Animals
The experimental design is shown in Fig. S1 (Supporting Information). In brief, 147 rats were surgically prepared and intrathecally catheterized. Two rats displayed locomotor dysfunctions and had not recovered on POD 3. Therefore, 145 rats were used in the study.
Time-dependent changes of spinal VEGF protein levels in CCI rats
To determine whether angiogenesis occurs in the spinal cord, we evaluated VEGF expression by Western blot and immunohistochemistry. VEGF immunoreactivity was significantly upregulated in the ipsilateral dorsal horn of the spinal cord (SCDH) after CCI at POD 7, 14, 21, and 28 (22.4-fold, p < 0.001; 23.3-fold, p < 0.001; 15.1-fold, p = 0.0017; and 4.1-fol, p = 0.002; respectively), but not in the contralateral side (1.2-, 1.5-, 1.0-, and 1.1-folds, respectively; all p > 0.05), compared to VEGF staining in controls (POD 7 after sham operation; Fig. 1 A,B). In addition, VEGF immunoreactivity was also significantly upregulated in the ipsilateral SCDH after CCI at POD 7, 14, 21, and 28 (22.4- vs. 1.2-fold, p < 0.001; 23.3- vs. 1.5-fold, p < 0.001; 15.1- vs. 1-fold, p = 0.0015; and 4.1- vs. 1.1-fold, p < 0.001; respectively) compared to it in the contralateral side at the same time point (Fig. 1 A,B). Western blot analyses also revealed increased VEGF protein levels in the ipsilateral lumbar spinal cord dorsal part on POD 7, 14, and 28 (1.4-fold, p = 0.019; 2.7-fold, p = 0.0019; and 2.0-fold, p = 0.0243; respectively) but not on POD 3 (1.2-fold, p = 0.18) after CCI (Fig. 1 C,D) compared to that in control rats. There was no significant difference in VEGF protein content between the contralateral spinal cord of CCI rats at POD 3, 7, 14, and 28 (1.1-, 1.0-, 1.2-, and 0.9-folds, respectively; all p > 0.05) and the control group (Fig. 1 C,D). Similarly, VEGF proteins was also significantly upregulated in the ipsilateral SCDH after CCI at POD 7, 14, and 28 (1.4- vs. 0.9-fold, p = 0.0152; 2.7- vs. 1.2-fold, p =0.0029; and 2.0- vs. 0.9-fold, p = 0.0246; respectively), but not at POD 3 (1.2- vs. 1.1-fold, p = 0.684), compared to it in the contralateral side at the same time point (Fig. 1 C,D). These results demonstrate that CCI induces upregulation of VEGF expression in the ipsilateral spinal cord from POD 7 to 28 with a peak at POD 14.
CCI upregulates CD31 and vWF protein levels in the spinal cord in a similar time-dependent manner
CD31 immunoreactivity was markedly increased from POD 7 to 28 after CCI, with a peak at POD 14 in the ipsilateral SCDH compared to that in the control group (3.2-fold, p = 0.002; 5.2-fold, p = 0.002; 4.4-fold, p = 0.031; and 3.3-fold, p = 0.015 at POD 7, 14, 21, and 28, respectively), while it was not affected on the contralateral side (1.1-, 1.1-, 1.3-, and 1.3-folds; all p > 0.05) (Fig. 2). In addition, CD31 immunoreactivity was significantly upregulated in the ipsilateral SCDH after CCI at POD 7, 14, and 28 (3.2- vs. 1.1-fold, p < 0.001; 5.2- vs. 1.1-fold, p =0.0081; and 3.3- vs. 1.3-fold, p = 0.0299; respectively), but not at POD 21 (4.4- vs. 1.3-fold, p = 0.052), compared to it in the contralateral side at the same time point (Fig. 2).
As shown in Fig. 3, vWF immunoreactivity was significantly increased from POD 7 to 28 after CCI, with a peak at POD 14 in the ipsilateral spinal cord compared to that of controls (2.3-fold, p = 0.00125; 2.7-fold, p < 0.001; 2.4-fold, p < 0.001; and 1.7-fold, p < 0.001 at POD 7, 14, 21, and 28, respectively), while there was no difference on the contralateral side (1.1-, 1.2-, 1.2-, and 1.0-folds; all p > 0.05). Furthermore, vWF immunoreactivity was also significantly upregulated in the ipsilateral SCDH after CCI at POD 7, 14, 21, and 28 (2.3- vs. 1.1-fold, p < 0.001; 2.7- vs. 1.2-fold, p < 0.001; 2.4- vs. 1.2-fold, p = 0.023; and 1.7- vs. 1.0-fold, p = 0.0141; respectively) compared to it in the contralateral side at the same time point (Fig. 3). These data support the hypothesis that CCI activates angiogenesis in the ipsilateral lumbar spinal cord.
Neuroinflammation and enhances proinflammatory cytokine release in the ipsilateral lumbar spinal cord after CCI
In control rats, IL-1β concentration in ipsilateral lumbar spinal cord dorsal part was not significantly different to that in the contralateral side (118.2 ± 6.9 vs. 121.2 ± 9.8 pg/100 μg proteins, p = 0.807; Fig. 4A). However, IL-1β level was increased in the ipsilateral spinal cord at POD 7, 14, 21, and 28 after CCI compared to that in controls (149.9 ± 11.5, p = 0.0315; 154.8 ± 9.3, p = 0.00631; 146.8 ± 5.5, p = 0.00646; and 144.6 ± 3.3, p = 0.00493 respectively, vs. 118.2 ± 6.9 pg/100 μg proteins; Fig. 4A). There were no differences in IL-1β concentration between the contralateral spinal cord of CCI rats and the controls (119.4 ± 6.8, 122.3 ± 2.3, 122.0 ± 3.0, and 120.6 ± 7.7 at POD 7, 14, 21, and 28 , vs. 121.2 ± 9.8 pg/100 μg proteins, all p > 0.05; Fig. 4A). Furthermore, IL-1β level was significantly upregulated in the ipsilateral side after CCI at POD 7, 14, 21, and 28 (p = 0.037, 0.00384, 0.00102, and 0.0151, respectively) compared to it in the contralateral side at the same time point (Fig. 4A). Similarly, IL-6 concentrations were not significantly different between ipsilateral and contralateral sides of lumbar spinal cord dorsal part in control rats (107.3 ± 4.6 vs. 110.1 ± 3.8 pg/100 μg proteins, p = 0.652; Fig. 4B). IL-6 levels were significantly elevated after CCI in the ipsilateral spinal cord at POD 7, 14, and 21 compared to that in controls (140.8 ± 13.4, p = 0.0292 and 133.1 ± 10.5, p = 0.0394 respectively, vs. controls 107.3 ± 4.6 pg/100 μg proteins) but not at POD 7 and 28 (136.2 ± 13.0, p = 0.0526 and 111.2 ± 2.1 pg/100 μg protein, p = 0.457, respectively) (Fig. 4B). CCI had no effect on IL-6 concentration in the contralateral spinal cord at POD 7, 14, 21, and 28 compared to that of control rats (108.6 ± 3.3, 105.4 ± 3.0, 107.5 ± 3.5, and 106.8 ± 1.8, respectively, vs. 110.0 ± 3.8 pg/100 μg proteins, all p > 0.05; Fig. 4B). Additionally, IL-6 level was significantly upregulated in the ipsilateral side after CCI at POD 14 and 21 (p = 0.0187 and 0.0346 respectively), but not at POD 7 and 28 (p = 0.0567 and 0.136 respectively), compared to it in the contralateral side at the same time point (Fig. 4B). TNF-α concentrations were also not significantly different between ipsilateral and contralateral sides of lumbar spinal cord dorsal part in control rats (104.0 ± 4.0 vs. 104.7 ± 7.3 pg/100 μg proteins, p = 0.941; Fig. 4C). TNF-α levels in the ipsilateral spinal cord were significantly elevated at POD 7, 14, and 21 in CCI rats (135.8 ± 11.8, p = 0.0294; 138.6 ± 6.9, p < 0.001, and 116.3 ± 2.5, p = 0.018 respectively), but not at POD 28 (117.6 ± 5.0 pg/100 μg proteins, p = 0.0563), compared to that in controls (104.0 ± 4.0 pg/100 μg proteins; Fig. 4C). Similarly, TNF-α concentrations were not affected in the contralateral side after CCI at POD 7, 14, 21, and 28 (100.9 ± 5.6, 108.6 ± 5.6, 107.3 ± 5.0, and 117.3 ± 6.3, respectively, vs. controls 104.7 ± 7.3 pg/100 μg proteins, all p > 0.05; Fig. 4C). Furthermore, TNF-α level was significantly upregulated in the ipsilateral side after CCI at POD 7 and 14 (p = 0.0173 and < 0.001 respectively), but not at POD 21 and 28 (p = 0.128 and 0.967 respectively), compared to it in the contralateral side at the same time point (Fig. 4C).
IL-10 concentrations were also not significantly different between ipsilateral and contralateral sides of lumbar spinal cord dorsal part in control rats (104.2 ± 4.5 vs. 104.7 ± 7.3 pg/100 μg proteins, p = 0.956; Fig. 4D). IL-10 levels in the lumbar spinal cord dorsal part were unaffected by CCI at POD 7, 14, 21, and 28 compared to the controls (for the ipsilateral side 108.9 ± 7.4, p = 0.596; 106.2 ± 7.4, p = 0.823; 116.9 ± 8.0, p = 0.185; and 98.9 ± 4.9, p = 0.441 respectively, vs. 104.2 ± 4.5 pg/100 μg proteins; for the contralateral side 100.9 ± 5.6, 108.6 ± 2.7, 107.3 ± 5.0, and 117.3 ± 6.3, respectively, vs. 104.7 ± 7.3 pg/100 μg proteins; all p > 0.05; Fig. 4D). However, IL-10 level was significantly decreased in the ipsilateral side after CCI at POD 28 (p = 0.0359), but not at POD 7, 14, and 21 (p = 0.402, 0.766, and 0.324 respectively), compared to it in the contralateral side at the same time point (Fig. 4D). These data demonstrate that CCI induces a proinflammatory cytokine response in the ipsilateral lumbar spinal cord.
Intrathecal administration of fumagillin and anti-VEGF-A monoclonal antibodies attenuates CCI-induced neuropathic pain
The dose-response effect of fumagillin on CCI-induced pain behavior is shown in Fig. S2 (Supporting Information). Intrathecal injection of fumagillin per se had no analgesic effect in naïve and sham-operated rats for a 0.01–1 μg dose range. However, a trend toward decreased neuropathic pain in CCI rats was observed within 3 hours after a single intrathecal injection of 0.1 and 1 μg fumagillin. We also examined both narrow-beam walking and weight bearing tests to identify whether i.t. administration of anti-VEGF-A antibody has neurotoxic effect on motor function, and the results revealed that all rats were free from motor disabilities with the i.t. administration of anti-VEGF-A antibody at a dose of 0.3 μg/day for 14 consecutive days. In addition, i.t. administration of anti-VEGF-A antibody reduced the prolonged time to cross the beam and changed hindlimb weight distribution (Figure S3; Supporting Information) induced by CCI. Therefore, we used intrathecal administration of fumagillin (0.1 μg/day) or anti-VEGF-A antibodies (0.3 μg/day) once a day for 14 consecutive days after CCI to examine the angiogenesis’ role in CCI-induced neuropathic pain. The baseline nociceptive response to radiant heat and to a mechanical stimulus was comparable in all groups (p > 0.05, n = 6 in each group, Fig. 5). Time course studies showed a marked time-dependent reduction of the PWL in response to radiant heat (21.0 ± 1.0 vs. 29.9 ± 1.3 s, p = 0.008; 15.0 ± 1.7 vs. 30.0 ± 0.5 s, p < 0.001; 14.5 ± 2.1 vs. 29.0 ± 0.5 s, p < 0.001; 14.1 ± 1.4 vs. 29.8 ± 0.6 s, p < 0.001, 13.5 ± 1.9 vs. 29.5 ± 1.2 s, p < 0.001, and 12.9 ± 1.4 vs. 28.9 ± 1.9 s, p < 0.001, at POD 5, 7, 9, 11, 13, and 14, respectively; Fig 5A) from POD 5 to 14 and PWT triggered by a mechanical stimulus (8.2 ± 0.7 vs. 11.8 ± 0.6, p = 0.0127 at POD 3; 2.5 ± 0.5 vs. 11.8 ± 0.3, 2.1 ± 0.5 vs. 11.8 ± 0.9, 2.8 ± 0.7 vs. 12.2 ± 0. 5, 2.2 ± 1.0 vs. 11.5 ± 1.3, 2.2 ± 0.6 vs. 12.2 ± 1.0, and 2.0 ± 0.8 vs. 11.8 ± 0.9 g, at POD 5, 7, 9, 11, 13, and 14, respectively; all p < 0.001; Fig 5B) from POD 3 to 14 for the ipsilateral hindpaw of the CCI group compared to those of control rats. These data indicate that CCI progressively induces thermal hyperalgesia and mechanical allodynia within the 14-day postoperative observation period with the most remarkable effect at POD 14. Fumagillin significantly ameliorated the reduced PWL induced by CCI from POD 7 to 14 (24.1 ± 1.0 vs. 15.0 ± 1.7 s, p = 0.002; 25.5 ± 1.7 vs. 14.5 ± 2.1 s, p = 0.006; 24.5 ± 1.6 vs. 14.1 ± 1.4 s, p = 0.001; 24.5 ± 0.8 vs. 13.5 ± 1.9 s, p = 0.005; and 25.3 ± 1.0 vs. 12.9 ± 1.4 s, p < 0.001, at POD 7, 9, 11, 13, and 14, respectively) but the anti-VEGF-A antibody only significantly improved it at POD 7 and POD 11 (21.2 ± 1.1 vs. 15.0 ± 1.7 s, p = 0.026 and 20.0 ± 1.8 vs. 14.1 ± 1.4 s, p = 0.045) (Fig. 5A). Compared to the PWL of control rats, the PWL was significantly reduced in the CCI + fumagillin group at POD 7 (24.1 ± 1.0 vs. 30.0 ± 0.5 s, p = 0.007) and in the CCI + anti-VEGF group at POD 7, 9, 11, 13, and 14 (21.2 ± 1.1 vs. 30.0 ± 0.5 s, p < 0.001; 21.2 ± 1.1 vs. 29.0 ± 0.5 s, p = 0.016; 20.0 ± 1.8 vs. 29.8 ± 0.6 s, p = 0.006; 19.9 ± 2.7 vs. 29.5 ± 1.2 s, p = 0.015; and 18.7 ± 2.3 vs. 28.9 ± 1.9 s, p = 0.013, respectively; Fig. 5A). These results suggested that fumagillin was more efficient than the anti-VEGF-A antibody to suppress CCI-induced thermal hyperalgesia although the difference was not statistically significant (all p > 0.05; Fig. 5A).
Similarly, fumagillin significantly ameliorated the PWT from POD 5 to 14 (7.0 ± 1.0 vs. 2.5 ± 0.5 g, p = 0.006; 8.5 ± 1.0 vs. 2.1 ± 0.5 g, p < 0.001; 8.5 ± 0.5 vs. 2.8 ± 0.7 g, p < 0.001; 8.5 ± 0.5 vs. 2.2 ± 1.0 g, p = 0.002; 7.5 ± 1.5 vs. 2.2 ± 0.6 g, p = 0.004; 7.0 ± 1.3 vs. 2.0 ± 0.8 g, p = 0.008; at POD 5, 7, 9, 11, 13, and 14, respectively) but the anti-VEGF-A antibody only significantly improved it at POD 5 (7.1 ± 1.4 vs. 2.5 ± 0.5 g, p = 0.005), POD 7 (6.5 ± 0.7 vs. 2.1 ± 0.5 g, p = 0.003), POD 9 (6.8 ± 0.7 vs. 2.8 ± 0.7 g, p = 0.004), and POD 14 (5.7 ± 0.6 vs. 2.0 ± 0.8 g, p = 0.04) compared to the PWT of CCI rats (Fig. 5B). Compared to that of controls, the PWT was significantly decreased in the CCI + fumagillin group at POD 5, 9, 13, and 14 (7.0 ± 1.0 vs. 11.8 ± 0.3 g, p = 0.024; 8.5 ± 0.5 vs. 12.2 ± 0. 5 g, p = 0.001; 7.50 ± 1.5 vs. 12.2 ± 1.0 g, p = 0.039, and 7.0 ± 1.3 vs. 11.8 ± 0.9 g, p = 0.023, respectively; Fig. 5B) and in the CCI + anti-VEGF group from POD 5 to 14 (7.1 ± 1.4 vs. 11.8 ± 0.3 g, p = 0.027; 6.5 ± 0.7 vs. 11.8 ± 0.9 g, p = 0.008; 6.8 ± 0.7 vs. 12.2 ± 0.5 g, p = 0.008; 5.3 ± 0.7 vs. 11.5 ± 1.3 g, p = 0.005; 5.3 ± 0.4 vs. 12.2 ± 1.0 g, p = 0.007, and 5.7 ± 0.6 vs. 11.8 ± 0.9 g, p = 0.008, at POD 5, 7, 9, 11, 13, and 14, respectively; Fig. 5B). Fumagillin showed a tendency to be more efficient than the anti-VEGF-A antibody in suppressing the CCI-induced mechanical allodynia although the differences was only statistically significant at POD 11 (8.5 ± 0.5 vs. 5.3 ± 0.7 g, p = 0.0207; Fig. 5B).
In summary, intrathecal administration of fumagillin (0.01–0.1 μg) and anti-VEGF-A antibodies (0.3 μg) did not affect the behavior of control rats but improved CCI-induced pain behaviors. Because of its stronger effect on attenuating CCI-induced nociceptive behaviors, fumagillin (0.1 μg/day) was used in the following experiments to evaluate its effect on CCI-induced angiogenesis, astrocyte activation, and dysregulated pro-/anti-inflammatory cytokine balance in the spinal cord.
Intrathecal fumagillin attenuates CCI-induced angiogenesis in the lumbar spinal cord
The effects of the antiangiogenic therapy on the CCI-produced central sensitization were assessed at POD 14 since CCI-induced the most significant angiogenesis at this time point. Fumagillin significantly decreased the CCI-induced upregulation of VEGF and vWF immunoreactivities (12.4- vs. 27.0-fold, p = 0.0486 and 1.6- vs. 2.4-fold, p = 0.0266, respectively; Fig. 6 A-D), but not CD31 immunoreactivities (and 1.0- vs. 1.8-fold, p = 0.0841; Fig. 6 E,F), in the ipsilateral lumbar SCDH. The extent of fumagillin-reduced expression of angiogenic factor was strongest on VEGF (−54%), followed by CD31 (−48%) and vWF (−33%), whereas no significant difference was found in the CD31 expression (Figure 6 B,D,F). However, VEGF and vWF immunoreactivities in the CCI + fumagillin group were still more intense than that of the control group (12.4- vs. 1-fold, p = 0.01 and 1.6- vs. 1-fold, p = 0.0409; respectively), while there was no difference in CD31 staining (1.0- vs. 1.0-fold, p = 0.907) (Figure 6 B,D,F). These data demonstrate that fumagillin effectively improves CCI -induced angiogenesis in the ipsilateral lumbar spinal cord.
Fumagillin differentially modulates CCI-induced spinal cytokine production
IL-1β levels did not significantly change at POD 7 and 14 in the CCI + fumagillin group compared to that of the CCI group (146.7 ± 11.9 vs. 138.0 ± 7.1, p = 0.486 and 139.3 ± 7.23 vs. 143.5 ± 7.9 pg/100 μg proteins, p = 0.652; Fig. 7A). However, fumagillin dramatically reduced the amounts of IL-6 (76.2 ± 16.2 vs. 162.7 ± 7.4 and 124.7 ± 10.3 vs. 159.1 ± 2.0 pg/100 μg proteins at POD 7 and 14, respectively, both p < 0.001; Fig. 7B) and TNF-α (67.5 ± 7.9 vs. 133.2 ± 10.8 and 89.5 ± 5.0 vs. 130.2 ± 9.2 pg/100 μg proteins, at POD 7 and 14, respectively, both p < 0.001; Fig. 7C) at POD 7 and 14 compared to that of the CCI group. Furthermore, the effect of fumagillin in suppressing CCI-upregulated TNF-α and IL-6 concentrations at POD 7 tended to be stronger than that at POD 14, albeit not statistically significant (Fig. 7B and 7C). Unexpectedly, fumagillin significantly increased IL-10 levels at POD 7 and 14 compared to that of CCI rats (147.9 ± 20.6 vs. 110.3 ± 4.8, p = 0.0076 and 140.4 ± 12.7 vs. 114.6 ± 5.2 pg/100 μg proteins, p = 0.0332). The increased IL-10 production induced by fumagillin seemed greater at POD 7 than that at POD 14; however, the difference was not statistically significant (Fig. 7D). In summary, fumagillin modulates CCI-induced neuroinflammation by enhancing IL-10 amounts and inhibiting the production of TNF-α and IL-6 without affecting IL-1β levels.
Fumagillin intrathecal administration attenuates CCI-induced astrocyte activation in the ipsilateral lumbar spinal cord
CCI markedly upregulated GFAP immunoreactivities at POD 14 compared to that of the control group (1.7- vs. 1-fold, p < 0.001; Fig. 8). GFAP expression was decreased in the CCI + fumagillin group compared to that of the CCI group (0.6- vs. 1.7-fold, p = 0.00278) but not statistically significant compared to that of control groups (0.6- vs. 1-fold, p = 0.232) (Fig. 8 A,B). These data demonstrate that intrathecal administration of fumagillin effectively abolishes astrocyte activation in the ipsilateral lumbar spinal cord in CCI-induced neuropathic pain.