3.1 Lignin Properties
2D-HSQC analysis is a powerful method to determine lignin structures.24 13C1H correlation spectra in the 2D-HSQC analysis of the lignin samples were identified according to previous literatures.[24-26] The side-chain region (δC/δH 50.0–90.0/2.50–6.00) and aromatic region (δC/δH 100.0–150.0/5.50–8.50) of lignin are shown in Fig. 1A-1F and 1a-1f, respectively.
According to Fig. 1A, the original kraft lignin (control) showed prominent C-H correlations of β-O-4 linkages at Cα-Hα (δC/δH 71.3/4.76ppm), Cβ-Hβ (δC/δH 83.9/4.30ppm), and Cɤ-Hɤ (δC/δH 60.2/3.35-3.69ppm). A strong signal of methyl groups was observed at δC/δH 55.7/3.77 ppm. The presence of resinol (Fig. 3II) was confirmed by C-H correlations of Cα-Hα, Cβ-Hβ, and Cɤ-Hɤ at δC/δH 85.3/4.62, 53.7/3.04, and 71.0/3.76 and 4.16 ppm, respectively. The major component of the kraft lignin is G subunits, as dense signals at δC/δH 109.6/7.16, 115.4/6.74, and 119.6/6.97 ppm represent C2-H2, C5-H5, and C6-H6 of guaiacyl units (G), respectively (Fig. 1a). The light spot at δC/δH 103.5/6.63 of C2,6-H2,6 implies a low amount of S units in the kraft lignin. Correlations from C2,6-H2,6 in the p-hydroxyphenol structure were observed at δC/δH 128.2/7.17 ppm.
Acidic pH at 4.5 had a minor influence on lignin structure since similar spectra were observed between the control and L4.5. On the other hand, alkaline environment caused significant changes in lignin structure. A decrease in correlation intensity of the methoxyl groups (δC/δH 55.7/3.77 ppm) was observed in all alkali treated lignins. The C-H correlations of Cα-Hα and Cβ-Hβ of β-o-4 linkages disappeared, and a low response of Cɤ-Hɤ in L8.5, L12, and L12-4.5 indicated the cleavage of β-O-4 linkages, which resulted in higher phenolic hydroxyl groups.[27] Some β-O-4 linkages were reformed after pH shifting in alkali treatment (i.e., L8.5-4.5); however, this was not observed when the initial pH was extremely high (i.e., L12-4.5). A similar scenario was found in β-β linkage in resinol (Fig. 3II). The signal assigned to S units was invisible in all pH-treated lignin samples. The C-H correlations of p-hydroxyphenol (*) also disappeared in alkali and alkali-shifting treatments. Compared with untreated lignin, alkali treated lignin samples have a less intense C5-H5 signal and only a faint response at C2-H2, and C6-H6 positions. These findings confirm the replacement of H atoms in C2, C6, and C5 of G units.
Changes in lignin structure obtained from NMR were confirmed by morphological images. Unmodified lignin was scanned at 400x magnification as seen in Fig. 2a. A mixture of large particles, both rough and smooth, was observed, and the particles were separate from each other. Based on the NMR spectra, there was no significant change between L4.5 (Fig. 1B and b) and the control (Fig. 1A and a). But based on scanning electron images, L4.5 was depolymerized and contained smaller particles (Fig. 2b). These particles formed interactions and attached to each other as shown in Fig. 2b, indicating significantly smaller particle sizes and changes in electrostatic charges. On the other hand, alkaline treatments resulted in sharp-edged lignin as shown in Fig. 2c. Higher ratios of sharp-edged lignin were observed at the extremely high pH (Fig. 2e). Since acid hydrolysis broke down lignin into small particles, whereas base hydrolysis created sharp-edged lignin, the pH-shifting process accordingly created a mixture of both lignin characteristics (Fig. 2d and 2f). In addition, small linkages between 2 clamping particles were observed (Fig. 2d and 2f at the red arrows), which suggests additional lignin-lignin interactions and/or repolymerization after pH shifting.
Lignin is stable at high temperatures due to its condensed aromatic and highly branched structures. Thermal degradation of lignin occurs in a broad range of temperatures (Fig. 3) because lignin contains various aromatic side chains and branches attached to different positions on the aromatic units. The initial weight loss (25-200 °C) is mainly from moisture evaporation.[28] After that, pyrolysis reaction takes place and contributes to major weight loss. The first degradation stage is the dehydration of hydroxyl groups (150-275 °C) and breakage of ether bonds (150-300 °C). Then aromatic rings start to lose aliphatic side chains. Still higher temperatures (370-400 °C) break C-C linkage. After that, backbone degradation at temperatures between 500-700 °C results in gaseous products.[29] The degradation temperature is affected by fragmentation of inter-unit linkages and the content of C-C bonds.[30, 31] Lignin with higher degradation temperatures contains a higher portion of high molecular weight compounds.[32]
The thermal stability of unmodified lignin was slightly higher due to its high G content.[33] Tg of all modified lignins were in the range of 338.48-389.15 °C, whereas unmodified lignin Tg was 404.19 °C (Table 1), which suggests that acid, base, and pH shifting depolymerized lignin and resulted in lower molecular weight compounds. In Fig. 3, different derivative weight curve patterns suggested that different subunits were produced in different treatments. Acid modification caused a lower Tg but did not significantly change the thermal profile of unmodified lignin. On the other hand, a shoulder around 410-430 °C appeared in L8.5 and both pH-shifting samples, suggesting repolymerization of lignin. L12 presented a distinctive thermoresistance profile: the first peak indicating water evaporation and hydroxyl group dehydration was bigger than that of other lignins. L12 also showed a unique weight loss profile (Table 1). More than a quarter of total weight loss came from the dehydration of hydroxyl groups and evaporation, whereas only 3-8 % of those was released in other modified lignins. This indicates that extremely high pH treatment increased lignin’s hydroxyl groups and water binding capacity. Moreover, in L12, only about 55% of weight was lost between 200-500 °C, whereas 74-81% was lost in this range in other lignins. This suggests that in L12, some part of the lignin skeleton had broken in the extremely high pH environment, and some small compounds were released. These compounds likely contributed to the odor perceived during the sample preparation. As a result, a relatively small backbone degradation peak was observed in L12, and approximately 60% ash was recovered after the test. All lignins emitted approximately the same amount of gaseous products after 500 °C. L12 showed the highest thermoresistance properties with ~60% residual after 700 °C, indicating that lignins modified with a high pH are likely to be more thermostable than acid treated or untreated lignins.
3.2 Lignin-SP and SP Morphology
According to Fig. 4a, the average single agglomeration of pure SP was 60-70 nm. Larger agglomerations (less than 500 nm) also formed. Lignin linked protein agglomerations and resulted in a large clamp (Fig. 4b). The size of the lignin-protein network was larger than 500 nm. The quantitative analysis of particle sizes was shown in Table 2. Lignin particles became smaller and more uniformed after being treated with different pH values. In severe alkaline environment (pH 12), particles were undetectable, as all samples were completely soluble. Exposing protein to alkaline conditions before aggregation resulted in bigger clusters. The biggest cluster (131.559±32.526 μm) was observed after exposing protein to denaturing pH. This means that denatured protein can refold to looser structures that are different from the folding structures of non-denaturing protein.
The particle size of LSP8.5 was larger and more varied than that of SP8.5 and L8.5. The bigger complex may be caused by covalent bonds and cross-links between lignin and protein. Since the average particle size of LSP8.5 was bigger than the single-pair binding of L8.5-L8.5, L8.5-SP8.5, or SP8.5-SP8.5, multiple-interactions and lignin-protein networks could have formed. This suggests the existence of multiple active sides in lignin and protein and supports a model proposed in a previous study[7]. However, lignin-protein clusters at the isoelectric point of LSP4.5, LSP8.5-4.5, and LSP12-4.5 were smaller than that of SP4.5, SP8.5-4.5, and SP12-4.5, respectively. Since the active sites of protein already bonded to lignin, less active sites were available to form interactions with protein when the pH was brought to pI (pH 4.5). Moreover, the presence of lignin in the system disturbed the formation of large protein networks. As a result, the cluster of LSP was smaller than that of SP samples in all conditions.
Dried film morphology was shown in Fig. 5. At higher pH values, SP films turned yellow. LSP films also turned dark brown due to the brown color of lignin. All pH 4.5 and pH-shifting films were flat and had tendency to stick on the micro slide. At pH 8.5, film partially lifted from the slide and formed a corrugated film with air inside. Small bubbles were spread along the film in both SP and LSP samples. In contrast, a large bubble was formed in the middle of the SP12 film with cracks around (Fig. 5d), which suggests more cohesion at high pH values. Light reflection was seen in both SP12 and LSP12 dried films.
Film texture was observed under an optical microscope. More cracks were found in SP samples than LSP samples. Small bubbles, both broken and complete, were found in SP8.5, whereas in the LSP8.5 film, air was trapped and formed bigger unbroken bubbles (Fig.5 B and G). This indicates a stronger cohesion force in LSP films than in SP films. The higher strength of LSP films was also observed at pH 12: in the SP12 film, small fragments sprang from the main film at the edge of the bubble, whereas fragments were not observed in the LSP12 film (Fig.5 D and I). Therefore, lignin strengthened and improved the cohesiveness of SP films.
Film morphology under water strongly reflects a sample’s water resistance. The appearance of SP and LSP 4.5 were similar (Fig. 6a and f), while differences were observed in alkaline conditions. At pH 8.5, many small SP8.5 film pellets detached at the edge of the film, whereas the LSP8.5 film was intact with no free pellets (Fig. 6b and g). Conformational changes start to occur at pH values above 11,[34] and conformational changes and chemical changes influence each other. As protein denatured and unfolded, polar and non-polar groups were exposed and became more available to interact with lignin.[35] Thus, more remarkable improvement and protein-lignin interactions were detected at extreme alkaline pH values. The SP12 film floated freely and was completely dissolved (Fig. 6d). By contrast, the floated LSP12 film was thick and partially dissolved in water. The film absorbed water and formed a gelatinous pellet. The improvement in morphology after soaking in water demonstrates that lignin improves protein water resistance possibly by crosslinking and other lignin-protein interactions. As seen in Fig. 6c and h, mild pH shifting process (8.5-4.5) did not change the film appearance of SP and LSP samples. However, extreme alkali-shifting (12-4.5) affected protein solubility and led to the exposure of hydrophobic groups and cleavage of native S-S bonds.[36] With the interference of lignin on protein aggregation as mentioned in section 3.2, the LSP12-4.5 film showed many fractures, but the SP12-4.5 film stayed intact (Fig. 6e and j). Therefore, the water resistance of SP and SPL films is not only influenced by protein characteristics but also the strength of lignin-protein interactions as a consequence of pH and pH shifting.
3.3 Lignin-Protein Interactions
The syringe contained 84 µM lignin and the cell contained 3 µM soy protein. The 1st peak is a mock injection. Upper panel: raw data showing heat pulses obtained by multiple injections; Lower panel: integrated heat of binding as a function of molar ratio of ligand:macromolecule. The graph shown is after the subtraction of negative control where lignin was injected into the buffer.
The binding interaction of soy protein with lignin was studied using isothermal titration calorimetry (ITC). The exothermic peaks observed during the titration (Fig. 7) suggested that some binding took place. However, attempts to obtain reproducible ‘N’ values were unsuccessful, which led us to believe that the binding was of non-specific types. The interpretation of binding was further complicated by the lack of specific molecular weight data for the lignin and soy protein. Hence, for the current observation, the molecular weight was estimated to be the sum total of molecular weights of all the subunits.
The differences between SP and LSP samples mentioned in sections 3.1 and 3.2 were found at pH 8.5. No band was detected in the negative control lignin sample (Fig. 8a, lane 1). In the pure SP sample (lane 2a), the distribution pattern of α0, α, and β subunits from 7S around 75 and 50 kD was similar to that found in Qi et al. (2011). Acid and basic polypeptides of 11S also appeared at lower molecular weight bands. The variation in molecular sizes could be caused by differences in electrostatic interactions at different pH values.[37] A similar protein subunit distribution pattern was presented in the LSP sample (Fig. 8a, lane 3), indicating that some polypeptides did not interact with lignin. A significant band was detected in LSP8.5 (lane 3a) around 250 kD and >250kD but not in SP8.5 (lane 2a), suggesting that lignin-protein complex was formed. As discussed in section 3.1, the lignin-protein networks were formed with a broad range of particle sizes, which correspond with the extensive bands (ranging from ~30 to >250 kD). This phenomena was also reported in a previous study in which high molecular weight fractions of cross-coupling products, coniferyl alcohol (a lignin major subunit)-amino acids complex, were observed.[2] Lignin-protein complex fractions were also observed at room temperature. The same protein subunit distribution pattern was also found in fresh SP and LSP samples. There were no bands in lignin samples (Fig. 8b lane 1). The lignin-protein complex band was found in the same size range as in Fig. 8a, and the band intensity elevated from Fig. 8b lane 3 to lane 7 as lignin concentration increased from 10 to 50%, indicating that more lignin-protein complexes were formed with more lignin in the system.
Elemental composition analysis reflects the overall composition of samples. Table 3 shows (%) the ratio of oxygen to carbon atoms (O/C) of lignin, protein, and lignin-protein samples. The O/C of the lignin was 35.21% at pH 4.5. The (%) O/C increased to 84.51 and 75.28 for L12 and L12-4.5, respectively. The additional oxygen atom in the structure mainly resulted from the cleavage of β-O-4, which created more active groups such as hydroxyl group. This is especially the case in extreme alkali environments. These data confirmed the 2D-HSQC NMR results in section 3.1 that more reactions and changes occurred in extremer alkali treatments.
The (%) O/C of protein was consistent in SP4.5, SP8.5, and SP8.5-4.5. The higher (%) O/C was found at denaturing pH (pH 12). This may be caused by hydrolysis of the peptide bond (-CO-NH-) on the polypeptide chain, which resulted in –COOH and -NH2 groups. The additional oxygen atom in –COOH group could have contributed to the significantly higher (%) O/C in SP12 and SP12-4.5. The changes in LSP samples followed the same pattern.
The (%) O/C of Lignin+SP was calculated based on the blending ratio of lignin and SP with the assumption that there were no chemical and bonding interactions between lignin and protein that could change the compositions of the molecules. The difference between the actual value obtained from LSP samples and the calculated Lignin+SP value (Table 3) indicates changes in the oxygen content after reactions. The negative differences indicate an interaction between lignin and SP, since LSP samples have less oxygen than the calculated Lignin+SP values. Loss of oxygen could imply covalent crosslink interactions between lignin and protein samples. Since lignin and protein have active groups such as amino, hydroxyl, carbonyl, and carboxyl groups, the reactions that involved oxygen were likely to be condensation and esterification.
At the most extreme condition of pH 12-4.5, lignin, soy protein, and lignin-soy protein gave the clearest IR spectra. The IR spectrum of lignin shows aromatic skeleton vibrations at 1595, 1510, and 1422 cm-1. The shoulder at 1701 cm-1 corresponds to the C=O stretching in conjugated carbonyl compounds with aromatic rings. The absorption at 1457 is attributed to C-H bending and vibrations of aromatic groups. The absorption peak at 1213 and 1029 corresponds to C-O bond and bending vibrations of G subunit’s aromatic plane, respectively.[38]
Protein major absorption bands Amide I, II, and III were observed at 1640, 1520, and 1233, respectively. Amide I contains major stretching vibrations of the amide group (80%) and C-N bond. Amide II peak comes from N-H bending (60%) and C-N stretching (40%) vibrations. Amide III peak mainly describes C-N stretching vibrations and N-H in-plane bending vibrations.
According to Fig 9, compared to SP samples, LSP samples showed differences in all 3 amide peaks. First, the amide I absorption peak of LSP was slightly weaker than that of SP. This indicates the stronger H-bond near the C=O group. The stronger H-bond reduces electron density around C=O and results in a lower absorption intensity. In an extreme pH environment, extended peptide chains are aligned closer to neighboring chains after unfolding, which contributes to strong intermolecular hydrogen bonds. Second, the amide II peak absorption of LSP showed a slight shift in peak shape and position. This could be caused by the influence of the lignin adsorption band or changes in the chemical environment, because frequency shifts can be true shifts or caused by relative intensities of component band variations.[39] The frequency and absorption bands of amide I and II could be influenced by the strength of hydrogen bonds involving amide C=O and N-H groups. The stronger H-bonds indicate stronger secondary structures in LSP samples. Last, the amide III peak of LSP was broader, and the peak slightly shifted from 1233 to 1230 cm-1 with the shoulder appearing at 1273 cm-1. This change is possibly caused by the lignin absorption peak at the same position, 1273 cm-1. However, it is difficult to interpret the amide III absorption band as it could also be influenced by C=O in-plane bending, C-C stretching, and CH2 wagging vibrations. The absorption band at 1392 cm-1 is attributed to the stretching and vibration of the COO- group.[40, 41] Since SP and LSP samples have the same-size absorption peak at 1447 cm-1, while the LSP sample showed a relatively low 1392 cm-1 peak, it is highly possible that there were interactions between lignin and the COO- group of protein. In addition, different absorption patterns between SP and LSP samples were observed between 1200 and 1000 cm-1. As the spectrum in the 1300-1000 cm-1 region is related to the C-O stretching of alcohol, phenol, and carboxylic groups, changes in the LSP sample might be caused by the influence of lignin absorption, interactions of C-O, changes in the chemical environment, or a combination of these factors.
Fig. 9b is the difference spectrum where the spectrum of SP is subtracted from that of LSP and compared with the spectrum of lignin. Additional peaks at 1728 and 1642 cm-1 represent stretching vibration of unconjugated and conjugated carbonyl groups (C=O), respectively.[38, 42-44] Fig. 9c is the difference spectrum where the spectrum of lignin is subtracted from that of LSP and compared with the spectrum of SP. In Fig. 9c, the amide I peak shifted to higher frequency together with an increase in peak intensity. A conjugated carbonyl group represented inter- and intramolecular hydrogen bonds that generally occur between O-H and N-H, for example, the -COO- and -OH groups of protein and lignin.[45] Thus, these changes in both subtracted LSPs spectra in Fig. 9b and c indicate that strong molecular interactions were formed between lignin and the C=O group of protein. An ester bond was possibly formed by alcoholysis and ammonolysis of -COOH in acid condition. Ester, ether, and the bonds between C/N, C/S, S/S were previously observed between amino acid side chains such as lysine and phenolic rings, and disulfide bonds between lignin and protein.[2, 18, 46], [47] Moreover, change in aromatic skeleton vibration was also observed. First, the 1595 cm-1 peak indicated a mixed signal of aromatic skeleton vibration and C=O stretching. The reference lignin peak response was from aromatic skeleton vibration, whereas the subtracted LSP peak response was from both aromatic skeleton vibration and C=O stretching. The peak intensity of subtracted LSP was less than that of the referenced lignin, indicating that lower response came from aromatic skeleton vibration. Second, the spectrum at 1510 cm-1 corresponds directly to the aromatic skeleton vibration, which disappeared after subtraction. Third, the amide II absorption band shifted from 1520 to the higher wavenumber 1540 cm-1, and the relative intensity of amide II increased compared to that of amide I in the subtracted spectrum (Fig. 9c). These changes indicate a new C-N bond formation between protein and lignin.[48] The interaction between amino groups attached directly to C2 and C5 of phenolic compounds through quinine intermediate[18] limits molecular movement and vibration of the aromatic ring.
On the other hand, compared to the reference lignin spectrum, the subtracted LSP spectrum showed slightly shifted and stronger signals of C-O-C (1366), C-O of the G subunit (1273), C-C and C-O (1213), C-H in plane deformation (1143), C-O deformation of secondary alcohol and aliphatic ethers (1092), C-O of the guaiacyl group, and blending vibrations inside of the aromatic plane of the guaiacyl ring (1029). Stronger intensity in the range of 1000-1400 cm-1 indicated the formation of lignin-protein linkages. New absorption peaks at 1481, 900, and 836 cm-1 of the subtracted spectrum were observed. The new peaks likely indicate special covalent crosslinking between lignin and protein. However, further investigation is needed for more specific evidence. The changes in peak intensity in this range are likely caused by non-specific interactions. Strong cross-links are supported by many kinds of electrostatic interactions. H-bonds were claimed to be the dominant electrostatic interactions between protein and phenolic compounds in aqueous solutions[49] Hydrophobic interactions and electrostatic forces were also observed between hydrophobic regions of soy protein and close to the rings of lignin. In addition, when lignin and protein molecules are close to each other, attraction and repulsion between positive and negative regions take place and lead to ion-dipole and ion-induced-dipole interactions and dipole-dipole interactions.[23] However, the mechanism of lignin-protein interactions has not been clearly elucidated.
It is difficult to predict or quantify lignin-protein interactions and networks because the formation of bonds and forces depends highly on the nature of lignin and protein and environmental factors. Lignin has different structures, functional groups, and molecular weights depending on plant species and extraction methods. In addition, various kinds of plant protein also have diverse amino acid sequences, subunit profiles, and folding structures. The diverse nature and structure of lignin and protein significantly determine the nature of interactions.[50] Reaction type and rate also highly depend on environmental conditions[45, 50, 51] Organic solvents and pH not only impact protein but also lignin, affecting their solubility, degree of folding, denaturing stage, exposure of functional groups, hydrophobic/hydrophilic properties, and degree of depolymerization.