Evaluation of the histological and mechanical features of tendon healing in a rabbit model with the use of second-harmonic-generation imaging and tensile testing

Rabbit model of tendon healing

The experimental protocol used in this study was approved by the Bioethics Committee for Animal Experiments at The Tokushima University. A rabbit model of tendon healing was prepared following the method used in previous research by another group.¹³ In total, eight male Japanese white rabbits (eight to ten weeks old, weight 2.0 kg to 2.5 kg) were obtained from Kitayama Labes Co., Ltd. (Ina, Japan) and were individually housed in a climate-controlled animal care facility for one week in order to be acclimatised to a new environment. The rabbits were locally anesthetised with 1% lidocaine hydrochloride after inducing anaesthesia with the inhalation of isoflurane (5% for induction, 2% for maintenance, each with an air flow rate of 2 litres/minute). Following anaesthesia, the right leg was disinfected and shaved. The flexor digitorum tendon in the right leg was sharply transected by a No. 11 scalpel. The disrupted tendon was then repaired by intratendinous stitching with a looped 5-0 nylon suture. Post-operatively, the rabbits were individually housed in a climate-controlled animal care facility. To highlight the definite difference of collagen structure between the healing tendon and the control tendon, we performed the comparison at a single time-point of tendon healing. At four weeks post-operatively, the rabbits were euthanised by administering an overdose of sodium pentobarbitone (0.15 mg/kg BW). The flexor digitorum tendon in the right leg was excised as the healing tendon sample (n = 8), whereas that in the left leg was excised as the control (n = 8). Samples were kept frozen at -15 °C until evaluation. After being thawed at room temperature, each specimen was examined by SHG imaging, followed by tensile testing.

SHG imaging

Figure 1 shows the experimental setup of the inverted SHG microscope. We used a mode-locked Cr:Forsterite laser (CrF-65P, Avesta Ltd., Moscow, Russia, centre wavelength 1250 nm, pulse duration 70 fs, repetition rate 73 MHz) to enhance the depth of penetration in SHG imaging.¹⁴ The mean power of the laser output light was adjusted to 20 mW at the sample surface by using a combination of a half-wave plate (HWP; CWO-1250-02-10, Lattice Electro Optics Inc., California, retardation tolerance = λ/500) and a polariser (P; PGL8610, CASIX Inc., Fujian, China, extinction ratio < 1 × 10^-6). Its linear polarisation was converted into circular polarisation by a quarter-wave plate (QWP; CWO-1250-04-10, Lattice Electro Optics Inc., retardation tolerance = λ/500) in order to cancel the polarisation dependence of the SHG efficiency of the orientated collagen fibres. The focal point of the laser beam was two-dimensionally scanned on a sample by a combination of a galvanometer mirror (GM), relay lenses (RL1 and RL2), and an objective lens (OL; CFI Plan 50 × H, Nikon Corp., Tokyo, Japan, magnification ×50, numerical aperture 0.9, working distance 350 µm, oil-immersion type). Backscattered SHG light was collected by the same OL, was reflected by a harmonic separator (HS; LWP-45-Runp625-Tunp1250-B-1013, Lattice Electro Optics Inc., Fullerton, California, reflected wavelength 625 nm), passed through an optical filter with a sharp pass-band (BPF; 625/26 nm BrightLine, Semrock Inc., Rochester, New York, pass wavelength 612 nm to 638 nm), and was then detected by a photon-counting photomultiplier with Peltier cooling (PMT; H7421-40, Hamamatsu Photonics K.K., Hamamatsu, Japan). Using the above GM optics, SHG images of a 400 µm × 400 µm region, composed of 256 pixels × 256 pixels, were acquired at a rate of 0.5 images/second. The range of probing depth in the present system was 0 ~ 200 µm from the sample surface, which was limited by the working distance (350 µm) of the OL and the thickness (120 ~ 170 µm) of a cover glass between the OL and the sample, and not limited by the penetration depth (~several hundred µm) of the 1250 nm laser light. To expand the lateral imaging region, we scanned the sample position at intervals of 400 µm using a stepping-motor-driven 3-axis translation stage every time an SHG image was obtained by the GM. Finally, we obtained a large-area SHG image with a size of 3.2 mm × 3.2 mm by stitching together 64 SHG images in a matrix of eight rows and eight columns. The total image acquisition time was around 15 minutes for one sample. During the experiment for SHG imaging, the sample was immersed in the physiological saline solution to avoid tissue dehydration. Although the sample temperature was not controlled, we consider the assessment to be unaffected. There was no load on the tendon when SHG imaging was performed. For calculation of the mean SHG light intensity, we used the large-area SHG images without contrast enhancement. In addition, image analysis based on 2D Fourier transform was performed using the magnified SHG images without or with contrast enhancement. The presence or absence of contrast enhancement did not influence results because we used the shape of the 2D Fourier transform spectrum.

Fig. 1

Diagram showing the experimental setup of an inverted SHG microscope (HWP, half-wave plate; P, Polariser; QWP, quarter-wave plate; GM, galvanometer mirror; RL1 and RL2, relay lenses; HS, harmonic separator; OL, objective lens; BPF, optical band-bass filter; PMT, photon-counting photomultiplier).

Tensile testing

Tensile testing was undertaken immediately in order to avoid tissue degradation. We used a commercial tensile testing machine (EZ - S, Shimadzu Corp., Kyoto, Japan, load capacity 500 N, load cell precision ±1 %) for evaluating mechanical strength in the healing tendon. A single excised tendon fascicle was selected and the cross-section of the fascicle was estimated from the major and minor axes in the elliptically shaped cross-section of the fascicle. The tendon fascicle was then stretched at a tensile rate of 2 cm/min. Young’s modulus was calculated from the stress–strain curve, in which we used the nominal stress and assumed that the cross-section did not change during the tensile test.

Statistical analysis

The data for mean SHG light intensity were reported as mean and standard deviation (sd). P-values < 0.001 and < 0.05 represented statistical significance and were used for statistical tests of mean SHG intensity and Young’s modulus, respectively. Two group comparisons were made using Student’s t-test. For the correlation between the mean SHG intensity and Young’s modulus, the coefficient of determination (R²) was used.

SHG imaging

Since the SHG image indicated the highest intensity and the best contrast at a depth of 100 µm from the sample surface, we selected this depth for comparison of the collagen structure. First, we performed SHG imaging on the control samples. Figure 2a shows a comparison of the large-area SHG images without contrast enhancement (image size 3.2 mm × 3.2 mm, pixel size 2048 pixels × 2048 pixels) for three out of eight control samples (control (A), (B), and (C)) as representatives, although we performed the SHG imaging for all samples. The distribution of strong SHG light intensity was observed for all of them. Furthermore, thick collagen fibres were uniformly orientated parallel to the long axis of the tendon (vertical direction). When the region of the control (A) in Figure 2a (orange box) was magnified, as shown in Figure 2b (image size 400 µm × 400 µm, pixel size 256 pixels × 256 pixels), a crimp structure of thick collagen fibres was observed. Since these characteristics are consistent with the histological views of tendon tissue,^5,6 we can conclude that SHG imaging provides an accurate histological visualisation of collagen fibres in tendon without the need for histological sectioning and staining.

Fig.

a) Comparison of the large-area second-harmonic-generation images without contrast enhancement (image size 3.2 mm × 3.2 mm, pixel size 2048 pixels by 2048 pixels) for three out of eight control samples (control (A), (B), and (C)); b) magnified image of the region of the control (orange box) (A) in (a). Image size is 400 µm × 400 µm, composed of 256 pixels × 256 pixels.

Next, we performed SHG imaging on the healing tendon samples. The disrupted portion had reconnected four weeks post-operatively. After eliminating scar tissue around the disrupted portion using a scalpel, we visualised a collagen structure of the healing tendon tissue at a depth of 100 µm from the sample surface. Figure 3 shows a comparison of the large-area SHG images without contrast enhancement (left column, image size 3.2 mm × 3.2 mm, pixel size 2048 pixels × 2048 pixels), the large-area SHG images with contrast enhancement (centre column, the same size as the left column), and the magnified SHG images with contrast enhancement (right column, image size 400 µm × 400 µm, pixel size 256 pixels × 256 pixels, corresponding to the region in the centre column orange box) among six out of eight healing samples (healing (A), (B), (C), (D), (E), and (F)) as representatives. From a comparison of SHG images without contrast enhancement between the control and the healing samples (Fig. 2a and the left column in Fig. 3), the SHG light intensity considerably decreased in all of the healing samples. The contrast-enhanced SHG images in the centre column and their magnified SHG images in the right column enabled visualisation of the fibrous structure of the collagen in the healing area more clearly. In contrast to the control sample, the healing samples did not show clear crimp structures of thick collagen fibres. In addition, we observed that the collagen fibres in the healing samples were thin, somewhat oriented, and/or unevenly distributed.

Fig. 3

Comparison of the large-area second-harmonic-generation (SHG) images without contrast enhancement (left column, image size 3.2 mm × 3.2 mm, pixel size 2048 pixels × 2048 pixels), the large-area SHG images with contrast enhancement (centre column, image size 3.2 mm × 3.2 mm, pixel size 2048 pixels × 2048 pixels), and the magnified SHG images with contrast enhancement (right column, image size 400 µm × 400 µm, pixel size 256 pixels × 256 pixels, corresponding to the region in the centre column; orange box) among six out of eight healing samples (healing (A), (B), (C), (D), (E), and (F)).

The regenerated tissue in healing tendon mainly contains type-1 and type-3 collagen, whereas native tissue in the control tendon mainly contains that of type-1 collagen only.¹⁵ Although type-3 collagen has less higher-order fibre structure than type-1 collagen,¹⁶ there is no significant difference in SHG light efficiency between type-1 and type-3 collagen.¹⁷ We consider the SHG light intensity closely related with the degree of structural maturity, rather than the difference in type of collagen. The disorganised, less-oriented, thin collagen fibre structure in the healing tendon decreases the mechanical strength and SHG light intensity, whereas the well-organised, better-oriented, thick collagen fibre structure in the control tendon enhances mechanical strength and SHG light intensity.

We calculated the mean SHG light intensity of all pixels in the large-area SHG image without the contrast enhancement (see Fig. 2a and the left column in Fig. 3) for each sample to compare the SHG light intensity of the healing samples with that of the control samples quantitatively. Figure 4 shows a comparison of the mean SHG light intensity between the control samples (n = 8) and the healing samples (n = 8). The mean SHG light intensity was 201 (sd 37) counts for the control samples and 36 (sd 21) counts for the healing samples, respectively. The mean SHG light intensity in the healing samples was 17.9% of that in the control samples. These differences were statistically significant (p < 0.001) and may indicate that the collagen fibres were still immature even though the repaired portion had apparently united (shown through gross visualisation).

Fig. 4

Comparison of mean second-harmonic-generation intensity between the control (n = 8) and the healing samples (n = 8). Error bar shows the standard deviation of the data. *p < 0.001 (Student’s t-test).

Tensile testing

Immediately after the SHG imaging, tensile testing was performed on the same samples. Figure 5 shows the stress–strain curves for controls (A), (B), and (C) as representatives, although we performed the tensile testing for all samples. The curves of all the control samples presented similar profiles. The strain nonlinearly increased at low stress, and then linearly increased with the stress, indicating elastic deformation. The appearance of the peak after the linear region did not indicate the yield point but indicated slipping of the sample from the mechanical clamp. As a result, we did not include the other mechanical properties such as ultimate strength in this study.

Fig. 5

Stress–strain curves of controls (A), (B), and (C). Orange lines show the result of the linear fitting to the linear increasing region in the stress-strain curve.

Next, we performed tensile testing on the healing samples. Figure 6 shows the stress–strain curves for healing (A), (B), and (C), respectively. Unlike the curves of the control samples, the linear region of the healing samples was considerably less steep than that of the control samples.

Fig. 6

Stress-strain curves of healing (A), (B), and (C). Orange lines show the result of the linear fitting to the linear increasing region in the stress-strain curve.

Young’s modulus was calculated by applying linear fitting to the linear region of the stress–strain curve, as shown by the orange lines in Figures 5 and 6. Figure 7 shows a comparison between the Young’s modulus of the control samples (n = 8) and the healing samples (n = 8). Young’s modulus was 0.63 MPa (sd 0.15) for the control samples and 0.36 MPa (sd 0.23) for the healing samples. Young’s modulus in the healing samples was 57.1 % of that in the control samples, and this difference was statistically significant (p < 0.05). These results indicated that mechanical recovery of the healing samples was still imperfect at four weeks post-operatively.

Fig. 7

Comparison of Young’s modulus between the control samples (n = 8) and the healing samples (n = 8). Error bar shows the standard deviation of the data. *p < 0.05 (Student’s t-test).

Correlation between mean SHG light intensity and Young’s modulus

As both the mean SHG light intensity and Young’s modulus were obtained from the same sample, we calculated the correlation between the mean SHG light intensity and Young’s modulus. Figure 8 shows the relation between the two sets of data, in which square and circle plots, respectively, indicate the data for eight control samples and eight healing samples. The control samples were distributed in the area of high SHG intensity and large Young’s modulus, whereas the healing samples indicated low SHG intensity, with some variance in Young’s modulus. From the data plots in Figure 8, we determined the coefficient of determination (R²) to be 0.37 for both control and healing samples (see line in Figure 8).

Fig. 8

Correlation between mean second-harmonic-generation light intensity and Young’s modulus. Square plots show the control samples (n = 8), whereas circle plots show the healing sample (n = 8). Line shows the result of linear fitting to the plots of both control and healing samples, respectively.