INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

NITRIC OXIDE SYNTHASE (NOS) catalyzes the formation of nitric oxide (NO) and citrulline (Cit) from arginine (Arg) and oxygen. The cofactors include Ca²⁺ and the Ca²⁺-binding protein calmodulin (CaM), flavin mononucleotide and flavin adenine dinucleotide (FAD), and tetrahydrobiopterin (BH₄). All forms of the enzyme require reducing equivalents from NADPH, an additional cofactor. Two isoforms of the enzyme, neuronal NOS (NOS-1) and endothelial NOS (NOS-3), require Ca²⁺ for activity, whereas the inducible NOS (NOS-2) isoform does not require Ca²⁺ because it is already tightly bound to the enzyme with CaM (reviewed in Ref. 9).

We recently demonstrated that the renal vasodilation and hyperfiltration of pregnancy are mediated by NO via endothelin (ET) and the endothelial ET_B receptor in conscious rats (6, 8). Using these physiological observations to guide our investigations further at the cellular and molecular level, one of our objectives was to determine NOS activity in renal cortical homogenates prepared from nonpregnant and pregnant rats. Because over 90% of renal blood flow is distributed to the cortex (15), we focused specifically on this zone of the kidney.

Using a conventional "NOS activity" assay with L-[2,3-³H]Arg as the labeled substrate (2, 5, 7), we found that the major product(s) was not Cit. Initial clues that the major product(s) formed by renal cortical homogenates was not Cit and the enzyme activity was not NOS included complete independence of Ca²⁺, on the one hand, and of NADPH, on the other. The major product(s) looked like Cit, insofar as it passed freely through columns containing cation-exchange resin and comigrated with Cit on one- and two-dimensional straight-phase thin-layer chromatography systems. However, the compound(s) was clearly resolved from Cit by precolumn derivatization and reverse-phase HPLC.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Tissue collection. Renal cortex was isolated by blunt dissection from Long-Evans female rats (Harlan Sprague Dawley, Indianapolis, IN or Frederick, MD) and NFR/N mice (National Institutes of Health). Renal cortex was also obtained from human male kidneys provided by the National Disease Research Interchange (Philadelphia, PA). The human kidney tissue was snap-frozen in liquid nitrogen within 6 h of accidental death. Villous tissue was dissected from human placentas that were acquired from normal spontaneous vaginal deliveries or cesarean section, as previously described, and served as a positive control for NOS activity (5, 7).

Tissue homogenates. We previously reported in detail the preparation of tissue homogenates for the NOS activity assay (5, 7). Briefly, tissue was homogenized in a solution containing 50 mM sucrose, 25 mM HEPES, 1.0 mM dithiothreitol, and several protease inhibitors (10 µg/ml each of leupeptin, pepstatin A, chymostatin, antipain, and soybean trypsin inhibitor, as well as 100 µg/ml phenylmethylsulfonyl fluoride). The homogenate was then centrifuged at 10,000 g for 15 min at 4°C. The supernatant was next passed through a desalting column (10DG columns, BioRad, Richmond, CA) with a molecular exclusion size of <6 kDa in order to remove all substrates and cofactors except possibly CaM, which has a molecular mass of ~17 kDa. Finally, particulate and cytosolic fractions were prepared by a second centrifugation at 100,000 g for 60-90 min at 4°C (12).

NOS activity assay. Again, this procedure has been previously described (2, 5, 7). In brief, 100 µl of the homogenate was added to 100 µl of incubation buffer. Except when otherwise indicated, the following final concentrations of cofactors were used: L-Arg, 20 µM; CaM, 10 U/ml; FAD, 10 µM; NADPH, 0.5 mM; BH₄, 2 µM; and L-[2,3-³H]Arg, 2.5 µCi/ml or L-[guanido-¹⁴C]Arg, 2.5 µCi/ml (NEN, Boston, MA). When enzyme activity was tested in the presence of Ca²⁺, the final concentration of the cation was 10 µM as verified by Ca²⁺-sensitive electrode. To assess Ca²⁺-independent NOS activity, Ca²⁺ was omitted and EGTA was added to the incubation buffer (final reaction concentration 0.5 mM). Incubations were conducted at 25°C in a shaking water bath usually for 40 min. The enzyme reaction was stopped by adding 1.8 ml of ice-cold buffer containing 80 mM HEPES, 8 mM EDTA, pH 5.2, except for the samples prepared for HPLC, which were stopped by 1.8 ml of ice-cold distilled water. (In pilot experiments, we found that the binding of Arg and the passage of Cit through the cation-exchange columns were not affected by these different stop solutions.) Although nonenzymatic conversion was negligible, it was determined routinely in each experiment using homogenate that had been heated to 60°C for 30 min. The radioactive counts were then subtracted from those generated by the 25°C incubations. This step also accounted for the small, but not trivial, amounts of L-[2,3-³H]Arg or L-[guanido-¹⁴C]Arg that eluted from the cation-exchange column. The radiolabeled Cit (and other radiolabeled products; see below) produced in the enzyme reaction was separated from radiolabeled Arg by Dowex AG50W-X8 cation-exchange columns (Na⁺ form).

Thin-layer chromatography. Both one-dimensional (1-D) and two-dimensional (2-D) straight-phase thin-layer chromatography systems were used as previously reported (7). In both cases, SG-60 plates were utilized (Analtech, Newark, DE). For the 1-D system, a basic mobile phase of chloroform-methanol-ammonium hydroxide-water (5:45:20:10 vol/vol/vol/vol) was employed with clear separation of Arg, ornithine, and Cit R_f values of 0.48, 0.60, and 0.75, respectively). For the 2-D system, the basic mobile phase was used for both directions with excellent separation of Cit from Arg and ornithine, which were themselves poorly separated. Alternatively, excellent separation for Cit was also obtained using an acidic mobile phase (chloroform-methanol-acetic acid-water, 5:45:20:10 vol/vol/vol/vol) in the second direction while achieving some separation of Arg and ornithine.

Reverse-phase HPLC. The samples eluted from the cation-exchange columns were dried and reconstituted in 60 µl of 100 µM L-Cit. Samples were then mixed with an equal volume of o-phthaldialdehyde reagent solution (OPA; Sigma, St. Louis, MO) for a final volume of 120 µl. Cit was separated from the other radiolabeled products in the sample using reverse-phase HPLC similar to that described by Carlberg (4). Briefly, 100 µl of each sample was injected onto a C₁₈ Partisil 10 ODS-3 analytic column (4.6 × 250 mm) that had been equilibrated before injection for 1 h with mobile phase consisting of 11.5% methanol, 5% acetonitrile, and 1% tetrahydrofuran in 10 mM KH₂PO₄ buffer, pH 5.9. The flow rate for the pump was 1.5 ml/min, and the absorbance of the sample was monitored at a wavelength of 340 nm. After injection, the eluate was collected in 30-s fractions for 20 min. Each fraction was mixed with scintillation cocktail and counted for 10 min.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Table 1 depicts the V_max and K_m for enzyme activity in renal cortical homogenates from rats. (Because the relationship correlation between 1/V and 1/[substrate] was >0.99 for each homogenate, we presume the rather large SEs are due to biological variation among rats.) These enzyme characteristics were not significantly affected by deleting Ca²⁺ and adding 0.5 mM EGTA (or even a higher EGTA concentration of 5.5 mM). The lack of Ca²⁺ dependency by enzyme activity was also observed using renal cortical homogenates from cesarean section-derived, barrier-raised Long-Evans female rats (n = 2) and from a pool of 10 NFR/N female mice (data not shown).

To verify that the NOS enzyme assay was reliable, we used human term villous placental homogenates as a positive control for the endothelial form of NOS (5, 7). Table 2 shows that, as expected, by deleting Ca²⁺ and adding 0.5 mM EGTA, virtually all of the activity was inhibited. Yet, when tested side-by-side in the same assay under identical conditions, enzyme activity in renal cortical homogenates from rats was again virtually unaffected by deleting Ca²⁺ and adding EGTA (Table 2).

Additional enzyme characteristics for human term villous placenta and rat renal cortex are also presented in Table 2. By deleting NADPH and BH₄, the NOS activity in the placenta was abolished. In contrast, enzyme activity in rat renal cortex was not diminished. To be certain that all of the endogenous NADPH was removed, the kidney homogenates were subjected to two passages through the desalting columns or extensive dialysis and then reaction with nitroblue tetrazolium to consume any residual NADPH. The robust enzyme activity was the same irrespective of nitroblue tetrazolium treatment. In contrast, NOS activity of the human placental homogenates tested in the presence of NADPH was greatly diminished when subjected to the same nitroblue tetrazolium treatment (data not shown). To determine the CaM dependency of enzyme activity, trifluperazine or R-24571 was used. Although these CaM antagonists clearly reduced activity in the term human villous placenta (7), they did not affect the activity in the renal cortex conducted in the presence of 10 µM Ca²⁺ (n = 3 rats): control, 3.1 ± 0.1; trifluperazine, 3.2 ± 0.1; and R-24571, 3.3 ± 0.1 nmol · mg protein¹ · 40 min¹.

In the presence of Ca²⁺, the cytosolic and particulate activities in renal cortical homogenates from three rats were 498 ± 73 and 54 ± 20 pmol · mg protein¹ · 40 min¹, respectively. When factored for tissue wet weight, 96.5 ± 1.0% of the activity resided in the cytosolic fraction. In the absence of Ca²⁺, the cytosolic and particulate activities were 519 ± 66 and 81 ± 20 pmol · mg protein¹ · 40 min¹, respectively. When factored for tissue wet weight, 94.0 ± 2.1% of the activity resided in the cytosolic fraction.

Although NOS activity in the human term villous placenta was inhibited 90.3 ± 3.6% by L-NAME (n = 3 placentas), enzyme activity was only modestly reduced in the rat renal cortex by this Arg analog (Table 3). In experiments using an additional three rats, enzyme activity was inhibited by 13-47% using L-NAME-to-Arg concentration ratios of 20-50 to 1. The degree of inhibition did not vary consistently with the presence or absence of Ca²⁺. Table 4 depicts the data for human male kidney cortex. The enzyme activity was again mostly independent of NADPH and BH₄ and was little affected by L-NAME. Furthermore, the enzyme activity was lower than in the rat kidney cortex, which may reflect the longer interval before freezing (see METHODS) or inherently lower activity in human kidneys.

To test whether the product(s) was Cit, we performed both 1-D and 2-D straight-phase thin-layer chromatography. The enzyme assay was conducted in the presence of all cofactors, and the reaction was first passed through a cation-exchange column before thin-layer chromatography. Using 1-D thin-layer chromatography, 95.5 and 96.3% of total counts, respectively, comigrated with unlabeled Cit for the human placenta and rat renal cortex. In the 2-D system using the basic mobile phase in both directions, 93.1% of applied counts comigrated with Cit for the human placenta. When the assay was carried out both with and without Ca²⁺ for the rat renal cortex, 81.4 and 86.0% of the total counts comigrated with Cit, respectively. In the 2-D system using the basic mobile phase in the first direction and the acidic phase in the second direction, 89.1 and 81.4% of the counts comigrated with Cit.

Further testing was conducted using precolumn derivatization and reverse-phase HPLC (Fig. 1). In Fig. 1A, most of the radioactive product generated from the placental homogenates eluted with unlabeled Cit with a peak at fraction 15. A minor peak detected at fraction 7 was independent of heat inactivation. By contrast, for the renal cortical homogenates from rats, most of the radioactivity eluted much earlier, with a peak at fractions 6-8, regardless of the presence (Fig. 1B) or absence (Fig. 1C) of Ca²⁺, NADPH, and BH₄. In the presence of Ca²⁺ and other cofactors, an additional small amount of radioactivity eluted with unlabeled Cit. Interestingly, this peak persisted, albeit to a smaller degree, in the absence of Ca²⁺ and cofactors. Comparable results were obtained using another reverse-phase HPLC system and derivatization with phenyl isothiocyanate as described by Buzzigoli and co-workers (Ref. 3; data not shown).

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Analysis of 3H-labeled products by reverse-phase HPLC in tissue homogenates from human term villous placenta (A) and from female rat renal cortex both with (B) and without Ca2+, NADPH, and tetrahydrobiopterin (BH4; C). OPA, o-phthaldialdehyde; see METHODS for details. Cit, citrulline; cpm, counts/min.

The radioactive product(s) generated by renal cortical homogenates was virtually unretained by the reverse-phase HPLC column irrespective of derivatization with OPA (Fig. 2A). (Similar results were observed when phenyl isothiocyanate was used as a derivatization reagent; data not shown.) As expected, the mobility of [³H]Cit generated by NOS activity in placental homogenates was shifted by derivatization (Fig. 2B). We observed that background counts associated with the heat-inactivated tissue homogenates inexorably increased over several months. As noted in METHODS, the counts associated with the heat-inactivated homogenate were subtracted from the enzyme activity measured at 25°C in each assay, so they were accounted for. Nevertheless, we were successful in reducing this background by incubating an aliquot of the stock L-[2,3-³H]Arg with a small amount of anion-exchange resin for 10 min at 25°C immediately before addition to the incubation cocktail. This spontaneously generated breakdown product(s) was also virtually unretained by the reverse-phase HPLC column and eluted with a peak at fraction 7 (Fig. 2C).

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Analysis of 3H-labeled products by reverse-phase HPLC in tissue homogenates from female rat renal cortex (A) and human term villous placenta (B). C: degradation 3H-labeled products formed spontaneously in the stock L-[2,3-3H]Arg (+ OPA). Because half of the reaction in both A and B was derivatized with OPA, and the other half was not, the cpm are approximately half of those portrayed in Fig. 1. The amount of labeled L-[2,3-3H]Arg used in C was the same as that used in Fig. 1 and was otherwise prepared the same as for assay of enzyme activity except 1 aliquot was not pretreated with anion-exchange resin.

To begin investigating the identity of the compound(s) in question, we tested several Arg metabolites by comparing their R_f values vs. Cit on 1-D straight-phase thin-layer chromatographs. Because the unidentified product(s) comigrates with Cit in this chromatographic system (see above), we reasoned that candidate molecules not comigrating with Cit could be excluded. Using this approach, we were able to rule out Arg (radiolabeled precursor); hydroxyarginine (NOS intermediate or P-450 product); ornithine (arginase); argininosuccinate (urea cycle); polyamines including spermine, spermidine, and putrescine (ornithine decarboxylase); proline (via ornithine transaminase and glutamic acid); and phosphoarginine (data not shown). Using 10 mM L-valine, we found a reduction in the activity of renal cortical homogenates from 170 to 32 pmol · mg protein¹ · 40 min¹, while activity in placental villous homogenate was mostly unaffected, 124 to 116 pmol · mg protein¹ · 40 min¹ (Cit).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A standard approach to assessing NOS activity in tissue homogenates is the use of a radioactive isotope of Arg in the assay and then quantification of labeled Cit after separation of Arg and Cit by cation-exchange column chromatography (2, 5, 7). By using this approach, we previously identified NOS activity in the human term villous placenta that was virtually all Ca²⁺ and calmodulin dependent, inhibited by L-NAME, and primarily located in the particulate fraction (5, 7). If fact, we used placental homogenate as a positive control for the current investigation of enzyme activity in the renal cortex.

Our goal was to quantify NOS activity in renal cortical homogenates from nonpregnant and midterm pregnant rats, because our physiological studies demonstrated that gestational renal vasodilation and hyperfiltration were mediated by NO (8). Unfortunately, our goal was not achieved because we found that the standard NOS activity assay when applied to renal cortical homogenates did not reflect NOS activity or yield Cit as the major product.

When we measured V_max and K_m of the putative NOS activity in renal cortical homogenates of nonpregnant female rats, we found that V_max was the same regardless of the presence or absence of Ca²⁺. By contrast, deletion of Ca²⁺ from the enzyme reaction for term placental villous homogenates virtually eliminated all activity as previously reported (5, 7). In a similar vein, whereas CaM antagonists greatly reduced NOS activity in the term human villous placenta (7), no such reduction was observed for the putative NOS activity in rat renal cortex.

We next considered the possibility of a Ca²⁺-independent isoform of NOS, perhaps one of the inducible NOS isoforms that had been identified in the kidney at least at the level of mRNA by RT-PCR (11). To eliminate the possibility of low-grade, chronic renal interstitial nephritis contributing to the induction of NOS, we tested the enzyme activity in cesarean section-derived, barrier-raised female rats. However, comparable activity was again observed regardless of the presence or absence of Ca²⁺. The enzyme activity observed in renal cortical homogenates prepared from female mice and male humans was also Ca²⁺ independent. Finally, the K_m was higher than most values published for the various NOS isoforms (9).

Another attribute of NOS isoforms is that they are inhibited by various analogs of L-Arg (9). In agreement with our previous work (5, 7), the NOS activity in homogenates from term human villous placenta was reduced, on average, by 90% using L-NAME. By contrast, the inhibition of the putative NOS activity in tissue homogenates prepared from rat or human renal cortex was at best partial, ranging from 8 to 47%.

An absolute requirement for NOS activity is the contribution of reducing equivalents from NADPH (9, 11). When NADPH (and BH₄) was deleted from the enzyme reaction for term human villous placenta, virtually all of the NOS activity was abolished. However, omission of these cofactors hardly affected the enzyme activity of either rat or human renal cortex.

Both the Ca²⁺- and NADPH-independent characteristics of the putative NOS activity in the renal cortex prompted us to further investigate whether the major product formed was indeed Cit. When the radioactive product(s) eluting from the cation-exchange column was applied to either 1-D or 2-D straight phase thin-layer chromatography, the product consistently comigrated with unlabeled Cit, which was well-separated from both Arg and ornithine in these chromatographic systems. However, because the cofactor requirements did not fit with NOS, we further performed reverse-phase HPLC. Using two different precolumn derivatization procedures and reverse-phase HPLC systems, the major radioactive product(s) formed by the rat kidney cortex and eluted from the cation-exchange columns was clearly not Cit.

Although we tried to identify the product(s) in question using gas chromatography-mass spectrometry, we were unsuccessful. Thus we do not currently know its identity, although the L-[2,3-³H]Arg may spontaneously decay into the compound(s), which is partly adsorbed by anion-exchange resin. Because the retention (or lack thereof) of the product was mostly unaffected on reverse-phase HPLC by derivatization either with phenyl isothiocyanate or o-phthaldialdehyde, it is probably not a primary or secondary amine. Moreover, the product did not comigrate with many of the common Arg metabolites on 1-D straight-phase thin-layer chromatography. In collaboration with Drs. J. Roberts and S. Sladek of the Magee-Womens Research Institute, Univ. of Pittsburgh, and R. Hoffman of the Dept. of Surgery, Univ. of Pittsburgh, the product was also identified in rat uterus and mouse jejunum, respectively, using L-[¹⁴C]Arg substrate labeled at all of the carbon atoms (personal communications, unpublished data).

Using L-[guanido-¹⁴C]Arg and rat renal cortical homogenates, we again found comparable activity in the presence and absence of NADPH/BH₄, i.e., 784 vs. 826 pmol · mg protein¹ · 60 min¹, respectively. Inexplicably, these values were twice those observed using the same homogenate run concurrently with L-[2,3-³H]Arg, i.e., 484 vs. 366 pmol · mg protein¹ · 60 min¹. In the former, however, we were uncertain of the ability of the cation-exchange columns to completely retain any urea that might be formed.

To recapitulate, the enzyme activity that we have identified in the renal cortex is unlikely to be NOS for three main reasons: 1) the activity is independent of NADPH, 2) it is poorly inhibited by Arg analogs, and 3) the major product is not Cit. Interestingly, both R. Hoffman (unpublished observation) and we (see RESULTS) observed reduction of product(s) formation by L-valine, suggesting involvement of the arginase pathway. It is conceivable, however, that L-valine may not be a specific arginase inhibitor. Moreover, arginase is supposed to be heat stable (10), and in our hands, activity was completely lost with incubation at 55-60°C for 30 min. The kidney is rich in L-amino acid oxidase (1) and diamine oxidase (13), which may contribute to the formation of the unidentified Arg metabolite(s) observed herein. Finally, we cannot conclusively rule out that the unidentified compound(s) arises from a labeled contaminant within the Arg preparation rather than from the Arg itself, although this possibility seems unlikely for several reasons. First, cold Arg readily competes for the formation of radiolabeled product. Thus we were able to derive Michaelis-Menten kinetics (see RESULTS). The cold Arg was >99% pure by thin-layer chromatography as reported by the manufacturer (Sigma). Second, we consistently observed the unidentified compound(s) using many different lots of L-[2,3-³H]Arg as well as with [¹⁴C]Arg. Third, we often ran our labeled Arg on 1-D straight-phase thin-layer chromatographs and verified the high purity claimed by the manufacturer.

A recent report by Singh et al. (14) suggested the presence of a new NOS isoform in the whole rat kidney that was Ca²⁺, CaM, and NADPH independent, as well as relatively resistant to inhibition by L-NAME and S-ethyl-isothiourea and mainly present in the cytosolic fraction. These enzyme characteristics mirror our own. However, the work of Singh et al. (14) and our own do not entirely agree, insofar as they (14) demonstrated the major product to be Cit by ion pair reverse-phase HPLC, whereas we showed that it was not Cit, also by reverse-phase HPLC, but using precolumn derivatization procedures. We considered two potential explanations for this discrepancy. The first and most likely was that our procedures of precolumn derivatization and reverse-phase HPLC permitted resolution of Cit from the unknown product(s), while the HPLC method of Singh and colleagues (14) did not. In support of this possibility, we observed that both unlabeled Cit standard and labeled Cit generated by placental homogenates, when they were not derivatized, eluted in the same fractions as the unknown compound(s) produced by renal cortical homogenates. Second, we also observed generation of Cit by renal cortical homogenates that was inhibited by heat, but it was a minor product. Interestingly, this Cit formation was not completely inhibited by deleting NADPH/BH₄, similar to the work of Singh et al. (14). However, perhaps significant counts associated with the unknown compound(s) were eluted in the void volume and discounted by Singh and colleagues (an HPLC profile of radioactive products was not presented). Nevertheless, it is difficult to understand how the enzyme responsible for the residual Cit formation could have been a NOS isoform as Singh and co-workers (14) proposed, since the activity was independent of NADPH.

Perspectives