Proof of Concept: Pharmacological Preconditioning with a Toll-like Receptor Agonist Protects against Cerebrovascular Injury in a Primate Model of Stroke

Introduction

Cerebral ischemic injury is the third leading cause of death and the leading cause of serious, long-term disability in the United States. Therapies for the treatment of stroke are most often targeted at acute neuroprotection immediately after ischemia, but antecedent neuroprotective strategies designed to reduce the damage caused by ischemic injury are also highly feasible. Despite demonstrated efficacy in preclinical studies, most candidate neuroprotective compounds have failed in clinical trials. Reasons thought to contribute to clinical trial failures of promising therapeutics include an absence of preclinical evaluation in a nonhuman primate (NHP) model of stroke before clinical trial development and testing. Nonhuman primates offer greater relevance to humans as they have similar neuroanatomy and vasculature and unlike rodents, they possess gyro-encephalic brains. Moreover, NHP models of stroke offer the potential for more advanced neurologic assessment compared with rodents, offering a second meaningful outcome measure in addition to infarct volume. Humans and NHPs also have a more similar ratio of gray-to-white matter, as well as similar thresholds to ischemic injury of gray and white matter (Arakawa et al, 2006; Bristow et al, 2005; Falcao et al, 2004; Marcoux et al, 1982; Pantoni et al, 1996) as compared with rodents, although differences do still exist among different primate species. As white and gray matter tissues differ in their neurochemical response to ischemia, amelioration of damage in each type of tissue may require different therapeutic approaches to test candidate treatments. Finally, studies performed in NHPs better extrapolate efficacious and toxic dose ranges for human subjects.

In many instances, patients at high risk for an ischemic event can be identified. For example, approximately half of the patients undergoing coronary artery bypass surgery suffer long-term cognitive decline from intraoperative focal or global ischemia (Barber et al, 2008). For these patients, prophylactic neuroprotection through antecedent therapy would be extremely beneficial. In addition, individuals who have experienced a stroke are at substantial risk of experiencing a recurrent stroke. Treatment of high-risk populations with preconditioning agents, either by repeated administration in anticipation of a subsequent stroke event or by a single acute treatment preoperatively, has the potential to improve stroke outcomes.

In response to the paucity of adequate NHP models specifically involving cortical stroke, our laboratory recently developed a novel model in the rhesus macaque (Macaca mulatta) (West et al, 2009). In this model, the right middle cerebral and both anterior cerebral arteries are transiently occluded using a minimally invasive transorbital approach, resulting in a highly reproducible stroke with good survivability. This remains the only NHP model of stroke that results in an infarct primarily involving cortical gray matter, allowing for targeted study of therapies protecting this region of the brain. Additionally, by producing a reliable pattern of damage including the primary motor cortex, the resulting neurologic deficit in this model is plegia of the upper and lower extremities. This type of motor deficit is quantifiable in the rhesus macaque, whereas changes in alertness or cognition are not as readily ascertained due to the demeanor of this species as compared with baboons or humans.

We, and others, have shown that Toll-like receptor (TLR) activation has great potential in prophylactic neuroprotection (Rosenzweig et al, 2004; Stevens et al, 2008; Tasaki et al, 1997). Toll-like receptors are activated by molecules that contain pathogen-associated molecular patterns, both natural (i.e., bacterial lipopolysaccharide, bacterial DNA, and single-stranded RNA) and synthetic (i.e., unmethylated cytosine-guanine-rich DNA oligonucleotides (CpG ODNs), imidazoquinolines and polyinosinic:-polycytidylic acid (poly I:C)). In addition to acting as sensors of invading pathogens, TLRs also act as sentinels of tissue damage and mediate inflammatory responses to aseptic tissue injury. When TLR ligands are given in advance of ischemic injury in rodent models of cerebral ischemia, decreased infarct volume is observed, most likely through the reprogramming of the innate immune response to injury. Prophylactic lipopolysaccharide has also been shown to salvage hippocampal neurons from ischemic death in a pig model of cardiopulmonary bypass (Hickey et al, 2007), indicating the potential for broad utility of TLR ligands as neuroprotectants in a diverse range of conditions that result in cerebral ischemic injury.

Importantly, TLR ligands are being developed for use in humans for a variety of clinical indications, both as monotherapies and as components of various FDA-approved and novel vaccine technologies. Several CpG ODNs have shown reasonable safety profiles in humans and have been explored in numerous human clinical trials (Krieg, 2006). The three different CpG ODNs used in our study (referred to as Kmix) were selected based on their potential to broadly stimulate an immune response in peripheral blood cells from a majority of individuals comprising the human population. Previous work by Klinman and colleagues showed that the variation in the individual response of primate immune cells was due to differences in CpG ODN sequence and length (Leifer et al, 2003). Further, structural differences between various classes of CpG ODNs have been shown to stimulate distinct cell populations, allowing these compounds to be used selectively to achieve specific therapeutic goals. K-type (B-type) ODNs used in our study predominantly directly stimulate plasmacytoid dendritic cells and B cells in primates, while ultimately affecting additional bystander cell populations (i.e., NK cells and monocytes) (Krug et al, 2001; Verthelyi et al, 2001).

The present study evaluated the potential of a mixture of K-type CpG ODNs to act as a prophylactic neuroprotectant using the aforementioned clinically relevant NHP stroke model. These data satisfy the Stroke Therapeutics Academic Industry Roundtable recommendations, a potentially valuable precursor to the advancement of these compounds to human clinical trials (Fisher et al, 2009). Further, this work demonstrates efficacy in stroke models in two different species, mouse and monkey, suggesting a similar mechanism of action of CpG ODNs. The use of mice in this dual-model approach could expedite the knowledge gained regarding TLR-induced neuroprotection. To our knowledge, this is the first demonstration that a pharmacological preconditioning agent can protect the central nervous system from ischemic damage in the NHP.

Materials and methods

Reagents

ODN1826 (tccatgacgttcctgacgtt), phosphorothioate CpG ODN was obtained from Invivogen (San Diego, CA, USA). Endotoxin levels were determined to be negligible (<0.125EU/mg). Kmix CpG ODN, a 1:1:1 mixture of three phosphorothioate K-type ODNs (K3: ATC GAC TCT CGA GCG TTC TC; K23: TCG AGC GTT CTC; K123: TCG TTC GTT CTC) (Leifer et al, 2003) was kindly provided by Dr Dennis Klinman for mouse studies and material used for NHP studies was manufactured under compliance with good manufacturing practices guidelines by Oligos, etc. (Wilsonville, OR, USA).

Drug Treatments

For mice, a volume of 200μL of CpG ODN1826, Kmix CpG ODNs, or saline vehicle was administered by intraperitoneal injection at indicated doses 72 hours before middle cerebral artery occlusion (MCAO). For NHP, a fixed dose volume of 1mL of Kmix CpG ODN solution or saline vehicle was administered by intramuscular injection 72 hours before MCA and anterior cerebral artery occlusion (ACAO). Animals were randomized to receive 0.06 mg/kg Kmix CpG (low dose), 0.3mg/kg Kmix CpG (high dose), or saline. Investigators were blinded to treatments and the experimental timeline is depicted in Figure 2A.

OPEN IN VIEWER

Figure 1

Reduction in infarct volume following middle cerebral artery occlusion (MCAO) in mice preconditioned with CpG1826 or Kmix cytosine-guanine-rich DNA oligonucleotides (CpG ODNs). CpG ODNs or saline control was administered in a final volume of 200 μL via intraperitoneal injection 72 hours before the onset of surgical occlusion. All animals were subjected to 35 minutes of MCAO and infarct volumes were measured 24 hours after reperfusion. (A) Absolute infarct volume and (B) percent infarct of total hemisphere are displayed as mean ± s.e.m. of n = 4 to 7 per group and *P< 0.05 by one-way analysis of variance (ANOVA) and Bonferroni post hoc test.

OPEN IN VIEWER

Figure 2

Experimental timeline and surgical occlusion. (A) Animals were assessed for physical health, blood was drawn for baseline prescreening and they were transferred to their home cages 2 weeks before surgery for acclimation. Kmix cytosine-guanine-rich DNA oligonucleotides (CpG ODNs) or saline control was given 72 hours before surgery. Just before surgery, blood was drawn to assess drug-induced changes. After 48 hours of reperfusion, animals were taken to magnetic resonance imaging (MRI) for imaging of final infarct volume. Following MRI, animals were taken directly to necropsy and tissues were collected. (B, C) Reproduction of schematic of the surgical site and vasculature of the rhesus two-vessel occlusion model from West et al, (2009). (B) In situ surgical view of the right orbit of the rhesus macaque, illustrating the exposure of major cerebral arteries and positioning of aneurysm clips for occlusion of the right middle cerebral artery and bilateral anterior cerebral arteries. (C) Ex vivo illustration of the Circle of Willis with one clip on the middle cerebral artery distal to the orbitofrontal artery (lateral clip) and a second clip on both anterior cerebral arteries (medial clip) just before the vessels join to form a single pericallosal artery.

Surgery

Mouse middle cerebral artery occlusion model: Cerebral focal ischemia was induced by MCAO as published previously (Stevens et al, 2008). Mice were briefly induced with 3% isoflurane and maintained with 1.5% to 2% isoflurane throughout the surgery. Briefly, the MCA was blocked and the filament was maintained in place for 35 minutes under anesthesia; the filament was removed subsequently and blood flow was restored. Cerebral blood flow was monitored with laser Doppler flowmetry (Transonic System Inc., Ithaca, NY, USA). Temperature was monitored with a rectal thermometer and maintained at 3°C ±0.5°C, with a controlled heating pad and lamp (Harvard Apparatus, Holliston, MA, USA). Mice were survived for 24 hours with access to soft food and water until euthanasia.

Two-vessel occlusion protocol in nonhuman primate: Two weeks before surgery, animals were screened for general health, endemic disease, and neurologic disorders. The right MCA (distal to the orbitofrontal branch) and both anterior cerebral arteries were exposed and occluded with vascular clips for 60 minutes, as previously described (West et al, 2009). Surgical procedures were conducted by a single surgeon. Briefly, animals were given ketamine (~10mg/kg, intramuscular injection) and then intubated and maintained under general anesthesia using 0.8% to 1.3% isoflurane vaporized in 100% oxygen. A blood sample was taken and a venous line was placed for fluid replacement. An arterial line was established for blood pressure monitoring throughout the surgery and to maintain a mean arterial blood pressure of 60 to 80 mm Hg. Endtidal CO₂ and arterial blood gases were continuously monitored to titrate ventilation to achieve a goal Paco₂of 35 to 40 mm Hg. Postoperative analgesia consisted of intramuscular hydromorphone HCl and buprenorphine.

Infarct Measurements

Mouse: Mice were deeply anesthetized with isoflurane, and were then perfused with ice-cold saline containing 2U/mL heparin at 24 hours after stroke. Brains were removed rapidly, placed on a tissue slicer, and covered with agarose (1.5%). The olfactory bulbs were removed and the remainder of the brain was sectioned into 1 mm slices beginning from the rostral end into a total of seven slices. The area of infarction was visualized by incubating the sections in 1.5% 2,3,5-triphenyltetrazolium chloride (TTC; Sigma Aldrich, St Louis, MO, USA) in phosphate-buffered saline for 15 minutes at 37°C. The sections were then transferred to 10% formalin (Sigma Aldrich). Images of the sections were scanned, and the hemispheres and areas of infarct were measured using NIH ImageJ v1.38 software (Bethesda, MD, USA). The measurements were multiplied by the section thickness and summed over the entire brain to yield volume measurements. Data of ischemic damage were calculated using the indirect method to minimize error introduced from edema. Percent infarct = (contralateral hemisphere volume—volume of noninfarcted tissue of the ipsilateral hemisphere)/(contralateral hemisphere volume) × 100 (Stevens et al, 2008).

Nonhuman primate: Measurement of infarct volume was performed using T₂-weighted magnetic resonance (MR) images taken after 48 hours of reperfusion (West et al, 2009). All scans were performed on a Siemen's 3T Trio system, housed near the surgical suite at Oregon National Primate Research Center. Because of the small filling capacity of the rhesus macaque head, a human extremity coil was used to achieve better image quality of the brain. Animals were given ketamine (10 mg/kg intramuscular injection) and a blood sample was obtained. Animals were then intubated and administered 1% isofluorane vaporized in 100% oxygen for anesthesia maintenance. Animals were scanned in the supine position, and most of the animals also received baseline scans before surgery. During the scans, animals were monitored for physiologic signs, including pulse oximetry, end-tidal CO₂, and respiration rate. All animals received anatomical MR imaging (MRI) scans (T1- and T₂-weighted), high-resolution time of flight scans, and a diffusion-weighted scan. The T₁ scan was an Magnetization-Prepared Rapid Acquisition Gradient-Echo (MPRAGE) protocol, with repetition time (TR) = 2500 ms, echo time (TE) = 4.38 ms, number of averages = 1, and the flip angle = 12°. Full brain coverage was attained at a resolution of 0.5 mm isovoxel. The T₂ scan was a turbo spin-echo protocol, with TR = 5280 ms, TE= 57 ms, number of averages = 4, an echo train length of 5, and a refocusing pulse flip angle of 120°. The entire brain was imaged with a 0.5 ± 0.5 mm² in-plane resolution and a slice thickness of 1 mm. For visualization of the region of infarction, 48 hours after stroke, brain sections were prepared and immediately placed in 1.5% (TTC; Sigma Aldrich) in 0.9% phosphate-buffered saline and stained for 15 minutes at 37°C, as appropriate. Images from T₂-weighted MRIs and TTC-stained sections were examined for the location of infarction, and the total affected area measured using ImageJ, as previously described (West et al, 2009). Each of the techniques (MRI, TTC) analyzed comparable anatomical regions and sampled ~15 slices (4 mm each). Measurements of infarct volume as a percentage of the ipsilateral hemisphere or cortex were made using the following formula: (volume of infarcted tissue of the ipsilateral hemisphere/total volume of the ipsilateral hemisphere) × 100%.

Neurologic Assessment in Nonhuman Primate

Neurologic assessments were performed by a single observer, as previously described (Spetzler et al, 1980). The scale evaluates motor function, behavior (mental status), and cranial nerve deficit, and higher scores represent better functional outcomes. Motor function scores from 1 to 70, according to severity of hemiparesis in the left extremities, are presented. A score of 10 = severe hemiparesis, 25= mild hemiparesis, 55= favors normal side, or 70 = normal ability. Behavior and alertness scores are presented, ranging from 0 to 20, with 0 = dead, 1= comatose, 5 = aware but inactive, 15 = aware but less active, and 20 = normal.

Tissue Collection in Nonhuman Primate

Two days after stroke and immediately after MRI, animals were taken to necropsy under sedation followed by 25 mg/kg pentobarbital. Blood samples were drawn before exsanguinations and perfusion of the brain was performed with cold heparinized saline (2 U/mL) through the ascending aorta. Brains were rapidly removed, placed in a rhesus brain matrix (ASI, Warren, MI, USA), and cut into 15 consecutive, 4 mm thick coronal slabs per brain.

Inclusion Criteria and Data Exclusions

The inclusion criteria established a priori for this study include (1) animal survival to 48 hours MRI procedure, (2) a surgical procedure that was without complications, and (3) no evidence of preexisting clinical conditions discovered at necropsy. Two NHPs died during the surgical procedure. One animal from the 0.3mg/kg Kmix CpG treatment group died when the dura mater was ruptured before occlusion. One animal in the saline treatment control group died 24 hours after surgery, and was found at necropsy to have an abnormally enlarged heart, deemed the likely cause of death. Two additional saline treatment control animals were excluded from the study, one due to intracranial bleed that continued postsurgically and one animal due to abnormal blood clotting. No drug-related adverse events were apparent. To maintain the consistency of terminal measurements at 48 hours after occlusion, a third saline-treated animal was excluded from the data set due to early termination at 24 hours after occlusion. Exclusion of data for this animal did not alter the significance of the findings.

Inclusion of Additional Untreated Controls in Nonhuman Primate Study

Five untreated animals were included to increase the power of the study in the control group and balance the excluded animals in the saline-treated control group. Untreated controls differed from saline-treated controls in that they did not receive an intramuscular injection of saline 72 hours before the onset of surgical occlusion. There was no statistically significant difference in infarct volumes between saline-treated (41.6%; ± 2.0% of hemisphere) and untreated (33.3% ± 4.1%) animals (P=0.12 with Student's t-test) and variances were similar by Bartlett's test. Importantly, we included data from all untreated controls from that prior experiment, resulting in a higher sample size for the control group compared with treated groups in this study, but this prevented selection bias. The analysis of variance performed used weighted means to account for unequal sample size.

Plasma and Blood Analyses in Nonhuman Primate

Levels of granulocyte macrophage-cerebrospinal fluid (CSF), interferon-γ, interleukin (IL)-1β, IL-2, IL-6, IL-8, IL-10, IL-12p70, and tumor necrosis factor (TNF)-α were assayed using a 9-plex Proinflammatory Multiplex Human Cytokine ELISA Array (Meso Scale Design, Gaithersburg, MD, USA) internally validated for NHP cytokine detection. The limits of detection of plasma levels of all cytokines were reported as 0.6 pg/mL; levels of granulocyte macrophage-CSF, interferon-γ, IL-1b, IL-2, IL-β, IL-10, IL-12p70, and TNF-a were below detectable limit for most samples and group means were <5pg/mL in most cases. The manufacturer reports that TNF-alpha and IL-10 may be underestimated at very low concentrations in rhesus macaques using this kit. C-reactive protein (CRP) was measured by human-specific Active ELISA (Beckman Diagnostics, Brea, CA, USA). Hematology parameters were determined using an ABX Pentra 60 analyzer (Horiba Medical, Irvine, CA, USA).

Statistical Analyses

All statistical analyses were performed using Prism (Graphpad Software, La Jolla, CA, USA). Group means were compared using a one-way analysis of variance with Bonferroni's multiple comparison post hoc test. For data with repeated measurements, data were compared using a two-way analysis of variance with Bonferroni's post hoc test. Data represent mean ± standard error of the mean (s.e.m.), unless otherwise noted. Differences were considered statistically significant when P<0.05.

Results

Prophylactic Administration of Kmix Cytosine-Guanine-Rich DNA Oligonucleotides Significantly Reduces Cortical Damage in Nonhuman Primates

Using the optimal time window determined from mouse studies (Stevens et al, 2008), Kmix CpG ODNs or vehicle was administered to NHPs 72 hours before surgical occlusion (Figure 2A). All doses of CpG were well tolerated by the rhesus macaque with no apparent toxicities. Prophylactic systemic administration of Kmix CpG ODNs significantly reduced cortical damage, resulting from surgical occlusion (depicted in Figures 2B and 2C) in a dose-related manner, as assessed by T₂ MRI and TTC staining 48 hours after reperfusion (Figures 3 and 4). Compared with the control group (37.0% ± 2.64% infarct volume), doses of 0.06 and 0.3mg/kg Kmix CpG ODNs reduced cortical damage to 27.3% ± 4.8% (P>0.05) and 21.3% ± 5.0% (P< 0.05), respectively (Figure 3A). Similar results were obtained when data were expressed as absolute infarct (Figure 3B). In addition to cortical stroke volume measurements using T₂MRI, TTC staining of coronal sections was also performed and the area of infarct as a percent of ipsilateral hemisphere was determined (Figure 3C). Quantitation of infarct by TTC is presented as percent of the ipsilateral hemisphere, as opposed to percent of cortex, because discernment of the cortical boundaries was unreliable. A 42% reduction in mean infarct volume was observed between saline and the 0.3 mg/kg treated groups, when determined as a percent of ipsilateral hemisphere using TTC staining (Figure 3C) and absolute infarct (Figure 3D); therefore, the magnitude of neuroprotection (i.e., reduction in infarct volume) was identical to that obtained for percent of cortex using the serial MRI approach (Figure 3A). Our previous study showed a significant correlation between percent of hemisphere infarct volumes measured using TTC staining and T₂ MRI methods (West et al, 2009). Similarly, the current study showed a significant correlation between percent of hemisphere measurements derived using TTC and MRI approaches (r=0.86 by Spearman correlation analysis, P<0.0001; data not shown).

OPEN IN VIEWER

Figure 3

Reduction in infarct volume following middle cerebral artery (MCA) and anterior cerebral artery (ACA) occlusion in nonhuman primates (NHPs) preconditioned with Kmix cytosine-guanine-rich DNA oligonucleotides (CpG ODNs). Kmix ODNs or saline control was administered 72 hours before the onset of surgical occlusion via intramuscular injection in a final volume of 1 mL. Infarct volume was measured 48 hours after reperfusion using T₂ magnetic resonance imaging (MRI) scans to determine (A) percent infarct of total ipsilateral cortex and (B) absolute cortical infarct volume in mm³. 2,3,5-Triphenyltetrazolium chloride (TTC) staining was used to determine (C) infarct as a percent of ipsilateral hemisphere and (D) absolute hemispheric infarct volume in mm³. Data are expressed as mean ± s.e.m. *P<0.05 using one-way analysis of variance (ANOVA) with Bonferroni post hoc test.

OPEN IN VIEWER

Figure 4

Reduction of ischemia-induced cortical damage following pretreatment with Kmix cytosine-guanine-rich DNA oligonucleotides (CpG ODNs). T₂-weighted magnetic resonance imaging (MRI) was performed 48 hours after 60 minutes of surgical occlusion of the right middle cerebral artery (MCA) and anterior cerebral arteries (ACAs). Images from a representative control and pretreated brains are shown at three horizontal levels, with the left side of each image representing the right (ischemic) hemisphere. Edema is apparent in affected regions (e.g., insular cortex; arrows), appearing as a hyperintense signal in expanded cortical areas. Damage is distinguishable from the contralateral (unaffected) hemisphere where signal is observed from cerebrospinal fluid (CSF) in the sulci (horizontal arrowhead) and ventricles (oblique arrowheads). Cortical damage was measured and was reduced in a dose-dependent manner throughout the hemisphere that received ischemia.

Prophylactic Administration of Kmix Cytosine-Guanine-Rich DNA Oligonucleotides Improves Neurologic Function in Nonhuman Primates

Animals were also examined for neurologic function using a modification of the Spetzler neurologic scale adapted for evaluating stroke in the rhesus macaque (Spetzler et al, 1980). The Spetzler scale has been further validated in stroke efficacy studies in NHPs (Mori et al, 1995). For motor function, we have excluded facial weakness scores, resulting in a range of 1 to 70 for our scale, as apposed to 1 to 75 in original Spetzler scale. Groups receiving either 0.06 or 0.3 mg/kg Kmix CpG ODNs before surgical occlusion demonstrated an ~40% higher motor function score (27 ± 7 and 27 ± 8, respectively) compared with the control group (19 ± 3) (Figure 5A). Although this increase is very likely to be relevant functionally, the improvement observed in Kmix-treated animals was not dose dependent or statistically significant compared with saline-treated animals (P>0.05). In addition, no significant differences in behavior and alertness were observed between groups (Figure 5B), although these parameters were more difficult to assess.

OPEN IN VIEWER

Figure 5

Neurologie improvement following preconditioning with Kmix cytosine-guanine-rich DNA oligonucleotides (CpG ODNs). (A) Motor function and (B) behavior and alertness were assessed using the modified Spetzler neurologic scale at 48 hours after surgical occlusion, before sedation for magnetic resonance imaging (MRI). To minimize the sedative effects of postoperative analgesics, animals were assessed in the morning, at least 6 hours after their last dose of pain medications. Data displayed are mean ± s.e.m.

Kmix Cytosine-Guanine-Rich DNA Oligonucleotides have no Effect on Blood Cell Composition After Stroke

To monitor for potential effects of prophylactic CpG ODN administration on blood composition, basic hematological parameters were assessed at baseline (2 weeks before surgery), 72 hours postdrug administration (just before surgery), and at the time of necropsy (48 hours after surgery). No significant changes were observed between baseline and perisurgery values for any of the treatment groups, suggesting that CpG treatment appeared to have little or no effect on the measured parameters at the time of stroke induction (Table 1). At 48 hours after surgery, there were no significant differences in mean white blood cell counts among control or treatment groups. In contrast, a statistically significant reduction in mean lymphocyte count was observed 48 hours after stroke in all groups compared with their respective prescreen values (Table 1). When compared with blood taken just before surgical occlusion, mean neutrophil and monocyte counts showed a trend for elevation in samples taken at the time of necropsy in all groups (Table 1). The mean blood monocyte count at the time of necropsy was significantly elevated in the control group compared with the time of surgery value, and a trend for increase was also apparent in the Kmix-treated groups compared with their time of surgery values.

OPEN IN VIEWER

Table 1

Complete blood count summary

Treatment	WBC (103/mm3)	Neutrophils (103/mm3)	Lymphocytes (103/mm3)	Monocytes (103/mm3)	Eosinophils (103/mm3)	Basophils (103/mm3)	RBC (106/mm3)	Platelets (103/mm3)	HCT (%)
Prescreen
Control	8.63 ± 2.12 a	4.16 ± 2.21	3.94 ± 1.09	0.33 ± 0.14	0.10 ± 0.05	0.08 ± 0.02	5.61 ± 0.42	255.2 ± 81.1	40.5 ± 2.5
0.06 mg/kg Kmix CpG	7.61 ± 2.21	4.24 ± 1.71	2.68 ± 0.73 e	0.31 ± 0.13	0.16 ± 0.11	0.08 ± 0.05	5.61 ± 0.42	322.4 ± 61.3	40.0 ± 2.3
0.3 mg/kg Kmix CpG	9.63 ± 5.05	4.37 ± 3.39	4.87 ± 2.40 c	0.38 ± 0.22	0.30 ± 0.29	0.011 ± 0.08	5.64 ± 0.42	317.3 ± 96.9	41.1 ± 2.2
Surgery
Control	6.35 ± 3.04	2.82 ± 1.81	3.12 ± 1.18	0.24 ± 0.21	0.10 ± 0.09	0.05 ± 0.06	5.35 ± 0.42	300.5 ± 72.9	38.6 ± 2.6
0.06 mg/kg Kmix CpG	6.79 ± 2.95	3.54 ± 2.43	2.30 ± 0.68	0.28 ± 0.10	0.09 ± 0.04	0.03 ± 0.02	5.23 ± 0.25	295.3 ± 83.4	38.1 ± 2.3
0.3mg/kg Kmix CpG	7.04 ± 2.37	3.07 ± 1.44	3.45 ± 1.22	0.31 ± 0.09	0.14 ± 0.09	0.09 ± 0.09	5.53 ± 0.29	284.0 ± 48.5	39.9 ± 1.6
Necropsy
Control	6.91 ± 3.41	4.27 ± 1.97	1.79 ± 0.94a,b	0.52 ± 0.38 a	0.09 ± 0.07	0.06 ± 0.07	4.77 ± 0.48	280.2 ± 70.9	34.9 ± 1.0
0.06 mg/kg Kmix CpG	6.44 ± 2.40	4.72 ± 2.17	1.67 ± 0.81	0.47 ± 0.19	0.09 ± 0.08	0.05 ± 0.03	4.98 ± 0.57	301.1 ± 45.6	36.4 ± 4.3
0.3 mg/kg Kmix CpG	7.60 ± 2.32	4.55 ± 1.72	2.34 ± 1.11 d	0.62 ± 0.26 c	0.10 ± 0.07	0.07 ± 0.04	5.00 ± 0.90	267.6 ± 68.3	36.2 ± 2.6

ANOVA, analysis of variance; CpG, cytosine guanine; HCT, hematocrit; RBC, red blood cell; WBC, white blood cell.

Data represent mean ± s.d. of n = 7 for Kmix groups and n = 11 for control group. Prescreen denotes analysis of blood drawn 2 weeks before occlusion, surgery denotes blood drawn at the time of surgery (72 hours posttreatment and before occlusion), and necropsy denotes 48 hours after reperfusion. Two-way ANOVA comparisons of time points with Bonferroni post hoc test indicated significance of ^aP < 0.05 versus control surgery value, ^bP < 0.001 versus control prescreen, ^cP < 0.05 versus 0.3 mg/kg surgery value, and ^dP < 0.001 versus 0.3 mg/kg prescreen value; and comparisons of group means are indicated ^eP < 0.01 versus 0.3 mg/kg prescreen value.

Plasma Interleukin-6 and C-Reactive Protein are Elevated After Stroke in Nonhuman Primates

Plasma levels of key proinflammatory cytokines (granulocyte macrophage-CSF, interferon-γ, IL-1β, IL-2, IL-6, IL-8, IL-10, IL-12p70, and TNF-α) were measured at baseline (2 weeks before surgery), 72 hours postdrug administration (just before surgery), and at the time of necropsy (48 hours after surgery) to investigate the effects of stroke and/or potential mechanisms of action of CpG ODNs in neuroprotection. Following stroke, systemic levels of IL-6 were detectable at 48 hours after stroke in all animals (Figure 6A). Mean plasma levels of both IL-6 and CRP (Figures 6A and 6B) were significantly increased after stroke in all groups when compared with prestroke values. All other cytokines measured were either below the detectable limit or present at very low levels, with no differences observed between the time points examined.

OPEN IN VIEWER

Figure 6

Plasma interleukin (IL)-6 and C-reactive protein (CRP) are elevation following stroke in nonhuman primates (NHPs). Blood samples for plasma cytokine analysis were taken at 48 hours after vessel reperfusion. (A) Plasma IL-6 and (B) CRP levels are presented. *P < 0.05 for all groups at the time of necropsy versus all other time points. One outlier was excluded for IL-6 values from 0.06 mg/kg group, as animal demonstrated 1 log higher values for IL-6 (range 166 to 170 pg/mL) and tumor necrosis factor (TNF)-α (range 650 to 857 pg/mL) at all time points, with no deviation after ischemia. Data are displayed as mean ± s.e.m. of n = 6 to 7 per group. Using Two-way analysis of variance (ANOVA) with Bonferroni post hoc test, * P < 0.05 versus prescreen and surgery time points and # denotes significance versus saline (P < 0.05) and 0.06 mg/kg (P<0.01) groups at necropsy.

Discussion

Cytosine-Guanine-Rich DNA Oligonucleotide Treatment and Functional Improvement of Neurologic Deficits

Neurologic assessments performed following stroke revealed a trend for improved functional outcome in animals given CpG ODN treatment in advance of stroke. Although these data were not statistically significant, there was a marked trend for improvement in motor function. Additionally, no differences in behavioral outcome were seen between control and CpG ODN-treated animals. There are several factors that may have contributed to the limited improvement of neurologic parameters. All animals were maintained on postoperative analgesia for 48 hours after stroke, limiting our ability to accurately assess behavior and alertness of animals. Longerterm survival of animals and eventual discontinuation of analgesics would allow for a more accurate behavioral and motor deficit assessment. Additionally, the scale used to assess motor deficit does not clearly differentiate proximal versus distal functional deficits in the upper and lower extremities. Based on location of the infarct induced in this model, animals showing less severe deficits are characterized primarily by distal limb weakness, while dysfunction in more severely affected animals extends more proximally. Control animals in our study demonstrated symptoms ranging from mild hemiparesis to paralysis (no movement of limb), with distal plegia or paralysis in the upper limb commonly present. The lower limb deficits tended to be more variable with weakness or mild paresis often observed. Animals generally demonstrate good mentation and are usually quiet and alert, but less active. Animals often demonstrate facial paresis, although this can be difficult to assess.

Finally, assessing behavioral changes in rhesus macaques is challenging compared with baboons or humans, given their demeanor and overriding tendency to mask physical deficits as a survival mechanism. Rhesus macaques tend to exhibit fewer manipulative behaviors in the presence of an observer as compared with baboons (Iredale et al, 2010). Therefore, we have developed a testing regimen tailored for the examination of rhesus macaques, a species requiring specific manipulations to adequately assess deficits when an observer is present. Prestroke evaluations were performed using treats as positive stimuli to attempt to acclimate animals to the presence of observers. Similar direct and indirect stimulations were then performed after stroke to elicit specific motor functions for evaluation. Our method of neurologic assessment using a modified Spetzler scale correlates well with infarct volumes in this rhesus macaque model (West et al, 2009) over a wide range of infarct volumes (unpublished data). Food treats tend to be motivating and allow sufficient interaction with observers for behavior and motor function assessment.

A more extensive neurologic scale designed to assess the specific deficits seen in this model is under development by our laboratory, which may provide greater insight into functional outcomes in future studies. Despite these complexities, cortical infarct volume was directly related to motor function outcome (P = 0.04 using Spearman correlation; data not shown). Specifically, as the infarct volume increased, the extent of motor deficit observed increased, as indicated by a decrease in the motor score. The Spetzler scale was developed using a different stroke model in the baboon. In contrast to the baboon stroke model, our rhesus stroke model has little to no damage to the caudate/putamen regions, while the cortical damage can involve frontal, parietal region, especially the motor cortex. Hence, the Spetzler scale may not adequately capture deficits, resulting from the extent and location of neurologic damage in our model.

While the demonstration of drug efficacy in NHPs is promising, this may not necessarily ensure success in humans. The animal models used herein achieve an optimal balance of damage and survival that allow robust assessment of therapeutic efficacy for cortical stroke. As with many other stroke model systems, these models have not been designed to accurately reflect clinical strokes that are typically treated, in that full reperfusion within a 30- to 60-minute timeframe is distinctly rare in human stroke patients. Previous efficacy studies of tirilazad mesylate, a synthetic, lipid-soluble, nonglucocorticoid, 21-aminosteroid, were conducted in mouse, rat, hamster, and baboon models of ischemic stroke. When tirilazad was given before reperfusion after the onset of focal cerebral ischemia, infarct volume was reduced and neurologic outcome improved in baboon studies (Mori et al, 1995). Importantly, efficacy of tirilazad was independently validated in two baboon models from different laboratories (Mori et al, 1995; Suzuki et al, 1999). Despite these efforts and validation in both rodents and NHPs, clinical studies with tirilazad were unsuccessful when administered acutely after stroke (Committee, 2000). Human studies will be necessary to fully evaluate the therapeutic potential of Kmix ODNs.

Inflammatory Cytokines, Acute Phase Proteins, and Hematology: Potential Clinical Stroke Biomarkers

Elevations in neutrophil count, plasma CRP, IL-6, TNF-α, and IL-1β levels, as well as reduced lymphocyte counts, have been reported in stroke patients. These parameters were monitored in our studies as potential indicators of stroke magnitude and/or efficacy. The magnitude of poststroke lymphopenia was comparable in all treatment groups—a finding that mirrors the results seen in humans with stroke. A transient immediate loss of T-lymphocytes reaches a nadir at 12 hours after onset of stroke in human patients, which recovers between days 7 and 14 after stroke (Vogelgesang et al, 2008). In our study, NHPs did not demonstrate significant neutrophilia after stroke. Conversely, in one clinical study, the neutrophil count was increased > 50% after stroke and changes were maintained for over 14 days after stroke (Buck et al, 2008). Another study in human stroke patients showed that the magnitude of early neutrophilia correlated well with cerebral infarct size, as determined by MRI (Akopov et al, 1996), although a direct cause—effect relationship could not be determined.

The role of neutrophils in damage after ischemic injury has been described in multiple species. Neutrophils are proposed to be important mediators of postischemic cerebral damage, although more recent findings suggest that the presence of neutrophils does not always correlate with the extent of pathology (Emerich et al, 2002). Importantly, the time points chosen for hematology evaluation in the current study may not have been optimal or the composition of the neutrophil population may be different but absolute numbers unchanged. In future studies, it would be highly informative to perform kinetic and functional characterization of neutrophils in this system.

Finally, postsurgical neutrophilia has been observed in humans and monkeys and anesthesia can induce alterations in hematological parameters, thus the observed changes could be an artifact, due in part, to the surgical model itself. In contrast to mice, the effects of surgical preparation on granulocyte function NHP focal ischemia in the baboon are considerable (Ember et al, 1994) and presumably the same could be said for the rhesus macaque. Ember et al, showed that their surgical implantation procedure before occlusion in their baboon stroke model resulted in significant increases in granulocyte numbers with a return to baseline at 7 days postop. These effects were attributed to stress and anesthesia, although they also saw significant increases above this baseline level after stroke, validating these changes are potential preclinical biomarkers in NHPs. A systematic study of neutrophil distribution and phenotype in sham surgery animals compared with stroked animals will be necessary to determine the contribution of surgery on the observed hematological changes.

Tumor necrosis factor-α, IL-1β, CRP, and IL-6 are four biomarkers that have been detected in the plasma of some stroke patients and correlations with prognosis or infarct severity have been observed in some cases (reviewed by Tuttolomondo (2008)). Interestingly, levels of CRP and IL-6 were elevated at 2 days after stroke in all treatment groups irrespective of infarct size. While plasma CRP elevations are not disease specific, they can be a sensitive marker for tissue injury and inflammation. Increases in CRP and IL-6 are seen acutely after stroke in humans, and these markers are considered potential clinical biomarkers for infarct severity (Smith et al, 2004; Song et al, 2009). Importantly, CpG preconditioning had no impact on the induction of plasma IL-6 or CRP levels at 2 days after stroke. In fact, animals treated with the high dose of Kmix showed significantly higher IL-6 values. This is not likely to reflect drug activity given that Kmix ODNs are known to induce IL-6 very acutely after administration and in this study, 3 days after treatment (at the time of surgery) revealed undetectable IL-6 levels. These data imply a lack of correlation between IL-6 levels and infarct reduction. However, findings by Smith et al, (2004) show a positive correlation between plasma IL-6 levels (dynamic range of 10 to 1000 pg/mL) and infarct volume 1 week after ischemic stroke in humans, which also correlated with clinical outcome. These findings are confirmed in a later prospective cohort study by Whiteley et al, (2009). However, the value of IL-6 as a biomarker of stroke is complex, as mice deficient in IL-6 have similar stroke volume and disability at 24 hours after ischemia as wild-type mice, suggesting that IL-6 may be part of the inflammatory response to stroke, and not directly pathogenic in mice (Clark et al, 2000).

As IL-6 and other cytokines are known to contribute to the induction as well as the resolution and recovery phases of inflammatory processes, it is tempting to assign a protective role for IL-6 in this context. Interleukin-6 has been shown to exert neuroprotective (Brenneman et al, 1992) and neurotrophic effects (Hama et al, 1989). It is important to note that the dynamic range for IL-6 was very small in our study. Mean IL-6 values changed by ~5 pg/mL between control and 0.3 mg/kg groups, respectively. While these are above the level of detection of our assay and statistically significant differences were observed, these values represent relatively low concentrations in the circulation. Similarly, serum levels of IL-6 in stroke patients were significantly lower than CSF levels of IL-6 and did not correlate with the size of the brain lesion in some studies (Tarkowski et al, 1995). Also, increase in intrathecal, but not systemic production of IL-1β was observed early during the stroke in that study.

Low levels of TNF-α were detected and no differences were observed between treatment groups or between time points in individual animals (data not shown), suggesting that the levels observed were not related to stroke or Kmix treatment. Some studies have shown detectable systemic TNF-α levels after stroke in humans, whereas other studies showed predominantly central TNF-α expression, but little to no systemic levels in the majority of patients as early as 24 hours after stroke (Sairanen et al, 2001). When TNF-α levels were detected acutely, CSF samples tend to show more incidents of positivity compared with plasma with relatively low levels present, but not in all patients. Tumor necrosis factor-α appears more often detected in CSF from a subset of patients with pronounced white-matter lesions (Tarkowski et al, 1997). Our model produces mostly cortical injury and this could have contributed to the limited detection of TNF-α in our study. Alternatively, it is likely that 2 days after stroke may not have been an optimal peak expression time point for the detection of differences for these cytokines in our study. Moreover, the surgical model itself or the stress involved in animal manipulations are likely to induce some changes in the levels of these acute phase reactants or their inducers, as our model involves manipulation of the dura mater and cerebral vessels.

Modest Systemic Effects Elicited by Cytosine-Guanine-Rich DNA Oligonucleotide Treatment

Our study design did not allow for acute sample collection and this limited our ability to evaluate pharmacological biomarkers typically present within the first 24 hours after Kmix administration in primates. Although Kmix was administered systemically, no significant changes were seen in hematological parameters 72 hours after drug treatment, just before stroke induction or after stroke. This result was not entirely unexpected, as TLR preconditioning in mice appears to induce only a transient change in gene expression (Gunzer et al, 2005). Also, a TLR7 ligand given systemically in mice has been shown to induce transient reversible leukocyte depletion from the blood (Gunzer et al, 2005). These studies showed that peripheral blood leukocyte levels were reduced as early as 1 hour following TLR7 ligand treatment due to increased endothelial adhesiveness, contributing to increased leukocyte tissue residence time and were returned to normal by 24 to 48 hours. Overall, the lack of modulation of blood cell distribution observed between drugtreated (protected) and control (nonprotected) animals just before stroke may indicate that cells in the peripheral circulation are not responsible for the protective phenotype. It is also possible that peripheral blood cells have been reprogrammed by prior CpG treatment leading to diminished damage.

Following administration of Kmix CpG ODNs in our NHP study, very little detectable systemic TNF-α was observed in plasma at 72 hours postadministration, with the exception of a single animal (data not shown). This is not surprising given that the systemic increase seen in vivo in mice following CpG administration occurs rapidly (within 1 to 2 hours after drug administration), and returns to undetectable levels by 3 hours postadministration. Although it was not possible to obtain samples at early times following the administration of CpG ODNs, in the future, it will be important to examine the early acute effects of CpG ODN to examine the kinetics of potential informative biomarkers.

Preliminary Toxicology of Kmix Oligonucleotides

No human clinical studies have been conducted with Kmix and Dmix CpG ODN sequences, although historical data show that the in vivo behavior of a given CpG ODN is largely sequence independent with reasonable consistency among species (Geary et al, 2001). We have evaluated several relevant clinical parameters in our study based on existing toxicology and clinical experience with CpG ODNs. Doses used in our study were relatively conservative in nature. The reported lowest observed adverse effect level for ProMune (Coley Pharmaceuticals Inc., Wellesley, MA, USA; CpG 7909; Class B or K-type ODN) was 0.5 mg/kg per day in NHPs in a 26-week repeated dose study with daily subcutaneous or intravenous dosing according to the material safety data sheet (http://media.pfizer.com/files/products/material_safety_data/PROMUNE.pdf). These data suggest that singledose acute toxicity of K-type ODN is likely to be much > 0.5 mg/kg. Reported dose-related toxicities of Promune involved kidneys, blood, immune system, cardiovascular system, and liver as target organs in NHPs.

To evaluate potential acute dose toxicities, renal and liver toxicities were evaluated at 3 days postdrug dose and at necropsy 2 days after stroke by clinical chemistry analysis and gross organ pathology. We also monitored blood cell composition (complete blood count, differential count) and plasma acute phase proteins before and up to 5 days after drug dosing. No significant drug-associated changes in any of these parameters were observed. We monitored liver enzyme levels (aspartate amino transferases/ alanine amino transferases (AST/ALT)) in animals before and following CpG ODN treatment and did not see elevations at any time point examined. Studies currently evaluating higher doses of Kmix ODN up to 0.9 mg/kg also reveal no significant changes in drug-related toxicity parameters 3 days postdrug dose and up to 7 days after stroke (10 days postdose).

Summary

These data provide important insights for the process of stroke drug development. The NHP stroke model described here demonstrates key clinical phenotypes observed in human stroke patients. The model has good predictive value, serves as a valuable preclinical tool to validate potential therapeutic targets, and will allow for more adequate establishment of efficacious dose ranges and regimens for clinical studies in humans. The studies herein advance a TLR9 ligand into NHPs and show for the first time that antecedent therapy using CpG ODNs provides robust neuroprotection against stroke injury in this relevant animal model. In addition, CpG ODNs are highly promising therapeutics for clinical stroke prevention due to their favorable safety profile in humans. Thus, future studies are necessary and warranted to address the optimal dosing and timing of CpG ODNs, as well as neurologic improvements after stroke.