 20.34a	5.62 ± 1.55a,b
15	>1, <2	0.397 ± 0.004c	−29.23 ± 0.55b	192.78 ± 2.09a	25.60 ±1.46a	7.54 ± 0.36a,b
20	2	0.336 ± 0.006d	−32.27 ± 0.12c	246.31 ± 3.46b	47.31 ± 3.92a	5.24 ± 0.53a,b
25	>2, <3	0.284 ± 0.007e	−28.56 ± 1.05b	239.07 ± 21.67b	31.65 ± 7.89a	7.80 ± 1.47b

Mean ± Standard deviations (n = 3)

F_max, maximum of fluorescence intensity; K_d, dissociation constant PSH, protein surface hydrophobicity. Values followed by the same superscript letter in the same column represent no significant difference at p < 0.05.

Preparation of stirred yogurt Reconstituted milk was mixed with 0.015 % (w/w) lactic acid bacteria culture and incubated at 43 °C for 5 h using a low-temperature water bath (NCB-2100, Tokyo Rikakikai Co., Ltd.). Afterward, yogurt was transferred to a water bath, holding at 72 °C for 10 min to ensure inactive bacteria. The yogurt was then cooled in an ice bath, and a glass rod was used to destroy the curd structure. Next, a mixer (MAZELA NZ-1000, Tokyo Rikakikai Co., Ltd.) with an anchor stirrer (Teflon, W 60 × H 40 mm, T–1A, Tokyo Rikakikai Co., Ltd.) was used at 500 rpm for 5 min to produce the stirred yogurt. Stirred yogurt was stored at 5 °C until measurement.

Measurement of particle size distribution The particle size distribution of reconstituted milk was measured using a laser diffraction and scattering size analyzer (LS-230, Beckman Coulter Inc., Brea, CA, USA) at room temperature for 90 s. Before measurement, the flow cell was washed and filled with deionized water to obtain polarization intensity differential scattering (PIDS) < 15 %. The sample was then added until PIDS reached 45–55 %.

Measurement of zeta potential The zeta potential measurement of reconstituted milk was measured using a zeta potential and particle size analyzer (ELSZ-1000ZS, Otsuka Electronics Co., Ltd., Osaka, Japan) at 25 °C. A standard flow cell unit was used and filled with the sample, diluted 20-fold with deionized water to obtain a suitable concentration for the test.

Measurement of protein surface hydrophobicity (PSH) The protein surface hydrophobicity of reconstituted milk was determined following the method described by Bonomi et al. (1988), using 8-anilino-1-naphthalene sulfonic acid (1,8-ANS, λ_ex = 390 nm, λ_em = 480 nm, MP Biomedicals, Inc., Santa Ana, CA, USA) as the fluorescent probe. A total of 10 % (w/w) reconstituted milk samples diluted with 50 mM sodium phosphate buffer solution (pH 6.8) were mixed with 1,8-ANS at final concentrations of 12.5, 25, 50, 100, and 200 µM for 15 min in the dark. A fluorescence spectrophotometer (RF-5300, Shimadzu Corp., Kyoto, Japan) was used to determine the relative intensity of the fluorescent signal. The titration curves were fitted using OriginPro 2020 (OriginLab Corp., Northampton, MA, USA). Sample PSH was calculated as follows:

  

where F_max represents the maximum fluorescence intensity and K_d represents the concentration of 1,8-ANS at which the fluorescence intensity is half of F_max.

Transmission electron microscopy (TEM) Reconstituted milk microstructure was observed via TEM (JEM-2100, JEOL Ltd., Tokyo, Japan) at 20 kV. The reconstituted milk was encapsulated in a 3 % (w/w) agar gel capsule following Kalab (Kalab, 1988). For sample preparation, reconstituted milk in an ager capsule was chemically fixed overnight using 2.5 % (v/v) glutaraldehyde, then fixed with 2 % (w/w) osmium tetroxide for 1 h, both in 0.1 M sodium phosphate buffer solution (pH 7.4). Next, the samples were dehydrated using an increasing ethanol series (50 %, 70 %, 80 %, 90 %, 95 %, 100 %, 100 %, and 100 %, v/v) for 15 min at each concentration, replacing the void with epoxy resin. Before observation, the samples were cut into 100 nm slides using a microtome (Ultra Cut S, Leica Microsystems GmbH, Wetzlar, Germany) and stained successively with 2 % (w/w) uranium acetate for 15 min and lead stain solution for 3 min.

Atomic force microscopy The surface structure of the fat globule was observed using an atomic force microscope (NX-10, Park Systems Corp., Suwon, South Korea). The sample preparation followed Obeid et al. (Obeid et al., 2019). Reconstituted milk samples were mixed with 50 % (w/w) sucrose solution at a ratio of 2:1 (w:w) at 50 °C to adjust the density of the sample. Thereafter, 5 mL of sample mixed with sucrose solution was injected gently into the bottom of 50 mL centrifuge tube with 20 g of 5 % (w/w) sucrose solution, to ensure an interface between the sample and the clear solution. Centrifugation was performed at 2,600 × g for 15 min at 20 °C. The cream layer was collected and coated onto a glass slide, then stored in a desiccator for >24 h at 20–25 °C before observation. Image processing and statistical analysis were completed using XEI image analysis software (Park Systems Corp.).

Syneresis Approximately 20 g of stirred yogurt sample was added to a 50 mL centrifuge tube and centrifuged at 100 × g for 15 min at 5 °C. The syneresis of stirred yogurt sample was determined as follows:

  

Rheological properties

Flow curve The flow curves of stirred yogurt were determined at 5 °C using a rotational rheometer (ARES-G2, TA Instruments Inc., New Castle, DE, USA) with cone-plate geometry (Φ 25 mm, cone angle of 1 deg, cutting tip of 0.023 mm). For the flow ramp test, the shear rate was increased from 0.1 s⁻¹ to 200 s⁻¹ linearly over 100 s, then immediately reduced from 200 s⁻¹ to 0.1 s⁻¹ linearly over 100 s. The relationship between shear rate ( ) and shear stress (σ) was characterized using Origin 2020 (OriginLab Corp.) with a power law flow equation ( , where k is the consistency constant, and n is the flow behavior index). Given that yogurt was known as a kind of thixotropic fluid (Vélez-Ruiz, 2019), we evaluated hysteresis, the temporal change in yogurt structure, by calculating the thixotropic index (TI), the ratio of the loop area encircled by flow curve (A_hys) and the integral of the upward flow curve (A_up).

Dynamic viscoelasticity The dynamic viscoelasticity of stirred yogurt was determined at 5 °C using the rheometer described in the above section of flow curve. For the oscillation-strain test, angular frequency (ω) was set at 6.28 rad·s⁻¹ (1 Hz), and shear strain was changed from 1.0 × 10⁻⁴ to 2.5 to obtain the linear viscoelastic strain range for each sample. For the oscillation-frequency test, 80 % of the maximum strain of the linear viscoelastic range was loaded onto the sample to test in its linear viscoelastic range, and the angular frequency was changed from 0.1 to 100 rad·s⁻¹ (1.59 × 10⁻³ to 15.9 Hz). The storage modulus (G′), loss modulus (G″), and loss tangent (tan δ = G″/ G′), obtained in the oscillation-frequency test, were used to characterize the dynamic viscoelasticity of the samples. To characterize the behavior of G′ as angular frequency increases, the data were fitted using a power-law equation (G′ = Aω^B, where A is the critical constant representing gel strength and B is the relaxation index) (Gabriele et al., 2001).

Confocal laser scanning microscopy To observe the mesoscopic structure of stirred yogurt, a confocal laser scanning microscope system (C2si, Nikon Corp., Tokyo, Japan) equipped with an inverted microscope (Nikon Eclipse Ti, Nikon Corp.) was used. For fluorescent dye labeling, FITC (Fluorescein isothiocyanate, λ_ex = 488 nm, λ_em = 525 nm, 1 mg/mL in ethanol) was added to stain proteins at 1 % (v/v, dye solution/yogurt).). Nile red (λ_ex = 561 nm, λ_em = 595 nm, 1 mg/mL in ethanol) was added to stain lipids at 1 % (v/v, dye solution/yogurt). Then, for each sample, micrographs (H 127.28 µm × W 127.28 µm × T 5 µm) were taken from three different observation regions, using the image analysis plugin BoneJ2 in ImageJ 1.52p (National Institutes of Health, USA) to calculate the average particle volume ( , where i is the number of object particles from 1 to N, is the volume of particle i, and N is the total number of particles); specific surface area ( , where and is the surface area of particle i); volume fraction ( , where V_Total is the volume of the observation region (81 000 µm³ in this study)); and the fractal dimensions (D_f) of the protein and lipid. The fractal dimension was calculated using the box-counting algorithm (Fazzalari and Parkinson, 1996).

Statistical analysis IBM SPSS Statistics Ver. 27 (IBM Corp., USA) was used for statistical analysis. Analysis of variance (ANOVA) followed by Tukey’s HSD test (p < 0.05) was performed for comparison of the means between sample groups. Statistical significance was based on a significance threshold of p < 0.05.

Results and Discussion

Particle size distribution of reconstituted milk Three peaks were observed in the particle size distribution of the reconstituted milk samples (Fig. 1). Single peak at 0.1 µm in the particle distribution of reconstituted skim milk probably reflects the particle size distribution of casein micelles. Like reconstituted skim milk, a peak of casein micelles at 0.1 µm could be found in reconstituted milk samples. The highest peak obtained for reconstituted milk (except for the sample homogenized under 5 MPa) was at 0.5 µm; this may be the primary peak for milk fat globules. However, in the sample homogenized under 5 MPa, the highest peak was at 2 µm, reflecting large milk fat globules in reconstituted milk. The peak areas of the peaks at 0.1 and 0.5 µm increased as homogenization pressure increased, while the area of the peak at 2 µm decreased until the peak vanished. There was a negative correlation between reconstituted milk mean volume diameter and homogenization pressure (Table 1): as homogenization pressure increased, the number of large milk fat globules (∼ 2 µm) decreased, whereas the number of fat globules close to 0.5 µm increased rapidly. This suggests that, in reconstituted milk, the fat globule distribution changes discontinuously and the lipid-protein complex structure changes with increasing homogenization pressure.

Fig. 1

Particle size distribution of milk samples: commercial milk for reference (dashed line), reconstituted skim milk (solid line), reconstituted milk with milk fat homogenized at pressure of 5 (square), 10 (diamond), 15 (triangle), 20 (cross), and 25 MPa (circle).

Zeta potential of reconstituted milk In raw milk, casein micelles have a negative charge (−20 mV) and are dispersed in the water phase by electrostatic interactions (McMahon and Brown, 1984). For our samples, the apparent zeta potential exceeded −20 mV in absolute value (Table 1). In reconstituted milk homogenized at 5 MPa, the zeta potential was −25.91 ± 0.74 mV, close to the value for casein micelles (McMahon and Brown, 1984). As the homogenization pressure changed from 5 to 20 MPa, the apparent zeta potential decreased to -32.27 ± 0.12 mV. It is considered that the specific surface area of milk fat globules increases as their particle size decreases, providing more binding sites to associate with casein micelles and other milk proteins which decreased the apparent zeta potential of samples. However, for reconstituted milk homogenized at 25 MPa, the absolute value of apparent zeta potential was 28.56 ± 1.05 mV, and significantly lower than that generated at 20 MPa; this could relate to the adsorption state of casein micelles onto the milk fat globule interface, or to the changes in casein micelle structure.

Protein surface hydrophobicity It was known that heat treatment of milk induced the denatured whey proteins, which can self-aggregate interact with the casein micelle as well as fat globule membrane and contribute to an enhanced surface hydrophobicity (Wiking et al., 2022). The reconstituted milk with fat globules showed higher PSH than reconstituted skim milk (Fig. 2). F_max reveals the extent of potential hydrophobic interactions in the sample. For most of the reconstituted samples, F_max was higher than for the reconstituted skim milk samples, indicating that the milk protein was exposed to a structure with more hydrophobic regions due to the milk fat globules (Table 1). However, the F_max of reconstituted milk homogenized at 15 MPa was lower than that of the other samples. This suggests that complexation of casein micelles and fat globules shifted to another state, thereby reducing F_max. There was no significant difference between samples in K_d, which reflects the strength of hydrophobic interactions between 1,8-ANS and milk protein in reconstituted milk. PSH tended to increase as the homogenization pressure increased, which was beneficial to form a strong milk protein gel network after fermentation. It was possible that homogenization led to more available binding sites for denatured whey proteins on the membrane of the fat globules and therefore, the sample produced by high homogenization pressures showed higher values of PSH.

Fig. 2

Fluorescence intensity changes as a function of 1,8-ANS concentration added in reconstituted milk: skim milk (circle, open) and milk with milk fat homogenized at pressure of 5 (square), 10 (diamond), 15 (triangle), 20 (cross), and 25 MPa (circle, close). Milk samples diluted with pH 6.8 phosphate buffer solution and fitting curves were shown as solid lines.

Microstructure of reconstituted milk In the TEM images, proteins stained with heavy metals appear as dark patches while lipids appear as bright patches (Fig. 3). There were many dark spheres, which we considered to be casein micelles. In the reconstituted milk, most of the casein micelles were observed with a diameter approximately 200 nm. Similarly, many of the fat globules had a diameter close to 200 nm. The images reveal the protein layer, including casein micelles, adsorbed onto the interface of the milk fat globules.

Fig. 3

Microstructure of reconstituted milk homogenized at pressure of 5 (A), 10 (B), 15 (C), 20 (D) and 25 MPa (E) observed by transmission electron microscope, and typical fat globules were shown in (F) to (H). CM represents casein micelles, and FG represents fat globules, respectively. Scale bars represent 200 nm in (A, B, C, D, E, G and H) and 1 000 nm in (F).

Based on the TEM images of the reconstituted milk samples, the interactions between milk fat globules and casein micelles can be summarized as follows: a) for fat globules much larger than the average casein micelle, the fat globules were deformed spheres with a jagged surface, and many casein micelles were observed associating with a single fat globule; b) for fat globules close in size to the average casein micelle, fat globule shape was closer to spherical, and few casein micelles were associated with the same fat globule; c) for fat globules much smaller than the average casein micelle, the fat globules were almost perfectly spherical, and several fat globules were associated with the same casein micelle. As the homogenization pressure increases, the number of large fat globules decreases and that of small fat globules increases, causing the specific interface area of the fat globules to increase; this causes the principal interaction between fat globules and the associating casein micelles to shift from interaction type a to type b. Furthermore, those casein micelles associating with fat globules via interaction type b were deformed, suggesting that casein micelle deformation increases PSH. Fat globules associated with casein micelles through interaction type c were rarely observed; however, the number of small fat globules increased as homogenization pressure increased (at >15 MPa), thus potentially reducing the number of casein micelles associated with fat globules and causing a lower zeta potential in reconstituted milk homogenized at 25 MPa condition.

Surface structure of milk fat globules Many spherical fat globules with a diameter of 0.2 to 2 µm were observed (Fig. 4), consistent with the results described in Particle size distribution of reconstituted milk. Moreover, finer protrusions (Φ < 0.2 µm) were observed on the surfaces of the milk fat globules; these were considered to be milk proteins adsorbed onto the milk fat globules. Phase mode, in which the phase lag is measured, is useful for evaluating interactions between the probe and sample surface (i.e., harder surfaces and those with less adsorbate have a smaller lag) (Ruggeri et al., 2019). While milk fat comprises various types of fatty acids, it typically melts fully at 40 °C and has a high solid-fat content at 20 °C. For the current samples at 20 to 25 °C, the milk fat tended to crystallize. Therefore, the milk fat globule surface showed a small phase lag, as shown in our atomic force microscopy images.

Fig. 4

Surface structure of fat globules of reconstituted milk homogenized at pressure of 5 (A), 10 (B), 15 (C), 20 (D), and 25 MPa (E) observed by atomic force microscope scanning forward in a 10×10µm² region. Left column shows the height difference, and right column shows the phase lag angle images.

Interestingly, the bright area, showing a large phase lag, increased with homogenization pressure (Fig. 4, right). The increase in the phase lag of the milk fat globule surface structure could be due to an increase in protein adsorption as the homogenization pressure increases. In contrast, the melting point of milk fat globules may be lower following higher-pressure homogenization, because of their smaller mean diameter, as described by the Gibbs-Thomson equation (Alcoutlabi and McKenna, 2005). Based on this, the higher homogenization pressure may cause more milk protein to adsorb strongly onto the globule’s interface, increasing phase lag of sample.

Water holding capacity The syneresis of the stirred yogurt prepared from reconstituted milk and milk fat was significantly lower than that of yogurt prepared using reconstituted skim milk (Fig. 5). As the homogenization pressure increased, syneresis showed a decreasing tendency. A protein network with high water holding capacity forms effectively when milk fat is dispersed in milk as small globules.

Fig. 5

Syneresis of stirred yogurt samples (n = 3). Dotted line on the top represents the syneresis of stirred yogurt sample prepared with reconstituted skim milk. Dots with same letter represent no significant difference at p < 0.05.

Rheological properties

Flow behavior The stirred yogurt samples exhibited shear thinning flow behavior (Fig. 6), in which the apparent viscosity decreases as the shear rate increases, reflecting the dissociation of aggregates during shearing. Interestingly, shear-stress overshoot was observed at 10 s⁻¹ in the upward flow curve for all samples. It is likely that the aggregates of the stirred yogurt are reconstructed under low-shear rates, causing a rapid growth on shear stress. However, these reconstructed structures dissociate rapidly due to excessive shear stress as the shear rate keeps increasing. The power-law equation was used to fit the upward flow curve of the stirred yogurt: the consistency constant, reflecting the viscosity of samples under low-shear conditions, tended to increase with homogenization pressure, while no significant difference was observed between samples in the flow behavior index (Table 2). The loop area of the flow curves indicates that hysteresis occurred after shearing, tending to increase with homogenization pressure. In contrast, the thixotropy index, here defined as the ratio of the flow curve loop area to the area under the upward flow curve, showed no significant differences between samples. Therefore, for reconstituted milk, homogenization pressure mainly influenced the apparent viscosity of the stirred yogurt: as the homogenization pressure increased, the consistency of the yogurt improved, and there was greater aggregate dissociation during shearing.

Fig. 6

Flow behavior of stirred yogurt prepared from reconstituted milk homogenized at pressure at 5, 10, 15, 20, and 25 MPa.

Table 2. Flow properties of stirred yogurt.

Homogenization pressure [MPa]		k [Pa·sn]	n [-]	Hysteresis [Pa·s−1]	TI [%]
1st stage	2nd stage
5	<1	11.9 ± 0.79a	0.229 ± 0.0318a	2472.3 ± 521.78a	37.9 ± 5.54a
10	1	11.8 ± 3.17a	0.271 ± 0.0496a	3277.0 ± 349.28a,b	43.1 ± 3.56a
15	>1 <2	13.9 ± 2.81a,b	0.238 ± 0.0403a	3424.7 ± 518.36a,b	43.8 ± 5.39a
20	2	18.2 ± 12.30a,b	0.250 ± 0.0901a	4099.4 ± 1415.30a,b	41.8 ± 4.79a
25	>2, <3	28.6 ± 2.82b	0.167 ± 0.0084a	5398.2 ± 835.46b	45.4 ± 1.11a

Mean ± Standard deviations (n = 3)

k, consistency coefficient; n, flow behavior index; TI, thixotropic index.

Values followed by the same superscript letter in the same column represent no significant difference at p < 0.05.

Dynamic viscoelasticity The oscillation-strain test revealed that the samples had a linear viscoelastic limit close to 0.003 with no significant differences between samples (Fig. 7). As homogenization pressure increased, G′ and G″ tended to increase. G′ was significantly higher in samples prepared from reconstituted milk homogenized at 25 MPa than at other pressures.

Fig. 7

Storage modulus (G′, close) and loss modulus (G″, open) of stirred yogurts made from different reconstituted milk homogenized at pressure at 5 (a), 10 (b), 15 (c), 20 (d), and 25 MPa (e) as a function of shear strain.

The oscillation-frequency test revealed all of the samples to be viscoelastic solids with tan δ ≈ 0.3. There were no significant differences between samples in terms of G′ and G″ (Fig. 8). When sample viscoelasticity was evaluated using the power-law equation, all of the samples exhibited relaxation indexes close to 0.16, indicating that the stirred yogurt was a simple weak gel with a continuous distribution of relaxation modes (Table 3). The critical constant, which is related to gel strength, tended to increase with homogenization pressure. However, the increase was not significant. This indicates that the gel network connectivity increased by incorporating more fat globules. Dynamic viscoelasticity did not differ significantly between samples, because stirring process caused the yogurt aggregates to dissociate into small fragments.

Fig. 8

G′ (close) and G″ (open) of stirred yogurts made from different reconstituted milk homogenized at pressure of 5 (a), 10 (b), 15 (c), 20 (d), and 25 MPa (e) as a function of angular frequency.

Table 3. Rheological properties of stirred yogurts made from different reconstituted milk.

Homogenization pressure [MPa]		γ0	A	B
1st stage	2nd stage	[×10−3]
5	<1	3.03 ± 0.994a	80.01 ± 18.298a	0.152 ± 0.0085a
10	1	2.93 ± 0.374a	74.14 ± 16.832a	0.166 ± 0.0144a
15	>1, <2	3.16 ± 1.606a	110.69 ± 28.611a	0.156 ± 0.0025a
20	2	3.48 ± 0.845a	99.96 ± 21.002a	0.159 ± 0.0049a
25	>2, <3	2.72 ± 0.374a	115.25 ± 20.280a	0.155 ± 0.0033a

γ₀, limit of linear viscoelastic range; A, critical constant; B, relaxation exponent.

Mean ± Standard deviations (n=3)

Values followed by the same superscript letter in the same column represent no significant difference at p < 0.05.

Mesoscopic structure The milk protein aggregates in stirred yogurt were irregular (Fig. 9). Fat globules were observed on the protein aggregate surface and inside the aggregates. For yogurt homogenized at 5, 15, and 25 MPa, the protein aggregate network structure became finer as homogenization pressure increased. In contrast, at 10 and 20 MPa, the network structure was denser than that at the other pressures, possibly due to the pressure used during the second stage of homogenization.

Fig. 9

Mesoscopic structure of stirred yogurts made from different reconstituted milk homogenized at pressure of 5 (A), 10 (B), 15 (C), 20 (D), and 25 MPa (E) observed by confocal laser scanning microscope with a 100x oil objective lens. Red represents protein, and green represents lipid, respectively. Scale bars represent 25 µm.

The protein aggregate volume ranged from 20.23 to 34.07 µm³, with no significant difference between the samples (Table 4). Interestingly, the volume fraction of protein aggregates was negatively associated with syneresis. This indicates that the volume fraction of protein is an efficient parameter for evaluating the water holding capacity of yogurt.

Table 4. Mesoscopic structural properties of stirred yogurt prepared with reconstituted milk homogenized at different pressure.

Homogenization pressure [MPa]			Protein aggregate				Fat globule
1st stage	2nd stage			ϕvol[-]	Df[-]			Df[-]
5	<1	20.23 ± 6.644a	4.27 ± 0.246a	0.10 ± 0.017a	2.56 ± 0.019a, b	1.16 ± 0.098a	3.37 ± 0.112a	2.02 ± 0.066a
10	1	24.54 ± 9.790a	4.15 ± 0.261a, b	0.10 ± 0.025a	2.56 ± 0.040a, b	0.73 ± 0.118b, c	3.64 ± 0.037a	2.20 ± 0.090b
15	>1, <2	34.07 ± 1.904a	3.97 ± 0.064a, b	0.11 ± 0.002a	2.50 ± 0.015a	0.49 ± 0.069c	3.51 ± 0.068a	2.22 ± 0.009b, c
20	2	25.21 ± 9.318a	4.20 ± 0.162a, b	0.13 ± 0.030a	2.63 ± 0.056b	0.47 ± 0.174c	2.81 ± 0.361b	2.37 ± 0.070c, d
25	>2, <3	32.90 ± 1.150a	3.76 ± 0.088b	0.14 ± 0.018a	2.58 ± 0.044a, b	1.04 ± 0.173a, b	3.37 ± 0.118a	2.42 ± 0.053d

Mean ± Standard deviations (n = 3), , average volume of aggregates; , specific surface area; ϕ_vol, volume fraction; D_f, fractal dimension

Values followed by the same superscript letter in the same column represent no significant difference at p < 0.05.

As the homogenization pressure increased, the fat globules became difficult to detect on the protein aggregate surface. It suggests that the fat globules contribute to forming the protein network, thus improving the elasticity of the yogurt microgel. With increasing pressure, the average fat globule volume decreased up to 15 MPa, consistent with the change in the mean diameter of fat globules of reconstituted milk. However, at pressures > 20 MPa, the average fat globule volume was higher, probably due to milk fat globule flocculation. The fat globule fractal dimension increased with increasing pressure, indicating that high pressure produced a complex distribution of fat globules, reinforcing the milk protein network.

Krzeminski et al. (Krzeminski et al., 2011) reported that adding whey protein causes larger particle size, higher viscosity, and firmer network structure in stirred yogurt. However, high-level whey protein concentration could induce large interstitial spaces and huge amount of whey protein aggregates, therefore, leads to a coarse gel network structure, which can be crushed easily. In present work, the results of zeta potential and PSH suggested that more whey protein is absorbed on the interface of milk fat globules as homogenization pressure rising. This indicates that high pressure homogenization leads to more binding sites for whey protein on milk fat globule surface, which inhibits the formation of whey protein aggregates and large particles. Furthermore, since whey protein on milk fat globules provides binding sites of covalent bond, interaction between milk protein and fat globule becomes stronger and build firmer milk protein network structure.

The interactions between milk protein and fat globules are influenced by the properties of the oil or fat, such as the solid-fat content, interface tension, and triacylglycerol composition. Therefore, further research is required into the impact of the type of oil or fat dispersed in the reconstituted milk on the physicochemical properties of resulting stirred yogurts. It would be useful to apply small angle x-ray scattering to quantify the changes in milk protein-fat globule complexes that we observed via TEM.

Conclusion

This study evaluated the colloidal properties of reconstituted milk samples containing differently sized fat globules, and assess the physicochemical properties of stirred yogurt prepared from these reconstituted milk samples. As homogenization pressure increased, fat globule size decreased, and the interactions between milk proteins and fat globules changed. Evaluation of the interface and colloidal properties revealed that increased pressure increased the zeta potential and PSH of the reconstituted milk samples. Consequently, reconstituted milk with high-pressure homogenization tends to produce stirred yogurt with higher viscosity under the same conditions (Fig. 10). This could be associated with the consistency felt during oral processing, and enhance the smoothness of stirred yogurt.

Fig. 10

The schematic diagram of stirred yogurt gel structures influenced by the milk fat globules homogenized using different pressures.

Acknowledgements The authors thank Meiji Co., Ltd. for providing the material for the yogurt preparation of this work. We thank the staff of the Electron Microscope Room, Iwate University, for assisting with the microscopy.

Conflict of interest There are no conflicts of interest to declare.

References

•  Alcoutlabi,  M. and  McKenna,  G. (2005). Effects of confinement on material behaviour at the nanometre size scale. J. Phys.: Condens. Matter, 17, R461. doi: 10.1088/0953-8984/17/15/R01.
•  Arab,  M.,  Yousefi,  M.,  Khanniri,  E.,  Azari,  M.,  Ghasemzadeh-Mohammadi,  V., and  Mollakhalili-Meybodi,  N. (2022). A comprehensive review on yogurt syneresis: effect of processing conditions and added additives. J. Food Sci. Technol., 60, 1656–1665. doi: 10.1007/s13197-022-05403–6.
•  Bonomi,  F.,  Iametti,  S.,  Pagliarini,  E., and  Peri,  C. (1988). A spectrofluorimetric approach to the estimation of the surface hydrophobicity modifications in milk proteins upon thermal treatment. Milchwissenschaft (Germany, F.R.), 43, 281–285. (in German)
•  Fazzalari,  N.L., and  Parkinson,  I.H. (1996). Fractal dimension and architecture of trabecular bone. J. Pathol., 178, 100–105. doi: 10.1002/(SICI)1096-9896(199601)178:1<100::AID-PATH429>3.0.CO;2-K.
•  Gabriele,  D.,  de Cindio,  B., and  D’Antona,  P. (2001). A weak gel model for foods. Rheol. Acta, 40, 120–127. doi: 10.1007/s003970000139.
•  Haque,  A.,  Richardson,  R., and  Morris,  E. (2001). Effect of fermentation temperature on the rheology of set and stirred yogurt. Food Hydrocolloids, 15, 593–602. doi: 10.1016/S0268-005X(01)00090-X.
•  Hati,  S.,  Mandal,  S., and  Prajapati,  J. (2013). Novel starters for value added fermented dairy products. Current Research in Nutrition and Food Science Journal, 1, 83–91. doi: 10.12944/CRNFSJ.1.1.09.
•  Herve-Jimenez,  L.,  Guillouard,  I.,  Guedon,  E.,  Boudebbouze,  S.,  Hols,  P.,  Monnet,  V.,  Maguin,  E., and  Rul,  F. (2009). Postgenomic analysis of Streptococcus thermophilus cocultivated in milk with Lactobacillus delbrueckii subsp. bulgaricus: involvement of nitrogen, purine, and iron metabolism. Appl. Environ. Microbiol., 75, 2062–2073. doi: 10.1128/AEM.01984–08.
•  Kalab,  M. (1988). Encapsulation of viscous foods in agar gel tubes for electron microscopy. Food Struct., 7, 12. Retrieved from https://digitalcommons.usu.edu/foodmicrostructure/vol7/iss2/12.
•  Kopp-Hoolihan,  L. (2001). Prophylactic and therapeutic uses of probiotics: a review. J. Am. Diet. Assoc., 101, 229–241. doi: 10.1016/S0002-8223(01)00060–8.
•  Kratz,  M.,  Baars,  T., and  Guyenet,  S. (2013). The relationship between high-fat dairy consumption and obesity, cardiovascular, and metabolic disease. Eur. J. Nutr., 52, 1–24. doi: 10.1007/s00394-012-0418–1.
•  Krzeminski,  A.,  Großhable,  K., and  Hinrichs,  J. (2011). Structural properties of stirred yoghurt as influenced by whey proteins. LWT - Food Science and Technology, 44, 2134–2140. doi: 10.1016/j.lwt.2011.05.018
•  Laguna,  L.,  Farrell,  G.,  Bryant,  M.,  Morina,  A., and  Sarkar,  A. (2017). Relating rheology and tribology of commercial dairy colloids to sensory perception. Food Funct., 8, 563–573. doi: 10.1039/C6FO01010E.
•  Lee,  W.J., and  Lucey,  J.A. (2004). Structure and physical properties of yogurt gels: Effect of inoculation rate and incubation temperature. J. Dairy Sci., 87, 3153–3164. doi: 10.3168/jds.S0022-0302(04)73450–5.
•  McMahon,  D.J., and  Brown,  R.J. (1984). Composition, structure, and integrity of casein micelles: a review. J. Dairy Sci., 67, 499–512. doi: 10.3168/jds.S0022-0302(84)81332–6.
•  Mutahi,  A., and  Miura,  M. (2020). Effects of Different Chemically Modified Starches on the Rheological Properties of Stirred Non-fat Yoghurt. Food Sci. Technol. Res., 26, 177–184. doi: 10.3136/fstr.26.177.
•  Nakamura,  A.,  Yoshida,  R.,  Maeda,  H., and  Corredig,  M. (2006). The stabilizing behaviour of soybean soluble polysaccharide and pectin in acidified milk beverages. Int. Dairy J., 16, 361–369. doi: 10.1016/j.idairyj.2005.01.014.
•  Obeid,  S.,  Guyomarc’h,  F.,  Francius,  G.,  Guillemin,  H.,  Wu,  X.,  Pezennec,  S.,  Famelart,  M. H.,  Cauty,  C.,  Gaucheron,  F., and  Lopez,  C. (2019). The surface properties of milk fat globules govern their interactions with the caseins: role of homogenization and pH probed by AFM force spectroscopy. Colloids Surf., B, 182, 110363. doi: 10.1016/j.colsurfb.2019.110363.
•  Routray,  W. and  Mishra,  H.N. (2011). Scientific and technical aspects of yogurt aroma and taste: a review. Compr. Rev. Food Sci. Food Saf., 10, 208–220. doi: 10.1111/j.1541-4337.2011.00151.x.
•  Ruggeri,  F.S.,  Šneideris,  T.,  Vendruscolo,  M., and  Knowles,  T.P.J. (2019). Atomic force microscopy for single molecule characterisation of protein aggregation. Arch. Biochem. Biophys., 664, 134–148. doi: 10.1016/j.abb.2019.02.001.
•  Sato,  T. (1968). A modified method for lead staining of thin sections. J. Electron Microsc., 17, 158–159.doi: 10.1093/oxfordjournals.jmicro.a049610.
•  Savaiano,  D. A., and  Hutkins,  R.W. (2021). Yogurt, cultured fermented milk, and health: A systematic review. Nutr. Rev., 79, 599–614. doi: 10.1093/nutrit/nuaa013.
•  Sodini,  I.,  Remeuf,  F.,  Haddad,  S., and  Corrieu,  G. (2004). The relative effect of milk base, starter, and process on yogurt texture: a review. Crit. Rev. Food Sci. Nutr., 44, 113–137. doi: 10.1080/10408690490424793.
•  Tremblay,  A., and  Panahi,  S. (2017). Yogurt consumption as a signature of a healthy diet and lifestyle. J. Nutr., 147, 1476S–1480S. doi: 10.3945/jn.116.245522.
•  Tribst,  A.A.L.,  Falcade,  L.T.P.,  Carvalho,  N.S.,  Cristianini,  M.,  Júnior,  B.R.D.C.L., and  de Oliveira,  M.M. (2020a). Using physical processes toimprove physicochemical and structural characteristics of fresh and frozen/thawed sheep milk. Innovative Food Sci. Emerging Technol., 59, 102247. doi: 10.1016/j.ifset.2019.102247.
•  Tribst,  A.A.L.,  Falcade,  L.T.P.,  Carvalho,  N.S.,  Júnior,  B.R. D. C. L., and  de Oliveira,  M. M. (2020b). Using stirring and homogenization to improve the fermentation profile and physicochemical characteristics of set yogurt from fresh, refrigerated and frozen/thawed sheep milk. LWT-Food Sci. Technol., 130, 109557. doi: 10.1016/j.lwt.2020.109557.
•  Vélez-Ruiz,  J.F. (2019). Rheological Characterization and Pipeline Transport Needs of Two Fluid Dairy Products (Flavored Milk and Yogurt). In “Milk-based Beverages,” ed. By  A. M.  Grumezescu, and  A. M.  Holban. Woodhead Publishing, pp. 427–472. doi: 10.1016/b978-0-12-815504-2.00013-x.
•  Walstra,  P. (1983). Physical Chemistry of Milk Fat Globules. In “Developments in Dairy Chemistry—2,” ed. by  P.  Fox. Springer, pp. 119–158. doi: 10.1007/978-94-010-9231-9_4.
•  Wiking,  L.,  Gregersen,  S.,  Hansen,  S., and  Hammershøj,  M. (2022). Heat-induced changes in milk fat and milk fat globules and its derived effects on acid dairy gelation–a review. Int. Dairy J., 127, 105213. doi: 10.1016/j.idairyj.2021.105213.
• i)  https://www.who.int/news-room/fact-sheets/detail/obesity-and-overweight (Dec. 14, 2021)