2.1.

Tissue Sample Preparation and Crush Injury

For tissue sample preparation, 6-week old Sprague–Dawley rats (Shimizu Laboratory Supplies, Kyoto, Japan) were used. The animals were maintained on a 12-h light/dark schedule with free access to food and water. The Committee for Animal Research, Kyoto Prefectural University of Medicine approved all surgical and experimental procedures. All surgical and experimental procedures were carried out in accordance with the guidelines of the National Institutes of Health on Animal Care.

All rats were deeply anesthetized using a mixture of oxygen and 2.0% isoflurane (Abbott Japan Co., Osaka, Japan). The right sciatic nerve was exposed and ligated at the proximal-thigh level using 3–0 silk to create a crush injury. Following ligation for 5 min, the suture was carefully released to produce a regenerating axonotmesis model.^14,15 Samples were collected in the intact condition (day 0) at 7, 14, 21, and 28 days after crush injury ($n=3$ for each time point).

For analysis of sciatic nerve tissue, the animals were deeply anesthetized and transcardially perfused with 0.9% NaCl, followed by 4% paraformaldehyde in 0.1 M phosphate-buffered saline (pH 7.4). After postfixation with 25% sucrose and cryoprotection, 15-mm segments of nerve tissue were sectioned by cryostat (14 μm; CM3050S; Leica, Nussloch, Germany), including regions proximal and distal to the injury site in the longitudinal plane. The frozen sections were then mounted on the glass slides.

2.2.

Raman Spectroscopy

Raman spectra were measured using a LabRam Aramis instrument (Horiba Jobin-Yvon, Kyoto, Japan) with a focal length of 460 mm and the entrance slit of the spectrometer set to 100 μm. The dispersive element had a grating of $600\text{lines}/\mathrm{mm}$ and spectral resolution was $<3{\mathrm{cm}}^{-1}$. Raman spectra were produced by excitation with a 532-nm laser focused on the sample through $50\times$ and $100\times$ microscope objectives (NA 0.75 and NA 0.9; Olympus, Japan). The lateral ($x\text{-}y$) spatial resolution was 2 μm and the depth ($z$) spatial resolution was 4 μm using the $50\times$ objective lens. The lateral spatial resolution was 1 μm and the depth spatial resolution was 2 μm using the $100\times$ objective lens. The objectives provided a laser spot of 1–2 μm on the sample. Laser power was set at 6.0 and 0.6 mW for analysis of sciatic nerve tissue and cells, respectively. Raman scattered light was collected in 180 deg back-scattering mode and recorded on a liquid nitrogen-cooled charge-coupled device camera (Symphony CCD detector; Horiba Scientific, Japan).

The serial mapping technique was used for the analysis of Raman nerve tissue spectra. Nerve tissue was placed on an automated motorized $X$, $Y$ mapping stage and spectra were obtained at different points on the tissue by moving it under the microscopic objective. For mapping in the $X$, $Y$ planes at chosen spatial resolutions, the sample was moved in both spatial dimensions ($X$ and $Y$) and a spectrum was recorded at each ($X$, $Y$) position.¹⁶ The measured area of sciatic nerve tissue was $140\times 140\mu \text{m}$, with a 10-μm step size in both dimensions.

To evaluate the peak changes quantitatively, three fields in each sample were randomly detected ($n=9$ for each time points). The area of the $\mathrm{C}─\mathrm{H}$ stretching band between 2770 and $3200{\mathrm{cm}}^{-1}$ was analyzed using a Gaussian Lorentzian fit with the center wavenumbers of the six components at 2853, 2890, 2940, 2976, 3010, and $3062{\mathrm{cm}}^{-1}$. The ratio of band intensity (I2940/I2853) was calculated, in which the intensity at $2940{\mathrm{cm}}^{-1}$ was divided by that at $2853{\mathrm{cm}}^{-1}$.

2.3.

Distribution Analysis of Raman Spectra

To identify the components that contribute to the spectrum of intact sciatic nerve tissue, the classical least squares (CLS) method in LabSpec5 software (Horiba Jobin-Yvon)^17,18 was used. This method is generally used for analysis of the components in a substance.¹⁹ In the current analysis, spectra of neurites of DRG and sphingomyelin were used as representative of axons and myelin, respectively. These spectra were summed and compared with the original spectrum of the intact sciatic nerve. To visualize the distribution of axons and myelin in sciatic nerve tissue, color images were constructed using mapping data for the intact sciatic nerve. The distribution of axon and myelin components at each mapping point of the sciatic nerve was expressed in separate spectra and a merged image was calculated. In the separate spectra, the greater the contribution to the mapping point, the denser the color. Residual components (excluding axons and myelin) were also determined, with the color density expressing the intensity of these components in the tissue. The residual components were within 4.5% of the total intensity, which was regarded as an acceptable error.

Changes in the distributions of axon and myelin were also measured. The mapping data at each time point after crush injury were collected. The contributions of axon and myelin were obtained by dividing the mean density of each component by that of the total area in each mapping image and expressed as C(axon) and C(myelin). The axon/myelin ratio was calculated by dividing C(axon) by C(myelin). Before the application of chemometric method, spectra were baseline-corrected using piecewise linear correction with the same baseline points for all maps and reference spectra.

2.4.

Primary Culture of Dorsal Root Ganglion Neurons and Schwann Cells

Dorsal root ganglion neurons (DRGs) were dissected from spinal cords of embryonic day 13.5 mice (Shimizu Laboratory Supplies) and treated with trypsin for 20 min in Hanks’ buffered salt solution. Cells were dissociated mechanically and the trypsin reaction was terminated by the addition of Neurobasal medium (Gibco, Carlsbad, California). After centrifugation, cells were resuspended in DRG medium [Neurobasal medium (Gibco), penicillin/streptomycin (Gibco), 2% B27 (Gibco), 0.4% glucose, 2 mM l-glutamine, and $100\mathrm{ng}/\mathrm{ml}$ nerve growth factor (Sigma–Aldrich, St. Louis, Missouri)] and plated on dishes coated with Matrigel (BD Biosciences, San Jose, California). Mitosis of non-neuronal cells was inhibited by 4-day treatment with $13\mathrm{mg}/\mathrm{ml}$ 5-fluoro-2′-deoxyuridine initiated 24 h after plating. These cultures can be maintained indefinitely with half-volume changes of medium every 3 days. Primary cultures of Schwann cells were obtained from the sciatic nerve of P2 rats and plated onto 60-mm tissue culture dishes. The resulting cultures were maintained at 5% ${\mathrm{CO}}_2$ at 37°C in Dulbecco’s modified Eagle’s medium containing 10% fetal bovine serum. To analyze phospholipids incorporated in myelin, we detected the presence of sphingomyelin using a Phospholipid kit (Avanti Polar Lipids, Inc. Alabaster, Alabama).²⁰

2.5.

Immunocytochemistry

Cultured DRGs and Schwann cells were incubated overnight at 4°C with primary antibodies: mouse monoclonal antibody against Neurofilament 200 (Sigma–Aldrich) and mouse monoclonal antibody S100 (1:200; Abcam, Cambridge, Massachusetts). After rinsing off primary antibodies, the cells were incubated for 2 h at room temperature with a secondary antibody for staining (goat anti-mouse IgG labeled with Alexa 488; Molecular Probes, Eugene, Oregon) diluted $1:1000$. The slides were then coverslipped with Gelvatol (20% polyvinyl alcohol and glycerol mixture) and imaged on a confocal laser-scanning microscope (Fluoview 300; Olympus). The detailed procedures for immunohistochemistry are described elsewhere.^11,15

2.6.

Immunohistochemistry

Mounted sections were incubated overnight at 4°C with primary antibodies: rat monoclonal antibody against myelin basic protein (MBP; 1:2000; Millipore, Temecula, California) and rabbit monoclonal antibody against NF200 (Sigma–Aldrich). After rinsing off primary antibodies, the sections were incubated for 2 h at room temperature with secondary antibodies for double staining (goat anti-rabbit IgG labeled with Alexa 488 and goat anti-rat IgG labeled with Alexa 548; both Molecular Probes) diluted $1:1000$. Slides were analyzed with the procedure used for immunocytochemistry. In semiquantitative analysis of axonal and myelin changes, microscopic fields in each sample were randomly selected. Images were digitized on a gray scale using an automatic threshold tool linked to morphometry software (ImageJ, National Institutes of Health, Bethesda, Maryland). The mean number of axons was counted in the field (area: $1.25\times {10}^2{\mathrm{mm}}^2$) and expressed as the number per ${\mathrm{mm}}^2$ ($\mathrm{N}/{\mathrm{mm}}^2$). The myelinated area was expressed as the MBP-immunopositive area divided by the total area (%). To evaluate the relative changes of axon to myelin, the number of axons was divided by the myelinated area and expressed as the axon/myelin ratio.

2.7.

Statistical Analysis

All results are expressed as $\text{means}\pm \text{standard deviation}$. Nonparametric one-way analysis of variance was used within the same group to evaluate the significance of differences at multiple time points, with a Tukey-Kramer post hoc test. All statistical analyses were performed using JMP6 software (SAS Institute, Cary, North Carolina).

3.1.

Raman Spectra of Intact and Injured Sciatic Nerves

Raman spectra of an intact rat sciatic nerve tissue were examined. An epi-illumination image of longitudinally sectioned intact sciatic nerve segments is shown in Fig. 1(a). Raman mapping of the rectangular region of Fig. 1(a) at 1-μm intervals gave the mean spectrum shown in Fig. 1(b). Prominent peaks in the range were 1091, 1302, 1450, 1660, 2853, 2885, and $2940{\mathrm{cm}}^{-1}$. The Raman bands at 1091, 1302, 1450, and $1660{\mathrm{cm}}^{-1}$ correspond to $\mathrm{C}─\mathrm{C}$ stretching vibration, amide Ш, CH deformation, and amide І, respectively (Table 1).^21,22 These bands are mainly derived from the lipid and protein contents. The bands at 2853 and $2885{\mathrm{cm}}^{-1}$ arise from symmetric and asymmetric ${\mathrm{CH}}_2$ stretching modes. The ${\mathrm{CH}}_2$ vibration modes mainly reflect lipid contents. The band at $2940{\mathrm{cm}}^{-1}$ is assigned to ${\mathrm{CH}}_3$ symmetric stretching, which is derived from methyl groups that are abundant in proteins.

Fig. 1

Raman spectra of sciatic nerve tissue. (a) Epi-illumination image of Raman spectra in longitudinally sectioned sciatic nerve segments. (b) Mean spectrum collected from the Raman mapping of intact sciatic nerve in the rectangular region of a. Prominent Raman peaks are at 1091, 1302, 1450, 1660, 2853, 2885, and $2940{\mathrm{cm}}^{-1}$. (c) Time-dependent Raman spectra of injured sciatic nerve in the range $2800–3000{\mathrm{cm}}^{-1}$ at 0, 7, 14, 21, and 28 days after crush injury. The peak at $2853{\mathrm{cm}}^{-1}$ showed a decrease after injury and the peak at $2940{\mathrm{cm}}^{-1}$ had a relative increase. (d) Quantitative analyses of the band intensity ratio (I2940/I2853) for each time point. The ratio increased at 21 days after injury, but then decreased at 28 days. Values are expressed as $\text{means}\pm \text{standard deviation}$. *Significant difference between time points ($P<0.05$).

Table 1

Representative Raman spectra collected from sciatic nerve tissue and the chemical bond corresponding to each Raman band.

To examine changes in Raman bands in injured sciatic nerve compared with the intact sciatic nerve, spectra were collected at 7, 14, 21, and 28 days after crush injury. The mapping area was collected from a 5-mm portion distal to the crush site [Fig. 1(c)]. The three peaks from 2800 to $3000{\mathrm{cm}}^{-1}$ exhibited clear postinjury changes. Changes were not observed in the region at $1000-1800{\mathrm{cm}}^{-1}$, which reflects chemical structures including amides І and III. Therefore, we focused on the range $2800–3000{\mathrm{cm}}^{-1}$. The peak at $2853{\mathrm{cm}}^{-1}$ showed a decrease after injury and that at $2940{\mathrm{cm}}^{-1}$ showed a relative increase. At 28 days after injury, the peak at $2940{\mathrm{cm}}^{-1}$ showed an apparent decrease compared with that at $2885{\mathrm{cm}}^{-1}$.

To evaluate these peak changes quantitatively, the band intensity ratio (I2940/I2853) was calculated [Fig. 1(d)]. This ratio showed a significant increase after injury that continued for 21 days. At 28 days after injury, the ratio decreased significantly.

3.2.

Raman Spectra of Cultured DRG Neurons and Schwann Cells

Transmission images of DRG neurons and Schwann cells [Figs. 2(a) and 2(c)] and immunofluorescent images [Figs. 2(b) and 2(d)] of these cells were obtained after collection of Raman spectra [Fig. 2(e)]. In these spectra of neurites of DRGs, a prominent peak was located at $2940{\mathrm{cm}}^{-1}$, which is due to ${\mathrm{CH}}_3$ symmetric stretching. Prominent peaks for Schwann cells occurred at 2851, 2885, and $2909{\mathrm{cm}}^{-1}$, which are associated with ${\mathrm{CH}}_2$ vibration modes reflecting lipid contents. The peak pattern of axons was distinct from that of Schwann cells in the $2800–3000{\mathrm{cm}}^{-1}$ range. The Raman spectra of pure sphingomyelin, one of the main phospholipids in myelin, also showed prominent peaks at $2800–3000{\mathrm{cm}}^{-1}$, similar to those for Schwann cells. This finding suggests that the peaks observed for Schwann cells reflect the composition of myelin.

Fig. 2

Raman spectra of cultured DRG neurons and Schwann cells. Transmission images (a and c) and immunofluorescent images (b and d) of DRG neurons (a and b) and Schwann cells (c and d). $\text{Scale bars}=50\mu \text{m}$. (e) Raman spectra of neurites of a DRG neuron and cell bodies of Schwann cells. Prominent Raman peaks of neurites are at 2886 and $2940{\mathrm{cm}}^{-1}$. Raman Schwann cell peaks are at 2851, 2885, and $2909{\mathrm{cm}}^{-1}$. Raman spectra of sphingomyelin were also detected, in which Raman peaks are at 2849, 2885, and 2919 ${\mathrm{cm}}^{-1}$. The Schwann cell Raman spectrum between 2800 and $3000{\mathrm{cm}}^{-1}$ showed a similar pattern to that of sphingomyelin.

3.3.

Distribution Analysis of Raman Spectra

To compare Raman bands of cells with those of sciatic nerve tissue, we focused on the CH stretching bands in the $2800–3000{\mathrm{cm}}^{-1}$ range. In this region, neurites of DRGs showed a prominent peak at $2940{\mathrm{cm}}^{-1}$ [Fig. 3(a)], while peaks for sphingomyelin were found at 2853 and $2885{\mathrm{cm}}^{-1}$ [Fig. 3(b)]. The neurites of DRGs and sphingomyelin are representative of the axon and myelin components, respectively. To analyze how each component contributes to the spectrum of intact sciatic nerve tissue, spectra of neurites of DRGs and sphingomyelin were summed using the CLS method. The summed spectra [shown in black in Fig. 3(c)] was similar to the spectrum for intact sciatic nerve tissue [shown in red in Fig. 3(c)].

Fig. 3

(a–c) Spectra of intact sciatic nerve tissue shown in Fig. 1(b) were separated into two spectra, based on spectra for DRG neurites (a) and sphingomyelin (b). A summed spectrum of DRG neurites and sphingomyelin is shown in black (c). The spectra of intact sciatic nerve are shown in red (c). The summed spectrum and the spectrum for nerve tissue have similar patterns. (d–f) The distribution of axon (represented by DRG neuritis) and myelin components (represented by sphingomyelin) in intact sciatic nerve tissue is expressed in colored images: axons in green (d), myelin in red (e), and a merged image (f). These figures show that myelin was aligned parallel to axons. (g) Residual component image. (h) Time-dependent changes in the distribution ratio of axons divided by myelin in injured sciatic nerve. The changes in the distribution of axon and myelin were similar to the band intensity ratios.

To investigate how axons and myelin were distributed in sciatic nerve tissue, color images were constructed using mapping data for intact sciatic nerves. With reference to the spectra in Figs. 3(a) and 3(b), the distributions of axon and myelin components at each mapping point of the sciatic nerve are shown in green [Fig. 3(d)] and red [Fig. 3(e)], respectively. The merged image is shown in Fig. 3(f). The green and red areas are aligned in parallel, similar to the typical structure of peripheral nerves in longitudinal slices [Figs. 4(a)–4(c)]. The image of the residual components is shown in Fig. 3(g).

Fig. 4

Immunohistochemical analyses of serial sections of sciatic nerve. Axonal and myelin expressions are shown in injured sciatic nerves double-labeled with NF200 and MBP. $\text{Scale bars}=100\mu \text{m}$. Expression of NF200 (a, d, g, j, m), MBP (b, e, h, k, n), and merged images (c, f, i, l, o) in intact (a–c) and crush-injured sciatic nerves (d–o). The number of immunopositive axons (NF200) and myelin (MBP) are decreased at 7 days after injury. The immunoreactivity recovered by 14 days and gradually increased until 28 days. The number of axons (p) and the myelinated area (q) are shown. Relative axon and myelin changes are expressed as the axon/myelin ratio (r).

To examine how the distribution of axons and myelin changes in injured sciatic nerve, the ratio of the contributions of axon and myelin was calculated [Fig. 3(h)]. The results showed an initial trend for an increase followed by a reduction at 28 days after injury, similar to the changes in the band intensity ratio.

3.4.

Immunohistochemical Analysis in Intact and Crush-Injured Sciatic Nerves

For comparison with the Raman spectral changes, we performed immunohistochemical analysis using the serial sections used in the Raman spectroscopic analysis [Figs. 4(a)–4(o)]. In intact sciatic nerve (d0), axons staining positively for NF200 [Fig. 4(a)] and myelin staining positively for MBP [Fig. 4(b)] are aligned in parallel, reflecting the typical structure in which myelin wraps around the axon. After crush injury, immunopositive axons and myelin decreased because of Wallerian degeneration. At 14 days after injury, immunopositive nerve fibers were observed. The increase in regenerated axons and myelin with the restructuring of peripheral nerve tissue was confirmed at 28 days after injury. Quantitative analysis showed that the number of axons and the myelinated area were significantly decreased at 7 days after crush injury, and then increased until 28 days after injury [Figs. 4(p) and 4(q)]. To analyze the relative change, the axon to myelin ratio was calculated. This ratio increased until 21 days after injury and then showed a significant reduction at 28 days [Fig. 4(r)]. This change was correlated with the changes in the band intensity ratio [Fig. 1(d)] and distribution [Fig. 4(h)].