ANESTHETICS have been used in surgical procedures for well over a hundred years, but the molecular mechanism that underlies inhalational anesthesia is still poorly understood. Ion channels and receptors at synapses of the central nervous system (CNS) have been highlighted as potential targets for inhaled anesthetics.^1–5  In the CNS, these ion channels and receptors are linked to their downstream signaling pathways through postsynaptic density protein-95, Drosophila disc large tumor suppressor, and zonula occludens-1 (PDZ) domain–mediated protein–protein interactions. The PDZ domain can recognize and bind to specific amino acid sequences, including those at the C-termini of receptors and ion channels.^6–8  Our previous studies have shown that knockdown of PDZ domain–containing scaffolding protein postsynaptic density protein-95 (PSD-95) facilitates isoflurane anesthesia⁹  and inhibits chronic pain behaviors.^10,11  PSD-95, a member of the membrane-associated guanylate kinase family of proteins that assemble protein complexes at synapses, is composed of three tandem PDZ domains (PDZ1–3) at the N-terminus and guanylate kinase and Src homology 3 domains.^12–14  The first two PDZ domains of PSD-95 can bind the NR2A and NR2B subunits of N-methyl-d-aspartate (NMDA) receptors, neuronal nitric oxide synthase, Shaker-type potassium channel Kv1.4, and Src family tyrosine kinase.^7,15–17  Furthermore, we have demonstrated that PDZ domain–mediated interactions between PSD-95 and NMDA receptors or neuronal nitric oxide synthase are disrupted by clinically relevant concentrations of inhaled anesthetic halothane.¹⁸  By injecting cell-permeable peptide Tat-PSD-95 PDZ2 (comprising the second PDZ domain of PSD-95) intraperitoneally into mice, we also showed that disrupting PSD-95 PDZ domain–mediated protein–protein interactions reduces the threshold for halothane anesthesia¹⁹  and diminishes chronic inflammatory pain.²⁰  Taken together, these results suggest that the PDZ domain at synapses of the CNS may be an important molecular target for inhaled anesthetics and that alteration of PDZ domain–mediated protein–protein interactions may contribute to the molecular mechanism of inhaled anesthesia. However, it is unclear whether and how inhaled anesthetics interact with different PDZ domains in the CNS.

In the current study, we used nuclear magnetic resonance (NMR) and other state-of-the-art techniques to examine in greater detail the nature of the anesthetic interaction with the PDZ protein–binding domains.

Glutathione S-transferase (GST) and GST fusion proteins (GST-PSD-95 PDZ2 and GST-Kv1.4 C-terminus 100) were prepared in BL21 (Novagen, Germany) with glutathione sepharose 4B (Amersham Pharmacia Biotech Inc., USA) as an affinity resin for purification. All animal experiments were conducted with the approval of the Animal Care and Use Committee at Johns Hopkins University, Baltimore, Maryland, and were consistent with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Membrane-bound proteins were extracted from mouse cerebral cortex and solubilized. Mouse cerebral cortex was lysed in homogenizing buffer (containing 10 mM Tris-HCl [pH 8.0], 5 mM MgCl₂, 250 mM sucrose, 2 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol, 1 mM phenylmethanesulfonylfluoride, 10 µg/ml leupeptin, 10 µg/ml aprotinin, and 20 µg/ml pepstatin A) with a rotor-stator Polytron homogenizer (Brinkmann Instruments, USA). The homogenized solution was centrifuged at 900g for 15 min, and the supernatant was removed and centrifuged again at 37,000g for 40 min. The precipitate was solubilized with resuspension buffer (containing 500 mM Tris-HCl, 1% [vol/vol] TritonX-100, and 1% [wt/vol] sodium deoxycholate) and centrifuged at 37,000g for 20 min. The supernatant contained the membrane-bound proteins. For binding experiments, glutathione sepharose 4B (40 µl) was preincubated with 1 nmol of GST or GST fusion protein for 30 min and then the membrane-bound proteins (400 µg) were mixed with the GST fusion protein at room temperature for 1 h in the presence of different concentrations of halothane delivered in an incubation chamber (described under Yeast Two-hybrid Analysis). After the resin was washed five times with washing buffer (500 mM NaCl, 0.1% Triton X-100, 1 mM phenylmethanesulfonylfluoride, 10 µg/ml leupeptin, 10 µg/ml aprotinin, and 20 µg/ml pepstatin A in phosphate-buffered saline), the bound proteins were eluted in sodium dodecyl sulfate polyacrylamide gel electrophoresis sample buffer, separated by electrophoresis, and detected by immunoblotting with anti-PSD-95 (Chemicon, USA), anti-Kv1.4 (Chemicon), and anti-GST (Santa Cruz Biotechnology Inc., USA) antibodies.

The membrane-bound proteins were prepared as described above. The affinity-purified rabbit anti-Kv1.4 antibody (4 μg) was incubated with 100 μl of protein A–sepharose slurry for 1 h. The solubilized membrane proteins (400 μg) were then added to the sepharose beads, and the mixture was incubated with different concentrations of isoflurane for 2 h at room temperature. The mixture was washed once with 1% Triton X-100 in immunoprecipitation buffer (containing 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na₂HPO₄, 1.4 mM KH₂PO₄, 5 mM EGTA, 1 mM sodium vanadate, 10 mM sodium pyrophosphate, 50 mM NaF, and 0.1 mM phenylmethylsulfonyl fluoride plus 20 U/ml Trasylol), twice with 1% Triton X-100 in immunoprecipitation buffer plus 300 mM NaCl, and three times with immunoprecipitation buffer. The bound proteins were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis and detected by immunoblotting with anti-PSD-95 and anti-Kv1.4 antibodies. As a positive control (input), 40 μg of the solubilized membrane proteins were loaded onto the gel. The specificity of the anti-Kv1.4 antibody was verified by preincubation with Kv1.4 fusion peptide.

The two partners in each interaction were subcloned into EcoRI and BamHI sites of the yeast vectors pGADT7 and pGBKT7 (Clontech Laboratories Inc., USA), respectively. The yeast reporter strain AH109 (Clontech Laboratories Inc.) was transformed with the pGBT9 and pGAD424 plasmids. Protein–protein interactions were evaluated as described in our previous study¹⁸  by growth of yeast on selective agar plates sealed in incubator chambers containing different concentrations of isoflurane, which was introduced into the chambers by its conventional calibrated anesthesia vaporizer (OHIO Medical Co., USA) for 30 min of equilibrium with air at 3 l/min. Anesthetic concentration within the chamber was determined with a Capnomac Ultima analyzer (Datex-Engstrom, Finland), and the yeast medium/gas partition coefficient for isoflurane at 30°C was determined by gas chromatography (Hewlett Packard 5890 series II plus; Agilent Technologies, Inc., USA) according to the method described previously.^18,21  Concentration of halothane or isoflurane in the medium was calculated in millimolar units. The amount of yeast growth was quantified by measuring the average intensity of the yeast on the selective agar plates with Image-J software (National Institute of Mental Health, USA).

The PDZ123 domain of Rat source PSD-95 protein (residues 61 to 393,²²  PDZ1 61 to 151, PDZ2 155 to 247, PDZ3 302 to 393) was incorporated into the pET28 vector and expressed within Escherichia coli. The N-terminus of the protein was extended with His Tag for facilitating the Ni affinity column purification. The ¹⁵N-labeled sample was at concentration of 5 to 6.8 mg/ml or 125 to 170 μM in elution buffer (containing 50 mM NaH₂PO₄/Na₂HPO₄ [pH 8.0], 300 mM NaCl, and 250 mM imidazole). The samples were aliquoted approximately 1 ml and saved at −80°C. For NMR experiments, the aliquoted sample was exchanged to NMR buffer (containing 50 mM NaH₂PO₄/Na₂HPO₄ [pH6.7] and 50 mM NaCl) through diafiltration; the protein concentration was approximately 100 μM. The peptides composed of the last 20 amino acids from the C-terminal of NR2A (LNSCSNRRVYKKMPSIESDV) and NR2B (FNGSSNGHVYEKLSSIESDV) were synthesized and purified by United Peptide Corporation as lyophilized powder. The peptides were dissolved with NMR buffer to 10 mM (approximately 2.5% w/v) in clean glass vial and visually checked for solubility. The reported binding affinity for PDZ domain C-terminal peptide normally ranges from 10⁻⁸ to 10⁻⁶ M.²³  A peptide containing the C-terminal nine amino acid residues of NR2B binds PDZ1, PDZ2, and PDZ3 of SAP102 (a PSD-95 family protein) with EC₅₀ values of 30 nM, 6 nM, and 1 μM, respectively.²⁴  In our experiments, the peptides were titrated up to 310 μM. The isoflurane concentration was calibrated by ¹⁹F NMR with an external reference of trifluoroacetic acid; 20 μl D₂O (deuterium oxide) and 2 μl 100 mM DSS (4,4-dimethyl-4-silapentane-1-sulfonic acid) were added to each sample (approximately 500 μl) for NMR signal locking and reference, respectively. NMR spectra were recorded at 298K on a Bruker-BioSpin Avance 600 spectrometer (Bruker, USA) using a cryoprobe. The chemical shift perturbation was quantified as the combined ¹H and ¹⁵N chemical shifts change with the following equation:

The synthetic peptide corresponding to the C-terminal tail of NR2B was synthesized (Invitrogen, USA) with an additional cysteine residue for thiol coupling to a biosensor CM5 chip (BIAcore, Sweden).^18,25  The peptide was immobilized to the biosensor CM5 chip surface by using the standard maleimide coupling with m-maleimidobutyryloxy-sulfosuccinimide ester (sulfo-GMBS [N-maleimidobutryloxysulfosuccinimide ester]). First, the matrix surface was activated by injecting a 1:1 mixture of 0.1 M N-hydroxysuccinimide and 0.4 M 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide at a flow rate of 10 µl/min at 25°C for 10 min. Second, amine groups were introduced with 10 M ethylenediamine solubilized in 0.1 M sodium borate (pH 8.5) for 10 min. Third, maleimide groups were introduced with 50 mM sulfo-GMBS solubilized in 0.1 M sodium borate (pH 8.5) for 20 min and then 100 µl of NR2B peptide (100 µM) was injected and immobilized onto the matrix. The remaining active sites of the matrix were blocked with 50 mM cysteine. PSD-95 PDZ2 was subcloned into pET28a (Stratagene, USA) and expressed in BL21. Expressed proteins were purified with nickel NTA agarose (Qiagen, USA). The interaction between PSD-95 PDZ2 and NR2B C-terminal peptide was analyzed on a BIAcore 2000 system (BIAcore). PSD-95 PDZ2 (20 µM) was injected with different concentrations of isoflurane over the NR2B peptide–coupled surface at a flow rate of 5 µl/min. Anesthetic concentrations were determined by head-space sampling from an aliquot of the anesthetic-containing solution. Sensorgrams were recorded and normalized by subtracting the baseline resonance unit value. The baseline sensorgram was obtained by injecting PSD-95 PDZ2 over a nonprotein, blocked surface. Between successive measurements, the surface was regenerated with 50 mM phosphoric acid (3 min contact time). The analysis of kinetic parameters was performed with BIAevaluation version 3.0 software according to the manufacturer’s instructions.

Data from the yeast two-hybrid experiments were expressed as means ± SD and statistically analyzed with SigmaStat version 3.1 software (Systat Software Inc., USA) by using one-way ANOVA followed by Student–Newman–Keul method. The nature of hypothesis testing is two tailed. The sample sizes were selected based on our previous study.¹⁸  Statistical significance was set at P value less than 0.05.

Our previous studies showed that PDZ domain–mediated protein–protein interactions between PSD-95 and NMDA receptors or neuronal nitric oxide synthase are disrupted by clinically relevant concentrations of inhaled anesthetics.¹⁸  In this study, we used an yeast two-hybrid approach to investigate whether halothane and isoflurane also disrupt PDZ domain–mediated interactions between PSD-95 and potassium channel Kv1.4. In the absence of anesthetic, the second PDZ domain of PSD-95 (PSD-95 PDZ2) interacted with the C-terminal tail of Kv1.4 (Kv1.4 CT100), as evidenced by the growth of the yeast cells harboring both pGADT7-PSD-95 PDZ2 and pGBKT7-Kv1.4 CT100 fusion protein plasmids in synthetic dropout agar plates lacking adenine, leucine, tryptophan, and histidine (fig. 1). However, the yeast cell growth was slowed by halothane (fig. 1, A and B) and isoflurane (fig. 1, C−E) in a dose-dependent manner, indicating that these anesthetics dose-dependently inhibit PDZ domain–mediated protein–protein interactions between PSD-95 and Kv1.4. Compared with isoflurane, halothane disrupted the interactions with a stronger potency. A high but still clinically relevant concentration of halothane (0.95 mM) completely blocked the yeast cell growth (fig. 1, A and B). We verified that halothane itself was not cytotoxic to yeast cells by growing the same strain on low-stringency media (synthetic dropout agar plates that lack leucine and tryptophan and are selective for the yeast cells cotransformed with pGADT7 and pGBKT7) in the presence of the same concentrations of halothane (fig. 1A); halothane did not inhibit the growth of the control yeast cells.

Next, we used GST pull-down and coimmunoprecipitation assays to further confirm the yeast two-hybrid results. GST-PSD-95 PDZ2 and GST-Kv1.4 CT100, but not GST alone, precipitated Kv1.4 and PSD-95 (fig. 2A), in the absence of halothane. However, halothane dose-dependently inhibited the binding of Kv1.4 to GST-PSD-95 PDZ2 and the binding of PSD-95 to GST-Kv1.4 CT100 (fig. 2A). To determine whether inhaled anesthetics disrupt PDZ domain–mediated protein–protein interactions in a physiological setting, we used a coimmunoprecipitation assay to detect in vivo binding of PSD-95 to Kv1.4. We found that immunoprecipitation of Kv1.4 by its specific antibody resulted in coprecipitation of PSD-95 under normal conditions but that isoflurane dose-dependently inhibited the coprecipitation of PSD-95 (fig. 2B). These results suggest that inhaled anesthetics may disrupt the physiological complex of PSD-95 and Kv1.4 in the CNS.

To test whether inhaled anesthetics also disrupt AMPA receptor–involved PDZ domain–mediated protein–protein interactions, we investigated the effects of halothane and isoflurane on interactions between the AMPA receptor GluA2 subunit and glutamate receptor–interacting protein or protein interacting with c kinase 1 in an yeast two-hybrid system. Similar to our previous studies¹⁸  and the results described above for PSD-95, halothane and isoflurane dose-dependently inhibited the growth of yeast cells harboring both pGADT7-GluR2 C-terminus and pGBKT7-glutamate receptor–interacting protein PDZ4,5 or pGBKT7-protein interacting with c kinase 1 PDZ fusion protein plasmids (fig. 3). As in our other studies, halothane was more potent that isoflurane at disrupting these PDZ domain–mediated protein–protein interactions (fig. 3).

We also used an yeast two-hybrid approach to determine the effect of isoflurane on the interaction of PSD-95 PDZ2 with NR2B C-terminus (C-terminal 20 amino acids). In the absence of isoflurane, PSD-95 PDZ2 interacted with the C-terminal tail of NMDA receptors NR2B subunit, as evidenced by the growth of the yeast cells harboring both pGADT7-PSD-95 PDZ2 and pGBKT7-NR2B C-terminus fusion protein plasmids in synthetic dropout agar plates lacking adenine, leucine, tryptophan, and histidine (fig. 4A). However, the yeast cell growth was slowed by isoflurane (0.33, 0.66, and 0.99 mM) in a dose-dependent manner (fig. 4, A and B), indicating that isoflurane dose-dependently inhibits PDZ domain–mediated protein–protein interactions between PSD-95 and NR2B. To verify that isoflurane itself is not cytotoxic to yeast cells, we grew the same strain on low-stringency media (synthetic dropout agar plates that lack leucine and tryptophan and are selective for the yeast cells cotransformed with pGADT7 and pGBKT7) in the presence of the same concentrations of isoflurane. It was shown that isoflurane did not inhibit the growth of the control yeast cells.

We further used a surface plasmon resonance–based BIAcore assay to examine the ability of isoflurane to inhibit PSD-95 PDZ domain–mediated protein–protein interactions in real time. Binding of PSD-95 PDZ2 to NR2B C-terminus was measured in real time as an increase in resonance units. In the absence of isoflurane, 10 μM PSD-95 PDZ2 formed a complex with immobilized NR2B C-terminal peptide as indicated by a large increase in resonance unit level (fig. 5). Interestingly, the inclusion of isoflurane in the mobile phase inhibited the equilibrium binding in a dose-dependent manner (fig. 5). These results suggest that isoflurane reduces the number of sites available for the binding between PSD-95-PDZ2 and NR2B C-terminus, thus preventing the formation of the PDZ domain–mediated protein–protein complex.

Different PDZ domain–mediated protein–protein interactions exist in the CNS. To define whether inhaled anesthetics disrupt PDZ domain–mediated interactions between the γ-aminobutyric acid type B (GABA_B) receptor subunit 2 (GABABR2) and GABA_B receptor–interacting proteins PAPIN and Mupp1, we investigated the effects of halothane and isoflurane on these interactions in an yeast two-hybrid system. We used the PDZ interaction between PSD-95 PDZ2 and NR2B as a positive control. Although halothane and isoflurane dose-dependently inhibited the growth of yeast cells harboring both pGADT7-PSD-95 PDZ2 and pGBKT7-NR2B C-terminus fusion protein plasmids (figs. 6 and 7), treatment with the same concentrations of the two inhaled anesthetics had no effect on the growth of yeast cells harboring pGADT7-GABABR2 and pGBKT7-PAPIN PDZ1 or pGBKT7-Mupp1 PDZ13 fusion protein plasmids (figs. 6 and 7). Likewise, the yeast cells grew normally in the absence of the inhaled anesthetics (figs. 6 and 7).

Figure 8A shows a representative ¹⁵N-¹H Heteronuclear Single Quantum Coherence (HSQC) NMR spectrum of PDZ1–3, in which 230 peaks are assigned to individual residues based on previous publications.^26,27  Those residues without reliable assignments or with weak intensity are labeled in lower case. To reveal an underlying cause of isoflurane perturbation on PSD-95 PDZ domain–mediated protein–protein interactions, we studied the interaction of isoflurane to the three PDZ domains of PSD-95 (PDZ1, PDZ2, and PDZ3) using NMR chemical shift as a probe. For each residue, its ¹⁵N and ¹H chemical shifts can be affected by changes in the local chemical environment, such as ligand-binding or the binding-induced alteration in protein conformations. Upon titrating different concentrations of isoflurane into a PDZ1–3 sample, a number of residues showed chemical shift changes in an isoflurane-concentration–dependent manner in the ¹⁵N-¹H HSQC spectra (fig. 8B). The fitting of chemical shift changes as a function of isoflurane concentrations provided a disassociation constant (Kd) of approximately 3 mM, suggesting a low affinity binding of isoflurane to PDZ1–3. The combined chemical shift changes of ¹⁵N and ¹H quantified the anesthetic perturbation. We found that isoflurane affected all three PDZ domains, but PDZ2 had the most affected residues, whereas PDZ3 had the least (fig. 8C). The affected residues in PDZ1 and PDZ2 were mostly in, or close to, their peptide-binding groove, whereas very few residues close to the binding groove in PDZ3 were affected (fig. 8C). These findings are valuable for understanding how anesthetics perturb PDZ domain–mediated protein–protein interactions.

To further determine whether and how isoflurane perturbs peptide binding to PDZ1−3, we characterized NMR spectral changes of the PDZ1–3 domains of PSD-95 produced by binding of peptides NR2A-c20 or NR2B-c20 in the absence and presence of isoflurane. We found that NR2A-c20 and NR2B-c20 disturbed the motion of PSD-95 PDZ1−3. Upon peptide binding, dozens of NMR peaks in the ¹H-¹⁵N HSQC NMR spectra of PDZ1−3 exhibit considerable intensity changes and some of them even disappeared from the spectra (Supplemental Digital Content 1, figs. 1 and 2, http://links.lww.com/ALN/B134; fig. 9, A and B). The NMR peak intensity changes are often associated with changes in protein motions. The reduced intensity or disappearance of peaks signified that the peptide binding probably drove the residues into intermediate time-scale motions.^28–32  Significant chemical shift changes along with peptide binding were also observed in the ¹H-¹⁵N HSQC NMR spectra of PDZ1−3 (Supplemental Digital Content 1, figs. 1 and 2, http://links.lww.com/ALN/B134; fig. 9, C and D). The combined ¹H and ¹⁵N chemical shift changes were mapped onto the structures of PDZ1−3 (fig. 9, E−G).

In the presence of peptides NR2A-c20 or NR2B-c20, isoflurane disturbed more extended regions of PSD-95 PDZ1−3 than in the absence of the peptides as evidenced by chemical shift changes in PDZ-peptide complexes upon adding isoflurane of approximately 3 mM (Supplemental Digital Content 1, figs. 3–6, http://links.lww.com/ALN/B134). Furthermore, the combined chemical shift showed that isoflurane introduced larger disturbance on the NR2A-c20 complex, but more extended disturbance on the NR2B-c20 complex (fig. 9, E−G). Several residues, including A106 and S116 of PDZ1 and S188, A201, and L241 of PDZ2, even disappeared in the spectrum of the NR2A-c20 complex. The disturbance of isoflurane on the PDZ-peptide complex could be beyond the binding groove (βB and αB). For the NR2A-c20 complex, the residues most perturbed by isoflurane were not within the binding groove, as was the case for the NR2B-c20 complex. Collectively, isoflurane interacts with residues outside of the peptide-binding region once either one of the peptides bound to PDZ1−3.

Cumulative evidence suggests that inhaled anesthetics act on multiple targets in the CNS.^33–35  It has been hypothesized that to achieve their anesthetic effect, they primarily modulate membrane-associated proteins that are involved in synaptic transmission.³  Within synapses, ion channels regulate ionic flow across the cellular membrane, thereby influencing the presynaptic release of neurotransmitters and altering postsynaptic excitability. Several ion channels have been reported to contribute to the physiological actions of anesthetics.³  Some, including nicotinic acetylcholine, serotonin type 3, GABA type A, glycine, NMDA, and AMPA receptors, are sensitive to inhaled anesthetics at clinically effective concentrations.^33,36–38  The alteration of potassium channel function has also been reported to be involved in the central effects of inhaled anesthetics.^39,40  In the CNS, PDZ domain–mediated protein–protein interactions have been identified in different signaling complexes at synapses and are critical to synaptic organization and for the activity of several excitatory receptors.^16,41–45  Our previous studies have shown that clinically relevant concentrations of inhaled anesthetics dose-dependently inhibit PDZ domain–mediated interactions between PSD-95 and NMDA receptors or neuronal nitric oxide synthase but have no effect on the non-PDZ domain–mediated interaction between the guanylate kinase domain of PSD-95 and Src homology 3 domain of SAP102.¹⁸  We have also shown that disrupting PDZ domain–mediated protein–protein interactions by systemic injection of the cell-permeable peptide Tat-PSD-95 PDZ2 reduces the threshold for halothane anesthesia.¹⁹  In the current study, we show that halothane and isoflurane inhibit PDZ domain–mediated interactions between PSD-95 and Shaker-type potassium channel Kv1.4 and between AMPA receptor subunit GluR2 and glutamate receptor–interacting protein or protein interacting with c kinase 1 and that isoflurane inhibits PDZ domain–mediated interactions between PSD-95 and NMDA receptors. A previous study⁴⁶  has shown that mutations of the homologous Shaker channel in flies causes anesthetic resistance. The discrepancy may be due to the difference of properties of potassium channels between Drosophila and mammalian species. For instance, it has been reported that the potassium channel in Drosophila differs kinetically from mammalian species by exhibiting a faster inactivation time course.⁴⁷  We further demonstrate that isoflurane mostly affects the residues close to or in the peptide-binding groove of PDZ1 and PDZ2 of PSD-95, while barely affecting the peptide-binding groove of PDZ3. Among three PDZ domains of PSD-95, PDZ2 domain has the most residues affected by isoflurane. Given that our current studies are investigating PDZ domain–mediated protein–protein interactions in in vitro preparations, our experiments are only suggestive of conditions in synapses, and actions in neuronal milieu could be different. Yet our combined studies of plasmon resonance, yeast two-hybrid, GST pull-down, and coimmunoprecipitation as well as our previous in vivo minimum alveolar anesthetic concentration studies are all consistent with a functional effect of anesthetic disruption of PDZ domain–mediated protein–protein interactions being important to the anesthetic state.

As binding partners of PDZ domains, NR2B-c20 and NR2A-c20 interact with the PDZ domains of PSD-95. Due to the long length of both peptides, their interactions with the PDZ domains extend beyond the peptide-binding groove, which is formed by αB, βB, and the carboxylate-binding loop⁷  of the PDZ domain. The extended region includes βA (E65, K162), βB-βC link (T83, I88, G89, G177, G179), βC (I100, S339), αA (K202, G345, L349), βD (G209, V215, V362), and βE (L367, r368). The “disappeared” and “shifted” residues partially overlapped with the residues forming the three peptide-binding groves of PSD-95 PDZ1−3, which include R70, G74, L75, G76, F77, S78, I79, G81, T97, H130, L137, and G141 in PDZ1; K165, G169, L170, G171, F172, S173, I174, G176, T192, H225, L232, and Y236 in PDZ2; R318, G322, L323, G324, F325, N326, I327, G329, S339, H372, L379, and G383 in PDZ3. This divergence from the peptide-binding groove could come from at least three different sources. First, the real structure could be different from the model used here, the simple combination of PDZ1, PDZ2, and PDZ3. Second, the peptides used in the model are shorter than what we used in our experiments, which should cause the extended disturbance on the PSD-95 PDZ1−3. Third, some “disappeared” and “shifted” residues, which were not assigned by the current ¹H-¹⁵N HSQC NMR spectra, could be other disturbed residues in the groove. About half of the residues in the groove of PDZ1 were affected by the peptides binding, whereas three quarters of the residues of PDZ2 were affected. These results suggest that the binding of NR2A-c20 and NR2B-c20 to the three PDZ domains of PSD-95 is not equivalent, and PDZ2 may be the primary target of the peptides.

Our yeast two-hybrid analysis and surface plasmon resonance assay further demonstrate that the inhaled anesthetic isoflurane disrupts the PDZ domain–mediated interaction between PSD-95 and NMDA receptors. PDZ domain–mediated protein–protein interactions provide a framework for the assembly of multiprotein signaling complexes at synapses. These PDZ proteins coordinate and guide the flow of regulatory information and regulate receptor and ion channel activities. One of the best-understood PDZ domain proteins at synapses is PSD-95, a modular protein highly enriched in the postsynaptic density. In our yeast two-hybrid experiments, the growth of the yeast cells harboring both pGADT7-PSD-95 PDZ2 and pGBKT7-NR2B C-terminus was slowed by isoflurane (0.33, 0.66, and 0.99 mM) in a dose-dependent manner. Our surface plasmon resonance assay showed that the inclusion of isoflurane in the mobile phase inhibited the equilibrium binding of PSD-95 PDZ2 to NR2B C-terminus in real time. These results suggest that the inhaled anesthetic isoflurane can inhibit the binding of PSD-95 with NMDA receptors and thereby interrupt relevant downstream signaling. Taken together, our data indicate that the ability of anesthetics to disrupt neuronal signaling pathways through inhibition of PDZ domain interactions plays a significant role in anesthesia itself. Different experimental conditions certainly affect quantitative data for peptide or isoflurane binding. For example, a protein-crowding effect may exist in cells but not in NMR sample tubes; the isolated PDZ2 is not comparable to the intact PDZ1−3; plasmon resonance experiments need to have the peptide immobilized. All of these may render variances in quantification of binding affinities.

We also investigated the effects of halothane and isoflurane on PDZ domain–mediated protein–protein interactions in GABA receptor signaling. We found that inhaled anesthetics had no effect on PDZ domain–mediated interactions between the GABA_B receptor and its interacting proteins PAPIN and Mupp1, suggesting that inhaled anesthetics have differential effects on different PDZ domain–mediated protein–protein interactions in the CNS.

In conclusion, PDZ domain–mediated protein–protein interactions in NMDA/AMPA receptor signaling and Kv1.4 channels, but not GABA receptor signaling, can be inhibited by inhaled anesthetics. Our data also indicate that PSD-95 PDZ1–3 domains can interact with isoflurane, of which PDZ2 is the most affected domain. Thus, inhaled anesthetics may affect synaptic transmission by binding to PDZ domains, which can be considered as a new molecular target for inhaled anesthetics. The common inhibition seen on pull-down, yeast two-hybrid, and NMR (all controlled experiments) as well as a functional change in minimum alveolar anesthetic concentration provide support for our hypothesis and for a functional role for the anesthetic–PDZ interaction in anesthesia.

The authors thank Randy A. Hall, Ph.D., Department of Pharmacology, Emory University School of Medicine (Atlanta, Georgia), for generously providing GABABR2, PAPIN PDZ1, and Mupp1 PDZ13 cDNA constructs.

This work was supported by the National Institutes of Health (Bethesda, Maryland) grants R01 GM049111 (to Dr. Johns) and R01 GM056257 (to Dr. Tang).

The authors declare no competing interests.