The existence of binding sites for benzodiazepines (BZs), the most widely prescribed and therapeutically used drugs for their anxiolytic, sedative, and muscle relaxant properties, has prompted several teams to search for endogenous ligands of the BZ receptors. Thus, the team of Erminio Costa has isolated from rat brain extracts an 11-kDa polypeptide able to competitively displace tritiated diazepam on synaptosomes, which has been called diazepam binding inhibitor (DBI) (1). Two major compounds, generated from DBI, have been identified in rat brains, the triakontatetraneuropeptide (TTN, DBI₁₇₋₅₀) and the octadecaneuropeptide (ODN, DBI₃₃₋₅₀). DBI and its processing products, collectively grouped under the generic term of endozepines (2), are currently considered as the endogenous ligands of BZ receptors. DBI is also known as an acyl-coenzyme A-binding protein (ACBP) due to its ability to bind acyl-coenzyme A esters and to stimulate fatty acid synthesis in different cell types (3).

In this review, which will only refer to endozepines as natural ligands of BZ receptors, the acronym DBI will be used. In addition, we will focus on ODN, which is the major form of endozepines produced in the brain and whose structure has been well-preserved during evolution, suggesting that it exerts important biological functions in the central nervous system (CNS). Supporting this hypothesis, high concentrations of DBI mRNA and DBI-derived peptides are detected in the rat brain during ontogenesis. Furthermore, it has been reported that in the cerebellar cortex, ODN is expressed by Bergmann glia, which controls cerebellar granule neuron migration (4), indicating that ODN may act as a neurotrophic factor during brain development.

Cloning and characterization of cDNAs encoding DBI were first performed from a rat brain cDNA library (5). Since then, DBI has been cloned from other tissues (adrenal, liver, and testis) and from various animal species, including human (6), cattle (7), frog (8), and carp (9). The cDNA encoding DBI has also been cloned in invertebrate species such as the fruit fly Drosophila melanogaster (10), the butterfly Bombyx mori (11), and the tobacco maggot Manduca sexta (12), different plant species such as Arabidopsis thaliana (13), and unicellular organisms such as the yeast Saccharomyces cerevisiae (14) and the bacteria Escherichia coli (15, 16).

Southern blot analysis of genomic DNA from human (6), rat (5), and mouse (17) revealed the existence of several genes encoding DBI, but only one of which is functional. In human, the active gene is located on the q12-21 region of the long arm of chromosome 2 (18) and the pseudogenes are located on chromosomes 5, 6, 11, and 14 (19). The human and murine genes are organized into four exons (E1–E4), coding for regions 1–2, 3–41, 42–62, and 63–86 of the DBI, respectively (20, 21), while in Drosophila, there are only three exons (10). In human and murine adipocytes, a novel exon, named E1c, has been identified in the DBI gene, which codes with the three other exons for a longer transcript. Genomic DNA analysis shows that the E1c exon is located in the intron between exons E1 and E2, and that its expression would be under the control of a second promoter located in intron 1 (22, 23). These data reinforce the work of Helledie et al. (24), which showed that the DBI gene, cloned from human and murine adipocytes, contains a functional consensus sequence on intron 1, the peroxisome proliferator-activated receptor (PPAR)-response element (PPRE), capable of binding the transcription factors PPARγ and retinoid X receptor α. All these data indicate that the expression of the DBI gene is under the control of two promoter regions (P1 and P2) located upstream and downstream of the E1 exon initially described in human and rat (Figure 1).

The absence of TATA and CCAAT boxes, the presence of several element initiators of transcription, and the high content of nucleotide acids C and G in a promoter are often considered as the hallmarks of a domestic gene. Nevertheless, the expression level of DBI gene is regulated in different cell types (25). In addition, the P1 and P2 regions of the DBI gene have several consensus sequences for transcription factors including a glucocorticoid-response element (GRE), a CAAT-binding transcription factor/nuclear factor 1 (CTF/NF1), a hepatocyte nuclear factor (HNF), a sterol regulatory element-like sequence (SRE), and a PPAR sequence. Moreover, it has been reported that glucose and hormones, such as insulin or androgens, regulate the expression of the DBI gene (21, 26–28). Several research groups have also shown a modification of the levels of DBI gene expression in the brain in various experimental conditions. For instance, a decrease in DBI mRNA levels is observed after starvation (29) or castration (30) in rat, while an increase of DBI expression is shown after high-frequency vestibular stimulation in rabbits (31). Drug addiction (alcohol, morphine, and nicotine) also leads to an increase in DBI gene expression in the rodent brain (32, 33), and abrupt interruption of the consumption of these substances further increases DBI mRNA levels (33). In vitro studies on cultured astrocytes have shown that DBI gene expression is stimulated by different cell stress inducers such as peptide β-amyloid (34) and hydrogen peroxide (H₂O₂) (35). All these observations reveal that the DBI gene cannot be considered as a “housekeeping gene.”

Ultrastructural studies revealed that DBI immunoreactivity is diffused throughout the cytoplasm (39), excluding its packaging in secretory vesicles. This is consistent with the absence of signal peptide in the primary sequence of DBI that could direct the protein into the endoplasmic reticulum/Golgi system. Nevertheless, mass spectrometry analysis of rat astroglial cell secretome revealed the presence of DBI in incubation media (46). The endozepine secretion is insensitive to brefeldin A, an inhibitor of Golgi vesicular transport (46), and can be stimulated by induction of autophagy by rapamycin (47). These data indicate that DBI and its derived peptides may be released by a mechanism independent of the conventional vesicular exocytosis pathway. The translocation of endozepines to the extracellular space could be provided by carriers called ATP-binding cassettes (ABCs), as demonstrated for the release of interleukin (IL)-1β by astrocytes (48) and annexin 1 by follicle-stellar cells (49), two proteins whose precursors are devoid of a signal peptide sequence. Consistent with this hypothesis, a site-directed mutagenesis study conducted with the amoeba Dictyostelium discoideum has shown the existence of a direct interaction between DBI and ABC-serine protease transporter (TagA) (50).

Although the mechanism of secretion of ODN and other DBI-derived peptides is not yet clearly elucidated, the secretion of endozepines is modulated in different pathophysiological situations and regulated by many factors. An assay using antibodies against ODN shows that the level of ODN-like immunoreactivity (ODN-LI) is increased in the plasma of septic shock patients (51). In vitro studies reveal that the release of ODN-LI is positively regulated by numerous neuroactive compounds, including neuropeptides [pituitary adenylate cyclase-activating polypeptide (PACAP) (52), urotensin-II (UII), UII-related peptide (53), and peptide β-amyloid (54)], steroid hormones [cortisol, pregnenolone, or progesterone (55)], cytokines [tumor necrosis factor-α and IL-1β (56)], or elevated extracellular potassium concentrations (57), and conversely repressed by the neuropeptide somatostatin (58) and the neurotransmitter gamma aminobutyric acid (GABA) (59). Pharmacological studies indicate that the release of ODN-LI from astroglial and retinal cells is mediated by ABC transporter activity through a mechanism that is dependent on the phosphorylation by protein kinases A and C (PKA and PKC) (52, 54, 60) (Figure 2), and blockage of the phosphorylation of ABC transporter inhibits the basal release of ODN-LI by about 50%. A phosphorylated form of ODN is also released by astrocytes but appears less active than the non-phosphorylated peptide, suggesting the existence of a posttranslational regulation mechanism (61).

In culture media from astrocytes exposed to moderate oxidative stress, the quantity of authentic ODN is significantly higher than that detected in culture media from control astrocytes (35). Clinical studies have shown that the levels of endozepines, ODN, and DBI, increase in the plasma or cerebrospinal fluid (CSF) of patients suffering from pathological disorders related to oxidative stress, such as systemic inflammation, hepatic encephalopathy, and neurodegenerative diseases (25). Such an increase in ODN release could be responsive for the stimulation of antioxidative defenses (induction of Mn-superoxide dismutases, catalase, glutathione peroxidase-1, sulfiredoxin-1 gene transcription, and glutathione biosynthesis) in glia and neurons (62, 63) in agreement with the emerging concept indicating that ODN acts as a potential cytoprotective actor preventing the deleterious action of oxidative insults.

As endogenous ligands of BZ receptors, ODN interacts with the central-type BZ receptors, which is an integrated component of the GABA-A receptor (GABA-A-R)/Cl⁻ channel complex (64, 65). Like DBI, ODN has been shown to displace diazepam and modulate GABAergic transmission via an allosteric reaction (1). Electrophysiology and cell functional studies indicate that ODN may act as an agonist, i.e., a positive allosteric modulator (PAM) promoting the action of GABA, or as an inverse agonist, i.e., a negative allosteric modulator (NAM) reducing the action of GABA, depending on the subunit composition of the GABA-A-R (57, 66). Studies conducted on cultured astrocytes have demonstrated that ODN is also the endogenous ligand of a G protein-coupled receptor, which is functionally and pharmacologically different from the conventional BZ receptors (67, 68). In particular, ODN stimulates the metabolism of polyphosphoinositides (PIPs) in rat astrocytes and cerebellar granule neurons via a phospholipase C (PLC) coupled to a G protein sensitive to pertussis toxin (63, 69). In addition, in cultured astrocytes, ODN increases the intracellular concentration of calcium from intracellular pools (68, 70). In the same cell type, ODN stimulates the formation of cAMP through activation of adenylyl cyclase (AC) activity (61, 71). The ability of ODN to stimulate phosphorylation of extracellular signal-regulated kinase 1 and 2 (ERK 1 and 2) is also found in both astrocytes and cerebellar granule neurons (63, 71). These data collectively indicate that ODN is the natural and specific ligand of a G_i/0 or G_S protein-coupled receptor, which activates complementary transduction mechanisms depending on the cell type.

The structure–activity studies of ODN in relation to its ability to activate the metabotropic receptor expressed in rat astrocytes show that the C-terminal octapeptide of ODN (OP, ODN₁₁₋₁₈) is the shortest biologically active fragment of the peptide (72). OP is able to mimic the effects of ODN on calcium mobilization and cAMP level increase observed in astrocytes (71, 72). Conversely, OP has no effect on the binding of [³H]flumazenil in cerebellar granular cells or of [³H]PK11195 in rat astrocytes (73). The effects of ODN on the activation of the cAMP/PKA, PIPs/calcium/PKC, and MAPK-ERK transduction pathways in astrocytes and neurons are completely suppressed by a specific antagonist of this receptor, the cyclo₍₁₋₈₎[DLeu⁵]OP (63, 68, 72).

Pharmacological studies have shown that central BZ receptor agonists are unable to mimic the protective effects of ODN (25). Similarly, central-type BZ receptors and translocator protein (TSPO) antagonists fail to suppress ODN-induced protection of neurons and glial cells from injuries, indicating that “classical” BZ receptors are not involved in the protective activity of ODN. However, OP and cyclo₍₁₋₈₎OP, two agonists of the ODN metabotropic receptor, exhibit a protective activity when used in the same range of concentration as ODN. In addition, cyclo₍₁₋₈₎[DLeu⁵]OP, a potent antagonist of this metabotropic receptor, blocks the protective effects of ODN on neurons and astrocytes (62, 63, 71, 82).

Downstream of the receptor, several transduction pathways can mediate the protective effect of ODN according to the cell type. In astrocytes, activation of ODN metabotropic receptor implies stimulation of the activity of the AC/cAMP/PKA pathway (71). The neuroprotective effect of ODN against 6-OHDA-induced cerebellar granule cell death involves activation of the PLC/IP3/PKC pathway (63). Downstream to PKA and PKC activations, ODN stimulates the phosphorylation of the ERK1 and ERK2 proteins in astrocytes and neurons (79). The phosphorylation of ERK1 and ERK2 in turn stimulates the expression of the Bcl-2 protein and concomitantly inhibits the expression of Bax, thereby preventing mitochondrial dysfunction and blocking cell death via the intrinsic apoptotic pathway (84, 94). The stimulating effect of ODN on the expression of antioxidant enzyme genes and their enzymatic activities is dependent on the activity of the cAMP/PKA transduction pathway as well as that of MAPK–ERKs in astrocytes (62, 77, 82). ODN promotes cerebellar granule cell survival from apoptosis through the PKC pathway (63), while another neuropeptide, PACAP, protects the same neurons from apoptosis through the PKA pathway (95). It would thus now be of interest to investigate if combining the two molecules leads to an enhanced and/or prolonged antiapoptotic effect.

Clinical studies have revealed that endozepine levels are significantly increased in the CSF of patients suffering from neurological disorders such as epilepsy, hepatic encephalopathy (96, 97), or Alzheimer's and Parkinson's diseases (98). Mass spectrometry, radioimmunoassay (RIA), and q-RT PCR analyses reveal that moderate oxidative stress induces stimulation of the endogenous production of ODN, as well as the expression of the ODN precursor gene DBI from cultured astrocytes (76). Furthermore, induction of endogenous ODN production is responsible for (i) the stimulation of the expression and activity of antioxidant enzymes and thus rapid resorption of ROS, (ii) the protection of bio-macromolecules from oxidative damages, and (iii) the prevention of astrocytes and neuron cell death induced by oxidative stress. In contrast, inhibition of ODN release by the PKA inhibitor H89 or blockage of the effects of ODN with an antagonist of its metabotropic receptor, the cyclo₍₁₋₈₎[DLeu⁵]OP, exacerbates damages induced by oxidative stress on astroglial cell viability (76). Consistently, knockdown of ODN precursor expression, by DBI siRNA, induces morphological alterations with loss of membrane integrity and the formation of apoptotic bodies and increases the vulnerability of oxidative stress inducing cell death (76). It has also been shown, by using DBI-KO animals, that ODN precursor deficiency increases brain sensitivity to MPTP toxicity, highlighting the neuroprotective role of endogenous ODN against neuronal degeneration (35). Taken as a whole, these data indicate that the induction of ODN production in pathological conditions reproducing early stages of neurodegenerative processes may represent a compensatory mechanism, initiated by reactive astrocytes, to reduce their sensitivity, as well as that of their surrounding neurons, to oxidative aggression and to limit the progression of the neurodegeneration process in some neurological disorders.

OM-K, AN, GL, JL, M-CT, and DV provided valuable editorial comments and relevant bibliographic references. YH, SB, IG, and DV provided valuable editorial comments. OM-K wrote the manuscript with the contribution of AN, GL, and DV. TG, JC, and BL contributed to the review and editing revised version. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.