Single-cell analyses of nitrergic neurons in simple nervous systems

In contrast to other intercellular signal molecules, gaseous nitric oxide (nitrogen monoxide or NO) is synthesized and released without special storage and transfer mechanisms. It is one of the smallest and most diffusible signal molecules known, and it freely crosses membrane barriers to exert its effects through direct binding and/or by reaction with its target proteins (Stamler et al., 1992). Its diffusion from a point source can affect targets a relatively long distance away from its origin (Wood and Garthwaite, 1994). NO is also a highly reactive radical that can give rise to toxic secondary radical species (Stamler et al., 1992), and overproduction of NO contributes to numerous pathological conditions, such as septic shock, and to postischemic brain and heart injury (Dawson and Dawson, 1996; Moncada and Higgs, 1995). The reasons for the evolutionary choice of this versatile and dangerous molecule as a major modulator of physiological processes in animal tissues are not yet apparent, suggesting that a large part of our understanding of nitrergic physiology is yet to come.

To date, NO, its synthesis and its actions have been most extensively investigated in mammalian systems in literally scores of laboratories devoted to various aspects of biochemistry, molecular biology, physiology and behavior. Studies performed on the nervous system are a considerable portion of these efforts, as it has become obvious that NO is a prominent neuromodulatory molecule with significant relevance to neural excitability, synaptic transmission, network function arousal and learning and memory mechanisms (Garthwaite and Boulton, 1995).

Essential questions for understanding the biochemistry and physiology of NO in neurons and glia concern the actual concentrations of NO, its precursors and its breakdown products in single cell types. However, the complexity of the CNS, the equally great heterogeneity of cell types and the small sizes of the cells have for the most part greatly impeded the development and application of techniques necessary to answer such questions, despite the large amount of effort directed at mammalian studies.

In the study of the manifold roles of NO in the CNS, those studies reported from invertebrate model systems are relatively few in number. However, despite their small number, a certain depth of understanding may be emerging from the versatile nature of the preparations, their accessibility to varied experimental approaches and the high rates of data extraction in studies on small nervous systems with individually identifiable neurons.

Among the many invertebrate groups tested, reliably identified and relatively large NOS-containing neurons have only been demonstrated in annelids and molluscs: the medicinal leech Hirudo medicinalis (Leake et al., 1996; Leake and Moroz, 1996), the freshwater pond snail Lymnaea stagnalis (Moroz et al., 1994a) and the marine sea slugs Aplysia californica (Jacklet and Gruhn, 1994b; Moroz et al., 1996a) and Pleurobranchaea californica (Moroz et al., 1996a; Moroz and Gillette, 1996b). The gastropod molluscs are especially useful models for single-neuron studies of NOS regulation and NO action. The large size and robustness of their neurons make these cells easily accessible to isolation and penetration with multiple electrodes, while their identifiability has allowed the roles of specific neurons in neural networks and behavior to be characterized. The few dozen identified 100–600μm diameter nitrergic neuronal somata openly displayed on the surfaces of ganglia in Pleurobranchaea, Lymnaea and Aplysia (Fig. 1) offer a far more approachable situation than the thousands of 10–60 μm diameter diffusely distributed nitrergic neurons, surrounded by millions of other cells, in the mammalian brain. The genetic sequences of neuronal NOS isoforms of the fruitfly Drosophila melanogaster (Regulski and Tully, 1995), the hawkmoth Manduca sexta (Nighorn et al., 1998) and the pond snail Lymnaea stagnalis (Korneev et al., 1998) share substantial homology with mammalian NOS.

Histochemical location of NOS-containing neurons is generally a useful first step in studying the physiology of NO in any system. Immunohistochemistry using mammalian NOS antibodies has been helpful in some invertebrate preparations (Cooke et al., 1994; Martinez et al., 1994; Moroz et al., 1994b; Scholz et al., 1998). However, invertebrates may well have NOS isoforms that are quite different from mammalian ones, and caution must be exercised because of the absence of specific cross-reactivity of some antibodies. As a general NOS indicator, fixative-resistant NADPH-diaphorase (NADPH-d) histochemistry has proved useful for both mammals and invertebrates (Bredt et al., 1991; Dawson et al., 1991; Hope et al., 1991; Huang et al., 1997; Matsumoto et al., 1993; Muller, 1994). The appropriate fixation protocols for NADPH-d labeling vary among species, and some procedural comparisons of different conditions are required for a new preparation.

Using NADPH-d histochemistry, putative nitrergic neurons have been examined in more that 30 molluscan species (Moroz and Gillette, 1996a), including gastropods, amphineurans and cephalopods, and such information has been published in detail for Pleurobranchaea californica (Moroz and Gillette, 1996b), Aplysia californica (Jacklet and Gruhn, 1994a), Helix aspersa (Cooke et al., 1994; Huang et al., 1997) and Lymnaea stagnalis (Moroz et al., 1994a), with examples shown in Fig. 1. Almost all these species have a few NADPH-d-positive giant neurons, located mainly in the buccal and/or cerebral ganglia, and many relatively small NADPH-d-reactive cells at the periphery. Since many putative NOS-containing cells have been identified as characterized neurons of feeding networks (Jacklet, 1995; Moroz et al., 1994a; Moroz and Gillette, 1996a; Park et al., 1998), it is no surprise that NO may have significant roles in the activation of the feeding motor patterns (Moroz et al., 1993) and in chemosensory pathways (Elphick et al., 1995; Gelperin, 1994). In contrast to many other molluscs tested, the CNS of Pleurobranchaea contains several dozen large intensely-NADPH-d-positive cells distributed among all the ganglia (Moroz and Gillette, 1996a), whereas peripheral NADPH-d-reactive neurons have not been observed.

Among the many characterized NADPH-d-positive neurons, some suitable for single-cell analysis have been identified in the feeding network, including (i) five pairs of giant radular protractor motoneurons (the buccal A-group) in the buccal ganglion of Pleurobranchaea californica (Fig. 1A), (ii) a pair of giant serotonergic metacerebral (C1) cells in Pleurobranchaea californica, (iii) a symmetrical pair of esophageal motoneurons (B2) in the buccal ganglia of Lymnaea stagnalis (Fig. 1C) and (iv) the cerebral histamine-containing C2 neurons in Aplysia californica (Fig. 1B) (see Chiel et al., 1990; Jacklet, 1995). It is of interest that only in Pleurobranchaea were the serotonergic metacerebral cells labeled by NADPH-d, whereas their homologs in Lymnaea and Aplysia were not (Moroz and Gillette, 1996a,b). Curiously, the metacerebral neurons in Lymnaea are among several neurons exhibiting mRNA for a putative NOS, as shown by in situ hybridization with a Lymnaea-derived specific probe (Korneev et al., 1998). Potentially, the NOS mRNA message is either untranslated in the soma or the enzyme is inactive or only present at a low concentration.

Since expression of NADPH-d reactivity is species-dependent, one must relate the observed staining to NOS activity using other methods. One of the most important tests is a functional verification of NO-mediated transmission. It has recently been shown that electrical stimulation of NADPH-d-positive neurons in the feeding network of Aplysia (C2 cells) and Lymnaea (B2 cells) induced a slow excitatory postsynaptic potential (EPSP) in follower neurons which was suppressed by both NOS inhibitors and NO scavengers (Jacklet, 1995; Park et al., 1998). Buccal B2 motoneurons also expressed NOS-immunoreactivity (Moroz et al., 1994b) and preserved histochemical properties in cell culture (Moroz et al., 1994a) (see Fig. 1C,D). It is of interest that NO-mediated transmission has been reconstructed in vitro, as shown in recent experiments on co-cultured B2/B7nor neurons of Lymnaea (Park et al., 1998).

In spite of the notable progress in our understanding of NO-mediated signaling, nitrergic neurons themselves are still a poorly characterized population in the nervous system. Intracellular concentrations of NO-related metabolites have not been reported, and verification of the nitrergicity of many other NADPH-d-reactive neurons has been difficult. Direct, single-cell analysis of the NO/NOS-associated metabolites is highly desirable.

NO has a short half-life and is difficult to measure directly in a complex chemical microenvironment such as nervous tissue; however, the NOS coproduct L-citrulline and the breakdown products of NO oxidation can be used as indicators of NOS activity. A number of analytical methods have recently been developed for measurements of NO and related metabolites in biological tissues (Archer, 1993), but problems arise when they have to be applied at the single-cell level.

Of the various methods employed in NO analyses, only three can be used directly for single-cell analysis: (i) chemiluminescence (Archer, 1993; Kikuchi et al., 1996; Leone et al., 1996) and fluorescence detection of NO formation using fluorescent probes (Kojima et al., 1997, 1998; Nakatsubo et al., 1998), (ii) NO-sensitive electrodes (Malinski and Czuchajowski, 1996; Malinski et al., 1996; Malinski and Taha, 1992; Shibuki and Okada, 1991), and (iii) capillary electrophoresis (CE) (Beale, 1998; Camilleri, 1993; Landers, 1997).

The first two methods have been developed to detect authentic NO gas (Archer, 1993). The NO electrode was successfully applied to isolated individual cells with useful results (Malinski and Taha, 1992), and some commercially available electrodes indicate surprisingly high signal levels both at nitrergic cell bodies and in the neuropil of molluscan ganglia (Moroz et al., 1996b). However, the large sensor tip diameter of the commercially available NO electrodes (>30 μm) as well as the limited selectivity of both NO electrodes and chemiluminescence methods against various redox species (e.g. monoamines, H₂O₂) and pH restrict their applications for single-cell analysis within heterogeneous neuronal populations. Recent fluorescent indicators of NO show promise (Kojima et al., 1997, 1998; Nakatsubo et al., 1998). Useful methodologies for these applications are not yet available, and control studies are still required to gauge their effectiveness. If effective, they promise the advantage of multi-cell optical assays for NO and may offer the ability to follow its time-dependent generation.

In contrast, CE cannot be used for direct NO detection but, as a microseparation technique, it allows analysis of many essential metabolites associated and interacting with the NO–citrulline cycle. We emphasize the versatility of CE for single-neuron measurement in greater detail; it has not been reviewed previously, and CE involves significantly different methodology from the more familiar electrochemical and fluorescent approaches.

Separation of metabolites by CE is based on the differential electrophoretic migration of compounds in an electric field. CE is similar to conventional gel electrophoresis except that the separation takes place in a small-diameter (10–75 μm) fused-silica capillary, allowing the use of a higher separative voltage (up to 30 kV) because of the efficient removal of current-induced heat (Fig. 2). This provides effective and rapid separation of metabolites from pico- and nanoliter samples. Recent reviews provide more details of the capabilities and operation of CE (Beale, 1998; Camilleri, 1993; Jones, 1997; Khaledi, 1998; Landers, 1997; Righetti, 1996).

Detection using CE can be challenging owing to the small path length and small volumes. However, detection systems have been developed for most types of biologically active molecules and ions (Bryant et al., 1998; Cruz et al., 1998; Swanek et al., 1997; Tomer et al., 1998). Laser-induced fluorescence detection in CE has the best performance characteristics compared with other detection modes in terms of sensitivity, limits of detection and linearity (Lillard and Yeung, 1997). Most metabolites are not fluorescent; however, they can often be derivatized using various fluorescent or fluorogenic reagents. CE has been successfully used to assay amino acids, neurotransmitters, peptides and some enzymes in small biological samples, including individual cells such as erythrocytes (Yeung, 1994), chromaffin cells (Chang and Yeung, 1995) and single neurons (Jankowski et al., 1995; Kennedy et al., 1989). Examples of the detection of low femtomole, attomole and even zeptomole amounts of selected low-molecular-mass species and some peptides in cells, as well as descriptions of methods, have been reported elsewhere (Fuller et al., 1998; Gilman and Ewing, 1995; Jankowski et al., 1995; Kennedy and Jorgenson, 1989; Kennedy et al., 1989; Lillard and Yeung, 1997; Lillard et al., 1996; Shippy et al., 1995; Swanek et al., 1997; Tong and Yeung, 1997). Thus, the application of CE to analyses of NO metabolites is a natural evolution in studies of NO-mediated signaling at the level of the single cell.

On the basis of the mechanisms of enzymatic synthesis of NO and its degradation (see Fig. 3), three groups of assays of NO metabolism in single cells have been developed recently. These are (1) quantification of the nitrite/nitrate breakdown products of NO by indirect ultraviolet absorbance detection after CE separation; (2) quantification of L-arginine, L-citrulline and related metabolites following fluorescent derivatization, CE and fluorescence detection; and (3) quantification of intrinsically fluorescent NOS cofactors (NADPH, biopterin, FAD and FMN) by CE and ultraviolet fluorescence detection. These methods are particularly useful when combined with NADPH-d histochemistry for identifying putatively nitrergic somata for analyses. The following examples illustrate such assays in selected individual molluscan neurons. In all cases, single cells were manually dissected, placed in a nanovial or microvial, and homogenized for analysis. Multiple NOS species have been studied both to provide a confirmation of NOS activity and to estimate the levels of NOS-related metabolites.

The products of NO oxidation, NO₂⁻ and NO₃⁻, which are not appreciably membrane-permeant, are potentially accumulated in NOS-containing cells (Salter et al., 1996). Although dietary factors can also contribute to nitrite/nitrate levels (Rhodes et al., 1995), these anions can be used, with appropriate controls, as stable and reliable markers of NO synthase (NOS) activity (Meulemans and Delsenne, 1994; Meulemans et al., 1995; Salter et al., 1996; Yokoi et al., 1996). Single isolated neuron somata from the CNS of Pleurobranchaea californica were directly injected into a 75 μm diameter capillary, disrupted and, after CE separation, intracellular NO₂⁻/NO₃⁻ levels were measured by ultraviolet absorption at 214 nm (Fig. 4; see also Cruz et al., 1997). The limits of detection for NO₂⁻ and NO₃⁻ were less than 200 fmol (less than 4 μmol l⁻¹ for the neurons under study; volumes were determined optically), which was significantly lower than the actual intracellular concentrations. The NO₂⁻ and NO₃⁻ levels in individual neurons varied from 2 mmol l⁻¹ (NO₂⁻) and 12 mmol l⁻¹ (NO₃⁻) in neurons histochemically positive for NADPH-d activity (e.g. A-group buccal motoneurons) down to undetectable levels in many NADPH-d-negative cells (e.g. buccal C-group cells), confirming the previous identification of NADPH-d-reactive buccal cells in Pleurobranchaea as nitrergic neurons. NO₂⁻ was not detected in whole-ganglion homogenates or in hemolymph, while hemolymph NO₃⁻ concentration averaged 1.8×10⁻³±0.2×10⁻³ mol l⁻¹. Hemolymph NO₃⁻ concentration in Pleurobranchaea was appreciably higher than values measured for the freshwater pulmonate Lymnaea stagnalis (3.2×10⁻⁵±0.2×10⁻⁵ mol l⁻¹). The high levels of intracellular nitrite measured in these cells suggest that the nitrites might function as a source of non-enzymatic NO production. It was shown that nitrites themselves generate NO in acidified and reducing environments (Benjamin, 1994; Moroz et al., 1997; Weller et al., 1996; Zweier et al., 1995).

L-Arginine/L-citrulline conversion is a second complementary and widely used assay for NOS activity (Archer, 1993). However, both L-arginine and L-citrulline interact with other metabolic pathways (see Fig. 3) such as the urea cycle and phosphagen synthesis (Bredt and Schmidt, 1996; Hecker and Billiar, 1996), which can significantly complicate the detailed analysis of NOS activity in vivo. CE provides reliable simultaneous detection of major intermediates of both NOS-associated and other pathways not related to NO (Fig. 5A; see also Floyd et al., 1998). In this type of assay, the primary or secondary amine metabolites from single-cell samples were converted into highly fluorescent species by either pre-or on-column derivatization with fluorescamine. The products were separated by CE and measured by laser-induced fluorescence (Fig. 5A). The limits of detection for L-arginine, L-citrulline, L-argininosuccinate, L-ornithine and L-arginine-phosphate ranged from 50 amol to 17 fmol, which corresponds to 5 nmol l⁻¹ to 17 μmol l⁻¹ in the neurons under study. Thus, the detection limits for this analytical technique are significantly lower than actual intracellular amounts of the metabolites. This sensitivity allows the use of only a small (1–5 %) fraction of the individual neurons for determination of intracellular metabolite concentrations (Fig. 5B, see also Floyd et al., 1998). Levels of NOS metabolites in individual neurons varied from 6 mmol l⁻¹ (arginine) and 4 mmol l⁻¹ (citrulline) in putative NOS-containing neurons down to less than1 μmol l⁻¹ (undetectable levels) in many NADPH-d-negative cells. These results have also upheld the use of NADPH-d staining as a useful indicator of nitrergic cells in the molluscan CNS.

The millimolar concentrations of intracellular L-arginine measured in these cells has immediate significance for experiments in which competitive inhibitors of NOS are to be used: appropriate concentrations of inhibitory compounds can also be millimolar in many cases. Moreover, care must be used in the selection of these compounds (Greenberg et al., 1995), because some of these inhibitors can actually act as substrates for the enzyme-independent formation of appreciable amounts of NO in the presence of cellular reducing compounds such as NADPH, L-cysteine and L-ascorbate (Moroz et al., 1997, 1999).

While CE is quantitative, measuring cell volumes accurately can be problematic. Taking the ratio of amounts of arginine:citrulline (Arg/Cit) has the advantage that it is independent of cell volume. Low values of Arg/Cit (0.2–4) were found to be characteristic of NADPH-d-reactive cells, in accordance with expected NOS activity; in NADPH/NOS-negative cells, however, the ratio was very high (9–40), and in many cells L-citrulline was not detected (Floyd et al., 1998). The Arg/Cit ratio was strongly enhanced by the presence of NOS inhibitors, mostly because of a decrease in the level of the NO coproduct citrulline, consistent with depressed NOS activity. The intermediate of the restorative conversion of L-citrulline to L-arginine, L-argininosuccinate (Fig. 3), could also be readily identified. The substrate for citrulline production in reactions of the urea cycle (Fig. 3), L-ornithine, was not detectable in the neurons assayed, although it was found in the hemolymph. This excludes false positive conclusions of NOS activity in the molluscan CNS due to NOS-independent L-citrulline production by the urea cycle enzyme. Levels of arginine and citrulline were also determined in blood samples of Pleurobranchaea and Aplysia, and these values were orders of magnitude lower than the measured intracellular concentrations (Floyd et al., 1998).

Several cofactors, NADPH, FMN, FAD and biopterin, are required for NOS activity, and their availability may be involved in regulation of the enzyme (Griffin and Stuehr, 1995). Thus, the technique for single-cell detection of NOS cofactors is a complementary assay of the metabolism of a nitrergic neuron. Although the cellular sample matrix is chemically complex, the combination of electrophoretic migration time and fluorescence spectral information allows positive identification of aromatic monoamines, aromatic amino acids and peptides containing them, flavins, adenosine-and guanosine-nucleotide analogs, and other natively fluorescent compounds (Chang and Yeung, 1995; Lillard et al., 1996; Timperman et al., 1995; Tong and Yeung, 1997). Most of the NOS cofactors possess unique spectral properties and can be directly characterized after CE separation without additional derivatization. Using wavelength-resolved native fluorescence detection, direct quantitative and simultaneous assays of NADPH, biopterin, FAD and FMN can be performed on samples amounting to only 1 % of the total single-neuron extract (Fuller et al., 1998). This methodology was also applied to single molluscan neurons for analysis of potential cotransmitters, including serotonin, dopamine, noradrenaline, adrenaline, octopamine and neuropeptides containing aromatic amino acids, which can be performed simultaneously with measurements of the NOS cofactors. Single-cell assays of these cofactors may provide a potent tool to probe their roles in dynamic regulation of NOS activity.

Combinations of different methodologies can be brought to bear on single cells for powerful cross-correlation of physiological and biochemical parameters, both to analyze the physiology of the nitrergic neuron and to estimate possible differences among various types of NOS-containing cells. As the obvious next step in this direction, live, isolated or intact neurons can be voltage-clamped for measurement of ion currents while NO production is monitored via NO electrode/chemiluminescence techniques, a real-time measure of NOS activity. Following pharmacological or voltage manipulations and recording of the effects on current and NO production, the cell could be rapidly frozen and prepared for multiple measures of indices of NOS activity, reflected in NO breakdown products, NO precursor and coproducts and cofactor molecules.

We now know that neurons have distinct and specific complements of transmitters, with which NO might be acting as a cotransmitter or accessory modulator. Understanding the functional consequences of this complexity is likely to be best approached through the development and combination of techniques and instrumentation for multifactorial analyses in physiology and biochemistry. In this endeavor, molluscs continue to contribute their large and hardy nerve cells as usefully as they have done since the days of Hodgkin, Huxley, Young, Chalazonitis and Arvanitaki.

This research was supported in part by INTAS (93-3504) and International Research Programs (75195-540101; HHMI) grants to L.L.M., by NIH grant (NS31609) and by grant CHE 96-22663 to J.W.S.