Unraveling Water-Mediated 31P Relaxation in Bone Mineral.

Bone is a dynamic tissue composed of organic proteins (mainly type I collagen), inorganic components (hydroxyapatite), lipids, and water that undergoes a continuous rebuilding process over the lifespan of human beings. Bone mineral is mainly composed of a crystalline apatitic core surrounded by an amorphous surface layer. The supramolecular arrangement of different constituents gives rise to its unique mechanical properties, which become altered in various bone-related disease conditions. Many of the interactions among the different components are poorly understood. Recently, solid-state nuclear magnetic resonance (ssNMR) has become a popular spectroscopic tool for studying bone. In this article, we present a study probing the interaction of water molecules with amorphous and crystalline parts of the bone mineral through 31P ssNMR relaxation parameters (T 1 and T 2) and dynamics (correlation time). The method was developed to selectively measure the 31P NMR relaxation parameters and dynamics of the crystalline apatitic core and the amorphous surface layer of the bone mineral. The measured 31P correlation times (in the range of 10-6-10-7 s) indicated the different dynamic behaviors of both the mineral components. Additionally, we observed that dehydration affected the apatitic core region more significantly, while H-D exchange showed changes in the amorphous surface layer to a greater extent. Overall, the present work provides a significant understanding of the relaxation and dynamics of bone mineral components inside the bone matrix.

Introduction

Bone is a living composite mineralized biological tissue that is composed of organic mass such as proteins, lipids, and polysaccharides; water; and an inorganic component or mineral.[1,2] It supports the body structurally, protects internal organs, and provides movement and stability.[3] The bone components, i.e., type I collagen, noncollagenous proteins (NCPs), water, and lipid molecules, play an essential role in the processes of normal and pathological bone resorption and formation, collagen fibrillogenesis, and mineralization.[4] The bone mineral (i.e., the crystalline apatite core and the amorphous hydrated surface layer) contains elongated platelet-like carbonated calcium phosphate[5,6] that is constantly deposited within a proteinaceous organic matrix and affects the collagen fibril architecture in various orthopedic diseases such as osteoporosis, ontogenesis imperfecta, hypophosphatasia, and bone cancer.

The mechanical properties, i.e., the stiffness, strength, and toughness of bone, are mainly associated with the degree of hydration[7,8]. Previously, it was suggested that mineralized bone tissues contain abundant water components (∼20% of the volume) and that water has a considerable effect on the biological and mechanical properties of the bone.[8−10] Water plays a major role in bone mineralization and is present in both mobile and bound forms in the mineral lattice[8] with varying degrees of motional restriction.[11] In addition, the bound water content is known to provide plasticity to the bone,[12] varying significantly between individuals and changing dramatically during aging.[12,13] Recently, there has been a growing interest in studying the effects of hydration and dehydration on the bone. Both dehydration and chemical exchange (H–D) greatly reduce bone’s mechanical properties by changing its interactions inside the collagen matrix.[7] Solid-state NMR spectroscopy (ssNMR) is one of the few techniques for studying the structure of bone and is used to probe minerals, mineralized collagen, and water.[14−21] More specifically, 31P NMR provides a unique opportunity to study the mineral components in bones over the organic matrix without any intervention or structural perturbation. Previously, it has been suggested that the water layer is essential for the ordering and orientation of mineral platelets in bone.[15] Moreover, water and its role in bone have been extensively studied by NMR spectroscopy and the relaxation methodology.[9,14,22] Recent advancements in ssNMR, including BioSolids CryoProbe and high-field dynamic nuclear polarization (DNP) based solid-state NMR instrumentation and methodologies, have solved the problem of sensitivity enhancement in bone-like complex materials to a great extent.[23−26] However, exact knowledge about the water-mediated changes inside the bone mineral is lacking, and further work is needed to characterize the role of water in bone mineralization. Additionally, the development of new solid-state NMR methodologies has become a major necessity to understand the relaxation mechanism and dynamics of bone minerals.

In the current study, we have applied an advanced ssNMR technique to study the relaxation and dynamics of water-mediated structural changes in bone minerals. Herein, for the first time, we developed a ssNMR methodology to investigate the relaxation behavior and dynamics of mineral components associated with the crystalline apatitic core (unprotonated phosphate, OH–/PO43–) and the amorphous surface layer (water or protonated phosphate, HPO42–) individually. For this we utilized the developed pulse-edited 1H–31P heteronuclear–correlation (HETCOR) experiments in goat cortical bone samples (fresh, H–D-exchanged, and dehydrated). The developed ssNMR methodology has the ability to measure the relaxation parameters and dynamics of bone mineral components independently. Our study indicates a significant change in the dynamic behavior of both bone mineral components. We present the results from one particular bone sample under different conditions with the intention of establishing an experimental procedure for the further rigorous investigation of various phosphorus environments of healthy and diseased human bone samples. This study may help to explain the relaxation and dynamics of bone minerals inside the bone matrix.

Experimental Section

Sample Preparation

The corical fibular bone of an Indian goat (Capra hircus, 2–3 years old) was taken from a local slaughterhouse and used as a sample in the current study. The procured bone samples were cleaned of soft tissues, bone marrow, and periosteum. Further, the bone samples were washed with double-distilled water and kept wrapped in aluminum foil at ambient temperature until further use. To avoid any sample variability, the same animal bone sample was used for the ssNMR experiments. Previously, it was suggested that grinding of bone alters the uniformity and hydration properties of the bone matrix.[27] Therefore, in this study we have obtained small flakes of intact bone by filing them with the help of a bistoury. These flakes had a morphology similar to that of the intact bone as they were not subjected to any kind of processing such as grinding.[17,28,29] In the present study, we used three types of bone samples, i.e., fresh hydrated native bone, dehydrated bone, and H–D-exchanged bone. Fresh hydrated native bone was not subjected to any treatment and was analyzed within 1 h of its extraction from the animal. The dehydrated sample was prepared by lyophilizing the bone for 48 h, and the H–D-exchanged bone sample was prepared by soaking the fresh bone sample in the D2O solution (Sigma-Aldrich, U.S.) for 72 h at 200 rpm using an incubator shaker (Scigenics Biotech, India), which allowed for maximum water–D2O exchange. All samples were packed inside a 3.2 mm zirconium rotor for further ssNMR experiments.

Solid-State NMR Experimental Parameters

All the ssNMR experiments were carried out on a 14.1 T Bruker Biospin Avance III spectrometer operated at 600.124 (1H) and 242.934 MHz (31P) Larmor frequencies. The spectrometer was equipped with a 3.2 mm DVT probe (Bruker). The magic-angle spinning (MAS) speed was controlled by a MAS pneumatic unit (Bruker) with an accuracy of ±2 Hz. All one-dimensional (1D) and two-dimensional (2D) spectra were recorded at room temperature (298 K) at the MAS speed of 10 kHz. π/2 pulse lengths for 1H and 31P were calibrated at 3 and 3.4 μs, respectively. To determine the relaxation delays of the fresh native, dehydrated, and H–D-exchanged bone samples, 1D 31P one pulse experiments were carried out that were optimized with eight scans and 700, 1100, and 900 s recycle delays, respectively.

Since the relaxation delay times of 31P are more than hundreds of seconds, we used the cross-polarization (CP) inversion recovery (CPir) experiment[30] to measure the 31P spin–lattice relaxation (T1), which makes the 31P relaxation measurements significantly faster. Each spectrum was recorded under high-powered proton decoupling with 8 scans, a π/2 pulse length of 3 μs, a recycle delay of 5 s, and 16 relaxation intervals from 1 to 260 s. To measure the 31P spin–spin relaxation (T2), the Carr–Purcell–Meiboom–Gill (CPMG) spin–echo pulse sequence was carried out. The pulse length was calibrated to the 31P π/2 pulse length of 3.4 μs and a π-pulse of length 6.8 μs and used with the respective recycle delays in all three bone samples. Each spectrum was recorded using eight transients with various echo times from 0 to 400 ms. Signal intensities were measured for each spectrum, corresponding to the different echo times. The observed signal intensities were plotted as a function of the echo time and fitted to a (multi-) exponential curve using MATLAB.

1H–31P CPMAS spectra were recorded with 16 scans, the pulse length was calibrated to 3.0 μs for the 1H π/2 pulse with a 5 s recycle delay, and the ramp CP sequence with the Hartman–Hahn match condition was used with spinal-64[31]1H decoupling of 58.13 kHz with short (1 ms) and long (10 ms) contact times. The 2D 1H–31P HETCOR experiments (Figure A) were performed using phase-modulated Lee–Goldberg (PMLG)[32]1H homonuclear decoupling (Supporting Information). Each spectrum was recorded with eight scans, 64 T1 increments, a 5 s recycle delay, and different contact times (1 and 10 ms). In this sequence, the transverse 1H magnetization is allowed to evolve under the PMLG pulse sequence during the T1 period, which is used to suppress 1H–1H homonuclear dipolar couplings. The CP sequence was then performed, during which the information on the 1H spin magnetization was transferred to the 31P spins, and then the observation of the 31P spin was carried out during time T2.

Figure 1

(A) Pulse sequences used in 2D 1H–31P HETCOR spectroscopy under MAS. (B) 2D HETCOR pulse sequence used for the 31P measurement of the relaxation time T1. We introduced two consecutive π/2 pulse lengths with delay τ in the 31P channel. The black rectangles represent the π/2 pulse. (C) 2D 1H–31P HETCOR pulse sequence used for the 31P measurement of the relaxation time T2. We introduced delay τ with a π-pulse length in the 31P channel. The black rectangles in the proton channel represent π/2 pulse lengths, and those in the 31P channel represent π-pulses. The gray boxes in panels B and C show the introduction of π/2 and π-pulse lengths with delay τ.

For site-specific relaxation measurement of T1, we incorporated a π/2 pulse just after the CP transfer on the 31P channel, followed by evolution delay τ and another π/2 pulse just before acquisition in the 2D 1H–31P HETCOR pulse sequence (Figure B) (Supporting Information). These experiments were performed with various τ delays (1 to 150 s), and the phase of the 90° pulse (introduced just after the CP) was altered. Similarly, for the site-specific relaxation measurement of T2, we introduced a spin–echo by introducing a π-pulse on the 31P channel after the CP and varied the spin–echo by increasing the number of loops in the 2D 1H–31P HETCOR pulse sequence (Figure C) (Supporting Information). We kept all the parameters the same as those for the 2D 1H–31P HETCOR experiments (CP contact time of 10 ms). Further, T1 and T2 values were used for the calculation of the site-specific 31P correlation times (τc) of both the mineral components.

Results and Discussion

1H and 31P 1D and 2D HETCOR NMR of the Bone Mineral in Native Bone

Bone is a dynamic tissue that undergoes a continuous rebuilding process throughout the life span of a vertebrate. Water plays a crucial role in various interactions inside the bone matrix and also affect the mechanical properties of bone. Previously published reports suggested that dehydration and H–D exchange diminished both the strength and quality of bone.[7] In order to map the structural changes, we carried out a set of ssNMR experiments on fresh native, dehydrated, and H–D-exchanged bone samples isolated from goat bone tissues. One-pulse 1H MAS spectra of all the three bone samples were recorded at 10 kHz MAS, as shown in Figure . The presented 1H MAS spectra measures the relative water content in the sample because the 1H NMR spectra of the bone samples are largely dominated by water molecules relative to lipid molecules. In agreement with the previous reports, we observed hydroxide, lipid (narrow lines between 0 and 1.3 ppm), and water (5.2 ppm) in the 1H MAS spectrum of fresh native bone (green).[16] In agreement with the previous study,[28] we also observed a change in the relative distribution of water when the fresh native bone was subjected to dehydration and H–D exchange (Figure ). Upon chemical exchange, the line width and intensity of structural water (5.2 ppm) were found to change, suggesting that some of the bone water was replaced by D2O in the H–D exchange process (Figure , blue color). The study of bone by 1H NMR is complicated due to the overlapping of signals from both the mineral and the organic matrix; therefore, the 1H–31P CP NMR technique was used. The cross-polarization experiment is more effective for rigid systems than mobile components, as the 1H–31P dipolar interactions are reduced for the highly mobile components due to reduced cross-polarization. Thus, to study the mineral component, 1D 1H–31P CP MAS spectra were recorded at short (1 ms) and long (10 ms) contact times; (Figure S1A–C). We observed that the line shape and the line intensity vary as a function of the CP time in all the three-bone samples (Figure S1A–C). The narrower line width and good peak intensity (i.e., 10 ms contact time) was used in further studies.

Figure 2

1D 1H NMR MAS spectra of fresh native, H–D-exchanged, and dehydrated bone samples of goat cortical bone. Spectra were recorded at 600 MHz at room temperature.

The 31P ssNMR spectrum of bone contains a broad signal with a maximum at ∼3.2 ppm, which is a combination of the two components (crystalline apatitic core (orange color) and amorphous surface layer (cyan color)) (Figure A) as observed in the previous studies.[5,15,33] Such an overlap of these two lines could be differentiated by 2D 1H–31P HETCOR (Figure A), as can be seen in the spectrum in Figure B. 2D 1H–31P HETCOR spectra provide structural information on protonated species from the mineral matrix of bone that are in close proximity to phosphate ions. The mineral components of bone tissues mostly consist of phosphate ions. Therefore, phosphorus nuclei in bone tissue are primarily used to probe the mineral component, and the 2D 1H–31P HETCOR technique is widely used to study the local environment of 31P and 1H in bone apatite crystals. Therefore, we also conducted similar experiments to observe the mineral environment of fresh native, dehydrated, and H–D-exchanged bone samples. In line with earlier studies,[5,15] we also observed that the phosphorus NMR signal is dominated by the mineral rather than by phosphorus associated with the organic matrix, and we identified two different phosphate environments for all three types of bone samples. For the fresh native and dehydrated bone samples, two different 31P–1H signal correlation peaks were observed in the 2D 1H–31P HETCOR spectra, which exhibited a 31P chemical shift (PO43– at 3.2 ppm) similar to that correlated with OH–/PO43– at 0 ppm and water/HPO42– resonances spanning from 4 to 10 ppm in the bone matrix (Figures B and S2 A). However, the addition of D2O to the fresh bone sample affects both the mineral components (OH–/PO43–) and the water/HPO42– resonances associated with the crystalline apatitic core and the surface layer component (Figure S2B). Therefore, the H–D exchange in bone tissues confirmed that the 1H could be exchanged with 2H (resulting in the reduction of the water signal). The previous report[15] suggested that H–D exchange affected the surface layer component of bone minerals, but our results indicated that water molecules were involved in a proton chemical exchange with the mineral apatitic component as well as the surface layer component.

Figure 3

(A) Quantification of 31P cross-polarization MAS ssNMR spectrum of a fresh bone tissue sample (blue line) and its corresponding fitting (magenta dashed line). The two peaks correspond to the PO43–-containing crystalline apatitic core (orange peak) and the HPO42–-containing nonapatitic environments in the form of an amorphous surface layer (cyan peak). (B) 2D 1H–31P HETCOR spectrum of fresh native bone at a contact time of 10 ms. The two peaks of shown are the OH– signal at 0 ppm corresponding PO43– and HPO42– and water resonance between 4 and 10 ppm.

Effect of Hydration, Dehydration, and Chemical Exchange on the Relaxation Parameters: T1 and T2

To understand the intimate molecular interactions between the organic matrix and the mineral components in bone tissues, the study of atomic-level structure and dynamics is necessary. Therefore, recent developments in ssNMR methodologies represent indispensable methods for the quantitative elucidation of dynamic properties in bone tissues. Previously, NMR relaxation measurements were used to study bone quality by measuring the mobile and bound water components.[10] They were also used to determine the bone size distribution and the porosity of bone tissues.[34] In addition, 31P ssNMR spectroscopy was shown to be a valuable method for selectively analyzing the crystalline apatitic and amorphous surface layer mineral compartments in bone tissues.[35] Earlier reports suggested that the hydration–dehydration process affected the crystalline apatitic core and the amorphous surface layer.[15] However, the relaxation behavior and dynamics of the two mineral components were not studied individually in the previous literature. Therefore, to know the exact mechanism of bone mineralization and the motional behavior of bone minerals, it is imperative to study the relaxation properties and dynamics of both the mineral components associated with crystalline apatitic core and amorphous surface layer separately. A vast array of methods and sequences has been developed so far to calculate the T1 and T2 relaxation times of various biological tissues. However, the pulse sequences for measuring the relaxation and dynamic properties of individual mineral components (crystalline apatitic core and amorphous surface layer) in bone tissues were not identified until recently. Thus, in the current study, to understand the relaxation behavior of the inorganic mineral component individually, we measured site-specific 31P T1 (spin–lattice) and T2 (spin–spin) relaxation parameters in combination with 2D 1H–31P HETCOR (Figure B and C, respectively).

To date, very few publications have performed 31P NMR relaxation measurements of bone at a high magnetic field.[36] The inorganic mineral component of the bone tissues interacts with the organic mass, NCPs, and water molecules. The observed changes in these interactions also affect the T1 and T2 parameters. As described earlier, the T1 relaxation is dependent on the molecular dynamics, the phosphorus local environment, and the crystallinity of the bone crystal lattice;[37] therefore, any changes near the phosphorus nuclei in the bone crystal lattice perturbs the 31P T1 and T2 parameters. Thus, probing the relaxation behavior of phosphorus nuclei will provide information about the inorganic components of bone tissues. In addition to this, 31P T1 is also affected by the concentration of proton nuclei, which also depends on the aging and hydration of bone tissues. Therefore, we conducted a CP inversion recovery (CPir) experiment for the measurement of 31P T1 to identify the individual inorganic mineral components of the bone sample. An earlier report suggested that 31P T1 depends on the magnetic field, and a higher magnetic field is associated with a longer T1 time.[36]31P T1 was found to be equal to 100 s at 400 MHz, increasing by 15% upon dehydration.[38] Similarly, we observed a 31P T1 relaxation time of 125 s in the fresh native bone at 600 MHz, which increased significantly by approximately 48% upon dehydration and 15% upon H–D exchange (Figure B and Table S1). This increase in the 31P T1 values of dehydrated bone samples is possibly due to the presence of fewer 1H nuclei in the proximity of 31P because the majority of the water is removed upon dehydration. Similarly, higher 31P T1 values were obtained when 1H was replaced with 2H in the H–D-exchanged sample. The reason for this increase is due to the replacement of hydrogen by deuterium because deuterium is a spin 1 quadrupolar nucleus with a negligible dipole; therefore, the dipolar coupling would be smaller than that for the spin 1/2 proton, which has a large dipole. The obtained result suggested that the dipolar interaction between 1H–31P is the major mechanism for the T1 relaxation of phosphorus in the bone, as was also observed in the earlier report.[39] The 31P T1 measurements of all three bone samples are shown in Figure A.

Figure 4

(A) Spin–lattice (T1) measurements of fresh native, H–D-exchanged, and dehydrated bone samples taken using the CPir pulse sequence and (B) the corresponding graphical representation of the T1 values of the fresh native, H–D-exchanged, and dehydrated bone samples. (C and D) Graph-fitting curves of spin–lattice (T1) measurements combined with 2D 1H–31P HETCOR of fresh native, H–D-exchanged, and dehydrated bone samples at 0 and 5.2 ppm, respectively, and (E) the corresponding graphical representation of the fresh native, H–D-exchanged, and dehydrated bone samples with time at 0 (orange) and 5.2 ppm (blue) respectively.

Further, to individually measure the 31P T1 relaxation of the apatitic core (OH–/PO43– at 0 ppm) and the surface layer component (water/HPO42– at 5.2 ppm), pulse-edited 2D 1H–31P HETCOR experiments were carried out. For this we incorporated a π/2 pulse just after the CP transfer on the 31P channel, followed by an evolution delay τ and anoter π/2 pulse just before acquisition in the 2D 1H–31P HETCOR pulse sequence (Figure B). These experiments were recorded with various evolution delays. The alternating experiments create ±z magnetization. The phase cycling is done accordingly to record the difference between the two scans. For short relaxation delays, a maximum signal was obtained, while at longer relaxation delays the signal was found to decay exponentially as a function of the relaxation delay. We plotted a graph between the relative intensity (Figure C and D and Table S2) and the relaxation delay for both the peaks (0 and 5.2 ppm) in all three types of bone samples and observed that both the peaks followed the single-exponential decay (eq ) pattern with different T1 values, as shown in Table S2.

Little variation was observed in the 31P T1 values in the crystalline apatitic and surface layer components (0 and 5.2 ppm) in the fresh native bone sample (Table S2). Furthermore, upon dehydration we observed a significant change in the 31P T1 values of OH–/PO43– at 0 ppm associated with the crystalline apatitic core compared with the 31P T1 values of water/HPO42– (at 5.2 ppm) (Figure E). Previously it was suggested that 31P resonance is sensitive to hydration–dehydration and chemical exchange processes, and phosphate ions located near the surface layer of the bone mineral were found to be affected.[15] Additionally, it is also known that the T1 relaxation rate increases with the surface-to-volume ratio.[39,40] Later on, Kaflak et al. suggested that there is a transfer of water molecules between the amorphous surface layer and the crystalline apatitic core region.[35] In line with these studies, our experimental results showed a significant increase in the 31P T1 value for the apatitic core (0 ppm) in comparison with that for the amorphous surface layer component (5.2 ppm) upon dehydration. We speculate that these changes are due to the diffusion of water molecules from the crystalline apatitic core region to the amorphous surface layer component, thereby increasing the surface-to-volume ratio of the crystalline apatitic core and hence the T1 values.

Further, to determine the contribution of the 1H–31P dipolar interaction to 31P T1 of the bone mineral, relaxation measurements were repeated on the bone tissue sample after deuterium oxide (D2O) exchange. Accordingly, the majority of the water-binding sites on bone tissue were occupied by deuterium. Earlier it was determined that dipolar interactions in deuterated solutions are weaker and their contribution to relaxation can be expected to be reduced.[41] In a similar way, our results also confirmed the higher 31P T1 values in deuterated bone tissues when compared to those of the fresh native bone samples. This increase in the 31P T1 values was observed for the apatitic core region (0 ppm), which might be due to the fact that the mineral platelets are so thin (4 nm)[5] that deuteration at the surface would be sufficient to reduce the effect of the 1H spin diffusion experienced by the crystalline apatitic core and would thus impact the 31P T1 values.

In addition to 31P T1 measurements, we performed the CPMG spin–echo pulse sequence to measure the 31P transverse relaxation parameter T2. T2 of the 31P NMR spectrum of bone provides information about the local environment around the phosphate (PO43–/HPO42–) moiety in the bone matrix. Therefore, any variations in the 31P T2 values will reflect the changes in the environment near the phosphorus-containing sites of native bone upon dehydration or H–D exchange. In the present study, each spectrum of the three bone samples was recorded with various echo times from 0 to 400 ms (Figure A). The observed signal intensities were plotted as a function of echo time, which gave a biexponential fit with two T2 parameters, namely the short and long components (Table S1) (eq ).

Figure 5

(A) Spin–spin (T2) measurements of fresh native, H–D-exchanged, and dehydrated bone samples taken using the Car–Purcell–Meiboom–Gill (CPMG) pulse sequence and (B) the corresponding graphical representation of T2 values; the slow relaxation component of the fresh native, H–D-exchanged, and dehydrated bone samples is shown in blue, and the fast relaxation component is shown in pink. (C) and (D) Graph fitting curves of spin–spin (T2) measurements combined with 2D 1H–31P HETCOR of the fresh native, H–D-exchanged, and dehydrated bone samples at 0 and 5.2 ppm, respectively, and (E) the corresponding graphical representation of T2 values at 0 (orange) and 5.2 ppm (blue).

All three bone samples showed decaying behavior for both the short and long T2 components as the interpulse delay increased (Figures A and B). Hydration network dependencies of phosphorus nuclei affect the motional behavior. In line with previous studies, we also observed changes in both components of the T2 values of dehydrated and H–D-exchanged bone; however, a more conspicuous change was observed in the slow-relaxing component (long T2, Table S1).[28] The slow- (long T2) and fast-relaxing (short T2) T2 components represented the 31P sites close to and distant from the water molecules, respectively. Among all three bone samples, dehydrated bone showed the largest T2 value compared to those of native and H–D exchanged bone. We assumed that dehydration leads mineral platelets to become closer, thereby restricting the motion of minerals that show longer T2. Here, we refrain from saying whether the obtained two values of T2 are from the apatitic core and the surface layer of the bone mineral.

Further, we have performed pulse-edited 1H–31P HETCOR experiments for site-specific T2 measurements of all three bone samples (Figure C and D). For these measurements we introduced a π-pulse on the 31P channel after CP with various spin–echo by increasing the number of loops in the 2D 1H–31P HETCOR pulse sequence (Figure C) during the time [τ/2 – π – τ/2]. Transverse magnetization decay in a solid-state sample generally depends on the chemical shift anisotropy (CSA) and the homonuclear and heteronuclear dipole–dipole interactions.[42] The π-pulse introduced after the contact pulse reverses the dephasing induced by all these phenomena and eliminates their effects. The decaying spin echo amplitude was plotted as a function of time, which yielded a curve that decayed exponentially with 31P T2. 31P T2 measurements were carried out for both peaks in all three bone samples (0 and 5.2 ppm) individually (Figure C and D and Tables S1 and S2). At 0 ppm, the relative intensity and spin–echo time plot was fit by a single-exponential curve (eq ), while at 5.2 ppm the plot was fit by a biexponential curve (eq ) (Figure C and D and Table S2).

Regarding the apatitic core at 0 ppm (OH–), no significant changes in 31P T2 were observed between fresh native and dehydrated bone samples, while very few changes were observed in deuterated bone. The change in the 31P T2 deuterated bone sample is due to spin diffusion, as illustrated previously. At 5.2 ppm, the 31P T2 of the amorphous surface layer exhibited two components, one with a slower relaxation rate and another with a faster relaxation rate. The slow-relaxing component showed the relaxation of 31P sites in close proximity to water molecules, whereas the fast-relaxing component represented the 31P sites distant from the water molecules. This was further proved when we dehydrated the fresh bone sample and found a larger variation in the slow relaxing component but only an insignificant change in the fast relaxing component. In the deuterated bone sample, both slow- and fast-relaxing components showed significant changes, which were due to anisotropic interactions. Overall, the combination of 31P T1 and T2 parameters provides details of the relaxation behavior of the two distinguished peaks (OH–/PO43– and water/HPO42–) associated with the crystalline apatitic core and the amorphous surface layer component independently.

Water-Mediated 31P Dynamics in the Bone Mineral

Over the past few decades, ssNMR has emerged as a crucial technique for elucidating molecular motion with atomic resolution and dynamics in biological systems. Earlier, the experimental analysis for measuring the relaxation parameters in MAS ssNMR was developed, providing information about time scales and amplitudes of motions ranging from picoseconds to milliseconds.[42] The NMR relaxation parameters T1 and T2 studied in the current study mainly provide useful information about the dynamics, i.e., the rotational correlation time τc, and further structural details of bone minerals. In general, anisotropic interactions, i.e., chemical-shift anisotropies, dipolar–dipolar interactions, quadrupolar couplings, scalar relaxation, chemical exchange, and paramagnetic relaxation, are major mechanisms that give direct information about the magnitude of the molecular motions. Among these various relaxation mechanisms, CSA and dipolar relaxation can be directly related to probe the rotational motion. Calcium hydroxyapatite (CHA) is one of the best models resembling bone, and its 31P CSA value is very low, ∼20 ppm.[43] In the proton-decoupled 31P static NMR experiment (Figure S3), we also observed that the 31P static NMR spectrum did not show the powder pattern and was symmetric, exhibiting a very small CSA contribution. Therefore, the contribution of CSA to relaxation is omitted in the present study. Thus, the dominant relaxation mechanism was dipolar interactions, i.e., the relaxation in the heteronuclear spin system is almost always dominated by directly attached 1H nuclei. Herein, we focused our attention on investigating the behavior of spin relaxation to probe the molecular dynamics as a function of the rotational correlation time τc, which characterizes the time scale of the molecular motion. We used the Lipari–Szabo model-free approach to obtain the correlation time of the bone mineral components. The structural model of the bone minerals suggested the internal crystalline core was coated by an amorphous layer.[5] The platelet-shaped bone minerals were found to be very thin (4 nm; the crystalline apatitic core is ∼2 nm thick and the amorphous surface layer is ∼0.8 nm thick).[5,44] Therefore, the environments near the 31P nuclei in both the crystalline apatitic core and the amorphous surface layer would be nearly isotropic, and the time scales of the dynamics would be of the same order. Therefore, for an isolated heteronuclear two spin system undergoing isotropic rotational tumbling, the spectral density function[42,45,46] in the model-free approach is given by

We assume here that 31P relaxation is mainly governed by dipole–dipole interactions between 31P and 1H nuclei and the axially symmetric chemical shift anisotropy tensor, i.e., no asymmetry. Therefore, the correlation time equation for dipole–dipole spin–lattice relaxation (T1)[42] of 31P is given by

Similarly, the correlation time equation for the dipole–dipole spin–spin relaxation (T2) of the 31P nuclei is given bywhere DIS is dipolar coupling between I and S spins and ωI and ωS the Larmour frequencies of I and S spins, respectively. In the present study, the measured 31P T1 and T2 relaxation times of the apatitic core and the amorphous surface layer can further be used to determine the rotational correlation time for both components separately in all three bone samples. For larger molecules, we assumed that the spectral density at frequency zero and ωS were important, giving the relation

Table S3 shows the molecular correlation times for all three bone samples, which were calculated using eq . Two 31P correlation times (on the order of 10–7 s) were obtained for all the three bone samples (Table S3), which suggested the presence of different molecular motions in the mineral component of the bone sample. For fresh native bone, the fast correlation time showed the molecular motion (0.19 ± 0.01 μs; Figure A, cyan) of the 31P nuclei that were in close proximity to water molecules with a greater degree of freedom; conversely, the larger correlation time (0.59 ± 0.03 μs; Figure A, purple) suggested the water molecules distant from the 31P nuclei as discussed in the preceding section. We did not observe any significant changes in the correlation times (0.60 ± 0.03 and 0.19 ± 0.01 μs) in the chemical exchange process. However, upon dehydration, both the 31P correlation times increased, confirming the slower dynamics of the 31P nuclei. The significant increase in the long correlation time (0.68 ± 0.04 μs; Figure A, purple) is due to the fact that the removal of water make the structures of minerals more compact, with slow or restricted motions around 31P sites. Similarly, decreased motional freedom due to a decrease in the number of 31P sites (in close proximity to water molecule) was observed in the short correlation time (0.21 ± 0.01 μs; Figure A, cyan).

Figure 6

(A) Measurement of the 31P correlation time of the combined bone mineral (core and surface). The short component is shown in cyan, and the long component is shown in purple. (B) Measurement of the 31P correlation time of the crystalline apatitic core component at 0 ppm. (C) Measurement of the 31P correlation time of the amorphous surface layer component in fresh native, dehydrated, and H–D -exchanged bone samples at 5.2 ppm. The short component is shown in sky blue, and the long component is shown in navy blue.

Further, to determine the exact dynamics of the crystalline apatitic core (0 ppm) and the amorphous surface layer components (5.2 ppm) individually, we calculated the site-specific 31P correlation times of both the components. At 0 ppm, we observed a significant change in 31P correlation times of the crystalline apatitic core upon the removal of water and chemical exchange. Upon dehydration by lyophilization, a increase in the 31P correlation time (0.60 ± 0.05 μs) (Figure B) was greater than that in the fresh native sample (0.42 ± 0.16 μs; Figure B). This may be due to the water molecules diffusing from the apatitic core to the surface layer component in a similar way as discussed in the previous section. In the H–D exchange process, few significant changes were observed in the 31P correlation time (0.47 ± 0.04 μs; Figure B). Further, to observe the motional changes in the amorphous surface layer component (at 5.2 ppm), two 31P correlation times (in the range of 10–6–10–7 s) were obtained for all three-bone samples (Figure C). After dehydration, we observed that the values of both short (0.47 ± 0.05 μs; Figure C, sky blue) and long (1.40 ± 0.25 μs; Figure C, navy blue) correlation times were similar to those of the fresh native bone sample, suggesting there was similar molecular motion at the surface layer. However, a more pronounced effect on 31P correlation time was observed for the H–D exchange process (0.96 ± 0.11 and 0.36 ± 0.09 μs; Figure C, navy blue and sky blue, respectively). This change in the 31P correlation time is due to the exchange of 1H with 2H at the surface sites, weakening the hydrogen bonding network and hence increasing the molecular motion; therefore, a shorter correlation time was observed. Overall, we demonstrated the noticeable effect of dehydration-induced changes in 31P correlation times for the apatitic core region in comparison with the amorphous surface layer components (Table S3 and Figure S4), which is due to the diffusion of water molecules from the core to the surface in the dehydration process affecting core more significantly. On the contrary, the H–D exchange process produced more significant changes in the 31P correlation times of the surface layer component, which shows that the surface proton sites are more likely to be replaced than the core.

Conclusions

In this work, we measured 31P T1 and T2 relaxation in combination with the 2D 1H–31P HETCOR pulse sequence and the 31P correlation time in goat cortical bone. For the first time, we indicated the site-specific relaxation and dynamics of different mineral components associated with crystalline apatitic core and the amorphous surface layer components individually. In the present study, we observed a great extent of variation in the 31P T1 and T2 values among all three bone samples. The combination of the 31P T1 and T2 parameters provides insight into details of the water-mediated, i.e., dehydration and H–D exchange, relaxation behavior of 31P nuclei associated with the amorphous surface layer (water/HPO42– at 5.2 ppm) and the crystalline apatitic core (OH–/ PO43– at 0 ppm) independently. Along with these, 31P correlation times indicated the different dynamic behaviors of the apatitic core and amorphous surface layer mineral components. We found two different 31P environments in the surface layer component, representing the different motional behaviors of the mineral platelets. Additionally, we observed the that effect of dehydration on the 31P correlation times was more pronounced for the apatitic core region, whereas more changes of the amorphous surface layer components were observed in the H–D exchange process. In summary, based on the ssNMR studies, we have demonstrated that the two mineral components (OH–/ PO43– and water/HPO42–) associated with the crystalline apatitic core and the amorphous surface layer behave distinctively with different dynamic behaviors. This study provides unprecedented insights into the site-specific relaxation dynamics and structural features of bone minerals. It has been postulated that bone water greatly influences the mechanical properties of the bone matrix and that hydration can be changed dynamically in various bone-related diseases, including osteoporosis and osteogenesis imperfecta. Therefore, understanding the water-mediated changes in the structure and relaxation dynamics of bone minerals will help in the development of improved architectures and mechanically strong bone substitutes.