Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Many neurodegenerative diseases are accompanied by inflammation and permeability alterations in the blood-brain barrier (BBB) (Stewart, 1994; Claudio et al., 1995; McGeer and McGeer, 1995; Power and Johnson, 1995; Tomimoto et al., 1996; Zhang et al., 1996). Localized or global disruption of the BBB is suspected to contribute to the pathology of neurological diseases with an identifiable inflammatory component, viz., Alzheimer's disease (McGeer and McGeer, 1995; Tomimoto et al., 1996), multiple sclerosis (Claudio et al., 1995; Stone et al., 1995), HIV-1 dementia (Power and Johnson, 1995; Giovannoni et al., 1998), cerebral ischemia (Zhang et al., 1996), brain tumors (Stewart, 1994), and meningitis (Tunkel and Scheld, 1993; Boje, 1995a). Additionally, transient loss of BBB integrity is a putative initiating event in Rasmussen's epilepsy (Rogers et al., 1994).

A survey of the literature suggests that excessive production of nitric oxide (NO) contributes to the pathophysiology of many central nervous system (CNS) diseases (Boje, 1997; de Vries et al., 1997; Merrill and Murphy, 1997). Overproduction of endogenous CNS NO during meningitis was identified as a mediator of meningeal inflammation in both humans (Milstien et al., 1994; Visser et al., 1994) and animals (Boje, 1995b, 1996; Koedel et al., 1995). Recent studies involving the administration of NO synthase (NOS) inhibitors further implicated a significant contributory role for NO modulation of blood-cerebrospinal fluid and BBB permeabilities (Boje, 1995b, 1996; Koedel et al., 1995). Moreover, lipopolysaccharide application to pial cerebral arterioles caused an NO-dependent increase in cerebrovascular permeability (Mayhan, 1998).

The administration of pharmacological agents that release NO (i.e., NO prodrugs) coupled with an examination of their effects on the BBB provides another paradigm to understand the contribution of NO modulation of the BBB. Because NO can biochemically exist in one of several redox forms, NO^·, NO⁺, and NO (Stamler and Feelisch, 1996), it is important to explore the effects of NO redox congeners on BBB permeability. There is plausible reason to hypothesize that distinct NO redox species may have differential effects on modulation of the BBB. In the present studies, an in situ rodent brain perfusion preparation (Takasato et al., 1984) was used to administer NO prodrugs into the cerebrovascular circulation. Various classes of NO prodrugs were selected based on the liberated NO redox form (NO^·, NO⁺, and NO) and the rate of NO redox release (Feelisch, 1998). BBB integrity was then assessed by brain perfusion with [¹⁴C]sucrose, a tracer that is poorly permeable across the normally intact BBB (Takasato et al., 1984).

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. [¹⁴C]Sucrose (600 mCi/mmol in 90% ethanol), [³H]methyldiazepam (85 Ci/mmol in ethanol), [¹⁴C]iodoantipyrine (55 mCi/mmol in sterile saline), and [³H]inulin (300 mCi/g, solid) were purchased from American Radiolabeled Chemicals (St. Louis, MO). All isotopes were >99% pure. [³H]Inulin was dissolved in buffer and prepurified with Sephadex G-15 gel filtration chromatography. S-nitroso-N-acetylpenicillamine (SNAP) was synthesized and characterized as described by Field et al. (1978). Other NO prodrugs were procured from Cayman Chemical (Ann Arbor, MI). 2-Hydroxypropyl--cyclodextrin (CD; mol. wt. = 1423; average degree of substitution = 5.5) was purchased from Research Biochemicals International (Natick, MA). Soluene 350 was obtained from Packard (Meriden, CT), and Solucint-O was obtained from National Diagnostics (Atlanta, GA). Halothane was purchased from J.A. Webster (Sterling, MA).

Animal Surgery. All procedures involving animals were approved by the University of Buffalo Institutional Animal Care and Use Committee and performed according to the guidelines set forth in the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

NO Prodrug Infusion Protocol. Rats were randomly allocated to one of several treatment groups (n = 5-7 rats/group). NO prodrugs were infused into the common carotid artery (via retrograde infusion of the right external carotid artery) at 42.5 µl/min, a flow rate which constitutes <5% of the normal cerebral blood flow (Preston et al., 1995). This infusion flow rate was chosen so as to not disrupt or dilute normal CBF. The duration of the NO prodrug infusion was 1 h.

Determination of BBB Integrity after NO Prodrug Administration: Tracer Perfusion Protocol. Animals were prepared for in situ brain radiotracer perfusion after NO prodrug administration. In brief, the heart was surgically exposed and cut, the common carotid was immediately ligated, and the brain was perfused via the right external carotid artery cannula at 5 ml/min for 1 min with warmed (37°C), oxygenated [¹⁴C]sucrose perfusate (0.5 µCi/ml; a marker of BBB integrity) in bicarbonate-buffered physiological saline (Takasato et al., 1984) with or without [³H]diazepam (0.5 µCi/ml; a marker of perfusate flow) or [³H]inulin (0.5 µCi/ml; a marker of intravascular volume). The brain was harvested and bluntly dissected into various brain regions: right frontal, parietal and occipital cortices, hippocampus, hypothalamus/thalamus, and striatum. The tissues were weighed and digested with Soluene 350 overnight at 50°C. The digested samples were solubilized with 5 ml of Solucint-O and counted in a Packard 1900CA liquid scintillation analyzer (Packard Instrument Co., Downers Grove, IL). The counting efficiencies of the tissue samples were typically ~90% for ¹⁴C and ~40% for ³H. An aliquot of the perfusate also was obtained for liquid scintillation counting.

Drug Treatment Groups. Groups I and II consisted of naive rats perfused for 1 min at 5 ml/min with [³H]inulin and [¹⁴C]sucrose or [³H]diazepam and [¹⁴C]sucrose to determine the intravascular brain volume or cerebral perfusate flow, respectively. Groups I and II did not undergo a 60-min perfusion of saline before tracer perfusion. Group III animals were perfused for 1 min with hyperosmotic mannitol (1.6 M) (Ziylan et al., 1984), followed by a 1-min perfusion (at 5 ml/min) with [¹⁴C]sucrose. This group served as a positive control for BBB disruption.

Determination of Regional Cerebral Blood Flow. The [¹⁴C]iodoantpyrine method of Sakurada et al. (1978) was used to assess rCBF after NO prodrug infusion. In brief, [¹⁴C]iodoantpyrine (10 µCi/ml in saline) was infused at 1.33 ml/min via the femoral vein cannula. Immediately after the start of the iodoantipyrine infusion, the femoral artery catheter was cut to 2 cm and arterial blood samples were collected serially in preweighed vials containing absorbent filter paper. At 36 s into the infusion, the rat was decapitated immediately and the various brain regions were harvested for liquid scintillation counting as previously described. Blood samples were reweighed and allowed to dry overnight, then solubilized for 24 h with 5.0 ml of Scintiverse cocktail and 0.5 ml of water.

Data Analysis. Total [¹⁴C]sucrose accumulation in each brain region was calculated as the observed tissue disintegrations per minute (representing tracer entrapped within the parenchyma and vasculature) normalized for tissue weight and perfusate concentration (Takasato et al., 1984). Perfusion with the intravascular marker [³H]inulin yielded regional cerebrovascular inulin volumes that were used to correct each brain region for vascularly entrapped [¹⁴C]sucrose tracer. Thus, a [¹⁴C]sucrose influx clearance (Cl_in; corrected for tracer entrapped within the vasculature) was calculated as the total tissue [¹⁴C]sucrose accumulation less the cerebrovascular inulin volume.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Figure 1 depicts the changes in occipital cortex [¹⁴C]sucrose permeability after cerebrovascular infusion with various NO prodrugs. Table 1 presents the [¹⁴C]sucrose Cl_in data for the remaining brain regions. Of the thionitrite formulations, only cerebrovascular infusion of the SNAP-CD formulation elicited significant [¹⁴C]sucrose penetration across the BBB (Table 1). The control formulations, saline or CD, did not alter BBB permeability. Systemically infused SNAP-CD (via the femoral artery) also failed to alter BBB permeability. Angeli's salt was the only nitroxyl agent that caused a significant increase in [¹⁴C]sucrose Cl_in. With the exception of DETA NONOate, all other tested NONOates produced statistically significant increases in BBB permeability in each of the examined brain regions (Table 1).

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Occipital cortex [14C]sucrose Clin after NO prodrug infusion. Cerebrovascular infusion of saline and CD, as well as no drug infusion, served as negative control treatments. Infusion with hyperosmotic mannitol served as a positive control treatment. *P < .05, **P < .01, and ***P < .001 from other treatments by ANOVA with Newman-Keuls; mean ± S.E. (n = 5). cbv, cerebrovascular administration via the external carotid artery; fem, systemic administration via the femoral artery.

View this table: [in this window] [in a new window]   TABLE 1 Ipsilateral forebrain regional [14C] sucrose Clin after cerebrovascular NO prodrug administration Data are mean ± S.E. (n = 5-11 rats) and are corrected for intravascularly entrapped isotope. Clin units: cm3/g of wet wt/min × 103. Statistical significance was assessed by two-way ANOVA with Newman-Keuls post hoc test. Data with dissimilar (or no) superscripts are statistically different from other drug treatments and are statistically similar to values with the same superscript. The infusion rates for the SNAP formulation are indicated, and the infusion rate for each remaining NO prodrug was 1.7 mmol/min.

Nitric oxide (also known as endothelium-derived relaxant factor) was originally identified by its vasorelaxant effect on vascular smooth muscle. In vivo NO release from the various NO prodrugs was indirectly assessed by monitoring systemic hemodynamic responses (Tables 2 and 3). An ANOVA of the untransformed MAP after 60 min of infusion did not detect significant treatment effects (Table 2), in part due to interanimal variability and the number of comparisons made among all treatments. However, visual inspection of the data suggested potential treatment trends in the MAP response. Statistically significant NO prodrug effects on MAPs were observed when MAP was re-expressed as MAP, the percentage change in MAP from preinfusion baseline (Table 3). Irrespective of the route of administration, MAP was significantly decreased by SNAP-CD infusion (Table 3). The nitroxyl-releasing agent Angeli's salt evoked a pronounced ~40% decrease in MAP from baseline (P < .01), in contrast to the negligible MAP response observed with Piloty's acid. The MAP response to Piloty's acid was not different from infusion with control vehicle, which strongly suggests negligible NO release from this putative NO prodrug. The NONOate analogs MAHMA, diethylamine, and spermine elicited an ~40 to 50% decrease in MAP from baseline (P < .01). MAP for DETA NONOate was not statistically different, probably due to variable responses.

View this table: [in this window] [in a new window]   TABLE 3 Comparative analysis of BBB permeability after NO prodrug infusion Data are mean ± S.E. of three to six rats. The drug infusions were: each nitroxyl agent, 1.7 mmol/min; each NONOate agent, 1.7 mmol/min. Statistical significance was assessed by one-way ANOVA with Newman-Keuls post hoc test. Data with dissimilar (or no) superscripts are statistically different from other drug treatments and are statistically similar to values with the same superscript. The %osmotic opening (relative to mannitol) was calculated as (test drug Clin  mean control Clin)/(mean mannitol Clin  mean control Clin) · 100%.

Table 2 also presents cerebrovascular hemodynamic parameters after NO prodrug infusion. Importantly, there were no significant differences in CBF (average ipsilateral forebrain) or ICP when either one-way ANOVA or unpaired t test was used. Of note, rCBF data were analyzed by two-way ANOVA and no significant effects attributable to treatment or specific brain regions were observed (data not shown). In addition, NO prodrug treatment effects were not observed for CPP or CVR (Table 2) by one-way ANOVA.

For each NO prodrug, the [¹⁴C]sucrose Cl_in was averaged for all ipsilateral forebrain regions and compared with the [¹⁴C]sucrose Cl_in induced by the positive control treatment hyperosmotic mannitol (Table 3). Of the NO prodrugs, cerebrovascularly infused SNAP-CD_cbv caused the greatest BBB disruption, causing permeability changes that were ~45% of hyperosmotic mannitol. SNAP-CD_fem infused systemically failed to elicit significant BBB disruption. Angeli's salt caused an ~18% disruption, relative to hyperosmotic mannitol. The NONOates, with the exception of DETA NONOate, elicited comparatively small, yet significant BBB permeability increases that were ~5 to 8% of the mannitol treatment.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Depending on a variety of factors, NO can exist biochemically as NO^·, NO⁺, and/or NO (Stamler and Feelisch, 1996). Although not universally accepted, endogenous NO biosynthesized by constitutive NOS was suggested to be released in the nitroxyl form (i.e., NO) (Schmidt et al., 1996), and may or may not account for some of the biological activities ascribed to NO^·. The redox form(s) of NO synthesized by inducible NOS are presently unknown. However, literature evidence suggests that NOS activity during a CNS pathology may evoke permeability alterations of the BBB (Koedel et al., 1995; Boje, 1996; Merrill and Murphy, 1997; Giovannoni et al., 1998; Mayhan, 1998). The objective of the present work was to assess the permeability effects of NO redox forms (NO^·, NO⁺, and NO), as delivered from various NO prodrugs, on the integrity of the BBB.

The selection of NO prodrugs was based on the liberated NO redox form(s) and the rate of NO release. Among the selected NO prodrugs were the NONOates, which release NO^· over a range of first-order release rates. The reported NO release half-lives (in 0.1 M phosphate buffer, pH 7.4, at 37°C) of MAHMA and diethylamine NONOates are ~1 to 2 min, 39 min for spermine NONOate, and >1200 min for DETA NONOate (Anonymous, 1997). Angeli's salt, a nitroxyl-releasing agent (NO^· and NO), has a reported degradation t_1/2 of ~2 min (in 0.1 M phosphate buffer, pH 7.4, at 37°C) (Anonymous, 1997). SNAP is among the S-nitrosothiol prodrugs that release NO^·, NO⁺, and NO at rates dependent on oxygen tension, other nucleophiles (such as thiols), and trace metals (Feelisch and Stamler, 1996). The NO release half-life from SNAP is ~2.5 min at 37°C in whole, oxygenated arterial blood (K. M. K. Boje, unpublished data).

Significant BBB permeability alterations were observed in a number of brain regions after cerebrovascular infusion of various NO prodrugs (Tables 1 and 2). The NO prodrug rank order of BBB disruption was SNAP-CD_cbv > Angeli's salt > MAHMA ~ diethylamine > spermine > DETA ~ Piloty's acid ~ saline. SNAP-CD_cbv evoked the largest increase in part due to release of NO^·, NO⁺, and NO, as well as the corresponding thiol congeners, which also may participate in BBB disruption. The cerebrovascular exposure to the total NO redox congener concentrations was estimated (Table 3). Of note, the NO prodrug rank order of BBB disruption does not mirror the rank order of the absolute amount NO redox species release. For example, MAHMA NONOate released the greatest amount of total NO release (as NO^·) within the cerebrovasculature, yet only modest BBB disruption was observed. However, when viewed from the perspective of specific NO redox forms, there is a casual rank ordering of the NO redox forms with BBB disruption: prodrugs releasing a single specie, NO^· (MAHMA, diethylamine NONOates) produced a mean Cl_in of 11.1 (cubic centimeters per gram per minute × 10³); prodrugs releasing two species, viz., NO^· and NO (Angeli's salt) increased the mean Cl_in by an additional 3.8 U; and prodrugs releasing all three species, NO^·, NO⁺, and NO (SNAP-CD_cbv) increases the Cl_in by an added 15.5 U (referenced to NO^· and NO).

The systemic and cerebrovascular hemodynamic data (Table 2) suggest that disturbances in cerebral perfusion are not a likely explanation for NO prodrug-mediated BBB disruption. As expected, NO prodrugs have systemic nitrovasodilator activity with predictable effects on MAP and CVR. In spite of a general nitrovasodilator effect, ICP data were within the normal range for rodents (Kawamura and Yasui, 1994; Morimoto et al., 1996). Because the ICP data reflect central venous pressures (Miller and Bell, 1987), the stable ICP data alleviate concerns about elevated cerebral venous pressures and venule overdistention as mechanisms of BBB disruption.

It is also unlikely that BBB permeability alterations derive from changes in cerebral perfusion and CPP secondary to decreases in MAP. NO prodrug treatment did not alter rCBF in the face of diminished CPP, most likely due to an intact autoregulatory control of cerebral perfusion. Cerebrovascular autoregulatory mechanisms assure the constancy of blood flow with moderate alterations of MAP and CPP. It is known that nitrovasodilators cause the CBF-MAP curve to shift leftwards such that CBF remains constant at lower-than-normal MAPs (Moss, 1995). It is for this reason that nitrovasodilators are used for controlled hypotension during surgical procedures (Moss, 1995). The observed cerebrovascular data argue against BBB disruption attributable to cerebral hypoperfusion.

The last piece of experimental evidence to argue against systemic and cerebrovascular disturbances in cerebral perfusion is the comparison of SNAP-CD infused cerebrovascularly versus systemically. Despite similar responses in systemic and cerebrovascular hemodynamics, cerebrovascular infusion of SNAP-CD_cbv elicited significant BBB disruption, whereas systemic infusion of SNAP-CD_fem did not.

The results of these studies imply, but do not prove, that NO redox congeners exert a direct, disruptive effect on cerebral capillaries. Under pathological conditions, NO redox species are produced by glia and endothelial cells. Moreover, endothelial cells may convert NO species into other NO species. This study is limited to NO redox specie presentation to the luminal side of the BBB. Because the BBB is polarized, there may be differences between abluminal versus luminal NO exposure and response.

The precise biochemical mechanisms for NO-mediated disruption of the BBB are presently undefined. NO activates guanylate cyclase to stimulate the formation of cGMP, which mediates dose-dependent disruption of cerebral pial arteries (Chi et al., 1999). Alternatively, NO redox forms may act directly on cerebrovascular endothelial cells by modulating membrane fluidity or by attacking critical cellular or tight junctional targets. NO redox forms and ONOO (generated by the reaction of NO^· with O₂) may exert toxicity through an attack of transition metals, free sulfhydryl residues or the nucleophilic centers of deoxyribonucleic acids and tyrosine residues. NO redox species also act to depress mitochondrial respiration by nitrosylation of the iron-sulfate centers of several essential enzymes, thereby promoting a diminished rate of ATP generation and loss of intracellular iron (Stamler et al., 1992; Stamler, 1994).

The current work presents pharmacological evidence of a "proof-of-concept" to support the notion that excessive amounts of NO redox forms exert deleterious effects on the BBB. Specifically, the results of this work identify that NO^· alone exerts substantially modest BBB disruption compared with a combination of NO^· with NO or NO plus NO⁺. This inference is consistent with the comparative in vitro reactivity of NO⁺ and NO relative to NO^· (Stamler and Feelisch, 1996) and cytotoxicity studies of NO in cultured fibroblast cells (Wink et al., 1998).