Glyoxalase 1 increases anxiety by reducing GABA_A receptor agonist methylglyoxal

Glyoxalase 1 (Glo1) expression has previously been associated with anxiety in mice; however, its role in anxiety is controversial, and the underlying mechanism is unknown. Here, we demonstrate that GLO1 increases anxiety by reducing levels of methylglyoxal (MG), a GABA_A receptor agonist. Mice overexpressing Glo1 on a Tg bacterial artificial chromosome displayed increased anxiety-like behavior and reduced brain MG concentrations. Treatment with low doses of MG reduced anxiety-like behavior, while higher doses caused locomotor depression, ataxia, and hypothermia, which are characteristic effects of GABA_A receptor activation. Consistent with these data, we found that physiological concentrations of MG selectively activated GABA_A receptors in primary neurons. These data indicate that GLO1 increases anxiety by reducing levels of MG, thereby decreasing GABA_A receptor activation. More broadly, our findings potentially link metabolic state, neuronal inhibitory tone, and behavior. Finally, we demonstrated that pharmacological inhibition of GLO1 reduced anxiety, suggesting that GLO1 is a possible target for the treatment of anxiety disorders.

Anxiety disorders, such as post-traumatic stress disorder, panic disorder, social phobia, specific phobia, and generalized anxiety disorder, comprise the most common psychiatric diseases in the United States (1). Mouse genetic studies have identified associations between glyoxalase 1 (Glo1) and anxiety-like behavior. Our previous work identified a copy number variant (CNV) in mice that causes a duplication of Glo1 (2). The Glo1 duplication was associated with increased Glo1 expression and increased anxiety-like behavior (2). This CNV underlies differential Glo1 expression and anxiety-like behavior previously reported among inbred mouse strains (3). Nevertheless, discrepant findings (4) have made the role of GLO1 in anxiety controversial (5). Here, we report the generation of mice with a Tg bacterial artificial chromosome (BAC) containing Glo1. We used these Tg mice to investigate the effect of Glo1 on anxiety-like behavior. The BAC transgene allows overexpression of Glo1 using the endogenous cis-regulatory elements on an isogenic background (6). BAC transgene expression is often copy-number dependent (7), providing an appropriate model of the Glo1 CNV.

Importantly, the molecular mechanism underlying the effect of GLO1 on anxiety has yet to be identified. GLO1 is a ubiquitous cytosolic enzyme that catabolizes acyclic α-oxoaldehydes, particularly methylglyoxal (MG) (8). MG is mainly formed from the nonenzymatic degradation of the glycolytic intermediates dihydroxyacetone phosphate and glyceraldehyde-3-phosphate (9). In vitro studies have demonstrated a critical role for GLO1 in detoxifying MG: pharmacological inhibition of GLO1 results in MG accumulation (10), while overexpression of Glo1 prevents MG accumulation (11). When MG accumulates to high levels, it has cytotoxic effects, including protein and nucleotide modification (advanced glycation end products [AGEs]), generation of reactive oxygen species, and cell death (8, 12). Despite the well-known cellular effects of MG, there is no known mechanism linking MG to anxiety. In this study, we used BAC Tg mice to explore the mechanism by which GLO1 modulates anxiety. We focused on the possibility that GLO1 increases anxiety by regulating MG concentration in the brain. First, we established that low doses of MG are anxiolytic and that high doses of MG produce pharmacological effects similar to those of GABA_A receptor agonists. We then assessed electrophysiological effects of MG on GABA_A receptors in vitro.

Glo1 copy number regulates Glo1 mRNA and protein expression. To investigate the effect of GLO1 on anxiety, we obtained a BAC containing the Glo1 gene and its cis-regulatory elements (Figure 1A). The BAC contained partial copies of 2 additional genes, Dnahc8 and Btbd9. We ablated the transcriptional start sites of these flanking genes using recombinogenic engineering (13) by inserting ampicillin and kanamycin cassettes into the first exons of each gene, respectively (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI61319DS1). Therefore, this BAC should only express Glo1. Using this modified BAC, we generated 3 lines of Tg mice on a C57BL/6J (B6) background. We measured BAC copy number in each line using quantitative real-time PCR (qPCR): Tg lines had 2, 8, and 10 copies of the BAC (Figure 1B). Tg animals were fertile, healthy, and did not display any grossly discernible physical or behavioral abnormalities (Supplemental Table 1).

Figure 1

Glo1 copy number regulates Glo1 expression. (A) The BAC (RP23-247F19) contained Glo1 and partial copies of Dnahc8 and Btbd9. The first exons of Dnahc8 and Btbd9 were ablated by inserting ampicillin (AMP) and kanamycin (KAN) cassettes, respectively. (B) BAC copy number was estimated by qPCR on genomic DNA (n = 8 WT and 3–5 Tg per line). P < 10^–10, 1-way ANOVA. (C) Glo1 mRNA was estimated in whole brain by qPCR (n = 12 WT and 4–7 Tg per line). P < 10^–14, 1-way ANOVA. (D and E) GLO1 protein was estimated in whole brain by immunoblot (n = 16 WT and 3–6 Tg per line). P < 10^–11, 1-way ANOVA. Data are mean ± SEM. *P < 0.05; **P < 0.0001.

Next, we measured Glo1 mRNA expression using qPCR. Tg mice showed a copy number–dependent increase in Glo1 mRNA in the brain (Figure 1C) and peripheral tissues (Supplemental Figure 2). Similarly, Tg mice displayed a copy number–dependent increase in GLO1 protein in the brain, as measured by immunoblot (Figure 1, D and E). Thus, the BAC dose-dependently increased Glo1 mRNA and protein in Tg mice. Using gene-expression microarrays, we confirmed that Tg mice overexpressed Glo1, but we did not observe any other changes in gene expression (Supplemental Table 2). In particular, we did not observe increased expression of Dnahc8 or Btbd9, confirming successful ablation of their transcription from the BAC.

Glo1 overexpression increases anxiety-like behavior. We used male Tg mice to investigate the effect of Glo1 overexpression on anxiety-like behavior in the open field (OF) test. The OF test is sensitive to anxiogenic and anxiolytic agents (14) and is associated with brain regions and neurotransmitters involved in human anxiety (15, 16). Further, the OF test was sensitive to differential Glo1 expression in previous studies (2, 3). Tg mice displayed a significant, copy number–dependent decrease in time in the center of the OF (referred to throughout as center time) compared with that of WT mice (Figure 2A), reflecting increased anxiety-like behavior (17). WT and Tg mice did not significantly differ in total distance traveled during the test (Figure 2B). These data provide direct evidence that GLO1 increases anxiety-like behavior.

Figure 2

Glo1 overexpression increases anxiety-like behavior. WT and Tg littermates were tested in the OF test (n = 74 WT [pooled across all lines] and 14–21 Tg per line). (A) Center time. P < 0.001, 1-way ANOVA. (B) Total distance. P > 0.1, 1-way ANOVA. Data were pooled from multiple individual experiments. Data are mean ± SEM. *P < 0.05.

We generated 5 additional lines of Tg mice on a FVB/NJ (FVB) background to assess the generalizability of the anxiogenic effect of GLO1. FVB Tg mice showed increased Glo1 copy number (Supplemental Figure 3A) and expression (Supplemental Figure 3B) compared with WT mice. In the OF test, FVB Tg mice displayed decreased center time compared with that of WT mice (Supplemental Figure 3C). These results replicate our findings in B6 Tg mice and demonstrate that GLO1 increases anxiety-like behavior in multiple genetic backgrounds.

In order to more thoroughly assess the role of GLO1 in anxiety, we tested WT and Tg mice in 2 additional well-established tests of anxiety-like behavior, the light-dark (LD) box test (18) and the elevated plus maze (19). In the LD box test, Tg mice spent less time in the light compartment compared with WT mice (Supplemental Table 3). WT and Tg mice did not significantly differ in the number of transitions between the 2 compartments, indicating normal locomotor activity (18, 20). In the elevated plus maze, Tg mice made fewer entries into the open arms compared with WT mice (Supplemental Table 4). WT and Tg mice did not differ in the number of total arm entries, again indicating normal locomotor activity. Together, these data demonstrate that Glo1 overexpression increases anxiety-like behavior across different genetic backgrounds and multiple behavioral tests.

Glo1 increases anxiety by reducing MG levels. We used the B6 Tg line with the highest copy number (B6 line 3) to explore the molecular mechanism underlying GLO1’s anxiogenic effect. Given known role of GLO1 in clearing MG (8), we hypothesized that GLO1 increases anxiety by regulating MG concentration. First, we measured MG metabolism in WT and Tg mice by a GLO1 enzymatic activity assay. Brain tissue from Tg mice metabolized a hemithioacetal substrate, formed from MG and glutathione, more rapidly than brain tissue from WT mice (Figure 3A). This reflects an increased capacity for clearing MG. We next measured MG concentration in the brains of WT and Tg mice by HPLC. Tg mice had an approximately 10% reduction of MG concentration in the brain compared with WT mice (Figure 3, B and C). These results indicate that Glo1 overexpression reduces MG concentration in the brain.

Figure 3

MG regulates anxiety-like behavior. (A) GLO1 enzymatic activity in whole brain (n = 3 WT and 3 Tg). (B and C) HPLC measurement of MG concentration in whole brain. (B) Representative chromatograms from WT (solid) and Tg (dashed) mice. An enlarged view of the relevant peak is shown in the inset. (C) Average MG concentration (n = 9 WT and 8 Tg). (D) HPLC measurement of MG concentration in whole brain after i.p. treatment with MG. Assay order had a significant effect on MG concentration and was used as a covariate in a 1-way analysis of covariance (ANCOVA) for the factor treatment (n = 4–7 per group). P = 0.009. (E and F) MG decreased anxiety-like behavior in the OF test. MG (50 mg/kg) (E) increased time in the center of the OF but (F) did not change total distance traveled (n = 18 per group). mAU, milli absorbance units. Data are mean ± SEM. *P < 0.05, **P < 0.0005.

We hypothesized that MG is anxiolytic and that GLO1 increases anxiety by reducing endogenous MG concentration. To test this hypothesis, we administered MG i.p. to male WT mice and found that MG treatment dose-dependently increased MG concentration in the brain (Figure 3D and Supplemental Table 5). We then treated mice with MG (50 mg/kg) and tested them in the OF test 10 minutes after injection. MG treatment increased center time by approximately 30% (Figure 3E), without affecting distance traveled (Figure 3F).

We then used a 3-pronged approach to establish the validity and robustness of the anxiolytic effect of MG. First, we performed 6 replication studies to independently confirm the anxiolytic effect of MG in the OF test. A meta-analysis of the replication studies demonstrated that MG treatment significantly reduced anxiety-like behavior (Cohen’s d = 0.78; meta-analysis z-score = 5.0; P < 0.0001) (Supplemental Figure 4 and Supplemental Table 6). Second, we tested CD-1 mice treated with MG (50 mg/kg) in the OF test. Again, MG treatment increased center time in the OF, without affecting distance traveled (Supplemental Figure 5). This demonstrates that the anxiolytic effect of MG is not strain-specific. Third, we tested mice treated with MG (50 mg/kg) in the LD box test. MG treatment increased time spent in the light compartment, without affecting number of transitions (Supplemental Figure 6). This bolsters our initial finding and demonstrates that MG is anxiolytic across different behavioral tests of anxiety. Together, these results provide robust evidence that MG is acutely anxiolytic and that reduced MG levels mediate the anxiogenic effect of GLO1.

At high concentrations, MG has been shown to induce neuronal apoptosis (21). Therefore, we used in situ TUNEL staining to assess apoptosis in brains of mice treated with MG (Supplemental Figure 7). We found no evidence of increased apoptosis in mice treated with the anxiolytic dose of MG (50 mg/kg). Notably, the time course for the anxiolytic effects of MG is distinct from the time course for its cytotoxic effects: MG is anxiolytic within minutes of administration, while high doses of MG require hours to induce apoptosis (21). Given the disparity in time course and the lack of apoptosis in the brains of mice treated with 50 mg/kg MG, we conclude that the anxiolytic effect of MG is independent of cytotoxicity.

MG has GABAergic properties in vivo. To further investigate the mechanism of MG’s anxiolytic effect, we examined pharmacological properties of MG at higher doses. At 100 mg/kg, MG caused decreased locomotion (Figure 4A). At 300 mg/kg, MG caused ataxia (Figure 4B), hypothermia (Figure 4C), and sedation (data not shown). This behavioral profile is similar to those of known GABA_A receptor agonists, such as ethanol, barbiturates, and benzodiazepines (22–24). GABA is the primary inhibitory neurotransmitter in the mammalian CNS and acts through 2 receptor subtypes, GABA_A and GABA_B. The GABA_A receptor subtype is a ligand-gated Cl^– channel; when activated, it triggers increased intracellular Cl^– concentration, membrane hyperpolarization, and reduced neuronal excitability (25). Positive modulators of GABA_A receptors have anxiolytic properties (26) and are a mainstay of the clinical treatment of anxiety (27). Based on the pharmacodynamic profile of MG, we hypothesized that MG is a novel endogenous GABA_A receptor agonist.

Figure 4

MG has GABAergic effects in vivo. (A) 100 mg/kg MG caused locomotor depression (n = 8 per group). (B) 300 mg/kg MG increased foot slips on the balance beam (n = 14 per group). Mann-Whitney U test, U = 26; P = 0.0002. (C) 300 mg/kg MG caused hypothermia (n = 10 per group). Data are mean ± SEM. *P < 0.05, **P < 0.0005.

MG activates GABA_A receptors. To test this hypothesis, we studied the electrophysiological effects of MG on primary cerebellar granule neurons (CGNs) using current-clamp recording to assess neuronal membrane potential (V_m) and voltage-clamp recording to measure ionic currents. In whole-cell current-clamp mode, extracellular application of MG rapidly depolarized CGNs in a concentration-dependent manner (Figure 5A). Initial, rapid depolarization was followed by repolarization of the V_m. In whole-cell voltage-clamp mode, application of MG produced inward currents at negative membrane potentials, characteristic of Cl^– channel activation under our recording conditions (Figure 5B). The MG-evoked current diminished rapidly (Figure 5B), consistent with activation and desensitization of ligand-gated ion channels. These electrophysiological effects were similar to those elicited by application of GABA (Figure 5, A and B). Current density was concentration-dependent and similar in character to that evoked by GABA, while the relative amplitude was approximately one-third that of GABA. To establish that MG acts at GABA_A receptors, responses were studied in the presence of the selective GABA_A receptor antagonist SR-95531 (SR). SR inhibited more than 95% of the V_m (Figure 5C) and current-density changes (Figure 5D) induced by both MG and GABA.

Figure 5

MG is a GABA_A receptor agonist in CGNs. (A) CGNs are depolarized by MG (white circles) or GABA (black circles) (EC₅₀, 10.5 ± 0.5 μM MG; Hill coefficient, 1.17). The relative amplitude of depolarization is shown normalized to the response of each cell to 100 μM MG. (B) MG evokes inward currents in a concentration-dependent manner (EC₅₀, 12 ± 0.7 μM; Hill coefficient, 1.13). The amplitude of currents is shown normalized to the peak response of each cell (I/Imax). (C) Depolarization evoked by 10 μM MG (left) or GABA (middle) was blocked by 10 μM SR. Mean data are plotted as a histogram (right). (D) Inward currents evoked by 10 μM MG (left) or GABA (middle) were also blocked by 10 μM SR. Mean data are plotted as a histogram (right). (E) Currents evoked by 100 μM GABA were reduced by coapplication of MG (left). Scale bars: 200 pA/pF, 25 s. Mean data are normalized to the current evoked by 100 μM GABA in each cell (right). (F) Currents observed approximately 40 seconds after the application of 10 μM MG to the inside of macropatches excised from CGNs were blocked when 10 μM SR was included in the pipette. Application of 10 μM GABA to the inside of macropatches did not evoke a current. Mean data are plotted as a histogram. The bar above each trace shows the duration of drug application. Data are mean ± SEM. n = 6–12 cells or macropatches per condition.

We next established that electrophysiological effects of MG were specific to GABA_A receptor activation. MG did not affect the major sodium currents (Supplemental Figure 8A) or potassium currents (Supplemental Figure 8, B and D–F) in CGNs. Moreover, MG did not activate GABA_B receptors (Supplemental Figure 8C), glycine receptors (Supplemental Figure 8G), or glutamate receptors (Supplemental Figure 8, H and I). Together, these results demonstrate that activation of GABA_A receptors is both necessary and sufficient for electrophysiological activity of MG.

We also compared the baseline electrophysiological properties of CGNs cultured from WT and Tg mice. Neurons from Tg mice were more excitable, displaying a diminished potassium leak current (I_Kso), a depolarized V_m, and increased cellular input resistance (Supplemental Figure 9).

MG is a competitive partial agonist at GABA_A receptors. Next, we investigated whether MG acts competitively or noncompetitively with GABA. To do this, we coapplied MG and GABA to CGNs at 100 μM, concentrations that saturate GABA_A receptor response (Figure 5B), indicating that binding sites are maximally occupied. Consistent with data shown in Figure 5B, 100 μM MG activated a current one-third the magnitude of that elicited by 100 μM GABA. In contrast, coapplication of 100 μM MG and 100 μM GABA activated a current approximately 60% of the magnitude elicited by 100 μM GABA alone (Figure 5E). The fraction of maximal current evoked by coapplication of MG and GABA increased as the concentration of MG was reduced (Figure 5E). These findings suggest that MG is a partial agonist at GABA_A receptors and competes with GABA for the same binding site.

MG diffuses across the plasma membrane to activate GABA_A receptors. Endogenous MG is generated intracellularly; therefore, we investigated intracellular effects of MG on GABA_A receptors using macropatches of plasma membrane (PM). When applied to the intracellular face, MG elicited an inward current following a latency of approximately 40 seconds; this current was inhibited by SR applied to the extracellular face, suggesting that MG crosses the PM by diffusion and then activates GABA_A receptors (Figure 5F). In contrast, GABA, which is unable to cross the PM, did not elicit a current when applied to the intracellular face of the PM macropatch (Figure 5F). These data suggest that endogenously produced MG accesses GABA_A receptors by diffusing across the PM.

MG activates multiple GABA_A receptor subtypes across multiple neuronal types. We next investigated the ubiquity of GABAergic effects of MG, specifically in different neuronal cell types and GABA_A receptor subtypes. We demonstrated that electrophysiological effects of MG in CGNs were similar to those in primary cultures of hippocampal neurons (HNs) (Figure 6A), a cell type relevant to anxiety (28, 29). Specifically, MG dose-dependently evoked inward currents in HNs, with peak responses of approximately one-third of those evoked by GABA. Coapplication of 100 μM MG and 100 μM GABA activated a current of approximately 60% of the magnitude elicited by 100 μM GABA alone; the portion of the maximal current evoked by coapplication increased as the concentration of MG was reduced (Figure 6A). These data implicate MG as a partial GABA_A receptor in HNs as well as CGNs. More importantly, they suggest that MG activates GABA_A receptors to a similar degree in different neuronal cell types, including those relevant to anxiety.

Figure 6

MG is a GABA_A receptor agonist in HNs. (A) Both GABA and MG evoked concentration-dependent inward currents in HNs. The EC₅₀ of currents evoked by MG was 9.5 ± 0.9 μM, with a Hill coefficient of 1.05. The relative amplitude of the currents was normalized to that cell’s response to 100 μM MG. Currents evoked by 100 μM GABA were reduced by coapplication of MG in a concentration-dependent manner. Scale bars: 1 nA, 10 s. Mean data for competition experiments are shown normalized to the current evoked by 100 μM GABA in each cell. (B) Currents evoked by 10 μM MG or GABA were augmented by coapplication of 500 nM diazepam, midazolam, or zolpidem. Scale bars: 1 nA, 10 s. Mean data are plotted as histograms. The bar above each trace shows the duration of drug application. Data are mean ± SEM for 8–10 cells per condition.

We investigated the action of MG at specific GABA_A receptor subtypes by co-applying 10 μM MG to HNs with subtype-selective positive modulators of GABA currents. MG-evoked currents were increased by coapplication of diazepam, midazolam, and zolpidem (Figure 6B). Diazepam, which acts at GABA_A receptors containing α1, α2, α3, or α5 subunits, increased MG-evoked currents by approximately 60% and GABA-evoked currents by approximately 45%. Midazolam also modulates GABA_A receptors containing α1, α2, α3, or α5 subunits in addition to γ2 subunits. Coapplication of 500 nM midazolam increased MG-evoked currents by approximately 40% and GABA-evoked currents by approximately 27%. Zolpidem selectively modules GABA_A receptors containing α1 and γ2 subunits. Coapplication of 500 nM zolpidem increased MG-evoked currents by approximately 15% and GABA-evoked currents by approximately 17% (Figure 6B). Diazepam, midazolam, or zolpidem did not activate currents when applied alone (data not shown), indicating negligible concentrations of endogenous MG and GABA in our experimental conditions. Taken together, these data demonstrate that MG operates a broad range of GABA_A receptor subtypes, similar to GABA. Further, MG acts at GABA_A receptors in HNs as well as CGNs, which have a diverse receptor population containing high levels of α1, α6, β2, β3, γ2, and δ subunits (30).

GLO1 inhibition reduces anxiety-like behavior. Finally, we evaluated the suitability of GLO1 as a target for anxiolytic drugs by synthesizing S-bromobenzylglutathione cyclopentyl diester (BrBzGCp2), a previously described GLO1 inhibitor (31, 32). We confirmed that BrBzGCp2 inhibited GLO1 enzymatic activity in vitro (Figure 7A). We then administered BrBzGCp2 (30 mg/kg) by i.p. injection to WT B6 mice and allowed MG levels to accumulate for 2 hours (ref. 10 and Figure 7B). GLO1 inhibition by BrBzGCp2 increased center time in the OF test (Figure 7C), without changing distance traveled (Figure 7D). These data suggest that GLO1 inhibition increased MG concentration, thus reducing anxiety-like behavior. We obtained similar results in CD-1 mice treated with BrBzGCp2 (Supplemental Figure 10).

Figure 7

Pharmacological inhibition of GLO1 reduces anxiety-like behavior. (A) Enzymatic activity of 10 ng purified GLO1 protein treated with vehicle or 100 μM BrBzGCp2 (n = 3 vehicle and 4 BrBzGCp2). (B) HPLC measurement of MG concentration in whole brain 2 hours after i.p. treatment with vehicle or BrBzGSHCp2. Assay order had a significant effect on MG concentration and was used as a covariate in a 1-way ANCOVA for the factor treatment (n = 8 per group). P < 0.0001. (C and D) B6 mice were treated with i.p. vehicle or 30 mg/kg BrBzGCp2 and then tested in the OF test 2 hours after injection (n = 25 vehicle and 24 BrBzGCp2). (C) Center time. (D) Total distance. mU, milliunits (μmol/min). Data are mean ± sem. *P < 0.05, **P < 0.0001.

Finally, we investigated whether increasing MG concentration can rescue the anxiogenic effect of Glo1 overexpression. In CD-1 mice, a naturally occurring Glo1 duplication increases Glo1 expression and anxiety-like behavior (ref. 2 and Supplemental Figure 11). Treatment with exogenous MG (50 mg/kg) reduced anxiety-like behavior in mice with the Glo1 duplication (Supplemental Figure 11A), as did treatment with BrBzGCp2 (Supplemental Figure 11B). Together, these results indicate that increasing MG concentration reverses the anxiogenic effect of Glo1 overexpression. Further, they illustrate the therapeutic potential of GLO1 inhibitors for the treatment of anxiety disorders.

In the present study, we investigated the effect of GLO1 on anxiety and elucidated the underlying molecular mechanism. GLO1 had no obvious mechanistic relationship to anxiety. Rather, GLO1 was best characterized for detoxifying MG, a cytotoxic by-product of glycolysis (8, 9). Using BAC Tg mice, we demonstrated that Glo1 overexpression increased anxiety-like behavior. Tg mice had decreased MG levels in the brain, and administration of MG reduced anxiety-like behavior, suggesting that MG is an endogenous anxiolytic agent.

We next characterized MG as an endogenous partial agonist at GABA_A receptors, which are well-established mediators of anxiety (26). Physiologically relevant concentrations of MG elicited electrophysiological effects characteristic of GABA_A receptor activation in CGNs and HNs (Figures 5 and 6). MG-evoked currents were approximately one-third the magnitude of those elicited by GABA (Figure 5B and Figure 6A), and coapplication with MG reduced the magnitude of GABA-evoked currents (Figure 5E and Figure 6A). These data implicate MG as a partial competitive agonist at GABA_A receptors. While we did not identify the specific GABA_A receptor subtypes activated by MG, potent activation of currents in CGNs suggests that MG operates receptors containing α6 subunits, which are prevalent in CGNs (33). Further, MG-evoked currents were enhanced by coapplication of the benzodiazepines diazepam and midazolam as well as by zolpidem (Figure 6B). Diazepam and midazolam operate GABA_A receptors that contain α1, α2, α3, and α5 subunits, while zolpidem targets receptors with α1 and γ2 subunits. Taken together, these data suggest that MG acts at a broad range of GABA_A receptors, likely acting on neurons throughout the CNS. Future studies will be critical in further elucidating the pharmacological and biophysical responses of specific GABA_A receptor compositions to MG.

MG is generated intracellularly by both neurons and glia and then crossed the PM (Figure 5F) to activate GABA_A receptors in a paracrine manner. Thus, MG is expected to activate synaptic GABA_A receptors, which regulate phasic inhibitory signaling, as well as somatic, dendritic, and axonal GABA_A receptors, which contribute to tonic inhibitory signaling (34). Tonic inhibition via GABA_A receptors decreases cellular input resistance, which diminishes the propagation of excitatory currents and decreases the probability that an action potential will be initiated by excitatory inputs (35). Many therapeutic agents, including those for anxiety disorders, target extrasynaptic GABA_A receptors with the aim of modulating tonic inhibition (36). We speculate that the actions of MG at extrasynaptic GABA_A receptors mediate tonic inhibition and anxiety. Indeed, midazolam, which is known to increase tonic conductances (37), enhanced MG-evoked currents (Figure 6B).

The coordinated release of synaptic vesicles ensures that GABA concentrations in the synaptic cleft peak in the millimolar range (38). In contrast, concentrations of GABA in the extracellular space are in the submicromolar range (39). The concentration of MG in the brain was approximately 5 μM (Supplemental Table 5), reflecting concentrations both at the synaptic cleft and in the extracellular space. Therefore, future studies must directly measure the effect of MG on extrasynaptic receptors in order to determine the relative contribution of MG to phasic and tonic inhibitory activity in the CNS. In particular, measuring inhibitory postsynaptic currents in slice preparations may further elucidate the role of MG in neural physiology.

We provided evidence that physiological concentrations of MG produce robust electrophysiological effects in vitro. A concentration of 5 μM MG was on the steep segment of the concentration-response curve, at which a small change in concentration elicits large changes in current and membrane potential (Figure 5, A and B, and Figure 6A). Accordingly, we showed that small changes in MG concentration in vivo substantially altered behavior. Tg mice exhibited less than a 10% decrease in MG concentration (Figure 3C) but exhibited a marked anxiety-like phenotype, spending up to 30% less time in the center of the OF than did WT mice (Figure 2A). Similarly, exogenous administration of MG (50 mg/kg) increased MG concentration by only 15% (Figure 3D and Supplemental Table 5) but increased center time in the OF by almost 30% (Figure 3E). Thus, GABA_A receptors and downstream behaviors are exquisitely sensitive to relatively small changes in MG concentration, consistent with an important role for MG under normal physiological conditions. Finally, because MG levels increase under conditions of high metabolic load (8, 40), our findings suggest a mechanism whereby metabolic state regulates neuronal inhibitory tone and behavior.

To our knowledge, activation of GABA_A receptors represents a novel physiological role for MG. Previous studies have established the role of MG in AGE formation and cytotoxicity as well as diseases in which these effects are relevant to pathogenesis, such as diabetic complications, cancer, and aging (40–44). Because supraphysiological levels of MG (100–1,000 μM) induce AGEs, reactive oxygen species, and apoptosis in neurons (21, 45, 46), these effects have been thought to underlie the role of MG in CNS disorders, including anxiety. For instance, chronic intracerebroventricular administration of MG was reported to reduce anxiety-like behavior, coincident with AGE accumulation (47). In contrast, our data demonstrate that physiological levels of MG have a selective effect on GABA_A receptors and elicit behavioral effects within minutes, suggesting a novel role for MG in the CNS that is independent of cytotoxicity.

Our findings put the previous controversy regarding the role of GLO1 in anxiety-like behavior into context. Hovatta et al. reported a positive correlation between Glo1 expression and anxiety-like behavior among inbred mice and proposed a causal role for GLO1 in anxiety (3). Our previous work identified a CNV in mice that causes a duplication of Glo1 and was associated with increased Glo1 expression and anxiety-like behavior (2). This CNV represents the molecular polymorphism likely responsible for the prevalence of differential Glo1 expression among inbred mice. However, discrepant findings have cast doubt on the association between GLO1 and increased anxiety. In particular, Kromer et al. generated 2 inbred mouse lines selected for high and low anxiety-like behavior; the low anxiety line showed greater Glo1 expression than the high anxiety line (4). Results from these selected lines are directionally opposite of results from Hovatta et al. and Williams et al. (2, 3). In this study, we generated BAC Tg mice with increased copies of Glo1 to model the naturally occurring CNV. To isolate the effect of Glo1, Tg mice overexpressed Glo1 but not the other 3 genes in the CNV. Using these Tg mice, we provided direct evidence that Glo1 copy number regulates Glo1 expression and anxiety. This effect was strain-independent, as Glo1 was anxiogenic on both B6 and FVB backgrounds. We note that, of the 3 B6 Tg lines, only the 2 with the highest Glo1 expression exhibited increased anxiety-like behavior (Figure 2A). FVB Tg lines showed a similar pattern (Supplemental Figure 3C), suggesting that the anxiogenic effect of GLO1 is dose-dependent. Taken together, our data indicate that increased Glo1 copy number increases Glo1 expression and anxiety-like behavior. This study helps resolve the apparent discrepancy in the literature: it appears that the bidirectionally selected and subsequently inbred lines of Kromer et al. (4) differentially fixed the Glo1 CNV. The low anxiety line carried the duplication, while the high anxiety line did not (47). The difference in Glo1 expression is likely offset by numerous other alleles that collectively underlie the response to selection. Our data unequivocally support the hypothesis that increased Glo1 expression causes increased anxiety-like behavior and are strengthened by the use of isogenic backgrounds.

The present study may also provide mechanistic insight into GLO1’s proposed associations with other CNS diseases, such as autism (48, 49), affective disorders (50, 51), panic disorder (52), and schizophrenia (53). Human genome-wide association studies have identified an association between restless legs syndrome (RLS) and a haplotype containing GLO1 (54, 55). Our results suggest that GLO1’s role in GABAergic signaling could underlie its association with RLS. Indeed, GABA_A receptor agonists are used to treat RLS (56). Finally, we established that GLO1 is a target for the treatment of anxiety disorders and demonstrated the anxiolytic effect of a GLO1 inhibitor. Similar drugs may be useful in the treatment of anxiety and other diseases linked to GABA signaling, including epilepsy and sleep disorders (57, 58).

View Supplemental data

This work was funded by NIH grants R01MH079103 and R01DA021336 and The University of Chicago Diabetes Research and Training Center grant P60 DK020595 (to A.A. Palmer). M.G. Distler was supported by NIH grant T32GM07281. L.D. Plant was supported by NIH grant R01NS058505 (awarded to S. Goldstein). We thank Linda Degenstein for pronuclear injection, Peter Pytel for histological analysis, the Human Tissue Resource Center at The University of Chicago for performing TUNEL staining, and Stephanie Dulawa, Steve Goldstein, Jason MacLean, Kathy Millen, Gopal Thinakaran, William Zeiger, and Xiaoxi Zhuang for their scientific advice and feedback.

Address correspondence to: Abraham A. Palmer, 920 E. 58th St., CLSC 507D, Chicago, Illinois 60637, USA. Phone: 773.834.2897; Fax: 773.834.0505; E-mail: aap@uchicago.edu.

Conflict of interest: The authors have declared that no conflict of interest exists.

Reference information: J Clin Invest. 2012;122(6):2306–2315. doi:10.1172/JCI61319.

1. Bystritsky A. Treatment-resistant anxiety disorders. Mol Psychiatry. 2006;11(9):805–814.
   View this article via: PubMed CrossRef Google Scholar
2. Williams Rt, et al. A common and unstable copy number variant is associated with differences in Glo1 expression and anxiety-like behavior. PLoS One. 2009;4(3):e4649.
   View this article via: PubMed CrossRef Google Scholar
3. Hovatta I, et al. Glyoxalase 1 and glutathione reductase 1 regulate anxiety in mice. Nature. 2005;438(7068):662–666.
   View this article via: PubMed CrossRef Google Scholar
4. Kromer SA, et al. Identification of glyoxalase-I as a protein marker in a mouse model of extremes in trait anxiety. J Neurosci. 2005;25(17):4375–4384.
   View this article via: PubMed CrossRef Google Scholar
5. Thornalley PJ. Unease on the role of glyoxalase 1 in high-anxiety-related behaviour. Trends Mol Med. 2006;12(5):195–199.
   View this article via: PubMed CrossRef Google Scholar
6. Heintz N. BAC to the future: the use of bac transgenic mice for neuroscience research. Nat Rev Neurosci. 2001;2(12):861–870.
   View this article via: PubMed CrossRef Google Scholar
7. Chandler KJ, Chandler RL, Broeckelmann EM, Hou Y, Southard-Smith EM, Mortlock DP. Relevance of BAC transgene copy number in mice: transgene copy number variation across multiple transgenic lines and correlations with transgene integrity and expression. Mamm Genome. 2007;18(10):693–708.
   View this article via: PubMed CrossRef Google Scholar
8. Thornalley PJ. Glyoxalase I--structure, function and a critical role in the enzymatic defence against glycation. Biochem Soc Trans. 2003;31(pt 6):1343–1348.
   View this article via: PubMed CrossRef Google Scholar
9. Thornalley PJ. The glyoxalase system in health and disease. Mol Aspects Med. 1993;14(4):287–371.
   View this article via: PubMed CrossRef Google Scholar
10. Kuhla B, et al. Pathological effects of glyoxalase I inhibition in SH-SY5Y neuroblastoma cells. J Neurosci Res. 2006;83(8):1591–1600.
    View this article via: PubMed CrossRef Google Scholar
11. Shinohara M, et al. Overexpression of glyoxalase-I in bovine endothelial cells inhibits intracellular advanced glycation endproduct formation and prevents hyperglycemia-induced increases in macromolecular endocytosis. J Clin Invest. 1998;101(5):1142–1147.
    View this article via: JCI PubMed CrossRef Google Scholar
12. Kalapos MP. The tandem of free radicals and methylglyoxal. Chem Biol Interact. 2008;171(3):251–271.
    View this article via: PubMed CrossRef Google Scholar
13. Copeland NG, Jenkins NA, Court DL. Recombineering: a powerful new tool for mouse functional genomics. Nat Rev Genet. 2001;2(10):769–779.
    View this article via: PubMed Google Scholar
14. Belzung C, Griebel G. Measuring normal and pathological anxiety-like behaviour in mice: a review. Behav Brain Res. 2001;125(1–2):141–149.
    View this article via: PubMed CrossRef Google Scholar
15. Hohoff C. Anxiety in mice and men: a comparison. J Neural Transm. 2009;116(6):679–687.
    View this article via: PubMed CrossRef Google Scholar
16. LeDoux JE. Emotion circuits in the brain. Annu Rev Neurosci. 2000;23:155–184.
    View this article via: PubMed CrossRef Google Scholar
17. Bourin M, Petit-Demouliere B, Dhonnchadha BN, Hascoet M. Animal models of anxiety in mice. Fundam Clin Pharmacol. 2007;21(6):567–574.
    View this article via: PubMed CrossRef Google Scholar
18. Bourin M, Hascoet M. The mouse light/dark box test. Eur J Pharmacol. 2003;463(1–3):55–65.
    View this article via: PubMed CrossRef Google Scholar
19. Dawson GR, Tricklebank MD. Use of the elevated plus maze in the search for novel anxiolytic agents. Trends Pharmacol Sci. 1995;16(2):33–36.
    View this article via: PubMed CrossRef Google Scholar
20. Bouwknecht JA, Paylor R. Pitfalls in the interpretation of genetic and pharmacological effects on anxiety-like behaviour in rodents. Behav Pharmacol. 2008;19(5–6):385–402.
    View this article via: PubMed CrossRef Google Scholar
21. Di Loreto S, Zimmitti V, Sebastiani P, Cervelli C, Falone S, Amicarelli F. Methylglyoxal causes strong weakening of detoxifying capacity and apoptotic cell death in rat hippocampal neurons. Int J Biochem Cell Biol. 2008;40(2):245–257.
    View this article via: PubMed CrossRef Google Scholar
22. Carter LP, Koek W, France CP. Behavioral analyses of GHB: receptor mechanisms. Pharmacol Ther. 2009;121(1):100–114.
    View this article via: PubMed CrossRef Google Scholar
23. Kumar S, et al. The role of GABA(A) receptors in the acute and chronic effects of ethanol: a decade of progress. Psychopharmacology (Berl). 2009;205(4):529–564.
    View this article via: PubMed CrossRef Google Scholar
24. Rudolph U, Mohler H. Analysis of GABAA receptor function and dissection of the pharmacology of benzodiazepines and general anesthetics through mouse genetics. Annu Rev Pharmacol Toxicol. 2004;44:475–498.
    View this article via: PubMed CrossRef Google Scholar
25. Sieghart W. Structure and pharmacology of gamma-aminobutyric acidA receptor subtypes. Pharmacol Rev. 1995;47(2):181–234.
    View this article via: PubMed Google Scholar
26. Kalueff AV, Nutt DJ. Role of GABA in anxiety and depression. Depress Anxiety. 2007;24(7):495–517.
    View this article via: PubMed CrossRef Google Scholar
27. Kent JM, Mathew SJ, Gorman JM. Molecular targets in the treatment of anxiety. Biol Psychiatry. 2002;52(10):1008–1030.
    View this article via: PubMed CrossRef Google Scholar
28. Crestani F, et al. Decreased GABAA-receptor clustering results in enhanced anxiety and a bias for threat cues. Nat Neurosci. 1999;2(9):833–839.
    View this article via: PubMed CrossRef Google Scholar
29. Canteras NS, Resstel LB, Bertoglio LJ, Carobrez Ade P, Guimaraes FS. Neuroanatomy of anxiety. Curr Top Behav Neurosci. 2010;2:77–96.
    View this article via: PubMed Google Scholar
30. Sieghart W, Sperk G. Subunit composition, distribution and function of GABA(A) receptor subtypes. Curr Top Med Chem. 2002;2(8):795–816.
    View this article via: PubMed CrossRef Google Scholar
31. Vince R, Daluge S, Wadd WB. Studies on the inhibition of glyoxalase I by S-substituted glutathiones. J Med Chem. 1971;14(5):402–404.
    View this article via: PubMed CrossRef Google Scholar
32. Thornalley PJ, Ladan MJ, Ridgway SJ, Kang Y. Antitumor activity of S-(p-bromobenzyl)glutathione diesters in vitro: a structure-activity study. J Med Chem. 1996;39(17):3409–3411.
    View this article via: PubMed CrossRef Google Scholar
33. Nusser Z, et al. Alterations in the expression of GABAA receptor subunits in cerebellar granule cells after the disruption of the alpha6 subunit gene. Eur J Neurosci. 1999;11(5):1685–1697.
    View this article via: PubMed CrossRef Google Scholar
34. Hablitz JJ, Mathew SS, Pozzo-Miller L. GABA vesicles at synapses: are there 2 distinct pools? Neuroscientist. 2009;15(3):218–224.
    View this article via: PubMed CrossRef Google Scholar
35. Kullmann DM, Ruiz A, Rusakov DM, Scott R, Semyanov A, Walker MC. Presynaptic, extrasynaptic and axonal GABAA receptors in the CNS: where and why? Prog Biophys Mol Biol. 2005;87(1):33–46.
    View this article via: PubMed CrossRef Google Scholar
36. Brickley SG, Mody I. Extrasynaptic GABA(A) receptors: their function in the CNS and implications for disease. Neuron. 2012;73(1):23–34.
    View this article via: PubMed CrossRef Google Scholar
37. Yeung JY, Canning KJ, Zhu G, Pennefather P, MacDonald JF, Orser BA. Tonically activated GABAA receptors in hippocampal neurons are high-affinity, low-conductance sensors for extracellular GABA. Mol Pharmacol. 2003;63(1):2–8.
    View this article via: PubMed CrossRef Google Scholar
38. Farrant M, Nusser Z. Variations on an inhibitory theme: phasic and tonic activation of GABA(A) receptors. Nat Rev Neurosci. 2005;6(3):215–229.
    View this article via: PubMed CrossRef Google Scholar
39. Vithlani M, Terunuma M, Moss SJ. The dynamic modulation of GABA(A) receptor trafficking and its role in regulating the plasticity of inhibitory synapses. Physiol Rev. 2010;91(3):1009–1022.
    View this article via: PubMed CrossRef Google Scholar
40. Brownlee M. Biochemistry and molecular cell biology of diabetic complications. Nature. 2001;414(6865):813–820.
    View this article via: PubMed CrossRef Google Scholar
41. Ahmed N, Thornalley PJ. Advanced glycation endproducts: what is their relevance to diabetic complications? Diabetes Obes Metab. 2007;9(3):233–245.
    View this article via: PubMed CrossRef Google Scholar
42. Fleming TH, Humpert PM, Nawroth PP, Bierhaus A. Reactive metabolites and AGE/RAGE-mediated cellular dysfunction affect the aging process: a mini-review. Gerontology. 2011;57(5):435–443.
    View this article via: PubMed Google Scholar
43. Morcos M, et al. Glyoxalase-1 prevents mitochondrial protein modification and enhances lifespan in Caenorhabditis elegans. Aging Cell. 2008;7(2):260–269.
    View this article via: PubMed CrossRef Google Scholar
44. Thornalley PJ. Protecting the genome: defence against nucleotide glycation and emerging role of glyoxalase I overexpression in multidrug resistance in cancer chemotherapy. Biochem Soc Trans. 2003;31(pt 6):1372–1377.
    View this article via: PubMed CrossRef Google Scholar
45. Di Loreto S, Caracciolo V, Colafarina S, Sebastiani P, Gasbarri A, Amicarelli F. Methylglyoxal induces oxidative stress-dependent cell injury and up-regulation of interleukin-1beta and nerve growth factor in cultured hippocampal neuronal cells. Brain Res. 2004;1006(2):157–167.
    View this article via: PubMed CrossRef Google Scholar
46. Li G, Chang M, Jiang H, Xie H, Dong Z, Hu L. Proteomics analysis of methylglyoxal-induced neurotoxic effects in SH-SY5Y cells. Cell Biochem Funct. 2011;29(1):30–35.
    View this article via: PubMed CrossRef Google Scholar
47. Hambsch B, et al. Methylglyoxal-mediated anxiolysis involves increased protein modification and elevated expression of glyoxalase 1 in the brain. J Neurochem. 2010;113(5):1240–1251.
    View this article via: PubMed Google Scholar
48. Junaid MA, Kowal D, Barua M, Pullarkat PS, Sklower Brooks S, Pullarkat RK. Proteomic studies identified a single nucleotide polymorphism in glyoxalase I as autism susceptibility factor. Am J Med Genet A. 2004;131(1):11–17.
    View this article via: PubMed Google Scholar
49. Barua M, Jenkins EC, Chen W, Kuizon S, Pullarkat RK, Junaid MA. Glyoxalase I polymorphism rs2736654 causing the Ala111Glu substitution modulates enzyme activity—implications for autism. Autism Res. 2011;4(4):262–270.
    View this article via: PubMed CrossRef Google Scholar
50. Fujimoto M, et al. Reduced expression of glyoxalase-1 mRNA in mood disorder patients. Neurosci Lett. 2008;438(2):196–199.
    View this article via: PubMed CrossRef Google Scholar
51. Benton CS, et al. Evaluating genetic markers and neurobiochemical analytes for fluoxetine response using a panel of mouse inbred strains [published online ahead of print November 24, 2011]. Psychopharmacology (Berl). doi: 10.1007/s00213-011-2574-z .
52. Politi P, Minoretti P, Falcone C, Martinelli V, Emanuele E. Association analysis of the functional Ala111Glu polymorphism of the glyoxalase I gene in panic disorder. Neurosci Lett. 2006;396(2):163–166.
    View this article via: PubMed CrossRef Google Scholar
53. Arai M, et al. Enhanced carbonyl stress in a subpopulation of schizophrenia. Arch Gen Psychiatry. 2010;67(6):589–597.
    View this article via: PubMed CrossRef Google Scholar
54. Kemlink D, et al. Replication of restless legs syndrome loci in three European populations. J Med Genet. 2009;46(5):315–318.
    View this article via: PubMed CrossRef Google Scholar
55. Stefansson H, et al. A genetic risk factor for periodic limb movements in sleep. N Engl J Med. 2007;357(7):639–647.
    View this article via: PubMed CrossRef Google Scholar
56. Trenkwalder C, Paulus W. Restless legs syndrome: pathophysiology, clinical presentation and management. Nat Rev Neurol. 2010;6(6):337–346.
    View this article via: PubMed CrossRef Google Scholar
57. Treiman DM. GABAergic mechanisms in epilepsy. Epilepsia. 2001;42 suppl 3: :8–12.
    View this article via: PubMed CrossRef Google Scholar
58. Winsky-Sommerer R. Role of GABAA receptors in the physiology and pharmacology of sleep. Eur J Neurosci. 2009;29(9):1779–1794.
    View this article via: PubMed CrossRef Google Scholar
59. Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc. 2008;3(6):1101–1108.
    View this article via: PubMed CrossRef Google Scholar
60. Allen RE, Lo TW, Thornalley PJ. A simplified method for the purification of human red blood cell glyoxalase. I. Characteristics, immunoblotting, and inhibitor studies. J Protein Chem. 1993;12(2):111–119.
    View this article via: PubMed CrossRef Google Scholar
61. Dhar A, Desai K, Liu J, Wu L. Methylglyoxal, protein binding and biological samples: are we getting the true measure? J Chromatogr B Analyt Technol Biomed Life Sci. 2009;877(11–12):1093–1100.
    View this article via: PubMed Google Scholar
62. Plant LD, et al. Amyloid beta peptide as a physiological modulator of neuronal ‘A’-type K+ current. Neurobiol Aging. 2006;27(11):1673–1683.
    View this article via: PubMed CrossRef Google Scholar
63. Bax L, Yu LM, Ikeda N, Tsuruta H, Moons KG. Development and validation of MIX: Comprehensive free software for meta-analysis of causal research data. BMC Med Res Methodol. 2006;6:50.
    View this article via: PubMed CrossRef Google Scholar
64. Storey JD, Tibshirani R. Statistical significance for genomewide studies. Proc Natl Acad Sci U S A. 2003;100(16):9440–9445.
    View this article via: PubMed CrossRef Google Scholar