ETD and sequential ETD localize the residues involved in D2-A2A heteromerization

Abstract

Certain amino acid residues and posttranslational modifications play an important role in the formation of noncovalent complexes (NCXs) by electrostatic interactions. Electrospray ionization mass spectrometry (ESI-MS) is the most widely used MS technique for the study of NCXs, due to its softer ionization process and compatibility with the solution phase of NCX mixtures. In order to locate the site where interactions are forming in NCXs involving phosphopeptides and adjacent arginines, tandem mass spectrometry studies using collision-induced dissociation (CID) and electron transfer dissociation (ETD) were performed on NCXs at different charge states. CID fragmentation revealed two dissociation pathways: one in which the electrostatic interaction is disrupted and another in which the covalent bond attaching the phosphate group to the amino acid residue is cleaved, while the electrostatic interaction is maintained. ETD and sequential ETD/ETD, and CID/ETD allow the determination of the NCX interaction site. These results confirmed the involvement of the phosphorylated amino acid and at least two adjacent arginines as the binding site.

---

Introduction

Noncovalent complexes (NCXs) are formed from a wide variety of interactions that include hydrophobic interactions,¹ electrostatic (coulombic) interactions,^2,3 and hydrogen bonding,^4,5 which is actually a weak electrostatic interaction.⁶ Electrostatic interactions are of particular importance in determining the quaternary structure of interacting molecules and determining their conformation which gives specificity to the proteins involved⁷ such as in receptor heteromerization.^8,9 Certain amino acid residues and posttranslational modifications play an important role in the formation of NCXs by electrostatic interactions.^6,10,11 The use of mass spectrometry to study NCXs in biological systems has sky rocketed in recent years.^6,12,13 Mass spectrometry is well-suited to study electrostatic interactions as they are strengthened and stabilized in the gas phase when compared to solution phase, due to the dielectric constant of the solvent, which weakens electrostatic interactions in solution phase.^6,14 Electrospray ionization mass spectrometry (ESI-MS) is the most widely used MS technique for the study of NCXs, due to its softer ionization process and compatibility with the solution phase of NCX mixtures. ESI-MS has been successfully used to analyze protein complexes,^14–16 protein–peptide complexes,^17,18 NCXs of peptides,^19–21 NCXs of DNA and RNA,^22–24 and NCXs in drug discovery.²⁵

Several studies using mass spectrometry have probed the electrostatic interaction forming NCXs to determine its location and stability/strength. First, NCXs enzymatic digests have been conducted to determine the possible binding sites for peptide–peptide^2,26 and protein–peptide²⁷ interactions. However, the interpretation of results can be complicated due to the plethora of peaks and background ions resulting from the digest, although it narrows down the possible location of the electrostatic interaction. Tandem MS experiments using collision-induced dissociation (CID) have analyzed the gas-phase stability of NCXs of different molecules, in order to compare the strength of the noncovalent bond holding complexes together.^20,21,28,29 Furthermore, results from studies analyzing the fragmentation of NCXs formed with phospho/sulfopeptides, using CID revealed two dissociation pathways for these complexes: one in which the electrostatic interaction is disrupted leading to the two intact partners and another in which the covalent bond attaching the phosphate/sulfate group to the amino acid residue is cleaved, while the electrostatic interaction is maintained.^19,21,30 Similar dissociation pathways in which the noncovalent bond was maintained, has been observed in NCXs between oligonucleotide/polybasic compounds,³¹ ribonuclease–nucleotide³² and DNA/peptides.³³ While these results confirm the role of certain amino acid residues and chemical groups (including phosphate/sulfate), in the formation, stability and strength of NCXs, they do not allow the assignment of the specific sites involved in the electrostatic interactions, especially when more than one possible binding site is present on the domains involved.

CID fragmentation can be complemented by using either electron capture dissociation (ECD) or electron transfer dissociation (ETD), which cleave the peptide backbone while side chains and post-translational modifications, such as phosphorylation, are left intact.^34–36 In one previous study, ECD was conducted on weakly bound polypeptide complexes and produced some intramolecular fragmentation ions without weak bond dissociation.³⁷ Our group showed how ETD can be an effective tool for the identification of the specific amino acid residues involved in electrostatic interactions (NCXs) between peptides.³⁸ ECD has recently been used to elucidate the site of binding between ATP and protein.³⁹ In this study, we investigated the interaction between epitopes from D2 and A2A receptors, known to be biologically involved via receptor heteromerization.^9,40 Both CID and ETD were conducted on the NCX, and the use of sequential ETD experiments on NCX fragments will be introduced to identify the site of interaction when multiple sites of interaction are possible.

Materials and methods

Chemicals

The basic epitope RRRRKRVNTKRSSR (R₄), [C₇₃H₁₄₂N₃₈O₁₉, 1855.13 u] a linear motif in the third intracellular loop of the D2 receptor, and the acidic epitope SAQEpSQGNT (pS), [C₃₄H₅₉N₁₂O₁₉P, 1000.35 u] a linear motif located in the carboxy terminus of the Adenosine A2A receptor, where p denotes a phosphorylated residue, were synthesized at the Johns Hopkins School of Medicine Sequencing and Synthesis Laboratory (Baltimore MD). EtOH was purchased from the Warner-Graham Company (Cockeysville, MD).

Sample preparation and analysis

Peptides stock solutions were prepared in water at a concentration of 1 nmol μL⁻¹. A mixture of 1 : 4 (basic : acidic) ratio was prepared and diluted in 50% ethanol prior to mass analysis.

Mass spectrometer

Analysis was conducted on a LTQ Orbitrap Velos mass spectrometer (Thermo Electron, San Jose, CA) with a static nanospray source in positive ion mode. A spray voltage of 1.5 kV was used with 2 μm nanospray tips (New Objective, Woburn, MA). A mass selection window of 2 mass units was used for MSⁿ analysis. CID experiments were conducted with collision energies of 12 to 17. ETD experiments were conducted via a modification involving a chemical ionization source at the rear of the instrument. Fluoranthene ions were used as the ETD reagent. The anion target was 5e⁶ and the activation time for the ETD reaction was 100 ms. All mass spectra in this study, are the sum of 25 to 50 scans. The mass resolution was set at 60 000 for an m/z 400. The MS³ sequential experiments were performed in the linear ion trap. To simplify notations, protonated species [M + nH]ⁿ⁺ are labeled as Mⁿ⁺.

Results and discussion

In order to investigate the stability of the interaction involved in NCX formation, a mixture of two peptide R₄ and pS, known to form complexes, was analyzed by ESI-MS. Fig. 1A shows an ESI-MS spectrum for a sample mixture of R₄ and pS in positive ion mode. The major peaks observed correspond to the peptide R₄ with 3 to 6 charges and pS with one charge. The multiple charges on R₄ are expected since it contains seven arginines residues. The NCX formed between R₄ and pS is also observed with 4 to 6 positive charges. Fig. 1B shows the CID MS/MS spectrum for the NCX⁴⁺observed in Fig. 1A at m/z 715.13 Da. The dissociation of the NCX⁴⁺ peak resulted in a singly charged ion corresponding to pS (m/z 1001.36 Da) and a very intense triply charged ion attributed to R₄³⁺ (m/z 619.38 Da). This result suggests an uneven sharing of charges between the two partners of the complex when disrupted by CID. The positive charges stay mainly on the basic epitopes. These fragment ions result from the disruption of the electrostatic interaction between the two peptides. A second pathway, cleaving the covalent bond between the oxygen and the phosphate lead to the formation of the ion fragments at m/z 646.04 Da corresponding to [R₄ + HPO₃]³⁺, and [pS-HPO₃]⁺ at 921.39 Da. This type of fragmentation for NCXs has been reported in previous studies.^19,21,30,41 An additional fragment ion from the acidic [pS-H₃PO₄]⁺ at 903.38 Da can be observed as a consecutive loss of water from [pS-HPO₃]⁺ or the loss of H₃PO₄ from pS⁺.

Fig. 1 Ion mass spectrum (A) from a peptide mixture of SAQEpSQGNT(pS) and RRRRKRVNTKRSSR(R₄) and product-ion spectra of the [NCX + 4H]⁴⁺with (B) CID and (C) ETD.

The ETD product-ion spectrum from the NCX⁴⁺, in Fig. 1C, shows a very different fragmentation of the complex compared to the CID. Unlike CID, the NCX did not dissociate into the parent peptides, but mainly produced c and z ions, fragmenting along both peptides. Intramolecular fragments which conserved the electrostatic bond have been marked with an asterisk *. These fragment ions allowed the identification of the amino acids involved in the binding. For the basic peptide (R₄) the ions of the c serie (c^*₄–c^*₁₃, color purple) and the z serie (z^′*₁₂–z^′*₁₄, color pink), that were interacting with the phosphate, all contained the two adjacent arginines (Arg₃ (3^rd arginine of the peptides) and Arg₄). Likewise, the fragment ions from the acidic peptide (pS): c^*₅–c^*₈ (color orange) and z^′*₅–z^′*₉ (color green) that were interacting with R₄, contained the phosphoserine. Other complementary fragment ions from R₄ are also present in the spectrum (c and z′ in blue) but do not offer any information on the interaction site of the complex.

CID and ETD experiments were also conducted on both NCX⁵⁺ and NCX⁶⁺ (Fig. 2). The CID spectra again show two fragmentation pathways. The first one is the complex dissociation, leading to pS at m/z 1001.36 and R₄ at m/z 464.79 in Fig. 2A (corresponding to R₄⁴⁺) or at m/z 372.03 in Fig. 2B (corresponding to R₄⁵⁺). As observed with NCX⁴⁺, the charges are not evenly shared in this complex as most of them are on the basic epitope. For every charge state tested, pS is singly charged and R₄ is multiply charged (one charge less than the complex), which is understandable considering the number of arginines in R₄. Finally, as for the NCX⁴⁺, another fragmentation pathway is observed, in which an additional mass of 79.97 u (corresponding to HPO₃) is seen on the basic peptide [R₄ + HPO₃]. The relative intensity of this peak (R₄ + HPO₃), compared to R₄, shows a net decrease when the number of positive charges increase. An explanation, that we proposed,¹⁹ is that as the number of charges on the complex increases the stabilization of the guanidinium group and decreases the deprotonation of the phosphate group, thus weakening the electrostatic bond.

Fig. 2 CID-product-ion spectra from a peptide mixture of SAQEpSQGNT and RRRRKRVNTKRSSR of the (A) [NCX + 5H]⁵⁺ and the (B) [NCX + 6H]⁶⁺. ETD-product-ion spectra of (C) [NCX + 5H]⁵⁺and ⁺(D) [NCX + 6H]⁶⁺.

ETD spectra were also recorded for NCX⁵⁺ (Fig. 2C) and NCX⁶⁺ (Fig. 2D). Similar to the ETD mass spectrum of the NCX⁴⁺ ion (Fig. 1C), numerous ion fragments are observed from the cleavage of peptide bonds in both epitopes (ESI†). Ion fragments from the basic peptide which is still interacting with pS (c^*₄–c^*₁₃ and z^′*₁₂–z^′*₁₄ respectively in purple and pink) and their complementary fragment ions (c₁–c₂ and z^′₂–z^′₁₀) are observed. Basic peptides fragments still involved in the interaction always contain Arg₃ (3^rd arginine of the peptides) and contain at least two adjacent arginines. To form the complex, we previously showed that two arginines were required,¹⁰ these results confirm it and show that they are essential for the complex stability. Overall ETD fragments from R₄ are more intense with the NCX⁶⁺ and NCX⁵⁺ than with the NCX⁴⁺, this can be explained by the fact, that as observed in CID, and as expected, most of the charges are on the basic epitopes. These charges allow us to better see the fragment ions from R₄ by mass spectrometry and make the cleavage by ETD easier. However, when looking at the acidic peptides, ETD results in fewer fragments (Fig. 2C and D). Ion fragments (c^*₈ and z^′*₆–z^′*₉) still interacting with R₄ are seen in both spectra. The diagnostic ion c^*₇ is present only in the ETD spectrum of NCX⁶⁺. Surprisingly, we have less coverage on the acidic peptides than with the NCX⁴⁺. This lack of fragmentation from the acidic peptide is an issue in determining the binding site and is explained by the higher number of charges on the basic epitope.

To have access to the binding site, MS³ sequential ETD (ETD/ETD) were attempted on the ETD-fragment [c₄ + SAQEpSQGNT + 2H]²⁺ at m/z 821.90 from the [NCX + 5H]⁵⁺ (Fig. 3A). The parent ion is only doubly charged, for a better visibility of the intensity of the fragments the mass range 1000–1700 has been multiplied by 3. c- and z-ions fragments from the acidic peptide (c^*₅–c^*₈ and z^′*₅–z^′*₈) where the interaction was intact with c₄ from R₄ are seen. This pattern correspond exactly to the one obtained with the [NCX + 4H]⁴⁺ (Fig. 1C), confirming that the phosphoserine is the preferred interaction site with the c₄ fragment. c₄ is able to lose one of the arginine from each side of the chain. The same experiment was conducted on the ETD-fragment [c₃ + SAQEpSQGNT + 2H]²⁺ at m/z 743.85 from the [NCX + 6H]⁶⁺ confirming the phosphoserine as the preferred site of interaction (data not shown). Sequential ETD of the ion fragment (R₄ + HPO₃)⁴⁺ from the CID of NCX⁵⁺ was also performed to confirm the binding site on the basic epitope. The CID/ETD spectrum is presented in Fig. 3B, and shows the cleavage of peptide bonds on R₄. Ion fragments still interacting with the phosphate (c^*₃–c^*₁₃ and z^′*₁₂–z^′*₁₄ respectively in purple and pink) and their complementary fragment ions (c₁–c₃ and z^′₂–z^′₁₀) are observed. The same series of ions are observed in Fig. 2A and B, confirming the Arg₃ as the most favorable site on the basic epitope. All ETD fragment ions, conserving the interaction, possess at least two adjacent arginines.

Fig. 3 (A) Sequential ETD/ETD-product-ion spectra from [c₄ + SAQEpSQGNT + 2H]²⁺ at m/z 821.90 from the ETD of the [NCX + 5H]⁵⁺. (B) Sequential CID/ETD-product-ion spectra of the [RRRRKRVNTKRSSR + HPO₃ + 4H ]⁴⁺ at m/z 484.78 from the CID of the [NCX + 5H]⁵⁺.

Conclusions

CID experiments show that the electrostatic interaction is so stable that often a covalent bond will break before the electrostatic bond is disturbed .This work demonstrates the utility of ETD for studying NCX by generating intermolecular fragments from both partners. Similar to CID, ETD spectra are highly dependent on the charge states of the parent ions. The comparison of the product-ion mass spectra at different charge states shows less information at higher charges as it leads to less fragmentation from the acidic partner. To overcome this issue, the determination of the interaction site of the NCX has been done using sequential ETD/ETD. We confirmed the involvement of the phosphorylated amino acid at every charge state. ETD and sequential CID/ETD confirmed that the arginines’ guanidinium groups are the binding site and that at least two adjacent arginines are needed. Future work will include using ETD and sequential ETD to probe for the interaction sites in more complex NCXs involving proteins.

References

1. D. Chandler, Interfaces and the driving force of hydrophobic assembly, Nature, 2005, 437, 640–647.
2. A. S. Woods and M. A. Huestis, A Study of Peptide-Peptide Interaction by MALDI, J. Am. Soc. Mass Spectrom., 2001, 12, 88–96.
3. A. S. Woods and A. Zangen, A direct Chemical Interaction between dynorphin and Excitatory Amino acids, Neurochem. Res., 2001, 26, 395–400.
4. E. Arunan, G. R. Desiraju, R. A. Klein, J. Sadlej, S. Scheiner, I. Alkorta, D. C. Clary, R. H. Crabtree, J. J. Dannenberg, P. Hobza, H. G. Kjaergaard, A. C. Legon, B. Mennucci and D. J. Nesbitt, Defining the hydrogen bond: An Account, Pure Appl. Chem., 2011, 83, 1619–1636.
5. A. S. Woods, J. C. Buchsbaum, T. A. Worrall, R. Cotter and J. M. Berg Matrix, Assisted Laser Desorption Ionization of Non-Covalently Bonded Compounds, Anal. Chem., 1995, 67, 4462–4465.
6. K. A. Schug and W. Lindner, Noncovalent Binding between Guanidium and Anionic Groups: Focus on Biological- and Synthetic-Based Arginine/Guanidinium Interactions with Phosph[on]ate and Sulf[on]ate Residues, Chem. Rev., 2005, 105, 67–114.
7. A. R. Fersht, Basis of biologically specificity, Trends Biochem. Sci., 1984, 9, 145–147.
8. A. S. Woods, F. Ciruela, K. Fuxe, L. F. Agnati, C. Lluis, R. Franco and S. Ferre, Role of Electrostatic Interaction in Receptor-Receptor Heteromerization, J. Mol. Neurosci., 2005, 26, 125–132.
9. A. S. Woods and S. N. Jackson, How adenylate cyclase choreographs the pas de deux of the receptors heteromerization dance, Neuroscience, 2013, 238, 335–344.
10. A. S. Woods, The Mighty Arginine, the Stable Quaternary Amines, the Powerful Aromatics and the Aggressive Phosphate: Their Role in the Noncovalent Minuet, J. Proteome Res., 2004, 3, 478–484.
11. A. S. Woods, J. Koomen, B. Ruotolo, K. J. Gillig, D. H. Russell, K. Fuhrer, M. Gonin, T. Egan and J. A. Schultz, A Study of Peptide-peptide Interactions Using MALDI Ion Mobility o-TOF and ESI-TOF Mass Spectrometry, J. Am. Soc. Mass Spectrom., 2002, 13, 166–169.
12. K. Breuker, The study of protein-ligand interactions by mass spectrometry-a personal view, Int. J. Mass Spectrom., 2004, 239, 33–41.
13. M. Sharon, How Far Can We Go with Structural Mass Spectrometry of Protein Complexes?, J. Am. Soc. Mass Spectrom., 2010, 21, 487–500.
14. J. A. Loo, Electrospray Ionization Mass Spectrometry: A Technology for Studying Noncovalent Macromolecular Complexes, Int. J. Mass Spectrom., 2000, 200, 175–186.
15. J. A. Loo, Studying Noncovalent Protein Complexes By Electrospray Ionization Mass Spectrometry, Mass Spectrom. Rev., 1997, 16, 1–23.
16. J. L. P. Benesch and C. V. Robinson, Mass spectrometry of macromolecular assemblies: preservation and dissociation, Curr. Opin. Struct. Biol., 2006, 16, 245–251.
17. T. L. Pukala, T. Urathamakul, S. J. Watt, J. L. Beck, R. J. Jackway and J. H. Bowie, Binding studies of nNOS-active amphibian peptides and Ca²⁺ calmodulin, using negative ion electrospray ionization mass spectrometry, Rapid Commun. Mass Spectrom., 2008, 22, 3501–3509.
18. A. N. Calabrese, L. A. Speechley and T. L. Pukala, Characterisation of Calmodulin Structural Transitions by ion Mobility Mass Spectrometry, Aust. J. Chem., 2012, 65, 504–511.
19. S. N. Jackson, H.-Y. J. Wang and A. S. Woods, Study of the Fragmentation Patterns of the Phosphate-Arginine Noncovalent Bond, J. Proteome Res., 2005, 4, 2360–2363.
20. S. N. Jackson, H.-Y. J. Wang, A. Yergey and A. S. Woods, Phosphate Stabilization of Intermolecular Interactions, J. Proteome Res., 2006, 5, 122–126.
21. S. N. Jackson, S. C. Moyer and A. S. Woods, The Role of Phosphorylated Residues in Peptide-Peptide Noncovalent Complexes, J. Am. Soc. Mass Spectrom., 2008, 19, 1535–1541.
22. F. Rosu, S. Pirotte, E. De Pauw and V. Gabelica, Positive and negative ion mode ESI-MS and MS/MS for studying drug-DNA complexes, Int. J. Mass Spectrom., 2006, 253, 156–171.
23. A. Burmistrova, V. Gabelica, A.-S. Duwez and E. De Pauw, Ion Mobility Spectrometry Reveals Duplex DNA Dissociation Intermediates, J. Am. Soc. Mass Spectrom., 2013, 24, 1777–1786.
24. S. A. Hofstadler and R. H. Griffey, Analysis of Noncovalent Complexes of DNA and RNA by Mass Spectrometry, Chem. Rev., 2001, 101, 377–390.
25. S. A. Hofstadler and K. A. Sannes-Lowery, Applications of ESI-MS in drug discovery: interrogation of noncovalent complexes, Nat. Rev. Drug Discovery, 2006, 5, 585–595.
26. A. S. Woods, R. Kaminski, M. Oz, Y. Wang, K. Hauser, R. Goody, H.-Y. J. Wang, S. N. Jackson, P. Zeitz, K. P. Zeitz, D. Zolkowska, R. Schepers, M. Nold, J. Danielson, A. Gräslund, V. Vukojevic, G. Bakalkin, A. Basbaum and T. Shippenberg, Decoy Peptides that Bind Dynorphin Noncovalently Prevent NMDA Receptor-Mediated Neurotoxicity, J. Proteome Res., 2006, 5, 1017–1023.
27. A. S. Woods, D. Marcellino, S. N. Jackson, R. Franco, S. Ferre, L. F. Agnati and K. Fuxe, How Calmoduline Interacts with the Adenosine A2A and the Dopamine D2 Receptors, J. Proteome Res., 2008, 7, 3428–3434.
28. M. A. Raji, P. Frycak, M. Beall, M. Sakrout, J.-M. Ahn, Y. Bao, D. W. Armstrong and K. A. Schug, Development of an ESI-MS screening method for evaluating binding affinity between integrin fragments and RGD-based peptides, Int. J. Mass Spectrom., 2007, 262, 232–240.
29. A. Delvolvé, J.-C. Tabet, S. Bregant, C. Afonso, F. Burlina and F. Fournier, Charge Dependent Behavior of PNA/DNA/PNA Triplexes in the Gas Phase, J. Mass Spectrom., 2006, 41, 1498–1508.
30. A. S. Woods, H.-Y. J. Wang and S. N. Jackson, Sulfation, the up-and-coming post-translational modification: its role and mechanism in protein-protein interaction, J. Proteome Res., 2007, 6, 1176–1182.
31. P. Terrier, J. Tortajada and W. Buchmann, A study of noncovalent complexes involving single-stranded DNA and polybasic compounds using nanospray mass spectrometry, J. Am. Soc. Mass Spectrom., 2007, 18, 346–358.
32. S. Yin, Y. Xie and J. A. Loo, Mass Spectrometry of Protein–Ligand Complexes: Enhanced Gas-Phase Stability of Ribonuclease–Nucleotide Complexes, J. Am. Soc. Mass Spectrom., 2008, 19, 1199–1208.
33. S. Alves, A. Woods, A. Delvolvé and J. C. Tabet, Influence of salt bridge interactions on the gas-phase stability of DNA/peptide complexes, Int. J. Mass Spectrom., 2008, 278, 122–128.
34. R. A. Zubarev, Electron-Capture Dissociation Tandem Mass Spectrometry, Curr. Opin. Biotechnol., 2004, 15, 12–16.
35. J. J. Coon, J. E. P. Syka, J. Shabanowitz and D. F. Hunt, Tandem Mass Spectrometry for Peptides and Protein Sequence Analysis, BioTechniques, 2005, 38, 519–523.
36. R. A. Zubarev, A. R. Zubarev and M. Savitski, Electron Capture/Transfer versus Collisionally Activated/Induced Dissociations: Solo or Duet?, J. Am. Soc. Mass Spectrom., 2008, 19, 753–761.
37. K. F. Haselmann, T. J. D. Jorgensen, B. A. Budnik, F. Jensen and R. A. Zubarev, Electron Capture Dissociation of Weakly Bound Polypeptide Polycationic Complexes, Rapid Commun. Mass Spectrom., 2002, 16, 2260–2265.
38. S. N. Jackson, S. Dutta and A. S. Woods, The Use of ECD/ETD to Identify the Site of Electrostatic Interaction in Noncovalent Complexes, J. Am. Soc. Mass Spectrom., 2009, 20, 176–179.
39. S. Yin and J. A. Loo, Elucidating the Site of Protein-ATP Binding by Top-Down Mass Spectrometry, J. Am. Soc. Mass Spectrom., 2010, 21, 899–907.
40. K. Fuxe, S. Ferré, M. Canals, M. Torvinen, A. Terasmaa, D. Marcellino, S. R. Goldberg, W. Staines, K. X. Jacobsen, C. Lluis, A. S. Woods, L. F. Agnati and R. Franco, Adenosine A2A and dopamine D2 heteromeric receptor complexes and their function, J. Mol. Neurosci., 2005, 26, 209–220.
41. A. S. Woods and S. Ferre, The Amazing Stability of the Arginine-Phosphate Electrostatic Interaction, J. Proteome Res., 2005, 4, 1397–1402.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra04757e

---