PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Proteome Res. Author manuscript; available in PMC 2010 September 24.
Published in final edited form as:
PMCID: PMC2945258
NIHMSID: NIHMS236775

Amazing Stability of the Arginine–Phosphate Electrostatic Interaction

Abstract

Electrostatic interactions between a basic epitope containing adjacent arginine residues and an acidic epitope containing a phosphorylated serine are involved in receptor heteromerization. In the present study, we demonstrate that this arginine–phosphate electrostatic interaction possesses a “covalent-like” stability. Hence, these bonds can withstand fragmentation by mass spectrometric collision-induced dissociation at energies similar to those that fragment covalent bonds and they demonstrate an extremely low dissociation constant by plasmon resonance. The present work also highlights the importance of phosphorylation–dephosphorylation events in the modulation of this electrostatic attraction. Phosphorylation of the acidic epitope, a casein kinase one consensus site, makes it available to interact with the basic epitope. On the other hand, phosphorylation of serine and/or threonine residues adjacent to the basic epitope, a protein kinase A consensus site, slows down the attraction between the epitopes. Although analyzed here in the frame of receptor heteromerization, the arginine-phosphate electrostatic interaction most likely represents a general mechanism in protein–protein interactions.

Keywords: electrostatic interaction, receptor heteromers, phosphorylation, CK1, PKA

1. Introduction

Different mechanisms were suggested for the underlying protein–protein interaction in receptor heteromerization.1,2 We have recently demonstrated that heteromerization of adenosine A2A with dopamine D2 receptors35 and glutamate NMDA receptors (NR1–1 subunit) with dopamine D1 receptors6,7 depend on epitope–epitope electrostatic interactions and suggested that this is a general mechanism for receptor heteromerization.810 Striking similarities were found between the epitopes involved in the A2A-D2 and NMDA-D1 receptor heteromerization:810 (1) an arginine (Arg)-rich epitope localized in the N-terminal portion of the third intracellular loop of the D2 receptor and the C-terminus of the NR1-1 subunit of the NMDA receptor; (2) on the opposing epitopes of the C-termini of both A2A and D1 receptors, a serine (Ser) residue susceptible of phosphorylation by casein kinase one (CK1) (pSer374 and pSer397, respectively); (3) a protein kinase A (PKA) consensus site adjacent to the Arg-rich epitope of the D2 receptor and the NR1-1 subunit of the NMDA receptor. In the present study, by using mass spectrometric collision-induced dissociation and surface plasmon resonance techniques, we demonstrate that the Arg-phosphate electrostatic interaction possesses a “covalent-like” stability and that its formation can either be induced or impaired as a result of phosphorylation.

2. Materials and Methods

Mass Spectrometry

A Quadrupole-Time-of-Flight (Q-TOF) Global Ultima mass spectrometer (Waters, Milford, MA) was used for ElectroSpray Ionization (ESI) analysis. A flow rate of 5 µL/min was used to introduce the sample into the mass spectrometer. The mass spectrometer was operated in positive mode with a capillary voltage of 2.9 kV, a sampling cone voltage of 50 V, a source temperature of 100 °C, a desolvation temperature of 200 °C, a desolvation gas flow rate of 650 L/Hr, and a cone gas flow of 100 L/Hr. For MS/MS analysis, a selection mass window of 6 Da with a collision gas pressure of 7 mbar was employed. Collision energies between 5 and 40 V were used for ion fragmentation. Mass spectra presented are the sum of 50 consecutive 1-s scans. Average intensity values reported in this study are the result of three replicate sample runs. Both a DE-pro Matrix-Assisted Laser Desorption/Ionization (MALDI) and a MALDI–Time-of-Flight-Time-of-Flight (MALDI-TOF-TOF) 4700 (both MALDI are fromApplied Biosystems Framingham, MA) were used. All spectra were acquired in positive ion-mode. Both instruments were used to acquire spectra in linear mode and the MALDI-TOF-TOF was used for the collision-induced dissociation work (selection mass window of 5 Da). The MALDI matrix, 6-aza-2-thiothymine (ATT) was purchased from Aldrich (Milwaukee, WI) and prepared fresh daily as a saturated solution in 50% ethanol. A 0.1 or 0.3 µL peptide mixture + a 0.1 or 0.3 µL matrix (ATT) were applied to the MALDI target and allowed to air-dry prior to introduction into the mass spectrometer.

We chose the acidic peptide for normalizing the complexes because the basic ones always ionize better and have a better relative abundance mainly 100%. Thus, to partially nullify the effect of suppression and for lack of a better way to normalize we chose to divide the RA of the complex by the RA of its acidic peptide.

Peptides

VLRRRRKRVN, HELKGVCPEPPGLDDPLAQDGAVGS, SAQEpSQGNT, RRRRKRVNpTKRpSSR, SFKRRRSSK, SFKRRRSpSK, GpSSEDLKKEEA, and biotinylated VLRRRRKRVN were synthesized by the Sequencing and Synthesis Laboratory at the Johns Hopkins School of Medicine.

Sample Preparation

The amount deposited in the experiment shown in Figure 1 are VLRRRRKRVN (25 pmoles), HELKGVCPEPPGLDDPLAQDGAVGS (50 pmoles) are SAQEpSQGNT (50 pmoles). The spectrum was aquired in positive reflectron mode on an ABI TOF-TOF and is the average of 400 Laser shots. The amount deposited in the experiment shown in Figure 2b are VLRRRRKRVN (25 pmoles), SAQEpSQGNT (50 pmoles). The spectrum was aquired on an ABI TOF-TOF, the CID was done using air and is the average of 1000 laser shots. Figure 2C and D data was obtained on the Q-TOF using a flow rate of 5 µL/min, each microliter contained VLRRRRKRVN (1 pmole) and SAQEpSQGNT (15 pmoles). The amount deposited in the experiment shown in Figure 4a SFKRRRSSK (10 pmoles), SFKRRRSpSK (10 pmoles), GpSSEDLKKEEA (10 pmoles). The spectrum was aquired in linear positive mode on an ABI DE-pro and is the average of 50 Laser shots. The amount deposited in the experiment shown in Figure 4b are SAQEpSQGNT (100 pmoles), VLRRRRKRVN (25 pmoles), RRRRKRVNpTKRpSSR (25 pmoles). The spectrum was aquired in positive reflectron mode on an ABI TOF-TOF and is the average of 400 Laser shots.

Figure 1
MALDI spectrum showing the interaction of the basic epitope from the third intracellular loop of the D2 receptor VLRRRRKRVN with the two possible acidic epitopes from the carboxyl terminus of the A2A receptor, SAQEpSQGNT and HELKGVCPEPPGLDDPLAQDGAGVS ...
Figure 2
Mass spectrometric collision-induced dissociation (CID) of the NCX formed by the D2 receptor-epitope VLRRRRKRVN with the A2A receptor-epitope SAQEpSQGNT. (a) Cartoon showing the electrostatic interaction between the guanidinium group of Arg (blue) and ...
Figure 4
MALDI spectra showing the formation of noncovalent complexes (NCXs) between the epitopes of the D1 and A2A receptors with the phosphorylated and nonphosphorylated epitopes of the NR1–1 subunit of the NMDA and D2 receptors, respectively. (a) The ...

Plasmon Resonance

Surface plasmon resonance measurements were made using Biacore 3000 with sensor chip SA or custom sensor chip Neutravidin CM4. The running buffer was HBS-P (Biacore, BR-1001–87) containing 10 mM HEPES, pH 7.4, 150 mM NaCl, 0.005% Surfactant P20. The regeneration buffers were 0.05% SDS, and HBS-P adjusted to 500 mM NaCl. Neutravidin (Pierce, Rockford, IL) was immobilized to sensor chip CM4 using amine coupling via an immobilization wizard. Sensor chip CM4 (BR-1000–39) has a carboxymethylated dextran hydrogel containing approximately 30% less carboxymethlyation than sensor chip CM5. Reactive N-hydroxysuccinimide (NHS) esters were introduced on the surface through modification of the carboxymethyl groups by an 8 min injection of a mixed solution of 0.4 M EDC (1-Ethyl-3-(3-dimethylaminopropyl)-carbodiimide) and 0.1 M NHS (BR-1000–50). The Neutavidin was diluted in 10 mM sodium acetate pH 5.0 (BR-1003–51) to a final concentration of 50 µg/mL and injected for 10 min. After the coupling step, any remaining reactive esters on the surface were blocked with a 7-min injection of 1 M ethanolamine, pH 8.5. The flow rate was 10 µl/min and the detection temperature was 25 °C. Using this procedure, immobilization densities for Neutravidin ranged from 9000 to 10 000 RU. The biotinylated-VLRRRRKRVN peptide was diluted to 100 µg/mL in HBS-P and injected for 5 min over one flow cell. The amount of peptide captured using this condition was 1500 RU (resonance Units). Biotin binding sites on the reference surface and any remaining free sites on the active surface were blocked with free biotin. A kinetics application wizard was employed for measuring the peptide–peptide interactions. Before running the method, the newly prepared sensor chip surface was preconditioned by three 30-s injections of the regeneration buffers. The preconditioning step ensures that any change in the ligand density or activity due to regeneration occurs before the assay is initiated. SAQEpSQGNT was diluted into HBS-P to concentrations of 0.3, 0.612, 1.25, 2.5, 5, and 10 µM. Each peptide concentration was injected at a flow rate of 30 µL/min over the reference and active surfaces for 3 min, and the complex decay was measured for 3 min. Regeneration of the surface was achieved by a 30-s injection of 0.05% SDS followed by a 30 s injection of 500 mM NaCl HBS-P at a flow rate of 30 µL/min. The detection temperature was 25 °C. All data processing and analysis were performed using Biacore’s BIA-evaluation software version 4.1.

3. Results

Basics of the Arg-Phosphate Electrostatic Interaction

We previously used soft ionization techniques (MALDI and ESI–MS) to demonstrate the ionization of noncovalent complexes (NCXs) made by salt bridge formation between peptides containing two or more adjacent basic residues (e.g., Arg-Arg) and peptides containing two or more adjacent acidic residues, such as glutamate-glutamate (Glu-Glu) or aspartate-aspartate (Asp-Asp) or a phosphorylated Ser or threonine (Thr) residues.1113 We also showed that a phosphate group forms a more stable interaction with the Arg guanidinium group than two adjacent carboxyl groups from either Asp or Glu.11 Although we previously suggested that adjacent Asp (DD) in the C-terminus of the A2A receptor could possibly be the site of interaction with the D2 receptor, the site was not conserved and as shown in the mass spectrum in Figure 1, the relative abundance (RA) of the NCXs that the Arg-rich D2 receptor epitope (215VLRRRRKRVN224) did form with the A2A receptor epitope containing the pSer374 (370SAQEpSQGNT378) was far higher than that formed with the A2A receptor epitope containing the adjacent Asp residues (388HELKGVCPEPPGLDDPLAQDGAVGS412), even though the epitope containing the adjacent Asp is more electronegative than the phosphorylated epitope (total charge at physiological pH, 7.4: −4 and −3, respectively; GPMAW 6.1 software, Denmark).

The guanidinium group at the terminus of the Arg side chain defines a plane in which the central carbon, three nitrogen and five hydrogens all reside (Figure 2a). It has a delocalized positive charge which is distributed over the entire group, resulting in triple resonance stabilization. Its pKa is 12.5 and hence it is ionized over the entire pH range in which proteins exist in nature. The two highest occupied nonbonding molecular orbitals of the guanidinium group are degenerate in energy, thus distributing two pairs of electrons evenly over the three nitrogen atoms as shown by its resonance. The net positive charge can be neutralized by removing a proton. On the other hand, the phosphate in phospho-Ser δπ molecular orbital systems spreads the negative charge over two or more oxygens because their acid dissociation constants are quite acidic (pKa1 = 0.9 and pKa2 = 6.1) due to inductive electron withdrawal. Guanidine-anion binding is often hydrogen bond driven, an enthalpically favorable event, and offers a balance of favorable attraction between the epitopes (Figure 2a). Another possible arrangement could take into account the fact that the guanidinium cation is flat and trigonal, and the phosphate ester is tetrahedral about the phosphorus, with three equivalent oxygen atoms sharing one or two negative charges. Looking at the P atom exactly opposite (180°) from the P-0 ester bond to serine, the three negatively charged oxygens appear to be 120° apart, exactly matching the 120° disposition of the partially positive N atoms of the guanidinium ion. So possibly the best positioning of these groups might be an arrangement in which one oxygen sits exactly above one nitrogen.

Stability of the Arg-Phosphate Electrostatic Interaction

In previous work, we have shown that the residues involved in the electrostatic interactions are resistant to enzymatic digests as they are not accessible for cleavage.12 A mixture of the D2 receptor epitope 215VLRRRRKRVN224 (1351.9 Da) and the A2A receptor epitope 370SAQEpSQGNT378 (1000.3 Da) was analyzed using both Matrix Assisted Laser Desorption/Ionization (MALDI) and ElectroSpray Ionization (ESI) Mass Spectrometry (MS). The molecular ion (MH+) of the NCX was fragmented by collision-induced dissociation (CID) in the MALDI Time-of-Flight-Time-of-Flight (TOF-TOF) and the ESI Quadrupole-Time-of-Flight (Q-TOF) instruments. The MALDI CID spectrum showed no fragments of the parent peptides, just their MH+ at m/z 1001.34 and 1352.87 as well as the NCX at m/z 2353.23. The base peak (RA 100%) in that spectrum was that of the Argrich epitope. The spectrum was very similar to the one previously obtained in linear mode without CID,8 although the NCX peak had a lower RA, and a new peak with a RA of 63% appeared at m/z 1433.25 (Figure 2b). The MH+ of the new peak is 80 units larger than the MH+ of the basic peptide, which corresponds to the D2 receptor epitope + HPO3 from the pSer of the A2A receptor epitope (Figure 2a). The ESI CID spectrum of the NCX showed the same MH+ as the ones seen in the MALDI spectrum (data not shown). An increased dissociation of the NCX was observed in the MALDI by progressively increasing the fluency of the laser (data not shown). While the RA of both D2 receptor-epitope and D2 receptor-epitope + HPO3 increased for a while, the former kept increasing while the later eventually started to decrease. With ESI, it was observed that there was very little dissociation of the NCX below 25 V, from 25 to 30 V the dissociation increased and the RA of the D2 receptor-epitope and D2 receptor-epitope + HPO3 kept increasing. At 35 V, the RA of the D2 receptor epitope remained the same while that of D2 receptor-epitope + HPO3 started to decrease, and at 40 V the RA of both the D2 receptor epitope and the D2 receptor epitope + HPO3 was significantly decreased (Figure 2c and 2d). Thus, these MALDI and ESI CID data show that the epitopes forming the NCX do not fragment into amino acid residues, but instead they dissociate into the parent peptides. When subjected to CID, the parent peptides totally fragmented at a voltage of 25 V. Thus, we show that the complexes stability was such that at voltages higher than the ones that completely fragment the parent peptides, the complex only dissociates. Thermodynamics principles dictate that a covalent bond is far more stable than a noncovalent electrostatic bond. However, our experimental data showed that the phosphate–guanidinium interaction was so stable that the covalent bond between the oxygen from the phosphorylated serine and HPO3 fragmented before the noncovalent complex between the guanidinium and the phosphate was disrupted (Figure 2a,b), as shown by the MH+ at m/z 1433.25.

The D2 receptor-epitope was immobilized on a neutravidin surface and solutions of the A2A receptor-epitope at concentrations of 10, 5, 2.5, 1.25, 0.612, and 0.3 µM were injected (plasmon resonance technology). Association and dissociation responses were each measured for 3 min. A specific and stable interaction between the A2A receptor epitope and the D2 receptor epitope was observed. The KD (k(off)/k(a)) was 55 µM, the on rate was relatively slow (k(a): 11(M−1s−1)) and the off rate was extremely slow (kd: 6.1 e−4 (s−1)). The binding responses were fitted to the 1:1 Langmuir model and the data implied that the binding stoichiometries could be greater than 1:1, or bi-phasic association (Figure 3). Considering that the Arg-rich epitope contains 4 adjacent Arg, it is likely that more than one acidic peptide could have interacted with each basic one. Thus using plasmon resonance technology, we have been able to confirm specific and stable interactions between epitopes whose binding appears to be driven by electrostatic attraction.

Figure 3
Plasmon Resonance sensogram. The D2 receptor epitope VLRRRRKRVN was biotinylated and attached to the chip surface, while increasing concentrations (10, 5, 2.5, 1.25, 0.612, and 0.3 µM) of the phosphorylated A2A receptor epitope SAQEpSQGNT were ...

Phosphorylation-Mediated Regulation of the Arg–Phosphate Electrostatic Interaction

The main signal transduction pathway of D1 receptors involves PKA activation and it has been shown that D1 receptor stimulation can modulate NMDA receptor function through PKA-mediated phosphorylation of Ser897 of the NR1-1 subunit.14 Lee et al.6 showed that D1 receptor stimulation decreases the NMDA-D1 receptor heteromerization involving the NR1-1 subunit. Since Ser897 is adjacent to the Arg-rich epitope of the NR1-1 subunit, we tested the possibility that phosphorylation of Ser897 could decrease the epitope–epitope electrostatic interaction involved in NMDA-D1 receptor heteromerization. In fact, as shown in Figure 4a, the addition of a phosphate group to Ser897 significantly decreased the ability of the Arg-rich epitope of the NR1-1 subunit (890SFKRRRSpSK898) to form NCXs with the D1 receptor epitope containing pSer397 (396GpSSEDLKKEEA408). The analysis of the total charge of these peptides (at physiological pH, 7.4) showed the significant electronegativity of the D1 receptor epitope (total charge: −4) and the significant electropositivity of the Arg-rich NR1-1 subunit epitope (total charge: +5, without phosphorylation of Ser397), which most probably determines the ability of these epitopes to approach each other and form heteromers through epitope-epitope electrostatic interaction. The reduced number of NCXs with the phosphorylated NR1-1 subunit epitope is most probably due to the reduction of the total electropositive charge after phosphorylation (total charge: +3). These results strongly suggest that D1 receptor-mediated phosphorylation of Ser897 decreases NMDA-D1 receptor heteromerization by decreasing the epitope–epitope electrostatic attraction between the C termini of the D1 receptor and the NR1-1 subunit of the NMDA receptor (see below).

Close inspection of the residues adjacent to the Arg-rich epitope of the D2 receptor involved in A2A-D2 receptor heteromerization uncovered a Thr residue (Thr225) and a Ser residue (Ser228) both susceptible to phosphorylation by PKA (Swiss Prot “Net Phos” Program).15 As expected, the addition of a phosphate group to Thr225 and Ser228 significantly decreased the ability of the Arg-rich epitope of the D2 receptor to form NCXs with the A2A receptor epitope containing pSer374 (Figure 4b). Again, the decreased number of NCXs might be due to the phosphorylation-induced decrease of the electropositivity of the Arg-rich domain of the D2 receptor (the total charges for 370SAQEpSQGNT378, 215VLRRRRKRVN224 and 213RRRRKRVNpTKRpSSR230 were −3, +6 and +5, respectively). As for the D1 receptor, the main signal transduction pathway of A2A receptors involves the activation of PKA.16 Although it remains to be demonstrated, our results predict that stimulation of A2A receptors leads to a decrease in the formation of A2A-D2 receptor heteromers. In conclusion, its covalent-like stability and its regulation by phosphorylation makes the Arg-phosphate bond a very likely general phenomenon involved in protein–protein interactions.

Proposed Mechanistic Model for Receptor Heteromerization Involving the Arg–Phosphate Electrostatic Interactions

The A2A, D2 and D1 receptors belong to the of GPCR family, with seven transmembrane domains.2,16 The A2A and D1 receptors have similar structures, with a long C terminus, and the D2 receptor is characterized by a long third intracellular loop (Figure 5). The NR1-1 subunit of the NMDA receptor (a ligand-gated ion channel) contains four transmembrane domains (although the second can only be considered partially transmembrane) and long C and N termini17 (Figure 5). The epitopes involved in A2A-D2 and NMDA-D1 receptor heteromerization are situated in their largest intracellular domains (Figure 5). The analysis of the total charge of these domains (at physiological pH, 7.4) showed the D2 receptor domain to be significantly electropositive with a total charge of +15 for the first 90 amino acids in the N-terminal portion of the third intracellular loop, while the NR1-1 subunit of the NMDA receptor has a total charge of +12 for the last 90 amino acids of the C terminus. On the other hand, a significant electronegativity was found in the corresponding domains of the A2A receptor with a total charge of −9 for the last 90 amino acids of the C-terminus (including pSer374) while the D1 receptor has a total charge of −7 for the last 90 amino acids of the C-terminus (including pSer397). This creates optimal conditions for a mutual approach of the domains containing the interacting epitopes. PKA-mediated phosphorylation of the Ser or Thr residues located in the vicinity of the Arg-rich epitopes of the D2 receptor and NR1–1 unit of the NMDA receptor engenders a decrease in the total electropositivity of the domain (2 less positive charges per phosphorylated residue), which would decrease the attraction to the respective electronegative domains of the C termini of the A2A and D1 receptors, thus decreasing A2A-D2 and NMDA-D1 receptor heteromerization.

Figure 5
Scheme of the structure of the GPCRs A2A, D2, and D1 and the NR1-1 subunit of the ligand-gated ion channel NMDA receptor. The acidic epitopes of both the A2A and D1 receptors (dark red) are localized in an electronegative domain (light red) of their characteristic ...

4. Discussion

Most physiological events require proteins to interact, as in receptor-G protein cascades and receptor heteromerization. Protein folding is paramount for such functions and, hence, the importance of attractive forces in protein–protein interactions. We have previously shown that the mechanism involved in some receptor heteromerizations is due to the linear alignment of the guanidinium groups of adjacent Arg on one epitope and a phosphorylated amino acid residue on the opposing epitope, resulting in salt bridge formation through electrostatic attraction. The present work highlights the importance of phosphorylation–dephosphorylation events in the modulation of this electrostatic interaction. Thus, phosphorylation of the CK1 consensus site on the acidic epitope makes it available to interact with the Arg-rich basic epitope. On the other hand, phosphorylation of Ser and Thr residues adjacent to the basic epitope, a PKA consensus site, will slow the attraction between the epitopes. In addition, the collision-induced dissociation data show that once the noncovalent bond is established, it leads to a highly thermodynamically favorable conformation, as the noncovalently bound phosphate breaks its covalent bond to the Ser and remains noncovalently bound to the basic epitope. Hence, under the right electrostatic conditions, the Arg-phosphate noncovalent bond possesses a “covalent-like” stability. Although in the present study such bond is analyzed in the frame of receptor heteromerization, epitope–epitope electrostatic interactions most likely represent a general mechanism underlying many biological processes. Emphasizing the importance of Arg and Ser are the following facts: Ninety percent of phosphorylated residues are Ser, and the genetic code shows that Arg and Ser share the highest level of degeneracy with Leucine (six codons).

Acknowledgment

We would like to thank Drs. Roy Wise, Barry Hoffer, Jerold Meinwald, Shelley Jackson and HayYan Wang for intellectual input, Dr. Irina Orlova and Michael Murphy from Biacore Inc., for their help with the plasmon resonance data and ONDCP for instruments funding, without which this and other projects could not have been done.

References

1. Bouvier M. Nat. Neurosci. 2001;2:274–286. [PubMed]
2. Agnati LF, Ferré S, Lluis C, Franco R, Fuxe K. Pharmacol. Rev. 2003;55:509–550. [PubMed]
3. Hillion J, Canals M, Torvinen M, Casado V, Scott R, Terasmaa A, Hansson A, Watson S, Olah ME, Mallol J, Canela EI, Zoli M, Agnati LF, Ibanez CF, Lluis C, Franco R, Ferré S, Fuxe K. J. Biol. Chem. 2002;277:18091–18097. [PubMed]
4. Canals M, Marcellino D, Fanelli F, Ciruela F, de Benedetti P, Goldberg SR, Neve K, Fuxe K, Agnati LF, Woods AS, Ferré S, Lluis C, Bouvier M, Franco R. J. Biol. Chem. 2003;278:46741–16749. [PubMed]
5. Kamiya T, Saitoh O, Yoshioka K, Nakata H. Biochem. Biophys. Res. Commun. 2003;306:544–549. [PubMed]
6. Lee FJ, Xue S, Pei L, Vukusic B, Chery N, Wang Y, Wang YT, Niznik HB, Yu XM, Liu F. Cell. 2002;111:219–230. [PubMed]
7. Fiorentini C, Gardoni F, Spano P, Di Luca M, Missale C. J. Biol. Chem. 2003;278:20196–20202. [PubMed]
8. Ciruela F, Burgueno J, Casado V, Canals M, Marcellino D, Goldberg SR, Bader M, Fuxe K, Agnati LF, Lluis C, Franco R, Ferré S, Woods AS. Anal. Chem. 2004;76:5354–5363. [PubMed]
9. Woods AS, Ciruela F, Fuxe K, Agnati LF, Lluis C, Franco R, Ferré S. J. Mol. Neurosci. 2005;26:125–132. [PubMed]
10. Ferré S, Ciruela F, Canals M, Marcellino D, Burgueno J, Casado V, Hillion J, Torvinen M, Fanelli F, de Benedetti P, Goldberg SR, Bouvier M, Fuxe K, Agnati LF, Lluis C, Franco R, Woods AS. Parkinsonism. Relat. Disord. 2004;10:265–271. [PubMed]
11. Woods AS. J. Proteome. Res. 2004;3:478–484. [PubMed]
12. Woods AS, Huestis MA. J. Am. Soc. Mass. Spectrom. 2001;12:88–96. [PubMed]
13. Woods AS, Koomen JM, Ruotolo BT, Gillig KJ, Russel DH, Fuhrer K, Gonin M, Egan TF, Schultz JA. J. Am. Soc. Mass. Spectrom. 2002;13:166–169. [PubMed]
14. Dudman JT, Eaton ME, Rajadhyaksha A, Macias W, Taher M, Barczak A, Kameyama K, Huganir R, Konradi C. J. Neurochem. 2003;87:922–934. [PubMed]
15. Blom N, Gammeltoft S, Brunak S. J. Mol. Biol. 1999;294:1351–1362. [PubMed]
16. Ferré S, Fredholm BB, Morelli M, Popoli P, Fuxe K. Trends Neurol. Sci. 1997;20:482–487. [PubMed]
17. Dingledine R, Borges K, Bowie D, Traynelis SF. Pharmacol. Rev. 1999;51:7–61. [PubMed]