BFGF-induced proliferation and differentiation of MSCs or cartilage precursor cells in ear cartilage in vivo model
Representative findings from the evaluation of the in vivo rabbit model with a single bFGF treatment are shown in Fig. 1. One to 3 days after the injection of 10 μg of bFGF (100 μg/ml), infiltration of mononuclear cells, comprising mostly macrophages, and an expansion in the volume of the perichondrium layer in the superficial perichondrium and the subcutaneous layer near the perichondrium were observed. One week after the injection, proliferation of perichondrium cells with spindle and spider shapes, and extracellular matrix with type 1 collagen fibers were observed. Two weeks after the injection, the superficial perichondrium layer and the deep neo-chondrium layer with neocartilage cells were differentiated from pericartilage cells. Additionally, by 2 weeks, type 2 collagen fibers in the extracellular matrix were observed. From one to 3 months after the injection, neocartilage cells with hypertrophy of the cytoplasm and abundant extracellular matrix were present. The new cartilage was morphologically similar to mature tissue.
The average thickness of the cartilage stained with an anti-type 2 collagen antibody (Fig. 1) was 171 ± 26 μm for the control, 177 ± 35 μm at 1 day, 206 ± 30 μm at 3 days, 213 ± 58 μm at 1 week, 703 ± 100 μm at 2 weeks, 835 ± 84 μm at 1 month (not shown in Fig. 1), and 933 ± 18.8 μm at 3 months. The differences in the thickness of cartilage between 1 week and 2 weeks were statistically significant (P < 0.0001).
CD31-positive cells were seldom observed around the perichondrium layer in the noninjected group (Fig. 2). Mononuclear cells and small vascular endothelial cells were observed on day 3 following the administration of bFGF. During the first week, proliferation of vascular endothelial cells was noted and the vessels extended vertically into the perichondrial region. Vascular endothelial cells were not observed in the deep layer of perichondrium, but were located in the superficial layer during week 2.
As shown in Fig. 2, a small number of MMP-1-positive cells in the perichondrium was observed in the noninjected control group. During day 3, a large number of mononuclear cells and pericartilage cells stained positive for MMP-1 in the perichondrium and the outer layer of perichondrium. Only mononuclear cells stained positive for MMP-1 in the superficial layer of the perichondrium during the first week following bFGF treatment.
CD44 and CD90 are markers for MSCs of auricular cartilage, whereas CD34 is not and served as a negative control. The protein Ki-67 is a cellular marker for proliferation.
As shown in Fig. 3, a small number of CD44-positive and CD90-positive cells was observed in the noninjected group, but a large number of CD44-positive and CD90-positive cells was observed 1 week after the injection of bFGF. Almost all the observed spindle- and spider-shaped cells in the perichondrium layer stained positive for CD44 and CD90. These positively stained cells decreased in number 2 weeks after the injection. No CD34-positive cells were detected in any of the experimental groups. Ki-67-positive cells were scarcely observed in noninjected controls at 1 day after the injection. During week 1, Ki67-positive perichondrial cells were observed in the outer region of the perichondrial inner-layer/outer-layer boundary. In the superficial layer of perichondrium during week 2, only perichondrial cells stained positive for Ki-67.
Immunohistochemical staining with the anti-MMP-1 antibody was performed to evaluate type 1 collagenolytic activity. The average number of positive cells per microscopic visual field was 2.4 ± 2.0 for the control, 12.0 ± 3.3 at 1 day, 40.1 ± 14.4 at 3 days, 18.2 ± 4.1 at 1 week, 13.5 ± 2.7 at 2 weeks, 3.6 ± 2.1 at 1 month, and 1.7 ± 1.6 at 3 months. The differences in the number of MMP-1-positive cells from the control to 1 day, 1 day to 3 days, and 3 days to 1 week were statistically significant (P < 0.0001) (Fig. 4a). Immunohistochemical staining with the anti-CD31 antibody was performed to evaluate angiogenesis. The average number of positive cells per visual field was 2.0 ± 0.8 for the control, 2.9 ± 1.3 at 1 day, 12.3 ± 4.6 at 3 days, 35.3 ± 8.2 at 1 week, 17.8 ± 3.6 at 2 weeks, 4.3 ± 2.7 at 1 month, and 1.2 ± 0.7 at 3 months. The differences in the number of CD31-positive cells from 1 day to 3 days, 3 days to 1 week, and 1 week to 2 weeks were statistically significant (P < 0.0001) (Fig. 4b). Immunohistochemical staining with the anti-Ki67 antibody was performed to evaluate proliferation. The average number of positive cells per visual field was 0.1 ± 0.3 for the control, 0.6 ± 0.7 at 1 day, 1.4 ± 0.8 at 3 days, 35.3 ± 8.3 at 1 week, 10.4 ± 3.3 at 2 weeks, 0.2 ± 0.4 at 1 month, and 0.1 ± 0.4 at 3 months. The differences in the number of Ki67-positive cells from 3 days to 1 week, and 1 week to 2 weeks were statistically significant (P < 0.0001) (Fig. 4c). Immunohistochemical staining using the anti-CD90 antibody was conducted to evaluate MSCs. The average numbers of positive cells were 1.6 ± 0.8 for the control, 1.8 ± 1.1 at 1 day, 2.8 ± 1.0 at 3 days, 35.5 ± 7.7 at 1 week, 10.4 ± 3.3 at 2 weeks, 2.4 ± 1.2 at 1 month, and 2.3 ± 1.0 at 3 months. The differences in the number of CD90-positive cells from 3 days to 1 week, and from 1 week to 2 weeks were statistically significant (P < 0.0001) (Fig. 4d).
A single administration of bFGF to the perichondrium resulted in the proliferation of precursor cells, and cartilage tissue formation with prolonged maintenance of cartilage morphology for at least 3 months. Inflammatory cells comprising mononuclear cells migrated into the treatment site by day 3 following administration of bFGF; moreover, by this time the proliferation of MMP1-positive cells peaked. During week 1, thickening of the perichondrium occurred and proliferation of vascular endothelial cells in the perichondrial region was observed. The bFGF treatment stimulated CD44-positive and CD90-positive cartilage MSCs or progenitor cells in the perichondrium to proliferate. Subsequently, neocartilage was formed after 2 weeks, and after 3 months hypertrophied mature cartilage was formed and maintained.
It has been reported that bFGF is a growth factor for fibroblasts present in bovine-brain extract [8]; it promotes the proliferation, differentiation, and migration of various cells and is a growth factor with strong angiogenic action [6]. The growth-promoting effect of administration of bFGF on mature chondrocytes has been previously reported, but the effects on perichondrial tissue or perichondrial cells have not been elucidated. Moreover, it is unknown which cells in the perichondrium are involved in proliferation and differentiation into cartilage [9, 10]. The bFGF-stimulated cartilage-proliferation rabbit model in this study showed that perichondrial cells, rather than chondrocytes, were more likely to proliferate and differentiate into cartilage. Based on the proliferation activity of perichondrial cells analyzed by Ki67 labeling, it was suggested that the perichondrial cells in the outer layer of the perichondrial inner-layer/outer-layer boundary may have been involved in the observed proliferation. The perichondrium is composed of two layers: an outer layer in which small fibrocyte-like cells are interspersed between histologically sparse collagenous fibers, and an inner layer in which somewhat rounded cells in compact fibers have an irregular layer structure composed of three to four layers [9]. Kobayashi et al. demonstrated by immunostaining for the auricular-perichondrial MSC markers CD44 and CD90 in an in vitro study using human auricular perichondrial cells that in the inactive perichondrium, MSCs exist as elongated spindle-shaped cells located at the inner layer-layer/outer-layer boundary [2]. It has been shown that CD44-positive and CD90-positive MSCs in the perichondrial inner-layer/outer-layer boundary are activated by bFGF, become active chondrocyte precursor cells with short spindle shapes to star-like shapes, and participate in proliferation and chondrocyte differentiation [2]. It has also been shown that MSCs in the proliferated cartilage membrane are able to maintain long-term morphology by reforming the perichondrial membrane, with one of the roles of MSCs in the regeneration process of the perichondrium having been elucidated in this in vivo study.
Verifying the effect of chondrogenesis following bFGF administration
The effect of bFGF concentration on auricular chondrogenesis was shown in Fig. 5a. The average neogenesis cartilage rates were 0.046 ± 0.025 (control with no bFGF), 0.040 ± 0.020 (1 μg/ml bFGF), 0.28 ± 0.064 (5 μg/ml), 0.38 ± 0.11 (10 μg/ml), 0.70 ± 0.06 (25 μg/ml), 0.71 ± 0.061 (50 μg/ml), and 0.78 ± 0.07 (100 μg/ml). Kruskal-Wallis analysis revealed a highly significant difference among the groups (P < 0.0001). Post hoc comparisons by Fisher test indicated significant differences between the treatment groups that received more than 5 μg/ml of bFGF and the control group. However, the difference between the group that received 1 μg/ml bFGF and the control group was not statistically significant (P = 0.51).
The blockage of bFGF-induced chondrogenesis with an anti-bFGF neutralizing antibody was shown in Fig. 5b. The average cartilage neogenesis rates based on the ratio of new cartilage to total cartilage were 0.061 ± 0.024 in G1 (anti-bFGF neutralizing antibody injected immediately after bFGF injection), 0.055 ± 0.021 in G2 (anti-bFGF neutralizing antibody injected immediately after bFGF injection and at 1 week post injection), 0.046 ± 0.019 in G3 (anti-bFGF neutralizing antibody injected immediately after bFGF injection and at 2 weeks post injection), 0.44 ± 0.13 in G4 (anti-bFGF neutralizing antibody injected at 1 week post-bFGF injection), 0.33 ± 0.07 in G5 (anti-bFGF neutralizing antibody injected at 2 weeks post-bFGF injection), and 0.27 ± 0.06 for controls not receiving any anti-bFGF neutralizing antibody injections after the bFGF injection. Kruskal-Wallis analysis revealed a significant difference among the three groups immediately after injection of bFGF (G1–3) and the control group (P < 0.01), but suppression of chondrogenesis at 1 week and 2 weeks after bFGF injection was not recognized in comparison with the control group. The difference in the neogenesis rates between G1, G2, and G3, and between G4 and G5 were not significant.
The results from the current study confirmed that bFGF-induced chondrogenesis was concentration-dependent, with the highest concentration tested of 100 μg/ml yielding the highest levels of cartilage formation. It has been reported in vitro that the bFGF concentration at which chondrocytes show proliferation-promoting activity is 10 ng/ml [11, 12]. In our study, the concentration of bFGF that induced the highest degree of chondrogenesis was the same as that used in clinical practice for intractable ulcers and burn ulcers, and appeared to be a reasonable concentration for cell stimulation in vivo. The results from the experiments using anti-bFGF neutralizing antibodies revealed that chondrogenesis was inhibited in the groups that received the neutralizing antibody immediately after bFGF treatment, whereas in the group that received the neutralizing antibody at 1 week and 2 weeks post bFGF-injection, cartilage formation was not inhibited. Therefore, the process by which injected bFGF binds to and releases proteoglycans in vivo and the possibility that bFGF from cells in the perichondrial region is continuously produced is unlikely. The concentration-dependent bFGF-induced cartilage proliferation and suppression of cartilage proliferation by anti-bFGF neutralizing antibodies suggests that the stimulation of perichondrial cells by the injection of high concentrations of bFGF may have turned on a switch that promoted sustained proliferation of tissue stem cells or perichondrial progenitor cells present in the perichondrium, as well as their differentiation into chondrocytes. Nonetheless, it is unlikely that the initial time lag for significant proliferation of CD44-positive and CD90-positive MSCs or cartilage precursor cells of the perichondrium that was observed after bFGF stimulation was due to the direct growth-stimulating effect of bFGF on MSCs. We speculated that this effect was caused by angiogenesis induced by bFGF. Therefore, we examined the effects of an MMP inhibitor that inhibits angiogenesis, and anti-VEGF neutralizing antibodies.
Verifying the inhibition of chondrogenesis through the inhibition of angiogenesis
The CD31 positive cells, the perichondrium and neocartilage thickness by administration of MMP inhabitation are shown in Fig. 6a. Immunohistochemical staining with the anti-CD31 antibody was performed to evaluate angiogenesis (Fig. 6a, graph panel a). The average numbers of positive cells per visual field in the control groups that received only an injection of the carrier DMSO were 19.8 ± 5.4 at 1 week and 20.8 ± 6.8 at 2 weeks, compared to 1.2 ± 0.8 at 1 week and 1.8 ± 1.6 at 2 weeks in the experimental groups that received the injections of a monoclonal antibody inhibiting MMP. The differences in the numbers of CD31-positive cells between the controls and the corresponding experimental groups were statistically significant (P < 0.001 at 1 week and < 0.001 at 2 weeks post treatment). The average perichondrium thickness per microscopic visual field in the control group was 127 ± 3 μm at 1 week, and 92 ± 26 μm at 2 weeks compared to 60 ± 17 μm at 1 week and 29 ± 15 μm at 2 weeks in the experimental group (Fig. 6a, graph panel b). The differences between the controls and the corresponding experimental groups were statistically significant (P < 0.01 at 1 week and < 0.01 at 2 weeks). The average neocartilage thickness per visual field in the control group was 20 ± 5 μm at 1 week and 183 ± 47 μm at 2 weeks, compared to 17 ± 10 μm at 1 week and 19 ± 9 μm at 2 weeks per field in the experimental groups (Fig. 6a, graph panel c). The differences between the controls and the experimental groups at 2 weeks post-treatment were statistically significant (P < 0.001). This differed from the results obtained at 1 week post-treatment, in which there was no significant difference between the control and experimental groups.
The CD31 positive cells, the perichondrium and neocartilage thickness by administration of VEGF neutralization are shown in Fig. 6b. The average number of positive cells per visual field in the control group that received injections of only the carrier DMSO was 20.0 ± 4.4 at 1 week and 20.6 ± 4.6 at 2 weeks post treatment, compared to 12.4 ± 4.6 at 1 week and 13 ± 3.4 at 2 weeks post treatment in the experimental groups (Fig. 6b, graph panel a). The differences in the numbers of CD31-positive cells between the controls and the corresponding experimental groups were statistically significant (P < 0.05 at 1 week and < 0.01 at 2 weeks post treatment). The average perichondrium thickness per visual field in the control groups was 152 ± 47 μm at 1 week and 110 ± 29 μm at 2 weeks post treatment, compared to 102 ± 25 μm at 1 week and 125 ± 39 μm at 2 weeks post treatment in the experimental groups (Fig. 6b, graph panel b). The differences between the control groups and the corresponding experimental groups at 1 week post-treatment were statistically significant (P < 0.05), but this did not continue, as no significant difference was detected at 2 weeks post treatment. The average neocartilage thickness per visual field in the control group was 28 ± 10 μm at 1 week and 163 ± 80 μm at 2 weeks post-treatment, whereas in the experimental groups it was 19 ± 6 μm at 1 week and 64 ± 25 μm at 2 weeks post treatment (Fig. 6b, graph panel c). The differences between the controls and the experimental groups at 2 weeks post-treatment were statistically significant (P < 0.001), but not at 1 week post-treatment where no significant difference was observed. Angiogenesis induced by bFGF was sensitive to an angiogenesis inhibitor, which suppressed bFGF-induced perichondrium proliferation and neocartilage formation.
The results from the MMP inhibitor and VEGF neutralization experiments showed that the proliferation of new blood vessels 1 week after bFGF stimulation and the proliferation of perichondrial cells 1 week after bFGF stimulation were both inhibited, similarly to cartilage formation 2 weeks after treatment. Therefore, it appears that new blood vessels were strongly involved in perichondrial proliferation and subsequent cartilage formation. Humanized anti-VEGF monoclonal antibody (common name, bevacizumab) utilizes VEGF as a target molecule and inhibits angiogenesis [13, 14]. It has also been reported that mononuclear cells express and secrete IL-1β, and induce VEGF genes of the vascular endothelium and fibroblasts [15]. Because mononuclear cells accumulated in the perichondrial region 1 day after the administration of bFGF, it is possible that bFGF induced early-stage mononuclear cell infiltration that stimulated VEGF, which in turn may have been involved in the observed cartilage proliferation. In addition, the MMP inhibitor batimastat (chemical name) inhibits MT1-MMP, MMP-2, MMP-9, and angiogenesis [16], apparently resulting in inhibition of cartilage proliferation. Additionally, the MMP inhibitor suppressed the enzymatic activity of MMP-1, which degrades type I collagen, the main extracellular-matrix component of the perichondrium [17]. Based on the results of our study, the reason for the observed strong inhibition of perichondrial proliferation and cartilage formation by the MMP inhibitor compared to the VEGF-neutralizing antibody may have been not only the inhibition of angiogenesis, but also the inhibition of invasion of MMP-1-positive cells on days 1 to 3 post bFGF-treatment. Takebe et al. found that early interactions with endothelial cells in establishing avascular tissues from human specific progenitors trigger the initial expansion of cartilage progenitor cells and promote the self-aggregation of a 3D condensation of progenitors without any scaffold materials in vitro, and the introduction of MSCs into immunodeficient mice results in angiogenesis within 3 days of grafting [4]. They also found that cartilage precursor cells proliferated from day 2 to 7 post-grafting, and that the grafted cells differentiated into chondrocytes from days 10 to 20 [4]. This is similar to the gradual changes observed in the bFGF-stimulated cartilage proliferation model analyzed in our study. They also reported that the onset of angiogenesis during the early stage of grafting is consistent with the timing of proliferation of MSCs. In fact, blocking of angiogenesis strongly inhibits the proliferation of cartilage progenitor cells and cartilage formation, and angiogenesis is essential for the proliferation of MSCs (cartilage precursor cells) derived from the perichondrium and their differentiation into chondrocytes [4]. In the current study, the timing of angiogenesis and that of proliferation of MSCs were consistent, whereas cartilage formation and the proliferation of the cartilage membrane were suppressed as a result of angiogenesis inhibition. Therefore, it appears that MSCs were activated in vivo by angiogenesis induced by the administration of bFGF and that the activated MSCs caused perichondrial proliferation and cartilage formation.