Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

There is now a significant amount of information that supports a role of oxygen radical-induced LP in the pathophysiology of acute CNS injury and ischemia (Braughler and Hall, 1989; Hall and Braughler, 1989; Siesjo et al., 1989). The 21-aminosteroid (lazaroid), tirilazad, has been demonstrated to be a potent inhibitor of LP that acts by a combination of chemical radical scavenging and membrane stabilization mechanisms. It has been shown to reduce traumatic and ischemic damage in several experimental models, and a correlation has been demonstrated between attenuation of oxygen radical levels and/or LP and the neuroprotective effect in several instances (see review by Hall et al., 1994). Currently, tirilazad is being actively investigated in phase III clinical trials in head and spinal cord injury, ischemic stroke and SAH. Results from a multinational European/Australian/New Zealand trial in SAH have demonstrated a highly significant reduction in 3-month mortality and improvement in the incidence of "Good" recovery (Glasgow Outcome Scale) in patients treated with tirilazad (Kassell et al., 1996).

Tirilazad appears to act, in large part, on the CNS microvascular endothelium (Audus et al., 1991; Raub et al., 1993; Hall et al., 1994) and consequently has been shown to protect the BBB, to maintain cerebral or spinal cord blood flow autoregulatory mechanisms and/or to reduce delayed vasospasm in multiple models (Hall et al., 1994). Therefore, its ability to protect neural tissue from traumatic or ischemic insult in many models may be largely indirect. Indeed, tirilazad, most likely because of its limited penetration into brain parenchyma, has generally failed to affect delayed neuronal damage in the selectively vulnerable hippocampal CA1 and striatal regions (Beck and Bielenburg, 1990; Buchan et al., 1992; Sutherland et al., 1993), although it has some ability to protect cortical neurons (Sutherland et al., 1993). Moreover, in models of permanent focal ischemia in which microvascular effects may be less important than in temporary ischemia paradigms, the compound's ability to affect infarct size, although demonstrated in some experiments (Beck and Bielenburg, 1991; Park and Hall, 1994), has been inconsistent (Xue et al., 1992).

Thus, we reasoned that LP-inhibiting (antioxidant) compounds with improved brain penetration might possess certain advantages over the microvascularly localized tirilazad in certain CNS injury situations. Recently, we have discovered a new group of compounds, the pyrrolopyrimidines (fig. 1; Bundy et al., 1995), which are equal or better antioxidants than tirilazad, but with significantly improved ability to penetrate the BBB and gain direct access to neural tissue. The present report details the effects of these in regard to inhibition of iron-dependent lipid peroxidative injury to cultured mouse spinal cord neurons and neuroprotective activity in models of focal and global cerebral ischemia and concussive head injury.

View larger version (16K): [in this window] [in a new window]   Fig. 1.   Chemical structures of the 21-aminosteroid or "lazaroid" tirilazad (U-74006) and selected pyrrolopyrimidines. Tirilazad is a nonglucocorticoid 21-aminosteroid. The primary chemical antioxidant portion of the molecule is the amino moiety bound to the 21 position of the steroid side chain. Physicochemical studies indicate that the highly lipophilic (i.e., hydrophobic) steroid moiety orients itself within the fatty acid core of the membrane and is largely responsible for the high affinity of the compound for cell (e.g., endothelium) membranes (Hall et al., 1994). The more hydrophilic amino substitution exists closer to the surface in juxtaposition to the phosphate head groups of the phospholipids. In contrast, the pyrrolopyrimidines lack the highly lipophilic steroid moiety, but bear some structural resemblance to the bis-pyrrolidinylpyrimidinyl piperazine 21-amino substitution of tirilazad. Lack of the steroid serves to lessen the high affinity for and retention in lipid bilayers.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

All experiments received prior approval by the Institutional Animal Use and Care Committee of Pharmacia & Upjohn, Inc. to ensure that they were performed in strict compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Iron-dependent lipid peroxidative injury to cultured fetal mouse spinal neurons. Fetal mouse spinal neurons were cultured as described previously (Hall et al., 1991). Spinal neurons were dissected from 13- to 15-day-old embryos of CD-1 virus-free mice. The tissue was chemically and mechanically dissociated. The cells were resuspended in minimum essential media plus additional nutrients and transferred to 24-well, Falcon Primaria culture plates. The cells were plated at a density of approximately 6 × 10⁵/ml (total volume in well, 1 ml). Plates were maintained at 35°C in 8% CO₂/95% humidity. Cytosine arabinoside was added at a final concentration of 5 µM on day 6 and removed on day 8. Assays were run on day 15.

Determination of oxidation potential. Electrochemical oxidation followed by HPLC with UV detection was used to study the oxidation potential of the test compounds. The mobile phase consisted of 89% water, 10% methanol and 1% ammonium acetate. A 10-µl solution (0.3 µg/µl) was injected onto an HPLC system. At 1 ml/min, the mobile phase flowed into an ESA (Millipore Corp.; Milford, MA) guard cell. The guard cell was set at electrochemical potentials ranging between 0 and 1000 mV. Next, the mobile phase flowed into a Waters (Chelmsford, MA) C8 symmetry column where oxidative by-products could be separated. Flow carried column effluent into a UV detector that monitored light absorption at 230 nm. For each sample, 1 ng of pyrrolopyrimidine was injected onto the column. One minute after injection, the electrochemical potential across the guard cell was turned off until the next injection. During this 1-min period of electrochemical oxidation, electrical current output was observed to increase rapidly and then diminish to base line. The magnitude of current output was directly proportional to the electrochemical potential setting. This suggests that the pyrrolopyrimidines were oxidized in proportion to voltage across the guard cell. Quantitation of the test compound was based on the area of peaks which co-eluted with a nonoxidized sample of the test compound. Peak areas associated with oxidation were normalized to a peak that was measured under conditions where the test compound was not oxidized. The potential that resulted in a 50% loss of starting material was determined.

Determination of brain uptake in mice. Female CF-1 mice (Charles River; Portage, MI) were injected via a tail vein with 23 µmol/kg (approximately 10 mg/kg) of test compound in propylene glycol (5 mg/ml). At either 5 or 60 min after the dose, the mice were deeply anesthetized with methoxyflurane followed by intracardiac perfusion with 10 ml of 0.9% saline. The brain samples were placed in a volume of acetonitrile equal to five times the brain weight and kept on ice during processing. The samples were sonicated for 20 sec, followed by centrifugation at 2060 × g for 15 min at 4°C. The supernatants were analyzed by HPLC on a reverse phase column (Zorbax SB-CN, 250 × 4.6 mm, 5 µm particle size, Dupont, Chadds Ford, PA) with an autoinjector (Perkin Elmer ISS 100) with 100-µl injections and a flow rate of 1 ml/min. The isocratic mobile phase used consisted of various proportions of 100:1 (v/v) acetonitrile/trifluoroacetic acid and 100:0.115 (v/v) water/trifluoroacetic acid. UV detection (Applied Biosystems Spectroflow 783 Foster City, CA) was used for all compounds. However, fluorescence detection (Applied Biosystems 980) was found to be more sensitive for quantifying some of the pyrrolopyrimidine compounds, with excitation set at 320 nm and an emission cutoff filter at 389 nm. For others that were not fluorescent, electrochemical detection (Waters 460 with a glassy carbon electrode and Ag/AgCl reference electrode) was used at 700 mV.

Determination of oral bioavailability in rats. Male Sprague-Dawley rats (Charles River; Portage, MI) weighing 250 to 300 g were surgically implanted with a dosing/sampling cannula in their superior vena cava and allowed a 1-week recovery period. The animals had access to water ad libitum and were fasted overnight before and for 4 hr after each treatment. Three animals were used for the determination of oral bioavailability. They were administered approximately a 10 mg/kg i.v. bolus and a 25 mg/kg oral solution by gavage in a cross-over design, with a minimum of a week between doses. Serial blood samples (250 µl) were obtained at defined time points from before dosing through 48 hr postdosing.

Gerbil forebrain ischemia model. Male Mongolian gerbils (Tumblebrook Farms; West Brookfield, MA) weighing 45 to 55 g were anesthetized with methoxyflurane. A 1- to 2-cm midline throat incision provided access to both carotid arteries, which were occluded with microaneurysm clips. After 5 min of near-complete ischemia, the clips were removed and reperfusion allowed for 5 days. The animals were placed in a warming box with an ambient temperature maintained at 37°C during the ischemic insult and then until the animals regained their righting reflex after reperfusion. Previous studies from this laboratory have demonstrated that this warming procedure successfully maintains both rectal and brain temperatures of the animals at 35.5°C while the animals are kept in the chamber (Hall et al., 1993). Typically, the test compound was administered p.o. (3, 10 or 30 mg/kg) 30 min before induction of ischemia and again at 2 hr after reperfusion, followed by additional single oral doses at 24, 48 and 72 hr. The concentration of drug in the vehicle (40% hydroxypropyl cyclodextrin) was adjusted such that the oral administration volume was held constant at 0.1 ml.

Mouse focal cerebral ischemia model. Male CD-1 mice (Charles River; Portage, MI), weighing 18 to 22 g, were used. They had access to water and chow ad libitum. Under methoxyflurane anesthesia, an incision was made over the temporoparietal region, and the skull was exposed by retraction of the musculature and parietal gland. A small burr hole exposed the MCA. The MCA was cauterized in two places 2 mm apart and then cut, just above the origin of the lenticulostriate arteries by use of bipolar diathermy.

Mouse head injury model. Details of the mouse head injury model have been published in detail elsewhere (Hall et al., 1988). Male CF-1 mice (Charles River; Portage, MI) weighing 17 to 20 g were used. Each mouse that survived the injury received vehicle (0.05 N HCl) or test drug in a 0.1-ml volume as an i.v. bolus within 5 min postinjury. Groups of 20 to 25 mice were injured in rapid succession. Each trial was carried out in a blinded fashion, with a vehicle group and three groups receiving either 0.1, 1 or 10 mg/kg of test compound.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Comparison of inhibition of iron-dependent lipid peroxidative neuronal injury. Table 1 shows the IC₅₀ values and maximum percent protection of cultured fetal mouse spinal neurons from iron-induced lipid peroxidative injury (system described in detail elsewhere; Hall et al., 1991; Zhang et al., 1996) for tirilazad and the pyrrolopyrimidines. Protection of cellular viability was measured in terms of preservation of amino acid uptake (i.e., uptake of ³H-aminoisobutyric acid) as described and validated by Buxser and Bonventre (1981). As seen, the pyrrolopyrimidines are consistently more potent and efficacious in this in vitro model. Figure 2 shows full concentration-response graphs for the pyrrolopyrimidines U-89843E and U-101033E versus tirilazad (U-74006F). U-89843D, and even more so U-101033E, are more potent and more efficacious than U-74006F. Figure 3 shows similar graphs for U-104067E and its para-hydroxylated metabolite U-106311E. U-106311E is significantly more potent and efficacious than U-104067E, which has a concentration-response relationship quite similar to that of U-74006F (compare with fig. 2). It should be noted that neither tirilazad nor any of the pyrrolopyrimidines affected the base-line level of ³H-aminoisobutyric acid uptake (i.e., in the absence of iron; data not shown).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Protection of the viability (uptake of 3H-AIB) of cultured fetal mouse spinal cord neurons against iron-induced lipid peroxidative injury by tirilazad (U-74006F) or the three pyrrolopyrimidines U-89843E, U-94430E or U-101033E. Values are means ± standard error for triplicate determinations. Cultures were pretreated with test compound in varying concentrations for 1 hr. The medium containing the compound was then removed and the treated cultures exposed to 200 µM ferrous ammonium sulfate for 40 min, followed by determination of 3H-AIB uptake. The absolute level of uptake-related radioactivity in iron-exposed, non-drug-treated cultures ranged from 729 ± 65 to 1640 ± 238 cpm. CON, control.

View larger version (16K): [in this window] [in a new window]   Fig. 3.   Protection of the viability (uptake of 3H-AIB) cultured fetal mouse spinal cord neurons against iron-induced lipid peroxidative injury by the pyrrolopyrimidines U-104067F or U-106311E. Values are means ± standard error for triplicate determinations. Cultures were pretreated with test compound in varying concentrations for 1 hr. The medium containing the compound was then removed and the treated cultures were exposed to 200 µM ferrous ammonium sulfate for 40 min, followed by determination of 3H-AIB uptake. The absolute level of uptake-related radioactivity in iron-exposed, non-drug-treated cultures ranged from 483 ± 30 to 3629 ± 987 cpm. CON, control.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Probable antioxidant mechanisms of the pyrrolopyrimidines. Antioxidant mechanism I, which is relevant to U-87663, U-89843, U-94430 and U-101033, involves reaction of an alkoxyl (LO·), peroxyl (LOO·) or hydroxyl (·OH) radical in which the pyrrolidine nitrogen quenches the radical species via the transfer of an electron. The pyrrolopyrimidine, in turn, becomes a radical cation. The cationic center is transferred to the 5-position of the pyrrolopyrimidine which can then react with (i.e., trap) a second radical species, most likely a peroxyl radical. This particular scenario is unlikely in U-104067. Rather, its antioxidant capacity is probably dependent on para-hydroxylation of the aromatic ring, resulting in a phenolic-type electron-donating antioxidant (U-106311). Indeed, U-106311 has a much lower oxidation potential than U-104067 (see table 1).

Comparison of brain uptake. The comparative brain uptake of tirilazad (U-74006F) and the pyrrolopyrimidines in mice after i.v. administration of molar equivalent doses (approximately 10 mg/kg) is shown in figure 5. Each of the pyrrolopyrimidines produced significantly higher brain levels 5 min after injection compared with tirilazad. For U-87663E, U-89843E, U-94430E, U-101033E and U-104067F, the brain levels at the initial time point were 4.7, 3.2, 3.7, 2.8 and 5.1 times higher, respectively, than the levels of tirilazad. For U-87663E, the brain levels were still 2.3 times higher than tirilazad at 60 min postinjection. It is also clear that the brain levels seem to fall off faster for the pyrrolopyrimidines as a further reflection of their greater membrane permeabilities (Sawada et al., 1995a, b). In other words, they diffuse into the brain and, in the absence of repeated dosing, they may diffuse out of the brain more quickly than tirilazad. Additionally, we have taken advantage of the intrinsic fluorescent properties of U-87663E to demonstrate unequivocally by fluorescence microscopy that this prototype pyrrolopyrimidine efficiently penetrates the BBB in mice after i.v. dosing (fig. 6). Moreover, it concentrates in brain parenchyma and not the cerebrovascular endothelium (fig. 6, inset).

View larger version (34K): [in this window] [in a new window]   Fig. 5.   Comparison of the mouse brain levels of U-74006F and selected pyrrolopyrimidines at 5 or 60 min after a 23 µM/kg (approximately 10 mg/kg) i.v. bolus. Mice were perfused with saline at sacrifice to eliminate any contribution of blood to the measured levels. Values are means ± standard error for n = 3. Asterisk indicates P < .05 versus U-74006F at the same time point.

View larger version (K): [in this window] [in a new window]   Fig. 6.   Low magnification fluorescence photomicrograph of a slice of mouse cerebral cortex, with dura mater intact, showing penetration of the pyrrolopyrimidine U-87663 into brain parenchyma after a 23 µM/kg (approximate 10 mg/kg) i.v. bolus. The color scale shows relative drug-associated fluorescence pseudocolor intensity (white = highest concentration). The black holes represent brain microvessels. The circular structure on the right is a pial artery. Note that vessel walls show relatively low drug-related fluorescence compared to parenchyma (higher magnification inset).

Comparison of oral bioavailability. Table 2 shows the comparative oral bioavailability and terminal plasma half-lives of tirilazad (U-74006F) versus several pyrrolopyrimidines. The bioavailability of the included pyrrolopyrimidines ranges from a low of 20.7% (U-106311E) to a high of 84.6% (U-89843E). The terminal half-lives range from 2.7 hr (U-106311E) to 22.1 hr (U-87663E). These values are considerably higher than those for the 21-aminosteroid U-74006F.

Pyrrolopyrimidine neuronal protection in transient forebrain ischemia. As noted above, tirilazad has been demonstrated to have very limited ability to attenuate selective hippocampal CA1 vulnerability in models of transient forebrain ischemia (Beck and Bielenburg, 1990; Buchan et al., 1992; Sutherland et al., 1993). This is most likely caused by its limited BBB penetration in the context of models where BBB permeability is minimally compromised. In contrast, figure 7 illustrates that oral preischemic treatment (plus repeated postreperfusion treatment) with any of the BBB-permeable pyrrolopyrimidines (U-94430E, U-101033E and U-104067F) produces significantly more CA1 neuronal protection than that observed in vehicle-treated animals. Figure 8 shows the dose-response curve for the ability of U-101033E to protect the CA1 region. Dose levels of 10 or 30 mg/kg (×5) are significantly effective, but doses as low as 1 and 3 mg/kg appear to have some effect. Figure 9 displays the therapeutic window for the efficacy of U-101033E in regard to CA1 protection in the gerbil 5-min forebrain ischemia model. The initiation of dosing 30 min before ischemia (plus repeated postreperfusion dosing) is the most effective. However, a delay in dosing to 4 hr after reperfusion still provides a statistically significant neuroprotective effect.

View larger version (25K): [in this window] [in a new window]   Fig. 7.   Attenuation of hippocampal CA1 neuronal loss of the pyrrolopyrimidines U-94430E, U-101033E and U-104067F at 5 days after a 5-min episode of bilateral carotid occlusion in male Mongolian gerbils. n = 10 animals/group. Gerbils were dosed with 30 mg/kg p.o. preischemia plus 2 hr after reperfusion and once daily on days 2, 3 and 4. Asterisk indicates P < .05 versus the vehicle-treated group (Veh).

View larger version (32K): [in this window] [in a new window]   Fig. 8.   Dose response for the effect of U-101033E to salvage hippocampal CA1 neurons at 5 days after a 5-min episode of bilateral carotid occlusion in male Mongolian gerbils. n = 10 animals/group. Gerbils were dosed with each dose level p.o. preischemia plus 2 hr after reperfusion and once daily on days 2, 3 and 4. Asterisk indicates P < .05 versus the vehicle-treated group (Veh).

View larger version (37K): [in this window] [in a new window]   Fig. 9.   Therapeutic window for the effect of U-101033E to salvage hippocampal CA1 neurons at 5 days after a 5-min episode of bilateral carotid occlusion in male Mongolian gerbils. n = 10 animals/group. Gerbils were dosed beginning at each time point with 30 mg/kg p.o. plus 2 hr later and on the subsequent days. Asterisk indicates P < .05 versus the vehicle-treated group (Veh).

Pyrrolopyrimidine neuroprotection in permanent focal ischemia. U-101033E has also been examined for the ability to limit brain infarct volume in mice subjected to permanent MCA occlusion. Mice received two doses of test compound, one at 5 min and a second at 60 min postocclusion. As observed in figure 10, U-101033E potently reduced infarct volume by as much as 27% at a dose of only 0.1 mg/kg (×2) i.v. In addition, U-87663E and U-89843E have also been found to be significantly effective in the mouse permanent MCA occlusion model, whereas tirilazad (0.1-3.0 mg/kg × 2) has shown only nonsignificant trends toward infarct reduction (data not shown).

View larger version (79K): [in this window] [in a new window]   Fig. 10.   Dose-response for U-101033E reduction of 6 hr infarct size in male CF-1 mice subjected to permanent unilateral occlusion of the right MCA. The mice received one dose of compound i.v. at 5 min postocclusion and a second dose at 1 hr. Values are percent reduction in mean infarct volume (summation of eight cross-sections through the infarct) compared with a vehicle-treated group of animals. n = 8 to 13 animals/group. Asterisk indicates P < .05 versus vehicle.

Pyrrolopyrimidine enhancement of early neurological recovery in brain-injured mice. Figure 11 demonstrates the ability of acutely administered (5 min postinjury) intravenous U-101033E to enhance the early (1 hr) neurological recovery of male CF-1 mice subjected to a severe concussive head injury. A dose-related effect is seen over the range of 0.1 to 10 mg/kg. A similar magnitude of effect has been reported for several other pyrrolopyrimidines (Bundy et al., 1995) and for tirilazad (Hall et al., 1988, 1992).

View larger version (45K): [in this window] [in a new window]   Fig. 11.   Dose-response for the effect of U-101033E to improve the early (1 hr) neurological recovery of male CF-1 mice after a severe concussive head injury. n = 6 to 9 mice in the U-101033E-treated groups and 23 mice in the vehicle-treated group (V). The data are presented as percent decrease in infarct size. However, the statistical analysis was actually performed using the raw percent infarct size values.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

This report describes a novel group of potent inhibitors of iron-dependent LP in neural tissue, the pyrrolopyrimidines (e.g., U-87663E, U-89843D, U-101033E). Members of the series have also been reported to inhibit in vitro neuronal injury by peroxynitrite (Fici et al., 1996). With both iron and peroxynitrite-induced cellular injury, the pyrrolopyrimidines show increased potency and efficacy compared with the 21-aminosteroid tirilazad. In addition, the pyrrolopyrimidines possess improved BBB permeability and brain parenchymal penetration compared with tirilazad which remains largely localized in the cerebral microvascular endothelium (Raub et al., 1993; Hall et al., 1994). In previously reported experiments, the comparative brain uptake of tirilazad and selected pyrrolopyrimidines has also been evaluated in rats in terms of first-pass extraction of radiolabeled compounds after intracarotid injection (Sawada et al., 1995b; Hall et al., 1995). With this technique, tirilazad is only 6% extracted, which is not much better than the extraction of sucrose or thiourea to which the BBB is essentially nonpermeable. In contrast, the pyrrolopyrimidine U-87663E is 83% extracted on the first pass through the cerebral circulation, and U-89843E is 88% extracted. These extraction values are very near that measured for the freely BBB-diffusible butanol. Thus, the pyrrolopyrimidines appear to permeate the BBB quite readily. Confocal laser microscopy has also been used to show that U-87663E readily gains access to the intracellular space of cultured cells (Sawada et al., 1995a, b). As a result of the greater brain and perhaps intracellular penetration, compounds like U-101033E have neuroprotective efficacy in attenuating selective neuronal damage in highly vulnerable regions, such as the CA1 area of the hippocampus, after a transient episode of forebrain ischemia. In contrast, tirilazad is largely ineffective in salvaging CA1 neurons in rodent forebrain ischemia paradigms (Beck and Bielenburg, 1990; Buchan et al., 1992; Sutherland et al., 1993; Hall et al., 1995).

Other purportedly brain-penetrable antioxidants have been described previously with neuroprotective efficacy in transient forebrain ischemia models in either gerbils or rats, including LY-178002 (Clemens et al., 1991), PBN (Phillis and Clough-Helfman, 1990) and dimethylthiourea (Pahlmark et al., 1993). However, much higher doses of all of these compounds are required to achieve neuroprotection, whereas U-101033E is significantly effective at oral dose levels as low as 10 mg/kg. This suggests that these earlier described and studied compounds either may not be as brain-penetrable as thought or perhaps they are not as effective in attenuating oxygen radical-induced, iron-catalyzed LP as the pyrrolopyrimidines. In addition, U-101033E has been shown to have at least a 4-hr postischemic therapeutic window. In contrast, PBNs ability to protect CA1 neurons in the identical gerbil forebrain ischemia model, which requires higher mg/kg doses (100 mg/kg) than those presently used, is lost by 2 hr after reperfusion (Phillis and Clough-Helfman, 1990). Nevertheless, further study of U-101033E and other pyrrolopyrimidines is necessary before an exact assessment of their neuroprotective potency and efficacy can be established in comparison with the earlier described antioxidant compounds.

The pyrrolopyrimidines similarly outperform tirilazad in the context of permanent focal cerebral ischemia. In the face of permanent vascular occlusion, a successful neuroprotective compound must intuitively be able to penetrate the underperfused ischemic penumbral zone to be optimally effective in salvaging the still viable, but potentially doomed, neural tissue. Although not directly determined in the current study, it is likely that U-101033E is able to penetrate the ischemic penumbra more effectively than the microvascularly localized tirilazad. Although tirilazad has been reported to reduce infarct volume in the setting of permanent MCA occlusion in Sprague-Dawley (Park and Hall, 1994) and Fischer (Beck and Bielenburg, 1991) rats, it has not been shown to be efficacious in the same model in the spontaneously hypertensive rat strain (Xue et al., 1992). Likewise, tirilazad is only marginally effective in the mouse permanent MCA model (Hall et al., 1995). Similarly, another antioxidant, dihydrolipoate, has not shown activity in the mouse permanent MCA occlusion model (Prehn et al., 1992). In contrast, the nitrone spin-trapping agent PBN is effective in reducing infarct size in the rat permanent MCA occlusion model although high doses (100 mg/kg i.p.) are required (Cao and Phillis, 1994). However, PBNs protective efficacy in the mouse model has not been evaluated.

A greater efficacy of the pyrrolopyrimidines has also been seen in the context of temporary focal ischemia. Rats subjected to 90 min of MCA occlusion showed a 47% smaller brain infarct 7 days later when pretreated with a 3 mg/kg i.v. dose of U-101033E, plus 3 mg/kg i.v. 15 min before and 1 hr after reperfusion. Equivalent dosing with the 21-aminosteroid U-74389G (16-desmethyl tirilazad) only achieved a 17% mean infarct size reduction, which was not statistically significant. U-101033E, but not U-74389G, also improved early postreperfusion neurological recovery (Schmid-Elsaesser et al., 1996). Similarly, PBN has been reported to reduce infarct size significantly in the rat temporary MCA occlusion model (Zhao et al., 1994).

Interestingly, in regards to severe concussive brain injury, the degree of acute neurological recovery enhancement observed with U-101033E is similar to, but not greater than, that reported for tirilazad (Hall et al., 1988, 1992). However, in the mouse concussive injury model, significant BBB disruption is observed as early as 5 min postinjury (Hall et al., 1992). Thus, tirilazad is better able to penetrate into brain parenchyma and therefore gain access to the endangered central neurons, perhaps as well as U-101033E. Indeed, the uptake of tirilazad is greater in injured than in noninjured mouse brain (Hall et al., 1992). Nevertheless, despite this greater posttraumatic BBB penetration, much of tirilazad's neuroprotective action may still be mediated at the microvascular level (Hall et al., 1994). Secondary microvascular damage in brain injury, and its potential attenuation by the microvascularly localized tirilazad, may be as important as parenchymal neuronal injury mechanisms which would be most efficiently countered by the brain-penetrable pyrrolopyrimidines. In any event, it will be interesting to investigate the combination of tirilazad, which is largely localized in the brain microvascular endothelium (Audus et al., 1991; Raub et al., 1993; Hall et al., 1994), with U-101033E, which permeates the vascular wall and exerts direct neuroprotection. Although it is likely that, within the overall spectrum of traumatic and ischemic CNS insults, antioxidant compounds that localize in brain microvasculature or that penetrate the brain parenchyma will have specific therapeutic roles to play, in some cases they may be complementary.

It has been increasingly recognized that oxygen radical-induced neuronal damage may also have an important role in the pathogenesis of the major chronic neurodegenerative diseases. These include amyotrophic lateral sclerosis (Rosen et al., 1993; Gurney et al., 1996), Parkinson's disease (Cohen, 1986; Jenner et al., 1992; Youdim et al., 1993) and Alzheimer's disease (Subbarao et al., 1990; Smith et al., 1994). Thus, the pyrrolopyrimidines may be useful in slowing the progression of these disorders by virtue of their excellent oral bioavailability, brain penetrability and antioxidant neuroprotective efficacy.