OUP user menu

Rescuing vasculature with intravenous angiopoietin-1 and αvβ3 integrin peptide is protective after spinal cord injury

Shu Han , Sheila A. Arnold , Srinivas D. Sithu , Edward T. Mahoney , Justin T. Geralds , Phuong Tran , Richard L. Benton , Melissa A. Maddie , Stanley E. D’Souza , Scott R. Whittemore , Theo Hagg
DOI: http://dx.doi.org/10.1093/brain/awq034 1026-1042 First published online: 7 April 2010

This article has a correction. Please see:


Blood vessel loss and inflammation cause secondary degeneration following spinal cord injury. Angiopoietin-1 through the Tie2 receptor, and other ligands through αvβ3 integrin, promote endothelial cell survival during developmental or tumour angiogenesis. Here, daily intravenous injections with an αvβ3-binding peptide named C16 or an angiopoietin-1 mimetic following a spinal cord contusion at thoracic level 9 in mice rescued epicentre blood vessels, white matter and locomotor function, and reduced detrimental inflammation. Preserved vascularity and reduced inflammation correlated with improved outcomes. C16 and angiopoietin-1 reduced leukocyte transmigration in vitro. Growth factor receptors and integrins facilitate each others’ function. Therefore, angiopoietin-1 and C16 were combined and the effects were additive, resulting in almost complete functional recovery. The treatment had lasting effects when started 4 h following injury and terminated after one week. These results identify αvβ3 integrin and the endothelial-selective angiopoietin-1 as vascular and inflammatory regulators that can be targeted in a clinically relevant manner for neuroprotection after central nervous system trauma.

  • degeneration
  • endothelial cell
  • inflammation
  • neuroprotection
  • vascular


Currently, the only neuroprotective treatment for acute spinal cord injury in humans is methylprednisolone, but its use is controversial (Hurlbert and Hamilton, 2008). Spinal cord injury, particularly the common contusive and compressive types, causes progressive tissue loss in part secondary to blood vessel dysfunction and inflammation at the injury epicentre (Norenberg et al., 2004; Fleming et al., 2006; Bramlett and Dietrich, 2007). Reduced microvascular perfusion and haemorrhage, starting with small petechial bleeding at the epicentre soon after the injury, are key to the progressive degeneration and subsequent functional deficits (Allen, 1914; Tator and Fehlings, 1991; Tator and Koyanagi, 1997; Hall and Springer, 2004; Norenberg et al., 2004). For example, treatment with nimodipine plus systemic vasopressor can maintain spinal cord blood flow and axonal conduction during the acute injury phase in rats (Fehlings et al., 1989; Guha et al., 1989). Endothelial cells and blood vessels show degenerative changes within 30 min after spinal cord injury (Dohrmann et al., 1971; Casella et al., 2006) and are lost during the first 3 days, exacerbating the ischaemia (Koyanagi et al., 1993; Loy et al., 2002; Casella et al., 2006; Benton et al., 2008). This is caused in part by reactive oxygen species and lipid peroxidation of microvasculature (Hall, 1995; Hall and Springer, 2004). In fact, treatment with the non-glucocorticoid 21-aminosteroid tirilazad, anti-oxidants or selenium maintains spinal cord blood flow and reduces leakage following contusive spinal cord injury (Hall and Wolf, 1986; Hall, 1988; Hall et al., 1994). Transient receptor potential cation channel M4 and sulfonylurea receptor-1 expression by endothelial (and possibly other) cells contributes to degeneration as their inhibition reduces microvascular fragmentation, haemorrhage, white matter loss and functional deficits in rats and mice with cervical hemi-contusion injuries (Simard et al., 2007; Gerzanich et al., 2009). Surviving blood vessels become leaky (Noble and Wrathall, 1989; Whetstone et al., 2003; Benton et al., 2008), contributing to oedma, and mediate leukocyte infiltration, which contributes to loss of myelin and tissue (Donnelly and Popovich, 2008). Therapies targeting inflammatory responses partially improve tissue sparing and neurological function following spinal cord injury in animals (Popovich et al., 1999; Weaver et al., 2005). The inflammatory response in the injured human spinal cord is largely similar to that seen in rodents (Fleming et al., 2006). We were interested in testing additional therapeutic agents that are even more selective for endothelial cells to improve outcome following spinal cord injury.

Angiopoietin-1 signals through the Tie2/Tek receptor, promotes endothelial cell survival, stabilizes blood vessels and reduces leakiness, effects observed during both developmental and adaptive angiogenesis in tumours and experimental CNS pathology (Thurston et al., 1999, 2000, 2005; Gale et al., 2002; Uemura et al., 2002; Zhang et al., 2002; Carmeliet, 2003; Nambu et al., 2004; Mochizuki, 2009). Tie2 is potentially a selective therapeutic target as it is almost exclusively present in endothelial cells (Dumont et al., 1993; Suri et al., 1996). The αvβ3 integrin promotes endothelial cell survival during tumour angiogenesis (Brooks et al., 1994). Integrin binding to extracellular matrix molecules such as laminin is important for attachment and survival of various cells, including endothelial cells (Hynes, 1992; Giancotti and Ruoslahti, 1999). Endothelial cells detach rapidly following spinal cord injury (Goodman et al., 1979; Koyanagi et al.,1993), suggesting that their death might result from lack of integrin stimulation. The synthetic C16 peptide, representing a functional laminin domain, selectively binds to αvβ3 and α5β1 integrin, promotes angiogenesis in vitro and in the chick chorioallantoic assay in vivo (Ponce et al., 1999, 2001, 2003). Integrins have a reciprocal functional interaction with growth factor receptors (Eliceiri, 2001), and both αvβ3 and Tie2 receptors can regulate endothelial cells through the PI3K-Akt pathway (Zheng et al., 2000; DeBusk et al., 2004). This suggested that co-activation with angiopoietin-1 and C16 might further enhance their functions, possibly resulting in better outcomes following spinal cord injury. The roles of angiopoietin-1 and αvβ3 integrin in neurotrauma have not been investigated.

Material and methods

Animals and overall design

A total of 291 female C57BL/6 mice were used (7–11 weeks, 16–24 g at the time of spinal cord injury; Jackson Laboratory, Bar Harbor, ME, USA) and age- and weight-matched between groups within an experiment. All animal procedures were performed according to University of Louisville Institutional Animal Care and Use Committee protocols and the National Institutes of Health guidelines. All invasive procedures were performed under deep anaesthesia obtained by an intraperitoneal injection of 0.4 mg/g body weight Avertin (2,2,2-tribromoethanol in 0.02 ml of 1.25% 2-methyl-2-butanol in saline, Sigma-Aldrich, St Louis, MO, USA).

An overview of all experimental groups is presented in Table 1. We first determined whether the integrins were present in endothelial cells of the penumbra at 1 and 3 days following contusion at thoracic level 9 (T9). We next determined whether C16 could decrease the volume of injury and what its optimal dose was for protecting white matter at 7 days post-injury. A dose–response study for white matter rescue by angiopoietin-1 was performed at a later time. Next, we determined whether a 14-day treatment with C16 would improve locomotor function over 6 weeks. The duration of the treatment was based on the idea that the sub-acute injury phase of demyelination and maximal inflammation lasts 1–2 weeks (Donnelly and Popovich, 2008). Subsequently, we scaled back to a 7-day treatment for the 6-week experiments involving co-treatment with C16 and angiopoietin-1, because shorter treatments would ultimately be preferable in a clinical setting to reduce potential side-effects of the therapeutic reagents. Additional groups of mice were used to investigate the effects of C16 and angiopoietin-1 or their combination on vascular and inflammatory responses at 1, 3 and 7 days. Because of persistent inflammation at 6 weeks post-injury, we also tested whether a second treatment with C16 between 4 and 5 weeks would further improve the functional outcomes.

View this table:
Table 1

Mouse numbers in each of the experiments

Integrin presence 1, 3 day injuryNo treatment13
7 day injury volumeVehicle10C1610
7 day injuryNormal3
Vehicle 15C16: 3 doses19
Vehicle 25Ang-1: 3 doses15
1 day injuryVehicle5C165
    plus treatmentAng-15
7 day injuryVehicle 117C1613
    plus treatmentSP38
Vehicle 25C164
Vehicle 34Ang-15
1X vehicle81X C169
6 week injuryVehicle 18C169
    plus treatmentVehicle 213Ang-114
4–5 week second treatmentVehicle8C166
Luciferase 1,Vehicle8C1610
    3 day injuryNo luciferase8Ang-110
  • Ang-1 = angiopoietin-1

Surgeries, behavioural measurements and quantification of histological results were done by investigators blinded to the treatments. Treatment solutions were assigned in a randomized order and were prepared and coded by someone (T.H.) different than the surgeon (S.H., J.G.). Spinal cords of individual mice were randomly coded before histological processing and un-blinded only after analyses.

Spinal cord injury

The mice were anaesthetized and their backs shaved and cleansed with Betadine (Purdue Products L.P., Stamford, CT, USA). Lacri-Lube ophthalmic ointment (Allergen, Irvine, CA, USA) was placed on their eyes to prevent drying and 50 mg/kg of gentamicin (Boehringer Ingelheim, St. Joseph, MO, USA) was administered subcutaneously to reduce infection. After a midline incision and laminectomy of the T9 vertebra, spinal cord contusions were induced using the Infinite Horizon impactor with the force set at 50 kilodyne (PSI, Lexington, KY, USA) (Scheff et al., 2003). The vertebral column was stabilized in a frame with rigid steel clamps inserted under the transverse processes. Only mice with injury displacements from 400–800 µm were included to ensure similar injury severity between groups and to exclude aberrant injuries. The average displacement per group did not differ in this study. After the injury, the muscles were closed in layers, the skin incision was closed with 7-mm metal wound clips and 2 ml of lactated Ringer’s solution was given subcutaneously. Bacitracin zinc antibiotic ointment (Altana, Melville, NY, USA) was applied to the incision area. Food was placed on the bottom of the cage and water bottles with long sipping tubes were used. Buprenorphine (0.05 mg/kg) was given subcutaneously at 48 h post-injury to reduce pain. Bladders were manually expressed twice daily until the mice had regained partial voluntary or autonomic voiding, at which time they were reduced to once a day manual expression until full voluntary or autonomic voiding was obtained. Surgeries were performed at room temperature with the mice positioned on a heating pad to maintain body temperature. After the surgery, mice were placed on fresh Alpha-dry bedding with cages placed on a water circulating thermal pad (37°C) overnight before being returned to the animal care facility. Metal sutures were removed after 7 days.

Intravenous injections

For intravenous injection on the day of spinal cord injury, a midline incision was made in the ventral neck area and one of the jugular veins was exposed by blunt dissection and injected with 100 µl of sterile vehicle or vehicle containing sterile C16 peptide (KAFDITYVRLKF; synthesized by Peptides International, Louisville, KY, USA), SP3 peptide (RFSVAVSSHYPFWSR; synthesized by Sigma-Genosys, St. Louis, MO, USA) or an angiopoietin-1 mimetic [angiopoietin-1 tetra-fibrinogen-like domain (TFD) or angiopoietin-14FD; a gift from Regeneron Pharmaceuticals Inc., Tarrytown, NY, USA], subsequently referred to as angiopoietin-1. Angiopoietin-1TFD contains two human angiopoietin-1 fibrinogen-like domains fused to a human Fc domain, mimics the natural multimeric angiopoietin-1 and has biological activity in vivo (Uemura et al., 2002; Zhang et al., 2002; Nambu et al., 2004). C16 is very selective as it was shown by affinity chromatography and immunoprecipitation to bind only to αvβ3 and α5β1 integrins and not α1, α2, α3, α6, β4, αvβ5 (Ponce et al., 2001). Moreover, C16 activity is blocked by addition of αvβ3 and α5β1 antibodies in vitro. We could not use C16-based control peptides, as scrambled C16 acts as an αvβ3 and α5β1 antagonist, for example, in fibroblast growth factor 2-induced angiogenesis, and reverse C16 has some binding activity for endothelial cells in culture (Ponce et al., 1999, 2001). Instead, as a general peptide control we used SP3 peptide, which is an inactive scrambled form of a α6β1-binding peptide without effects on neural precursor migration in vitro (Jacques et al., 1998). It has no neuroprotective activity after intrathecal infusion following contusive spinal cord injury in rats (Baker and Hagg, unpublished data) or on monocyte transmigration in vitro (see Results section). The latter suggests that it does not bind to αvβ3 integrin, which is involved in that transmigration. To improve solubility of the peptides, they were dissolved in distilled water with 0.3% acetic acid. Afterwards, the peptide solution was sterilized through a 0.22-µm disc filter and neutralized with NaOH. This solution was buffered by adding an equal volume of sterile phosphate-buffered saline. The vehicle was prepared in the same manner without adding the peptide. After the jugular vein injection, the skin was closed with metal sutures. On the following 6 or 13 days, the solutions were injected via the tail vein. Ready access to the tail veins was achieved by starting injections at the caudal end of the base of the tail and into one vein. Injection sites were moved to the alternate left or right sides and increasingly rostral on subsequent days.

To pre-label perfused blood vessels, the other jugular vein was exposed and injected with 100 µg/100 µl fluorescein isothiocyanate-conjugated Lycopersicon esculentum (tomato) agglutinin lectin (LEA; which labels perfused vasculature) (Benton et al., 2008) 30 min before euthanasia. Following intravenous injection, LEA binds only to endothelial cells and only to those of perfused blood vessels (Mazzetti et al., 2004; Benton et al., 2008). This is a reliable method to detect endothelial cell survival in perfused blood vessels simultaneously.

To quantify blood–spinal barrier permeability to proteins, mice were injected with 80 µl of a 0.5 µg/µl solution of luciferase (L9506; Sigma) in 0.05 M phosphate buffered saline/0.001% bovine serum albumin into the jugular vein 30 min before processing (Whetstone et al., 2003). After flushing out blood by phosphate buffered saline perfusion, a 5-mm block (epicentre ± 2.5 mm) was collected and protein immediately extracted, from the tissue, in lysis buffer (E1500; Promega), centrifuged and supernatant used to measure enzyme activity with a luciferase assay kit (E1500; Promega) and a luminometer. Values were corrected for wet tissue weight and background values, obtained from injured but untreated mice injected with saline instead of luciferase, subtracted. All samples were analysed in triplicate.

To determine potential effects of C16 on the number of peripheral leukocytes, 1 ml blood was drawn from the heart just before perfusion-fixation in mice that had been treated with vehicle or C16 for 7 days. The blood was collected in ethylenediaminetetraacetic acid-coated tubes and blood counts were performed by Drew Scientific Inc. (Oxford, CT, USA).

Functional testing

Functional recovery after spinal cord injury was determined weekly by observers trained and certified by Dr Basso at Ohio State University using open-field overground locomotor performance according to the Basso Mouse Scale (BMS) (Basso et al., 2006). This test measures hindlimb performance in stepping, limb coordination and trunk stability on a 10-point scale: BMS scores of 0–2 involve hindlimb paralysis; 3–4 some ankle movement; 5–6 weight-bearing with some coordination and stepping; 7–9 high functioning with consistent coordinated stepping; and where 9 is normal. BMS subscores (Basso et al., 2006) are more sensitive and can be calculated when mice have a score of 5 and over. These represent fine motor skills including frequency of plantar stepping, interlimb coordination, paw position during stepping, as well as trunk stability and tail positioning during locomotion. Before each evaluation, the mice were examined carefully for perineal infections, wounds in the limbs, or tail and foot autophagia, as they could influence stepping. Such mice were excluded from the BMS test. The mice were acclimatized to the testing area for at least 25 min, including individual handling for at least 5 min, for 3 days and then tested for baseline values before surgery. The mice in the 7 day experiments were tested on the day of euthanasia and the mice in the 6 week chronic group were tested every week starting on Day 8 or 9 after the contusion.

Histological procedures

Mice were perfused transcardially with 10 ml phosphate-buffered saline followed by 20 ml of 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Afterwards, the spinal cords were carefully dissected out and 1-cm segments containing the injury site were post-fixed for 24 h, in the same fixative, at 4°C and then cryoprotected at 4°C in 30% phosphate buffered sucrose overnight. Up to 10 segments were embedded in TissueTek (Sakura Finetek, Torrance, CA, USA) with their injury sites aligned. This ensures that groups within an experiment are processed for histology in the same manner. Twenty consecutive 20-μm transverse sections per 1 mm rostrocaudal distance along the spinal cord axis were cut on a cryostat and thaw-mounted onto charged microscope slides. In the set of mice used to determine effects of C16 on the volume of tissue loss, the cord was cut in the horizontal plane. The sections were stored in sequence at –20°C until further use.

To detect myelin in white matter tracts, one of every five of the transverse sections at each rostro-caudal 1 mm level were stained with a modified eriochrome cyanine staining protocol. After thawing and drying for 1–2 h in a slide warmer at 37°C, the slides were placed in xylene at room temperature for 2 × 30 min, then through graded ethanol solutions (twice each in 100 and 95% ethanol, once in 70% ethanol, and twice in double-distilled H2O and stained with 0.2% eriochrome cyanine RS in 0.5% sulphuric acid and 0.4% ferric ammonium sulphate for 30 min. Afterwards, the slides were gently washed in running tap water for 5 min, and then briefly rinsed in double-distilled water. The slide was differentiated in 5% ferric ammonium sulphate for 5–10 min, briefly rinsed in double-distilled water, dehydrated briefly through graded ethanol solutions, cleared through xylene and the sections coverslipped in Entellan (Electron Microscopy Sciences, Hatfield, PA, USA). The injury epicentre was determined for each mouse by the rostro-caudal level that contained the least amount of spared myelin per transverse section. This epicentre level was used to align all the other histological measurements for each mouse. In case of the 24 h post-injury mice, the greatest loss of LEA-labelled blood vessels was used to determine the epicentre.

Adjacent sections at each millimetre level were processed for double- or triple immunofluorescent staining to detect CD45 (leukocytes), CD68 (activated microglia/macrophages), platelet endothelial cell adhesion molecule-1 (endothelial cells), αv, β3 or α5β1 integrin, or in some cases laminin to define the area of tissue loss. In mice, lost tissue after spinal cord injury is replaced by a fibroblast-rich stroma, which has been identified as a heterodomain (Whetstone et al., 2003) and is rich in laminin (Ma et al., 2001). Slides were warmed for 20 min on a slide warmer, a ring of wax applied around the sections with a PAP pen (Invitrogen, Carlsbad, CA, USA), and the slides rinsed in 0.1 M Tris-buffered saline for 10 min. After blocking non-specific staining with 10% donkey serum in Tris-buffered saline containing 0.3% Triton X-100 for 1 h at room temperature, sections were incubated overnight at 4°C in Tris-buffered saline, Triton containing 5% donkey serum and rat anti-CD45 (1:500, catalogue number CBL1326, Chemicon International, Temecula, CA, USA), rat anti-CD68 (1:1000, catalogue number MCA 1957, ABD Serotec, Kindlington, Oxford, UK), rat anti-CD3 (1:400; BD Pharmingen/M070025) or rat anti-platelet epithelial cell adhesion molecule-1 (1:500, catalogue number 550274, BD Pharmingen, San Jose, CA, USA), rabbit anti-αv integrin (1:1000, catalogue number AB1930; Chemicon), rabbit anti-β3 integrin (1:1000, catalogue number AB1932, Chemicon), rat anti-α5β1 integrin (1:100, catalogue number MAB1984, Chemicon), or rabbit anti-laminin immunoglobulin G (1:300, catalogue number L9393, Sigma). As a control, purified rat or rabbit immunoglobulin G (IR-RB-IGG, IR-IB-IGG, Innovative Research, novi, MI, USA) was used at the same concentrations instead of the primary antibody. Next, the sections were incubated in Tris-buffered saline-Triton containing 5% donkey serum and 1:500 of appropriate secondary antibodies (donkey tetramethylrhodamine isothiocyanate-conjugated Fab’ fragments, Invitrogen; Alexa594) for 1 h at room temperature. Finally, the sections were cover-slipped with antifade Gel/Mount aqueous mounting media (Southernbiotech, Birmingham, AL, USA). In between steps, sections were washed 3 × 10 min in Tris-buffered saline. To measure the volume of tissue loss as determined by laminin staining in the horizontal sections, the sections were processed using a Vectastain ABC-diaminobenzidine staining protocol according to the manufacturer’s instructions (Vector Labs, Burlingame, CA, USA).

Five sham-operated mice received a laminectomy only and were analysed 7 days later. Compared to three normal uninjured mice, they did not show any effects on microglial activation (CD68) at the epicentre, a very sensitive indicator of damage, or differences in LEA-labelled blood vessels. These two groups were combined and used for normalizing all histological measurements.

Quantitative measurements and statistical analyses

Sections were examined using a Leica DMIRE2 brightfield and fluorescence microscope and images digitized with an attached Spot RTKE camera (Diagnostics Inc., Sterling Height, MI, USA). Eriochrome cyanine (myelin), CD45 and CD68 staining through the entire plane of the transverse sections was digitized using a 5 × objective and the area occupied by the staining calculated for the three sections per rostrocaudal level by using the threshold feature of Scion Image software (Scion Corporation, Frederick, MD, USA). This area measurement has been shown to be the least variable and most time-efficient method across different types of CNS injuries, including spinal cord injury, and between users for quantifying activated microglia/macrophages (Donnelly et al., 2009). Also as reported, single microglia/macrophages were too difficult to discern at the injury site. For analysis of LEA-labelled blood vessels, images of the dorsal column and adjacent grey matter and of the ventrolateral funiculus and adjacent grey matter were taken at the injury epicentre and at 1 mm rostral and caudal to it using a 20 × objective. This counts the microvessels relevant to function, as the region represents the lesion penumbra and not the developing fibrotic lesion core which has few LEA-perfused blood vessels (Benton et al., 2008). To provide a measure of the number of blood vessels, the number of LEA-labelled vessels intersecting 100-µm spaced horizontal (five) and vertical (six) lines were counted in each image. To determine the volume of the heterodomain, every fifth horizontal section was stained for laminin using 3,3′-diaminobenzidine as substrate, the heterodomain was circled in each section as well as the outline of the entire 10-mm length of the spinal cord segment using Neurolucida software (MBF Bioscience, Williston, VT, USA). The software calculated the volume based on the section interval and the area per section and this was expressed as a percentage of the total 10-mm segment.

Transendothelial migration assay

The leukocyte transmigration assay was performed as previously described (Sithu et al., 2007). Briefly, human aortic endothelial cells (BioWhittaker, Walkersville, MD, USA) were plated on permeable filters in Transwell culture plates (Costar, Cambridge, MA, USA) at 4 × 104 cells per well. They were grown for 72–96 h to reach confluency at 37°C in Dulbecco's modified Eagle's medium Ham's F-12 (DMEM F-12; BioWhittaker) plus endothelial growth medium-2 SingleQuots™ supplement (BioWhittaker) which contains hydrocortisone, hEGF, foetal bovine serum, vascular endothelial growth factor, hfgf-B, R3-insulin-like growth factor-1, ascorbic acid, heparin, and gentamicin/amphotericin-B and 10–20% foetal calf serum. Afterwards, the inner chamber of the Transwell system was loaded with 2 × 105 monocytic THP-1 cells (human acute monocytic leukaemia cell line; American Type Culture Collection, Manassas, VA, USA) on endothelial cells that were preincubated with or without 15 ng/ml of tumour necrosis factor-α (PeproTech, Rocky Hill, NJ) for 18 h at 37°C. C16 peptide was added into the medium of the inner chamber at 25, 50, 100, 200, 400 and 600 µm, and SP3 was added at 600 µm. Peptides were centrifuged to remove any precipitates before addition. The vehicle served as the control. THP-1 cells were allowed to transmigrate for 6 h at 37°C and the number of THP-1 cells that crossed the endothelial cell layer was counted. Values were derived from three wells per concentration and three independent experiments were performed. To assess the functions of integrins, azide-free blocking antibodies against human αvβ3 (catalogue number MAB1976Z, Chemicon; also known as LM609) (Cheresh and Spiro, 1987) and human α5 (catalogue number MAB1956Z, Clone P1D6 Chemicon) (Wayner et al., 1988) integrin were added in a separate experiment without C16. To test angiopoietin-1’s ability to increase endothelial cell maturation and resistance to transmigration under pathological conditions, endothelial cells grown on transwells were incubated overnight in regular growth medium containing different concentrations of angiopoietin-1. Angiopoietin-1 was removed and the cells incubated with or without 200 ng/ml fibrinogen for 1 h in 100 µl DMEM containing 1% foetal bovine serum (only in upper chamber). After washing once with 100 µl DMEM, THP-1 cells were added (200 000/100 µl) and the assay was continued for 6 h as described above.

Endothelial-monocyte adhesion assay

The leukocyte adhesion to endothelial cells was performed as previously described (Tsakadze et al., 2006). Endothelial cells grown on 96-well culture plates to confluence were incubated with 50 µl of phosphate buffered saline, C16 or SP3 peptide for 1 h at 37°C. THP-1 cells (1 × 105/0.2 ml, in Hank’s balanced salt solution containing 1% glucose, 1% bovine serum albumin and 1 mm each of CaCl2 and MgCl2) were added to the endothelial cells and further incubated for 1 h at 37°C. Conversely, THP-1 cells incubated with phosphate buffered saline or peptides were diluted 1:4 with Hank’s balanced salt solution/glucose/bovine serum albumin/CaCl2/MgCl2 and 250 µl containing 1 × 105 THP-1 cells were added to naive endothelial cell monolayer and incubated for another 1 h. After these incubations, non-adherent THP-1 cells were removed and adherent cells stained with 0.25% rose bengal. After washing, the stain was extracted with ethanol/Dulbecco’s phosphate buffered saline, and the absorbance was measured at a wavelength of 570 nm. The absorbance of naive endothelial cells in the absence of THP-1 cells or peptides was taken as the background and subtracted from the other values.


Statistically significant differences between two groups were determined by one-tailed t-test if an outcome was hypothesized beforehand, two-tailed if not, and paired when an individual animal’s response was compared before and after a treatment. Where applied, the two-tailed and paired t-test are indicated. One-way ANOVA followed by post hoc Tukey analysis was performed to compare groups of three or more. In cases where data were not normally distributed, the Kruskal–Wallis test was used. For analysis of differences in BMS scores between treatment groups over time, two way repeated measures ANOVA with post hoc Tukey was performed. Regression analyses were performed with a confidence interval set at 95%. Tests were performed using Excel (Microsoft Office XP Professional) and Sigmastat (Systat Software Inc., San Jose, CA, USA) software. A P ≤ 0.05 was considered statistically significant. Values for groups are presented as average ± SEM.


Intravenous C16 and angiopoietin-1 treatments greatly reduce locomotor deficits following spinal cord injury

To determine whether endothelial cells in the epicentre penumbra could respond to C16 peptide, αv, β3 and α5β1 integrin expression was assessed in adult female C57Bl/6 mice 1 and 3 days following a T9 contusion. Endothelial cells of perfused, LEA-labelled, blood vessels at the injury epicentre indeed had αvβ3 and α5β1 integrins at those times (Fig. 1). We determined effective doses of C16 and angiopoietin-1 using white matter sparing at the epicentre as a measure because that correlates with locomotor function (Basso et al., 2006; Li et al., 2006). Daily intravenous injections of C16 over 7 days at a dose of 30, 100 or 300 µg showed that 100 µg/day provided significantly better protection of white matter than the 30 µg dose and that the 300 µg/day dose did not further improve the outcome. Therefore, we used 100 µg/day C16 for subsequent experiments. Angiopoietin-1 was equally more effective compared to vehicle at 30, 100 and 300 µg/day. Another angiopoietin-1 mimetic has been shown to protect intestinal microvascular endothelial cells against radiation-induced death in adult mice injected intravenously at 100 µg/day (Cho et al., 2004). We chose 100 µg/day angiopoietin-1 for subsequent experiments, being within the maximally effective dosing range.

Figure 1

αvβ3 integrin is present on blood vessels after spinal cord injury. Twenty-four hours after a contusive spinal cord injury, immunostaining for αv (A), β3 (D) or α5β1 (G) integrin is seen in the penumbra at the epicentre on blood vessels identified by intravenous injection of LEA (B, E, H, respectively). The XZ and YZ views of these confocal images (C, F, I) confirm the co-localization of LEA and the integrins. Some neurons also stain for the integrins (arrows) that disappeared at the injury epicentre (not shown). Three days post-injury, the injury penumbra shows αv (J), β3 (K) or α5β1 (L) integrin staining in some neurons (arrows) and a few blood vessels, but staining for both αv and β3 or α5β1 integrin is not seen in many other cells, including the numerous inflammatory cells expected in these injured tissues. Scale bars are indicated.

C16 was first tested alone for its rescuing effects on overground locomotion using the BMS (Basso et al., 2006). Seven days following spinal cord injury and daily intravenous injections with C16 started immediately post-injury, the BMS score was 4.3 ± 0.3 compared to 1.8 ± 0.3 with vehicle or 0.7 ± 0.3 with SP3 control peptide (Fig. 2A).

Figure 2

An intravenous C16 plus angiopoietin-1 treatment provides superior and lasting improvement in locomotor function following spinal cord injury in mice. (A) Daily intravenous injections with C16 over 14 days (solid squares, n = 9) reduce locomotor deficits (measured by BMS), following a T9 contusion in mice compared to vehicle control injections (open squares, n = 8). Injections started 4 h post-injury and lasted 14 days (black horizontal bar). The benefit lasted beyond termination of the treatment and mice reached a score of 5 (horizontal line), indicating weight bearing and plantar stepping. Other groups of mice were analysed for histology at 7 days, showing that C16 (open triangle, n = 8) also causes better outcomes compared to vehicle (closed circle, n = 13) or SP3 peptide (open circle, n = 8) controls. Injections in these 7-day mice were started immediately following injury. (B) Daily intravenous injections over the first 7 post-injury days with angiopoietin-1 (solid triangle, n = 14) or angiopoietin-1 plus C16 (open diamonds, n = 10) greatly improve BMS scores compared to vehicle (open squares, n = 13). Injections started 4 h post-injury. The vehicle groups in (A and B) were not statistically different. (A and B) were analysed by two-way repeated ANOVA with post hoc Tukey. *P < 0.05; **P < 0.01; ***P < 0.001 compared to vehicle or as indicated by lines. BMS sub-scores (max = 11), which are particularly sensitive to fine motor function, showed that the group treated with C16 plus angiopoietin-1 was significantly better than the other treatment groups at 1 week post-injury (C) and over the last 2 weeks (D). These time points indicate the neuroprotective and lasting nature of the treatment, respectively. The BMS scores are not on a linear scale with regards to functionality. Therefore, mice were separated into different categories of hind-limb function where BMS scores of 0–2: paralysed, 3–4: some ankle movement, 5–6: some coordination and stepping, 7–9: high functioning with consistent coordinated stepping, where 9 is normal. The group treated with C16 plus angiopoietin-1 had more high functioning mice at 1 week post-injury (E) and over the last 2 weeks (F). (E and F) were tested by the z-test for binomial proportion. *P < 0.05; **P < 0.01, highest P-value in the given category of C16 plus angiopoietin-1 compared to the other three treatments. Ang-1 = angiopoietin-1; SCI = spinal cord injury.

To determine whether C16 would provide lasting benefits over 6 weeks, other contused mice received daily injections with vehicle or 100 µg C16 over the first 14 post-injury days, starting 4 h post-injury. The 4-h period was chosen because this is the time in which most human spinal cord injury cases are diagnosed and within which time intravenous treatments could start. C16-treated mice had a higher BMS score than vehicle-treated mice starting at 7 days (Fig. 2A) and continuing until Week 6 (5.4 ± 0.4 versus 4.3 ± 0.4). A score of 5 or higher indicates weight-bearing and consistent plantar stepping, showing that C16 treatment provided substantial functional improvement. We also tested whether a second C16 treatment during the chronic injury phase would further improve outcomes. Mice injected with C16 during the first week had higher BMS scores than vehicle-treated mice (5.8 ± 0.6 versus 1.4 ± 0.4 at 6 weeks; P < 0.001; data not shown). However, the second C16 (or vehicle) treatment from 28–34 days post-injury did not modify the BMS scores in mice injected with vehicle or C16 over the first week (P > 0.1; paired t-test; n = 4 for each of four groups).

Finally, C16 was combined with angiopoietin-1 to assess potential additive and lasting effects over a 6-week post-injury period. Mice received daily intravenous injections of vehicle, angiopoietin-1, or C16 plus angiopoietin-1 for 7 days, starting 4 h following contusion. Two separate experiments had similar results and were combined. At 8–9 days post-injury, the BMS score was higher with angiopoietin-1, and C16 plus angiopoietin-1 than with vehicle treatment (Fig. 2B). These differences were maintained after termination of the treatment, with all groups reaching a plateau. The average of the C16 plus angiopoietin-1 group was not significantly different from the angiopoietin-1 or C16 groups when tested by two-way repeated ANOVA and post hoc Tukey. However, separately, t-tests at the individual post-injury times showed significant differences. Also, the BMS sub-score (Basso et al., 2006), which is more sensitive to detect differences in fine lomocomotor skills, was better with the C16 plus angiopoietin-1 combination treatment at 1 and 5–6 weeks post-injury (Fig. 2C and D). The BMS is non-linear and to get better insight into the functionality of the mice, the scores were grouped in functional categories and compared for significant differences by z-test for binomial proportions. High-functioning mice with a score of 7 and above walk normally to the untrained eye. This occurred in 50% of the mice treated with C16 plus angiopoietin-1 at 1-week post-injury (Fig. 2E), versus 0% with vehicle (P < 0.001), 0% with C16 (P < 0.001) and 14% with angiopoietin-1 (P < 0.05). During the last 2 weeks (Fig. 2F) this was seen in 70% versus only 0% (P < 0.001), 11% (P < 0.005), and 29% (P < 0.05), respectively. These results reveal superior early and lasting effects on locomotor function by intravenous injection of an αvβ3 integrin-binding peptide together with the Tie2 ligand angiopoietin-1.

C16 and angiopoietin-1 reduce white matter loss following spinal cord injury

We also determined, in the above groups and additional ones, whether the treatments resulted in better white matter sparing at the epicentre, as this correlates with locomotor function. In an initial set of injured mice, horizontal sections through the spinal cord were used. Those treated with C16 for 7 days had a smaller lesion volume compared to mice treated with vehicle (Fig. 3). In all subsequent experiments, transverse (cross-sectional) sections were used for more reliable measurements of white matter sparing and enabling multiple histological stains in series of adjacent sections. The injury epicentre was determined by the minimum area of myelin along the spinal cord axis. In mice treated for 7 days with vehicle (Fig. 4B), only some of the ventral white matter remained present. With C16, much of the ventral and lateral white matter was spared (Fig. 4C). The area of white matter at the epicentre was greater with C16, angiopoietin-1 or C16 plus angiopoietin-1 than with vehicle at 7 and 42 days post-injury (Fig. 4D). As expected, epicentre white matter sparing correlated well with BMS scores at 7 and 42 days (Table 2). However, C16 plus angiopoietin-1 did not rescue more white matter than either C16 or angiopoietin-1 alone, suggesting that white matter sparing alone did not account for the better locomotor function seen for C16 plus angiopoietin-1. The white matter area was not significantly different between 7 and 42 days post-injury, irrespective of treatment group. Together with the locomotor improvement seen as early as 7 days, this suggests that C16 and angiopoietin-1 affect mechanism(s) such as blood vessel loss and/or inflammation during the acute and early sub-acute phase. This is also supported by the finding that one C16 injection (1 × C16) immediately following spinal cord injury rescued white matter at 7 days post-injury compared to 1 × vehicle (Fig. 4D; P < 0.001; two-tailed t-test) and improved BMS scores (3.8 ± 0.6 versus 2.0 ± 0.4; P < 0.05; two-tailed t-test). Also, the 1 × C16 treatment was as effective as the 7 day treatment in rescuing white matter and locomotor function at 7 days (not significant, P = 0.06, 0.13, respectively). C16 concentrations are expected to be much lower at 7 days after a single injection on Day 1 than with daily injections.

Figure 3

C16 reduces the volume of tissue loss 7 days after spinal cord injury. (A) Horizontal section shows a heterodomain characterized by deposits of laminin by invading mesenchymal cells which replace lost spinal cord tissue in a mouse contused 7 days before at T9. (B) C16 treatment over 7 days reduced the size of the heterodomains. With vehicle treatment, tissue loss occurred over half the diameter of the spinal cord, whereas with C16 treatment damage appeared to be less in the outer regions of the spinal cord, including white matter tracts. Note the preservation of blood vessels identified by laminin-positive basement membrane in the injury penumbra and the caudal and rostral (arrow pointing to ‘R’) regions which look normal. (C) The total volume of the laminin-positive heterodomains showed a 43% reduction after C16 treatment compared to vehicle treatment. Data are mean ± SEM; vehicle, n = 10, C16, n = 11. **P < 0.01. Lf = lateral funiculus; dh = dorsal horn; dc = dorsal column.

Figure 4

C16 and angiopoietin-1 treatments reduce white matter loss following spinal cord injury. Compared to a sham-operated (laminectomy only) mouse with normal white matter (A), a mouse injected intravenous with vehicle over 7 days (B) has extensive loss of white matter as shown by myelin staining with eriochrome cyanine in transverse sections at the injury epicentre. (C) Injections of C16 (shown here) or angiopoietin-1 or their combination increased the amount of spared white matter. Scale bar in (A) is 200 µm for (A–C). (D) The total area of white matter at the injury epicentre (as a percentage of sham) shows that C16, angiopoietin-1, and C16 plus angiopoietin-1 treatments improve white matter sparing compared to vehicle or SP3 controls at 7 and 42 days post-injury. A single bolus of C16 given immediately following the injury (1× C16) also results in white matter sparing seen at 7 days post-injury. Data are mean ± SEM. Group numbers are indicated in the bars. *P < 0.05; **P < 0.01; ***P < 0.001 compared to vehicle. Ang-1 = angiopoietin-1; SCI = spinal cord injury.

View this table:
Table 2

Regression analysis for sub-acute and chronic mice following T9 contusion

7 day
42 day
  • Data from individual mice at 7 and 42 days following contusive spinal cord injury that had all values for BMS, epicentre white matter (WM) area, CD45 and CD68 area, and LEA-labelled blood vessel counts were combined. The R2 values are shown as an indication of the potential contribution of one measure to the outcome of the other. A value of 1 would indicate a perfect correlation.

  • *P < 0.01, **P < 0.01, ***P < 0.001.

C16 and angiopoietin-1 improve vascularity at the injury site

To determine the effects of the treatments on endothelial cell survival, we measured the presence of perfused endothelial cell-lined blood vessels in the injury penumbra, which is relevant to tissue preservation. This was achieved by injecting an endothelial cell-binding fluorescein isothiocyanate-conjugated LEA lectin intravenously 30 min before euthanasia (Fig. 5A–C). At 24 h, 7 days and 42 days following contusion, C16- and angiopoietin-1-treated mice had more LEA-labelled blood vessels in a spinal cord segment from 1 mm rostral to 1 mm caudal from the epicentre than vehicle-treated mice (Fig. 5D). At 24 h, C16 and angiopoietin-1 rescued a proportion of injured LEA-labelled blood vessels (44% or 40%, both P < 0.05 versus vehicle, 27%; Fig. 5D). C16-treated mice had more blood vessels at 7 days than at 24 h (P < 0.01), suggesting that C16 stimulated angiogenesis, as it does in vitro (Ponce et al., 2001, 2003). This was not seen with angiopoietin-1, consistent with the fact that it does not induce endothelial cell proliferation during angiogenesis but plays a role in endothelial cell survival and maintaining integrity of the newly formed vasculature (Thurston et al., 1999, 2000, 2005; Gale et al., 2002; Mochizuki, 2009). At 7 days, the LEA values measured in the penumbra of the epicentre sections were also increased in the C16- or angiopoietin-1-treated groups (not shown), suggesting that the increase was not just due to a reduced lesion size. In a separate experiment, the LEA-labelled blood vessel counts at 7 days post-injury were not different between groups treated for 7 days with C16 (40 ± 5%) or C16 plus angiopoietin-1 (40 ± 4%) and both higher than with vehicle (26 ± 2%; P < 0.05). At 42 days, the number of vessels seen with a 14-day C16 treatment was greater than with a 7-day C16 plus angiopoietin-1 treatment from a subsequent experiment. The 14 days was initially chosen because demyelination and maximal inflammation lasts up to 2 weeks (Donnelly and Popovich, 2008). Subsequently, we shortened the treatment for the 6-week experiments because this might reduce the potential for side-effects. This finding raises the possibility that some of the new blood vessels were not maintained following termination of the 7-day C16 plus angiopoietin-1 treatment and/or the longer C16 treatment resulted in even more blood vessel growth or better preservation.

Figure 5

C16 and angiopoietin-1 treatments rescue blood vessels after spinal cord injury. To evaluate the extent of rescue of perfused blood vessels, LEA was injected intravenously 30 min before histological processing. (A) LEA-labelled blood vessels in sham-operated mice had a normal appearance. The box in the inset schematic represents the region presented in A–C. (B) 7 days following spinal cord injury, mice treated for 7 days with vehicle show few blood vessels, whereas (C) C16-treated mice have many more. Scale bar in (C) is 100 µm for (A–C). (D) A time course shows a reduction in the number of blood vessels in the injury penumbra in vehicle-treated mice by 24 h following spinal cord injury. Single injections of C16 or angiopoietin-1 rescues blood vessels after 24 h and with 7-day injections the numbers remain higher than with vehicle at the 7-day post-injury time point. This 7-day treatment with C16 increases the number of vessels compared to 24 h. At 42 days, mice treated for 14 days with C16 and those treated for 7 days with angiopoietin-1 or C16 plus angiopoietin-1 still had more blood vessels than those treated with vehicle. At both 7 (E and F) and 42 days (G and H), the number of LEA-labelled blood vessels correlated with locomotor performance as measured by BMS (E and G) and spared white matter at the epicentre (F and H). Data are mean ± SEM. Sham and normal mice, n = 8; 1 day vehicle, n = 5; 1 day C16, n = 5; 1 day angiopoietin-1, n = 5; 7 day vehicle, n = 13; 7 day C16, n = 8; 7 day angiopoietin-1, n = 5; 6 week vehicle, n = 22; 6 week C16, n = 12; 6 week angiopoietin-1, n = 7, C16+angiopoietin-1, n = 6. The numbers of mice are not the same as in Figs 2 and 4, as not all mice received LEA injections and not all mice with LEA injections were tested for BMS. *P < 0.05; **P < 0.01; ***P < 0.001 compared to vehicle or as indicated by the vertical line at 7 days Ang-1 = angiopoietin-1; SCI = spinal cord injury; dh = dorsal horn, dc = dorsal column; open squares = vehicle; filled squares = C16; filled triangles = angiopoietin-1; open diamonds = C16 plus angiopoietin-1.

It was reasonable to propose that improved vascularity might rescue white matter leading to improved function. Consistent with this idea was the finding that at 7 and 42 days post-injury, the number of LEA-labelled blood vessels correlated with BMS scores (Fig. 5E and G) and white matter sparing (Fig. 5F and H). However, the finding that at 7 days, C16-treated mice had similar BMS scores as the angiopoietin-1 treated mice, despite greater vascularity at 7 days, suggests that increased vascularity alone is not sufficient to explain the treatment-induced improvements. Also, the extent of correlation between blood vessels and BMS scores or white matter (R2 values) was reduced from 7 and 42 days post-injury (Fig. 5E versus Fig. 5G and Fig. 5F versus Fig. 5H). This suggests that additional non-vascular mechanisms not directly related to the C16 or angiopoietin-1 treatments (for example, plasticity, remyelination) contribute to outcomes during the chronic post-injury phase.

C16 and angiopoietin-1 reduce inflammation following spinal cord injury

Inflammation contributes to tissue loss after spinal cord injury (Fleming et al., 2006; Donnelly and Popovich, 2008). We used CD45 as a marker for extravasated leukocytes and CD68 as a marker for activation of resident microglia and extravasated macrophages. The latter are major contributors to demyelination (Shuman et al., 1997; Popovich et al., 1999, 2002; Gris et al., 2004). At 7 days post-injury and vehicle treatment, CD45-positive cells were abundant throughout the spinal cord at the epicentre (Fig. 6A) and several millimetres rostral and caudal to it. This infiltration was markedly attenuated by C16 (Fig. 6B). Sham-operated or normal mice had essentially no immunostaining (not shown). Similar results were seen with CD68 immunostaining (Fig. 6C and D). Peripheral leukocyte numbers were within the normal range in both the vehicle and C16 groups 7 days post-injury (Table 3), suggesting that C16 reduces extravasation into the spinal cord. The area of immunostaining was quantified at 1-mm distances through a spinal cord segment spanning 3 mm rostral to 3 mm caudal from the epicentre. This showed that C16, angiopoietin-1 or C16 plus angiopoietin-1 reduced inflammation at all post-injury times (Fig. 6). As early as 24 h post-injury and C16- or angiopoietin-1 treatment reduced the CD68 area by ∼25% compared to vehicle (Fig. 6E and F). This suggests that targeting very early mechanisms reduces secondary pathology following spinal cord injury, including vascular dysfunction and subsequent detrimental inflammation. However, although the single C16 injection (1 × C16) reduced CD68 (P < 0.05 versus 1 × vehicle, two-tailed t-test) it did not significantly affect CD45 (P = 0.12, two-tailed t-test). Moreover, the 1 × C16 treatment was less effective than the 7-day treatment in reducing CD45 and CD68 at 7 days post-injury (Fig. 6E and F), suggesting that 7-day treatments might have better neuroprotective effects. Inflammation at 7 or 42 days post-injury correlated with white matter loss and reduced BMS scores (Table 2), suggesting that the reduced inflammation following C16, angiopoietin-1 and C16 plus angiopoietin-1 treatments contributed to better outcomes. The CD45 and CD68 area was significantly less for C16 plus angiopoietin-1 than with C16 or angiopoietin-1 alone, suggesting that C16 and angiopoietin-1 target partially overlapping inflammatory mechanisms. In fact, treatment with angiopoietin-1, but not C16, reduced CD3+ T cells seen at 42 days (Fig. 7), a time when these cells were expected to be present (Kigerl et al., 2006).

Figure 6

C16 and angiopoietin-1 treatments reduce inflammation after spinal cord injury. (A) A transverse (cross) section through the entire spinal cord at the injury epicentre stained for CD45 shows extensive infiltration of leukocytes at 7 days post-injury in a mouse injected intravenously with vehicle over the same period. (B) Injections of C16 (shown here) or angiopoietin-1 or their combination greatly reduced the infiltration. (C and D) CD68, a marker for activated microglia and macrophages, was similarly reduced by C16. No sham or normal mice are shown as they had no immunostaining. Scale bar in (D) is 200 µm for (A––D) and in the higher magnification insets, 50 µm. (E) The total cross-sectional area of CD45-positive cells at the injury site shows that C16, angiopoietin-1 or C16 plus angiopoietin-1 treatments reduce infiltration compared to vehicle or SP3 controls at 7 and 42 days post-injury but not significantly at 24 h. The area is the sum of areas at 1 mm distances within a segment from 3 mm rostral to 3 mm caudal to the epicentre and is shown as a percentage of the vehicle group within the experiment. (F) The cross-sectional area of CD68-positive cells at the injury site shows that C16, angiopoietin-1 and C16 plus angiopoietin-1 treatments reduce microglia/macrophage activation at all post-injury times. Data are mean ± SEM. Group numbers are as in Figure 4 plus n = 5 each at 24 h. *P < 0.05; **P < 0.01; ***P < 0.001 compared to vehicle or as indicated by the horizontal line. Ang-1 = angiopoietin-1; SCI = spinal cord injury.

Figure 7

Angiopoietin-1 but not C16 reduces chronic presence of T cells after spinal cord injury. At 42 days following spinal cord injury, the number of CD3-positive T cells was counted in a spinal cord segment consisting of 3 mm on either side of the epicentre. Only daily intravenous treatments containing angiopoietin-1 over the first post-injury week reduced the chronic presence of T cells. The addition of C16 to angiopoietin-1 did not result in a further decrease. Data are mean ± SEM and calculated as a percentage of vehicle; vehicle n = 10, C16 n = 4, angiopoietin-1 = 7, C16 plus angiopoietin-1 n = 5. *P < 0.05 compared to vehicle. Ang-1 = angiopoietin-1; SCI = spinal cord injury.

View this table:
Table 3

C16 does not affect systemic leukocyte counts

 Normal rangeVehicleVehicleVehicleC16C16
12 312
Leukocytes total (K/µl)1.8–10.71.641.845.202.123.12
Neutrophils (K/µl)0.1–
Lymphocytes (K/µl)0.9–9.31.331.362.221.692.19
Monocytes (K/µl)0.0–
Eosinophils (K/µl)0.0–
Basophils (K/µl)0.0–
Erythrocytes (M/µl)6.36–9.425.804.577.777.837.51
Thrombocytes (K/µl)592–2972772261135572563
  • Mice were injected daily with vehicle (n = 3) or C16 (n = 2) for 7 days after a contusion at T9 and 1 ml blood drawn from the heart for blood cell analysis. (K/µl) = 1000 per microlitre; (M/µl) = 1 000 000 per microlitre. Low red blood cell and thrombocyte counts are probably due to bleeding caused by surgery.

Angiopoietin-1 but not C16 reduces blood vessel leakiness following spinal cord injury

We investigated additional potential differences between C16 and angiopoietin-1 that might contribute to better outcomes for C16 plus angiopoietin-1 treatment. We assessed their effects on increased permeability, as that may contribute to tissue loss following spinal cord injury. Angiopoietin-1 was known to reduce vascular permeability under inflammatory conditions (Thurston et al., 1999, 2000). Blood-spinal cord-barrier permeability was measured by the amount of luciferase that extravasated into the injured spinal cord after intravenous injection 30 min before analysis. With daily angiopoietin-1 injections following the contusion, luciferase values were not different from vehicle injections at 24 h following spinal cord injury (79 ± 16 versus 100 ± 46%) but were reduced to 61 ± 13% of vehicle (100 ± 17%) at 72 h (P < 0.05). Daily C16 injections had no significant effect at either time point (76 ± 10 and 77 ± 18%, respectively). This suggests that angiopoietin-1 treatment reduces pathological permeability during the sub-acute phase, possibly explaining why it can improve function despite lower vascularity than with C16 treatments.

C16 reduces αvβ3-dependent monocyte transmigration in vitro

To determine the cellular and integrin targets of C16 that might control leukocyte infiltration, we measured monocyte transmigration across an endothelial cell layer in a two-compartment culture system. Under control conditions (vehicle or 600 µM SP3 peptide), about 5% of monocytes crossed the endothelial cell barrier. C16 reduced transmigration by 48% at 600 µM (Fig. 8A). When the endothelial cells were stimulated with pro-inflammatory tumour necrosis factor-α, SP3 had no effect but C16 reduced the number of transmigrated cells already at 67 µM. Without C16 in the media no inhibition was observed, even if endothelial cells or monocytes were preincubated with C16 and then washed (data not shown), suggesting that C16 affects mechanisms during adhesion and/or transmigration. The latter is more likely, as C16 had no effect on monocyte adhesion to endothelial cells (Fig. 8B). Blocking αvβ3 integrin antibodies reduced monocyte transmigration by 36% but α5 integrin antibodies did not (Fig. 8C), suggesting that C16 reduces transmigration by blocking αvβ3 integrin. Lastly, in keeping with its known anti-inflammatory and anti-permeability role, angiopoietin-1 reduced monocyte transmigration across endothelial cells even when stimulated by fibrinogen, which is known to increase endothelial cell permeability (Fig. 8D).

Figure 8

C16 and angiopoietin-1 reduce monocyte transmigration across endothelial cells in vitro. (A) The number of monocytes (THP-1 cells) that had migrated across a monolayer of endothelial cells to a separate compartment in transwells was reduced by C16 in the absence or presence of the pro-inflammatory cytokine tumour necrosis factor-α. SP3 control peptide had no significant effect. Values are expressed as a percentage of vehicle (+ SEM). (B) Pre-incubation of endothelial cells or THP-1 cells with C16 did not affect adhesion of THP-1 cells to endothelial cells under continuing presence of C16 for one hour as shown by optical density measurements of Bengal Rose staining at 570 nm. (C) Transmigration was αvβ3 dependent as shown by blocking antibodies in the presence of tumour necrosis factor-α. The α5β1 integrin, which can also bind C16, was not involved, as α5 antibodies failed to affect transmigration. The extent of reduced transmigration was much less with the antibody than with C16. (D) Pre-incubation of endothelial cells with 100 ng/ml angiopoietin-1 reduced subsequent transmigration of monocytes and also prevented an increase in transmigration induced by pre-incubation with pro-inflammatory fibrinogen (FG, 200 ng/ml). Data are mean ± SEM, n = 3 each. *P < 0.05; **P < 0.01; ***P < 0.001 compared to vehicle or control. Ang-1 = angiopoietin-1.


The intravenous angiopoietin-1 and C16 treatments seem to target several vascular-related mechanisms such as endothelial cell death and dysfunction, and leukocyte extravasation which contribute to secondary tissue damage following spinal cord injury (Mautes et al., 2000). These neuroprotective effects are most likely achieved directly through endothelial cells by ligand binding to their Tie2 and integrin receptors. First, in keeping with its pro-survival role for endothelial cells (Fujikawa et al., 1999; Harfouche et al., 2002; Carmeliet, 2003; Mochizuki, 2009), angiopoietin-1 reduced loss of blood vessels at the injury site as early as 24 h post-spinal cord injury, probably reducing ischaemia and subsequent tissue loss. In fact, angiopoietin-1 treatment also permanently rescued white matter and improved locomotor function, both of which correlated with the number of perfused blood vessels. As angiopoietin-1 acts only through Tie2, which is only expressed by endothelial cells (Dumont et al., 1993; Suri et al., 1996), the current data add direct evidence that rescue of blood vessels results in improved outcome following spinal cord injury. This further strengthens the idea that vascular protection would be an important and viable component of a therapeutic strategy following CNS injuries (Tator and Fehlings, 1991; Zhang and Guth, 1997; Loy et al., 2002; Hall and Springer, 2004; Chen et al., 2007). In fact, treatments with nimodipine in the presence of systemic vasopressors can maintain spinal cord blood flow and, thereby, axonal conduction after spinal cord injury in rats (Fehlings et al., 1989; Guha et al., 1989). Treatments with tirilazad, antioxidants or selenium also maintain spinal cord blood flow and reduce leakage following experimental contusive spinal cord injury, and are likely to reduce lipid peroxidation that occurs in microvessels of the injured cord (Hall and Wolf, 1986; Hall, 1988; Hall et al., 1994; Carrico et al., 2009). Moreover, inhibition of transient receptor potential cation channel M4 or sulfonylurea receptor-1, which show increased expression predominantly by endothelial cells (and neurons) following spinal cord injury and involve ion channel mechanisms, reduces microvascular degeneration, white matter loss and functional deficits (Simard et al., 2007; Gerzanich et al., 2009).

Angiopoietin-1 also reduced permeability at 72 h, most likely via its capacity to preserve the integrity of endothelial cell tight junctions under pathological conditions (Thurston et al., 1999; Fiedler and Augustin, 2006). In vitro, angiopoietin-1 treatment reduced monocyte transmigration. In vivo, the reduction by angiopoietin-1 in overall inflammation coincided with reduced microglia/macrophage activation and infiltration. The latter also likely contributed to augmented white matter sparing at the injury epicentre as microglia/macrophages are known to be important for white matter damage (Blight, 1985; Popovich et al., 1999, 2002; Weaver et al., 2005). The extent of white matter sparing is directly related to locomotor function (Basso et al., 2006; Li et al., 2006), as again shown here using regression analyses.

C16, like angiopoietin-1, rescued blood vessels at the injury epicentre at 24 h following spinal cord injury. In vitro and in vivo in the chick chorioallantoic membrane assay, C16 acts as an agonist and binds only to αvβ3 and α5β1 integrins (Ponce et al., 1999, 2001, 2003). These integrins are well known to promote endothelial cell survival (Brooks et al., 1994). Therefore, it is likely that C16 promoted endothelial cells survival by activating their αvβ3 or α5β1 integrins, which are present on these cells following spinal cord injury. C16 represents a functional laminin sequence and might mimic the basement membrane attachment necessary for normal endothelial cell survival (Hynes, 1992; Giancotti and Ruoslahti, 1999), which is disrupted after spinal cord injury (Goodman et al., 1979; Koyanagi et al., 1993). C16 also seemed to have a second, pro-angiogenic effect. The αvβ3 integrin has a known role in promoting angiogenesis (Brooks et al., 1994) and C16 promotes angiogenesis in vivo in the chick chorioallantoic membrane assay (Ponce et al., 1999, 2001). This may explain why we find that the number of LEA-labelled blood vessels increased with C16 treatment between 24 h and 7 days following spinal cord injury. These new blood vessels were most likely to be beneficial, because the number of blood vessels correlated with both spared white matter and locomotor function at 7 days post-injury. We did not observe more LEA-labelled blood vessels at 7 days in the vehicle-treated mice, consistent with our previous finding in mice (Benton et al., 2008). Angiogenesis normally occurs at that post-injury time in mice (Whetstone et al., 2003; Benton et al., 2008), suggesting that without treatment few new vessels are normally perfused. Collectively, these data suggest that C16 is an αvβ3 agonist for endothelial cells and promotes therapeutic angiogenesis following spinal cord injury. Others have also found evidence for the beneficial effects of enhanced angiogenesis following spinal cord injury using vascular endothelial growth factor treatments (Widenfalk et al., 2003; Liu et al., 2009). However, it should be noted that vascular endothelial growth factor treatment can exacerbate tissue loss and vascular permeability following spinal cord injury (Benton and Whittemore, 2003; Patel et al., 2009).

C16 also seems to have a specific integrin-mediated and previously unknown anti-inflammatory activity. In vitro, C16 reduced monocyte transmigration across an endothelial cell layer to the same extent as blocking antibodies against αvβ3 but not α5 integrins. We observed αv or β3 immunoreactivity on endothelial cells but not on infiltrated cells following spinal cord injury, suggesting that αvβ3 integrin on endothelial cells plays a role in the transmigration. As C16 did not affect monocyte binding to endothelial cells in vitro and C16 was required during the transmigration phase of the assay, this suggests that αvβ3 integrin plays a dynamic role in reducing leukocyte entrance into the spinal cord. C16 is known as an αvβ3 agonist for angiogenesis in the chick chorio-allantoic membrane assay (Ponce et al., 1999, 2001). However, the fact that the αvβ3-blocking antibody also reduces transmigration suggests that C16, by binding to αvβ3, interferes with the binding of a leukocyte ligand required for transmigration. Despite the clear reduction in CD68 staining, we did not directly show effects on monocyte extravasation in vivo, and in the future it will be important to document that our treatments influence immune cell function.

Sub-classes of infiltrating leukocytes are thought to be detrimental following spinal cord injury, including the early arriving neutrophils and activated macrophages (Fleming et al., 2006; Donnelly and Popovich, 2008). General reduction of infiltration of such cells improves tissue sparing and function in rodents (Popovich et al., 1999; Weaver et al., 2005). Recent evidence suggests that certain sub-classes are involved in the healing response and can be beneficial, for example, the M2 type macrophages that are present during the first post-injury week (Kigerl et al., 2009) and certain lymphocytes that invade after 2 weeks (Donnelly and Popovich, 2008). Such cells may in fact release neurotrophic factors that promote cell survival, plasticity and regeneration. The detrimental M1 and beneficial M2 macrophages are both present during the first week when neuroprotection is most needed, suggesting that simple timing of treatments will not resolve this potentially complex issue. On the other hand, our results with the 1× bolus C16 injection show that the BMS scores and white matter sparing is similar to the 7 day treatment. This suggests that the more reduced inflammatory response with the 7 day treatment is not less neuroprotective and that longer treatments may be even better. This is also suggested by the potential regression of blood vessels in mice treated for 7 days with C16 and angiopoietin-1 combined compared to the ones treated with C16 alone for 2 weeks (Fig. 5D). In any case, it will be important to gain a better understanding of the respective mechanisms involved in infiltration of different sub-classes of leukocytes and test the effects of angiopoietin-1 and C16 on them.

Our data suggest that both C16 and angiopoietin-1 treatments regulate endothelial cell survival and leukocyte infiltration independently. Both angiopoietin-1 and C16 rescued endothelial cells at 24 h, when very little inflammation was seen in the cord, compared to later post-injury times, and without significant anti-inflammatory or anti-permeability effects. The reduced inflammation seen at later times could be due to additional effects such as reduced pathological permeability with angiopoietin-1 and reduced transmigration with C16, as also shown in culture. In vitro, C16 blocked transmigration through a well-surviving endothelial cell layer, again suggesting that endothelial survival and inflammation are two separate mechanisms affected by the treatments.

The C16 plus angiopoietin-1 combination treatment provided a superior and remarkable degree of locomotor recovery compared to the individual agents, despite the lack of better white matter sparing. However, C16 plus angiopoietin-1 reduced inflammation at 6 weeks post-injury more than the individual treatments, suggesting that chronic inflammation causes additional locomotor dysfunction besides that caused by tissue loss. This is also suggested by the lower correlation between the vascularity and outcomes at 42 days post-injury than at 7 days. During the first week we did not see more reduced inflammation for C16 plus angiopoietin-1, raising the possibility that the function of the inflammatory cells at the injury epicentre is most important for causing dysfunction of spared white matter tracts of passage. Also requiring further study is the question how C16 and angiopoietin-1 treatment during the first week post-injury would have additive anti-inflammatory effects; that is, how the integrins and Tie2 might cooperate. Angiopoietin-1 is known to down-regulate intercellular adhesion molecule-1, which is necessary for leukocyte binding (Kim et al., 2001), probably by a different mechanism than the αvβ3 integrin-related inhibition of transmigration, perhaps explaining why angiopoietin-1 and C16 act in an additive manner. Among other potential differences worth investigating is angiopoietin-1's unique reduction of permeability and T cell infiltration, whereas only C16 enhanced angiogenesis. We do think that rescue of blood vessels by C16 and angiopoietin-1 during the acute post-injury phase is a critical component underlying the improved functional outcomes. It is also possible that longer treatments with angiopoietin-1 would result in even better stabilization of the vasculature, resulting in even better outcomes. Angiopoietin-1 is well-known for its role in establishing and maintaining vascular maturation, stabilization and integrity (Thurston et al., 1999, 2000, 2005; Gale et al., 2002; Carmeliet, 2003; Mochizuki, 2009).

The current results reveal novel vascular- and αvβ3 integrin-related mechanisms amenable to small peptide targeting, which can cooperate with angiopoietin-1. This new clinically relevant treatment, which targets blood vessels and detrimental inflammation, is probably also relevant to other acute neurological disorders. For example, spinal cord injury and traumatic brain injury (Bramlett and Dietrich, 2007), as well as stroke, are characterized by ischaemia, vascular dysfunction, detrimental inflammation and tissue loss. The intravenous route is readily translatable to a clinical setting and ensures that therapeutic doses are quickly reached. When dealing with acute injuries such as neural trauma and stroke, rapid intervention is probably most efficacious. If also true in humans, the ability to delay the treatment by 4 h after the injury and maintain efficacy will ensure successful treatment of most patients, particularly because the intravenous treatment can be started as soon as a diagnosis of spinal cord injury is made. Our data also suggest that C16 and angiopoietin-1 treatments reduce early post-injury vascular degeneration and detrimental leukocyte infiltration and document that such neuroprotective treatment can be limited to the first week and still have a maximal effect. The finding that a single injection of C16 is also neuroprotective suggests that such treatments could be shorter than 1 week while retaining full efficacy, thus further reducing the potential for detrimental side-effects. The pharmacodynamics and kinetics of C16 are unknown and will have to be addressed during preclinical development. Lastly, angiopoietin-1 affects vascular homeostasis exclusively through the Tie2 receptor which is almost exclusively expressed in endothelial cells (Dumont et al., 1993; Suri et al., 1996), potentially making it an ideal target for pharmacological intravenous treatments. Angiopoietin-1 and C16 act through specific transmembrane receptors, that is, different from the vascular-targeting anti-oxidants and ion channel blocking treatments discussed above, suggesting that additional combination treatments might result in even better outcomes.


Kentucky Spinal Cord and Head Injury Research Trust and National Institutes of Health (NS045734 and RR015576); Norton Healthcare; Commonwealth of Kentucky Challenge for Excellence (to S.R.W., T.H.).


We thank Sheher Sun and Rollie Reid for technical assistance and Christine Nunn, Aaron Puckett, Kim Fentress, William devries and Conner Means for animal care and behavioural testing; and Darlene Burke for advice and help with statistical analyses. Angiopoietin-1 was a generous gift from Regeneron Pharmaceuticals Inc., Tarrytown, NY, USA.


  • Abbreviations:
    Basso Mouse Scale
    αvβ3-binding peptide
    Lycopersicon esculentum agglutinin


View Abstract