2.1.

Cartilage Specimen Preparation

Cartilage plugs with full thickness ($\sim 1$ to 2 mm) were excised from two fresh mature bovine knee joints using a scalpel and a 4-mm diameter biopsy punch. The articular cartilage plugs ($N=6$) were then stored frozen at $-20°\mathrm{C}$ until use.

To tease out PG and collagen type II contributions to the OH band, Raman spectra of chondroitin sulfate (CS) powder (Sigma-Aldrich, St. Louis, Missouri) and collagen type II from calf joints (Elastin Products Company, Owensville, Missouri) were recorded from both intact and gradually hydrated specimens. Briefly, these specimens were placed on an aluminum slide and Raman spectra were collected at room temperature. Then, these specimens were gradually hydrated until they reached fully hydrated state (no water absorbed anymore by the specimens). Raman spectra were simultaneously collected during the progressive hydration.

2.2.

Serial Dehydration and Gravimetric Measurements

The full depth plugs of cartilage were allowed to equilibrate at room temperature in phosphate buffered saline. Superficial water on specimen surface was gently blotted and wet weight of specimens was measured (Model XS64, Mettler Toledo, Columbus, Ohio). The specimens were then left at room temperature in air for 60 min and weighed again. Following air drying, the specimens were heated in an incubator in air at a rate of $\sim 15°\mathrm{C}$ per minute, held at 37°C for 48 h, and weighed again. The treatment duration was chosen because specimen weights did not change for longer durations. The temperature was selected to stay below the denaturation temperature of organic matrix macromolecules (i.e., collagen and PG) around 50°C to 70°C.^30,31 The oven drying conditions are, therefore, expected to remove predominantly mobile water and have minimal effect on tightly bound water compartment, which requires much higher temperature to remove from the matrix.^30,31

The specimens were kept in a sealed plastic bag except when gravimetric and Raman analyses were carried out to avoid rehydration by the moisture in air. The percent water loss was calculated as

2.3.

Hydrogen–Deuterium (H/D) Exchange

The long-term deuterium exchange beyond oven drying provides information on the rate of dissociation of different moieties of bound water through observation of the disappearance of OH-related Raman peaks.²⁷ Thus, a group of cartilage specimens dried in an oven were immersed into 1 ml of deuterium oxide (${\mathrm{D}}_2\mathrm{O}$, 99.9% isotopic purity, Sigma-Aldrich., St. Louis, Missouri) for 60 min, 24 h, and 7 days,²⁷ with periodic replacement of ${\mathrm{D}}_2\mathrm{O}$ at room temperature. Raman spectra were collected immediately after each step of deuterium treatment.

2.4.

Shortwave-Infrared Raman Spectroscopy and Data Processing

Quantum efficiency of the CCDs of standard 785-nm Raman systems decline in the higher wavenumber OH-stretch range; therefore, a refined analysis of the water compartments in biological tissues is challenging with standard RS when using with near-infrared laser wavelength. We recently customized a SWIR-RS system^25,27 to survey the OH-stretch range of biological tissues while accommodating the protein-related background fluorescence. Our SWIR-RS has been previously described in detail.^25–27 Briefly, this SWIR-RS system involved excitation at 847 nm (Axcel Photonics) and signal collection using a shortwave InGaAs IR spectrometer (BaySpec, Inc., California) specifically optimized for 1000- to 1400-nm wavenumber range that provided data collection in a spectral range of $\sim 2550$ to $4770{\mathrm{cm}}^{-1}$, which covered the CH and OH-stretch bands sufficiently.

Measurement locations at the articular surface were positioned equidistantly to cover the region of interest and Raman spectra were collected from the same location at a $100\text{-}\mu \mathrm{m}$ laser spot size. Signal collection time was set at 10 s with an accumulation of 10 scans. Laser power was set to $\sim 25\mathrm{mW}$ on the sample to avoid possible water loss and cartilage matrix damage associated with excessive laser heating.³² Representative spectra for any given treatment condition were obtained by taking the average of three spectra collected from each six cartilage specimens at room temperature. Raman spectra were analyzed using LabSpec 5 (Horiba Scientific, Edison, New Jersey) and OriginLab (OriginLab Corporation, Northampton, Massachusetts) softwares to determine the position and intensity of the peaks. The background was removed by subtracting the fitted fluorescence baseline from the raw spectra using piecewise linear segments. Then, the averaged spectra were smoothed to improve signal-to-noise ratio via denoising algorithm of LabSpec 5 software. The spectra were then fitted through second-derivative analysis to identify peak locations. All spectra were normalized to attain the same intensity for the CH-stretch band that is emerging from the organic matrix. Therefore, band intensity values reported in this study are normalized to total amount of organic matrix in the cartilage specimens.

2.5.

Statistical Analysis

All statistical analyses were performed using Minitab 16 statistical software (Minitab, Inc., State College, Pennsylvania). Data were presented as $\text{mean}\pm \text{standard deviation}$. Linear regression analyses were performed between Raman peak intensities versus gravimetrically measured water loss during progressive dehydration. Statistical significance was considered at a $p$-value $<0.05$ for the resulting $R^2$ value.

3.1.

Determination of Peak Locations in OH-Stretch Band of Wet Cartilage and Bulk Water

The presented spectra result from the average of 18 Raman spectra taken at different locations from six cartilage specimens. The standard deviation was within %13 of the intensity. All cartilage spectra in this study are scaled to obtain the same intensity at the CH-stretch peak centered at $\sim 2930{\mathrm{cm}}^{-1}$ [Fig. 1(a)]. The wavenumber range of 2700 to $3050{\mathrm{cm}}^{-1}$ manifests CH-stretch band associated with collagen and PG phases of cartilage matrix, whereas OH-stretch band extends across 3100 to $3800{\mathrm{cm}}^{-1}$.

Fig. 1

Raman OH stretch band of cartilage. Cartilage spectra are normalized with respect to the CH-stretch intensity that is representing the amount of protein. (a) Raman intensity changes of wet cartilage during sequential dehydration in the ambient for up to 60 min and in an oven for 48 h along with the Raman spectra of bulk water. The standard deviation is within 13% of the intensity and not shown in the spectra for the sake of clarity. (b) Second derivative spectra of wet cartilage and bulk water. Red dashed line indicates bulk water-related peaks and black dashed line indicated wet cartilage related peaks. (c) OH band of wet and oven-dried cartilage and their difference spectra, showing predominance of mobile like water. (d) $R^2$ for pairwise correlations between water loss measured by central three OH peaks and water loss by weight. All correlations are significant at $p<0.05$.

The comparison of Raman spectra of cartilage specimens with bulk water is shown in Fig. 1(a). OH-bands of fully hydrated cartilage spectrum and bulk water displayed prominent peaks at $\sim 3250$ and $\sim 3450{\mathrm{cm}}^{-1}$, with the former peak being more pronounced for bulk water [Fig. 1(a)]. A direct comparison of Raman spectrum of bulk water with wet and oven-dried cartilage specimens indicated the prominence of mobile water-related Raman intensity in wet cartilage [Fig. 1(a)]. Raman difference spectrum calculated by subtracting the spectrum of wet cartilage from oven-dried cartilage specimens confirmed that mobile water fraction accounted for about 95% of the Raman intensities of wet cartilage [Fig. 1(c)].

Using second-derivative method, locations of six peaks (three central peaks with their three shoulders) in OH-stretch bands of wet cartilage were revealed [Table 1 and Fig. 1(b)]. For wet cartilage, central peaks are located at $\sim 3200$, $\sim 3450$, and $\sim 3630{\mathrm{cm}}^{-1}$ and their shoulders at $\sim 3250$, $\sim 3520$, and $\sim 3650{\mathrm{cm}}^{-1}$, respectively. For bulk water, there were five peaks located at $\sim 3015$, $\sim 3220$, $\sim 3410$, $\sim 3565$, and $3630{\mathrm{cm}}^{-1}$, respectively [Table 1 and Fig. 1(b)]. The changes in intensities of major cartilage peaks were significantly correlated with the gravimetric water loss obtained by subtracting wet weight from oven-dried weight [Fig. 1(d)], indicating these Raman peaks were sensitive to dehydration and were associated with water content in cartilage.

Table 1

Raman band assignments and origin of the subpeaks. DAA-OH, single donor–double acceptor; DDAA-OH, double donor–double acceptor; DA-OH, single donor–single acceptor; DDA-OH, double donor–single acceptor; PG, proteoglycan; collagen: collagen type II; water, bound and mobile water.

3.2.

Changes in OH-Stretch Band of Cartilage Under Sequential Dehydration

There was a stepwise decline in the Raman intensities of cartilage’s OH-stretch band upon sequential drying [Fig. 2(a)]. We observed the peak intensities to stabilize after 48 h of oven drying, and the changes in peak intensities were significantly correlated with the gravimetric water loss [Fig. 1(d)].

Fig. 2

Raman OH stretching bands of deuterium oxide-treated specimens. (a) OH intensity change in dehydrated cartilage due to progressive deuterium oxide treatment. The standard deviation is within 13% of the intensity and not shown in the spectra for the sake of clarity. (b) Second derivative spectra of oven-dried and deuterated cartilage. Pink dashed lines indicate OH peaks associated with macromolecules of collagen and PG. Blue dashed lines indicate OH peaks associated with mainly bound water molecules.

The comparison of second-derivative Raman spectra of wet cartilage with oven-dried cartilage specimens showed that the observed locations of peaks and shoulders in wet cartilage shifted to lower or higher frequencies [Figs. 1(b) and 2(b)]. Following oven drying, a peak centered at $\sim 3330{\mathrm{cm}}^{-1}$ was revealed, which is likely to be associated with NH-stretching band along with tightly bound water since it is also sensitive to deuterium oxide treatment [Fig. 2(a)].

Spectral changes associated with bound water were revealed by Raman spectra of oven-dried cartilage specimens under H/D exchange treatment [Fig. 2(a)]. Comparison of deutrated spectra with spectra from oven-dried samples indicated Raman signal after oven drying was associated mainly with bound water compartments, which accounted for 3 to 4% of the total Raman intensities along with the contributions of organic matrix [Fig. 2(a) and Table 2]. We observed that Raman intensities stabilized upon 7 days long H/D exchange treatment, indicating there is no water fractions left in the matrix.²⁷ The second-derivative analysis showed the locations of six peaks in OH-stretch bands of deuterated cartilage specimens (Table 1) at $\sim 3195$, $\sim 3235$, $\sim 3345$, $\sim 3450$, $\sim 3500$, and $\sim 3660{\mathrm{cm}}^{-1}$, respectively [Fig. 2(b) and Table 1], and should be associated with the organic matrix of cartilage that is left after deuteration of water molecules.

3.3.

Raman High-Wavenumber Spectra of Pure Collagen Type II and Chondroitin Sulfate

Raman spectra of CS powder and pure collagen type II are shown in Fig. 3(a). Using second-derivative analysis, the locations of seven peaks were revealed from the spectra of collagen and CS powders. These peaks were located at $\sim 3195$, $\sim 3250$, $\sim 3345$, $\sim 3450$, $\sim 3500$, $\sim 3640$, and $\sim 3660{\mathrm{cm}}^{-1}$, respectively [Fig. 3(b)]. Among these peaks, both collagen and CS spectrum shared the same peaks located at $\sim 3195$, $\sim 3345$, and $\sim 3660{\mathrm{cm}}^{-1}$. The peaks at $\sim 3250$, $\sim 3450$, and $\sim 3640{\mathrm{cm}}^{-1}$ were exclusively associated with collagen, whereas the only peak associated exclusively with CS was located at $\sim 3500{\mathrm{cm}}^{-1}$ [Fig. 3(b)]. Importantly, the peaks identified for pure collagen II and CS were in close agreement with those identified for the organic matrix of native cartilage [Fig. 2(b)].

Fig. 3

Raman OH stretching bands of pure collagen type II and CS. (a) Comparison of Raman spectrum of collagen type II and CS. (b) Second derivative spectra of collagen type II and CS. Red dashed lines indicate exclusively collagen-related OH peaks while black dashed lines indicate exclusively PGs-related OH peaks. Blue dashed lines indicate the peaks are associated with both collagen and PGs. (c and d) OH intensity changes of collagen and CS, respectively, after just moistened the samples, showing the peak locations of bound water molecules. (e and f) OH intensity changes of collagen and CS, respectively, during progressive hydration. The standard deviation is within 13% of the intensity and not shown in the spectra for the sake of clarity. (g and h) 3450/3250 Ratio changes of collagen and CS, respectively during progressive hydration.

Gradual hydration of collagen and CS powders from moistened to solvated and then fully hydrated conditions, respectfully, allowed observation of the emergence of bound and mobile water fractions. Our results suggested that collagen-bound water-related peaks were located at $\sim 3450$ and $\sim 3640$, whereas PG-bound water-related peaks were located at $\sim 3500$ and $\sim 3660{\mathrm{cm}}^{-1}$, respectively [Figs. 3(c) and 3(d)]. Furthermore, the peaks at $\sim 3195$ and $\sim 3345{\mathrm{cm}}^{-1}$ emerged from both collagen- and PG-bound water [Figs. 3(c) and 3(d)]. Continued rehydration of collagen and CS reached equilibrium [Figs. 3(e) and 3(f)]. We calculated $3450/3250{\mathrm{cm}}^{-1}$ ratio that was recently reported as an indicator of water bonding states in skin³³ [Figs. 3(g) and 3(h)]. This value decreased for moistened collagen type II and CS standards compared with baseline but then increased with further rehydration [Figs. 3(g) and 3(h)], indicating a possible translation for water bonding states from bound to mobile during gradual hydration.³³

4.1.

Bulk Water Displays Five Subpeaks Located Between 3000 and 3800 cm⁻¹ in the OH-Stretch Region of Cartilage Spectrum

In hydrogen bonding, the proton (donor) interacts with an acceptor that is usually an electronegative atom such as oxygen or nitrogen. Therefore, an individual water molecule can interact with neighboring molecules establishing as few as no hydrogen bonds (dangling) to multiple donor and acceptor bonds^39,40,42,45 [Fig. 5(c)]. In accordance with previous studies on Raman analysis of liquid water,^39,40,42,45 the peak at $\sim 3015{\mathrm{cm}}^{-1}$ is associated with single donor–double acceptor (DAA-OH) hydrogen bonds, $\sim 3220{\mathrm{cm}}^{-1}$ is associated with double donor–double acceptor (DDAA-OH) hydrogen bonds, $\sim 3410$ is associated with single donor–single acceptor (DA-OH) hydrogen bonds, $\sim 3565$ is associated with double donor–single acceptor (DDA-OH) hydrogen bonds,^39,40,42 and $\sim 3630{\mathrm{cm}}^{-1}$ is associated with free-OH [nonhydrogen-bonded (dangling) OH group]⁴⁵ [see Figs. 4(a) and 4(b) for schematically the hydrogen bonding types of water molecules].

Fig. 4

A schematic depiction of the hydrogen bonding types of water molecules. (a) The Raman OH stretching band of bulk water includes five subpeaks centered at $3015{\mathrm{cm}}^{-1}$ (DAA-OH, single donor–double acceptor), $3220{\mathrm{cm}}^{-1}$ (DDAA-OH, double donor–double acceptor), $3410{\mathrm{cm}}^{-1}$ (DA-OH, single donor–single acceptor), $3565{\mathrm{cm}}^{-1}$ (DDA-OH, double donor–single acceptor), and $3630{\mathrm{cm}}^{-1}$ (free-OH). (b) The Raman OH stretching band of wet cartilage includes six subpeaks centered at $3200{\mathrm{cm}}^{-1}$ (DDAA-OH, double donor–double acceptor), $3250{\mathrm{cm}}^{-1}$ (DDAA-OH, double donor–double acceptor), $3450{\mathrm{cm}}^{-1}$ (DA-OH, single donor–single acceptor), $3500{\mathrm{cm}}^{-1}$ (DDA-OH, double donor–single acceptor), $3630{\mathrm{cm}}^{-1}$, and $3650{\mathrm{cm}}^{-1}$ (free-OH). (c) The percent of water types in cartilage normalized by total water content.

Fig. 5

A schematic depiction of cartilage structure and composition, and hydration layers surrounding collagen fibrils and PGs in the water environment. (a) Illustration of cartilage, (a) structure, and (b) composition. (c) Illustration of the interaction of collagen fibrils and PGs with an individual water molecules establishing as few as no hydrogen bonds (dangling) to multiple donor and acceptor bonds (e.g., DDAA-OH, DA-OH, or DDA-OH). This interaction most likely creates the first hydration layer surrounding to collagen fibrils and PGs, which represents tightly bound water in cartilage. Then, other water molecules bind to these bound water molecules and to each other with the schema of different hydrogen bonds network [e.g., DDAA-OH, DA-OH, DDA-OH or Free-OH (dangling)], creating other water states around the collagen fibrils and PGs, and results an increase in the intensity of corresponding Raman peaks located between 3100 and $3800{\mathrm{cm}}^{-1}$.

4.2.

Majority of Water Type in Cartilage is Mobile Water

Our results revealed that wet cartilage spectrum is dominated by mobile water, accounting for $\sim 95\%$ of total Raman signals and the remaining Raman intensities are associated with bound water and solid matrix macromolecules [Fig. 1(c) and Table 2]. This finding is in agreement with the literature that bound water fraction of cartilage was found to be $\sim 4\%$ of total mass using differential scanning calorimetric (DSC) experiment.⁴⁶ Furthermore, most water fraction in cartilage is mobile water state and has little interaction with the matrix.⁴⁶ High content of mobile water in cartilage explains the general similarity between wet cartilage spectrum and bulk water spectrum. However, there is also difference between bulk water and wet cartilage spectra in that $3250{\mathrm{cm}}^{-1}$ band is less pronounced relative to $3450{\mathrm{cm}}^{-1}$ in cartilage spectrum. This difference may result from the fact that the donor–acceptor interactions are limited to water molecules only in bulk water, whereas additional donor–acceptor interactions occur between water-matrix phases. We further calculated the percent of DDAA-OH, DA-OH, DDA-OH, and free-OH water amount normalized by total water content. As can be seen, the most concentrated water type in the cartilage is DA-OH water molecules ($\sim 35\%$), followed by DDAA-OH and DDA-OH [Fig. 4(c)]. Free-OH water molecules only account for $\sim 10\%$ of total water in cartilage [Fig. 4(c)].

Table 2

The breakdown of band intensities in wet cartilage between mobile water, bound water, and cartilage matrix.

4.3.

Peaks Located at 3200 and 3250 cm⁻¹ in Wet Cartilage Spectrum are Associated with DDAA-OH Hydrogen Bonded Water Molecules that are Bound to Collagen and PG (3200 cm⁻¹), and Bound to Specifically Collagen (3250 cm⁻¹) at the State I Hydration Layer

Since it was not possible to distinguish between bound water and mobile water in wet cartilage spectrum, we employed a sequential dehydration/deutration as well as a rehydration process to reveal bound water and organic matrix phases. The air drying followed by oven drying at 37°C employed in this study is expected to remove mobile water and has a minimal effect on bound water fraction since removing bound water from the matrix requires temperature at 70°C without denaturation³² or temperatures well in excess of 70°C, which would denature the organic matrix as well.^47,48 After oven drying for 48 h, we observed the weight of cartilage stabilized. The long-term deuterium exchange beyond oven drying provides information on the rate of dissociation of different moieties of bound water through observation of the disappearance of subpeaks. Overall, our sequential water removal protocol allowed the interpretation of the locations of Raman bands associated with varying water fractions.

In the past, based on DSC experiments, researchers found that there are up to five discrete water fractions observed around collagen structure reportedly ranging from low (state I) to high (state V) concentration.^49–51 The first hydration shell (state I) corresponds to tightly bound water and state V corresponds to bulk water.^49–51 To interpret molecular origins of the subpeaks in cartilage, we followed the findings of these DSC studies.^49–51 As elucidated, peaks in the range of $\sim 3200$ to $\sim 3250{\mathrm{cm}}^{-1}$ in wet cartilage spectrum are assigned as DDAA-OH hydrogen-bonded water molecules^39,40,42,45 [Fig. 4(b)]. These water molecules have stronger hydrogen bonds and shorter bond lengths compared with DAA-OH, DDA-OH, or DA-OH. Our findings (Figs. 2 and 3) along with the help of the earlier calorimetric studies^49–51 suggest that water molecules at $\sim 3200{\mathrm{cm}}^{-1}$ in wet cartilage spectrum are likely to establish state level I bonds with collagen and PG as the first hydration shell (Fig. 5) since both pure collagen and PG provide a peak at this frequency [Fig. 3(b) and Table 1] and the peak at $\sim 3195{\mathrm{cm}}^{-1}$ is observed in deuterated cartilage spectrum [Fig. 2(b)]. Other water molecules may start to bind to state level I molecules and to each other creating other water state levels (e.g., levels II, III, IV, or V) around the organic matrix (Fig. 5), resulting in the sequential increase of the intensity at $\sim 3200$ and $\sim 3250{\mathrm{cm}}^{-1}$ upon hydration of type II collagen and CS [Figs. 3(e) and 3(f)]. It is likely that the band at $\sim 3250{\mathrm{cm}}^{-1}$ establishes another set of level I hydrogen bonding network with collagen as evidenced build-up of intensity of this band for collagen [Figs. 3(c) and 3(e)], but not for CS [Figs. 3(d) and 3(f)].

4.4.

Peak Located at 3450 cm⁻¹ in Wet Cartilage Spectrum is Associated with DA-OH Hydrogen-Bonded Water Molecules that are Bound to Collagen (Hydroxyproline) at the State I Hydration Layer

The peak at $3450{\mathrm{cm}}^{-1}$ is assigned as DA-OH hydrogen-bonded water molecules [Fig. 4(b) and Table 1], these water molecules have weaker hydrogen bonds and taller bond lengths compared with DDAA-OH, and thus they are moderately hydrogen bound water molecules. Our findings show that the origin of this peak is associated with collagen molecules [Fig. 3(b)]. More specifically, the origin of the peak at $3450{\mathrm{cm}}^{-1}$ is the hydroxyl group of hydroxyproline according to our previous molecular dynamic simulation of single amino acid chain.²⁷ Briefly, we calculated theoretical/computational peak locations between 3000 and $3800{\mathrm{cm}}^{-1}$ for proline, hydroxypoline, and Glycine–Proline–Hydrxyoproline (Gly–Pro–Hyp) single amino acid chain. The results showed that only hydroxproline displayed a peak at $\sim 3450{\mathrm{cm}}^{-1}$.²⁷ The intensity associated with hydroxyl group of hydroxyproline in the present study only accounts for $\sim 1.5\%$ of total intensity (Table 2), whereas the majority of intensity is associated with DA-OH water molecules (Table 2). Cumulatively, our results suggest that, at state level I, the hydrogen donor atoms of hydroxyproline bind to acceptor oxygen atoms with the schema of DA-OH hydrogen bonds network. Then, other water molecules bind to these bound water molecules and to each other with the schema of DA-OH hydrogen bonds network [Fig. 5(c)], creating other water states around the collagen and results in an increase in the intensity of this peak [Fig. 3(e)].

4.5.

Peak Located at 3520 cm⁻¹ in Wet Cartilage Spectrum is Associated with DDA-OH Hydrogen-Bonded Water Molecules that are Bound to PG at the State I Hydration Layer

The peak at $3520{\mathrm{cm}}^{-1}$ is assigned as DDA-OH hydrogen-bonded water molecules and our results indicate that the origin of this peak at state level I is associated with the PG phase since this peak is only found in the spectra of CS but not in collagen spectra (Fig. 3), as evidenced build-up of intensity of this band for CS [Figs. 3(d) and 3(f)] but not for collagen [Figs. 3(c) and 3(e)]. These water molecules probably have weaker hydrogen bonds and longer bond lengths than those of Raman bands at lower wavenumbers.

4.6.

Peaks Located at 3630 and ∼3650 cm⁻¹ in Wet Cartilage Spectrum are Mostly Associated with Nonhydrogen-Bonded Water Molecules

The subpeaks at $\sim 3630$ to $\sim 3650{\mathrm{cm}}^{-1}$ are assigned as free-OH [nonhydrogen bonded (dangling) OH] water molecules [Fig. 4(b) and Table 1]. These nonhydrogen-bonded water molecules are typically observed in bulk water when donor and acceptor atoms for creation of hydrogen bonds are relatively limited. Although the major intensity of these peaks are associated with nonhydrogen-bonded water molecules, accounting for $\sim 95\%$ of total intensity (Table 2), our results suggest that collagen molecules also contribute to band at $\sim 3630{\mathrm{cm}}^{-1}$ at state level I as suggested by the spectrum of pure collagen standard [Fig. 3(a)]. Spectrum of CS standard implies that CS does not contribute to $3630{\mathrm{cm}}^{-1}$ band [Fig. 3(b)]. Deuteration showed that organic phase contribution was limited to $\sim 3\%$ of total Raman intensities (Table 2). The origin of this dangling OH vibration at state level I may be associated with hydroxyl groups of amino acids, such as hydroxyproline, serine, or threonine. Although the peak at $3650{\mathrm{cm}}^{-1}$ is also assigned as free-OH water molecules [Fig. 4(b) and Table 1], accounting for $\sim 95\%$ of total intensity (Table 2), our results indicate that the origin of this peak at state level I has contributions from both collagen [Figs. 3(b) and 3(c)] and PG [Figs. 3(b) and 3(d)], accounting for $\sim 4\%$ of total intensity (Table 2) and also as reflected by deuterated cartilage spectrum (Fig. 2). The origin of this peak at state I level may be associated with hydroxyl group of certain amino acids and CS that hydrogen atoms most likely create dangling OH vibration rather than creating a hydrogen bond. However, further studies are needed to identify the moieties of collagen and PG to which these peaks are originating from and to which water molecules are attracted. This dangling OH vibration may also originate from DA-OH hydrogen bonds network, where one of the hydrogen atoms is free to create dangling OH vibration [Figs. 4 and 5(c)]. Though we did not observe a peak at $3345{\mathrm{cm}}^{-1}$ in wet cartilage spectrum based on second-derivative analysis [Fig. 1(b)], this peak clearly appears after oven drying [Fig. 2(b)], and is associated with the NH-stretching band of both collagen and PG; however, its intensity is also sensitive to dehydration and deuteration [Fig. 2(a)] indicating contributions of OH vibrations as well.

4.7.

3450/3250 cm⁻¹ Ratio can be Used to Study on Water Bonding States in Cartilage

Previously, $3450/3250{\mathrm{cm}}^{-1}$ ratio (DA-OH/DDAA-OH) was reported as an indicator of water bonding states in polymer⁵² and skin.³³ The lower value of this ratio indicates the increased number of hydrogen bonds between water molecules and biological tissue.³³ Thus, we calculated this ratio for pure collagen type II and CS standards during gradual rehydration [Figs. 3(g) and 3(h)]. Our results showed that this ratio decreased for moistened samples for both collagen type II and CS, compared with baseline [Figs. 3(g) and 3(h)], suggesting that water molecules start to bind to collagen and CS surfaces creating tightly bound water around them during this stage of hydration [Fig. 5(c)]. Continued rehydration of collagen and CS [Figs. 3(e) and 3(f)] caused an increase in this ratio [Figs. 3(g) and 3(h)], suggesting that water molecules start to bind each other, creating mobile water layers around collagen and CS [Fig. 5(c)]. Therefore, this ratio can be used as an indicator of water bonding states in cartilage.

This present study has limitations. First, we measured water content in cartilage after a single freeze-thaw cycle at $-20°\mathrm{C}$, which could affect the amount of water content because such freeze-thaw cycle was reported to potentially influence the overall material properties of cartilage.⁵³ Thus, it would be better to assess water content in cartilage from fresh samples or samples refrigerated at 4°C in future studies because such storage condition does not affect the overall material properties of cartilage,⁵³ thereby, expecting to not change the amount of water content in cartilage as well. Second, in this study, Raman data were collected from the surface of articular cartilage. Therefore, it is not represented to depth-dependent water distribution in cartilage. Future studies are needed to qualify depth-dependent water distribution in cartilage.