Andreas Jurik

Amphetamine actions at the serotonin transporter rely on the availability of phosphatidylinositol-4,5-bisphosphate

Nerve functions require phosphatidylinositol-4,5-bisphosphate (PIP2) that binds to ion channels, thereby controlling their gating. Channel properties are also attributed to serotonin transporters (SERTs); however, SERT regulation by PIP2 has not been reported. SERTs control neurotransmission by removing serotonin from the extracellular space. An increase in extracellular serotonin results from transporter-mediated efflux triggered by amphetamine-like psychostimulants. Herein, we altered the abundance of PIP2 by activating phospholipase-C (PLC), using a scavenging peptide, and inhibiting PIP2-synthesis. We tested the effects of the verified scarcity of PIP2 on amphetamine-triggered SERT functions in human cells. We observed an interaction between SERT and PIP2 in pull-down assays. On decreased PIP2 availability, amphetamine-evoked currents were markedly reduced compared with controls, as was amphetamine-induced efflux. Signaling downstream of PLC was excluded as a cause for these effects. A reduction of substrate efflux due to PLC activation was also found with recombinant noradrenaline transporters and in rat hippocampal slices. Transmitter uptake was not affected by PIP2 reduction. Moreover, SERT was revealed to have a positively charged binding site for PIP2. Mutation of the latter resulted in a loss of amphetamine-induced SERT-mediated efflux and currents, as well as a lack of PIP2-dependent effects. Substrate uptake and surface expression were comparable between mutant and WT SERTs. These findings demonstrate that PIP2 binding to monoamine transporters is a prerequisite for amphetamine actions without being a requirement for neurotransmitter uptake. These results open the way to target amphetamine-induced SERT-dependent actions independently of normal SERT function and thus to treat psychostimulant addiction.

A Binding Mode Hypothesis of Tiagabine Confirms Liothyronine Effect on γ-Aminobutyric Acid Transporter 1 (GAT1)

Elevating GABA levels in the synaptic cleft by inhibiting its reuptake carrier GAT1 is an established approach for the treatment of CNS disorders like epilepsy. With the increasing availability of crystal structures of transmembrane transporters, structure-based approaches to elucidate the molecular basis of ligand–transporter interaction also become feasible. Experimental data guided docking of derivatives of the GAT1 inhibitor tiagabine into a protein homology model of GAT1 allowed derivation of a common binding mode for this class of inhibitors that is able to account for the distinct structure–activity relationship pattern of the data set. Translating essential binding features into a pharmacophore model followed by in silico screening of the DrugBank identified liothyronine as a drug potentially exerting a similar effect on GAT1. Experimental testing further confirmed the GAT1 inhibiting properties of this thyroid hormone.

Identification of the First Highly Subtype-Selective Inhibitor of Human GABA Transporter GAT3.

Screening a library of small-molecule compounds using a cell line expressing human GABA transporter 3 (hGAT3) in a [(3)H]GABA uptake assay identified isatin derivatives as a new class of hGAT3 inhibitors. A subsequent structure-activity relationship (SAR) study led to the identification of hGAT3-selective inhibitors (i.e., compounds 20 and 34) that were superior to the reference hGAT3 inhibitor, (S)-SNAP-5114, in terms of potency (low micromolar IC50 values) and selectivity (>30-fold selective for hGAT3 over hGAT1/hGAT2/hBGT1). Further pharmacological characterization of compound 20 (5-(thiophen-2-yl)indoline-2,3-dione) revealed a noncompetitive mode of inhibition at hGAT3. This suggests that this compound class, which has no structural resemblance to GABA, has a binding site different from the substrate, GABA. This was supported by a molecular modeling study that suggested a unique binding site that matched the observed selectivity, inhibition kinetics, and SAR of the compound series. These compounds are the most potent GAT3 inhibitors reported to date that provide selectivity for GAT3 over other GABA transporter subtypes.

Structure activity relationship of selective GABA uptake inhibitors.

A series of β-amino acids with lipophilic diaromatic side chain was synthesized and characterized pharmacologically on mouse γ-amino butyric acid (GABA) transporter subtypes mGAT1-4 in order to investigate structure activity relationships (SAR) for mGAT2 (corresponding to hBGT-1). Variation in the lipophilic diaromatic side chain was probed to understand the role of the side chain for activity. This yielded several selective compounds of which the best (1R,2S)-5a was more than 10 fold selective towards other subtypes, although potency was moderate. A docking study was performed to investigate possible binding modes of the compounds in mGAT2 suggesting a binding mode similar to that proposed for Tiagabine in hGAT1. Specific interactions between the transporter and the amino acid part of the ligands may account for a reverted preference towards mGAT2 over mGAT1.

Classification of High-Activity Tiagabine Analogs by Binary QSAR Modeling

Termination of GABA-ergic signaling requires fast uptake of the neurotransmitter by highly selective transporter proteins. Four subtypes of sodium- and chloride-dependent GABA transporters exist, GAT-1 being the most prominent one in the brain. The only marketed drug targeting this transporter system is the anticonvulsant tiagabine.1 It is highly GAT-1 selective, consisting of R-nipecotic acid as a GABA mimetic moiety and a diaryl region attached by a linker chain.2–3 Its development roots back to the work of Yunger et al. in the early 1980s, introducing a lipophilic diaromatic region to the amino acid, thus tackling the fact that nipecotic acid, already a potent inhibitor of GABA transport, is not able to penetrate the blood brain barrier.4 This resulted in the so-called SK&F tool compounds, which were subsequently optimized towards IC50 values in the nanomolar range and simultaneously rising GAT-1 selectivity. Lots of synthetic effort focusing on modifications in linker length and polarity, and substitutions on the (mainly di-) aromatic region has been put into the structural optimization of the compound class, as summarized by Madsen et al.5 Modifying the amino acid region is less tolerated, but might be the key for stepping towards other GAT subtypes. Likewise, the introduction of a third aryl ring goes along with an increase in selectivity for hGAT-3.6–7 It also turned out that ortho-substitution of at least one of the aromatic rings has a beneficial effect. In addition, introduction of a polar region at the distal side of the aliphatic linker, which is connected to the cyclic amino acid at its protonable nitrogen atom, increases activity. This is usually achieved by introducing a diaryloxime or a diarylvinyl ether group. Isolated investigation of the preferred carboxy group configuration in this GABA mimetic moiety showed a clear superiority of R-configuration to the non-racemic guvacine scaffold, itself being more potent than compounds containing S-nipecotic acid.8,9 Despite the considerable number of structure-activity relationship observations that have been described,6,10 a quantitative summary of their respective contributions has not been performed yet.​yet.11 Figure 1 Chemical structures of GABA, R-nipecotic acid, the lipophilic derivatives SK&F 89976-A and tiagabine. In the present work, we describe a ligand-based approach to summarize SAR information derived from a dataset of published lipophilic aromatic GAT inhibitors. A dataset of 162 consistently tested compounds was collected from the literature.8,9,11–19 Two classes of 2D and internal 3D descriptors were calculated using the software package MOE2012.10.20 The 2D class, not depending on the molecule conformation, consisted of 188 descriptors belonging to 7 categories, namely physicochemical properties, subdivided surface area, atom and bond counts, Kier & Hall connectivity and kappa shape Index, adjacency and distance matrix, pharmacophore feature and partial charge descriptors. Out of the available 3D descriptors, the ′x3D′-class was discarded as it depends on external coordinates as a frame of reference. The remaining ′i3D′ class consisted of 138 features, describing shape, potential energy and partial charges of the dataset. In addition, indicator variables were introduced for the three scaffolds of the amino acid mimicry, namely R- and S-nipecotic acid and guvacine. Three columns were added to the dataset, one for each scaffold. Presence or absence of the respective scaffold in the chemical structure was indicated by 1 and 0, respectively. The full data matrix is given in the supplementary material. Surprisingly, although the data set seems ideal for Hansch-anaylsis and PLS, all attempts to retrieve statistically significant models failed. Therefore, the strategy was adjusted towards binary QSAR. The according method implemented in the QuaSAR module of MOE2012 uses a biased Bayesian inference technique in order to predict the probability of a compound to be active or inactive, even for small and unbalanced data sets.21 A pIC50 activity threshold of 7.0 was defined for discrimination between highly active and inactive compounds. Four different descriptor sets were used for building the binary models: 16 physicochemical descriptors, 32 binned VSA descriptors plus the three indicator variables, one set of features chosen by contingency analysis (see Table ​Table1),1), and the first 10 principal components of the internal 3D descriptors. In order to assess the quality of the models, internal validation by leave-one-out cross-validation and prediction of an external test set was performed. For the latter, two procedures were applied to split the 162 compounds into 147 (90 %) for training and 16 (10 %) for testing. In order to achieve maximum diversity in the test set, primary splitting of the compounds was done on the basis of maximum diversity, calculated by MACCS fingerprint clustering. The second method used ten times repeated random selection. Upon selection of the test set compounds, it was also taken care of preserving the ratio between active and inactive compounds in both subset populations. Table 1 Descriptors chosen by contingency analysis for the two training sets and their (mean) relative importance Using the whole panel of 2D-descriptors followed by backward selection as well as the sole use of the 16 physicochemical descriptors did not provide any reasonably good model. Also 3D descriptors performed poorly and were thus discarded. Nevertheless, a set of 32 binned van der Waals surface area (VSA) descriptors turned out to be well suited to describe the dataset. Introducing the indicator variables outlined above increased both positive and negative predictive power for the external test set from 42.9 % and 77.8 % to 60.0 % and 81.8 %, respectively, clearly justifying their use (Table ​(Table2).2). The second feature selection method applied used the descriptor contingency calculation available in the MOE package (Table ​(Table1).1). For the diversity split, it were 9 mainly atom/bond count, adjacency matrix and polarity descriptors, performing equally well when compared to the VSA descriptors for the training set, but exhibiting inferior positive predictive power for the test set (Table ​(Table22). Table 2 Accuracy of the binary QSAR models for training and test sets (%) For the diversity splits, the best model showed an overall accuracy on the training set of 89.7 %, with 98.1 % for active and 85.1 % for inactive compounds. The external test set was predicted with an overall accuracy of 75.0 % (60 % on actives and 81.8 % on inactives), as summarized in Table ​Table2.2. Accordingly generated models for the ten random splits achieved similar values, performing slightly better on the training sets but exposing lower accuracy for identifying the active instances of the external test sets. Analyzing the misclassified compounds revealed several insights. VSA descriptors exhibited some difficulties in handling molecules with asymmetrical biaromatic moieties, which are often classified as false positives. Main challenges for 3D descriptors included long linker compounds, large tricyclic moieties and S-configuration of the carboxy group. Just two compounds of the dataset, which is provided in the Supporting Information, were misclassified by at least two models: Cpd. 100, was the only active one bearing a 7 heavy atom long linker. The other, cpd. 37, often was assigned to the active class due to its favorable combination of ortho-substitution and an oxime moiety in the linker, yet having an S-configured carboxy group. Nevertheless, in both cases the pIC50 value was close to the threshold of 7.0 (see Figure ​Figure22 for comparison with the most active compound 69). Figure 2 Comparison of the most active compound with most often misclassified compounds. For the most active compound 69, the optimal linker length and polarity, ortho-substitution and R-configuration of the carboxy group are present. Frequent false negative cpd. ... The surprisingly low importance of the indicator variables during the model generation of the training sets might be explained by the underrepresentation of S-configured representatives in the dataset. New insights about activity-determining features of GABA uptake come from the importance of the two descriptors wienerPol and opr_brigid. Suggested along with three other descriptors in the contingency selection for the random split training set, they perform surprisingly well even if taken alone. Taking just the two for model generation yields Matthews correlation coefficients of 0.53 and 0.63 for training and test set, respectively. Even though the performance is by far weaker when applied to the diverse split, this nicely demonstrates that the degree of rigidity and the polarity distribution play a significantly larger role for activity than expected so far. In contrast, taking just the indicator variables for model generation did not lead to any significant model. In conclusion, BQSAR is a versatile method for capturing SAR information from consistent datasets, when classical QSAR models do not yield sufficient predictive power.

Development of Refined Homology Models: Adding the Missing Information to the Medically Relevant Neurotransmitter Transporters

Neurotransmitter:sodium symporters are located on presynaptic neurons and terminate neurotransmission by removing the monoamine substrates from the synaptic cleft. Until very recently, only several conformational snapshots/structures of a bacterial homolog of neurotransmitter:sodium symporters, namely, the leucine/alanine transporter LeuT from Aquifex aeolicus, were available. However, this transporter shares only 21b % overall sequence identity with its human homologs. In this chapter, we describe how a model can be developed from a template with such low identity. The effort of model building will strongly depend on the purpose. We discuss this process and focus on the important steps that allowed us to obtain a model which can be used for molecular dynamics simulations. Furthermore, we also highlight the inherent limitations of the proposed approaches. Prediction of ligand binding brings in additional complexity. Therefore, experimental scrutiny of the resulting models is a key component to successful validation. We describe two specific examples: model building of the dopamine transporter and ligand docking to the serotonin transporter. We evaluate our modeling approach by direct comparison of our models to the recently published first eukaryotic neurotransmitter:sodium symporter, the drosophila melanogaster DAT transporter.

Development of potential selective and reversible pyrazoline based MAO-B inhibitors as MAO-B PET tracer precursors and reference substances for the early detection of Alzheimer's disease.

Since high MAO-B levels are present in early stages of AD, the MAO-B system can be designated as an appropriate and prospective tracer target of molecular imaging biomarkers for the detection of early AD. According to the preceding investigations of Mishra et al. the aim of this work was the development of a compound library of selective and reversible MAO-B inhibitors by performing bioisosteric modifications of the core structure of 3-(anthracen-9-yl)-5-phenyl-4,5-dihydro-1H-pyrazoles. In conclusion, 13 new pyrazoline based derivatives have been prepared, which will serve as precursor substances for future radiolabeling as well as reference compounds for the investigation of increased MAO-B levels in AD.

Towards an understanding of the psychostimulant action of amphetamine and cocaine

Cocaine and amphetamine are psychostimulant drugs that are illicitly used; they affect sensory perception by targeting the neurotransmitter: sodium symporters (NSS) at the synapses between neurons. They both increase the concentration of the neurotransmitter in the synaptic cleft but by different means.

Amphetamine actions rely on the availability of phosphatidylinositol-4,5-bisphosphate

Background Neuronal functions, such as excitability or endoand exocytosis, require phosphatidylinositol-4,5-bisphosphate (PIP2) since ion channels and other proteins involved in these processes are regulated by PIP2. Monoamine transporters control neurotransmission by removing monoamines from the extracellular space. They also display channel properties, but their regulation by PIP2 has not been reported. The psychostimulant amphetamine acts on monoamine transporters to stimulate transportermediated currents and efflux and thereby increases the levels of extracellular monoamines.

Pairwise structural comparison of tiagabine analogs gives new insights into their protein binding modes

Tiagabine (Gabitril®) is a selective inhibitor of the human gamma-aminobutyric acid (GABA) transporter 1 (hGAT-1), a transport protein belonging to the family of neurotransmitter-sodium-symporters (NSS). It is a marketed drug, used for treatment of epilepsy. However, the molecular basis of protein-ligand interaction remains obscure due to the lack of a 3D structure of the target protein. In order to identify activity-determining structural features of a series of tiagabine analogs taken from literature [1-3], we chose an approach combining traditional methods of molecular modeling with exhaustive sampling of docking poses, and a pairwise comparison of structural features and their respective bioactivity values. We determined a common binding mode of tiagabine analogs, which is in nice agreement with literature [4]. Further, we were able to trace back considerable differences in inhibitory activities to distinct molecular attributes of the analogs. Our study revealed the molecular explanation for the importance of a polar linker region and thus paves the way for subsequent screening efforts in the search for novel GAT-1 inhibitors.