Role of catechin on collagen type I stability upon oxidation: a NMR approach

Abstract The study focuses on the understanding, at molecular level, the mechanism of interaction between protein and flavonoids. Collagen and catechin interactions were investigated by NMR in solution and solid state. The effect of catechin on the stability of collagen to oxidation was also explored. Collagen was treated with two concentrations of catechin solutions. Oxidation was carried out by incubation of collagen solution with three oxidation systems: Fe(II)/H2O2, Cu(II)/H2O2, and NaOCl/H2O2. The effects of oxidation systems were evaluated by high resolution 1 D and 2 D proton spectroscopy and solid state NMR (13C CP MAS) experiments. Interactions between collagen and catechin preferentially occur between catechin B ring and the amino acids Pro and Hyp of collagen. Results showed that both iron and copper oxidation systems were able to interact with collagen by site specific attack. Moreover, catechin protects collagen proline from oxidation by metal/H2O2 systems, preventing copper and iron approach to collagene molecule;this behaviour was more evident for the copper/H2O2 system.


Introduction
The attack of free radicals in biological systems is directed mainly toward lipids, proteins, and nucleic acids, although the main target of oxidants are proteins both for their abundance and the rapid reaction rate (Davies 2016;Gebicki 2016).
The damage caused to proteins by oxidative processes involves amino acids modification, increased susceptibility to proteolysis, fragmentation, cross-linking and aggregation phenomena (Davies 2012;Luna and Est evez, 2018). Although iron and copper are important for life, the metal ions are able to mediate the formation of ROS via the Fenton reaction and can also bind to protein amino acids inducing site-specific damage (Leonard et al. 2004). On the other hand, the oxidation system that generate singlet oxygen are systems that attack proteins following a random mechanism (Davies 2016). Similar to lipids, proteins can be protected from oxidation by antioxidants and free radical scavengers like flavonoids and other compounds, although the mechanism whereby antioxidant work on protein is not yet defined.
Focus of our efforts was to shed light on the mechanism of action of natural substances, such as phenolics, in protecting proteins from oxidation (Reddy et al. 2015;Santini et al 2017;Santini and Novellino 2018;Durazzo and Lucarini In press). In our experiments, collagen was chosen as the target protein and catechin as the antioxidant molecule. Studies demonstrated that collagen is degraded when exposed to the action of ROS (Vissers and Winterbourn 1986;Monboisse et al. 1988;Olszowski et al. 2012) and that it is mostly sensitive to Cu (II)/H 2 O 2 oxidation system (Hawkins and Davies 1997). The structure of collagen is stabilized by inter and intra-chain hydrogen bond (Privalov 1982;Bhattacharjee and Bansal 2005) and phenolic compounds in vegetables are able to establish cross-link with collagen by multiple hydrogen bonds (Tang et al. 2003). Catechin belongs to the family of flavonol-3-ols and, in the hierarchy of flavonoid antioxidant activities, it appears to be one of the most active flavonoids (Rice-Evans et al. 1996;Raab et al. 2010;Shay et al. 2015). It contains five hydroxyl functions attached to the three ring structures, designed A, B and C ( Figure  S1). Since catechin has a molecular size smaller than the pore dimensions of collagen pentafibrils, it can fit into collagen domain and results in inter and intra-chain crosslinks that promote collagen resistance to collagenase attack (Madhan et al. 2005). Catechin, alone or in combination with others compounds, may act as protective agents in chronic diseases such as cancer, cardio-vascular diseases, diabetes by a variety of mechanisms (Matsui 2015;Shai 2015;Mangels and Mohler 2017;Bernatoniene and Kopustinskiene 2018).
In the current study, an in vitro system based on a protein-antioxidant complex was developed, using the NMR analysis as probe. The works aims at investigating the structural features of collagen and its complexes with catechins by NMR spectroscopy techniques in solution and in solid state. Furthermore, the effect of catechin on collagen towards three in vitro oxidation systems was evaluated. Nowadays NMR has become a valuable tool in research in food science and in food authentication assessment (Rotondo et al. 2011Mallamace et al. 2014;Cicero et al. 2015;Corsaro et al. 2015Corsaro et al. , 2016Dugo et al. 2015;Albergamo et al. 2017;Li et al. 2017;Consonni and Cagliani 2018;Costa et al. 2018;Ebrahimnejad et al. 2018;Salvo et al. 2016). The most commonly employed solution-state NMR techniques including one-dimensional (1D) and two-dimensional (2D), 1 H and solid-state cross-polarization/magic angle spinning (CP/MAS) 13 C NMR will be discussed throughout the study.

NMR measurements of collagen
The 1 H NMR spectrum of collagen in D 2 O is reported in Figure S2A. The assignments of the signals of the most common amino acids in collagen, Gly, Ala, Pro and Hyp, were made on the basis of literature data (Fan 1996;Melacini et al. 1996). For large proteins, such as collagen, the 1 H NMR peaks were broadened due to very large dipolar interactions. Thus, resonances of the different functional groups on the protein cannot be resolved; moreover, signals in the region between 4.5 and 5.0 ppm, were masked by the residual water peak. For instance, half height peak widths of Alanine and Glycine detected respectively at 4.32 and 4.13 ppm, were about 60 Hz.
The typical 13 C CP MAS spectra of collagen is reported in Figure S2B. At 25 C almost all signals of collagen can be resolved. The assignment of the NMR signals of the most common amino acids in collagen, Gly, Ala, Pro and Hyp, is known from the literature (Saito et al. 1984;Saito and Yokoi 1992;Aliev 2005). The region from 165 to 185 ppm is relevant to the carbonyl and carboxylic carbons of amino acid residues. The most intense peak at 174 ppm is assigned to the Gly carbonyl, as it is the most abundant amino acid residue in collagen. The peak at 158 ppm is due to carboxylic group of Arg. Small peaks observed in Figure S2B between 127 and 138 ppm are assigned to aromatic carbons of Phe. In the aliphatic region of the spectrum only the peak at 71 ppm, assigned to Hyp-c, is fully resolved. In the region between 60 and 10 ppm a complex envelope of resonances due to aliphatic side chains is present. In collagen, carbons b and c of Pro resonate at 30.4 and 25.7 ppm, respectively. Similarly, carbons b and c of Hyp resonate at 38.9 and 71.1 ppm. Studies on peptides (Aliev 2005) show that the cis X-Pro signal overlaps signals from Glu C-b and Leu c, at ca. 25 ppm, while trans X-Pro signal overlaps Arg b and Lys b at ca. 30 ppm.
The chemical shift of Ala b methyl is 18.1 ppm in collagen ( Figure S2B). The 13 C chemical shift of Ala methyl is sensitive to the polypeptide chain conformation. In particular, the measured value is similar to that in the triple-helix and 3 1 -helix conformations (Saito et al. 1984;Saito et al., 2000). Another example of the conformational dependence is the 13 C chemical shifts of Pro/Hyp b and c, which have been used to differentiate between cis and trans-rotomers along the peptide bond (X-Pro) preceding prolines in unfolded protein (Dorman and Bovey, 1973;Sarkar et al. 1987).

Characterization of type I collagen in the presence of catechin
The 1 H NMR spectrum of collagen in the presence of catechin 0.01 M is shown in Figure 1A. Catechin resonances are detectable and show defined multiplicity and linewidth relatively sharp. The interaction of catechin with collagen is evidenced by broadening of the NMR peaks of collagen. This behaviour could be ascribed to catechin-induced cross-link leading to a reduction in collagen chain mobility. Differences in the 13 C CP MAS spectra of native collagen and collagen treated with catechin 0.01 M and 0.001 M were highlighted in Figure 1B: no significant changes of peak positions and no new peaks appeared as a result of catechin treatments.
No differences in the linewidth among the three spectra were observed. The amplitude of methyl Ala b signal increased as catechin concentration raised. This means that the addition of catechin led to an increase in resolution, because the molecular dynamics resulted enhanced when the concentration of catechin increased. Concerning the 13 C chemical shift of Pro b and c signals (respectively at 30.4 and 25.7 ppm), the relative ratio Pro c/Pro b was 1/1 in non-modified collagen. After catechin treatments, significant increases were observed in Pro c signal; collagen treated with 0.001 M catechin showed 6% increase in the Pro c/Pro b ratio, while a treatment with catechin 0.01 M led to an 8% increase. Also, the ratio Hyp c/Hyp b changed with catechin treatments but it was not possible to be quantified because of the intense glycine peak at 43.30 ppm.
The 13 C CP MAS spectrum of catechin was also considered to verify if changes in Pro c/Pro b ratio were ascribed to the interaction between catechin and collagen and not to the catechin structure.
CP-MAS spectrum of catechin with the carbon assignment is reported by Mart ınez-Richa and Joseph-Nathan (2003). No peak from catechin was identified in the CP-MAS spectrum of collagen-catechin complex, due to the very low levels of catechin in the complex with collagen.
Interactions between catechin and collagen were also investigated by means of 2 D NMR spectroscopy. NOESY and ROESY spectra were acquired (at different mixing times) to this purpose. NOESY spectrum (acquired at 150 ms of mixing time) highlighted a weak correlation spot among catechin and collagen protons as evidenced in Figure S3. ROESY spectra (data not reported) showed the same correlation spot. The interaction among catechin protons was found at 6.80 ppm and that among collagen protons at about 2 ppm. According to previous assignments, the correlation spot was to be ascribed to the interaction between B ring protons of catechin (probably throughout the 2' and 5' protons) and hydroxyproline and proline protons of collagen. The B ring interaction was most likely occur because of its minor sterical hindrance compared to A ring.  Figure 2A shows the high resolution 1 H NMR spectra related to native collagen and collagen oxidized by the three-oxidation systems Fe(II)/H 2 O 2 , Cu(II)/H 2 O 2 , NaOCl/H 2 O 2 . A significant change in peak positions and in the line-width was observed as a result of oxidation mediated by metal systems. Results are summarized in Table S1.

Type I collagen oxidation
The comparison between non-modified collagen and collagen oxidized by Fe/H 2 O 2 and Cu/H 2 O 2 suggests that both oxidation systems triggered an increased mobility of Gly and Ala. Thus, these amino acids are most likely the main chelation or oxidation sites of metal-based oxidation systems that lead to radical formation. In fact, peak half height widths (PHHW), which were about 60 Hz in non-modified collagen, were greatly reduced in oxidized samples. Chemical shifts and peak half height widths were reported in Table S1 for Gly and Ala. Also, we noted that no proton peaks relative to Pro and Hyp (at 2.04, 2.27 and 5.30 ppm) were detected after oxidation by copper system (Figure 2A, spectra c). Compared to native collagen, the oxidized ones showed a strong chemical shift variation for the proton resonances of Gly a CH e Ala a CH (about 0.5 ppm for Fe(II)/H 2 O 2 system and about 1 ppm for Cu(II)/H 2 O 2 system). Concerning Gly a CH and Ala a CH, Cu(II)/H 2 O 2 system showed a larger peak half height width than Fe(II)/H 2 O 2 system. These paramagnetic shifts can be divided into two components, a scalar (through-bond) contact shift and a dipolar (through space) pseudocontact shift. Both paramagnetic shifts involve the formation of a complex between protein and the paramagnetic ions. Pseudocontact interaction is more interesting concerning proton spectroscopy and can be defined by the following equation (Bertini et al. 2005): Dv ax sin 2 h cos 2u The values of Dv ax known in literature for iron II and copper II (Bertini et al. 2005) and the broadening of Gly and Ala peaks may indicate that copper was closer than iron to the oxidized sites. These considerations were in accordance with studies that pointed out the higher specificity of Cu(II)/H 2 O 2 oxidation system compared to the iron system (Uchida et al. 1992;Hawkins and Davies 1997). Also, the singlet oxygen (Figure 2A, spectra d) induced damages to the polymer (from the literature mostly attributed to actions of cross-link), but, probably, this attack followed a random mechanism. In Figure 2B, typical 13 C CP MAS spectra of dry collagen fibrils (spectra a) and oxidized collagen fibril (spectra b-d) were plotted. There were not significant changes of peak positions between the four spectra but a new peak (at 63 ppm) appeared as a result of collagen oxidation mediated by iron system. CP MAS spectra demonstrated that collagen was sensitive to be damaged mostly by the Cu(II)/H 2 O 2 system rather than by the others. As a result of oxidation, change in the Pro c/Pro b ratio in collagen molecule was observed compared to the non-modified collagen.
As reported in Figure 2B (spectra c), Cu(II)/H 2 O 2 oxidation system interacted especially with Gly (peak at 48 ppm) and Pro (peaks at 30.4 and 25.7 ppm) amino acids identified between 20 and 50 ppm. Moreover, the CO signal exhibits little changes in peak shape when collagen was oxidized by metal/H 2 O 2 systems, whereas the aromatic carbons of Phe and the carboxylic group of Arg at 127-138 and 158 ppm respectively) were strongly reduced by both metal oxidation systems. In the Fe(II)/H 2 O 2 system ( Figure S4, spectra b), the Hyp a and Hyp d and Pro a signal resolution, at about 60 ppm, was improved by oxidation and additional resonance line could be resolved. This behaviour was probably due to the paramagnetic effect of iron on Pro or, in term of structural changes, to the order-disorder transition in collagen. This effect was not observed in collagen oxidized by copper system (Figure 2B, spectra c), probably because the peak broadening did not allow peak resolution. Figure S5a shows 1 H NMR spectra obtained from the oxidation experiments carried out on the collagen molecule modified with catechin 0.001 M (similar spectra profile was obtained from oxidation experiments performed on the collagen modified with 0.01 M catechins). Significant changes in Ala and Gly peak position after Cu(II)/H 2 O 2 oxidation were observed. A marked effect of catechin on copper oxidation was evident from Figure S5A (spectra b) compared to the non-modified collagen. The Gly and Ala peaks showed a variation in chemical shift and a narrowing of the NMR signals after Cu(II)/H 2 O 2 oxidation experiments compared to the results obtained for native collagen. In both collagen treated samples oxidized by copper system ( Figure S5A, spectra b) proton peaks ascribed to Pro and Hyp, not detectable in the non-treated sample, was observed. Chemical shifts and peak half height widths (PHHW) for Gly and Ala peaks of all the samples analysed were reported in Table S1. Peak half height widths decreased after both metal oxidation system treatments compared to non-modified collagen, especially when the Cu(II)/H 2 O 2 oxidation system was involved. This could be ascribed to an increased copper distance from collagen attack sites compared to the results obtained with the non-modified collagen. The variation in chemical shift of Cu(II)/H 2 O 2 and Fe(II)/H 2 O 2 was similar. Besides, the proline peak (at 5.4 ppm) was not completely destroyed in these conditions. No relationships between catechin concentration and oxidation level were found.

Oxidation of type I collagen modified with catechin
CP MAS spectra of collagen modified with 0.001 M catechin obtained before and after oxidation experiments are reported in Figure S5B. Not significant changes in peak positions among the four spectra were observed, however, a new peak (at 63 ppm) appeared as a result of collagen oxidation mediated by iron and copper system. 13 C CP MAS spectra confirmed results obtained from 1 H NMR analysis. Cu(II)/H 2 O 2 oxidation, as well as Fe(II)/H 2 O2 one, led to the Hyp a and Hyp d and Pro a signal resolution, at about 60 ppm. The protective effect against oxidation exhibited by catechin was much stronger in the case of copper oxidations. This was clearly notable in the Pro b and Pro c 13 C chemical shift. The 13 C CP MAS spectra showed that in the case of collagen treated with catechin, the initial triple-helix order of the native collagen is partly lost on oxidation mediated by metal system. Data from collagen oxidation experiments, highlighted the protective role played by catechin. The paramagnetic effect of iron evidenced on Pro a 13 C signal, was also showed by copper system on collagen treated sample ( Figure S5B, spectra c). This behaviour was probably induced by the increased metal-nucleus distance caused by the presence of catechin.

Conclusions
Data obtained from 1 D and 2 D proton NMR and 13 C CP MAS spectroscopy pointed out that interactions of collagen with catechin preferentially occurred between catechin B ring and the Pro and Hyp amino acids.
Oxidation studies carried out by metal/H 2 O 2 systems on non-modified and catechin-modified collagen showed that both iron and copper were able to interact with collagen by site-specific attacks. We have also investigated the binding of Cu(II) and Fe(II) ions to collagen substrate. The observation that proton NMR signals of Gly and Ala showed a pseudocontact shift, suggests that both copper and iron ions bind to these amino acids in collagen and that copper is closer than iron to the collagen attack site. The main effect of catechin in the collagen complex was to protect Pro from oxidation, preventing copper and iron approach to collagen molecule.

Disclosure statement
No potential conflict of interest was reported by the authors.