Structural Analysis of Breast-Milk αS1-Casein: An α-Helical Conformation Is Required for TLR4-Stimulation

Breast-milk αS1-casein is a Toll-like receptor 4 (TLR4) agonist, whereas phosphorylated αS1-casein does not bind TLR4. The objective of this study was to analyse the structural requirements for these effects. In silico analysis of αS1-casein indicated high α-helical content with coiled-coil characteristics. This was confirmed by CD-spectroscopy, showing the α-helical conformation to be stable between pH 2 and 7.4. After in vitro phosphorylation, the α-helical content was significantly reduced, similar to what it was after incubation at 80 °C. This conformation showed no in vitro induction of IL-8 secretion via TLR4. A synthetic peptide corresponding to V77-E92 of αS1-casein induced an IL-8 secretion of 0.95 ng/mL via TLR4. Our results indicate that αS1-casein appears in two distinct conformations, an α-helical TLR4-agonistic and a less α-helical TLR4 non-agonistic conformation induced by phosphorylation. This is to indicate that the immunomodulatory role of αS1-casein, as described before, could be regulated by conformational changes induced by phosphorylation.


Introduction
Human α S1 -casein is a breast-milk protein, which was described to be overexpressed in the synovia of patients with osteoarthritis or rheumatoid arthritis [1,2], the blood of patients with multiple sclerosis [3], in the tissue of patients with benign prostate hyperplasia [4] or breast cancer [5].Furthermore, α S1 -casein was reported to have an immunostimulatory role, as it appeared to induce a lifelong human immunoglobulin G (IgG)-antibody response due to breastfeeding [6] and secretion of several cytokines via Toll-like receptor 4 (TLR 4), e.g., granulocyte macrophage colony-stimulating factor, interleukin 1β, interleukin 6 and chemokine IL-8 (interleukin 8) [7,8].The α S1 -casein-induced cytokine secretion via TLR4 was blocked by phosphorylation with human protein kinase CK2 [8].Therefore, it was postulated that α S1 -casein could have two functions, a TLR4-agonistic, most likely immune modulatory function when unphosphorylated and a mere nutritional function when phosphorylated [9].
Human α S1 -casein and its primary structure were discovered lately after other caseins in breast milk [10,11].The studies focused on the quantification of α S1 -casein in breast milk, its posttranscriptional [11,12] and posttranslational modification.Hints for oligomers and disulfide bonds were mentioned.Furthermore, the authors described α S1 -casein in the matrix of breast milk as more stable than other breast-milk proteins, in particular with respect to acidic pH, breast-milk proteases and Holder pasteurization [11][12][13][14][15].A higher state of organization, however, in the structure of, e.g., purified human α S1 -casein was not investigated.A reason for this could be the low content of breast-milk α S1 -casein with less than 1% (3 mg/L to 537 mg/L) of total milk protein content [16].
The aim of this study was the characterization of human α S1 -casein structure in relation to its two functions, immune active (TLR4-agonistic) and nutritional (nonagonistic).The question to clarify was, can the two functions of α S1 -casein be related to two distinct conformations and is the switch between the two conformations eventually due to phosphorylation.To some extent, the structure of bovine and human α S1 -casein was compared to gain further insights into α S1 -casein species specificity.This could help to understand the role of human α S1 -casein, its phosphorylation-dependent modulation and regulation of immune activity in early infancy and pathogenesis.Moreover, it would give a prime example of how two completely different functions of a single protein are regulated by phosphorylation or dephosphorylation, respectively.

Results and Discussion
2.1.In Silico and Experimental Structure Analysis of Human α S1 -Casein 2.1.1.Amino Acid Sequence Analysis and Phylogenetic Relationships First, the amino acid sequence (AAS) of human α S1 -casein was compared to the AAS of α S1 -caseins from 17 other species using molecular evolutionary genetics analysis software [28], with results shown in Figures 1 and S1.Overall, the phylogeny of α S1 -casein was in accordance with the phylogenetic tree obtained by analysing 1% of the human genome [29].Exceptions were mouse and rat α S1 -casein, which showed only low AAS homology to rabbit α S1 -casein.AAS of human α S1 -casein had an identity of more than 92% to AAS of non-human hominids such as α S1 -caseins from chimpanzee (97%), western lowland gorilla C95%), and Sumatran orangutan (92%).AAS of primate Chlorocebus sabaeus (green monkey) had an identity of 82% to human α S1 -casein but exhibited no probability for a coiled-coil region.All other species had an identity below 50% to human α S1 -casein, e.g., rabbit (42%), mouse (29%) and rat (27%), donkey and dromedary (39%), cow (33%) and pig (35%).Furthermore, a coiled-coil domain was predicted for α S1 -casein in the same AAS region in all four hominids investigated (Figure S1).A coiled-coil region was also predicted for rabbit, guinea pig and donkey α S1 -casein.But here, the coiled-coil was predicted at a different position in the AAS compared to human α S1 -casein.In conclusion, high identity in the AAS and secondary structure elements as a coiled-coil in hominids could be a hint that these α S1 -casein share structure and, consequently, function which could be different from other mammals.
Figure 1.Phylogenetic tree as obtained by comparison of human αS1-casein amino acid sequence with that from 17 other species (red: hominids; blue: bovidae).It was calculated using molecular evolutionary genetic analysis.

Hydrophobicity Analysis of Human αS1-Casein and Comparison with Bovine αS1-Casein
Next, a hydrophobicity plot of human αS1-casein was calculated Expasy (SIB, Laussane, Switzerland) with Kyte and Doolittle coefficients [30], as illustrated in Figure 2 and compared to bovine αS1-casein.This hydrophobicity plot could give an indication of (1) which AA of human αS1-casein could be exposed, (2) whether S 33 , S 41 , S 71 and S 89 could be accessible for phosphorylation, and (3) which parts of αS1-casein were more hydrophobic and could be involved in intra-and/or intermolecular binding (e.g., to other human caseins).Human αS1-casein contained more hydrophilic AA at the N-terminus, especially at AAS 30-45, and the C-terminus (at AAS 170-185) compared to bovine αS1-casein.Human αS1-casein was more hydrophobic at positions 83-102 compared to bovine αS1-casein.This was the region a helix-loop-helix motive (SSS-X-EE) reported to be highly conserved in αS1-casein [11].Although this region is highly conserved, the neighbouring AA of this motive shows weak similarity between human and bovine αS1-casein.AAS 30-45 containing two phosphorylation sites (S 33 and S 41 ) could be exposed more to hydrophilic solutions compared to S 89 .Therefore, S 33 and S 41 could be more easily accessible for kinases.In accordance, it was found before that S 33 and S 41 of human αS1-casein [31] and stretch 85-102 of bovine αS1-casein were phosphorylated in significant amounts.Opposite to this, only low levels of phosphorylation were described for the hydrophobic stretch 83-102 of human αS1-casein, especially for S 89 [14,31].The low level of phosphorylation in stretch 83-102 of human αS1-casein indicated that it was less accessible for kinases, although it was found to be more hydrophobic than in bovine αS1-casein.Less accessibility of kinases could be caused by intra-or intermolecular binding of human αS1-casein stretch 83-102.In total, human αS1-casein contained more hydrophilic AA compared to bovine αS1-casein.The differences in hydrophobicity and the phosphorylation of human and bovine αS1-casein, as reported, suggested that they interact with other proteins or with themselves in a different manner.Next, a hydrophobicity plot of human α S1 -casein was calculated Expasy (SIB, Laussane, Switzerland) with Kyte and Doolittle coefficients [30], as illustrated in Figure 2 and compared to bovine α S1 -casein.This hydrophobicity plot could give an indication of (1) which AA of human α S1 -casein could be exposed, (2) whether S 33 , S 41 , S 71 and S 89 could be accessible for phosphorylation, and (3) which parts of α S1 -casein were more hydrophobic and could be involved in intra-and/or intermolecular binding (e.g., to other human caseins).Human α S1 -casein contained more hydrophilic AA at the N-terminus, especially at AAS 30-45, and the C-terminus (at AAS 170-185) compared to bovine α S1 -casein.Human α S1 -casein was more hydrophobic at positions 83-102 compared to bovine α S1 -casein.This was the region a helix-loop-helix motive (SSS-X-EE) reported to be highly conserved in α S1 -casein [11].Although this region is highly conserved, the neighbouring AA of this motive shows weak similarity between human and bovine α S1 -casein.AAS 30-45 containing two phosphorylation sites (S 33 and S 41 ) could be exposed more to hydrophilic solutions compared to S 89 .Therefore, S 33 and S 41 could be more easily accessible for kinases.In accordance, it was found before that S 33 and S 41 of human α S1 -casein [31] and stretch 85-102 of bovine α S1 -casein were phosphorylated in significant amounts.Opposite to this, only low levels of phosphorylation were described for the hydrophobic stretch 83-102 of human α S1 -casein, especially for S 89 [14,31].The low level of phosphorylation in stretch 83-102 of human α S1 -casein indicated that it was less accessible for kinases, although it was found to be more hydrophobic than in bovine α S1 -casein.Less accessibility of kinases could be caused by intra-or intermolecular binding of human α S1 -casein stretch 83-102.In total, human α S1 -casein contained more hydrophilic AA compared to bovine α S1 -casein.The differences in hydrophobicity and the phosphorylation of human and bovine α S1casein, as reported, suggested that they interact with other proteins or with themselves in a different manner.

Figure 2.
Comparison of the hydrophilicity plots of human (grey) and bovine αS1-casein (red) with signal peptide using Kyte Doolittle scale (hydrophobic > 0; hydrophilic < 0; window of nine amino acids; Expasy (SIB, Lausanne, Switzerland)).Black lines showing positions where human αS1-casein showed higher hydrophobicity and violet lines showing positions where bovine αS1-casein showed higher hydrophobicity.

In Silico Secondary Structure Analysis of Human αS1-Casein
Finally, secondary structures of human αS1-casein were predicted by RaptorX protein simulation [32,33], as shown in Figure 3.The probability for α-helical content of αS1-casein was highest at AAS S 41 -V 77 and AAS S 90 -E 121 , as well as the probability for random coils at AAS R 16 -E 40 , A 78 -S 89 , and V 158 -W 185 .Moreover, the probability for a β-sheet content was predicted at AAS I 123 -P 135 .In comparison to bovine αS1-casein (Figure S2), human αS1-casein appeared to have a higher α-helical content and longer α-helical stretches.For illustration, the first model of human αS1-casein based was constructed, using the PDB (protein data bank)-based web service RaptorX (Figure 3B).The model was in accordance with the secondary structure predictions, supporting a high α-helical and a high random coil content in αS1-casein.A p-value below 0.001 is an indicator that the model is accurate in describing an α-helical protein [32,33].However, the accuracy of the secondary structure prediction was low in our case because the template as used (PDB: 5DFZ) had low similarity to human αS1-casein, with a p-value of 0.019.Until now, the tertiary structure of αS1casein is unknown and lacks any better similarity to a known structure of the PDB or other structures described in the dark proteome databank [34].Finally, secondary structures of human α S1 -casein were predicted by RaptorX protein simulation [32,33], as shown in Figure 3.The probability for α-helical content of α S1 -casein was highest at AAS S 41 -V 77 and AAS S 90 -E 121 , as well as the probability for random coils at AAS R 16 -E 40 , A 78 -S 89 , and V 158 -W 185 .Moreover, the probability for a β-sheet content was predicted at AAS I 123 -P 135 .In comparison to bovine α S1 -casein (Figure S2), human α S1 -casein appeared to have a higher α-helical content and longer α-helical stretches.For illustration, the first model of human α S1 -casein based was constructed, using the PDB (protein data bank)-based web service RaptorX (Figure 3B).The model was in accordance with the secondary structure predictions, supporting a high α-helical and a high random coil content in α S1 -casein.A p-value below 0.001 is an indicator that the model is accurate in describing an α-helical protein [32,33].However, the accuracy of the secondary structure prediction was low in our case because the template as used (PDB: 5DFZ) had low similarity to human α S1 -casein, with a p-value of 0.019.Until now, the tertiary structure of α S1 -casein is unknown and lacks any better similarity to a known structure of the PDB or other structures described in the dark proteome databank [34].Finally, secondary structures of human αS1-casein were predicted by RaptorX protein simulation [32,33], as shown in Figure 3.The probability for α-helical content of αS1-casein was highest at AAS S 41 -V 77 and AAS S 90 -E 121 , as well as the probability for random coils at AAS R 16 -E 40 , A 78 -S 89 , and V 158 -W 185 .Moreover, the probability for a β-sheet content was predicted at AAS I 123 -P 135 .In comparison to bovine αS1-casein (Figure S2), human αS1-casein appeared to have a higher α-helical content and longer α-helical stretches.For illustration, the first model of human αS1-casein based was constructed, using the PDB (protein data bank)-based web service RaptorX (Figure 3B).The model was in accordance with the secondary structure predictions, supporting a high α-helical and a high random coil content in αS1-casein.A p-value below 0.001 is an indicator that the model is accurate in describing an α-helical protein [32,33].However, the accuracy of the secondary structure prediction was low in our case because the template as used (PDB: 5DFZ) had low similarity to human αS1-casein, with a p-value of 0.019.Until now, the tertiary structure of αS1casein is unknown and lacks any better similarity to a known structure of the PDB or other structures described in the dark proteome databank [34].(B) Prediction of tertiary structure of human α S1 -casein using RaptorX.

Prediction and Probability for Intrinsically Disordered Regions or Transmembrane Domains
A lack of similarity to any known structure can be due to an incomplete characterization of a protein.Nonpredicted transmembrane domains or intrinsically disordered regions could mislead structural comparison.Hence, the AAS of human α S1 -casein was analyzed for the probability of acquiring an intrinsically disordered conformation (Figure 4A) and whether such a region of disorder could be set into relation of a possible phosphorylation site using three algorithms of PONDR [35] (Molecular Kinetics Inc., Indianapolis, IN, USA).With high probability, human α S1 -casein was predicted to be an intrinsically disordered protein.All phosphorylation sites of human α S1 -casein known so far were located within AAS 16-125 with a high probability of being intrinsically disordered.A transmembrane domain could be excluded when the AAS of human α S1 -casein was analyzed with the algorithm of EXPASY (SIB, Lausanne, Switzerland), described by Zhao and London [36], as shown in Figure 4B.A lack of similarity to any known structure can be due to an incomplete characterization of a protein.Nonpredicted transmembrane domains or intrinsically disordered regions could mislead structural comparison.Hence, the AAS of human αS1-casein was analyzed for the probability of acquiring an intrinsically disordered conformation (Figure 4A) and whether such a region of disorder could be set into relation of a possible phosphorylation site using three algorithms of PONDR [35] (Molecular Kinetics Inc., Indianapolis, IN, USA).With high probability, human αS1-casein was predicted to be an intrinsically disordered protein.All phosphorylation sites of human αS1-casein known so far were located within AAS 16-125 with a high probability of being intrinsically disordered.A transmembrane domain could be excluded when the AAS of human αS1-casein was analyzed with the algorithm of EXPASY (SIB, Lausanne, Switzerland), described by Zhao and London [36], as shown in Figure 4B.
It can be assumed that αS1-casein present in colostrum breast milk is unphosphorylated because, in a former analysis, high ratios of peptides with phosphorylation sites were detected in their unphosphorylated form [31].

Secondary Structure Analysis of Human αS1-Casein by Spectroscopic Methods
Therefore, recombinant αS1-casein from E. coli, which is a priori not phosphorylated, was used for experimental verification of the structure obtained in silico, as described above.For this purpose, the secondary structure of recombinant human αS1-casein (12.5 µM in 10 mM NaH2PO4/Na2HPO4, pH: 7.4, 20 °C) produced in E. coli was analyzed by CDspectroscopy.As shown in Figure 5A, the spectrum as obtained showed typical minima for high α-helical content of a protein at 207 nm (−729,060 Grad•cm 2 •dmol −1 ) and 222 nm (−489,105 Grad•cm 2 •dmol −1 ) with a ratio (222 nm/208 nm) of 0.74 (Figure 5A).To estimate It can be assumed that α S1 -casein present in colostrum breast milk is unphosphorylated because, in a former analysis, high ratios of peptides with phosphorylation sites were detected in their unphosphorylated form [31].

Secondary Structure Analysis of Human α S1 -Casein by Spectroscopic Methods
Therefore, recombinant α S1 -casein from E. coli, which is a priori not phosphorylated, was used for experimental verification of the structure obtained in silico, as described above.For this purpose, the secondary structure of recombinant human α S1 -casein (12.5 µM in 10 mM NaH 2 PO 4 /Na 2 HPO 4 , pH: 7.4, 20 • C) produced in E. coli was analyzed by CDspectroscopy.As shown in Figure 5A, the spectrum as obtained showed typical minima for high α-helical content of a protein at 207 nm (−729,060 Grad•cm 2 •dmol −1 ) and 222 nm (−489,105 Grad•cm 2 •dmol −1 ) with a ratio (222 nm/208 nm) of 0.74 (Figure 5A).To estimate secondary structural features, the CD spectra obtained were compared to the protein data bank using K2D3 [37].The CD-spectrum of α S1 -casein showed a high proportion of 91.3% of α-helical content.Minima at 207 nm showed a more than 8 times more intense molar ellipticity compared to predicted spectra.Such high molar ellipticity suggests specific molecular interactions, which could be reasoned in intra-or intermolecular helix-helix interaction [38].Overall, CD-spectra of human α S1 -casein indicated a higher α-helical content in comparison to recombinant bovine α S1 -casein, which was reported before to have only a minimum at 205 nm [23], typical for proteins with low or no α-helical content.
a minimum at 205 nm [23], typical for proteins with low or no α-helical content.
In consequence, a further secondary analysis of human αS1-casein was performed by ATR-FTIR (Attenuated Total Reflection-Fourier Transformation Infra-Red spectroscopy, Figure 5B) because it is known that CD-spectroscopy tends to overrate α-helical content and to underestimate β-sheet content of proteins.In the spectra as obtained, the resolution of the amide band I (1600 to 1700 cm −1 ) showed a maximum at 1661 cm −1 , which was a strong indication for a 310 α-helical structure.Nevertheless, a maximum at 1625 cm −1 indicated that αS1-casein was at least partially β-sheet structured.To sum up, in silico structure prediction, CD and ATR-FTIR spectra concordantly indicated a high α-helical and partially random coil content of human αS1-casein, and in addition, that it is partially β-sheet structured.The in silico structure analysis suggested that αS1-casein is an intrinsically disordered protein.Moreover, the secondary structure, as well as degree and pattern of phosphorylation of human αS1-casein was different from bovine αS1-casein hypothetically, resulting in a difference in function. .By deconvolution, the typical peaks of the amide-I band were obtained (dotted line), which showed maxima at 1452 nm, 1557 nm, 1590 nm, 1625 nm and 1661 nm.

Oligomerization of Human αS1-Casein
Heteromers of human αS1-casein, e.g., with κ-casein [14] or milk micelles containing αS1-casein and other caseins [39] have been reported.In order to find out, at first, whether the high molar ellipticity would be a result of intermolecular interaction of αS1-casein with itself, we tried to substantiate oligomers of αS1-casein by SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis), MST (microscale thermophoresis) and SPR (surface plasmon resonance spectroscopy).
For oligomer analysis by SDS-PAGE, human αS1-casein (50 µM) was diluted in sample buffers under different denaturing conditions (Figure 6A): 0.2% SDS (20 min, RT), 2% SDS [20 min, RT] and 2% SDS (100 mM dithiothreitol, DTT, 20 min, 95 °C).After incubation at low SDS content and lower temperatures SDS-PAGE of αS1-casein led to prominent bands at molecular weights of 25, 55, 70 kDa, as well as several bands at even higher molecular weights.These bands were assigned to αS1-casein monomers, dimers, trimers and oligomers of higher order.The intensity of bands assigned to oligomers disappeared with In consequence, a further secondary analysis of human α S1 -casein was performed by ATR-FTIR (Attenuated Total Reflection-Fourier Transformation Infra-Red spectroscopy, Figure 5B) because it is known that CD-spectroscopy tends to overrate α-helical content and to underestimate β-sheet content of proteins.In the spectra as obtained, the resolution of the amide band I (1600 to 1700 cm −1 ) showed a maximum at 1661 cm −1 , which was a strong indication for a 3 10 α-helical structure.Nevertheless, a maximum at 1625 cm −1 indicated that α S1 -casein was at least partially β-sheet structured.To sum up, in silico structure prediction, CD and ATR-FTIR spectra concordantly indicated a high α-helical and partially random coil content of human α S1 -casein, and in addition, that it is partially βsheet structured.The in silico structure analysis suggested that α S1 -casein is an intrinsically disordered protein.Moreover, the secondary structure, as well as degree and pattern of phosphorylation of human α S1 -casein was different from bovine α S1 -casein hypothetically, resulting in a difference in function.

Oligomerization of Human α S1 -Casein
Heteromers of human α S1 -casein, e.g., with κ-casein [14] or milk micelles containing α S1 -casein and other caseins [39] have been reported.In order to find out, at first, whether the high molar ellipticity would be a result of intermolecular interaction of α S1 -casein with itself, we tried to substantiate oligomers of α S1 -casein by SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis), MST (microscale thermophoresis) and SPR (surface plasmon resonance spectroscopy).
For oligomer analysis by SDS-PAGE, human α S1 -casein (50 µM) was diluted in sample buffers under different denaturing conditions (Figure 6A): 0.2% SDS (20 min, RT), 2% SDS [20 min, RT] and 2% SDS (100 mM dithiothreitol, DTT, 20 min, 95 • C).After incubation at low SDS content and lower temperatures SDS-PAGE of α S1 -casein led to prominent bands at molecular weights of 25, 55, 70 kDa, as well as several bands at even higher molecular weights.These bands were assigned to α S1 -casein monomers, dimers, trimers and oligomers of higher order.The intensity of bands assigned to oligomers disappeared with stronger denaturation conditions, whereas the band assigned to the monomer appeared with stronger intensities.Only a single band at 25 kDa remained after treatment with sample buffer representing the strongest denaturation conditions (2% SDS, 100 mM DTT, 20 min, 95 • C).This was a clear indication that the bands detected at higher molecular weights under weaker denaturation conditions were indeed α S1 -casein oligomers.The intensity of the monomer band was lower, the weaker the denaturation conditions were as applied, whereas bands assigned to oligomers appeared with higher intensities.This seemed to be a clear indication for oligomerization of αS1-casein under nondenaturing and probably as well under native conditions [40].Therefore, α S1 -casein can be considered to form oligomers.The required addition of a reducing reagent as DDT for complete α S1 -casein monomerization could be an indication of disulfide bonds between different α S1casein monomers.Disulfide bonding has been described before for bovine α S2 -casein [41] but not for bovine α S1 -casein due to the overall lack of cysteines in this casein.
Int. J. Mol.Sci.2024, 25, × FOR PEER REVIEW 7 of 23 stronger denaturation conditions, whereas the band assigned to the monomer appeared with stronger intensities.Only a single band at 25 kDa remained after treatment with sample buffer representing the strongest denaturation conditions (2% SDS, 100 mM DTT, 20 min, 95 °C).This was a clear indication that the bands detected at higher molecular weights under weaker denaturation conditions were indeed αS1-casein oligomers.The intensity of the monomer band was lower, the weaker the denaturation conditions were as applied, whereas bands assigned to oligomers appeared with higher intensities.This seemed to be a clear indication for oligomerization of αS1-casein under nondenaturing and probably as well under native conditions [40].Therefore, αS1-casein can be considered to form oligomers.The required addition of a reducing reagent as DDT for complete αS1casein monomerization could be an indication of disulfide bonds between different αS1casein monomers.Disulfide bonding has been described before for bovine αS2-casein [41] but not for bovine αS1-casein due to the overall lack of cysteines in this casein.Next, FITC (fluorescein isothiocyanate)-labeled αS1-casein (25 nM) was added to different concentrations of nonlabeled αS1-casein (50 nM-175 µM), sonicated for 15 min, incubated for 1 h at 37 °C and analyzed by MST (Figure 6B).The binding of nonlabeled αS1casein to the fluorescence-labeled isoform was clearly detectable by this method, and a KD value of 2.2 µM was determined.As a second method for supporting the binding of αS1casein to itself, SPR was applied (Figure 6C).αS1-casein was immobilized on a CMDP-5 sensor chip, which could be monitored by a signal increase of 450 µRU.Subsequently, Next, FITC (fluorescein isothiocyanate)-labeled α S1 -casein (25 nM) was added to different concentrations of nonlabeled α S1 -casein (50 nM-175 µM), sonicated for 15 min, incubated for 1 h at 37 • C and analyzed by MST (Figure 6B).The binding of nonlabeled α S1 -casein to the fluorescence-labeled isoform was clearly detectable by this method, and a K D value of 2.2 µM was determined.As a second method for supporting the binding of α S1 -casein to itself, SPR was applied (Figure 6C).α S1 -casein was immobilized on a CMDP-5 sensor chip, which could be monitored by a signal increase of 450 µRU.Subsequently, different concentrations of unlabeled α S1 -casein were injected (0.63 to 5 µM).By this method, a K D value of 2 µM was determined.This was in good agreement with the K D value of 2.2 µM as determined by MST.As the signal increase is directly proportional to the mass, one would expect a signal increase of 450 µRU for an α S1 -casein dimer.The signal increased up to 1300 µRU and was, therefore, three times as high as with immobilized α S1 -casein.Therefore, SPR hints at the formation of tetramers.Due to the several types of α S1 -casein oligomers as identified by SDS-PAGE and by SPR, one has to consider that the K D value 2 µM could be due to a mixture of dimer, tetramer and higher oligomer.K D value was not obtained for dimer formation.The K D value of unphosphorylated α S1 -casein binding to itself was in the same order of magnitude as the K D value of 2 µM for dephosphorylated bovine α S1 -casein determined by SPR [21] and postulated K D values for all bovine caseins between 1 and 3 µM [21,42].
We observed that injection of α S1 -casein, which was stored longer than 12 h at 4 • C resulted in a loss of signal in the SPR sensograms and clogging of injection needles and capillaries.The formation of larger α S1 -casein oligomers could have been a reason for this.It would be in accordance with the results of the SDS-PAGE analysis, where higher α S1 -casein oligomers (e.g., octamers) were detected.Therefore, α S1 -casein oligomers were investigated for their diameter by PCS (photon correlation spectroscopy, Figure 6D).Using this method, α S1 -casein oligomers were detectable with a mean diameter (Ø) of 73.4 nm (polydispersion index [PI]: 0.6).
In summary, recombinant α S1 -casein, when unphosphorylated, formed higher-order oligomers with a moderate affinity of α S1 -casein to itself (K D value of 2 µM).The α S1 -casein oligomers had a considerable diameter of 73.4 nm.This could indicate that α S1 -casein functions could be impaired by its affinity to itself (protein-protein interaction) and/or influence the formation and diameter of micelles, e.g., in breast milk [39].

The Influence of Temperature, pH and Phosphorylation on α S1 -Casein Structure
In previous works, it was shown that α S1 -casein binding to TLR4 was abrogated by heat denaturation [8].Therefore, it was investigated here whether temperature-induced secondary structure changes of α S1 -casein were detectable by CD-spectroscopy (Figure 7A).In addition, the melting point of α S1 -casein was supposed to be determined by nano-DSF (nano-Differential Scanning fluorimetry, Figure S3), as well as a possible temperature dependence of the particle diameter by PCS (Figure 7B).At a temperature of 20 • C, α S1casein showed the highest content of α-helical structure (−834,946 Grad•cm 2 •dmol −1 at 208 nm) and the largest particle diameter of Ø 637 nm.The weakest molar ellipticity was detected at 30 • C to 40 • C (>720,892 Grad•cm 2 •dmol −1 at 208 nm), but α S1 -casein still exhibited high α-helical content.With temperature increase, the minima were altered from 208 (40 • C) to 205 nm (70 • C) and to 204 nm (100 • C).Such a bathochromic shift of minima indicated the loss of α-helical content, whereby a minimum at 204 nm is characteristic of a high random coil content of a structure.The decrease of α-helical-content was in line with a reduction of particle diameter from Ø 637 nm (20 • C) to Ø 467.6 nm (40 • C), Ø 189.4 nm (70 • C) and Ø 154 nm (90 • C) as determined by PCS.Normally, the ratio of fluorescence at 350/330 nm is expected to be constant in nano-DSF spectra but increases with protein unfolding.The ratio of fluorescence at 350/330 nm decreased for α S1 -casein from 0.68 (10 • C) to 0.44 (70 • C).This can only be explained by a higher absorbance of larger particles at 350 nm, whereas diameter reduction resulted in lower absorbance at 350 nm, which is in accordance with the results obtained by PCS.Nano-DSF spectra showed a turning point at 80.1 • C for α S1 -casein.This appears to be caused by a dramatic change in protein structure, e.g., by reaching the melting point [43].Melting of a protein could result in a loss of molar ellipticity and oligomer diameter.But α S1 -casein showed high molar ellipticity (−734,946 Grad × cm 2 × dmol −1 ) and oligomer size (154.9nm) at 90 • C. Therefore, we attributed the inflection point at 80.1 • C to a change in the conformation of α S1 -casein.In consequence, α S1 -casein was considered to be not heat denatureable, but a temperatureinduced change of conformation was detectable.As mentioned above, in a previous study, it was reported that α S1 -casein was no longer binding to TLR4 after heat treatment [8].As shown here, this non-TLR4-binding α S1 -casein conformation appears to have a lower α-helical content and a reduced particle diameter in comparison to α S1 -casein binding to TLR4.
Int. J. Mol.Sci.2024, 25, × FOR PEER REVIEW 9 of 23 that αS1-casein was no longer binding to TLR4 after heat treatment [8].As shown here, this non-TLR4-binding αS1-casein conformation appears to have a lower α-helical content and a reduced particle diameter in comparison to αS1-casein binding to TLR4.Noteworthy, human αS1-casein has higher α-helical content and higher temperature stability compared to bovine αS1-casein, which indeed was shown to melt at 65 °C before [17,23].However, the reduction in oligomer diameter between 20 and 37 °C was similar to that described for bovine αS1-casein [44].Noteworthy, human α S1 -casein has higher α-helical content and higher temperature stability compared to bovine α S1 -casein, which indeed was shown to melt at 65 • C be-fore [17,23].However, the reduction in oligomer diameter between 20 and 37 • C was similar to that described for bovine α S1 -casein [44].
This was supported by in silico structure analyses of α S1 -casein, showing that phosphorylation sites are in direct neighbourhood to an α-helix (Figure 3B).One known phosphorylation site is located in an α-helix-loop-α-helix motif [11].Phosphorylations in such a flexible region could lead to a destabilization of neighbouring α-helices due to the additional charge, as suggested by Jakob et al. [45].Therefore, the structure of P-α S1 -casein is supposed to have higher flexibility and could result in the exposure of other AA as in α S1 -casein, which could lead to an altered affinity to itself.
This was investigated by analyzing the binding of P-α S1 -casein to itself by SPR (Figure 7D).For this purpose, P-α S1 -casein was immobilized on a CMDP-5 sensor chip, which was monitored by a signal increase of 450 µRIU, and different concentrations of P-α S1 -casein were injected (0.625 to 5 µM, 5 min, 5 µL/min, RT).By this strategy, a K D value of 0.5 µM was determined for the binding of P-α S1 -casein to itself.This K D value was four times lower compared to the K D value obtained for the binding of unphosphorylated α S1 -casein to itself (K D value of 2 µM).P-α S1 -casein showed higher affinity to itself than α S1 -casein to itself, P-α S1 -casein, which does not bind to TLR4 as shown before, had a lower α-helical content, higher random coil content and higher affinity to itself than unphosphorylated α S1 -casein, binding to TLR-4.
In addition, the influence of the pH value, ranging from pH 2 to pH 10, on the secondary structure of α S1 -casein (12.5 µM) was investigated by CD-spectroscopy (Figure 7E) to elucidate any pH-dependent structural changes.α S1 -casein showed α-helical properties at all pH values applied.α S1 -casein at pH2 showed the typical minima of an α-helical protein at 208 and 222 nm.Furthermore, it showed the strongest molar ellipticity at this pH with −1,345,500 Grad•cm 2 •dmol −1 at 208 nm and −966,893 Grad•cm 2 •dmol −1 at 222 nm.Therefore, α S1 -casein should have the highest α-helical at a pH of 2. At alkaline pH values, the minimum was shifted to 205 nm (−1,162,139 Grad•cm 2 •dmol −1 ) at pH 9 (−1,162,139 Grad•cm 2 •dmol −1 ) and to 206 nm (−904,923 Grad•cm 2 •dmol −1 ) at pH 10.Therefore, the lowest α-helical content of α S1 -casein was detected at alkaline conditions.At pH 5, we detected the less intense minima with a molar ellipticity of −84,040 Grad•cm 2 •dmol −1 at a minimum of 208 nm.A decrease in molar ellipticity is associated with stronger interactions between the structural components of a protein and, hence, with a more rigid folding [38].The isoelectric point of human α S1 casein is at pH 5.1, at which it has its highest density.
In order to find out whether these differences in structure observed at different pH values have an influence on the oligomerization of α S1 -casein, the samples were analyzed by PCS at different pH as well (Figure 7F).The largest oligomers of α S1 -casein were detected at pH 2 (Ø 826.4 nm).The oligomers of α S1 -casein at pH 8 (Ø 482.9 nm) were significantly larger compared to oligomers between pH 4 and 7 (Ø 130-73 nm).This is interesting because in the gastrointestinal tract of infants, digestion of milk is supposed to take place under acidic conditions, with a pH range of 3-6 [46].At a pH of 2, α S1 -casein formed the largest oligomers and was able to induce an IL-8 secretion via TLR4, as shown in Figure S4.The diameter of oligomers at pH 7 was significantly smaller than that at pH 2. However, α S1 -casein changed its oligomer diameter dramatically from pH 7 to pH 8, i.e., from 73 (pH 7) to 482.9 nm (pH 8) at 37 • C and to 467.6 nm (pH 7.4) at 40 • C.
In summary, these results show that the α-helical content of αS1-casein was higher, and the oligomer diameter was larger at lower temperatures.Both gained maximum values at a pH of 2.Moreover, TLR4-binding α S1 -casein was shown to have a higher α-helical content compared to non-TLR4-binding P-α S1 -casein.The random coil content raised with temperature and phosphorylation.α S1 -casein did not induce an IL-8 secretion via TLR4 after incubation at 95 • C or after phosphorylation.Therefore, two distinct conformations of α S1 -casein were proposed: an α-helical TLR4-binding conformation and a less α-helical non-binding conformation.Phosphorylation sites of α S1 -casein were found in the neighborhood of α-helical regions (Figure 3A).As phosphorylation as well as loss of α-helical correlated with a loss of TLR4-binding of α S1 -casein, the TLR4 binding site could be located near the phosphorylation sites within the α-helical regions.The results as obtained indicate that α S1 -casein induced IL-8 secretion over a wide range of pH and in different oligomeric states.Until today, such pH-resistant activity was only described for immunologically associated proteins of breast milk such as IgG, sIgA and mucin-1, e.g., regulation of proliferation [14,47].In order to clarify the immunological relevance of these findings, the in vivo phosphorylation state of breast-milk α S1 -casein and possible ways to its dephosphorylation need to be investigated.
When human α S1 -casein (1 µM) was incubated for 6 d at RT in a bottomless 96-well plate glued on a glass support, fibrils of 80 nm in length and 20 nm in width were detectable by AFM (atomic force microscopy), and by the incorporation of thioflavin T, a reporter fluorescent dye know to specifically interact with amyloid fibrils.Such fibrils were not formed when P-α S1 -casein was treated similarly.

α S1 -Casein Contains a Coiled-Coil Domain
As mentioned in Section 2.1, high molar ellipticity of α S1 -casein appeared at typical minima of 208 and 222 nm.This phenomenon leads to the investigation of intermolecular interactions of α S1 -casein, as described in Section 2.2.Furthermore, α S1 -casein was shown to form oligomers, e.g., tetramers, and was heat stable, as shown in Section 2.3.Interestingly, all these observations are characteristic of a coiled-coil protein [48].Consequently, we analyzed α S1 -casein for coiled-coil domains.AAS of α S1 -casein was analyzed on the prediction of coiled-coil motifs with three different programs.The program COILS (Expasy) compared AAS of α S1 -casein with AAS of known coiled-coil peptides (Figure 8A).PCOILS (MPI Development Biology, Tübingen) predicted the homology and identity of the AAS of α S1 -casein to known AAS of coiled-coil motifs, as shown in Figure S5A [49 -51].MARCOIL (MPI for Development Biology, Tübingen) used a Hidden Markov Model for the identification of coiled-coil motifs (Figure S5B).In contrast to COILS and PCOILS, MARCOIL was not limited to a certain AAS but considered all amino acids of the protein [50,52].All three programs predicted a high probability for a coiled-coil motif in the AAS of human α S1 -casein between AS 102 and 130 ( 102 QFCRLNEYN QLQLQAAHAQ EQIRRMNENS 130 ).Due to this analysis, α S1 -casein appears to be the only human casein protein predicted to bind to itself, potentially by a coiled-coil domain (Figure S1).When α S1 -casein of 17 species was analyzed by the same programs, a high probability for a coiled-coil domain was predicted for hominids (orang utan, gorilla, chimpanzee).The coiled-coil domain within α S1 -casein of these species was located in the same stretch of the α S1 -casein AAS (Figure S1).Other regions of these α S1 -casein AAS did not show a probability for a coiled-coil domain.schematic illustration of secondary structure with RaptorX (pink: α-helix; yellow: β-sheet; white: random coil; blue: predicted coiled-coil motive.Homology of βand κ-casein to α S1 -casein was calculated using EMBUSS Needle-Wunsch algorithm (EMBL-EBI, Cambridge, UK).(B) CD-spectra of peptide S 91 -A 119 (5 µM, 2 mm path length) in 30% phosphate buffer/70% trifluorethanol (black) and 100% phosphate buffer (grey).(C) CD-spectra of peptide S 91 -A 119 ( 91 SEEMSLSKCA EQFCRL-NEYN QLQLQAAHA 119 , 5 µM, 2 mm path length) in 30% phosphate buffer/70% trifluorethanol at different temperatures (10-30 • C blue-light blue; 40-60 • C green-yellow; 70-95 • C orange-dark red).P: potential phosphorylation site; green: hydrophobic amino acids; yellow: polar amino acids; red: coiled-coil domain.
For in vitro investigation of this part of the human α S1 -casein AAS, a peptide corresponding to AA S 91 -A 119 ( 91 SEEMSLSKCA EQFCRLNEYN QLQLQAAHA 119 ) of α S1casein, most of the coiled-coil domain was synthesized.Furthermore, the peptide was designed in a way that it would be in direct neighborhood to the phosphorylation site S 89 of full-length α S1 -casein.The secondary structure of this peptide was analyzed via CD-spectroscopy (Figure 8B).The spectra showed a minimum below 204 nm, which would be expected for a random coil content.Furthermore, a smaller minimum at 222 nm was detected.The peptide contained parts of a coiled-coil domain and parts of an N-terminal sequence potentially random coil.To isolate the characteristics of the coiled-coil domain, the peptide was analyzed in 70% trifluorethanol (Figure 8C).Trifluorethanol is known to stabilize α-helical content of peptides [53,54].In trifluorethanol, the secondary structure of the peptide turned out to be merely α-helical with minima at 208 and 222 nm.
For investigating the coiled-coil part of α S1 -casein by CD spectroscopy, it was assumed that at a certain temperature, the supercoiled α-helix should unfold.The individual α-helices were expected to unfold as well at the same temperature.Thus, two states were expected, a native state and a denatured one, keeping in mind that a coiled-coil structure could also lead to oligomers.A clear indication of a coiled-coil structure is that the corresponding CD spectra at different temperatures show a common point of intersection at ~204 nm up to the complete denaturation of the protein [55].For this investigation, CD spectra of the peptide S 91 -A 119 were recorded at temperatures ranging from 10 • C to 95 • C (Figure 8C).Molar ellipticity decreased as temperature increased from −542,048 Grad × cm 2 × dmol −1 at 10 • C to −426,287 Grad × cm 2 × dmol −1 at 90 • C.Moreover, all CD-spectra showed fixed minima at 208 and 222 nm as well as an intersection at 204 nm and −318,760 Grad × cm 2 × dmol −1 .The conformation of the peptide was highly stable up to 95 • C.An unfolding of the peptide could not be investigated.Such stability was described for coiled-coil domains before [48].The intersection at 204 indicated that the peptide could form a coiled-coil structure, which is present in α S1 -casein as predicted above.This coiled-coil structure is located directly C-terminal to the phosphorylation site SS 89 SSEE of α S1 -casein and a helix-loop-helix motif as predicted [11].It, therefore, appears possible that destabilization of the coiled-coil structure by phosphorylation could result in a change of TLR4 agonisticity of α S1 -casein.

Identification of a TLR4-Stimulating Peptide Derived from α S1 -Casein
The results as obtained above indicate that the TLR4-binding domain of α S1 -casein is located in the predicted α-helical region ( 16 R-K 98 ) containing all known phosphorylation sites, including the most prominent ones S 33 , S 41 , S 71 , and S 89 [31].
To investigate this further, six variants of α S1 -casein, four truncated at the N-terminus (N1, N2, N3, N4) with an N-terminal His 6 -Tag and two truncated at the C-terminus (C1, C2) with a C-terminal His 6 -Tag.In consequence, these constructs all representing parts of the AAS of α S1 -casein (Figures 9A and S6, Table S1) were expressed in E. coli and purified by NTA column chromatography, analogous to the method described above for the full-size human α S1 -casein.
The binding of the truncated variants to TLR4 was analyzed by flow cytometry as described before [8].TLR4 + cells (HEK293 cells transfected with TLR4/MD2/CD14) and TLR4 − cells (HEK293 cells without TLR4) were incubated with full-length α S1 -casein and its truncated variants (500 nM), followed by the addition of a murine anti-His 6 IgG and a caprine Dylight633-antimurine IgG and were finally analyzed by flow cytometry.Cellular fluorescence resulting from the binding of full-size α S1 -casein and its truncated variants to TLR4 + cells is shown in Figure 9B.In addition, IL-8 secretion of TLR4 + cells and TLR4 − cells after incubation with full-length α S1 -casein and its truncated variants was analyzed and described before [9], and the results are shown in Figure 9C.Truncated variants N1, N2, C1 and C2 were shown to bind to TLR4 + cells, resulting in a 1.2 to 2 times higher fluorescence intensity as obtained after incubation with TLR4 − cells.N3 showed only marginal and N4 did not show any difference in fluorescence intensity when incubated with TLR4 + cells in comparison to incubation with TLR4 − cells (Table S1).Therefore, N3 and N4 appeared to be nonbinders of TLR4.Five of the six truncated variants of α S1 -caseins induced an IL-8 secretion (N1: 7.5 ng/mL IL-8; N2: 4.8 ng/mL; N3: 3.6 ng/mL; C1: 5.2 ng/mL and C2 5.2 ng/mL).The five truncated variants induced a significantly lower IL-8 secretion compared to full-length α S1 -casein (23.3 ng/mL IL-8).Truncated variant N4 did not induce any IL-8 secretion.
A clear loss in the induction of IL-8 secretion via TLR4 was shown for α S1 -casein variants truncated at the N-terminus (N1: 7.5 ng/mL IL-8; N2: 4.8 ng/mL; N3: 3.6 ng/mL; N4: no IL-8 secretion).Hereby, N3 induced an IL-8 secretion of 3.6 ng/mL and showed marginal hints for binding TLR4 + cells by flow cytometry as indicated by low fluorescence intensities of 71 for TLR4 + and 65 for TLR4 − cells.In contrast, N4 did not induce an IL-8 secretion at all (−2.5 ng/mL IL-8).This could be an indication that the AA of N3 missing in the AAS of N4 (V 77 -E 92 ) contains a binding motif of α S1 -casein to TLR4.C-terminal truncated variants C1 and C2 exhibited a similar binding to TLR4 as N1 and N2, as indicated by a 2.3 (C1) to 2.2 times (C2) higher fluorescence intensity when incubated with TLR4 + cells in comparison to TLR4 − cells.Both induced the secretion of identical amounts of IL-8 (C1, C2: 5.21 ng/mL IL-8) in TLR4 + cells.The C-terminal truncations as tested did not lead to a complete loss of binding and induced residual IL-8 secretion, and hence, could be involved in stabilizing the binding region of α S1 -casein to TLR4.
Therefore, a peptide corresponding to the AAS of N3 (TLR4 binder) absent in N4 (nonbinder) of α S1 -casein (V 77 -E 92 ) was synthesized and tested on induction of IL-8 secretion in TLR4 + cells.In addition, a second peptide was synthesized as control and was investigated on IL-8 secretion on TLR4 + cells as well.As shown in Figure 10, only peptide V 77 -E 92 induced an IL-8 secretion (0.95 ng/mL), whereas incubation with the control peptide was not significantly different in IL-8 secretion to the growth medium (0.23 ng/mL IL-8).A clear loss in the induction of IL-8 secretion via TLR4 was shown for αS1-casein variants truncated at the N-terminus (N1: 7.5 ng/mL IL-8; N2: 4.8 ng/mL; N3: 3.6 ng/mL; N4: no IL-8 secretion).Hereby, N3 induced an IL-8 secretion of 3.6 ng/mL and showed marginal hints for binding TLR4 + cells by flow cytometry as indicated by low fluorescence intensities of 71 for TLR4 + and 65 for TLR4 − cells.In contrast, N4 did not induce an IL-8 secretion at all (−2.5 ng/mL IL-8).This could be an indication that the AA of N3 missing in the AAS of N4 (V 77 -E 92 ) contains a binding motif of αS1-casein to TLR4.C-terminal truncated variants C1 and C2 exhibited a similar binding to TLR4 as N1 and N2, as indicated by a 2.3 (C1) to 2.2 times (C2) higher fluorescence intensity when incubated with TLR4 + cells in comparison to TLR4 − cells.Both induced the secretion of identical amounts of IL-8 (C1, C2: 5.21 ng/mL IL-8) in TLR4 + cells.The C-terminal truncations as tested did not lead to a complete loss of binding and induced residual IL-8 secretion, and hence, could be involved in stabilizing the binding region of αS1-casein to TLR4.
Therefore, a peptide corresponding to the AAS of N3 (TLR4 binder) absent in N4 (non-binder) of αS1-casein (V 77 -E 92 ) was synthesized and tested on induction of IL-8 secretion in TLR4 + cells.In addition, a second peptide was synthesized as control and was investigated on IL-8 secretion on TLR4 + cells as well.As shown in Figure 10, only peptide V 77 -E 92 induced an IL-8 secretion (0.95 ng/mL), whereas incubation with the control peptide was not significantly different in IL-8 secretion to the growth medium (0.23 ng/mL IL-8).Recombinant human αS1-casein [8] and P-αS1-casein [9] were purified as described before.αS1-casein was FITC-labeled as described [9].
The sequence of truncated variants of human α S1 -casein and rest of the plasmid was amplified from plasmid pET TS001 (coding for CSN1S1 with N-terminal His 6 -Tag) [8] with Phusion DNA polymerase (Thermo Fisher Scientific, Bonn, Germany) using forward (fw) and reverse (rv) oligonucleotides, as listed in Table 1.PCR products were purified according to the manufacturer's instructions with InnuPREP Plasmid Mini Kit (Analytik Jena, Jena, Germany).The resulting plasmids pET N1, pET N2, pET N3, pET N4, pET C1, pET C2 (Table 1) were transformed into Escherichia coli (E.coli) strain DH5α (Invitrogen, Carlsbad, CA, USA).Plasmid DNA replication, isolation, transformation into E. coli and purification of truncated variants of α S1 -casein were performed as described before for full-length-α S1 -casein [8].Protein purity was evaluated by Coomassie-stained SDS-PAGE.Protein concentration was determined by an indirect ELISA as described [8].
All samples were loaded onto an SDS-Gel containing 15% acrylamide with PAGE-ruler prestained protein marker (Fermentas, St. Leon-Roth, Germany) as molecular weight standard.After separation (80 V protein focus, 120 V protein separation), proteins were stained with Coomassie brilliant blue G250 (Serva, Heidelberg, Germany) for protein purity analysis and with silver staining for analyzing oligomers.For silver-staining, SDS-PAGE gel was fixed for 1 h (50% ethanol, 5% acidic acid).Gel was washed with 50% ethanol (10 min), twice with water (10 min), with sodium thiosulfate (0.02%) and twice with water (5 min).Gel was incubated for 30 min in 0.1% silver nitrate solution, quickly washed with water and developed in a solution of 0.04% formaldehyde and 2% sodium carbonate.The reaction was stopped in 5% acidic acid.

Secondary Structure Analysis by CD-and ATR-FTIR-Spectroscopy
All samples were transferred into a 10 mM NaH 2 PO 4 /Na 2 HPO 4 buffer (pH 7.2, chloridefree buffer) for CD measurements.CD spectra were conducted on a Jasco J-815 spectrometer with a Jasco PTC-348WI Peltier-type temperature control system (Jasco Corp, Hachioji, Japan) at constant nitrogen flow.Far-UV CD spectra were measured with 2 mm path length quartz cuvette.Human α S1 -casein was recorded at a concentration of 12.5 µM.Peptides were analyzed with a concentration of 5 µM in 30% phosphate buffer/70% trifluorethanol.Spectra were recorded from 190 to 260 nm with a resolution of 0.1 nm (100 nm/min).The final were corrected by subtracting the corresponding baseline spectrum and secondary analysis was estimated [38].

Photon Correlation Spectroscopy
α S1 -casein (1 mL, 50 µM) in phosphate buffer was incubated for 3 h at 37 • C. Samples were transferred into a quartz cuvette.Proteinoligomer diameter and oligomers' population PI were analyzed by PCS (Zetasizer NanoZS3600, Malvern Instruments, Worcestershire, UK).Before each measurement, samples were equilibrated for at least 2 min.The mean of 15 single measurements was recorded.Samples were recorded with an attenuator value of 8. Temperature denaturation was recorded by a heating rate resolution of 0.5 • C.
For AFM sample preparation, human α S1 -casein (25 µM) was incubated for 6 d at 37 • C in 1.5 mL Eppendorf reaction tube.Samples were centrifuged (20 min, 20,000× g, 4 • C) and suspended in 10 µL ddH 2 O.The suspension was transferred onto a freshly cleaved mica chip (NanoAndMore, Wetzlar, Germany).This chip coated with α S1 -casein was washed five times with 100 µL ddH 2 O and dried under a constant flow of nitrogen gas for 20 min.A Bruker Dimension 3100 atomic force microscope, equipped with a Nanoscope IIIa controller (Bruker, Karlsruhe, Germany) was used.Measurements were performed in tapping mode with n-type silicon cantilevers (HQ:NSC14/Al BS, nominal tip radius <10 nm, typical resonant frequency of about 160 kHz and a nominal spring constant of 5 N/m; manufactured by µmash, Sofia, Bulgaria).Nanoscope analysis software version 1.5 was used for data analysis.

Conclusions
Secondary structure analysis revealed that TLR4-agonistic α S1 -casein was mostly αhelical but also able to adopt a partial β-sheet structure.The increased α-helical content was associated with the binding of α S1 -casein to TLR4 and IL-8 secretion.Whereas the α-helical structure was stable over a wide range of pH and up to 80 • C (as well as IL-8 secretion), it was substantially altered by phosphorylation, which omitted TLR4-binding and IL-8 secretion.
Within α S1 -casein a TLR4-agonistic peptide (V 77 -E 92 ) was identified.Moreover, a coiledcoil domain was demonstrated at the same position of α S1 -casein from primates, such as human, orang-utan, gorilla or chimpanzee, but not from other mammals.Phosphorylation of α S1 -casein led to higher flexibility, higher affinity to itself (K D value: 0.5 µM, nonphosphorylated K D value: 2 µM), formation of random aggregates and loss in structural constraints in comparison to α S1 -casein.
The differences in conformation regulated by phosphorylation appear to be a kind of switch between two different states of α S1 -casein, which could be related to two different functions.On the one hand, a nutritional role of breast milk α S1 -casein and, on the other hand, an immunostimulatory role of α S1 -casein by binding TLR4, inducing proinflammatory processes and immune cell maturation, as has been shown before.
In a previous study, we were able to show that breast milk from breastfeeding mothers contained phosphorylated α S1 -casein and identified the phosphorylation sites by a targeted MS approach.In a further study, we could show that having been breastfed leads to a lifetime IgG response against unphosphorylated α S1 -casein.As synopsis with the data of the present study, it can be hypothesized that phosphorylated α S1 -casein in breast milk is dephosphorylated during or after breastfeeding and enters the intestine of the suckling.Not at least due to its pH stability, it is resorbed-in an unphosphorylated conformation-and can fulfill its immunostimulatory function.It may serve as a signal for the infant that he is out of the womb and, from now on, needs to take care on his own for his immune status.Systematic and continuous analyses of the α S1 -casein content in breast milk of breastfeeding mothers as well as a timely resolution following the onset of the suckling's immune system, would be the next steps to support this hypothesis, investigating formula-fed infants as control.The comparison of key immune parameters in breastfed and formula-fed persons of different ages could provide further elucidating insights.
Author Contributions: T.S. provided the concept and design, performed experiments, data analysis and prepared manuscript; M.F.S. prepared AFM samples, performed ThT-experiments; S.V. provided the concept and design and data analysis; F.C.H. performed AFM-measurements and analysis; J.B. isolated truncated variants of human α S1 -casein, performed flow cytometry and CD-analysis of the truncated variants; K.M. performed IR experiments and analysis; E.B. provided the concept and design; M.S. provided the concept and design; J.J. provided the concept and design, supervised the project and prepared manuscript.All authors have read and agreed to the published version of the manuscript.

Figure 3 .
Figure 3. Analysis of the secondary structure.(A) Secondary structure prediction of the AAS of human αS1-casein with signal peptide (grey: random coil structure; red: α-helix; blue: β-sheet).(B) Prediction of tertiary structure of human αS1-casein using RaptorX.

Figure 3 .
Figure 3. Analysis of the secondary structure.(A) Secondary structure prediction of the AAS of human αS1-casein with signal peptide (grey: random coil structure; red: α-helix; blue: β-sheet).(B) Prediction of tertiary structure of human αS1-casein using RaptorX.

Figure 4 .
Figure 4. (A) Prediction of human αS1-casein for its probability as intrinsically disordered using different training sets of PONDR (yellow: algorithm XL1 optimized to predict domains longer than 39 amino acids; red: algorithm VLXT valid for proteins being completely disordered; purple: Algorithms VSL2 combining both algorithms).Black line: region with high probability to be intrinsically disordered.(B) Probability of human αS1-casein to form transmembrane regions.

Figure 4 .
Figure 4. (A) Prediction of human α S1 -casein for its probability as intrinsically disordered using different training sets of PONDR (yellow: algorithm XL1 optimized to predict domains longer than 39 amino acids; red: algorithm VLXT valid for proteins being completely disordered; purple: Algorithms VSL2 combining both algorithms).Black line: region with high probability to be intrinsically disordered.(B) Probability of human α S1 -casein to form transmembrane regions.

Figure 7 .
Figure 7. Structural characteristics of human αS1-casein in dependence of temperature (A,B), phosphorylation (C,D) and pH value (E,F).A, C, E: CD-spectra of human αS1-casein (12.5 µM; 2 mm path length).B, F: Diameter detected of 50 µM αS1-casein by PCS at pH 7.4 and different temperatures (B) as well as at different pH at 37 °C (F).D: αS1-casein was immobilized on a SPR sensor chip and concentrations from 0.625 µM to 5 µM of αS1-casein were injected.Dots represent the difference in the signal intensity at steady-state (binding) compared to coated chip ([•] αS1-casein binding itself; [○] P-αS1-casein binding itself).

Figure 10 .
Figure 10.Testing synthetic peptides V 77 -E 92 and control peptide (each 1.5 µM) for induction of IL-8 secretion.Peptides were incubated with TLR4 + cells.Supernatants were analyzed for IL-8 as described above.

Figure 10 .
Figure 10.Testing synthetic peptides V 77 -E 92 and control peptide (each 1.5 µM) for induction of IL-8 secretion.Peptides were incubated with TLR4 + cells.Supernatants were analyzed for IL-8 as described above.

Table 1 .
List of α S1 -casein truncated variants, plasmids coding for these truncated variants and used oligonucleotides for construction of these plasmids.