Microscopy and Microanalysis (2022), 1–10 1

doi:10.1017/S1431927622000162 2
3
4
5
Original Article 6
7

Characterization of Structural Changes of Casein Micelles at 8

9

Different PH Using Microscopy and Spectroscopy Techniques 10

11
12
13
Liliana Edith Rojas-Candelas1, José Jorge Chanona-Pérez1* , Juan Vicente Méndez Méndez1,2, 14
José Antonio Morales-Hernández3 and Héctor Alfredo Calderón Benavides4 15
1 2 16
Escuela Nacional de Ciencias Biológicas, Instituto Politécnico Nacional, Av. Wilfrido Massieu Esq, Cda, Miguel Stampa s/n, C.P. 07738 Mexico City, Mexico; Centro de
Nanociencias y Micro y Nanotecnologías, Instituto Politécnico Nacional, Luis Enrique Erro s/n, Zacatenco, Gustavo A. Madero, C.P. 07738 Mexico City, Mexico; 3Centro 17
de Investigación y Asistencia en Tecnología y Diseño del Estado de Jalisco (CIATEJ), A. C. Av Normalistas #800, Guadalajara, Jalisco, México and 4Instituto Politécnico 18
Nacional. Escuela Superior de Física y Matemáticas, Av. Instituto Politécnico Nacional Edificio 9, U. Profesional Adolfo Lopez Mateos, Gustavo A. Madero, C.P. 07738 19
Mexico City, Mexico 20
21
22
Abstract 23
This study aimed to evaluate the influence of pH changes on morphometric parameters of casein micelles and a general overview of their 24
conformational structure through microscopy techniques, Raman spectroscopy and multivariate analysis. It was found that casein micelles 25
morphology and protein secondary structure depend strongly upon pH. The changes of arithmetic average roughness (Ra), size, and shape 26
of casein micelles at different pH are properly characterized by atomic force and cryo-transmission electron microscopy. Morphometric 27
changes of casein micelles were correlated correctly with folding and unfolding of casein molecules as evaluated by Raman spectroscopy 28
when the pH was varied. The novelty of this contribution consists in demonstrating that there is a close structure-functionality relationship 29
between the morphometric parameters of proteins and their secondary structure. Knowledge about casein micelles can help improve their 30
use of its diverse applications. 31
32
Key words: atomic force microscopy, casein micelle structure, cryo-transmission electron microscopy, multivariate analysis, Raman
33
spectroscopy
34
(Received 28 April 2021; revised 22 November 2021; accepted 23 January 2022) 35
36
37
38
Introduction (Pelton & McLean, 2000). The structure and amino acids caseins
39
give rise to the biological and functional value of food from their
Casein is a protein that represents 80% of all of milk protein and 40
emulsifying properties, foam, gelling, and solubility stabilizers.
contain carbohydrates, phosphate, and serine residues. The 41
The emulsifying property of proteins depends upon changing
caseins with phosphate calcium form colloidal particles called 42
the conformational state of the protein, that is, the fold–unfold
micelles, which provide fluidity to casein and solubilize phosphate 43
condition. When the protein is folded, there is a tendency for a
and calcium. These colloidal particles are made up of four com- 44
stable and fibrous structure that improves interaction with posi-
ponents αs1, αs2, β, and κ-casein. This provides fluidity to the 45
tively charged amino acid residues (Gómez et al., 2013).
casein proteins. Casein micelles are held together through hydro- 46
Moreover, by changing the pH of the protein, it can unfold and
phobic interactions, hydrogen bonding, and electrostatic interac- 47
improve its foaming properties. However, for the protein to be
tions, and are stable to moderate thermal treatments, and their 48
unfolded, it must be exposed to reactive groups (Liu et al.,
micelles are precipitated by changes in pH or by enzymatic cleav- 49
2012), or else to high temperature and pressure (Smeller, 2002).
age of kappa casein (Swaisgood, 1993; Qi, 2007). Casein micelles 50
In contrast, when there is an increase in α-helix conformations,
have a supramolecular arrangement with a secondary degree of 51
as is the case with casein itself, a decrease in micelle size and poly-
the protein structuring and without any tertiary structure (Bhat 52
dispersity index can be developed. This is due to the casein sec-
et al., 2016). Such a secondary structure is formed by regions 53
ondary structure, that is, a reduction of turn regions, that can
that give rise to folding or unfolding of the casein micelles. 54
lead to a disaggregation of micelles (Yang et al., 2015). It is
This is due to the fraction of peptide bonds on the turn, random 55
clear that the pH value is one of the important factors to control
coil, α-helix, and β-laminar regions (β-parallel and β-antiparallel 56
the structure and properties of the casein micelles. This can be
structures). These regions allow the protein to fold back on itself 57
explained as a consequence of interrupting electrostatic interac-
and become stabilized or else unfold and become disordered 58
tions between residues of the protein, that generates the formation
59
of salt bridges and, thus changing its native structure
*Corresponding author: José Jorge Chanona-Pérez E-mail: jorge_chanona@hotmail.com 60
(Chakraborty & Basak, 2007). Alternatively, the pH has been
Cite this article: Rojas-Candelas LE, Chanona-Pérez JJ, Méndez JVM, Morales- 61
Hernández JA, Benavides HAC (2022) Characterization of Structural Changes of
important in lacteous products because it regulates the solubility
62
Casein Micelles at Different PH Using Microscopy and Spectroscopy Techniques. and physicochemical properties of the casein (Jahaniaval et al.,
63
Microsc Microanal. doi:10.1017/S1431927622000162 2000). Thus, cheese can be produced by using pH from 5 to 6
64
© The Author(s), 2022. Published by Cambridge University Press on behalf of the Microscopy Society of America 65
2 Liliana Edith Rojas‐Candelas et al.

(Solowiej et al., 2008), there is gel formation in a pH range from 4 casein molecules. Knudsen & Skibsted (2010) reported micelles 66
to 4.4 for creating yogurt (Lucey & Singh, 1997) and casein curd at different high-pressure treatments. They have found that a 67
particles produce glue by a further increase of the pH (Mc larger number of small micelles are formed at high pressure 68
Sweeney & O Mahony, 2016). with sizes from 30 to 100 nm. 69
The development of microscopy, and spectroscopy techniques, Thereby, the use of different microscopy techniques makes 70
has improved knowledge about the properties of casein micelles possible to discover structural information at different length 71
properties by means of image modeling and molecular dynamics scales of the casein micelles and their corresponding related prop- 72
(Michael Byler et al., 1988; Horne, 2006; Moitzi et al., 2011; Pan & erties. Also, information regarding protein–protein interactions 73
Zhong, 2015; Yang et al., 2015). For decades, different spectro- and the corresponding process of stabilization and formation of 74
scopy techniques (X-ray diffraction, nuclear magnetic resonance, micelles has been derived from the use of AFM and cryo-TEM 75
Fourier transform infrared (FTIR), and Raman spectroscopy) (Uricanu et al., 2004; Knudsen & Skibsted, 2010; Moitzi et al., 76
have been widely useful to elucidate the molecular structure of 2011; Ouanezar et al., 2012; Yahimi Yazdi et al., 2014; Bahri 77
proteins and evaluate conformational changes produced by their et al., 2017). 78
processing and variations of pH (Horne, 2002; Wang et al., Nevertheless, there are only a few investigations that relate the 79
2017). Particularly, Raman spectroscopy has been adequate and morphometry of casein micelles with the folding and unfolding 80
efficient to study the conformational modifications of secondary properties of the secondary structure through a combination of 81
and tertiary structures of food origin proteins (Carbonaro & microscopy techniques, Raman spectroscopy, and multivariate 82
Nucara, 2010; Wang et al., 2017). Michael Byler et al. (1988) analysis. Finally, to the best of our knowledge, no relationship 83
investigated with FTIR the structure lyophilized of αs1 casein, β has been determined so far between the casein micelle morphom- 84
casein, and their mixture. They concluded that individual confor- etry and the casein secondary structure. For this reason, this 85
mations’ behavior is different from whole casein because when it research compares the morphology, overall roughness and confor- 86
has mixed whole casein, the protein tends to increase turn, helix, mational structural of the casein protein at different pH values in 87
and β-sheet conformations. Hussain et al. (2011) carried out a order to better understand the structural modifications and fur- 88
sorting study on casein micelles at different NaCl concentrations ther assembly of casein micelles. This is done by combining 89
(0–12%) using FTIR for analyze its secondary structure and trans- microscopy techniques, Raman spectroscopy and multivariate 90
mission electron microscopy (TEM) to visualize its shape. The analysis. 91
study confirmed changes in secondary structure and size of casein 92
micelles. They concluded that when NaCl concentration increase, 93
casein structure has to decrease β-sheet and size of micelles Materials and Methods 94
because the abundance of charged ions in solution destabilizes Sample Preparation 95
micelle formation. 96
The outstanding versatility of atomic force microscopy Purified micellar casein from bovine milk was acquired from the 97
(AFM) to characterize the structure at the nano and molecular supplier Santa Cruz Biotechnology (sc-397376, USA, CAS 98
levels, has led to in-depth knowledge of the structure as the 9000-71-9). It was used for casein solutions at 1% w/v dissolved 99
source of different properties of the protein. Ouanezar et al. in MiliQ water (initial pH 6.8, Riché et al., 2006) and adjusted 100
(2012) investigated by AFM the casein micelles, compared its to the corresponding pH values by using 0.1 M HCl, at room tem- 101
shape and size at different pH values, and reported that the perature (25°C) and shaken in a vortex by 5 min. The pH values 102
size distributions ranged from 50 to 220 nm. They concluded under analysis were 4, 5 (isoelectric point), and 6 (Xiao-zhou 103
that the casein micelles at pH 5.0 have smaller dimensions et al., 2014). At least four replicates were used in all 104
than at pH 6.7, and the volume of the average micelle decrease measurements. 105
by 50–75% a pH 5.0 with a reduction of net charge. Vié et al. 106
(2002) using AFM have shown that the casein micelles diameter Atomic Force Microscopy 107
becomes smaller due to its hydrophilic and lipophilic properties. 108
Uricanu et al. (2004) used AFM to study the relationship Overall, aspect and roughness (Ra) of micelles were evaluated by 109
between temperature, Young’s modulus, concentration, and the using an atomic force microscope (Bruker, Bioscope Catalysts 110
size of casein micelles. They found that the micelles are more ScanAsyst, USA) after preparation at different pH values (4, 5, 111
rigid at higher temperatures and low protein concentration, and 6). Each sample was prepared with a solution of at 1% w/v 112
and Young’s modulus increase with the micelle diameter. dissolved in MiliQ water, and 0.1 mL was placed on a glass 113
Bahri et al. (2017, 2018) studied with AFM the topography slide and dried out in a desiccator overnight for AFM imaging 114
and nanomechanics of casein micelles at different pH values, (Rojas-Candelas et al., 2019). Silicon cantilevers (DNP-10A) of 115
they associated the micelles size and their nanomechanical prop- spring constant 0.540 Nm−1 were used and a resonant frequency 116
erties with the complexity of the casein micelle structure. of 1 kHz. Eight images of each pH values under investigation and 117
Transmission electron microscopy (TEM) is a valuable tool scan at 2 × 2 μm2 were captured in ScanAsyst mode in atmo- 118
that provides a higher spatial resolution for characterization of spheric conditions. The arithmetic average roughness (Ra) was 119
the protein micelles. Several studies have carried out on casein measured using NanoScope software. 120
micelles and submicelles using transmission electron cryo- 121
microscopy (cryo-TEM) to obtain information about protein– 122
Transmission Electron Cryo-Microscopy (Cryo-TEM) and Image
protein interactions and morphometric parameters in submicelles 123
Analysis
and micelles. For instance, Yahimi Yazdi et al. (2014) have 124
removed casein and formed micelles with resveratrol and curcu- A transmission electron cryo-microscope (Jeol, JEM-2100, USA) 125
min. They have found larger micelles due to the formation of was used to analyze the size and shape of casein micelles at differ- 126
approachable hydrophobic regions left by the removal of the ent pH values. The sample preparation was made according to the 127
Microscopy and Microanalysis 3

method described by Waninge et al. (2004) and Yahimi Yazdi analyses are carried out by using the SigmaPlot software v.12 128
et al. (2014). A solution of protein and water MiliQ at 0.01% p/ (Systat Software Inc., USA). Principal component analysis was 129
v was prepared in a vortex for 5 min. An aliquot of 5 μL was used to classify proteins according to their morphometric and 130
extracted with a micropipette and placed on a copper grid. The structural properties at different pH values, by using a matrix of 131
grid with the sample was then cooled by using a Cryoplunge 3 nine parameters (Ra, MD, R, AR, β-antiparallel, b-parallel, turn, 132
(Cp3, Gatan, Inc. USA) with liquid ethanol at −180°C and trans- random coil, α-helix). In the PCA, the distance of separation 133
ferred to liquid nitrogen at −196°C. Twenty images of around 200 (d) between the proteins at different pH values has been estimated 134
micelles of each pH value were obtained at an accelerating voltage by using factorial scores of the first two principal components 135
of 200 kV and a magnification at 20,000×. The images were (PCs) and the equation of distance between two points in a 2D 136
changed to RGB/TIFF format and converted to grayscale. The space: 137
binary images were segmented with the plugin Mexican Hat avail- 

























138
able from ImageJ software (v. 1.46, National Institute of Health, 139
d(C1 , C2 ) = (x2 − x1 )2 + (y2 − y1 )2 , (1)
Bethesda, MD, USA). The micelle diameter (MD) or major diam- 140
eter of particles, roundness (R = (4Area)/(π(MD)2)) and its 141
aspect ratio (AR, ratio between the minimum and maximum where x and y are the average factorial scores of each principal 142
diameters) were obtained from segmented images. All image anal- component obtained from four replicates, and C1 and C2 are 143
ysis were performed in ImageJ. the pH values or points where d is estimated. The distances are 144
used to analyze the similarities between the proteins at different 145
pH values. Additionally, a Pearson analysis has been calculated 146
Raman Spectroscopy with all variables and produces a visualized correlation matrix. 147
Raman spectroscopy (LabRAM model HR800, Japan) was used to In these analyses, XLSTAT software (2020.1.3, Addinsoft, USA) 148
provide a general overview of caused changes on casein secondary was used. 149
structure of prepared micelles at different pH values. It is 150
equipped with an Nd:YAG laser (785 nm), at room temperature 151
(25°C), and integration time of 6 s. Origin Pro 8 (v8.0724, Results and Discussion 152
USA) software was used to set a baseline and apply a smoothing 153
Atomic Force Microscopy
procedure by using the Savitzky-Golay method with a polynomial 154
order 2 and 12 points of reference. Subsequently, the second Figure 1 shows a selection of AFM height images for the casein 155
derivative has been performed to find the minimum points that solution at different pH values. Micelles can be observed with 156
become the centroids for the band assignments. The amide I AFM for the different pH values under investigation (pH 4, 5, 157
band region is used to derive a baseline passing through the ordi- and 6). In all cases, their sizes were heterogeneous, micelle diam- 158
nates in 1,700, and 1,600 cm−1 and has been adjusted to calculate eters were evaluated manually (NanoScope software) and their 159
the absolute intensity of this band and the components of the sec- range from 60 to 500 nm. This shows that larger micelle agglom- 160
ondary structure. The resulting curve has been analyzed by con- erates were formed at pH 5 as compared to pH 4 and 6 (Figs. 1a– 161
sidering the expected position of bands for the secondary 1c). At pH 5, the protein is rather close to its isoelectric point and 162
structure. The assigned structures for the low-frequency compo- a neutral charge (Holt et al., 2013; Bhat et al., 2016). In turn, this 163
nent in the case of the β-laminar arrangements are the following: promotes both the precipitation of protein and the growth of the 164
1,612−1,640 cm−1, turns: 1,662−1,684 cm−1, α-helix: 1,648 micelles (Uricanu et al., 2004), that is, a large agglomerate is pro- 165
−1,662 cm−1, random coil: 1,640−1,650 cm−1, and to the high- duced. As for pH 4 and 6, the proteins could be soluble due to 166
frequency component for the laminar β arrangements 1,670 having more uniform surfaces and thus fewer micelle agglomer- 167
−1,694 cm−1 (Carbonaro & Nucara, 2010). The assigned bands ates (Figs. 1a, 1c). Also, electrostatic repulsive/attractive forces 168
to a given conformation have then been added and divided by have an important role on micelle stabilization, for example, at 169
the total area of amide I. In the amide II region, a straight baseline native pH (6.8) internal structural changes can be induced into 170
is used passing through the ordinates at 1,500 and 1,560 cm−1 and casein micelle as well as its surface layer, while that at acid condi- 171
adjusted to calculate the intensity of this band and the compo- tions (about pH 4.6, isoelectric point), casein micelles begin to be 172
nents of the secondary structure. The assigned element structures dispersed by dissolving colloidal calcium phosphate and solubili- 173
for the antiparallel laminar β arrangement is 1,510–1,530 cm−1 zation of individual caseins, due to that the net charge on micelle 174
and that for the parallel laminar β arrangement is 1,530– surface´s tend to zero and the charge repulsions among micelles 175
1,550 cm−1 (Pelton & McLean, 2000). The conformation assigned decrease. Consequently, an isoelectric precipitation of casein 176
bands have been added and divided by the total area of amide II micelles with large agglomerates are usually observed. Thus, cal- 177
and then related to the percentage of the high- and low-frequency cium phosphate plays essential role stability and structure of 178
component for the laminar β arrangement of the amide I band. micelle casein. By acidification, the calcium phosphate is released 179
and decreases the number of groups it could affect stiffness of the 180
micelles (Bahri et al., 2017). It could be attributed to kappa casein 181
Statistical and Multivariate Analysis
collapse due to nanoclusters of calcium and phosphate gradually 182
The measurements are expressed as average values with a stan- dissolve off the micelle (Ouanezar et al., 2012). 183
dard deviation accuracy. The data have been compared by using On the other hand, higher Ra values are found at a pH 5 of the 184
the ANOVA-Tukey test, and significant differences are considered casein (see Fig. 2). The Ra at pH 4 is 1.56 ± 0.58 nm and repre- 185
when p < 0.05. Frequency histograms of MD and AR have been sents the lowest found value. Furthermore, it can be seen the Ra 186
performed to fit the data with different distribution functions. at pH 6 oscillates around 2.59 ± 0.84 nm. The Ra values vary 187
The function with the highest R 2 value has been selected as the according to the number of micelle agglomerates that are not dis- 188
best model to describe the data distribution. Both statistical solved in the solution. This is consistent with the characteristic 189
 4 Liliana Edith Rojas‐Candelas et al.

190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
Fig. 1 - Colour online, Colour in print

208
209
210
211
212
213
214
215
216
217
218
Fig. 1. (a–c) Atomic force microscopy (AFM) height images of casein scanned at 2 × 2 μm and (d–f) cryo-TEM micrographs of the casein micelles at different 219
pH values. Scales correspond to 200 nm. Arithmetic average roughness (Ra). In AFM images, same letters indicate that there is a statistically significant difference
220
( p < 0.05).
221
222
223
found size for each pH condition. Thus, these results show that responsible for the total charge of the protein and, allows it to 224
the caseins size as well as the topographic characteristics depend solubilize or precipitate, as herein observed in AFM images. 225
upon the specific pH value used for its precipitation and affects Knudsen & Skibsted (2010) have reported a micellar diameter 226
the different properties of micelles for use in the food industry between 80 and 300 nm. Other works report a micelle diameter 227
(Moitzi et al., 2011). Therefore, the casein topographic structure of 300 to 220 nm at pH 6.7 and pH 5, respectively (Ouanezar 228
most likely can be associated with the differences in size, shape, et al., 2012). An internal structure of casein has been reported 229
and reassembly of the micelles as a result of pH changes. The dis- at pH 6.6 showing size values of around 111 to 264 nm (Bahri 230
tribution of the size and shape of micelles is discussed in detail in et al., 2017). As for the micelles size values observed in this 231
Cryo-TEM results, where the micelles dispersion is particularly work by AFM and cryo-TEM, they are clearly within the range 232
efficient. reported in the literature. 233
Furthermore, significant differences ( p < 0.05) have been 234
detected for a pH 6 in comparison with other pH values. 235
Characterization of Micellar Morphology with Cryo-TEM
Figure 3 shows that the most elongated micelles are produced 236
Table 1 shows the morphometric parameters (MD, R, and AR) with at pH 5. Because at pH 5 lose their surface heterogeneities, 237
and at least 200 micelles were measured for each pH studied. indicating that this structure is susceptible at the net charge 238
Regarding micelle shape, the Cryo-TEM images show clear ten- such as mentioned in Ouanezar et al. (2012) when they worked 239
dency toward circular micelles at pH 6 as compared to other milk casein micelles at different pH values. This condition yields 240
pH values. Table 1 shows roundness (R) and aspect ratio (AR) the highest AR and the lowest R values and also, having signifi- 241
values near to 1 indicating a perfect circular shape. Elongated cant differences ( p < 0.05). The AR correctly describes the elon- 242
micelles have AR values higher than 1. Cryo-TEM micrographs gated nature of the micelles, and thus, it is selected as the 243
and image analysis confirm that the largest micelles are found preferred parameter to describe the distribution of micelle elonga- 244
at a pH 5 (Figs. 1d–1g, Table 1, see values of MD and R). tion instead of the R descriptor. In addition, the AR is also more 245
Moreover, an ANOVA test indicates that there are significant dif- efficient to evaluate the differences between the micelles shape for 246
ferences ( p < 0.05) between these parameters for the different pH the pH values under investigation. Figure 3 shows the frequency 247
values under investigation. Figure 2 shows the frequency histo- histograms that describe the distribution of micelles elongation 248
grams of MD, together with a distribution fit to a Gaussian func- values (AR). They are used to determine a proper fit of the AR 249
tion with R 2 > 0.93 for different pH values. Casein micelles can data and R 2 > 0.92 in all cases by means of the Gaussian function. 250
have many different size values depending on its pH, as this is Nevertheless, micelle shapes turn out to be distributed more 251
 Microscopy and Microanalysis 5

described earlier, Figure 1 shows the micelles at different pH val- 252

ues. It is worth noticing that there are significant differences in 253
shape and size of the micelles (Table 1). The present analysis 254
shows the effects that a systematic variation of pH induces on 255
the casein micelles. 256
257
258

Characterization of Caseins Secondary Structures by Raman 259

Spectroscopy 260
261
The pH plays a pivotal role in the topographic and morphometric 262
characteristics of casein micelles and their structure. Nevertheless, 263
their correlation remains unknown to some degree. This work 264
deals the microstructural characterization of casein micelles and 265
the changes in the secondary structure of casein molecules com- 266
prised within the micelles after treatment at different pH values. 267
Nevertheless, the secondary structure prediction of casein micelle 268
should be used with caution due to interpretations done in the lit- 269
erature. However, the simultaneous identification of secondary 270
components (with the amide I and III bands) greatly enhances 271
our interpretation (Hussain et al., 2011). Figure 4 shows Raman 272
spectra with the characteristic bands corresponding to global sig- 273
nals of the main types of caseins (αs1, αs2, β, and κ) contented 274
within the micelles. There is information relating to the vibra- 275
tional amides bands as well as those for stretching, bending, 276
and other values. The first observed band on the spectrum is 277
due to stretching S-S bonding (760, 880, and 1,361 cm−1). The 278
Fig. 2 - B/W online, B/W in print

corresponding intensity is higher at pH 5 in comparison with 279

those at pH 4 and 6 (Gómez et al., 2013). Another band is related 280
to the tryptophan (830 and 850 cm−1). The lowest intensity is 281
found a pH 5, while for pH 4 and 6, a higher and similar intensity 282
is measured. The tryptophan band refers to the hydrophobicity 283
and only affects the change in the conformations of unfolded pro- 284
tein (Biswas et al., 2003). The next band is tyrosine (1,200 and 285
1,340 cm−1) and rather strong differences in intensity are found. 286
At pH 6, there is a lower intensity, while for pH 4, it is higher. 287

Fig. 2. (a–c) Frequency histograms of the micelle diameter (MD: average values) of
Tyrosine is related to hydrophilicity and its behavior is similar 288

casein at different pH values. Line indicates the fit with the Gaussian function. to that of the tryptophan band. This is due to a competition 289
that exists between hydrophobic and hydrophilic groups in the 290
same residue (Biswas et al., 2003). The characteristic band of pro- 291
asymmetrically than the MD distribution (compare Figs. 2, 3). teins is phenylalanine (1,004 cm−1). It is related to hydrophobic 292
This can be the result of micelles being a collection of heteroge- interactions, a non-covalent dispersion type, as well as interacting 293
neous particles with circular and elongated shapes. The size and with tryptophan (Biswas et al., 2003; Horne, 2017). The phenylal- 294
shape of micelles have been previously investigated, and the pre- anine band is similar to tyrosine for casein. The CH stretch and 295
sent results agree with these works. It has been reported in the lit- flex bands (2,800 and 3,000 cm−1) are similar and related to the 296
erature that the shape of the micelles tends to be spherical. hydrophobic groups of tyrosine (Gómez et al., 2013). The aro- 297
However, image analysis of cryo-TEM images shows that the matic band is related to emulsifying properties, while electroneg- 298
shape is not perfectly spherical or symmetrically distributed (see ativity has polar moieties that stabilize the protein (Biswas et al., 299
also Knudsen & Skibsted, 2010; Yahimi Yazdi et al., 2014). 2003; Gómez et al., 2013; Horne, 2017). 300
Consequently, the micelles have a larger elongation at pH 5 and Secondary structure of individual caseins has been determined 301
also the highest roughness and agglomerate size values. As by FTIR-Raman Spectroscopy, according to Carbonaro & Nucara 302
303
304
305
Table 1. Micelle Casein Morphometric Parameters Analyzed by Transmission Electron Cryo-Microscopy and Image Analysis at Different pH Values.
306
Morphometric parameters pH 4 pH 5 pH 6 307
308
Micelle diameter (MD, nm) 293.98 ± 177.83a,b 327.90 ± 151.47b 325.43 ± 105.24a 309

Roundness (R) 0.870 ± 0.073a,c 0.82 ± 0.08b,c 0.88 ± 0.08a,b 310

311
Aspect ratio (AR) 1.15 ± 0.11b,c 1.22 ± 0.14a,b 1.13 ± 0.12a,c
312
Results were expressed as mean ± standard deviation. In rows, the same letters indicate significant differences ( p < 0.05). 313
 6 Liliana Edith Rojas‐Candelas et al.

β-antiparallel, and turn conformations. Also, in values acid of 314

pH one of the caseins, k-casein expands, while at pH 7.5, the 315
k-casein is collapsed (Mirdha & Chakraborty, 2019). When the 316
protein is folded, it acquired energy transfer properties due to 317
the closeness of the protein–protein interaction (Chakraborty 318
& Basak, 2007). On the other hand, at pH 5, the protein mostly 319
unfolds. This can be seen from the detection of a random coil 320
arrangement structure, a decrease in the content of the α-helix 321
conformation and an increase in both the β-parallel and turn 322
conformations. The partial unfolding of the protein has aggre- 323
gates that are covalently bonded (Moitzi et al., 2011). 324
Furthermore, in the case of pH 5, there are hydrophobic interac- 325
tions that decrease the solubility of the protein (Wang et al., 326
2017). At pH 6, the results are different; the casein protein 327
retains a percentage of random coils but develops a higher con- 328
tent of the turn structure. There is also a reduction of the α-helix 329
conformation structure. This means most likely that the protein 330
is folded and disordered. Such values indicate that there is an 331
activation of sulfhydryl groups in the casein. This is accompa- 332
nied by the unfolding of the protein that improves specific bio- 333
logical functions (Bhat et al., 2016). Moreover, the unfolding 334
protein provides water retention and gelation properties (Li 335
et al., 2014). They found differences in the caseins secondary 336
structure at the different pH values under investigation are sig- 337
nificant because they are related to the size and shape of the 338
micelles. As the results at pH 5 show, the unfolded protein pro- 339
vides the largest micelles, mostly elongated shapes, and largest 340
agglomerates. Alternatively, at pH 4, there are fewer agglomer- 341
ates produced by spherical and smaller micelles due to the pro- 342
tein folding (see Table 1, AFM, cryo-TEM images). Additionally, 343
it is known that the protein is also immobilized at pH 4 (Moitzi 344
et al., 2011). Finally, at pH 6, despite the protein tending to 345
unfold (partially unfolded), it becomes charged which induces 346
a greater solubility as well as smaller agglomerates and micelles 347
Fig. 3 - B/W online, B/W in print

(Lahiri et al., 1999; Uricanu et al., 2004). 348

349
350
Multivariate Analysis
351
The data of mean roughness Ra and morphology of the casein 352
micelles and an overview of its secondary structure at different 353
pH values are integrated through PCA. The PCA can be an idea 354
general the relationship the morphometric parameters and sec- 355
ondary structures such as just one estimate. In Figure 6a, red- 356

Fig. 3. (a–c) Frequency histograms of the aspect ratio (AR: average values) of casein
colored points represent the average factorial scores of the 357
at different pH values. Line indicates the fit with a Gaussian function. two first principal components (PCs), by using the nine struc- 358
tural parameters evaluated for each pH value. They have been 359
reduced to a 2D space using PCA. The factorial scores of the 360
replicates (circles in black color) had not a large dispersion, 361
(2010). The results are shown in Figure 5 for each of the pH val- and average factorial scores values were used to estimate the dis- 362
ues under analysis. As for the amide I region at pH 4, there is an tance between each pH value studied (see black arrow lines in 363
overall reduction of the peaks height that gives rise to a valley in Fig. 6a). The separation between each pH value is noteworthy 364
the extreme left. In the case of pH 5 and 6, there is a heteroge- with higher observed displacements in PC1 and PC2. The dis- 365
neous distribution of peaks together with the generation of new tances (d ) between each of the pH values reveal similarities 366
valleys. The amide II region shows a homogeneous response for between them. Results at pH 5 are found to be closest to 367
pH 4. In the case of pH 5, there is an increase in peaks height, those at pH 6 (d(pH 6–pH 5) = 4.21). This is most likely due 368
for pH 6, a decrease is found most likely due to the signal expan- to charges in the protein at pH 6 and 5. For pH 4, there is 369
sion of the spectrum. a balance of positive and negative charges in order to reach 370
The final results for the characterization as a global overview the isoelectric point of the protein, which then produces 371
of the secondary structure of individual caseins contained in the precipitation of the protein. The distance of pH 4 with the 372
micelles as a function of pH are given in Table 2. At pH 4, the other pH values is rather long (d(pH 6–pH 4) = 5.67, d(pH 373
protein is partly unfolded, since the percentage for α-helix 5–pH 4) = 5.54). This means that most of its evaluated param- 374
confirmations is almost 50% as compared to β-parallel, eters show rather significant differences as compared to the 375
 Microscopy and Microanalysis 7

376
377
378
379
380
381
382
383
384
385
386
387
Fig. 4 - Colour online, Colour in print

388
389
390
391
392
393
394
395
396
397
398
Fig. 4. Raman spectra of caseins at different pH values. 399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
Fig. 5 - Colour online, Colour in print

425
426
427
428
429
430
431
432
433
434
435
Fig. 5. Changes in Raman spectra at indicated pH values for the amide II band (a) and amide band I (b) regions of casein. Represents peak-fitting of the second 436
derivative curves of the spectra. The peaks-fitting corresponding for each secondary structure identified (β-antiparallel, β-parallel, turn, random coil, and α-helix) in
437
the protein micelles are indicated with lines and arrows in different colors.
 8 Liliana Edith Rojas‐Candelas et al.

Table 2. Contribution Percentages Determined by Raman Spectroscopy of Also, a positive and moderate correlation (r = 0.776) between 438
Different Regions the Casein Secondary Structure at Different pH Values. random coil contribution and MD is determined. Random coil 439

Protein structures pH 4 (%) pH 5 (%) pH 6 (%) is an external region of the casein secondary structure, its abun- 440
dance is associated with unfolded caseins at its isoelectric point 441
β-Antiparallel 6.79 3.20 2.84 (Painter & Koenig, 1976; Pelton & McLean, 2000). Thus, for a 442

β-Parallel 4.14 3.56 6.62

high contribution of this region, large and unfold micelles can 443
be observed. Consequently, Ra and MD increase when they are 444
Turn 40.32 30.96 52.02 analyzed by AFM and TEM. At different pH values, the R of 445
Random coil 0 33.92 10.24 micelles shows a positive correlation with turn regions and a 446

α-Helix 48.74 28.33 28.28

negative correlation with random coil regions. By considering 447
that a higher contribution of the turn regions means that the 448
casein micelles tend to compact, then the direct observation of 449
a higher number of round micelles can be explained. 450
other values of pH (Fig. 6a). Thus, according to the current Additionally, the presence of more turn regions can also be 451
results, the microstructural parameters of micelles as well as linked to from a point of view overall of the secondary structure 452
an overall view of caseins secondary structure at different of folded caseins (Farrell et al., 1993; Pelton & McLean, 2000; 453
pH values are strongly related to the protein restructuration, Herrero et al., 2008; Wang et al., 2017). Conversely, when the 454
reassembly, and aggregation of the micelles through the effect random coil contribution increases then more relaxed and larg- 455
of pH. est micelles are generated. In addition and for the case of differ- 456
Additionally, a Pearson analysis is presented to elucidate the ent pH values, the conformational changes on β-antiparallel and 457
correlation between the changes observed at different pH values α-helix regions have had an important influence on the MD. 458
in the microstructural parameters of the micelles (Ra, DM, R, Negative correlations between MD and β-antiparallel and 459
AR) and on overall overview of the secondary structure of casein α-helix regions have been observed since these regions are an 460
regions (β-parallel, β-antiparallel, turn, random coil, α-helix). overview of folded regions the secondary structure of the pro- 461
Figure 6b shows an image of a Pearson correlation matrix, teins (Pelton & McLean, 2000). When their contribution is 462
where the range of its correlation coefficients (r) is shown. higher, folded micelles with lower diameters are produced. 463
According to the Pearson criterion, a value of r > 0.75 indicates Finally, a major contribution of random coil region yields elon- 464
a very high correlation (Asuero et al., 2006). Figure 6b shows in gated micelles (large AR values). Otherwise, higher contribu- 465
green or red the highest positive or negative correlations for the tions of turn regions produce folded and not elongated 466
present results, that is, positive or negative correlations are indi- micelles. In the case of the β-parallel region, there is a moderate 467
cated in green (r > 0.818) and red (r > −0.818) intensities in the positive correlation with R (r = 0.757) and negative with AR 468
Pearson matrix. This strict criterion has been used to select the (0.793). Thus, at pH 4 and 6, the contribution of β-parallel 469
variables with major correlation, and they are indicated with region is higher and more rounded micelles are produced due 470
crosses in Figure 6b. On the other hand, moderate correlations to major protein folding (Painter & Koenig, 1976). These find- 471
correspond to values between ±0.818 to ±0.636 (indicated in ings are associated with a high correlation between the pH val- 472
green and red color with lower intensity than those marked ues, an overall view of conformational restructuration of 473
with crosses). For example, a high positive correlation between different protein regions, and morphometric changes of studied 474
Ra and random coil contribution is found at different pH values. micelles. 475
476
477
478
479
480
481
482
483
484
485
Fig. 6 - Colour online, Colour in print

486
487
488
489
490
491
492
493
494
495
496
Fig. 6. (a) 2D plot of factorial scores obtained by analysis principal components using morphometric and secondary structural parameters of casein at different pH
497
values. (a–d) The replicates used. ave: average value. (b) Image of the Pearson correlation matrix using nine variables: arithmetic mean roughness (Ra), roundness
(R), micelle diameter (MD), aspect ratio (AR), and the contributions of the secondary conformational structures from different casein regions (β-antiparallel, 498
β-parallel, turn, random coil, and α-helix). Distances (d ) and principal component (PC). 499
Microscopy and Microanalysis 9

Conclusion Hussain R, Gaiani C, Aberkane L, Ghanbaja J & Scher J (2011). Multiscale 500
characterization of casein micelles under NaCl range conditions. Food 501
In this work, the microstructure of casein micelles at different pH Biophys 6, 503–511. 502
values are characterized quantitatively by AFM and Cryo-TEM. Jahaniaval F, Kakuda Y, Abraham V & Marcone MF (2000). Soluble protein
503
As evaluated by Raman spectroscopy, the variation in the Ra fractions from pH and heat treated sodium caseinate: Physicochemical and
504
and morphometry of micelles were explained in terms of the over- functional properties. Food Res Int 33, 637–647.
505
all conformational modifications of their caseins secondary struc- Knudsen JC & Skibsted LH (2010). High pressure effects on the structure of
casein micelles in milk as studied by cryo-transmission electron microscopy. 506
tures. By considering that, in the past, the study of caseins
Food Chem 119, 202–208. doi:10.1016/j.foodchem.2009.06.017 507
structures has been made by considering isolated parameters,
Lahiri J, Isaacs L, Tien J & Whitesides GM (1999). A strategy for the gener- 508
this work attempts to integrate spectroscopy, microscopy tech-
ation of surfaces presenting ligands for studies of binding based on an active 509
niques, and multivariate analysis to correlate straightforwardly ester as a common reactive intermediate: A surface plasmon resonance 510
the structural changes that occur at the microstructural and study. Anal Chem 71, 777–790. 511
molecular levels on the micelles and a partial view of the changes Li K, Kang Z-L, Zhao YY, Xu XL & Zhou G-H (2014). Use of high-intensity 512
in the secondary structure of the caseins contained in the micelles ultrasound to improve functional properties of batter suspensions pre- 513
studied at different pH values. In the future, research will require pared from PSE-like chicken breast meat. Food Bioprocess Technol 7,
514
to separate the casein to know who is attributed this behavior. 3466–3477.
515
This information can be valuable for understanding the behavior Liu Y, Zhao G, Zhao M, Ren J & Yang B (2012). Improvement of functional
properties of peanut protein isolate by conjugation with dextran through mail- 516
of the casein micelles and selecting adequate pH conditions to
lard reaction. Food Chem 131, 901–906. doi:10.1016/j.foodchem.2011.09.074 517
improve some quality properties in dairy products.
Lucey JA & Singh H (1997). Formation and physical properties of acid milk 518
Acknowledgments. Liliana Edith Rojas Candelas wishes to thank gels: A review. Food Res Int 30, 529–542. 519
CONACyT, BEIFI, and Instituto Politécnico Nacional (IPN) in Mexico City Mc Sweeney PLH & O Mahony JA (2016). Advanced Dairy Chemistry Volume 520
for the scholarship provided during her PhD studies, and the financial support 1B: Proteins: Applied Aspects, 4th ed. New York, Heidelberg, Dordrecht, 521
provided by CONACyT (239899, 268660) and Secretaria de Investigación y London: Springer. 522
Posgrado at IPN (20180455, 20195198, 20200505, 20200506) projects. Michael Byler D, Farrell HM & Susi H (1988). Raman spectroscopic study of
523
casein structure. J Dairy Sci 71, 2622–2629.
524
Mirdha L & Chakraborty H (2019). International journal of biological mac-
525
romolecules characterization of structural conformers of κ-casein utilizing
References fluorescence spectroscopy. Int J Biol Macromol 131, 89–96. doi:10.1016/ 526
j.ijbiomac.2019.03.040 527
Asuero AG, Sayago A & González AG (2006). The correlation coefficient: An
overview. Crit Rev Anal Chem 36, 41–59. Moitzi C, Menzel A, Schurtenberger P & Stradner A (2011). The pH 528
Bahri A, Martin M, Gergely C, Marchesseau S & Chevalier-Lucia D (2018). induced sol-gel transition in skim milk revisited. a detailed study using 529
Topographical and nanomechanical characterization of casein nanogel par- time-resolved light and X-ray scattering experiments. Langmuir 27, 2195– 530
ticles using atomic force microscopy. Food Hydrocoll 83, 53–60. 2203. 531
doi:10.1016/j.foodhyd.2018.03.029 Ouanezar M, Guyomarc’h F & Bouchoux A (2012). AFM imaging of milk 532
Bahri A, Martin M, Gergely C, Pugnière M, Chevalier-Lucia D & Marchesseau casein micelles: Evidence for structural rearrangement upon acidification. 533
S (2017). Atomic force microscopy study of the topography and nanome- Langmuir 28, 4915–4919.
534
chanics of casein micelles captured by an antibody. Langmuir 33, 4720–4728. Painter PC & Koenig JL (1976). Raman spectroscopic study of the proteins of
535
Bhat MY, Dar TA & Singh LR (2016). Casein proteins: structural and func- egg white. Biopolymers 15, 2155–2166.
Pan K & Zhong Q (2015). Amyloid-like fibrils formed from intrinsically dis- 536
tional aspects. In Milk Proteins - From Structure to Biological Properties and
ordered caseins: Physicochemical and nanomechanical properties. Soft 537
Health Aspects, pp. 1–18.
Biswas KM, DeVido DR & Dorsey JG (2003). Evaluation of methods for Matter 11, 5898–5904. doi:10.1039/C5SM01037C 538
measuring amino acid hydrophobicities and interactions. J Chromatogr, A Pelton JT & McLean LR (2000). Spectroscopic methods for analysis of protein 539
1000, 637–655. secondary structure. Anal Biochem 277, 167–176. 540
Carbonaro M & Nucara A (2010). Secondary structure of food proteins by Qi PX (2007). Studies of casein micelle structure: The past and the present. 541
Fourier transform spectroscopy in the mid-infrared region. Amino Acids Dairy Sci Technol 87, 363–383. 542
38, 679–690. Riché E, Carrié A, Andin N & Mabic S (2006). A P P L I C A T I O N N O T 543
Chakraborty A & Basak S (2007). pH-induced structural transitions of E New Products Crystallization System. www.aln20.com/newproducts.
544
caseins. J Photochem Photobiol, B 87, 191–199. Rojas-Candelas LE, Chanona-Pérez JJ, Méndez Méndez JV, Resendiz-Mora
545
Farrell HM, Brown EM & Kumosinski TF (1993). Three-dimensional molec- CA, Calderón-Benavides HA & Cervantes-Sodi F (2019). Microstructural
characterization of casein by microscopy and spectroscopy techniques. IEEE 546
ular modeling of bovine caseins: A refined, energy-minimized κ-casein
Access 25, 490–491. 547
structure. J Dairy Sci 76, 2507–2520.
Gómez AV, Ferrer EG, Añón MC & Puppo MC (2013). Changes in second- Smeller L (2002). Pressure-temperature phase diagrams of biomolecules. 548
ary structure of gluten proteins due to emulsifiers. J Mol Struct 1033, 51–58. Biochim Biophys Acta - Protein Struct Mol Enzymol 1595, 11–29. 549
Herrero AM, Cambero MI, Ordóñez JA, de la Hoz L & Carmona P (2008). Solowiej B, Mleko S & Gustaw W (2008). Physicochemical properties of acid 550
Raman spectroscopy study of the structural effect of microbial transglutami- casein processed cheese analogs obtained with different whey products. 551
nase on meat systems and its relationship with textural characteristics. Food Milchwissenschaft 63, 299–302. 552
Chem 109, 25–32. Swaisgood HE (1993). Review and update of casein chemistry. J Dairy Sci 76, 553
Holt C, Carver JA, Ecroyd H & Thorn DC (2013). Invited review: Caseins 3054–3061. doi:10.3168/jds.S0022-0302(93)77645-6
554
and the casein micelle: Their biological functions, structures, and behavior Uricanu VI, Duits MHG & Mellema J (2004). Hierarchical networks of casein
555
in foods. J Dairy Sci 96, 6127–6146. doi:10.3168/jds.2013-6831 proteins: An elasticity study based on atomic force microscopy. Langmuir
20, 5079–5090. 556
Horne D (2017). A balanced view of casein interactions. Curr Opin Colloid
Vié V, Moreno J, Beaufils S, Lesniewska E, Léonil J & Renault A (2002). 557
Interface Sci 28, 74–86. doi:10.1016/j.cocis.2017.03.009
Horne DS (2002). Casein structure, self-assembly and gelation. Curr Opin Interfacial behavior of goat kappa casein: Ellipsometry and atomic force 558
Colloid Interface Sci 7, 456–461. microscopy study. Single Molecules 3, 127–133. 559
Horne DS (2006). Casein micelle structure: Models and muddles. Curr Opin Wang K, Sun D-W, Pu H & Wei Q (2017). Principles and applications of 560
Colloid Interface Sci 11, 148–153. spectroscopic techniques for evaluating food protein conformational 561
10 Liliana Edith Rojas‐Candelas et al.

changes: A review. Trends Food Sci Technol 67, 207–219. doi:10.1016/ Yahimi Yazdi S, Corredig M & Dalgleish DG (2014). Studying the structure 562
j.tifs.2017.06.015 of β-casein-depleted bovine casein micelles using electron microscopy and 563
Waninge R, Kalda E, Paulsson M, Nylander T & Bergenståhl B (2004). fluorescent polyphenols. Food Hydrocoll 42, 171–177. doi:10.1016/ 564
Cryo-TEM of isolated milk fat globule membrane structures in cream. j.foodhyd.2014.03.022
565
Phys Chem Chem Phys 6, 1518–1523. Yang M, Cui N, Fang Y, Shi Y, Yang J & Wang J (2015). Influence of succi-
566
Xiao-zhou S, Hong-ru W & Mian H (2014). Characterization of the casein/ nylation on the conformation of yak casein micelles. Food Chem 179, 246–
567
keratin self-assembly nanomicelles. J Nanomater 2014, 1–7. 252.
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623