Acid gelation properties of fibrillated model milk protein concentrate dispersions

Whey proteins in milk are globular proteins that can be converted into fibrils to enhance functional properties such gelation, emulsification, and foaming. A model fibrillated milk protein concentrate (MPC) was developed by mixing micellar casein concentrate (MCC) with fibrillated milk whey proteins. Similarly, a control model MPC was obtained by mixing MCC with milk whey proteins. The resulting fibrillated model MPC and control model MPC contained 5% protein and a ratio of casein to whey proteins similar to milk. The objective of the current study was to understand the rheological characteristics of fibrillated and con-trol model MPC during acid gelation, using Förster resonance energy transfer (FRET) to assess small amplitude oscillation and casein–whey protein interaction. The results from the FRET index images showed greater interactions between caseins and whey proteins in fibrillated model MPC compared with the moderate and uniform interactions in control model MPC gels. Rheological study showed that the maximum storage modulus of acid gel of fibrillated model MPC was 546.9 ± 15.5 Pa, which was significantly higher than acid gel made from control model MPC (336.9 ± 11.3 Pa), indicating that fibrillated model MPC produced a firmer gel. Therefore, it can be concluded that acid gel produced from fibrillated model MPC was stronger than control model MPC. Selective fibrillation of the whey protein fraction in MPC can be used to improve gelation characteristics of acid gel type products.


INTRODUCTION
Major milk proteins are composed of caseins and whey proteins.In milk, casein micelles are present in a colloidal state and stabilized by surface (zeta) potential and steric stabilization due to protruding κ-CN hairs (Fox et al., 2015).When the pH of milk is reduced to 4.6 by gradual acidification, the surface potential on the casein micelles is diminished and casein micelles form a 3-dimensional network that entraps other constituents and consequently creates a homogeneous gel matrix (Donato et al., 2007).Acid gel properties such as firmness and mechanical strength are important for overall product quality and consumer acceptability (Peng et al., 2009).Acid gelation properties are best measured by rheological parameters such as storage modulus and complex viscosity (Lucey et al., 1997).
Currently, several processes are used extensively to improve the quality and stability of acid gels.Among them, the heating of milk before acid gelation is a common practice because it denatures whey proteins, allowing more water to be bound and the formation of a firm gel (Lucey et al., 1997;Anema et al., 2004;Lee and Lucey, 2004).Therefore, milk is heated to 90 to 95°C before acid gelation to denature whey proteins and to promote casein and whey protein interactions, specifically between κ-CN and β-LG (Gunasekaran and Solar, 2012).Casein-whey protein interactions during heating and acid gelation can be studied using the Förster resonance energy transfer (FRET) technique (Sahoo, 2011;Okamoto and Sako, 2017).Another approach to improving gelation characteristics is to incorporate additional milk proteins such milk protein concentrates and isolates and whey protein concentrates and isolate, as proteins are mainly responsible for the gel formation (Modler et al., 1983: Marafon et al., 2011;Karam et al., 2013).However, these additions contribute to increased costs.Hence, addition of gums and texture modifiers such as guar gum and low methoxyl pectin is another cost-effective method to improve acid gelation properties and gel strength (Baba et al., 2018;Khubber et al., 2021).The addition of gums and texture modifier improve gel by forming a dense network, which increases viscosity, water holding, and overall gel strength.
Converting globular proteins into fibrils has the potential to improve acid gelation characteristics because the fibrillar form has a good network-forming ability (Loveday et al., 2017).Most commonly, fibrils are prepared by heating an acidified protein solution to 80°C at pH 2 for 5 to 20 h.Unlike native globular proteins, fibrils have a high aspect ratio, with a length of a few micrometers and a diameter of a few nanometers.The conversion of native globular proteins such as whey proteins opens the protein structure and exposes more reactive groups that favor protein-protein interactions, enabling the formation of a close-knit acid gel (Loveday et al., 2012;Munialo et al., 2014;Mohammadian and Madadlou, 2016).In a previous study (Rathod and Amamcharla, 2021), milk whey protein isolates (mWPI) were converted into fibrillated mWPI, and a novel fibrillated model milk protein concentrate (MPC) was subsequently developed by selective fibrillation of whey proteins, followed by mixing with micellar casein concentrate (MCC) to achieve a ratio of caseins to whey proteins similar to milk.The fibrillated model MPC showed a higher viscosity and consistency compared with the control model MPC.In the current study, fibrillated model MPC was used for gelation studies along with control model MPC.The objective was to understand the rheological characteristics during acid gelation of the newly developed fibrillated model MPC.Further, the FRET technique was also used to understand the interactions between caseins and fibrillated whey proteins.

MATERIALS AND METHODS
No animals were used in this study, and ethical approval for the use of animals was thus deemed unnecessary.

Experimental Approach
Two lots of mWPI [4.3% moisture, 91.6% true protein (87.6% whey protein and 12.4% casein), 1.4% lactose, and 2.1% ash] and 1 lot of MCC [5.76% moisture, 85.5% true protein (7.6% whey protein and 92.4% casein)] were procured from a commercial supplier as powders.A simulated control model MPC was prepared by mixing reconstituted mWPI (2% protein) with reconstituted MCC (8% protein) to achieve a similar ratio of caseins to whey proteins as in milk.The fibrillated model MPC was likewise prepared by mixing pH-adjusted fibrillated mWPI (2% protein) with reconstituted MCC (8% protein) as described by Rathod and Amamcharla (2021).Control and fibrillated model MPC dispersions were heat treated and subsequently acidified to form a gel.Next, FRET analysis was carried out to understand protein-protein interactions between casein and whey proteins in control and fibrillated model MPC.Further, rheological properties of control and fibrillated model MPC were analyzed and compared.The experiment was replicated twice, using 2 independent lots of mWPI.

Preparation and Characterization of Simulated Control and Fibrillated Model MPC
Simulated control and fibrillated model MPC were prepared following the protocol described by Rathod and Amamcharla (2021).Initially, mWPI powders were reconstituted to 2% (wt/wt) solution on a protein basis using distilled water and stored overnight at 4°C to ensure complete rehydration.The following day, the mWPI solution was divided into 2 equal parts.One part was used as a control without any further treatment, while the pH of the second part was adjusted to 2.0 using 6 N hydrochloric acid and stored overnight at 4°C.The pH-adjusted mWPI solution was heated to 80°C and maintained at 80°C for 20 h under continuous gentle stirring using a magnetic stirrer to fibrillate the whey proteins, followed by immediate cooling to 4°C.Subsequently, the fibrillated mWPI solution was neutralized back to pH 6.7 using 3 N sodium hydroxide under gentle stirring to produce pH-adjusted fibrillated mWPI.
Simultaneously, MCC powder was also reconstituted to 8% (wt/wt) on a protein basis using distilled water at 50°C.Mixing was performed using an overhead stirrer (Caframo) at 750 rpm for the first 15 min, followed by continuous stirring at 500 rpm for an additional 30 min.The MCC dispersion was stored overnight at 4°C and then divided into 2 equal parts the following day.The first part of the reconstituted MCC was mixed with nonfibrillated mWPI using a magnetic stirrer (50 rpm), resulting in control model MPC with 5% (wt/ wt) protein and a casein-to-whey protein ratio similar to milk.Similarly, fibrillated model MPC was also prepared by mixing pH-adjusted fibrillated mWPI solution and the second part of the reconstituted MCC, resulting in fibrillated model MPC with 5% (wt/wt) protein similar to control model MPC.Sodium azide was added at the rate of 0.02% (wt/wt) in both types of MPC and stored in the refrigerator.

Confirmation of the Presence of Fibrils in Model MPC
Both model MPC dispersions were characterized in terms of thioflavin T fluorescence value, transmission electron microscopy (TEM), and viscosity, using the respective methods as described by Rathod and Amamcharla (2021), to confirm the presence of fibrils.

Acid-Base Titration Curve
Acid-base titration profiles of control and fibrillated model MPC were constructed by using the method described by Meletharayil et al. (2015).Briefly, both control and fibrillated MPC samples were titrated from their initial pH of 6.7 to pH 3.0 by adding 0.2 mL of 0.5 N hydrochloric acid in increments under constant stirring.The pH was recorded after each 0.2 mL incremental addition of hydrochloric acid.Once the pH reached 3.0, the same sample was back-titrated to pH 7.0 by 0.2-mL incremental additions of 0.5 N sodium hydroxide under constant stirring.The pH was also recorded after each 0.2-mL addition of sodium hydroxide.The pH was recorded using an Accumet AP110 portable pH meter (Fisher Scientific).Similarly, the acid-base titration curves of nonfibrillated mWPI and fibrillated mWPI solutions were constructed through incremental addition of 0.1 mL instead of 0.2 mL of acid and base for acidification and neutralization, respectively.Samples were analyzed in triplicate using 2 independent lots, and results were averaged.

FRET Microscopy
The FRET technique is widely used in biomedical and molecular biology to understand intermolecular or intramolecular interactions.The technique is based on the nonradiative (dipole-dipole) energy transfer from a donor fluorescent dye to an acceptor fluorescent dye that is bound to a molecule or molecules of interest (Hochreiter et al., 2019) when the donor and acceptor dyes are in close proximity (<10 nm). Figure 1 shows the working principle of the FRET technique.Jensen et al. (2015) used the FRET technique to study the function of the milk-clotting enzyme chymosin from bovine and camel sources.In the current study, caseins were labeled with a donor dye and whey proteins were labeled with an acceptor dye to determine the caseinwhey protein interactions in MPC.As shown in Figure 1, no energy transfer occurs between the donor and acceptor when the labeled casein and whey proteins are not interacting or are separated by a distance greater than the Förster radius.However, radiation-free energy transfer between the donor and acceptor dyes occurs when the labeled proteins are within the Förster radius (<10 nm).The Förster radius is defined as the distance at which 50% of the excited donors transfer energy to acceptors.As the labeled proteins come close during interaction, the acceptor and donor dyes are within their Förster radius, leading to the accepter dye emitting the accepter wavelength (560 nm) as shown in Figure 1.

Fluorescent Dye Labeling and Sample Preparation for FRET Analysis
The FRET analysis was carried out using the fluorescent dyes ATTO 488 NHS ester and ATTO 532 NHS ester (ATTO-TEC GmbH) as donor and acceptor dyes, respectively (Glover et al., 2020).For labeling of the proteins in the MCC, the pH of the MCC dispersion (8% protein) was increased to 8.3 using 1 M sodium bicarbonate and subsequently divided into 2 equal parts.ATTO 488 NHS ester (1 mg) was dissolved in dimethyl sulfoxide (20 µL) and added to the first part of MCC dispersion (1.2 mL) in Eppendorf tubes, as suggested by Glover et al. (2020), and then mixed thoroughly.Further, the mixture was intermittently vortexed every 15 min for up to 1 h.Fluorescent dye was not added to the second part of the MCC, which was considered unlabeled.Subsequently, the pH of both labeled and unlabeled MCC solutions was adjusted to pH 7 using 1 N hydrochloric acid.Similarly, the mWPI solution (2% protein) was divided into 2 equal parts.The first part (0.7 mL) was labeled with 0.15 mg ATTO 532 NHS ester dissolved in dimethyl sulfoxide (3 µL) and was termed as labeled.The second part did not receive a fluorescent dye and was termed as unlabeled.
For FRET analysis, the donor-only sample was prepared by mixing 1 part of ATTO 488 MCC solution (250 µL) with 1 part of unlabeled mWPI solution (250 µL); this sample was termed as the donor-only control model MPC.An accepter-only sample was prepared by mixing 1 part of unlabeled MCC solution with 1 part of ATTO 532 mWPI, and it was termed as the accepter-only control model MPC.The FRET sample was prepared by mixing 1 part of ATTO 488 MCC with 1 part of ATTO 532 mWPI solution, and it was termed as the FRET control model MPC.Similarly, donoronly fibrillated model MPC, accepter-only fibrillated model MPC, and FRET fibrillated model MPC were prepared by using fibrillated mWPI solution in place of the mWPI solution.Acid gels of all control and fibrillated model MPC were prepared by heating labeled and unlabeled model MPC dispersions to 90°C for 10 min, followed by cooling to 30°C.Then, a precalculated amount of glucono-δ-lactone was added and mixed for 1 min, followed by incubation at 30°C for 4 h.The acid gels were used to acquire confocal laser scanning microscopic images for FRET analysis.

Imaging of Acid Gel Samples for FRET Analysis
All acid gels prepared from control and fibrillated model MPC samples were carefully mounted on glass slides, avoiding gel disruption, and viewed on an Axioplan 2 microscope equipped with confocal laser scanning LSM 5 Pascal (Carl Zeiss) version 3.2 SP2.A Plan-Apochromat ×100/1.4oil objective was used for image acquisition.The donor-only acid gels were excited with 488 nm argon laser line with primary dichroic-HFT 488/543/633 to excite ATTO 488 NHS ester, while secondary dichroic NFT 545 was used to separate emission from the fluorescent dye.The emission signal was collected using a 505-to 530-nm band pass filter with the image collected by channel 2 [photomultiplier tube for detecting ATTO 488 NHS ester (green)].The acceptor-only acid gel was excited with a 543-nm helium-neon laser line with primary dichroic HFT 488/543/633 to excite ATTO 532 NHS ester, while secondary dichroic NFT 545 was used to separate emission from the fluorescent dye.The emission signal was collected using a 560-nm longpass filter with image collected by channel 1 [for detecting ATTO 532 NHS ester (red)].The FRET acid gels were excited with 488-nm argon laser line with primary dichroic-HFT 488/543/633 to excite the fluorescent dye, while secondary dichroic NFT 545 was used to separate emission from the fluorescent stain.The emission signal was collected using a 560-nm longpass filter with images collected by channel 1 [for detecting FRET (blue)].The captured 8-bit images were maximized using a dynamic range indicator associated with the acquisition software (Hochreiter et al., 2019).

Analysis of FRET Index
Donor-only, accepter-only, and FRET images were analyzed to derive the FRET index, using a FRET Analyzer plugin of ImageJ software (version 1.53f; ImageJ, National Institutes of Health).Donor and accepter images were analyzed for bleed-through analysis and checked for standard error and Pearson r value.The FRET images were analyzed for the FRET index, which was further analyzed using the color counter plugin of ImageJ software to get the hexacolor code with a pixel number to quantify the concentration of each color in the FRET index images to quantify the FRET index for both control and fibrillated model MPC.Hexacolor codes were converted to RGB (red, green, blue) combination and further clustered into black, blue, red, yellow, and white depending upon the dominance of the R, G, or B value in the RGB combination and their respective broad color classification.

Preparation of Acid Gels from Control and Fibrillated Model MPC
Both control and fibrillated model MPC dispersions were heated to 90°C for 10 min, immediately cooled to 4°C, and kept at refrigerated temperature overnight for further analysis.The following day, model MPC dispersions were tempered to 30°C, 2 g of glucono-δ-lactone (Fisher Scientific) was added per 100 mL, and the solutions were mixed using a magnetic stirrer for 1 min.Immediately afterward, part of the model MPC mixed with glucono-δ-lactone was transferred to a temperature-controlled water bath (Cole Parmer) maintained at 30°C for continuous pH measurement using the portable pH meter.Simultaneously, the remaining model MPC was used to monitor rheological characteristics during acid gelation.

Small Amplitude Oscillation
Acid gel formation was monitored using a stressstrain-controlled rheometer (MCR-92 Anton Paar) with a cup-and-bob geometry consisting of a coaxial cylinder (bob length, 60 mm; bob diameter, 38.7 mm, and cup diameter, 42 mm).The cup-and-bob geometry was preheated to 30°C.Vegetable oil was added to the model MPC surface to avoid evaporation during measurement.During acid gelation, the sample oscillated at a constant frequency of 1 Hz with an applied strain of 0.5%, which caused minimal disruption during the development of the gel network (Peng et al., 2009).Measurements were taken every 5 min until the pH reached 4.6.The derived parameters included storage modulus (G′) and loss tangent (LT).Gelation was arbitrarily defined as the moment when the G′ of gels was greater than 1 Pa (Peng et al., 2009).Samples were analyzed in duplicate using 2 independent lots, and results were averaged.

Statistical Analysis
Statistical analysis was performed using Excel (Microsoft Corp.).One-way ANOVA and Tukey's multiple range test were done using SAS Version 9.4 (SAS Institute Inc.).

Characteristics of Control and Fibrillated Model MPC
Thioflavin T values for control and fibrillated model MPC were 247.1 ± 11 AU and 367.7 ± 33.2 AU, respectively, and found to be significantly (P < 0.05) different.A higher thioflavin T value for the fibrillated model MPC indicated the presence of whey protein fibrils.In addition to a higher thioflavin T value, the presence of fibrils in the fibrillated model MPC was also confirmed using TEM (images not shown) and found to be similar to a previous study (Rathod and Amamcharla, 2021).Importantly, the commercial mWPI used in this study for fibrillation contained casein in addition to whey proteins.Coppola et al. (2014) and Carter et al. (2021) reported the possibility of a fraction of caseins (up to 1-20%) leaking through the microfiltration membrane along with whey proteins depending on the processing conditions, such as filtration temperature.However, concentrations of casein in WPI derived from cheese whey are low and are attributed to the higher selectivity of rennet enzyme compared with microfiltration membranes (Carter et al., 2021).Casein fractions such as α S1 -CN, α S2 -CN, β-CN, and κ-CN in mWPI intended for fibrillation of whey proteins can have a chaperone effect and can inhibit the whey protein fibrillation (Bhattacharyya and Das, 1999;Morgan et al., 2005;Treweek et al., 2011).However, previous studies were carried out under nearly neutral pH conditions.In addition, β-CN also acts as a molecular chaperone at high temperatures (>75°C) and neutral pH (O' Kennedy and Mounsey, 2006;Kehoe and Foegeding, 2011;Yong and Foegeding, 2008), but it is unclear whether the chaperone-like function persists during heating under acidic pH conditions.Recently, Dave et al. (2016) studied the chaperone-like effect of β-CN in the context of β-LG fibrillation at pH 2 and 80°C.The authors reported that the fibril formation was somewhat retarded at lower β-CN concentrations.However, at higher concentrations of β-CN, whey protein fibrillation was not further inhibited by the presence of β-CN and was attributed to the diversion of β-CN monomers, peptides, or both into self-assembled micelles.Reducing sodium dodecyl sulfate polyacrylamide gel electrophoresis images from a previous study (Rathod and Amamcharla, 2021) showed a clear band for β-CN in nonfibrillated mWPI.However, the β-CN band was absent in fibrillated mWPI at pH 2 as well as at pH 6.7.In addition, TEM images from a previous work by Rathod and Amamcharla (2021) showed the presence of visible fibrils in fibrillated mWPI as well as in fibrillated model MPC.Similarly, TEM imaging of whey protein fibrils (Figure 2) from the current work confirms the presence of fibrils in fibrillated model MPC.Further, control and fibrillated model MPC were further evaluated in terms of apparent viscosity (at 100 S −1 ) and found to be 12.69 ± 0.87 and 17.3 ± 2.54 mPaS, respectively; these values were also significantly (P < 0.05) different from each other.These results showed that the presence of fibrils increased the viscosity of the MPC dispersions.Similar findings were reported by Rathod and Amamcharla (2021) and other studies on whey protein fibrils (Loveday et al., 2012;Dave et al., 2013;Farrokhi et al., 2019).

Acid-Base Titration Curve
The acid-base titration curves of control and fibrillated model MPC are shown in Figure 3. Figure 3A shows the change in pH with respect to the addition of 0.5 N HCl.The control and fibrillated model MPC required the same amount of acid (2 mL) to reach pH 5.1 and thereby exhibited a similar buffering capacity.Both model MPC contained the same concentration of caseins, as they were prepared by mixing the same amount of MCC; therefore, both MPC contained the same colloidal calcium phosphate (CCP), which was in equilibrium with calcium and phosphate of the serum phase (Eshpari et al., 2016).Acidification of the model MPC dispersions to pH ~5.0 led to a complete solubilization of CCP (Corredig et al., 2019).However, below pH 5.0, the buffering capacity was mostly contributed by the caseins and their side groups as most of the CCP were already dissolved from casein micelles (Lucey et al., 1996).
The control model MPC required 4 mL of acid to reduce the pH from 5.1 to 3.0.Whey proteins in the control model MPC contributed to buffering capacity between pH 4.0 and 3.0, which was attributed to the presence of acidic amino acids (Salaün et al., 2005).However, the fibrillated model MPC needed 4.6 mL of acid to attain a similar drop in pH.During acidification between pH 5.1 and 3.0, the fibrillated model MPC required a higher amount of acid than the control model MPC to lower the pH, possibly due to the presence of fibrillated whey proteins in the fibrillated model MPC.
To further understand the differences in buffering capacity between native whey proteins and whey protein fibrils, nonfibrillated mWPI solution (2% wt/wt) and fibrillated mWPI solution (2% wt/wt) were analyzed in terms of their acid-base titration curves, similar to model MPC. Figure 3B shows that the fibrillated mWPI required a higher amount of acid to change the pH from 6.7 to 3 than nonfibrillated mWPI.
During back-titration from pH 3.0 to 7.0 (Figure 3C), the fibrillated model MPC needed more alkali (7.2 mL) to reach pH 7.0 from pH 3.0 than the control model MPC (6.2 mL).A higher alkali requirement for the fib-rillated model MPC could be attributed to the presence of whey proteins such as fibrils in this MPC.During the fibrillation process, native globular whey proteins are hydrolyzed and rearranged into multistranded twisted structures that are rich in β-sheets perpendicular to the fibril axis (Akkermans et al., 2008;Kroes-Nijboer et al., 2012;Loveday et al., 2017).The structural rearrangement changes the position and interactions of charged groups, possibly due to the higher buffering capacity of the fibrillated model MPC compared with the control model MPC.To confirm the role of whey protein fibrils in acid-base buffering curves, nonfibrillated mWPI and fibrillated mWPI solutions that were previously acidified to pH 3.0 during acid titration were back-titrated to pH 7.0 with incremental amounts of 0.5 N NaOH.Fibrillated mWPI needed a higher amount of alkali to reach pH 7.0 than nonfibrillated mWPI (Figure 3D).This observation proves that fibrillated mWPI is the reason for the higher buffering capacity of fibrillated model MPC. Figure 3C showed that no change occurred in the alkali addition rate during neutralization in the pH range of 5.1 to 7.0.During the acidification process, CCP continued to solubilize from the casein to the serum phase.When the same solution with destabilized CCP structure was neutralized during back-titration with alkali, the CCP structure was not fully recovered (Ezeh and Lewis, 2011); thereby, CCP may not have provided buffering capacity at pH 5.1.The buffering capacity of proteins can mostly be attributed to ionizable amino acid side groups, carboxyl, and amine groups liberated during enzymatic hydrolysis (Luo et al., 2018).Similarly, whey proteins were hydrolyzed (Akkermans et al., 2008) during acidification and subsequently rearranged into fibrils during the fibrillation process.Therefore, the differences in buffering capacity between control and fibrillated model MPC could be due to the presence of fibrils as functional groups rearranged during the fibrillation process, resulting in different buffering capacities at a lower pH range.

FRET Microscopy
Figure 4 shows images from the FRET analysis of control and fibrillated model MPC.The donor-in-donor only image (Figure 4A) shows a bright fluorescent green, which confirmed labeling of proteins in MCC and their distribution in the acid gel prepared from control model MPC.The FRET-in-donor only image (Figure 4B) shows a very low fluorescence signal, indicating negligible crosstalk or bleed-through of ATTO 488 dye in the FRET signal.The accepter-in-accepter only image (Figure 4C) shows a fluorescent red, which confirmed the labeling of whey proteins in mWPI in the acid gel.The color intensity of the FRET-in-accepter only image (Figure 4D) is also low, indicating a minimal bleedthrough of ATTO 532 in the FRET signal.Similar to the control model MPC, images of the fibrillated model MPC (donor-in-donor only [Figure 4H], FRET-in-donor only [Figure 4I], accepter-in-accepter only [Figure 4J], FRET-in-accepter only [Figure 4K)] also indicate minimal bleed-through and crosstalk.Additionally, bleed-through analysis of donor-FRET (Figures 4A and  4B; Figures 4H and 4I) and accepter-FRET (Figures 4C and 4D; Figures 4J and 4K) images in control and fibrillated model MPC, respectively, also showed a low standard error and Pearson r value greater than 0.95 (data not shown), indicating that the FRET interactions were highly significant (P < 0.05) in both MPC during the gelation process.These observations support that the proteins in MCC and mWPI were interacting in the acid gels.
The FRET index image of control model MPC (Figure 5A) was obtained by confocal laser scanning microscopic images (Figures 4E, 4F, and G) using ImageJ software.Similarly, the FRET index image of fibrillated model MPC (Figure 5B) was obtained by processing confocal laser scanning microscopic images (Figures 4L, 4M, and 4N).In the FRET index images (Figures 5A and 5B), the black-colored area indicates no FRET interaction between proteins in MCC suspension and mWPI solution, while the colors ranging from blue to red indicate a moderate to higher number of molecular interactions within the Förster radius for these 2 fluorochromes.However, white indicates saturation of the detector caused by an extremely high number of protein-protein interactions.Clustering of the color pixel of FRET index images showed 4.3% white, followed by 0.7% yellow, 92.9% red-orange, 1.4% blue, and 0.7% black for the control model MPC.However, the fibrillated model MPC showed 61.0% white, followed by 3.7% yellow, 28.5% red-orange, 0.1% blue, and 6.8% black.These findings suggest that the fibrillated model MPC had higher protein-protein interactions between MCC and fibrillated mWPI than the control model MPC.Fibrillated model MPC contained whey protein as fibrils with ordered β-sheets (Dave et al., 2013) and exposed disulfide bonds.Also, fibrils have a simpler secondary structure (Loveday et al., 2017) that is more open than the tertiary and quaternary structure of native proteins (Damodaran, 1997).Consequently, fibrils were more reactive and readily interacted with caseins during heating and acid gelation.
In addition, the FRET index image (Figure 5A) and the donor-in-donor only image (Figure 4A) collectively show that the acid gel of control model MPC was uniformly structured (denoted by a uniform red-orange area) with little or no porous structure (denoted by the lack of a blue-black area).However, the FRET index image (Figure 5B) and donor-in-donor only image (Figure 4H) for fibrillated model MPC show pores in the gel structure with protein dense area.Possible reasons could be higher interactions between caseins and whey proteins and the mounting of the acid gel on glass slides.

Rheological Properties of Acid Gel Prepared from Control and Fibrillated Model MPC
The gelation behavior of control and fibrillated model MPC is shown in Figure 6, and extracted gelation parameters are presented in Table 1.The gelation times for control model MPC and fibrillated model MPC were found to be 30.0± 1.3 and 28.8 ± 3.3 min, respectively, and they were not significantly different (P < 0.05) (Table 1).However, the gelation pH for the control model MPC and the fibrillated model MPC was 5.46 ± 0.03 and 5.55 ± 0.00, respectively, which were significantly different (P < 0.05).The possible reason for an early gelation could be attributed to fibrillated whey proteins and very high casein-whey protein interactions in the fibrillated model MPC.During fibril formation, whey proteins are hydrolyzed into peptides and self-arrange to form systematic β-sheet fibrils (Dave et al., 2013;Loveday et al.,2017) with disulfide bonds between the sheets.Fibrils have an open secondary structure with more exposed reactive groups in comparison to tertiary and quaternary structures of nonfibrillated whey proteins (Loveday et al., 2017).Higher interactions of caseins and fibrillated whey proteins were also evident from the FRET analysis.
Both control and fibrillated model MPC showed a gradual increase in storage modulus (G′) (Figure 6A).However, once gelation started, the increase in G′ was greater for fibrillated model MPC than for control model MPC.The possible reason for this behavior could be the presence of MCC, which was 80% of the total protein, and resistance to pH change through CCP until CCP was completely dissolved at ~pH 5.1.Gelation behavior is mainly governed by casein in MCC at a pH greater than 5.1.However, whey proteins or fibrils and their interactions with casein influence the gelation behavior at a pH below 5.1.Overall, the fibrillated model MPC showed a higher G′ over the entire gelation time until the pH reached 4.6.The G′ of the control model MPC increased to 336.9 ± 11.3 Pa (G′ max) at pH 4.79, which occurred at 158.8 ± 3.9 min; afterward, the G′ declined to 284.8 ± 10.3 Pa at pH 4.6.However, the G′ of the fibrillated model MPC increased to 546.9 ± 15.5 Pa (G′ max) at pH 4.64, which occurred at 209.3 ± 9.9 min; afterward, the G′ slightly declined to 541.8 ± 15.9 Pa at pH 4.6.The G′ of the fibrillated model MPC was significantly higher than that of the control model MPC, which indicates that the fibrils contributed to a firmer gel (Table 1).The G′ represents solid-like elastic behavior, hence a higher storage modulus implies higher solid-like behavior and elasticity of fibrillated model MPC (Cruz et al., 2013).
The plot between LT and pH (Figure 6B) shows that the control model MPC had higher LT than the fibrillated model MPC until pH 5.15, after which the fibrillated model MPC showed slightly higher LT till pH 4.8.Thereafter, the control model MPC showed an increased LT again until the final pH reached 4.6.For both MPC, the value of LT was below 1, which shows that the elastic behavior dominated (Mariotti et al., 2009) once gelation started in both MPC until pH 4.6.Figure 6B also shows that LT of the fibrillated model MPC was lower than the control model MPC.This observation suggests that acid gels of the fibrillated model MPC were more elastic (solid-like) than the control model MPC after the gelation point.Further, a maximum LT was seen for both MPC near pH 5.2, which is slightly above previously reported results for stirred yogurt (pH 5.0-5.1; Lee and Lucey, 2006).The gelation behavior of the control model MPC was found to be similar to earlier reported acid gels from MPC (Meletharayil et al., 2015(Meletharayil et al., , 2018)).However, the fibrillated model MPC formed a stronger elastic gel than the control model MPC owing to strong interaction between whey protein fibrils and casein, which can be seen in the FRET index image (Figure 5B).

CONCLUSIONS
The acid gelation properties of a control model MPC and a newly developed fibrillated model MPC were studied.The molecular interactions between casein and whey proteins were observed to be higher in the fibrillated model MPC compared with the control model MPC, and they consequently produced a firmer acid gel with improved microstructure.The fibrillated model MPC produced a more elastic and solid-like gel compared with the control model MPC.Further, acid gels of the fibrillated model MPC showed a higher gel strength than the control model MPC.It can be concluded that selective fibrillation of whey proteins in fibrillated model MPC can be used to improve acid gelation properties.a,b Means within a row with different superscripts differ (P < 0.05); n = 2.
1 From the plot of storage modulus (G′) versus pH (Figure 6).

Table 1 .
Rheological parameters obtained during gelation of acid gels of control and fibrillated model milk protein concentrates (MPC; means ± SD) 1