The Most Abundant Peptide Neurotransmitter in the Mammalian Central Nervous System


  • Abbreviations used : AMPA, α-amino-3-hydroxy-5-methylisoxazole-4-propionate ; cAMP, cyclic AMP ; DCG-IV, (2S, 2′R, 3′R)-2-(2′, 3′ -dicarboxycyclopropyl)glycine ; LTP, long-term potentiation ; mGluR, metabotropic glutamate receptor ; NAAG, N-acetylaspartylglutamate ; PMPA, (phosphonomethyl)pentanedioic acid.

Address correspondence and reprint requests to Dr. J. H. Neale at Department of Biology, Georgetown University, Washington, DC 20057-1229, U.S.A. E-mail : nealej@gunet.georgetown.edu


In the progress of science, as in life, timing is important. The acidic dipeptide, N-acetylaspartylglutamate (NAAG), was discovered in the mammalian nervous system in 1965, but initially was not considered to be a neurotransmitter candidate. In the mid-1980s, a few laboratories revisited the question of NAAG's role in the nervous system and pursued hypotheses regarding its function that ranged from a precursor for the transmitter pool of glutamate to a direct role as a peptide transmitter. Since that time, NAAG has been tested against nearly all of the established criteria for identification of a neurotransmitter. It successfully meets each of these tests, including a concentrated presence in neurons and synaptic vesicles, release from axon endings in a calcium-dependent manner following initiation of action potentials, and extracellular hydrolysis by membrane-bound peptidase activity. NAAG is the most prevalent and widely distributed neuropeptide in the mammalian nervous system. NAAG activates NMDA receptors with a low potency that may vary among receptor subtypes, and it is a highly selective agonist at the type 3 metabotropic glutamate receptor (mGluR3). Acting through this receptor, NAAG reduces cyclic AMP levels, decreases voltage-dependent calcium conductance, suppresses excitotoxicity, influences long-term potentiation and depression, regulates GABAA receptor subunit expression, and inhibits synaptic release of GABA from cortical neurons. Cloning of peptidase activities against NAAG provides opportunities to study the cellular and molecular mechanisms by which synaptic NAAG peptidase activity is controlled. Given the codistribution of this peptide with a spectrum of traditional transmitters and its ability to activate mGluR3, we speculate that one role for NAAG following synaptic release is the activation of metabotropic autoreceptors that inhibit subsequent transmitter release. A second role is the production of extracellular glutamate following NAAG hydrolysis.


In 1965, the neurochemistry community was welcoming GABA to the short list of putative neurotransmitters. With the exception of substance P, the compounds on this list were small amines. In the midst of the search for additional transmitter candidates, Curatelo et al. (1965) reported the presence of high micromolar to low millimolar concentrations of the acidic dipeptide, N-acetyl-aspartylglutamate (NAAG ; Fig. 1), in the brain and spinal cord. Confirmed a year later by Miyamoto et al. (1966), this discovery seems to have gone unexplored for several reasons. NAAG is present in concentrations far exceeding those of other established and putative transmitters, except glutamate. Peptides were not recognized as a significant category of neurotransmitters. NAAG appeared late in evolution with the highest concentration being found in the nervous system of mammals and very little, if any, being present outside of vertebrate species (Miyake et al., 1981). Initially, NAAG did not appear to alter membrane potential directly. Thus, immediately following its discovery, NAAG was thought to have a metabolic function, such as an intermediate in the biosynthesis of the transmitter pool of glutamate. This latter hypothesis remains untested today, although the extracellular hydrolysis of NAAG following synaptic release clearly results in the release of glutamate into the synaptic space.

Figure 1.

The acidic dipeptide transmitter, NAAG, is hydrolyzed in the extracellular space by one or more peptidase activities. This inactivation of the peptide results in the production of glutamate and N-acetylaspartate (NAA). Production of glutamate may lead to a second-order effect following the synaptic release of NAAG. The extent of this effect will be influenced by the local balance of peptide availability, hydrolysis rates, and glutamate transport/reuptake rates. NAA, one of the most prevalent amino acids in the nervous system, is taken up by glia. The mechanism of NAAG biosynthesis does not appear to involve posttranslational protein hydrolysis, but no synthetic enzyme has been identified to date.

FIG. 1.

Very little research on NAAG was published for almost two decades following its discovery. Not even the extraordinary expansion of interest in peptide transmitters that followed the discovery of the opioid peptides in the early 1970s was sufficient to return attention to NAAG. This intellectual resistance was first overcome by Joseph Coyle's laboratory. This group published a series of articles that provided additional data on the presence of the peptide in the nervous system and its interaction with a putative excitatory receptor (Zaczek et al., 1983 ; Koller and Coyle, 1984 ; Koller et al., 1984 ; Bernstein et al., 1985 ; ffrench-Mullen et al., 1985). Subsequent reports from the Coyle laboratory and others have brought into question some of the experimental designs and conclusions found in these initial articles (Riveros and Orrego, 1984 ; Blakely et al., 1988a ; Whittemore and Koerner, 1989). Nonetheless, this pioneering effort broke the inertial barrier, and several other laboratories entered the field to test the hypothesis that the peptide serves as a neurotransmitter. More than 30 years after its discovery, substantial data now support the hypothesis that NAAG is the most prevalent and widely distributed peptide neurotransmitter in the mammalian nervous system.

A structurally related molecule, N-acetylsuccinimidyl-glutamate, a cyclic form of NAAG, has been detected in the rat spinal cord and brainstem at concentrations three orders of magnitude lower than NAAG (Brovia et al., 1996). The biological relevance of this presumptive NAAG derivative is unknown.


Important to rigorous analysis of the cellular localization and distribution of NAAG in the nervous system was the development of highly specific antibodies. The presence of millimolar concentrations of glutamate and N-acetylaspartate in brain and spinal cord impose stringent specificity requirements on such antibodies. In 1985, Charles Cangro, a doctoral student at Georgetown, developed a method for the production of highly specific rabbit anti-NAAG antibodies and a parallel method that used a water-soluble carbodiimide for covalently coupling NAAG within cells during tissue fixation (Cangro et al., 1987). These antibodies and the fixation method were used to establish that the peptide was concentrated in nerve cells throughout the brain (Anderson et al., 1986). Over the next decade, these antibodies were used to define the neuronal distribution of NAAG in the rat (Cangro et al., 1987 ; Moffett et al., 1994 ; Moffett and Namboodiri, 1995 ; Renno et al., 1997), amphibian (Kowalski et al., 1987), cat (Tieman et al., 1987), primate (Tieman et al., 1991), and human (Tieman et al., 1988) nervous systems. In parallel, other preparations of polyclonal and monclonal antibodies against NAAG (Blakely et al., 1987 ; Forloni et al., 1987 ; Frondoza et al., 1990 ; Tsai et al., 1993) have been used by the Coyle laboratory to study the cellular distribution of NAAG in the rat brain and in studies of human brain pathology (Passani et al., 1997a).

These immunohistochemical studies demonstrated the extraordinarily broad distribution of the peptide throughout the mammalian brain, spinal cord, and sensory neurons. The peptide is present in a number of important projection pathways, such as ascending and descending spinal axons, spinal motoneurons, retinal ganglion cells, geniculo-cortical neurons, the nigrostriatal pathway, some cerebellar afferent neurons, neurons of the deep cerebellar nuclei, and large spinal sensory neurons (Cangro et al., 1987 ; Forloni et al., 1987 ; Tieman et al., 1988, 1991 ; Tsai et al., 1993 ; Moffett et al., 1994 ; Moffett and Namboodiri, 1995 ; Renno et al., 1997). The peptide also is concentrated in interneurons, including those in the cerebellar and cerebral cortices and hippocampus, and a spectrum of other GABAergic interneurons (Moffett et al., 1994 ; Moffett and Namboodiri, 1995). NAAG has been identified in neurons that use a variety of amine transmitters, including glutamate, GABA, serotonin, norepinephrine, dopamine, and acetylcholine.

At the ultrastructural level, the peptide is concentrated in synaptic vesicles (Williamson and Neale, 1988a ; Renno et al., 1997), whereas its compartment in neuronal cell bodies has not been resolved. By immunohistochemistry, NAAG appears to have a strictly neuronal localization. This may be due, however, to assay conditions that are optimized to identify the highest concentration of peptide. In fact, glial cells in culture contain low to moderate micromolar concentrations of the peptide when assayed by highly specific radioimmunoassay or by absorbance following resolution on HPLC. The appearance of 100-200 μM levels of NAAG in glia cultured from embryonic mouse cortex versus low micromolar levels in astrocytes cultured from neonatal rat cerebellum suggests that there are substantial quantitative differences in NAAG concentration among different types of glial cells (Cassidy and Neale, 1993a ; Passani et al., 1998 ; Wroblewska et al., 1998).

NAAG biosynthesis has not been well characterized. Cangro et al. (1987) demonstrated that the synthesis of NAAG by dorsal sensory ganglia in vitro was not dependent on protein synthesis, data that support the hypothesis that synthesis is mediated by an enzyme rather than by posttranslational processing. However, enzymemediated synthesis of NAAG from glutamate and N-acetylaspartate has not been demonstrated in a cell-free system.


The depolarization-induced, calcium-dependent release of endogenous NAAG was first demonstrated in slices and synaptosomes prepared from several regions of the rat brain, including cerebellum and hippocampus (Pittaluga et al., 1988 ; Zollinger et al., 1988, 1994). The cerebellar release was reduced by lesion of the climbing fiber pathway. In contrast, Sekiguchi et al. (1989a) were unable to demonstrate an increase above basal release of endogenous NAAG from cerebellar slices following depolarization.

Early immunohistochemical studies localized NAAG in amacrine cells of the amphibian retina (Kowalski et al., 1987) and ganglion cells of the mammalian retina (Anderson et al., 1987 ; Tieman et al., 1987, 1988), leading to the demonstration of NAAG release in several different visual system preparations. NAAG release was observed following depolarization in amphibian and avian retinas and in the rat superior colliculus (Tsai et al., 1988 ; Williamson and Neale, 1988b, 1992). NAAG was found to be concentrated in retinal ganglion cell bodies, axons, and terminal fields in avian (Williamson et al., 1991) and mammalian nervous systems (Anderson et al., 1987 ; Tieman et al., 1987, 1988). Following up on these reports, NAAG release was demonstrated from the optic tectum of chicks (Williamson et al., 1991) and superior colliculus of rats (Tsai et al., 1990) following stimulation of the optic nerve.


Initial equilibrium binding studies using NAAG radiolabeled on the glutamate moiety and binding displacement studies using unlabeled NAAG suggested the presence of a novel peptide receptor in the rat brain membranes (Zaczek et al., 1983 ; Koller and Coyle, 1984, 1985 ; Joëls et al., 1987 ; Schoepp and Johnson, 1989). However, an early attempt to use equilibrium binding of the peptide to rat brain slices (Riveros and Orrego, 1984) resulted in the discovery of membrane-bound peptidase activity that rapidly hydrolyzed NAAG to glutamate and N-acetylaspartate. This result led to a reconsideration (Blakely et al., 1988a) of the validity of those initial receptor binding studies that had failed to control for the consequences of peptide hydrolysis and the subsequent influence of radiolabeled or unlabeled glutamate in the assay systems. Receptor binding studies executed under conditions of peptidase inhibition demonstrated that NAAG had very little efficacy in displacing radioligand binding at the α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) and kainate subtypes of glutamate receptors, whereas NAAG displaced binding of CGS-19755 at NMDA receptor binding sites (IC50 values : glutamate, 0.4 μM ; NAAG, 8.8 μM) in rat forebrain membranes (Valivullah et al., 1994).

These receptor binding data are consistent with physiological studies in which NAAG was found to act as a low-potency agonist at NMDA receptors, but not kainate or AMPA receptors, in cultured spinal cord neurons [response threshold : glutamate, 0.5 μM ; NMDA, 1 μM ; NAAG, 300 μM (Westbrook et al., 1986)] and olfactory bulb neurons [EC50 values : NAAG, 666 μM ; NMDA, 29 μM (Trombley and Westbrook, 1990)]. Oocytes injected with NMDAR1 receptor subunit cDNA formed homomeric receptors that were activated by the peptide, whereas cells injected with AMPA and kainate receptor cDNAs were not (Sekiguchi et al., 1992). NAAG also activated NMDA-like receptors on guinea pig and chick cerebellar neurons (Mori-Okamoto et al., 1987 ; Sekiguchi et al., 1987) and elicited increases in intracellular calcium in neurons cultured from rat forebrain via NMDA-like receptors (Koenig et al., 1994).

Three questions derive from these observations. First, does the peptide reach sufficiently high synaptic concentrations to activate one or more subtypes of NMDA receptors ? Given the high micromolar to millimolar tissue concentration of NAAG in the rat brain and spinal cord (Koller et al., 1984 ; Guarda et al., 1988 ; Fuhrman et al., 1994) and its concentration in selected populations of neurons, this possibility cannot be excluded. The second question relates to neurons expressing NMDA receptors that differ in their subunit composition and pharmacological profiles. Does NAAG have differential potency at NMDA receptor subtypes ? In the cat dorsal lateral geniculate, for example, NAAG fails to mimic the excitatory response to NMDA (Jones and Sillito, 1992), yet it does so in spinal and olfactory neurons. Although this question remains to be more fully explored, preliminary data from Stefano Vicini's laboratory indicate that NAAG is much less potent than glutamate at NMDA receptors in cells transfected with cDNA encoding for the NMDAR1 and NMDA2B subunits, whereas the peptide has a substantially higher potency at the NMDA receptor in cells transfected with NMDAR1 and NMDA2D cDNAs (S. Vicini, personal communication). Similarly, Hess et al. (1999) found that NAAG potentiated the effect of glutamate on oocytes injected with NMDAR1/2D, but not NMDAR1/2A and NMDAR1/2B. Further, NAAG had an EC50 of 35 μM at NMDAR1/2D and 106 μM at NMDA1/2B. The third question derives from the observation that the NMDA receptor is inactivated as a consequence of agonist occupancy. Does NAAG inactivate one or more subtypes of NMDA receptors by receptor occupancy at concentrations that are below the threshold for receptor activation ? Similarly, one might speculate that if NAAG is less effective in activating the NMDA conductance than it is in occupying the ligand binding site, it also may function as a partial agonist when competing with glutamate for the receptor. It is possible that such a desensitizing or partial agonist effect of NAAG on some NMDA receptor subtypes may be responsible for reports of an inhibitory effect of the peptide on NMDA-induced events (Sekiguchi et al., 1989b ; Puttfarcken et al., 1993 ; Burlina et al., 1994), although its action at the type 3 metabotropic glutamate receptor (mGluR3) in these preparations also must be considered. The hypothesis that NAAG may act as a mixed agonist/antagonist at specific NMDA receptor subtypes is a significant and unresolved pharmacological and physiological issue.


The mGluRs are classified into three groups based on pharmacological and sequence similarities (for review, see Schoepp et al., 1999). The group II receptors consist of mGluR2 and mGluR3. These receptors share ~80% sequence homology, are negatively coupled to cyclic AMP (cAMP) levels, and are activated by similar agonists and antagonists. In cerebellar granule cells, Wroblewska et al. (1993) discovered that NAAG activated a group II mGluR, but not the other cAMP-coupled, group III, mGluRs or the phosphoinositide-coupled, group I mGluRs. NAAG was found to be a highly selective mGluR3 agonist with a potency that rivaled glutamate in cells transfected with mGluR3 cDNA. Assays of NAAG action at the highly related mGluR2 receptors in stably transfected cell lines indicate that NAAG concentrations as high as 300 μM did not activate this receptor significantly (Wroblewska et al., 1997 ; Hess et al., 1999). Additionally, in assays of chimeric receptors that contained either the extracellular domain of mGluR2 or mGluR3 coupled to the cytoplasmic and transmembrane domain of mGluR1α, only the mGluR3/1α chimera was activated by NAAG (Wroblewska et al., 1997). In contrast to these data, 300-1,000 μM NAAG is reported to reduce forskolin-stimulated cAMP levels in a cell line transfected with mGluR2 (Cartmell et al., 1998).

NAAG activated a group II, but not a group I or III, mGluR in astrocytes cultured from rat cerebellum. These cells expressed mRNA for mGluR3, but not mGluR2 (Wroblewska et al., 1998). β-NAAG, a synthetic analogue of NAAG in which the peptide bond is formed via the β-carboxyl group of aspartate, recently has been found to act as a selective mGluR3 antagonist (Wroblewska et al., unpublished observations), making it the first antagonist reported to discriminate between mGluR3 and mGluR2.

The consequences of group II receptor activation by the relatively selective agonists, trans-1-amino-1,3,-cyclopentanedicarboxylic acid (trans-ACPD) and (2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl)glycine (DCG-IV), include inhibition of L-type voltage-dependent calcium channels in rat cerebellar granule cells (Chavis et al., 1994), inhibition of L-type calcium current in acutely isolated neocortical neurons (Sayer et al., 1992), and inhibition of depolarization-induced, calcium-dependent GABA release from cultured rat cortical neurons (Schaffhauser et al., 1998). Central to understanding the role of NAAG in activation of the mGluR3 is the report of Bischofberger and Schild (1996) that NAAG, glutamate, and DCG-IV decrease voltage-activated calcium currents in amphibian olfactory bulb neurons.

These data lead to the hypothesis that one role for NAAG in the nervous system is the activation of mGluR3 autoreceptors on presynaptic endings (Fig. 2). Via reduction in cAMP or a direct action of G proteins on voltage-dependent calcium channels, NAAG has the potential to suppress subsequent synaptic release. Such a generic function is consistent with the condistribution of NAAG with the spectrum of small amine transmitters in neurons and with the identification of group II receptors on presynaptic endings (Ohishi et al., 1994). The group II agonist, DCG-IV, suppressed synaptic transmission at the mossy fiber-CA3 synapse (Kamiya et al., 1996) and monosynaptic excitation of spinal motoneurons (Ishida et al., 1993), actions that appeared to be mediated by presynaptic mGluR2 or mGluR3 receptor activation.

Figure 2.

Presynaptic NAAG is often colocalized with small amine transmitters, including GABA and glutamate. As an mGluR3 agonist, the peptide influences cAMP levels, calcium ion channel conductance, (co)transmitter release, GABAA receptor subunit expression, synaptic plasticity, and neurotoxicity. Astrocytes express NAAG, at least one type of NAAG peptidase, and mGluR3. They produce a neuroprotective effect, mediated by transforming growth factor-β (TGFβ), in response to NAAG. The consequences of NAAG's partial agonist activity at some NMDA receptor subtypes presently are unknown. LTD, long-term depression ; LTP, long-term potentiation ; PKA, protein kinase A.

FIG. 2.

Additional data consistent with this hypothesis come from physiological studies in the cerebellum, the nigrostriatal pathway, and the hippocampus. In guinea pig cerebellar slices, Sekiguchi et al. (1989b) observed that 30 μM NAAG decreased the excitatory postsynaptic potentials recorded in Purkinje cell dendrites following stimulation of climbing fibers. Climbing fiber axons and terminals have been shown to contain NAAG (Zollinger et al., 1994 ; Renno et al., 1997). Presumptive dopaminergic neurons in the pars compacta of the substantia nigra express a high concentration of NAAG (Fuhrman et al., 1994 ; Moffett et al., 1994). Infusion of NAAG into the cat striatum reduced the release of dopamine by a non-NMDA-related mechanism (Galli et al., 1991). Macek et al. (1996) demonstrated that group II mGluRs are partially responsible for decreasing inhibitory postsynaptic potentials in the midmolecular layer of the hippocampus.

In testing this hypothesis, Zhao et al. (2000) demonstrated that 30-100 μM NAAG decreases GABA release in cortical neurons in cell culture via mGluR3, protein kinase A, and L-type calcium channels (Fig. 2).


In the hippocampus, NAAG is present in interneurons (Anderson et al., 1987 ; Moffett and Namboodiri, 1995), and mGluR2/3 receptors have been identified immunohistochemically in this tissue (Petralia et al., 1996 ; Shigemoto et al., 1997). Two recent studies of the hippocampal medial perforant pathway provided evidence that NAAG acts through mGluR3 receptors to affect synaptic plasticity. John Sarvey's laboratory found that NAAG blocked long-term potentiation (LTP) and that this blockade was reversed by the mGluR3-selective antagonist, β-NAAG (Lea et al., 1999). Roger Anwyl's laboratory reported that NAAG and other group II agonists induced long-lasting depression of synaptic efficacy and that this effect was blocked by the coapplication of group II mGluR antagonists (Huang et al., 1999). In both of these studies, NAAG failed to alter paired-pulse facilitation significantly, suggesting that the peptide was acting postsynaptically. Additionally, 50 μM NAAG decreased LTP in the CA1 region of the hippocampus (Grunze et al., 1996). These results are consistent with the requirement for increased cAMP in the induction of LTP in CA1 and dentate gyrus and the ability of NAAG to decrease cAMP levels via mGluR3. The role of endogenous NAAG in synaptic plasticity remains to be elucidated. One approach to this question that may prove useful is the pharmacological manipulation of the extracellular peptidase activity that inactivates synaptically released NAAG, as well as the use of the antagonist, β-NAAG.

By activation of mGluR3 receptors in cerebellar granule cells, NAAG induced changes in GABAA receptor subunit expression and function (Ghose et al., 1997). The induction of the α6 subunit mRNA and protein expression by 10-30 μM NAAG was mediated by mGluR3-induced reduction in cAMP and was reversed by cotreatment with agonists for receptors that elevate cAMP. These results support the hypothesis that endogenous NAAG may modulate synaptic function by influencing transcription.


In 1984, Riveros and Orrego reported that an enzyme in rat brain cortical slices hydrolyzed NAAG and that the glutamate that was released in this process interfered with radioligand binding studies. This report was followed by a series of articles that characterized the pharmacology and biochemistry of an enzyme activity in rat brain that hydrolyzed NAAG (Robinson et al., 1987 ; Serval et al., 1990). The peptidase activity was found to be widely distributed throughout the nervous system, consistent with the distribution of NAAG (Blakely et al., 1988b ; Guarda et al., 1988 ; Fuhrman et al., 1994).

In a remarkable achievement by the laboratory of Joseph Coyle, glutamate carboxypeptidase II was purified and antibodies were produced against it (Slusher et al., 1990). These antibodies were used to obtain a partial cDNA sequence from an expression library, and the cDNA was found to be homologous to human prostate-specific membrane antigen (Carter et al., 1996). The ability of human prostate-specific membrane antigen cDNA (Israeli et al., 1993) to express NAAG peptidase activity (Carter et al., 1996) led to cloning of homologous enzymes from rat brain (Bzdega et al., 1997 ; Luthi-Carter et al., 1998) and porcine jejunum (Halsted et al., 1998). Another dipeptidyl peptidase IV cloned from human carcinoma cDNA possesses 67% amino acid identity to human carboxypeptidase II and is highly expressed in testis and ovary. It is not detectable in brain on northern analysis and expresses relatively weak activity against NAAG (Pangalos et al., 1999). Analysis of mRNA expression for this human carcinoma peptidase by RT-PCR suggests that it is expressed in the brain. Another structurally homologous peptidase was purified from blood in human placenta, and its cDNA was cloned (Gingras et al., 1999). The enzyme is secreted and found in blood plasma. When expressed in transfected cells, this peptidase actively hydrolyzes NAAG. However, neither mRNA for this peptidase nor the peptidase protein was identified on northern and western blots, although message and protein expression levels may have been below the level of detection of the reported assays. Preliminary PCR data (T. Bzdega, B. Wroblewska, and J. H. Neale, unpublished observations) suggest that human brain cDNA contains sequences that are homologous to both the carcinoma and placental peptidase cDNAs. A structurally related ileal peptidase has no activity against NAAG (Pangalos et al., 1999).

Consistent with the structure of glutamate carboxypeptidase II as a type 2 membrane protein, NAAG peptidase activity has been localized on the extracellular face of the plasma membrane of isolated retinal cells, cultures containing mouse brain neurons and glia, mouse brain glia cultured alone, and nonmyelinating Schwann cells (Williamson and Neale, 1992 ; Cassidy and Neale, 1993 a, b ; Berger et al., 1995 ; Berger and Schwab, 1996). Using antibody prepared against purified NAAG peptidase protein, immunohistochemical analysis suggests that the enzyme is concentrated, if not exclusively localized, in glia (Stauch-Slusher et al., 1992 ; Fig. 2). In situ hybridization, northern blots, and enzymatic assays also indicate that this peptidase does not appear to be expressed in neurons, but rather is found in most, if not all, astrocytes, Bergmann glia, Müller cells, and satellite cells in dorsal sensory ganglia (Luthi-Carter et al., 1998 ; Berger et al., 1999).

The rat brain membrane enzyme activity is blocked by quisqualate, β-NAAG, phosphate (Robinson et al., 1987 ; Serval et al., 1990), and (phosphonomethyl)pentanedioic acid (PMPA) (Jackson et al., 1996). PMPA possesses a Ki in the low nanomolar range (Jackson et al., 1996). It is reported to decrease the ischemia-induced elevation in extracellular glutamate and reciprocally to increase extracellular NAAG levels, a result that can be ascribed to reducing the rate of hydrolysis of extracellular NAAG (Vornov et al., 1999). The reduction in extracellular glutamate levels also could result from mGluR3-mediated actions of NAAG at glutamatergic synaptic endings, similar to that proposed for GABA (Fig. 2). PMPA also protects against neural injury in cell culture and in vivo models of ischemia (Slusher et al., 1999). Although PMPA is reported to be inactive at a broad spectrum of transmitter receptors and uptake systems (Slusher et al., 1999), it also functions as a potent (nanomolar) antagonist at mGluR3 (B. Wroblewska, T. Bzdega, and J. H. Neale, manuscript in preparation). As a result of NAAG's activity as an agonist at this metabotropic receptor, it may be difficult to discriminate the site of action of PMPA in systems where the peptide, peptidase, and receptor are present. Recently, another compound with structural similarity to NAAG has been found to have nanomolar potency as a peptidase inhibitor while being ~500-fold less potent as an mGluR3 agonist (Nan et al., 2000).


NAAG and peptidase activity against NAAG have been analyzed in tissue from human subjects with schizophrenia (Tsai et al., 1996) and three neurodegenerative conditions : amyotrophic lateral sclerosis (Tsai et al., 1991), Huntington disease, and Alzheimer disease (Passani et al., 1997b). Differences in peptidase activity and in NAAG concentration were observed, often restricted to specific human brain regions. These data support the conclusion that the metabolism of NAAG may be altered in these pathological states, particularly as a result of neuron loss, but provide no conclusive data in support of hypotheses as to the causal versus consequential roles of the peptide or peptidase in these disorders.

James Meyerhoff's laboratory has analyzed the influence of kindled seizures in rats on NAAG and the peptidase activity (Meyerhoff et al., 1985, 1989), observing an increase in NAAG concentration in the entorhinal cortex and a more generalized decline in the peptidase activity. In a strain of rats that are genetically prone to epilepsy, the peptidase activity was increased in several brain regions (Meyerhoff et al., 1992). Although it is difficult to propose a role for the peptidase in seizures, these studies clearly support the hypothesis that the peptidase activity is influenced by the excitatory events in the brain.

Ferdinando Nicoletti's laboratory demonstrated a two-fold neuroprotective effect of NAAG on NMDA-induced neurotoxicity in cortical cell cultures. First, when NAAG was coapplied with NMDA, the toxicity was blocked by an mGluR3-independent mechanism, possibly by NAAG acting as a partial NMDA receptor agonist. High concentrations of NAAG alone also induced NMDA receptor-dependent neuronal cell death. From a clinical perspective, the more important experimental result was that NAAG substantially reduced neuronal cell death when applied after the NMDA treatment. This effect was dependent on NAAG activation of mGluR3 on astrocytes in these cultures. It was mimicked by treatment of the glia with NAAG and subsequent transfer of the astrocyte medium to the neurons (Bruno et al., 1998a). Transforming growth factor-β released from astrocytes following mGluR3 activation mediates this neuroprotective effect (Bruno et al., 1998b ; Fig. 2). NAAG also reduced hypoxia- and NMDA-induced neuron death in cell culture models, and this effect was blocked by a group II mGluR antagonist (Yourick et al., 1998). Increasing the effective NAAG concentration by inhibition of peptidase activity with PMPA also is reported to protect cholinergic spinal cord neurons in culture against chronic glutamate-induced neurotoxicity (Corse et al., 1997).

NAAG coinjection with quinolinic acid reduced the size of striatal quinolinic acid-induced lesions (Orlando et al., 1997). Here again, NAAG may have been acting as a partial agonist at the NMDA receptor or via mGluR3 receptor activation. β-NAAG also was neuroprotective in this system, a startling result that awaits a fuller understanding of the pharmacology of this compound that is currently known to inhibit NAAG peptidase activity and act as an mGluR3 antagonist (Lea et al., 1999).


More than three decades after its discovery, NAAG has emerged as a peptide transmitter of substantial importance in influencing the excitatory balance in nervous system function as a highly selective mGluR3 agonist. Inactivation of the peptide by extracellular peptidase activity produces glutamate, providing the opportunity for a cascade of signal transduction following NAAG release. The physiological consequences of this glutamate will be dependent on the relative rates of NAAG hydrolysis and glutamate reuptake. Among the challenges that remain are defining the role of endogenous NAAG at identified synapses, the contribution of NAAG to the acute and tonic concentrations of glutamate (Sah et al., 1989) in the extracellular space, the role of glia in mediating NAAG's function via both the peptidase and mGluR3, the mechanisms that mediate plasticity and neuroprotection induced by NAAG, and the biosynthetic pathway for NAAG within neurons and glia.