All subjects were adult male Sprague–Dawley rats (Charles River Laboratories, Kingston, NY, USA) weighing 250–350 g. Animals were housed three per cage within a temperature-stabilized animal facility with food and water available ad libitum. Animals were maintained on a 12-h light/dark cycle (lights on 07:00 h). All experiments were approved by the Queens College Institutional Animal Care and Use Committee which operates under federal and state animal care guidelines. All experiments conformed to international guidelines on the ethical use of animals, and every effort was made to minimize both the number of animals used and their suffering.
The VEGF used for protein infusions was human recombinantVEGFA165
(a generous gift of Regeneron Pharmaceuticals, Tarrytown, NY, USA). VEGF was stored frozen until use and then diluted in sterile phosphate-buffered saline (PBS) (Sigma-Aldrich, St. Louis, MO, USA) to attain doses of 15 ng/day, 30 ng/day, 45 ng/day, and 60 ng/day when delivered at 0.5 μ
l/h via osmotic minipump. These doses were chosen based on pilot data (not shown) demonstrating no significant effect of human VEGF when infused at a dose of 15 ng/day or lower, as well as data demonstrating that infusion of more than 30 ng/day of mouse VEGF 164 (Croll et al., 2004b
) or 60 ng/day of human VEGF 165 (S. K. Shah and S. D. Croll, unpublished observations) resulted in overt angiogenesis in brain. Because we used human VEGF in the current experiments, no animal was treated with a dose over 60 ng/day in order to avoid the induction of overt angiogenesis. Flt-Fc, an immunoadhesin designed to sequester endogenous VEGF, was used at a dose of 12 μ
g/day to interfere with endogenous VEGF receptor binding (a generous gift of Regeneron Pharmaceuticals). This reagent is a forced dimer of regions 1–3 of fetal liver kinase (receptor) (Flt; VEGF receptor 1) fused to the Fc domain of human IgG (hFc) and dissolved in sterile PBS. It has been previously shown to interfere with VEGF’s effects in vivo
(Holash et al., 2002
). Equimolar hFc was used as a control protein for Flt-Fc. PBS was purchased in powder form, mixed with distilled water, sterilized, and used as a control. All protein reagents were continuously infused starting 5 days before seizure induction, with infusion continuing until the animals were killed.
Pump implantation and protein infusion
Animals receiving protein infusions were anesthetized using 6 mg/kg chlorpromazine injected intraperitoneally followed by 210 mg/kg ketamine (Sigma-Aldrich) administered intramuscularly. The scalp was shaved, cleaned with alcohol, and treated with iodine. Animals were placed into a stereotaxic apparatus and a longitudinal incision was made along the scalp. Two burr holes were drilled and anchor screws (Plastics One, Roanoke, VA, USA) were inserted. A sterile 4 mm cannula (Plastics One), with an attached heat-sealed polyvinyl catheter (Plastics One) containing sterile PBS, was implanted unilaterally into the dorsal hippocampus (3.8 mm posterior and 2.7 mm lateral as measured from bregma, so that the tip would be positioned in the lateral portion of the dentate hilus) of each animal. This location was chosen based on data demonstrating that VEGF diffuses over a 1.5 mm radius (Croll et al., 2004b
). Dental acrylic was then applied to secure the cannula and anchor screws in place. Nylon sutures were used to close the incision, topical antimicrobial ointment was applied, and animals were placed under a heat lamp to recover.
One week following cannula implantations, animals were re-anesthetized following the same procedure and an incision was made at the nape of the neck. The heat-sealed tip of the catheter was snipped and an Alzet osmotic minipump (Durect Corporation, Palo Alto, CA, USA), containing either rhVEGF165 (15, 30, 45, or 60 ng/day) or sterile PBS (Sigma-Aldrich), infusing 0.5 μl per hour was attached to the catheter and glued. Additional controls were used in some cohorts, which included the protein controls BSA (bovine serum albumin, to control for protein load), hFc (a recombinant human control protein), and inactivated VEGF. VEGF was inactivated by repeated freeze–thaw cycles, which has previously been shown to eliminate VEGF’s bioactivity (N. Papadopoulos, unpublished observations using human umbilical vein endothelial cell survival assays), rather than by heat, which results in a precipitate. As these control groups produced comparable results to PBS, the data were collapsed into a single category termed “controls.” The pump was inserted into the s.c. space at the nape of the neck and the incision was closed with wound clips. Animals were placed under a heat lamp to recover. illustrates the timeline for these experiments.
Experimental timeline showing timing of VEGF infusion and seizure induction.
Some animals received infusions of BowAng1, a fusion of four molecules of the vascular growth factor angiopoietin-1 with two molecules of hFc (Davis et al., 2003
), along with VEGF. Angiopoietin-1 has been shown to block VEGF’s vascular permeabilizing effects (Thurston et al., 2000
) but not its angiogenic effects in brain (Croll et al., unpublished observations), and was used to determine the role of vascular leak in VEGF’s effects after seizures.
Acute seizure induction
Five days following pump implantations for protein infusions, animals were pre-treated with 1 mg/kg atropine methylbromide (Sigma-Aldrich) injected s.c. 30 min prior to receiving 350 mg/kg pilocarpine hydrochloride (Sigma-Aldrich) intraperitoneally. Seizures were scored from stages 1–5 based on a previously published (Rudge et al., 1998
) modification of Racine’s (1972)
scale. Briefly, stages 1–4 were identical to Racine’s stages. However, the scale was expanded to define stage 5 as sudden, but transient, whole-body tonus. Stages 6 and 7 were status epilepticus, which was defined as over 5 min of continuous seizures without intervening return to normal behavior. Stage 6 involved continuous rearing and falling whereas stage 7 included the episodes of stage 5 seizures. Stage 8 was defined as death occurring during status epilepticus (Rudge et al., 1998
Status epilepticus was truncated with 10 mg/kg diazepam (Henry Schein, Melville, NY, USA) administered intraperitoneally after 60 min. Animals not reaching status epilepticus were excluded from all remaining analyses. Each control animal received an injection of atropine and subsequently diazepam concurrent with animals that had seizures, but 0.9% NaCl (saline) was injected instead of pilocarpine, using an equivalent volume.
VEGF enzyme-linked immunosorbent assays (ELISA)s
Rat VEGF ELISAs
Animals were killed with Euthasol veterinary euthanasia solution (Henry Schein, Inc.) 24 h following injection of saline or pilocarpine for analysis of VEGF protein levels by ELISA. Brains were removed and placed on ice for 3 min until firm. They were then placed in an acrylic brain matrix (MyNeuroLab, Inc., www.myneurolab.com
) and cut into 1.5 mm slabs with a thin razor. Two slabs containing the full medial–lateral extent of dorsal hippocampus were selected, and a 3 mm wide sample was cut from the center of dorsal hippocampus and the overlying cortex (selected to match the region of VEGF infusion in later studies). Tissue samples were frozen in Eppendorf tubes on dry ice. An Immuno Maxisorp plate (VWR Scientific, West Chester, PA, USA) was coated with 100 μ
l per well rhVEGF165
(Regeneron Pharmaceuticals) at 2 μ
g/ml in carbonate/bicarbonate buffer (Sigma-Aldrich) and incubated overnight at 4 °C. The plate was then washed with KPL (Kirkegaard and Perry Laboratories, Gaithersburg, MD, USA) buffer followed by 300 μ
l 0.2% I-blocking buffer (Tropix, Foster City, CA, USA) at room temperature for 1 h. The standard was diluted to 100 ng/ml and serially diluted in diluent with normal mouse serum. Samples were diluted and 100 μ
l was placed in each well, in duplicate, and incubated with VEGF antibody (R & D Systems, Minneapolis, MN, USA) for 2 h at room temperature. The plate was then washed four times with 300 μ
l wash buffer. Goat anti-human IgG Fc conjugated to horseradish peroxidase (Sigma-Aldrich) at 1:20,000 was added in diluent and incubated for 1 h at room temperature. The plate was washed again four times, followed by 100 μ
l per well of Tris–mercapto-ethanol buffer substrate (Sigma-Aldrich), and developed at room temperature for 30 min. Development was stopped by adding 100 μ
l per well 2 N H2
. The plate was read at 450–570 nm, and samples were normalized to standards where the range of the standard curve was 0.14–100 ng/ml.
Human VEGF ELISAs
After 5 days of continuous infusion of human recombinant VEGF or control protein (hFc) at 30 ng/day, rats were killed and tissue was collected and homogenized as described for the rat VEGF ELISAs. Human VEGF ELISAs were performed using the Quantikine Human VEGF ELISA kit from R & D Systems (RNDsystems.com) in accordance with manufacturer’s instructions.
Animals were deeply anesthetized with an overdose of a pentobarbital-based killing solution, and were subsequently exsanguinated with heparinized isotonic (0.9%) saline perfused through the heart. Following saline perfusion, animals were perfusion-fixed with 4% paraformaldehyde in first acetate and then borate buffer, as previously described (Croll et al., 1999
). The brains were removed and placed in 30% sucrose borate buffer at 4 °C until they were sectioned. After 3–7 days, brains were sectioned coronally at 40 μ
m and stored in cryoprotectant (Watson et al., 1986
) at −20 °C until they were stained. Sections were immunostained as previously described (Scharfman et al., 2000
) using a Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA, USA) and an anti-VEGF (goat polyclonal, 1:1000, R & D Systems) antibody; other sections were additionally double-immunostained with VEGF and a second antibody, glial fibrillary acidic protein (GFAP rabbit polyclonal, 1:60,000, Dako, Carpinteria, CA, USA), an astrocyte marker, to detect co-localization of VEGF protein with GFAP. Brains were also stained with an anti-RECA (rat endothelial cell antigen, 1:250, Serotec, Raleigh, NC, USA) antibody to visualize vasculature. Before staining brain tissue, we verified the specificity of the VEGF antibody by immunostaining sections adjacent to sections used for VEGF in situ
hybridization in developing embryos and adult rat ovaries. Expression patterns for immunostaining matched those for in situ
hybridization (data not shown). In addition, for all stains, the primary antibody was not added to some sections, and these sections showed no specific staining pattern. Sections were stained with Cresyl Violet for evaluation of cell damage using a previously described rating scale (Rudge et al., 1998
). Additional sections were stained with Methylene Blue for stereological quantification of cell damage. Sections were hydrated through graded ethanols and then stained with a 1.6%/1% Methylene Blue/Azure II solution following exposure to 1% periodic acid.
Quantification of vascular parameters
Vascular density and vascular diameters were measured in RECA-immunostained tissue sections as previously described (Croll et al., 2004b
). Briefly, images were viewed under a Nikon Eclipse E400 microscope (Morrell Instruments, Melville, NY, USA) captured with a digital video camera into SPOT software and imported into the public access image analysis program, NIH Image (National Institutes of Health, Bethesda, MD, USA). Vascular density was measured as proportion of area occupied by RECA-positive lumens in equatorial sections by point-count stereology using a randomly-oriented acetate grid overlay. Vascular diameters were measured by taking the smallest diameter across cross-sectional vascular profiles, and the perpendicular distance across longitudinally-oriented vessels. Both measures were taken using the NIH Image length function on those vessels randomly selected by the grid used in point-count stereology. All measurements were conducted by experimenters blind to the treatment groups of the animals.
Quantification of neuronal density
Precise estimates of cornus ammonis (CA) 1 density were determined using the optical fractionator method (West et al., 1991
). The total number of neurons was determined within a region of interest in area CA1 that was within the diffusion range, 1.5 mm radius, of the cannula tip (verified by VEGF immunostaining, data not shown). The region of interest was defined as the area of the CA1 pyramidal cell layer between area CA2 and the subiculum in the medial–lateral axis, and from the initial appearance of the CA1 pyramidal cell layer to the portion of the hippocampus where the dorsal and ventral portions of area CA3 united in the rostral caudal axis. Forty micron coronal sections in a one-in-six series, with a random starting point, were mounted on slides and stained with a 1.6%/1% Methylene Blue/Azure II solution following exposure to 1% periodic acid.
Sections were viewed with an Olympus BX-51 m icroscope and Optronics video camera. Using a stereological software package (Stereo Investigator, Microbrightfield Inc., Williston, VT, USA), a pre-determined counting frame (25 μm2) was systematically moved along a randomly placed grid (125 μm2), and the number of cell nucleoli that came into focus within a portion of the section (excluding 4 μm upper and lower guard zones) was counted. Only cells that had a darkly stained nucleolus surrounded by a lightly stained nucleus and cytoplasm were counted. In the event that two nucleoli could not be distinguished as belonging to two separate neurons, only one neuron was counted. Neurons that were pyknotic, i.e. had a dense, collapsed, profile, were not considered viable and were excluded from the analysis. The total number of neurons within the region of interest was estimated with the formula: N=sum Q−×1/tsf×1/asf/1/ssf where the number of neurons counted (sum Q−) was multiplied by the reciprocal value of the sampling probabilities based on the proportion of section thickness (tsf), cell layer area (asf), and total number of sections (ssf). All histological and stereological analyses were conducted by an independent evaluator blind to the treatment conditions for each animal.
A two-way factorial analysis of variance (ANOVA) was conducted on rat VEGF ELISA values to examine the effect of treatment (saline vs. pilocarpine) by region (hippocampus vs. cortex). A one-way independent groups ANOVA was conducted to compare CA1 neuronal densities for various VEGF doses versus vehicle-infused brains. Post hoc analyses were conducted to determine significant effective doses of VEGF. Student’s independent groups t-tests were used to compare vascular densities and cell loss scores for controls versus VEGF, hFc versus Flt-Fc, and/or VEGF versus VEGF+BowAng1. All analyses were conducted using an α value of 0.05.