Abstract
High-voltage activated calcium channels are modulated by a series of auxiliary proteins, including those of the α2δ family. Until recently, only a single α2δ subunit was known, but two further members, α2δ-2 and -3, have since been identified. In this study, the structure of these two novel subunits has been characterized and binding of the antiepileptic drug gabapentin investigated. Using antibodies directed against the amino terminal portion of the proteins, the gross structure of the subunits could be analyzed by Western blotting. Similar to α2δ-1, both α2δ-2 and -3 subunits consist of two proteins—a larger α2 and a smaller δ that can be separated by reduction. The subunits are also highly N-glycosylated with approximately 30 kDa of their mass consisting of oligosaccharides. α2δ-1 was detected in all mouse tissues studied, whereas α2δ-2 was found at high levels in brain and heart. The α2δ-3 subunit was observed only in brain. α2δ-1 and α2δ-2, but not α2δ-3, were found to bind gabapentin. TheK d value of gabapentin binding to α2δ-2 was 153 nM compared with the higher affinity binding to α2δ-1 (K d = 59 nM).
Voltage gated calcium channels are multisubunit complexes that permit the influx of calcium ions into cells after changes in the plasma membrane potential. A number of voltage gated calcium channels exist, which can be broadly grouped into high-voltage (HVA) and low-voltage activated (LVA) channels. The HVA channels can be further subgrouped as L-, R-, P/Q-, and N-type, depending on their biophysical characteristics. The channels consist minimally of an α1 pore protein that conducts current, contains the voltage sensor, and is the target of several drugs. Seven genes have been identified for the α1 subunits of HVA channels and three for LVA channels (for reviews, see Hofmann et al., 1999; Lacinová et al., 2000). Although it is known that HVA channels are modulated by accessory subunits, it remains to be unequivocally established whether the LVA channels are associated with other proteins. The HVA auxiliary subunits are β (four genes), α2δ (three), and γ (six). Most of these subunits have splice variants, giving rise to an even larger number of possible channel combinations and behaviors. Mutations of several channel proteins have been shown to be a causative factor in neurological disorders, making the calcium channel subunits target for therapeutic interventions (Burgess and Noebels, 1999).
The α2δ family consists of three genes. The first subunit identified was α2δ-1 in rabbit skeletal muscle (Ellis et al., 1988). Five tissue-specific splice variants exist (Angelotti and Hofmann, 1996), but a functional significance of the splicing has not been established. Two new α2δ family members were subsequently identified in human and mouse, and were named α2δ-2 and -3 (Klugbauer et al., 1999). The novel α2δ subunits are 56 and 30% homologous to α2δ-1 at the amino acid level and share a number of structural motifs. The subunits have similar hydrophobicity profiles and all contain several potential N-glycosylation sites. Northern analysis has shown that α2δ-1 is ubiquitously expressed, α2δ-2 is found in several tissues including brain and heart, and α2δ-3 is brain-specific (Klugbauer et al., 1999). An association of α2δ-2 with tumors has been suggested, and the mouse homolog is a candidate for theducky epileptic phenotype (Gao et al., 2000).
α2δ-1 consists of two proteins—a highly glycosylated α2 that is believed to be extracellularly located, and a smaller δ protein that anchors the α2 to the cell membrane (Brickley et al., 1995;Wiser et al., 1996). These proteins are coded for by a single gene, the product of which is translated as a precursor polypeptide that is post-translationally cleaved (De Jongh et al., 1990). The α2 and δ associate by disulfide bridges that form between the numerous cysteine residues found in both proteins.
The effects of α2δ-1on the biophysical properties of α1 subunits depend on the expression system and subunits used. The subunit invariably increases the current density of calcium channels by increasing the amount of functional channel at the cell surface, and has been reported to allosterically alter the activation and inactivation of several α1 subunits (Singer et al., 1991; Bangalore et al., 1996, Felix et al., 1997, Klugbauer et al., 1999). α2δ-1 is known to enhance dihydropyridine binding to L-type channels and ω-conotoxin GVIA to N-type channels (Brust et al., 1993; Felix et al., 1997). α2δ-2 and -3 significantly enhance and modulate the current through a number of HVA and LVA channels (Klugbauer et al., 1999; Gao et al.; 2000, Hobom et al., 2000). The action of these subunits depends on the α1subunit expressed and α2δ-2 and -3 may preferentially interact with Cav2.3 (α1E), Cav2.1 (α1A), or even Cav3.1 (α1G).
Gabapentin (GBP) is an antiepileptic drug that has also found application in pain and anxiolytic disorders (Welty et al., 1993;Beydoun et al., 1995). GBP binds to rat brain (Hill et al., 1993) and skeletal muscle homogenates (Gee et al., 1996), with lower binding seen in heart, lung, and pancreas (Suman-Chauhan et al., 1993). GBP was subsequently found to bind specifically to α2δ-1 of voltage activated calcium channels (Gee et al., 1996). Because voltage activated channels are involved in controlling the electrical excitability of neurons, it has been postulated that this drug reduces calcium current by modulating α1 indirectly through its association with α2δ-1 (Gee et al., 1996).
The purpose of this study was to characterize the gross structure and properties of the novel α2δ-2 and -3 subunits. It was found that α2δ-1, -2, and -3 are similarly processed and post-translationally modified. α2δ-1 and -2 bound GBP with differing affinities, but the more distantly related α2δ-3 did not.
Materials and Methods
Transfection of COS-7 Cells.
Plasmids containing the α2δ subunits, or pcDNA3 (Invitrogen, Groningen, The Netherlands) as control, were transiently transfected in COS-7 cells, using FuGene 6 (Roche, Mannheim, Germany). α2δ-1 and -3 had been previously cloned into the expression vector pcDNA3, and α2δ-2 into pcDNA3.1/V5/His-TOPO (Klugbauer et al., 1999). The cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Karlsruhe, Germany) containing 10% fetal calf serum, 100 U of penicillin, and 100 μg/ml streptomycin at 37°C, 6% CO2; the cells were seeded at 1.5 × 106 cells per 100-mm culture plate. After 24 h, the cells were transfected in antibiotic-free medium with 6 μg of DNA using 18 μl of FuGene reagent per plate. The medium was replaced with complete medium after 16 h and harvested 60 to 72 h after the start of transfection.
Membrane Preparations.
Tissues from BALB/c mice were crushed under liquid nitrogen using a mortar and pestle. The powder was homogenized in buffer A (20 mM MOPS, pH 7.4, 300 mM sucrose, 2 mM EDTA, 1 mM iodoacetamide, 1 mM orthophenanthroline, 0.2 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 1 μg/ml leupeptin, 1 μg/ml pepstatin A, and 1 μg/ml antipain) by two 20-s bursts with a Polytron homogenizer (Kinematica AG, Switzerland) and 8 strokes in a power-driven Potter pestle (Braun Biotech, Germany). Cell debris was pelleted by centrifugation at 5000 g for 10 min at 4°C. Low-speed pellets were re-extracted as above and the supernatants combined. The supernatants were centrifuged at 70,000 g for 30 min at 4°C. The pellets were resuspended in buffer A and stored in aliquots at −80°C. Membranes from transfected COS-7 cells were prepared similarly, with the following modifications: cells were harvested by scraping into 1× phosphate-buffered saline, pelleted by low speed centrifugation, resuspended in 5 mM Tris-Cl, 5 mM EDTA, pH 7.4, containing protease inhibitors (as above) and incubated on ice for 15 min before homogenization. The pellets resulting from ultracentrifugation were resuspended in 50 mM MOPS, pH 7.4, plus protease inhibitors. The protein content was determined by the Bradford method using BSA as a standard.
Anti-Peptide Antibodies.
Antibodies were raised in rabbits by Gramsch Laboratories (Schwabhausen, Germany) against peptides of the human α2δ-2 (amino acids 98–115) and murine α2δ-3 (amino acids 59–76) subunits (Klugbauer et al., 1999). These sequences were selected because they have low homology to each other and to α2δ-1, are found after the putative signal peptide sequence, and have no potential N-glycosylation sites. The antibodies were affinity purified using a SulfoLink coupling gel (Pierce, Rockford, IL) with bound antigen. Antibodies were eluted with 4.5 M MgCl2, and concentrated using Centricon YM-50 centrifugal filter devices (Millipore Corporation, Bedford, MA). The antibody was made to 1 mg/ml BSA in phosphate-buffered saline and stored at −20°C in aliquots.
Western Blotting.
Proteins were separated by discontinuous SDS-PAGE in 5 or 7.5% resolving gels. The membrane preparations were denatured in Laemmli sample buffer (with or without 0.1 M dithiothreitol) by boiling for 3 min. Typically, 50 to 100 μg of tissue or 3 μg of COS-7 microsomal membrane preparation was loaded per lane. After electroblotting and blocking in 3% BSA, the nitrocellulose membranes were incubated for 2 h at room temperature with primary antibodies used at 1/750 for α2δ-2 and 1/500 for α2δ-3. A commercially available anti-α2δ-1 monoclonal antibody (ABR, Golden, CO) was used at a concentration of 1/500. Incubation with the secondary goat anti-rabbit- or goat anti-mouse-horseradish peroxidase antibodies (Dianova, Germany) proceeded for 1 h at room temperature at a dilution of 1/7000. Antibody binding was detected using the ECL system (Amersham Pharmacia Biotech, Freiburg, Germany).
Deglycosylation Assay.
Membrane protein (70 μg) was denatured by boiling in 0.1 M β-mercaptoethanol, 0.5% SDS for 5 min. The reaction was brought to 15 mM Tris-Cl, pH 8.0, 20 mM orthophenanthroline, and 1% Triton X-100. Two units ofN-glycosidase F (Roche, Mannheim, Germany) were added and the reaction allowed to proceed for 5 h at 37°C. The protein was separated by SDS-PAGE gel and analyzed by Western blotting as described above.
Gabapentin Binding Assay.
COS-7 membrane preparations (10 to 30 μg) were incubated in 200 μl volumes with various concentrations of [3H]gabapentin (143 Ci/mmol, custom synthesized by Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, UK) in 10 mM HEPES/KOH, pH 7.4, for 30 min at room temperature. The protein was precipitated using 8 ml of ice-cold precipitation buffer (22.5 mM HEPES, pH 7.4; 11.25% polyethylene glycol 6000; 11.25 mM CaCl2) and filtered over GF/B filters soaked in the same buffer. Two further 8-ml volumes of precipitation buffer were used as washes. The activity of the filters was counted in a scintillation counter. Concentrations greater than 20 nM were achieved by adding nonradioactive gabapentin to the required concentration. The corrected binding was calculated using the equation, Total GBP bound = (specific dpm) × [1 + (concentration nonradioactive gabapentin/concentration [3H]gabapentin)].
Nonspecific binding was determined in the presence of 10 μM unlabeled gabapentin. The background binding was less than 20% of the counts without addition of unlabeled gabapentin. To assess the effect of the carbohydrates on GBP binding, 30 μg of native mouse brain membranes were incubated with 2 U of N-glycosidase F in 15 mM Tris-Cl containing protease inhibitors (as used for preparation of the membranes), for 18 h at 37°C without prior denaturing.
Results
Analysis of α2δ Protein Structure.
The specificity of the peptide antibodies was tested using membrane preparations of COS-7 cells transfected with the various α2δ proteins. The commercial α2δ-1 antibody did not recognize α2δ-2 and -3 preparations. Similarly, the peptide antibodies only recognized the α2δ against which they were raised (Fig. 1). The antibodies were also preincubated with an excess of peptide antigen before incubation with the protein blots to test for cross-reactivity. The bands corresponding to the α2δ proteins were not detected, indicating that the antibodies are specific (results not shown).
Using antibodies against all three α2δ proteins, various murine tissues were analyzed by Western blotting. In our study, α2δ-1 in brain had a mass of 200 kDa under nonreducing conditions and 140 kDa in a reducing environment (Fig. 2), which is in accord with the values reported previously (Jay et al., 1991). α2δ-2 in mouse brain had an apparent molecular mass of 190 kDa under nonreducing conditions, which shifted to 138 kDa when DTT was added. α2δ-3 in brain had the lowest mobility of the α2δ proteins, with a mass of 166 kDa under nonreducing and 131 kDa under reducing conditions (Fig. 2A). Our results indicate that, similar to α2δ-1, α2δ-2, and α2δ-3 consist of separate, disulfide linked α2 and δ proteins. Because the antibodies recognize the amino-terminal domains of the α2δ-2 and -3 subunits, which correspond to the putative α2, the shift in migration is due to the loss of the δ protein caused by reduction of the disulfide bonds. The mass estimated by subtraction for δ of α2δ-1 is 53 kDa, 57 kDa for α2δ-2, and 35 kDa for α2δ-3. A shift in electrophoretic mobility of both α2δ-2 and -3 upon reduction was also observed in COS-7 membranes expressing the proteins (Fig. 4, insert). The multiple bands detected are most probably caused by the glycosylation and processing of the overexpressed proteins by the COS-7 cells.
To test whether the novel subunits are N-glycosylated, mouse brain membrane preparations were incubated withN-glycosidase F and studied by Western blotting. Deglycosylation resulted in a loss of approximately 30 kDa for all α2δ proteins (Fig. 2B).
The tissue distribution of the three α2δ subunits in mouse was also analyzed by immunoblotting. As can be seen in Fig. 3, α2δ-1 is expressed in all tissues studied, namely brain, heart, skeletal muscle, liver, and lung. The highest expression was seen in skeletal muscle and brain; in the latter case only 40 μg of protein was used while 100 μg of the other preparations were loaded. The heart subunit had a larger apparent mass than that of the brain subunit −150 kDa as compared with 140 kDa (reducing conditions). The α2δ-1b splice variant is exclusively found in brain, whereas the predominant forms in heart are α2δ-1c and -d (Angelotti and Hofmann, 1996). Because the α2δ-1c and -d heart splice variants have 5 and 12 fewer amino acids than that of the brain isoform, respectively, the difference in mass is believed to be caused by glycosylation.
α2δ-2 was found to be expressed at high levels in brain and, to a lesser extent, in heart (Fig. 3). Weaker reactive bands of a similar size were seen in lung, liver, and skeletal muscle, in addition to bands at approximately 120 kDa in brain, heart, and lung. The apparent mass of the primary reactive band in heart was 153 kDa under reducing conditions, which is larger than that seen for α2δ-2 in brain (138 kDa). This may also be caused by differential glycosylation and/or splice variation, because four splice variants have been described so far for α2δ-2 (Gao et al., 2000; Hobom et al., 2000). To determine whether the protein detected in the tissues is in fact α2δ-2, the antibody was preincubated with an excess of antigen. The bands at approximately 160 and 120 kDa disappeared, indicating a specific reaction (data not shown). Because the antibody recognized both bands, the protein detected at lower masses is caused by the presence of more than one splice form, partial degradation, or incompletely processed forms of the highly glycosylated protein. The latter option is possible as the membrane preparations also contain endoplasmic reticulum with associated protein. α2δ-3 was only found to be expressed in brain, and the smears below and above the size expected for α2δ-3 in skeletal muscle are believed to be caused by spurious reactions with contractile elements (Fig. 3).
Gabapentin Binding Assays.
Binding of GBP to the α2δ subunits was assessed using COS-7 cells that overexpressed the proteins. Membrane preparations were incubated with [3H]gabapentin for 30 min. The protein was subsequently precipitated, captured on filters, and measured by scintillation counting. α2δ-1 and α2δ-2 were found to specifically bind GBP, whereas α2δ-3 did not show any binding (Fig.4). Cells transfected with pcDNA3 alone as a control did not interact with GBP. α2δ-1 bound GBP with a relatively high affinity, and a dissociation constant (K d) of 59 ± 6 nM (S.E.M.) was calculated by Scatchard analysis of data from three experiments (Fig.5). α2δ-2 was found to bind GBP with a K d of 153 ±14 nM (S.E.M.) in six independent experiments. In both cases, GBP seemed to bind to a single binding site, as judged by the Scatchard plots. The contribution of α2δ carbohydrate side chains to GBP binding was also investigated. The binding of [3H]GBP to native mouse brain membranes following deglycosylation for 18 h with N-glycosidase F was not significantly altered relative to controls incubated without enzyme (not shown).
Discussion
This study was undertaken to clarify the gross protein structure and characteristics of the novel α2δ-2 and -3 subunits. It was of particular interest to determine whether the characteristic traits of α2δ-1, namely its composition of two proteins derived from a single precursor protein, and its high level of glycosylation, are conserved. Because GBP, a drug increasingly used for neurological disorders, binds to α2δ-1, the novel subunits were also tested for drug binding.
Using antibodies directed against the putative α2 of the α2δ-2 and -3 subunits, it could be assessed whether precursor proteins are cleaved and form complexes via disulfide bridges. Both α2δ-2 and -3 in native tissues were found to be composed of an α2 and δ, which could be separated by reduction. The difference in mass of the nonreduced complex relative to the reduced protein roughly corresponds to that of δ alone and was approximately 50 kDa for each of the α2δ proteins. The cleavage site of α2δ-1 is between A934and A935, when the signal sequence is not taken as part of the protein (De Jongh et al., 1990; Jay et al., 1991). The alanine at position 934 in α2δ-1 is conserved in α2δ-2 and -3, although the rest of the sequence in the region diverges. Although the α2δ-1 cleavage site cannot be assumed to be the same as for the novel proteins, the approximate sizes of the δ proteins and the conservation of the alanine suggest that this could be the case. If the alanine is assumed to be the first residue of δ for both α2δ-2 and -3, then the mass calculated for δ of α2δ-2 based on its amino acid content is 17.2 kDa, and for α2δ-3, 15.1 kDa. The mass estimated indirectly is larger than that calculated using the amino acid sequence. This difference is likely to be due to the difficulty associated with inferring sizes based on shifts in SDS-PAGE migration of reduced and nonreduced proteins, as well as post-translational modifications such as N-glycosylation (both δ proteins contain potential N-glycosylation sites). A similar pattern of modification has been observed for the δ of α2δ-1, which has a calculated mass of 19 kDa, and a migration of 24 and 27 kDa using antibodies against the δ (De Jongh et al., 1990).
All three α2δ proteins were found to be highly glycosylated, with 30 kDa of the mass of the proteins consisting of oligosaccharides. Because glycosylation is essential for current stimulation by α2δ-1 of Cav2.1 (α1A) (Gurnett et al., 1996), this is likely to be the case for the novel subunits as well.
The distribution of α2δ protein in mouse tissue was compared with that of the mRNA expression pattern (Klugbauer et al., 1999). α2δ-3 was detected only in brain, which corresponds with the results of the Northern analysis. α2δ-2 mRNA was originally described to be ubiquitously expressed, with the highest levels in brain, heart, pancreas, and skeletal muscle (Klugbauer et al., 1999). Lower levels of the mRNA were seen in other tissues after longer exposures. In this study, α2δ-2 protein was detected at very low levels in skeletal muscle, although the Northern analysis had produced a signal comparable with brain. Gao et al. (2000) reported high levels of α2δ-2 mRNA in human lung and cloned the subunit from a lung library. Although α2δ-2 protein was observed at low levels in mouse lung in this study, it is possible that certain lung cell types express high levels. Whereas the Gao group showed overexpression of α2δ-2 protein in various tumor cell lines, no data was presented on the endogenous levels in healthy lung tissue.
The specific binding of gabapentin to α2δ-1 was the first described interaction between a regulatory subunit of voltage activated calcium channels and a pharmaceutical agent. TheK d of porcine brain α2δ-1 was reported as 9.4 nM (Brown et al., 1998) but as 37.5 nM for porcine α2δ-1 expressed in COS-7 cells (Brown and Gee, 1998) and 16 nM for rabbit α2δ-1 in COS-7 cells (Gee et al., 1996). In this study,K d values of 59 nM for α2δ-1 and 153 nM for α2δ-2 were determined. In a preliminary report by Su et al. (2000), α2δ-2 is described as having two binding sites (K dvalues of 147 and 25 nM). The reason for variations inK d values is not clear, but may be attributable to species differences and assay methods. Because GBP is used for a variety of neurological disorders, it is interesting to note that it binds to two auxiliary calcium channel subunits that have been found to exert differing modulatory effects on α1 pore subunits in heterologous expression systems.
The effect of GBP on the physiological activity of calcium channels is not clearly understood. In patch-clamp studies with hippocampal granule cells, no effect of GBP was reported (Schumacher et al., 1998). However, in other studies, modest to dramatic changes in calcium current were noted. A reduction in the calcium current in isolated neurons (Stefani et al., 1998) and in rat neocortical slices (Fink et al., 2000) upon application of GBP has been described. Calabresi et al. (2000) found GBP to reduce most excitatory properties of striatal spiny neurons, which could account for the anticonvulsant effect of the drug.Dooley et al. (2000) have also shown that GBP and a related compound, pregabalin, reduce the release of norepinephrine when stimulated by potassium and electrical pulses. The calcium channels affected are not known, but candidates are L-type (Stefani et al., 1998) and P/Q (Fink et al., 2000; Meder and Dooley, 2000). No consistent effect of GBP on Cav1.2 (α1C), Cav2.1 (α1A), and Cav3.2 (α1G) currents in the heterologous human embryonic kidney 293 expression system were observed in this study. The complexity of the interaction between GBP and α2δ is further illustrated by a temperature dependent influence of ruthenium red, MgCl2, and spermine on GBP binding (Taylor and Bonhaus, 2000). Initial studies on GBP binding using rat tissue homogenates showed strong binding in skeletal muscle and brain, where α2δ-1 is most highly expressed (Gee et al., 1996). A much lower binding of the drug was seen in liver and kidney, which express considerable levels of the protein, as judged by Western blotting. A possible explanation for these conflicting results is that the binding of GBP to α2δ is modulated by other subunits (e.g., the α1 pore protein). It is possible that the effects of GBP depend on the composition and environment of the channel. The lack of clinical side effects of the drug on skeletal muscle and other α2δ-1 expressing tissues supports this view (Beydoun et al., 1995). GBP has also been shown to be an agonist of certain γ-aminobutyric acidB receptors, and this has also been postulated to be involved in the clinical action of the drug (Ng et al., 2001).
Gabapentin binding is dependent on the presence of both α and δ subunits, which do not have to be translated as a single precursor protein (Wang et al., 1999). Their interaction is important, however, because neither α nor δ bind the drug when expressed alone (Wang et al., 1999). Cleavage of the precursor protein is also not required for binding (Brown and Gee, 1998). Mutation analysis of porcine α2δ-1 by Brown and Gee (1998) led to the identification of a region (960–994) in δ-1, containing a zinc-like finger, that is important for gabapentin binding. Another study identified residues 206 to 222, 516 to 537, and 383 to 603 as essential for binding (Wang et al., 1999). Because α2δ-2 has in this study been shown to bind gabapentin, it is instructive to compare the sequences identified as putative binding sites in previous reports (Fig.6). Mutations of charged amino acids in α2δ-1 was performed by Wang et al. (1999), and Arg217 found to be the most important. This residue is found in α2δ-2, but not in α2δ-3, supporting its role in the binding of GBP. It is, however, not clear whether these sites are essential for the association of α2 with δ, or whether they form a gabapentin-binding pocket.
Gabapentin analogs have been developed that bind to α2δ-1 with a higher affinity than gabapentin and are effective in an animal model of epilepsy (Bryans et al., 1998). Because α2δ subunits may associate preferentially with α1 subunits in the brain, these drugs could have differing therapeutic actions if more than one of α2δ family binds these drugs.
In summary, we present results indicating that α2δ-2 and -3 consist of two proteins, derived from a single transcript, that are associated by disulfide bonds. Both novel subunits are highly glycosylated. α2δ-2, but not α2δ-3, is capable of binding the antiepileptic drug GBP, which has a higher affinity for α2δ-1 than for α2δ-2. Further investigation into the molecular action of GBP and its analogs on voltage gated calcium channels should enable the fine-tuning of treatment of neurological disorders.
Acknowledgments
We thank Lubica Lacinová for electrophysiological experiments and Susanne Kamp for technical assistance.
Footnotes
- Received November 11, 2000.
- Accepted February 9, 2001.
-
Send reprint requests to: Dr. N. Klugbauer, Institut für Pharmakologie und Toxikologie der Technischen Universität München, Biedersteiner Strasse 29, 80802 München, Germany. E-mail: klugbauer{at}ipt.med.tu-muenchen.de
-
This research was supported by the Deutsche Forschungs Gemeinschaft and the Fond der Chemischen Industrie.
Abbreviations
- HVA
- high voltage activated
- LVA
- low voltage activated
- GBP
- gabapentin
- MOPS
- 3-(N-morpholino)propanesulfonic acid
- BSA
- bovine serum albumin
- PAGE
- polyacrylamide gel electrophoresis
- DTT
- dithiothreitol
- The American Society for Pharmacology and Experimental Therapeutics