The physical basis of fabrication of amyloid-based hydrogels by lysozyme

The fabrication of amyloid-based hydrogels has attracted remarkable attention within the field of materials science and technology. These materials have a multitude of potential applications in the biomaterials field such as developing scaffolds for tissue engineering, drug delivery and hygiene products. Despite the potential new applications of these materials, the physical nature of their assembly is not well understood. In this study, we have investigated how the conformation of the amyloid precursor state (I) is formed and correlated with the assembly of amyloid-based hydrogels. A transparent hydrogel was fabricated at pH 7.4 by cooling of the temperature-induced unfolded state of hen egg white lysozyme (HEWL). The completely unfolded state (U) at the gelation concentration of HEWL was obtained around 90 °C in the presence of tris(2-carboxyethyl)phosphine (TCEP), with a TCEP/HEWL molar ratio of 4 : 1. The characterization of the hydrogel showed that it was composed of an amyloid fibril-like material. The physical nature of its assembly was examined in detail and it was found that the hydrogel formation reaction was a three-step, four-states process (U → I → F → H). We concluded that the properties of the pre-molten globule state (I) of the protein correlated only with the fibrillation process, whereas the assembly of the fibrils into an hydrogel was found to be almost independent of the I state. Thus, the study presented here provides a complete biophysical insight into the pathway of lysozyme hydrogel assembly.


Introduction
The self-assembly of proteins and peptides into amyloid-like brils is mainly known for its notorious role in many pathological conditions such as Alzheimer's disease, Parkinson's disease, and type II diabetes. 1 However, protein brils are known to play functional roles in many organisms, including bacteria, yeasts and humans. 1,2 Since amyloid-like brils have been found to possess remarkable stability, tensile strength and tunable physicochemical properties, they can be fabricated into smart materials for various bio-nanotechnological applications. 3,4 Amyloid-based hydrogels (AbHs) are one of such smart materials and are very useful for various biomedical and nanotechnological applications. [5][6][7] Amyloid-based hydrogels (AbHs) are water-laden, threedimensional materials formed by cross-linking of protein brils. The ability of AbHs to hold large amounts of water comes from the fact that they have a large number of hydrophilic functional groups present in their polymeric chains. They resemble living tissues because of their viscoelasticity and biocompatibility. 8 They could be very useful in the development of materials for different biomedical applications such as superabsorption, drug delivery and tissue engineering. [5][6][7]9 Recently, Yang et al. have prepared a low-cost injectable AbH, which can carry embedded drugs. 10 Moreover, JD et al. have developed a stimuli-responsive AbH, which may be suitable as an injectable material for cell delivery and tissue engineering applications. 11 Several proteins and peptides are known to form hydrogels via the cross-linking of amyloid brils. [10][11][12][13] The conversion of brils into thermo-responsive hydrogels was found to be generally controlled by the concentration of proteins/peptides and the temperature of the solutions. Fmoc-protected amyloid beta peptides form AbH on cooling the heated peptides to room temperature. 14 Yan et al. have shown that adding 20 mM DTT to a 3.0 mM lysozyme solution, heating at 85 C for 10 min and cooling to room temperature lead to the formation of amyloidbased hydrogels. 15 Using multiple techniques, they have also demonstrated that the hydrogels were formed via network formation of amyloid brils during the cooling process. 16 However, it is not clear how these proteins behave during heating and cooling processes and what is the effect of protein concentrations on the process. Therefore, in this study, we have systematically answered the following questions: (1) can unfolding/refolding, aggregation and cross-linking reactions be studied independently? (2) How does HEWL in the presence of the reducing agent TCEP unfold and refold during heating and cooling processes? (3) What is the conformation of the protein at the heating and cooling end of the above three reactions? and (4) what is the correlation between the formations of the monomeric aggregation precursor state, the amyloid brils and hydrogel formation?
In this study, we found that HEWL at protein concentrations $300 mM preincubated with TCEP in a TCEP/HEWL molar ratio of 4 : 1 and heated at 90 C, converted into a self-healing hydrogel immediately aer cooling the solution to 25 C. We also standardized the concentrations of HEWL at which it unfolds and refolds into monomeric states under hydrogel fabrication conditions. We identied the monomeric aggregation precursor (I) state of the lysozyme formed by the refolding of the completely unfolded protein. We characterized the conformation of the I state using multiple spectroscopic techniques and found that it resembled a pre-molten globule state of proteins. We also standardized a range of HEWL concentrations at which it formed amyloid brils, but not the hydrogel. We established a correlation between these individual reactions. The combination of these individual steps reveals the global mechanism for the fabrication of this hydrogel. The mechanism provided here involves structures and the basis of the formation of on-pathway intermediate for hydrogel formation. We believe that this study will help in the design of desired hydrogels for various biomedical applications and materials development.

Materials
Hen egg white lysozyme (HEWL), Congo red (CR), tris(2carboxyethyl)phosphine (TCEP) and thioavin T (ThT) were purchased from Sigma Aldrich Co., USA. Rest of the chemicals used during the experiments were of analytical grade with purity greater than 99%. A PICO + Benchtop pH-meter (LabIndia Instruments Pvt Ltd) was used for pH measurements. A 0.2 mm Millex-LG syringe lter (Millipore, USA) was used to lter the buffer solutions before use. The concentrations of HEWL, CR, ANS and ThT solutions were determined using a UV/visible spectrophotometer (UV-1800, Shimadzu Corp, Japan), as described previously. 17

Preparation of the hydrogel
Hydrogel formation was initiated by a combination of heating and cooling steps. Briey, samples containing 700 mM of HEWL and various concentrations of TCEP in Tris buffer at pH 7.4 were heated at six different temperatures for a period of 2-10 min in a water bath. The heated samples were brought back to 25 C and were observed for the hydrogel formation using the test tube inversion test.

Transmission electron microscopy (TEM)
The morphology of the aggregates and hydrogel materials was analysed using transmission electron microscopy (TEM) imaging. The aliquots of ve-fold diluted samples were placed on a 200-mesh carbon-coated copper grid. Aer incubation for 2 min, the samples were stained using a 2% (w/v) uranyl acetate solution for 2 min. Excess stain was removed, and the samples were allowed to air-dry. The samples were analysed utilizing a Philips CM200 TEM operating voltage of 20-200 kV with a resolution of 2.4 A.

FTIR spectroscopy
FTIR spectra of native HEWL and the hydrogels were scanned between 1700 cm À1 and 1600 cm À1 (amide I band) using a Vertex-80 FTIR instrument (Bruker, Germany) equipped with a DTGS detector. 10 ml of each sample was placed and dried on a translucent KBr pellet, which was made as described previously. 14 10 ml of each blank sample (sampleprotein) was also put on another KBr pellet and used for the background spectrum. The spectra were deconvoluted by Fourier selfdeconvolution (FSD) method. The deconvoluted spectra were then subjected to the Lorentzian curve tting procedure using the opus-65 soware.

Far-UV circular dichroism (far-UV CD)
Far-UV spectra of the samples were recorded between 250 and 200 nm on a Jasco J-815 spectropolarimeter (Tokyo, Japan). For this purpose, a protein concentration of 5 mM, a 1 mm cuvette and a slit width of 1 nm were used. All the spectra were background subtracted with their respective blank. The data was converted into mean residue ellipticity [q] in deg cm 2 dmol À1 using eqn (1).
In this equation CD is the ellipticity in millidegrees, n is the number of amino acid residues, l is the path length of the cell in cm, and C is the molar concentration of the protein.
The contents of a-helices and b-strands of the monomeric states were determined using the K2D3 deconvolution soware. 18

Thioavin T (ThT) binding assay
For each sample, a xed ThT/protein molar ratio of 5 : 1 was used for ThT uorescence measurements. The uorescence was recorded on a Cary eclipse uorescence spectrophotometer at 25 C using a 1 cm path length cuvette. The emission spectra were recorded between 460 and 600 nm by exciting the samples at 444 nm. All the spectra were background subtracted with their respective blank.

Congo red (CR) binding assay
For each sample, a xed CR/protein molar ratio of 5 : 1 was used for CR difference spectra measurements. Aer 5 min of equilibration, absorption spectra were acquired from 400 nm to 700 nm. The difference spectra were obtained by subtracting the spectra of Congo red alone and hydrogel/aggregates alone from the individual spectrum of each CR-complex.

Sol-gel transition study
The transition of protein solutions into the gel form was measured by the ball-drop method on a home-made apparatus. A HEWL (700 mM) sample with a TCEP/protein molar ratio of 4 : 1 was prepared and heated to 90 C for 2 min. A 78 mg steel ball was placed on the surface of the samples using a sliding applicator with a hole in the middle. The sample tube was placed into a glass vessel lled with silicone oil and heated to the desired temperature using a thermo-controlled stirrer. The travel of the ball in the tube was video-recorded at every 5 degree intervals from 90 C to 20 C. The accurate time of the fall of the ball over a pre-marked distance was calculated using Filmora9 videoeditor (https://lmora.wondershare.com/video-editor/). The data was presented as ball-drop time (BDT) versus temperature. The sol-gel transition was also measured by 90 resonance light scattering (RLS), as described in the next section.
The fraction (f) of aggregate (f scat ) and gel (f gel ) was calculated from the BDT and RLS data using eqn (2): where y, y sol and y gel represent the observed BDT or RLS at temperature T ( C), of the solution phase and gel phase of HEWL, respectively. The y sol and y gel values in the transition region were obtained by the linear extrapolation of the signal values observed in the temperature regions before and aer the transition. In the case of RLS data, y gel was replaced with y scat . The transition mid points (T m ) of all the curves were determined using the sigmoidal eqn (3) where b is the slope factor of the transitions.

Aggregation study
Samples containing different concentrations of HEWL (<300 mM) preincubated with TCEP at a TCEP/protein molar ratio of 4 : 1 were heated at 90 C for 2 min. The heated samples were cooled at 3-5 degree intervals from 90 C to 25 C. At each temperature, the aggregation state was monitored using a 90 resonance light scattering (RLS) assays, as described recently by our group. 19 The aggregation prole thus generated was t using the following sigmoidal function (eqn (1)).
where F 0 and F are the observed RLS values at 90 C and T is temperature in C. t 1/2agg is the temperature at which aggregation reaches 50% of the maximum aggregation. m 0 and m 1 are the slopes of the pre-and post-transition regions of the aggregation proles, respectively. b is the slope factor of the transition region.

Unfolding/refolding study
For thermal unfolding, protein samples (3 mM) with a TCEP/ protein molar ratio of 4 : 1 were heated from 25 C to 90 C.
To examine the refolding process, samples were cooled from 90 C to 25 C. The process of unfolding and refolding was monitored using intrinsic uorescence, far-UV CD and ANS uorescence as previously described by our group. 17,20 The resulting transition curves were also analysed to calculate the T m of unfolding/refolding reactions as previously described by our group. 17,20

Sample space of hydrogel formation
Hydrogel formation by hen egg white lysozyme (HEWL) was induced by the heating/cooling processes. Fig. 1 shows the schematic of the sample space for the formation of the lysozyme hydrogel (H). First, HEWL (700 mM) samples preincubated with different concentrations of TCEP at pH 7.4 were heated at six different temperatures ranging from 40 C to 90 C. Second, these samples were allowed to cool to 25 C. Finally, the hydrogel formation of the protein was examined using the tube inversion test. 21 Samples with TCEP heated at and above 60 C were observed to form hydrogels immediately when the temperature was brought down to 25 C. It was also found that protein solutions preincubated with a TCEP/HEWL molar ratio of 4 : 1 formed transparent gel ( Fig. 1). However, the samples incubated with lower molar ratios of TCEP/HEWL converted into turbid gels (Fig. 1). The turbidity of the gels was observed to decrease with the increase in the concentrations of TCEP. HEWL has four intramolecular disulphide bonds, and when it was preincubated with TCEP in fourfold excess, it unfolded completely between 55 C and 90 C (as discussed later), indicating that all the disulphide bonds were broken during the heating of the samples. It has also been reported that all four disulphide bonds of HEWL can be reduced by TCEP under partial/full unfolding conditions. 22 The turbidity of the gel at the lower TCEP/HEWL ratios may be due to disulphide bond scrambling, which is known to promote amorphous aggregation. 23 From these results, it appears that the precursor state of the gel may be formed from the fully reduced, unfolded state of HEWL. The effect of protein concentration on hydrogel formation was also studied by cooling the preheated samples (90 C), containing TCEP/HEWL in a molar ratio of 4 : 1, to 25 C and using the tube inversion test. It was observed that at least 400 mM HEWL was needed to form a stiff gel. Below 400 mM of HEWL, the stiffness of the hydrogels was found to decrease with the decrease in protein concentrations. Hydrogel formation was not observed at protein concentrations below 300 mM even aer extended incubation at 25 C. We also observed that the gel formed under the above conditions had excellent self-healing properties. The hydrogel was found to reform within few seconds aer vortexing the tube for several seconds (https:// drive.google.com/le/d/ 1Ct6eU602Ylxk9qOGQHJlAgzaY_V6G344/view?usp¼sharing).

Characterization of the hydrogel constituting material (HCM)
To explore the nature of the cross-linked material constituting the HEWL hydrogel, the morphology of the HCM was determined by transmission electron microscopy (TEM). Fig. 2A shows the TEM image of the negatively stained hydrogel diluted ve-fold in double distilled water. The analysis showed the presence of well-dened brillar structures intertwined with each other. The TEM image of a further diluted (tenfold) hydrogel sample showed that the HCM comprises short straight brillar structures of around $10-20 nm in width and $200-400 nm in length (Section 3.4, Fig. 4D). The TEM image of these diluted samples shows the formation of several thinner brils or protolaments associated to each other laterally.
The secondary structures of the native protein and the hydrogel were studied using far-UV CD and FTIR spectroscopy. Fig. 2B shows the far-UV CD spectra of the native state and HCM. The spectrum of native HEWL displays strong negative bands in the range from 200 to 230 nm, and its negative ellipticity at 208 nm is greater than that at 222 nm (Fig. 2C), which is characteristic of an a + b protein. 24 The far-UV CD spectrum of the HCM indicates the presence of mainly an extended b-sheet conformation, as revealed by the single negative band at 215-218 nm. 24,25 The far-UV CD data were also analysed by calculating Kuhn's g-value, which is the ratio of a sample's CD and absorbance. 26,27 The g-value is an intensive property independent of the path length and concentration of the protein. The changes in g-values of spectra provide information about the change in the secondary structure of the protein. The absorbance and CD values were measured, as described in previous reports. 26,27 Fig. 2C and the inset of Fig. 2C show the absorbance and g-value spectra of the native form and HCM of lysozyme. The g-value spectra of N and HCM are similar to those of a + b and all-b proteins. 26 This also indicates that the HCM is mainly in beta sheet conformation. Far-UV CD seems to be less reliable for the determination of the secondary structure contents of aggregated proteins. Therefore, secondary structure contents of the native state and the HCM were investigated by FTIR spectroscopy. Fig. 2D and E show the FTIR spectra of the native protein and the HCM in the amide 1 region.
Amide I peak deconvolution of native protein using Fourier self-deconvolution (FSD) method and Lorentzian curve tting procedure using the opus-65 soware shows a secondary structure composition of 46% a-helices, 20% b-sheets and 34% turns and random structures. These values are in agreement with X-ray crystallography data. 28 However, the hydrogel has 53% b-sheets, 25% b-turns and 22% random structures. The peak at 1654 cm À1 corresponding to the a-helices of the native protein is not observed in the hydrogel spectrum. Thus, the material of hydrogel has dominant b-sheet structures and contains bands at 1616 and 1626 cm À1 , which are considered diagnostic for the b-sheet formation associated with amyloid brils formation. 29 Fig . 2F shows the uorescence spectra of ThT-dyed native protein and HCM. The ThT uorescence spectrum of the gel exhibits the characteristic peak of an amyloid bril-ThT complex at 485 nm, whereas the native state does not show any signicant ThT uorescence. We also examined the samples with the Congo red (CR) binding assay. The CR difference spectrum of hydrogel reveals a maximum at 540 nm (inset of Fig. 2F). Similar CR difference spectra have been observed for amyloid-like brils for many proteins. 17,19,30,31 ThT and CR interact with the cross-b motif and give information about the secondary structure of the samples. 17,19,31,32 The ThT uorescence and CR difference spectra analyses showed a marked ability of the hydrogel to interact with these dyes and cause the spectral changes associated with amyloid brils. The high content of b-sheet structures observed in the HCM is in agreement with the ability of the same samples to bind amyloiddiagnostic dyes, which suggests that such b-sheet structures are mainly intermolecular.  (Fig. 3B). The ball drop time (BDT), the time of travel of the ball between two points through the samples, and RLS values provide information about the rheological (ow) properties and aggregation of the protein, respectively. 17,18,33 The transitions monitored by BDT and RLS were found not to be superimposable, when normalised to the gel fraction (f gel ) and the aggregate fraction (f scat ) (Fig. 3C). This indicates that the formation of the aggregate and its cross-linking to form the gel are independent reactions. The aggregation reaction precedes the gelation reaction. We calculated and plotted the difference of f scat and f gel , which gave a fraction of aggregate in solution phase (fs agg ) population curve against temperature. This population curve gives information about the formation of fs agg and its subsequent incorporation into the hydrogel during the cooling process. As can be seen from Fig. 3D, Fs agg increased below 70 C and reached a maximum value at 57.3 C. Below 57 C, fs agg decreased and approached zero at $45 C. The temperature (57.3 C) at which fs agg showed its maximum value was calculated by tting the data in Fig. 3D using a Gaussian function. Moreover, it was also observed that the RLS values of the sample at 80-90 C were comparable to those for the native protein at identical concentrations and at the same temperature, which is only slightly higher than the buffer solution at pH 7.4. These results indicate that HEWL, in solutions with a TCEP/HEWL molar ratio of 4 : 1, exist essentially in a monomeric unfolded state (U) between 90 C and 75 C, a bril state (F) around 57 C and a gel state (H) between 45 C and 25 C. Moreover, the decrease of Fs agg below 57 C suggests that the hydrogel forms at the expense of Fs agg . We also examined the reversibility of U to F and F to H transitions by re-heating the H at 60 C and 90 C ( Fig. 3A and B, open red circles). We found that H completely converted into the solution phase at both temperatures. Instead, this reversibility monitored by RLS showed that the sample retained almost all F form at 60 C and only partially disaggregated at 90 C. This data suggests that the F-H transition is reversible, whereas the U-F transition is not reversible. On the basis of these results, it appears that the formation of the H state from the U state is a two-steps and three-states reaction, and the mechanism can be represented as follows (eqn (5)).

Temperature dependence of sol-gel transitions
In Fig. 3D, the region between temperature 90 C and 55 C showed the HEWL aggregation prole, whereas data between 55 C and 25 C corresponded to the F-H transition. Moreover, the data obtained by BDT also represented the F-H transition. All the three proles satisfactorily t (R 2 ¼ 0.994-0.997) by sigmoidal function t (eqn (3)) and provide information about the midpoints of U-F and F-H transitions. The midpoints of U-F and F-H transitions were thus calculated to be 63.3 AE 4.5 C and 51.5 AE 3.8 C respectively, from population curves data (Fig. 3D). The midpoint of the F-H transition as calculated from BDT data was found to be 51.6 AE 3.5 C, which is almost identical to the midpoint of the same transition calculated from the population diagram.

Effect of protein concentration on the U-F transition
Since HEWL did not form a hydrogel at protein concentrations below 300 mM, we studied the effect of protein concentration on U-F transitions using RLS at 400 nm. To see how the U-F transition is affected under gelation conditions, we also examined the same aggregation process at protein concentrations of 400 and 700 mM. Fig. 4 shows the effect of protein concentration on the aggregation of HEWL during cooling of the protein sample from 90 C to 25 C in the presence of a TCEP/HEWL molar ratio of  (3)) gave parameters such as the amplitude (A) and the midpoint temperature (T m ¼ T agg ) of the aggregation process. T agg and A give information about the temperature at which 50% aggregation is completed and the amount of aggregates at the end of the reaction, respectively. Both the parameters, T agg and A, were found to increase linearly up to 250 mM protein concentration, whereas they markedly deviated downward from linearity at the protein concentrations at which gelation occurs i.e. 400 and 700 mM of HEWL. The extrapolated line t to the data up to 250 mM concentration intercepted the x-axis at 10.3 mM of HEWL. This indicates that during the F-H transition, the determination of A or T agg may be underestimated. Moreover, the occurrence of zero amplitude at concentrations of protein #10.3 mM suggests that the protein did not aggregate below 10 mM. Fig. 4C shows the TEM images of aggregates formed on cooling the preheated sample from 90 C to 25 C at 250 mM HEWL. It can be noted that the diluted hydrogel (Fig. 4D) and the aggregates formed at non-gelation concentrations of the protein have very similar morphology. Both contain short brils of width $20 nm and length $200-400 nm. These results suggest that brils formed under either gelation or aggregation concentrations of the protein are similar in morphology. It is very likely that brils formed under aggregation conditions are on-pathway intermediates of H formation.

Unfolded HEWL converted into an intermediate state upon cooling
Since HEWL did not aggregate below 10 mM protein concentrations (Section 3.4) neither during heating or cooling, temperature induced unfolding and refolding of HEWL (4 mM) in the presence of a TCEP/protein molar ratio of 4 : 1 at pH 7.4 was followed using various spectroscopic techniques. Mean residue ellipticity at 222 nm ([q] 222 nm ) was used to probe the nature of the secondary structure (Fig. 5A). 34 The changes in tertiary conformation were monitored by uorescence intensities at 360 nm and 330 nm (FI ratio 360/330) (Fig. 5B). The FI ratio 360/330 gives information about the extent of solvation of the protein core and is used as a probe of the tertiary structure. 17,35 Maximum emission wavelength (l max ) of ANS uorescence was used to monitor the changes in exposed hydrophobic clusters. 36 All probes show that HEWL in the presence of a TCEP/HEWL molar ratio of 4 : 1 unfolds in the 35-50 C temperature range. All the transitions t satisfactorily (R 2 ¼ 0.992-0.997) with the two-state transition model. The midpoints of thermal transition (T m ) were calculated to be 38.8 AE 1.6, 39.6 AE 2.2, and 39.1 AE 2.3 for [q] 222 nm , FI ratio 360/330 and l max of ANS uorescence data, respectively. It was also found that the ANS uorescence intensity remained low and almost constant (6-9 a.u.) in the transition region. It has been reported that ANS emission intensity increases and l max shis to shorter wavelengths when

ANS interacts with certain intermediate states of a protein. 36,37
These observations suggest that the observed transition has occurred between two conformational states viz. N* and U, which exist between 25-30 C and 60-90 C, respectively.
In order to understand the initial conformation of the material composing the hydrogel, the effect of cooling was investigated by using all the three probing techniques mentioned above. It was found that the protein retains its U state between 90-60 C on cooling. However, the negative [q] 222 nm increases from $1900 deg cm 2 dmol À1 at 90 C to $4000 deg cm 2 dmol À1 at 25 C. FI ratio 360/330 decreased from $1.6 to $1.4. The l max of the ANS emission shied from 523 nm to 478 nm. An increase close to 95-fold in the ANS intensity was observed when the sample was cooled to 25 C from 90 C. These results suggest that U converts into a partially folded intermediate (I) on cooling.

The partially folded intermediate resembles a premolten globule state
The biophysical properties of N*, U and I states were examined by various probing techniques for secondary structure, tertiary structure and exposure of hydrophobic clusters at their surface. Fig. 6A shows the far-UV CD spectra of these states along with the spectrum of native HEWL at pH 7.4 (N). The far-UV CD spectrum of the N state is typical of a folded a + b protein with the negative ellipticity at 208 nm greater than that at 222 nm. 24 However, the decrease in negative ellipticity values for the N* state suggested that a signicant amount of the secondary structure was disrupted during the N-N* transition. The spectrum of the HEWL solution with a TCEP/HEWL molar ratio of 4 : 1 at 90 C (U) (Fig. 6A, red curve) is almost identical to the completely unfolded state of HEWL induced by 6 M guanidine hydrochloride and 3 mM TCEP (U G , spectrum not shown). This indicates that U is a random coil state without any signicant residual secondary structure. 24 The spectrum of the I state of HEWL, which accumulated by cooling of the U state to 25 C, showed signicantly negative ellipticity values between 208 and 222 nm compared to those of the U state. This indicates that the I state contains a signicant amount of secondary structures. Fig. 6B and its inset show the absorbance and g-value spectra of N, N*, I and U states of the lysozyme. The g-value spectra also showed signicant changes in secondary structure between different states of the protein. The higher negative g-value of the I state as compared to that of the U state again suggested the presence of some residual secondary structures. The amount of regular secondary structures viz. a-helices and b-strands for N, N* and I states were calculated by analysis of the CD spectra using the K2D3 deconvolution soware. 18 The secondary structural contents (a-helices and b-strands) of these states are given in Table 1. These analyses indicate that the I state, the precursor state of the brillar material of the hydrogel, contains around 8.6% a-helices and 26.7% b-strands. Since the occurrence of these beta structures in the I state is signicantly higher ($111%) than in the N state (12.6%), it is highly probable that the secondary structures of the I state are non-native in origin. Fig. 6C shows the intrinsic uorescence spectra of N, N*, U and I states of HEWL. As presented in Table 1, N* and U states of HEWL showed a red shi of $3.6 nm and $9 nm and a decrease of $9% and $69% in uorescence intensity at 340 nm, respectively, as compared to those of the N state. However, U G and I states showed a red shi of $10 nm with $55% decrease and 68% increase in the uorescence intensity, respectively, as compared to that of N state (Table 1). These results indicate that during the N-N* transition, Trp residues of the protein were only partially exposed to the solvent. In U and I states, Trp residues were found to be fully exposed to the solvent suggesting the absence of intact tertiary structures in both U and I states. However, differences in intensity compared to N indicate that the unfolded conformations of U and I are different in nature.
The changes in ANS uorescence spectral features such as l max and emission intensity upon binding with a protein molecule provide information about the protein's conformational states. [36][37][38] ANS preferentially interacts with partially folded intermediate states, causes a blue shi in l max and a marked increase in emission intensity compared to the relative aqueous solution. [36][37][38] ANS is a hydrophobic dye and strongly binds to hydrophobic clusters, which are exposed to the solvent in the PFI states. However, ANS does not bind, or binds weakly with, the native or unfolded states of a protein, because the hydrophobic clusters are generally buried in the native state and completely disrupted in the unfolded state. 39 We found that only the I state of HEWL produced spectral changes in ANS uorescence, meaning it corresponds to an intermediate state (Fig. 6D and Table 1). The enhanced ANS uorescence and the blue-shi from 526 to 478 nm of the I state compared to the U state suggest that HEWL had also formed solvent-exposed hydrophobic clusters during the cooling process.

Discussion
In this work we have attempted to understand the rationale of amyloid-based hydrogel formation by investigating the temperature-and protein concentration-induced state transitions for HEWL. As depicted in Fig. 7, the native state (N) of HEWL at all tested concentrations appears to convert into an unfolded state at 90 C in the presence of TCEP/HEWL molar ratio of 4 : 1. At low protein concentrations (<10 mM), HEWL was found to transform into the U state through a two-steps mechanism as shown by the phase diagram in Fig. 5D. The RLS values of the U state, even at higher concentrations, were found to be comparable to those corresponding to identical concentrations of the N state. This suggested that the U state, which exists between 90 C and 75 C at all the tested concentrations of HEWL, is essentially a non-aggregate. Interestingly, the U state was found to form the partially folded intermediate state (I) at low concentrations (<10 mM), amyloid-like short brils (F) at medium concentrations (10mM < HEWL > 300 mM) and a selfhealing hydrogel (H) at high (>300 mM) concentrations of HEWL on cooling to 25 C. However, the hydrogel formation occurred through an aggregate intermediate state, which occurred maximally around 55-60 C.
The I state is characterized by residual secondary structures, the absence of tertiary structures and the presence of signicant amounts of solvent-exposed surface hydrophobic sites. All these three properties of I are essential characteristics of a pre-molten globule state (PMG). 19,[39][40][41][42][43] Due to the ease of structural  Therefore, it appears that the I state is the aggregation precursor state for HEWL under the current study conditions. The aggregate formed under identical conditions, but at medium protein concentrations, possessed all the characteristic properties of amyloid brils. 1,17,19,44,45 However, the amyloid brils observed here are shorter in length than the typical brils of lysozymes. 1,17,19,45 These brils appear to contain several thinner protolaments associated with each other laterally. Fibrils formed by lateral association of proto-laments are very common in the lysozyme family. 17,19,45 Interestingly, hydrogel constituting materials formed at high protein concentrations are very similar in morphology to the amyloid-like brils formed at medium protein concentrations under identical conditions. From a close look at the TEM images, secondary structures and tinctorial properties, it appears that the F state rearranges into the HCM without much modications. Moreover, the reversible nature of the hydrogel with temperature suggests that cross-linking is mainly physical in nature. The extremely high self-healing property of the hydrogel may be due short length and lateral association of several constituting protolaments. Short laments may cross-link and delink easily without signicant breakage. However, larger brils are more prone to breakage. The detailed investigation of the mechanism of self-healing property of this hydrogel is out of the scope of this work.
In order to understand the connection between I, F and H states of HEWL, we established a correlation between the conformational properties of I, such as secondary structure and surface-exposed hydrophobicity, with F formation and H formation measured by RLS and BDT. Fig. 8 shows threedimensional graphs obtained by plotting the conformational properties of the I state viz. mean residue ellipticity at 222 nm (x-axis), ANS l max (y-axis) and RLS (z-axis) for F formation, H formation and BDT (z-axis) for H formation at z axis. The data points were tted with a plane using eqn (6).
where a and b are constants and determine the slope and y0 is an x-y intercept. We found that the data best t in case of F formation with R 2 > 0.99 and fairly t for H formation measured by RLS. The correlations are described by the following equations.
RLS ¼ À0.02[q 222 nm ] À 0.04(l max ) À 15.5 RLS ¼ À0.005[q 222 nm ] À 3.5(l max ) + 1610 However, H formation measured by BDT does not t (R 2 ¼ 0.667) with the plane equation. These correlations further support our conclusion that hydrogel formation by HEWL involves the I state as a precursor of the HCM, which physically cross-links and forms the hydrogel. Furthermore, the I state controls the formation of the HCM, whereas cross-linking of HCM into H is an independent process.   8 Correlation of properties for the formation of the aggregation precursor state, amyloid fibril and hydrogel. 3D graph showing mean residue ellipticity at 222 nm (x-axis), ANS l max (y-axis) and RLS values (z-axis) for fibril formation (A), for hydrogel formation (B) and BDT (z-axis) for hydrogel formation (C). Colored symbols represents data points at different temperatures. The grid represents the plane of best fit to all the data points and corresponds to eqn (6).

Conclusion
We have demonstrated that the individual reactions of a partially folded intermediate state, amyloid-like brils and an aggregated state during hydrogel formation by HEWL under identical solution conditions, but different protein concentrations, can gave rise to the global molecular mechanism of hydrogel fabrication. This conclusion cannot rule out some effects of protein concentration on the individual reactions. But given the difficulty in identication and characterization of the intermediates in the pathway of hydrogel fabrication due to high protein concentrations, this study provides a remarkable way to understand the mechanism of hydrogel fabrication. The difference in the structures of partially folded intermediates of a protein is known to generate polymorphism of amyloid brils. 46,47 Therefore, the understanding of "on pathway" intermediates of hydrogel formation can be exploited to engineer the properties of hydrogel for various biomedical applications.

Conflicts of interest
There are no conicts to declare.