NOS Inhibitors Decrease Hypoxia-induced ATP Reductions in Respiring Cerebrocortical Slices 

Our protocol for obtaining healthy, perfused, respiring neonatal cerebrocortical slices was approved by the University of California San Francisco Committee on Animal Research. This protocol was recently described, [14,15] along with details regarding the approximately 50-[micro sign]m-thick "injury layer" (where slices were cut from brain) and in situ stress gene responses to hypoxia. Protocols for maintaining perfused, respiring slices in a 20-mm NMR tube and protocols for³¹P NMR spectroscopy and representative NMR spectra also have been described for hypoxia and recovery. [16,17] In each experiment, 80 live cerebrocortical slices (350 [micro sign]m thick; approximately 3.2 g total wet weight) were obtained from twenty 7-day-old Sprague-Dawley rats (CD rats; Charles River Laboratories, Wilmington, MA) and perfused in a 20-mm diameter glass NMR tube. The rat pups were killed by decapitation, and each brain was removed from the cranial cavity within 30 s. Four cortical slices were obtained rapidly from the lateral surfaces of the left and right hemispheres by sliding the brain past a blade fixed 350 [micro sign]m above a flat lubricated surface. Once obtained, slices were rinsed twice in oxygenated artificial cerebrospinal fluid (oxy-ACSF) and then put in fresh oxy-ACSF and allowed to recover metabolically for 3 h. The ACSF consisted of a modified Krebs balanced salt solution containing 124 mM NaCl, 5 mM KCl, 1.2 mM KH²PO⁴, 1.2 mM MgSO⁴, 1.2 mM CaCl², 26 mM NaHCO³, and 10 mM glucose. The bicarbonate buffer for oxy-ACSF, maintained by continuous bubbling with a 95% oxygen and 5% carbon dioxide gas mixture, had constant values of carbon dioxide tension (P^CO(2);[approximate] 40 mmHg), oxygen tension (P^O(2); 600–650 mmHg), and pH ([approximate] 7.4). The bicarbonate buffer for hypoxic ACSF was made by equilibrating the ACSF with a gas mixture containing 95% nitrogen instead of 95% oxygen. During the 60-min hypoxia period, the gas region above the fluid of the tissue perfusion chamber was flushed gently with 100% nitrogen. After hypoxia, a third source of ACSF was used. It had no NOS inhibitors, and, similar to the prehypoxia ACSF, was equilibrated with a gas mixture that was 95% oxygen and 5% carbon dioxide.

After procurement, slices were rinsed in oxy-ACSF and transferred to a 20-mm diameter NMR tube that served as the perfusion chamber. The slice ensemble was perfused with fresh oxy-ACSF at 37 [degree sign]C at a flow rate of 20 ml/min. The perfusate was not recycled. Two peristaltic pumps, one on each side of the NMR tube, prevented hydrostatic pressure from accumulating in the tubes. A flow valve near the NMR magnet allowed rapid switching from one type of ACSF to another.

The NMR experiments began 120 min after metabolic recovery from decapitation ischemia. (This was defined as t = 0, a time when there was full recovery of intracellular pH [pHⁱ] to 7.1 and full recovery of the phosphocreatine:ATP ratio to approximately 1.5 [in vivo phosphocreatine:ATP]). Five-minute NMR acquisitions resulted in spectra from which we determined ATP, phosphocreatine, and intracellular pH values, using methods described previously. In previous studies, our slices were robust, and NMR spectra remained constant for more than 16 h if hypoxia was not induced (i.e., if slices were perfused only with oxygenated media). Using this same slice model, we also recently found in two molecular biology studies that there was no heat shock protein 70 mRNA induction in control slices, whereas heat shock protein 70 mRNA is induced in slices if they are made hypoxic subsequent to metabolic recovery. We also showed recently that postdecapitation c-fos mRNA induction decreases in control slices but increases again if subsequently slices are made hypoxic. [14] Our previous NMR and molecular biology studies thus indicate that slices in our model are carefully obtained and relatively healthy. [18]

After approximately 120 min of postdecapitation metabolic recovery (t = 0; i.e., when normal in vivo values for the phosphocreatine:ATP ratio and intracellular pH were detected), [16,17] the experiments began. The NMR spectra were obtained every 5 min. After 60 min of pretreatment with an NOS inhibitor in hyperoxic ACSF, the slices were subjected to 60 min of hypoxia (P^O(2) < 5 mmHg by polarographic oxygen electrode) and the same concentrations of NOS inhibitor that were present during hypoxia. At the end of the hypoxic period, the slices were perfused again with hyperoxic ACSF without NOS inhibitors present. Acquisition of NMR spectra continued for 120 min after the end of hypoxia, during hyperoxic reperfusion.

Preliminary studies of the three NOS inhibitors were performed at the following doses: 16–41 [micro sign]M 7-nitroindozole (7-NI), 16–35 [micro sign]M L-nitroarginine methyl ester (L-NAME), and 10–54 [micro sign]M L-nitroarginine (L-NOarg). For 7-NI, no protection against hypoxic energy failure was observed for concentrations less than 16 [micro sign]M. In addition, concentrations at or greater than 41 [micro sign]M resulted in energy failure during the pretreatment period, indicating toxicity. For L-NAME, no hypoxic protection was observed at concentrations less than 16 [micro sign]M, and normoxic toxicity was observed at doses greater than 35 [micro sign]M. No dose of L-NOarg provided hypoxic protection, and no dose showed toxicity up to 54 [micro sign]M. We did not study L-NOarg concentrations greater than 54 [micro sign]M. Based on these preliminary results, the same dose, 27 [micro sign]M, of each inhibitor was chosen for comparison studies. Preliminary studies were also conducted for L-NAME and 7-NI to choose the doses for increasing NO during the first 60 min of posthypoxic recovery. The NO donor S-nitroso-N-acetyl-penicillamine (SNAP) was tested at 9 [micro sign]M, 18 [micro sign]M, and 27 [micro sign]M; and the NOS substrate L-NOarg was tested at 9 [micro sign]M, 27 [micro sign]M, and 54 [micro sign]M. The final choices of SNAP and L-NOarg concentrations were 27 [micro sign]M and 54 [micro sign]M, respectively, for studies with 7-NI and were 9 [micro sign]M and 27 [micro sign]M for studies with L-NAME.

Each of the three studies with NOS inhibitors were performed using 27 mM of the inhibitor and repeated three times (i.e., on different days, using brain slices from different animals). In addition, the effects of SNAP and L-NOarg administration during the first 60 min of posthypoxic recovery included two more sets of three experiments performed at 27 [micro sign]M L-NAME and at 27 [micro sign]M 7-NI. To obtain extracts for the nitrotyrosine measurements, nine additional experiments were conducted. Thus, the number of NOS inhibitor slice experiments was 30, or 9 +(6 + 6)+ 9. In addition to these and the preliminary studies at other NOS inhibitor dose levels, nine control experiments were conducted: three with hypoxia, but without NOS inhibitors, and six with NOS inhibitors and no hypoxia.

Randomly chosen slices were removed at five different predetermined times in the protocol:(1) after decapitation, (2) at the end of decapitation recovery (the start of the NMR studies), (3) at the start of hypoxia, (4) at the end of hypoxia, and (5) at the end of recovery. The slices were fixed in 10% formalin, sectioned, stained, and examined as described below. As in previous studies, [15–17]³¹P metabolite concentrations were measured relative to corresponding NMR signal intensities in the control data (t = 0). Relative ATP concentrations were determined using the [Greek small letter beta]-ATP resonance at 16.3 ppm, and pHⁱwas calculated from the P^{i-phosphocreatine}chemical shift difference. [16,17]

Measurement of NOS activities in perfused brain tissue slices was based on the conversion of¹⁴C-L-arginine to¹⁴C-citrulline and adapted from published methods. [19,20] Briefly, individual slices, removed at the times listed before for NOS activity measurements, were rapidly weighed, washed, and homogenized in a buffer containing 25 mM TRIS, pH 7.4, 1 mM EDTA, and 1 mM EGTA. The homogenate was sonicated, transferred to microcentrifuge tubes, and centrifuged at 16,000g for 15 min at 4 [degree sign]C. Supernatants containing the enzyme extract were used for NOS assays. For each sample, an aliquot of 25 [micro sign]l enzyme extract was mixed with 100 ml reaction cocktail and 10 ml CaCl², 6 mM. The reaction cocktail contained 1 [micro sign]Ci/ml¹⁴C-L-Arginine, 10 [micro sign]M (Bio-Rad Life Science Research, Hercules, CA), 25 mM TRIS, pH 7.4, 3 [micro sign]M tetrahydrobiopterin, 1 [micro sign]M flavin adenine dinucleotide, 1 [micro sign]M flavin mononucleotide, and 100 nM calmodulin. Samples were incubated at room temperature (22 [degree sign]C) for 15 min. The reaction was stopped by adding 4 ml stop buffer, which consisted of 50 mM HEPES and 50 mM EDTA at pH 5.5. A 2-ml volume of reaction sample was applied to the top of a 50 x 8–400 Dowex column that were washed and equilibrated with stop buffer. A total of 5 fractions (1 ml each) from runoffs were collected into 7-ml borosilicate liquid scintillation vials filled with 6 ml scintillation cocktail (Ultima Gold; Packard Instrument Co., Meriden, CT). Quantitation was performed on a Packard Tricarb 1900 liquid scintillation counter using an appropriate background correction derived from boiled brain slice tissue and a¹⁴C-L-arginine reference that did not contain tissue extract.¹⁴C-L-arginine not consumed in the enzymatic reaction was retained in the column, allowing only¹⁴C-citrulline to be eluted from the column. The fractions collected contained the¹⁴C-citrulline.

Slices taken from the perfusion chamber at different times (before ischemia, at the end of ischemia, and at the end of reperfusion recovery) were transferred to tubes containing 10% formalin in phosphate-buffered 0.9% saline solution (pH [approximate] 7). After 24 - 72 h, the tissues were washed, dehydrated, and embedded in paraffin. Adjacent sections, 20 [micro sign]m thick, were cut in the plane of the slices, stained with cresyl violet (Nissl stain), and examined with a phase-contrast light microscope (model 20; Zeiss Axioskop, Thornwood, NY). Representative fields of 50 - 140 adjacent cells, corresponding to areas of approximately 10⁴[micro sign]m², were used to compare the shrunken and dead neurons.

Nitrotyrosine and tyrosine concentrations were measured in perchloric acid extracts that were lyophylized, frozen, and shipped frozen to the laboratory of Dr. M. Flint Beal. Reconstituted samples were analyzed using a high-performance liquid chromatography (HPLC) system that quantifies 3-nitrotyrosine using 16-electrode electrochemical detection. [21] The maximal oxidation potential was 840 mV, with a column retention time of 24 min (for injection, separation, and elution). Four HPLC determinations were made from each vial to ensure that instrument variability was minimal. Because we had data for three NOS inhibitors at each of three time points, it took nearly 3 days to conduct the HPLC measurements for the nine vials of pooled extracts. Practical considerations forced us to pool the extracts for each time point from three separate experiments, rather than measure nitrotyrosine separately for the three experiments that contributed to each time point. However, some of us recently measured brain nitrotyrosine levels in rodents, using similar tissue volumes and the same apparatus, and we found that biovariability was 10% or less. [22] The HPLC measurements are reported as the ratio of two ratios:(1) 3-nitrotyrosine:tyrosine (NT:T) measured at the time of interest, measured relative to (2) NT:T in the extract taken for the control group (t = 0) for that NOS inhibitor. [22] This yields a measure of the change in percentage of free, solubilized tyrosines that are nitrosylated. (Tyrosine nitrosylation in the precipitated protein of the pellet was not measured.)

Data are reported as the mean +/- SD. Average relative NMR metabolite values for different times were compared first with a repeated-measures analysis of variance to determine whether relative metabolite values appeared unchanged throughout time (i.e., if they were consistent with the null hypothesis). If the null hypothesis was rejected for a particular metabolite, data for that metabolite were subjected to a multiple comparisons test. For this we used Bonferroni corrected t tests, as in previous studies. [14–18] For the HPLC studies of tyrosine nitrosylation, t tests were used to compare data from the two final time points. The entire statistical analysis assumed that slices are identical and independent. Because one slice experiment consists of four brain slices from each of 20 animals (i.e., for a total of 80 slices per tube), and because physiologic conditions are so well controlled, biovariability is very low from experiment to experiment. Computations were performed using the computer program StatView 4.5 (Abacus Concepts, Berkeley, CA).

(Figure 1) shows the time dependence of intracellular energy failure during hypoxia and the recovery of intracellular energy after hypoxia. Relative values of intracellular ATP (Figure 1A), phosphocreatine (Figure 1B), and intracellular pH (Figure 1C), averaged for three experiments, are exhibited for three NOS inhibitors: 7-NI, L-NAME, and L-NOarg. The most striking result is that, during hypoxia, 60-min pretreatment with L-NAME resulted in significant preservation of ATP (only [approximate] 35% ATP loss, compared with total loss in untreated slices; P <or= to 0.01). In the group pretreated with 7-NI, there was less loss of ATP ([approximate] 75% loss compared with the total loss that occurred in untreated slices and slices treated with L-NOarg; P <or= to 0.01). The ATP losses with L-NAME and 7-NI were significantly different (P = 0.01). Full ATP and phosphocreatine recovery occurred in both of these groups (Figure 1, A and B). These data indicate that acute, hypoxia-induced reductions in ATP concentrations in cerebrocortical slices, which normally are rapid and complete, are reduced substantially by pretreatment with two NOS inhibitors, 7-NI and L-NAME, but not by pretreatment with the NOS inhibitor L-NOarg. In slices pretreated with 7-NI and L-NAME (Figure 1C), pHⁱdecreased significantly during hypoxia to approximately 6.8 from 7.1 (P = 0.003) and then returned to control values during hyperoxic recovery. However, the pH (ⁱ) of untreated slices and those pretreated with L-NOarg decreased to approximately 6.4 during hypoxia and returned only to approximately 6.7 during hyperoxic recovery (significantly different, P = 0.003).

(Figure 2, A and B) show that posthypoxia ATP recovery is retarded after administration of the NO donor SNAP or the NOS substrate L-NOarg during recovery from hypoxia. The top curve in Figure 2A, in which the x axis starts at the end of hypoxia (t = 120 min), shows the same data for slices treated with L-NAME alone that are shown in Figure 1A. Data for L-NAME-treated slices that also were treated with SNAP or L-NOarg during the first 60 min of recovery were added to the plot. Figure 2B shows ATP data from a similar set of experiments with 7-NI-treated slices. In slices pretreated with 27 [micro sign]M 7-NI, 27 [micro sign]M SNAP decreased ATP recovery by approximately 50%(P <or= to 0.003); lower concentrations of SNAP had no effect on ATP concentrations. Figure 2B also shows that administering 54 [micro sign]M L-NOarg during the first 60 min after hypoxia induced a partial reversal of the protection of 7-NI (P = 0.003). In contrast, as seen in Figure 2A, the ATP protection provided by L-NAME was eliminated with lower concentrations of SNAP (9 [micro sign]M) and L-NOarg (27 [micro sign]M; P = 0.003 for both). For 7-NI- and L-NAME-treated slices, metabolic recovery occurred at t = 240 min when SNAP was removed from the media, but not when L-arginine was removed (P = 0.03).

(Figure 3) shows representative cresyl-violet-stained, 20-[micro sign]m-thick, transverse sections from slices taken 60 min after the end of hypoxia. Compared with cells in slices treated with L-NAME (Figure 1A) and 7-NI (Figure 1B), shrunken nuclei and signs of cell swelling can be seen in sections from slices treated with L-NOarg (Figure 1C) and in untreated slices (Figure 1D). Qualitative comparisons show that histologic features of sections obtained from slices treated with 7-NI and L-NAME were minimally different from control normoxic slices.

(Table 1) summarizes the results of NOS activity assays obtained for homogenates of washed slices that were treated with an NOS inhibitor, before and after hypoxia. Total NOS activity in slice homogenates was almost completely eliminated by L-NAME, 7-NI, and L-NOarg, with rates always being less than approximately 3% of control rates. These results confirm that NOS inhibitors were present in the slice preparation during the experiment. Inhibition of enzyme activity in intact slices might differ from that in slice homogenates because inhibitors have easier access to the enzymes in homogenates.

(Figure 4) shows the results of HPLC measurements of NT:T. During hypoxia, the NT:T ratio increased by 52% in slices treated with 7-NI, and by 200–300% in the slices treated with 7-NI and L-NOarg. The difference in the NT:T change for the 7-NI group (close to zero), compared with the other groups, is significant (P < 0.05). Estimated errors for each data point are +/- 10%, based on our previous use of the technique. [22]Figure 4also shows that there was no detectable nitrotyrosine was present at the end of the pretreatment period in extracts of slices pretreated with 27 mM L-NAME, and that hypoxia did not change the NT:T ratio significantly for 7-NI-treated slices.

The significant finding of this study is that pretreatment with the NOS inhibitors L-NAME and 7-NI maintains higher concentrations of ATP in brain slices during hypoxia. This finding is consistent with several mechanisms noted previously and reviewed extensively. [23] It is also consistent with an earlier in vivo neonatal rat study in which NOS neurons previously were killed [24] and in vivo studies of nNOS knockout mice, which had smaller brain infarcts and injury after oxygen deprivation and N-methyl-D-aspartate receptor excitation than did non-knockout siblings. [25,26] However, our finding that NOS inhibition improves ATP preservation at first glance might seem to be inconsistent with a recent NMR cerebrocortical slice study b Tasker et al., [27] who studied slices from 21-day-old rats and 7-day-old rat pups. They induced "in vitro ischemia," using methods similar to ours and then administered high potassium concentrations to depolarize neurons. Tasker et al. [27] also evaluated the effects of several compounds during the post-ischemia recovery period, including the NO donor methylene blue. Methylene blue administration augmented postischemic recovery of ATP in slices from 7-day-old rats, which was the opposite of our finding, but methylene blue impeded ATP recovery in 21-day-old rats. However, it is not appropriate to compare our data directly with data from Tasker et al. [27] because our hypoxia protocol had a longer duration of in vitro ischemia (60 min compared with 20 min in the study by Tasker et al. [27]), and it also differed in other ways (e.g., Tasker et al. [27] stopped glucose administration during hypoxia, but we did not; and our protocol did not include potassium-induced depolarization). "End ischemia" values of ATP, phosphocreatine, and Delta pHⁱin the study by Tasker et al. [27] were 74 +/- 11%, 75 +/- 1.5%, and -0.14 +/- 0.03, respectively, compared with 0 (< 3%), 0 (< 3%), and (-0.67 +/- 0.03) in our study. We believe that our findings are not in conflict with those of Tasker et al. [27] because significant differences existed in their hypoxia protocol compared with ours. We also note that the L-NOarg and no-treatment groups not only demonstrated ATP, phosphocreatine, and pHⁱdecreases that were more rapid and extensive after the onset of hypoxia, but they also had more severe histologic signs of edema and early injury. Because pHⁱis a dependent parameter that is related to phosphocreatine and ATP via the creatine kinase reaction (phosphocreatine + ADP + H^+/-[left and right arrow] Cr + ATP), it is not possible to attribute injury to a low pH (ⁱ) value alone, but rather to a combination of low levels of pHⁱ, ATP, and phosphocreatine.

In their discussion of mechanisms that might explain why NO is protective in neonatal hypoxia yet injurious in adult hypoxia, Tasker et al. [27] again raised an important that was proposed originally by Lipton et al. [28] and recently reviewed by Iadecola. [3] This was that different NO donors and different physiologic states might cause NO to be in a redox state other than nitrogen monoxide, the form usually considered to be injurious. If NO existed as nitrosonium, the slices would be protected by downward regulation of N-methyl-D-aspartate receptors. [28] However, it was not apparent that this accounted for the differences found by Tasker et al. [27]

Our study does not differentiate among the many mechanisms proposed for neuronal NO modulations of ATP concentrations, some of which work in opposite ways compared with others. For example, although neuronal NO can decrease ATP synthesis, it also can inhibit ATPase and decrease ATP use. Nitric oxide is known to inhibit cerebrocortical Na-K-ATPase [29] and an H (⁺)- ATPase required for glutamate uptake. [30] In addition, reactive oxygen species and reactive nitrogen oxide species can arise from overstimulation of glutamate receptors and calcium activation of enzymes. Furthermore, reactive oxygen species, such as superoxide, and reactive nitrogen oxide species, such as peroxynitrite, directly inhibit several different types of glutamate transporter proteins in neuron and astrocyte plasma membranes. [31] Because many processes are affected simultaneously by excess NO, we cannot always predict whether ATP should increase or decrease.

Substantial tyrosine nitrosylation is expected in peroxynitrite-mediated injury. However, we found no correlation between the decreases in ATP concentrations during hypoxia and the changes in tyrosine nitrosylation of free, soluble molecules. During hypoxia, L-NAME preserved ATP concentrations better than all other NOS inhibitors, without significantly reducing tyrosine nitrosylation. In contrast, 7-NI impressively reduced tyrosine nitrosylation during hypoxia, despite providing less preservation of ATP concentrations. Interpretive limitations are clearly imposed by our having only one HPLC measurement for each of nine experimental conditions. However, we believe these data should be presented, because the experimental variability of the NT:T ratio for similar rodent brain tissue volumes was investigated in a recent study by two of the authors, [22] and they found it to be no more than +/- 10%, which is considerably less than the large difference we saw for 7-NI compared with other inhibitors.

L-NAME, the methyl ester of L-NOarg, diffuses easily into cells and is expected to inhibit all NOS isoforms. Although our measurements of enzyme activity in washed slices (Table 1) show that L-NOarg was present in slice homogenates, we do not know whether L-NOarg substantially entered cells or whether it simply remained in slice interstitium. Because we did not measure cyclic guanosine monophosphate in our study, we do not have direct measures of NOS modulation by the inhibitors.

In addition to neuronal NOS, there might have been some inducible or immunologic NOS present from macrophages and vascular endothelial cells, because hypoxia was not studied until 3 h after decapitation. However, a careful study of the induction of immunologic NOS in more severe ischemia found that 12 h of ischemia were necessary before significant detection of the activity of that enzyme occurred. [32] Nevertheless, a small immunologic NOS contribution might have occurred at 3 h in our system. [3,33]

The relative specificity of inhibitors of different NOS isoforms (nNOS, eNOS, iNOS, for neurons, endothelial cells, and immunologic macrophages, respectively) depends on pharmacokinetics and pharmacodynamics; that is, it depends on tissue penetration and molecular affinity of the inhibitor for the isoform. The three NOS isoforms have a bidomain structure composed of an N-terminal oxygenase domain and a C-terminal reductase domain. The oxygenase domain binds heme, tetrahydrobiopterin, and arginine, and it is the site where NO synthesis occurs. [34] The reductase domain, the chemical actions of which are also modulated by calmodulin, substrates, and enzyme modulators, is necessary and sufficient for superoxide production. [6] L-NAME, 7-NI, and L-NOarg bind to all isoforms of NOS, a homodimeric protein in approximately the same location, a heme domain in the arginine binding site. [35] Consequently, the different NOS inhibitors also might have affected superoxide production by NOS in different ways. For example, a discrepancy between ATP preservation and tyrosine nitrosylation could occur if 7-NI affected NO-producing neurons differently than L-NAME by causing them to produce less superoxide and peroxynitrite. It is also possible that if we had measured total tyrosine nitrosylation, no discrepancy would have been found between its decrease and increased ATP preservation. Other experiments using enzyme inhibitors with greater selectivity might be possible in the future. A newly discovered protein inhibitor of NOS, conveniently called PIN, binds to a molecular domain that is unique to nNOS. [36,37]

In conclusion, during hypoxia in respiring brain slices, neuronal NO appears to be a major cause, but not the only cause, of rapid reductions in brain slice ATP concentrations. Only when NOS inhibition was accomplished with 7-NI, a relatively selective inhibitor of the neuronal NOS isoform, was there no measurable increase of posthypoxic tyrosine nitrosylation. More studies are needed to identify molecular mechanisms of NO-related hypoxic injury and to define the possible roles of peroxynitrite.