VEGFR2 promotes central endothelial activation and the spread of pain in inflammatory arthritis

Abstract

Chronic pain can develop in response to conditions such as inflammatory arthritis. The central mechanisms underlying the development and maintenance of chronic pain in humans are not well elucidated although there is evidence for a role of microglia and astrocytes. However in pre-clinical models of pain, including models of inflammatory arthritis, there is a wealth of evidence indicating roles for pathological glial reactivity within the CNS. In the spinal dorsal horn of rats with painful inflammatory arthritis we found both a significant increase in CD11b⁺ microglia-like cells and GFAP⁺ astrocytes associated with blood vessels, and the number of activated blood vessels expressing the adhesion molecule ICAM-1, indicating potential glio-vascular activation. Using pharmacological interventions targeting VEGFR2 in arthritic rats, to inhibit endothelial cell activation, the number of dorsal horn ICAM-1⁺ blood vessels, CD11b⁺ microglia and the development of secondary mechanical allodynia, an indicator of central sensitization, were all prevented. Targeting endothelial VEGFR2 by inducible Tie2-specific VEGFR2 knock-out also prevented secondary allodynia in mice and glio-vascular activation in the dorsal horn in response to inflammatory arthritis. Inhibition of VEGFR2 in vitro significantly blocked ICAM-1-dependent monocyte adhesion to brain microvascular endothelial cells, when stimulated with inflammatory mediators TNF-α and VEGF-A₁₆₅a. Taken together our findings suggest that a novel VEGFR2-mediated spinal cord glio-vascular mechanism may promote peripheral CD11b⁺ circulating cell transmigration into the CNS parenchyma and contribute to the development of chronic pain in inflammatory arthritis. We hypothesise that preventing this glio-vascular activation and circulating cell translocation into the spinal cord could be a new therapeutic strategy for pain caused by rheumatoid arthritis.

Keywords: CD11b; Chronic pain; Glio-vascular activation; ICAM-1; Inflammatory pain; Mechanical allodynia; Microglia, mono-arthritis; Rheumatoid arthritis; VEGFR2.

Fig. 1

Articular inflammation drives secondary mechanical allodynia, glio-vascular activation and microglia reactivity in the spinal cord. Tibiofemoral joint diameter (a), weight borne on the ipsilateral hind paw (b), ipsilateral mechanical stimulus withdrawal threshold (c) and contralateral mechanical stimulus (d) withdrawal threshold after intra-articular CFA (single dose, 100 μg), n = 9 (CFA), n = 4 (non-inflamed controls). Immunoreactivity of GFAP and CD11b antibodies in the dorsal horn 8 and 11 days following induction of inflammatory arthritis (e-l). Representative images (e,g,k) showing the increase in GFAP (red) and CD11b (green) immunoreactivity in the dorsal horn in (e-f) non-inflamed control, (g-j) day eight and (k-l) day 11 following induction of articular inflammation with CFA (nuclei- blue). High magnification imaging revealed microvessels, identified by strings of flat (longitudinal) or curved (cross-sectional) nuclei (blue), surrounded by GFAP⁺ astrocytic end-feet (red, arrowhead) and associated with CD11b⁺ cells (green, arrows) at day eight (h,i). 2D image following z-stack reconstruction showing a CD11b⁺ cell with a cellular projection within the vessel lumen (j). Wide spread dorsal horn CD11b⁺ microglia-like staining within the dorsal horn parenchyma by day 11 (k-l). Representative images showing ICAM-1 expression in the dorsal horn of non-inflamed controls (m-r) and CFA-treated animals on day 8 (n) and 11 (m). ICAM-1⁺ microvessel fragments (p) were observed throughout the dorsal horn. ICAM-1⁺ immunoreactivity was associated with CD31⁺ vessels (q) and vessels wrapped in GFAP⁺ astrocytic end-feet (r). Statistical analysis: two-way repeated measures analysis of variance + Dunnett’s multiple comparisons test. CFA + vehicle vs. respective baseline: ***p,<0.001, ****p < 0.0001. Group size = 8–9. Data displayed as mean ± SD. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Systemic delivery of anti-VEGFR2 agents prevents the development of secondary mechanical allodynia. Tibiofmoral joint diameter after intra-articular CFA + two i.p. injections (arrows) of either a murine VEGFR2-specific neutralising antibody (DC101), VEGF-A₁₆₅b or PTK787 compared with control (Control IgG) (a). Weight bearing on the paw ipsilateral to CFA injection after treatment with either DC101 (murine neutralizing VEGFR2 mAb), PTK787 (VEGFR1&2 inhibitor) or VEGF-A₁₆₅b (competitive VEGF receptor antagonist) (b). Ipsilateral hind paw mechanical sensitivity after CFA and VEGFR2 inhibition (c). Contralateral hind paw sensitivity (d). Two statistical analyses were performed: two-way repeated measures analysis of variance + Dunnett’s multiple comparisons tests: CFA + Control IgG vs. respective baseline, *p < 0.05, **p,<0.01, ***p < 0.0001; or CFA + drug vs. CFA + Control IgG at given time point, ♯p < 0.05; ♯♯p < 0.01, ♯♯♯p < 0.0001, ♯♯♯♯p < 0.0001. Group size = 5–9. Data displayed as mean ± SD.

Fig. 3

Intra-articular (tibiofemoral) delivery of VEGF receptor inhibitors did not affect the development of pain-associated behaviors in the CFA model of inflammatory monoarthritis. Intra-articular (i.a) CFA (single dose) plus two i.a. vehicle injections (arrow heads) caused a significant reduction in ipsilateral mechanical stimulus withdrawal threshold (a,d) at all time points measured, weight borne on ipsilateral hind paw (b,e) (days 1–10) and a significant increase in tibiofemoral joint diameter (c,f) compared to respective baseline (day 0). CFA + two i.a. injections of VEGF-A₁₆₅b or PTK787 (arrow heads) had no effect on the development of the pain-associated behaviors and joint diameter (a-f). Two statistical analyses were performed: 2-way ANOVA + Dunnett’s multiple comparisons tests: CFA + vehicle vs. respective baseline, *p < 0.05, **p < 0.01, ***p < 0.0001; or CFA + vehicle vs. CFA + drug, #p < 0.05, ##p < 0.01, ###p < 0.0001. A-C n = 10, D-F n = 5. Data displayed as mean ± SD.

Fig. 4

The effect of anti-VEGFR2 agents on astrocytic and microglial reactivity to articular inflammation. Intra-articular CFA caused a significant increase in the number of GFAP⁺ end-feet-wrapped vessel fragments in the ipsilateral dorsal horn on both (a) day eight and (e) day 11. Treatment with DC101 (b,f), but not VEGF-A₁₆₅b (c,g) nor PTK787 (d,h) significantly reduced the number of ipsilateral and contralateral GFAP⁺ end-feet-wrapped vessel fragments on day eight (i), but not on day 11 (j). Intra-articular CFA caused a significant increase in microvessel-associated CD11b⁺ cells (ipsi- and contralateral) on day eight (k) and day 11 (l) with no difference observed between the CFA groups at both time points. On day eight there was no significant difference in the number of CD11b⁺ cells in the dorsal horn parenchyma compared to control (m), whereas on day 11 there was a significant increase in CD11b⁺ cells in the dorsal horn parenchyma compared to non-inflamed control animals. All three anti-VEGF treatments reduced this number although only VEGF-A₁₆₅b (ipsi & contalateral) and PTK787 (ipsilateral) reached statistical significance (n). The only significant difference between respective ipsi- and contralateral effects was observed in VEGF-A₁₆₅b group on the number of GFAP-wrapped vessels at day 11 (j). Three statistical analyses were performed: 2-way analysis of variance + Dunnett’s multiple comparisons test: control vs. non-inflamed *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001; anti-VEGFR2 agent vs. control (ipsi or contra): #p < 0.5, ##p < 0.01, ###p < 0.001, ####p < 0.0001 and ipsi vs. contra of respective group +p < 0.5. Day eight: n = 4; day 11: n = 4–5. Data displayed as mean ± SD.

Fig. 5

The effect of anti-VEGFR2 agents on dorsal horn vascular activation. Representative low magnification images of ICAM-1 immunoreactivity in the dorsal horn (a-h). Number of ICAM-1⁺ microvessels present in the dorsal horn (i,j). Three statistical analyses were performed: 2-way analysis of variance + Dunnett’s multiple comparisons test: control vs. non-inflamed *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001; anti-VEGFR2 agent vs. control (ipsi or contra): #p < 0.5, ##p < 0.01, ###p < 0.001, ####p < 0.0001 and ipsi vs. contra of respective group +p < 0.5. Day eight: n = 4; day 11: n = 4–5. Data displayed as mean ± SD.

Fig. 6

VEGFR2 inhibition reduces ICAM-1 expression in DRG. ICAM-1 and VEGFR2 immunoreactivity in the ipsilateral L3 dorsal root ganglion eight days following induction of inflammatory arthritis (a). ICAM-1 expression was localized to DRG microvessels expressing VEGFR2 (scale bar- 10 μm). Representative images showing ICAM-1 expression in L3 DRG 8 days following induction of inflammatory arthritis with treatment with control IgG, DC101, VEGF-A₁₆₅b or PTK787 (b-f). Arrow indicates ICAM-1⁺ vessel fragment. No primary and species- and concentration-matched isotype control IgG staining (g,h). Scale bar-100 μm. Quantification of microvessel ICAM-1 expression in ipsilateral and contralateral L3 DRG from non-inflamed, or CFA injected animals on day eight (i) and 11 (j) and with anti-VEGF treatments. Statistical analyses: g,h: one-way analysis of variance + Dunnett’s multiple comparisons test: vs. non-inflamed *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001; vs. CFA + IgG control: #p < 0.5, ##p < 0.01, ###p < 0.001, ####p < 0.0001 and ipsi vs. contra of respective group +p < 0.05, ++p < 0.01, +++p < 0.001, ++++p < 0.0001. Day eight: n = 3–4; day 11: n = 4–5. Data displayed as mean ± SD.

Fig. 7

Targeting VEGFR2 reduced monocyte adhesion to brain endothelial cells in vitro. Expression of endothelial associated protein von Willebrand Factor, and occludin-1, in two different primary human brain microvascular endothelial cells (HBMEC) cultures and HUVEC (a). Representative images of fluorescent monocytes attached to cultured brain endothelial monolayer following treatment with or without TNF-α or TNF-α+ ICAM-1 blocking antibody, scale bar = 100 μm (b). Effect of pre-treatment with a selective VEGFR2 inhibitor ZM883231 (ZM, 20 nM, 1 h), VEGF-A₁₆₅b (10 ng/ml, 24 h), the VEGFR1&2 inhibitor PTK787 (PTK, 100 nM, 1 h) or an ICAM-1 blocking antibody (conc, time) or non-specifc IgG on VEGF-A₁₆₅a-induced monocyte adhesion, n = 9 (c). Effect of pre-treatment with VEGF-A₁₆₅b (24 h), PTK787 (1h), an ICAM-1 blocking antibody or ZM323881 (1h) on TNF-α-induced monocyte adhesion (d). Statistical analyses: 1-way analysis of variance + Dunnett’s multiple comparisons test: vs. untreated *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001; vs. respective control: #p < 0.5, ##p < 0.01, ###p < 0.001, ####p < 0.0001. Abbrev. Ab – antibody; HBMEC – human brain microvascular endothelial cells; HUVEC – human umbilical vein endothelial cells; vWF – von Willebrand Factor; VEGF-A – vascular endothelial growth factor-A; TNF-α – tumor necrosis factor-α; ICAM-1 – intercellular adhesion molecule-1. Adhesion assays: n = 9–20 (individual wells collated from three independent experiments). Data displayed as mean ± SD.

Fig. 8

Endothelial VEGFR2 knock-out affected pain behaviors in comparison to uninduced control animals. Ipsilateral (a) and contralateral (b) hind paw mechanical stimulus threshold in control (uninduced) animals. In VEGFR2^ECKO mice, the reduction in mechanical stimulus threshold was significantly delayed in (a) the ipsilateral hind paw and significantly prevented in (b) the contralateral hind paw in comparison with uninduced control mice. A significant inhibition of the CFA-induced shift in weight bearing was also observed on day 2 (c) and by day 14 there was a significant reduction in joint diameter (d). Mechanical allodynia 2 and 5 days following the start of recombinant human VEGF-A₁₆₅a treatment (8ng/g i.p. biweekly) (g). Statistical analyses: b-f, h: 2-way repeated measures analysis of variances + Dunnett’s multiple comparisons test; vs. respective baseline (day 0): *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001; vs. respective control at given time point: ##p < 0.01, ###p < 0.001, ####p < 0.0001; f: Student’s t-test, *p < 0.05. a: n = 3; b: n = 10–11; c-f n = 5–6; g,h; n = 5. Abbrev. CFA, complete Freund’s adjuvant. Data displayed as mean ± SD.

Fig. 9

Endothelial VEGFR2 knock-out affected pain behaviors in comparison to tamoxifen-treated Tie2CreER^T2-wildtype control animals. In addition to the knock-out and uninduced experimental groups (refer to Fig. 8, a tamoxifen-treated Tie2CreER^T2-negative (wildtype) group was included in the peri-articular CFA experiment. A significant reduction in ipsilateral (a) and contralateral (b) mechanical stimulus threshold developed in these mice as was observed in the uninduced control group (refer to Fig. 8. These effects were significantly different from the knockout group (ipsi: day 2, contra: all time points). A significant inhibition of the CFA-induced shift in weight bearing was also observed on day 2 (c) while no difference was observed in joint swelling (d). Statistical analyses: two-way repeated measures analysis of variances + Dunnett’s multiple comparisons test; either control vs. respective baseline (day 0): *p < 0.5, **p < 0.01, ***p < 0.001, ****p < 0.0001; or test groups vs. respective control at given time point: ##p < 0.01, ###p < 0.001, ####p < 0.0001, n = 5–6. Data displayed as mean ± SD.

Fig. 10

VEGFR2^ECKO inhibited glio-vascular activation in the dorsal horn of CFA treated mice. Peri-articular CFA caused an significant increase in the number of dorsal horn CD31⁺ vessels associated with GFAP⁺ reactive astrocytic foot processes (glio-vascular response) compared with sham in uninduced mice (a,b, quantification in p). There was a significant increase in the number of vessels associated with GFAP⁺ astrocytic foot processes in sham-treated VEGFR2^ECKO compared with sham-treated uninduced controls, however CFA treatment in VEGFR2^ECKO mice did not significantly increase this further (c,d, quantification in p). CFA caused a significant increase in the number of CD11b⁺ cells associated with CD31⁺ vessels and the number of ICAM-1⁺ vessel structures in uninduced mice compared with sham (e,f,i,j quantification in q&r) but the same effects were not observed in VEGFR2^ECKO (g,h,k,l quantification in q&r). Higher magnification images of GFAP⁺ reactive astrocytic end feet (m, n, arrowheads denote GFAP⁺/CD31⁺ vessel structures), CD11b⁺ cells associated with vessel (n, arrows denote CD11b⁺ cells associated with CD31⁺ vessel structures, dotted arrow denotes a parenchyma CD11b⁺ cell) and ICAM-1⁺ vessels (o, arrowheads denote ICAM-1⁺/CD31⁺ vessel structures). Three statistical analyses were performed: 2-way ANOVA + Bonferroni multiple comparisons test: vs. uninduced sham con *p < 0.05, ** p < 0.01; KO CFA vs KO sham control – no significance; contra vs. ipsi of respective group – no significane; n = 3–6. Data displayed as mean ± SD.