Introduction

Vascular grafts (VG) prepared from expanded polytetrafluoroethylene or poly(ethylene terephthalate) (Dacron) have been used clinically for the revascularization of blood vessels with inner diameters ⩾6 mm. However, the replacement of smaller vessels (⩽5 mm) with these synthetic materials leads to the failure of the grafts, mainly because of early thrombosis formation.1, 2 In addition, the synthetic materials that are commonly used lack growth potential, and long-term results have revealed several material-related failures; for example, stenosis, thromboembolization, calcium deposition and infection.2 To overcome these limitations, various tissue-engineered VG have been developed, including poly-L-lactic acid, polyglycolic acid, polyurethane (PU), poly-e-caprolactone, cellulose, chitosan, poly(vinylalcohol) and their composites.1, 3, 4, 5, 6, 7, 8 However, at present, no materials satisfy the strict requirements of small-diameter grafts.

Silk fibroin (SF) from Bombyx mori is one of the candidates for small-diameter artificial grafts because the high strength and toughness of SF fiber have led to a long history of use in sutures.9 In our previous studies,10, 11, 12 SF grafts proved to be effective, as VG implanted in rat models. Owing to the success of these in vivo experiments, it has become important to obtain information on the state and mobility of water molecules around and inside of the graft with respect to thrombosis.

NMR imaging (MRI)13, 14 is a useful technique for non-destructively obtaining μm-scale spatial information about probe molecules, such as water in bulk matter. For this reason, MRI has been successfully applied to polymer gel systems.15, 16, 17, 18, 19, 20 The contrast of the resulting images is produced by the difference in the 1H spin density, 1H spin–spin relaxation time (T2) and diffusion coefficient (D). To develop an image, 1H MRI non-destructively applies information about the number of water molecules in the VG from the 1H spin density image and the mobility of water molecules in the VG from the 1H T2 and diffusion coefficient (D) images.

In this study, 1H NMR imaging was used to observe the spatial distribution of the 1H density, T2 and the diffusion coefficient (D) of water molecules in small-diameter silk fibroin VG coated with different materials (SF and PU) in water, and differences in the amount and mobility of water in the layers of the grafts were identified.

Experimental Procedure

Preparation of the VG from silk fibers

Raw silk fibers with silk sericin were used as the starting materials. The 4.0-mm diameter silk VG tube was prepared by a computer-controlled double-raschel knitting machine. The double-raschel-knitted silk tubes were degummed with a solution of sodium carbonate (0.1% w/v) and Marseille soap (0.2% w/v) at 95 °C for 120 min to remove silk sericin from the surface of the SF fiber. A Teflon rod was inserted into the SF graft tube so that the surface of the rod was covered with the SF fiber. The exteriors of the tubes were then coated with a SF aqueous solution or a PU solution. The outer-coated tubes were soaked in liquid N2 and frozen at −80 °C for 1 h. Then, the tubes were vacuum dried overnight. This outer-coating process was repeated two times, after which the tubes were kept in an autoclave at 120 °C. After autoclaving, the tubes were soaked in water, the Teflon rod was withdrawn and the outer-coated silk tubes were dried at 50 °C. Next, the interiors of the tubes were soaked in SF, PU and SF+PU complex solutions, coating them in the same manner as the outer-coating. This outer and inner coating method is summarized in Figure 1.

Figure 1
figure 1

Typical coating method for VG.

Full size image

We prepared several types of VG according to this coating method, as follows:

  1. 1)

    VG-S/S: Both sides (interior and exterior of the graft) were coated with SF.

  2. 2)

    VG-U/S: The exterior was coated with PU and the interior was coated with SF.

  3. 3)

    VG-U/SU: The exterior was coated with PU and the interior was coated with SF and PU (1:1).

  4. 4)

    VG-U/U: Both sides were coated with PU.

The final wall thickness of these grafts was 0.5 mm.

The cell-adhesive character was examined by cultivating human umbilical vein endothelial cells on plates coated with SF and PU, as shown in Figure 2. In this test, the surface of a 1.91-cm2 well in a 24-well tissue culture plate was coated with SF and PU (0.47 mg per well). After drying overnight, 70% ethanol was applied to the dishes to insolubilize and sterilize the SF. Then, the ethanol was removed and the coated dishes were washed twice with phosphate-buffered saline. Before seeding the cells, the dishes were incubated at 37 °C in 5% CO2. Then, 2 × 104 cells per well of human umbilical vein endothelial cells were added to each well. After a 3-h incubation in full growth medium (EBM-2) at 37 °C in a 5% CO2 atmosphere, the unattached cells were removed by gently rinsing the wells twice. The cells were stained with calcein-AM stock solution (Dojindo Co., Kumamoto, Japan) and counted with a fluorescence microscope with the appropriate filters. High adhesive character was observed for the SF coating, but essentially no cell-adhesive character was observed for the PU coating (Figure 2).

Figure 2
figure 2

The cell-adhesive character of human umbilical vein endothelial cells on silk fibroin- or PU-coated plates.

Full size image

MRI measurement

1H NMR imaging measurements were carried out in a Bruker BioSpin Avance DSX 300 NMR spectrometer (Bruker BioSpin, Rheinstetten, Germany) with an imaging system operating at 300 MHz and 290 or 300 K. For the acquisition and processing of the NMR data, the ParaVision program supplied by Bruker Biospin was used. Two sample configurations were prepared. In the first, the VG were set in a 5-mm NMR tube (with an internal diameter of 4 mm), the exteriors of the VG were in direct contact with the glass tube and pure water was added to fill only the interior of the VG (inside-water) (Figure 3a). In the second configuration, the VG were set in an 8-mm NMR tube (with an internal diameter of 7 mm) and both the interiors and exteriors of the VG were filled with pure water (both-side water) (Figure 3b).

Figure 3
figure 3

MR imaging sample tube of the VG in water used in this study.

Full size image

The 1H spin density, 1H spin–spin relaxation time (T2) and diffusion coefficient (D) images were constructed with the NMR2CALC program supplied by LSIC Co. Ltd, Kawasaki, Japan.

Results and Discussion

Spin-echo MR images of silk VG in inside-water case

Figure 4 shows spin-echo 1H images of the silk VG, in which (a) shows VG-S/S and (b) shows VG-U/S. The echo time is 13.4 ms and the repetition time is 1 s. The data points constituted a 512 × 360 array for an image and the field of view (FOV) was 1.11 cm. The slice thickness was 0.5 mm. The 1H signals in this experiment were mainly produced by the 1H in the water, because the T2 value of the SF is very short. The images in Figures 4a and b , which show the VG-S/S and VG-U/S samples that were inner-coated with SF, respectively, indicate that the brightness of the VG-S/S layer is greater than that of the VG-U/S layer. This result means that the VG-S/S layer contains a greater number of water molecules than the VG-U/S layer. This result shows that the outer-coating material also influences the amount of water in the VG.

Figure 4
figure 4

Spin-echo 1H images of silk VG, which are (a) VG-S/S and (b) VG-U/S. The echo time was 13.4 ms and the repetition time was 1 s. The data points numbered 512 × 360 for an image and the FOV was 1.11 cm. The slice thickness was 0.5 mm.

Full size image

The 1H spin density and T2 images of the silk VG the inside-water case constructed from several spin-echo 1H NMR images

To perform a more quantitative analysis of the 1H MRI data, we constructed the 1H spin density (S0) and T2 images of the silk VG from the various echo times of the spin-echo 1H NMR images. The echo times (τ) were 10.97, 21.94, 32.91, 43.88, 54.85, 65.82, 76.79, 87.76, 98.73 and 109.7 ms, and the repetition time was 1 s. The data points constituted a 256 × 256 array in an image and the FOV was 1.10 cm. The slice thickness was 0.5 mm. We used very long repetition time for T1; consequently, we neglected the T1 effect in the images. The signal intensity (Sτ) can be described by the following expression:

From this equation, the 1H spin density (S0) and T2 can be calculated from various echo time (τ) experiments.

Figure 5 shows the 1H spin density (S0) images of the silk VG, which are (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U. The 1H spin density in the VG-S/S layer is clearly greater than that of the VG-U/X layer (where X=S, SU and U). This result means that the amount of water in the VG-S/S layer is larger than that in the VG-U/X layer. From the SEM images, the outer coating material can be seen to permeate the silk graft layer.12 As SF is a hydrophilic material and PU is a hydrophobic material, the amount of water in the VG-S/S layer is greater than that in the VG-U/X layer.

Figure 5
figure 5

1H spin density images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U in the inside-water case.

Full size image

Figure 6 shows the T2 images of the silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U. The 1H T2 of bulk water is approximately 45 ms in Figure 6. The T2 images of VG-U/X (X=S, SU and U) show a T2 data defect area (black area), which means that the area contained little or no water, and the initial intensity of the 1H signal was very low. The data defect area increased as the content of PU in the coating increased. The T2 of the VG-S/S was 30–37.5 ms, and that of VG-U/X (X=S, SU and U) was 35–45 ms. Thus, the rotational mobility of water in VG-U/X is slightly greater than that of VG-S/S because T2 is associated with the rotational motion of water molecules. It is known that the intermolecular interaction between SF and water occurs through the formation of hydrogen bonds between the amide or carbonyl groups of SF molecules and water molecules, but that there is no intermolecular interaction between PU and water molecules. The rotational mobility of water molecules in a layer of VG-U/S is approximately the same as that of bulk water, but the rotational mobility of water molecules in a VG-S/S layer is restricted by the intermolecular interaction between the SF and water. These results show that the coating material affects both the amount and mobility of water in the layers of the VG.

Figure 6
figure 6

T2 images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U in the inside-water case.

Full size image

The diffusion coefficient images of the silk VG in the inside-water case are constructed from several pulse-field gradient spin-echo 1H NMR images

T2 is associated with the rotational motion of water molecules, but the diffusion coefficient (D) is associated with the center of gravity translation of water molecules. These two parameters indicate the different types of dynamics available to water molecules.

Figure 7 shows the diffusion coefficient (D) images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U, that are constructed from several pulse-field gradient spin-echo 1H NMR images with a field gradient pulse length of δ=10 ms, a gradient pulse separation of Δ=25 ms and field gradient strengths of G=0, 20, 40, 60, 80 and 100 mT m−1. The data points constitute a 128 × 128 array for an image and the FOV is 1.10 cm. The slice thickness is 0.5 mm.

Figure 7
figure 7

Diffusion coefficient (D) images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U in the inside-water case.

Full size image

The relationship between the echo signal intensity SG and the pulse-field gradient parameters in the pulse-field gradient spin-echo method is given by the following expression:

where S0 is the signal intensity in the absence of gradient pulses, γ is the 1H magnetogyric ratio and D is the diffusion coefficient. The plots of ln(SG/S0) against γ2G2δ2(Δ−δ/3) produce a straight line with a slope of –D. Therefore, the value D can be determined from this slope.

The value of D for bulk water is approximately 1.6 × 10−9 m2 s−1 at 290 K in Figure 7. The value of D in the VG-S/S layer is 0.9–1.2 × 10−9 m2 s−1 and it has a value of 0.4–0.5 × 10−9 m2 s−1 in VG-U/X (X=S, SU and U). Thus, the value of D in the VG-U/X layer is smaller than that in the VG-S/S layer. These results are inconsistent with the T2 results. However, as we mentioned before, T2 is associated with the rotational motion of water molecules, and the diffusion coefficient (D) is associated with the center of gravity translation of water molecules. These two parameters describe the different types of dynamics of water molecules. The T2 experiment shows that there is no intermolecular interaction between the PU and the water molecules. The smaller D value of the water in VG than that of bulk water shows that the water in the VG diffuses through pores. The D image shows that the water molecules are confined to pores, and the restricted diffusion21 of the water molecules occurs in the graft layer. The size of the pores in the VG-U/X is smaller than the pore size in VG-S/S.

1H spin density and T2 images of silk VG in both-side water case constructed from several spin-echo 1H NMR images

In the previous experiment, we filled only the inside of the VG with pure water. However, in the in vivo case, the outside of the grafts was also wet. We prepared both the interior and exterior of the grafts by filling with pure water (both-side water; Figure 3b). The 1H spin density (S0) and T2 images of the silk VG were constructed from the various echo time spin-echo 1H NMR images. The echo times (τ) were 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150 and 160 ms, and the repetition time was 1 s. The data points constituted a 256 × 256 array for an image and the FOV was 1.10 cm. The slice thickness was 0.5 mm.

Figure 8 shows the 1H spin density (S0) images of the silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U. The 1H spin density in the VG-S/S layer is clearly higher than that of the VG-U/X (X=S, SU, U) layer, and the amount of water in the VG-S/S layer is greater than that of the VG-U/X. This result is the same as that of the inside-water case. As SF is a hydrophilic material and PU is a hydrophobic material, the amount of water in the VG-S/S layer is larger than that of the VG-U/X layer.

Figure 8
figure 8

1H spin density images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U in the both-side water case.

Full size image

Figure 9 shows T2 images of the silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U. The 1H T2 of bulk water in this experiment is approximately 45 ms. The T2 of the VG-S/S layer is 30–37.5 ms, and that of VG-UX (X=S, SU and U) is 35–45 ms. The larger T2 area increases in size as the PU content of the coating increases. This trend means that the mobility of water in VG-U/X is higher than its mobility in VG-S/S. This result is the same as that of the inside-water case. It is known that intermolecular interactions exist between SF and water (an effect of the hydrogen bond between the amide or carbonyl groups of SF molecules and water molecules), but there is no intermolecular interaction between PU and water. The rotational mobility of water molecules in the VG-U/S layer is nearly equal to that of bulk water, but the mobility of the water molecules in the VG-S/S layer is restricted by the intermolecular interaction between the SF and water. These results show that the coating material affects the amount and mobility of water in the VG layers.

Figure 9
figure 9

T2 images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U in the both-side water case.

Full size image

The diffusion coefficient images of the silk VG in the both-side water case were constructed from several pulse-field gradient spin-echo 1H NMR images

Figure 10 shows the diffusion coefficient (D) images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U, that were constructed from several pulse-field gradient spin-echo 1H NMR images with a field gradient pulse length of δ=5 ms, a gradient pulse separation of Δ=10 ms, and field gradient strengths of G=0, 30, 60, 90, 120, 150, 180, 210, 240 and 270 mT m−1. The data points constituted a 128 × 128 array for an image and the FOV was 1.10 cm. The slice thickness was 0.5 mm.

Figure 10
figure 10

Diffusion coefficient (D) images of silk VG, (a) VG-S/S, (b) VG-U/S, (c) VG-U/SU and (d) VG-U/U in the both-side water case.

Full size image

The D value of bulk water is approximately 2.0 × 10−9 m2 s−1 at 300 K in Figure 10. The D value in the VG-S/S layer is 1.5–1.9 × 10−9 m2 s−1, and the value in VG-U/X (X=S, SU and U) is 0.7–1.3 × 10−9 m2 s−1. The D value in the VG-U/X layer is significantly smaller than that in the VG-S/S layer. The D imaging results show that the water molecules are confined in a closed porous medium, and restricted diffusion occurs in the graft layers. The pore size in VG-U/X is smaller than that in VG-S/S.

Conclusions

The spatial distribution of the 1H spin density T2 and the diffusion coefficient (D) of water molecules in the layers of small-diameter silk fibroin VG coated with different materials (SF and PU) in water were successfully identified by 1H NMR imaging. The water content of the SF-coated VG was higher than that of the PU-coated VG. The T2 of the water molecules in the SF-coated VG layer was smaller than that of the PU-coated VG, which means the rotational motion of the water in the SF-coated VG layer was restricted by the intermolecular interaction between the SF and the water molecules. The diffusion coefficient (D) of the water molecules in the PU-coated VG layer was smaller than that of the SF-coated VG, and the water molecules were confined to the pores in the graft layer.