Peptide Metal Binding: Donor Sets, Constants and Methods

A peptide chain is a polydentate ligand before it is anything else. The terminal amino group, the deprotonated amide nitrogens of the backbone and various side-chain donors are the most common metal binding sites in peptide complexes, and the imidazole nitrogen of histidyl residues is the side chain that dominates: its complexes combine outstanding thermodynamic stability with a wide range of accessible structures ^[1]. Carboxylates of aspartyl and glutamyl residues and thiolates of cysteinyl residues are among the other side chains strong enough to redirect the coordination ^[1]. A sequence designed for something else will therefore bind copper, nickel or zinc out of the buffer if its donors happen to be arranged correctly, and the resulting complex brings its own spectroscopy, its own charge and — for copper and iron — its own chemistry.

Which metals can recruit the backbone

Backbone amides are very weak acids, and their deprotonation is not observed in water within the measuring range of a glass electrode unless the proton is displaced by a metal such as Cu(II) or Ni(II) ^[2]. Model peptides carrying Ala-to-Pro substitutions — Ac-AAAHAAA-NH₂ compared with Ac-AAPHAAA-NH₂ and Ac-AAPHPAA-NH₂, proline removing an amide NH from contention — established that neither Zn(II) nor Fe(II) is able to deprotonate and bind amide nitrogens ^[2]. That asymmetry organises everything downstream: Cu(II) and Ni(II) can chelate along a stretch of backbone, while Zn(II) and Fe(II) are left with the terminal amine and side-chain donors.

Thiolate is the side chain that builds discrete structural sites. Across six zinc-finger model peptides carrying Zn(Cys)_4−x(His)_x sites — a treble-clef model, a model of a redox-switch protein site, and four variants of the classical ββα finger — those that adopt a well-defined structure bound zinc with constants around 10¹⁵ at pH 7.0 ^[3]. The same study makes a point worth carrying into any design work: packing hydrophobic residues into a well-defined core raised the stability constant by up to four orders of magnitude, and the half-life of zinc exchange with EDTA varied significantly across the set ^[3]. Affinity is not a property of the donor atoms alone.

Carboxylates matter more than their individual donor strength suggests, because they extend a chelate that has already started rather than nucleating a new one. In copper(II) complexes of the peptides FDAH and VIDAH the amino group and a β-carboxylate create a very effective site with (NH₂, N⁻, β-COO⁻) and (NH₂, N⁻, N⁻, β-COO⁻) coordination, leaving the histidine site free to form a separate (N_im, N⁻, N⁻) mode, so dinuclear complexes are preferred ^[4]. In the terminally protected two-histidine pentapeptides Ac-HHVGD-NH₂ and Ac-HVGDH-NH₂ the imidazole contribution shows up instead as enhanced stability of the 3N and 4N copper(II) species ^[4].

Geometry belongs to the oxidation state rather than to the peptide. In intrinsically disordered peptides the coordination is versatile because there is no stable fold and several possible anchors are present, so each oxidation state imposes its own favoured site: with the amyloid-β peptide, Cu(I) binds linearly between two histidines while Cu(II) takes a square-based pyramid built on the Asp1 amine and carbonyl groups with two histidines in the equatorial plane ^[5]. Because the two sites are not superimposable, direct electron transfer between them is prevented and an in-between state has to be invoked ^[5].

How confidently any of this transfers to a given sequence depends on how many histidines it carries. Complex formation by simple oligopeptides containing one histidyl residue has been clarified satisfactorily and reviewed; the coordination chemistry of multihistidine peptides — the class that includes prion protein fragments, amyloid-β, and fragments of copper(II) and zinc(II) transporter proteins — is understood less well ^[1].

ATCUN as the worked example

The amino-terminal Cu(II)- and Ni(II)-binding motif is the N-terminal sequence Xaa-Zaa-His bearing the N-terminal amine, with Zaa other than proline, and it binds Cu(II) among other transition metal ions in a square-planar geometry ^[6]. Two properties make it the reference case. It is portable: NH₂-Xxx-Zzz-His and NH₂-Xxx-His sequences form well-established high-affinity Cu(II) complexes — Asp-Ala-His at the N-terminus of serum albumin and Gly-His-Lys are the canonical examples — which opens a straightforward route to adding a high-affinity Cu(II) site to almost any peptide or protein by chemical or recombinant means ^[7]. It follows from the definition that acetylating or extending the N-terminus removes the anchor, and with it the site.

It is also not instantaneous. Cu²⁺ binding to ATCUN sequences is multi-step, with partially coordinated intermediate species present within time windows spanning hundreds of milliseconds to seconds ^[6]. Double-mixing stopped-flow on DAHK, the tetrapeptide representing the motif at the albumin N-terminus, showed that the transient two-nitrogen intermediate readily exchanged Cu(II) with histidine and was rapidly reduced to Cu(I) by glutathione, unlike the final four-nitrogen square-planar complex ^[6]. Mix and measure inside that window and you are characterising a different species from the one your stability constant describes.

A stability constant without its conditions is not a number

The metal dissociation constant expresses the thermodynamic strength of a metal–protein interaction, and obtaining a reliable one turns on rational probe selection combined with prudent experimental design ^[8]. In practice that means competition against a standard whose own constant is defined under the same conditions. A set of four dansylated peptide probes was built for exactly this. DP1–DP4 each bind one equivalent of Cu(II) with different, well-separated affinities at pH 7.4 — log K_D of −8.1, −10.1, −12.3 and −14.1 — and Cu(II) quenches their fluorescence at around 550 nm linearly until one equivalent is bound; each probe was characterised by its K_D, its pH dependence and the nature of its binding site ^[9]. That pH dependence is the reason a conditional constant is quoted with a pH attached: it is a single apparent number for one set of conditions, not a property of the peptide. Applied to Aβ16 peptides and to the GHK and DAHK motifs, the probes reproduced previously reported affinities ^[9].

What happens when the competitor is mis-specified is on record. Fluorescence quenching and circular dichroism, with competition studies against free glycine, L-histidine and nitrilotriacetate, gave an apparent (conditional) dissociation constant of 10⁻¹¹ M at pH 7.4 for Cu²⁺ with Aβ(1–42), the binding curves being almost identical for the monomeric and fibrillar peptide ^[10]. Previous studies of Cu–Aβ had typically found the affinity two or more orders of magnitude weaker, largely because the affinity of the competing ligands or buffers had been underestimated ^[10]. A reported K_D is interpretable only alongside pH, buffer identity and concentration, ionic strength, temperature, and the competitor together with the constant assumed for it.

Your buffer is a ligand

HEPES, routinely labelled non-coordinating, forms a CuL⁺ complex with logβ(CuL) = 3.22; that complex undergoes alkaline hydrolysis above pH 6 with precipitation of Cu(OH)₂, and 100 mM HEPES at pH 7.4 elevates the apparent binding constant determined for another ligand by a factor of 80 ^[11]. The recommendation from that work is still to prefer HEPES over common alternatives such as Tris and phosphate for Cu(II) studies, provided appropriate corrections and precautions are applied in quantitative experiments ^[11].

The copper affinity credited to HEPES is not straightforward either. Across a panel of zwitterionic buffers, MES and MOPS did not bind copper and were identified as good choices for metal speciation studies within their operational pH range; MOPSO bound weakly at log K_c 2.02; HEPES, HEPPS and HEPPSO gave log K_c values of 7.04 to 7.68, but those strong copper-binding ligands were present at concentrations much lower than the total buffer concentration ^[12]. Electrospray quadrupole time-of-flight analysis of the HEPES matrix identified several contaminants, prominently one at m/z 131, at levels similar to those obtained from modelling the copper titration data ^[12]. Much of the observed binding therefore sits with trace components rather than with the nominal buffer molecule.

Tris is stronger and less predictable. A reinvestigation by potentiometry, UV-vis, ESI-MS and EPR over a broad range of concentrations and ratios found a maximum Tris-to-Cu(II) stoichiometry of 2 with dimeric complexes at neutral pH, and an apparent affinity at pH 7.4, expressed as a competitivity index, varying between 2 × 10⁶ and 4 × 10⁴ M⁻¹ depending on the Tris and Cu(II) concentrations and their molar ratio ^[13]. There is no single figure to quote.

Kinetics are affected as well as equilibria. In stopped-flow experiments at pH 7.4, phosphate, Tris, HEPES and MOPS, together with MES and PIPPS as buffers considered non-binding, all formed Cu(II) complexes within the instrument dead time, and Cu(II)–peptide complexes were formed via ternary complexes with the buffers ^[14]. For Cu²⁺ reacting with Gly-Gly-His, the Good buffers affected the observed rate only slightly, but Tris acted as a competitive inhibitor and phosphate as a reaction catalyst ^[14].

What each measurement establishes, and what it does not

Potentiometry

A glass-electrode titration follows proton release, so it returns stoichiometries and stability constants for the set of species present, and for Cu(II) and Ni(II) peptide systems it is the technique behind the amide-deprotonation result described earlier ^[2]. What it does not deliver is the identity of the donor atoms: [CuH₋₂L]⁻ is a composition, not a structure. The systems that are properly resolved are resolved by combination — the histidyl/aspartyl peptides above were characterised by potentiometric, UV-vis, CD and EPR spectroscopic techniques used together ^[4], and the HEPES CuL complex was pinned down by combined potentiometric and spectroscopic study ^[11].

UV-visible absorption and circular dichroism

Visible CD is the more discriminating of the two because it responds to the chirality of the environment around the metal. In copper(II) and nickel(II) complexes of methionine-containing tripeptides it resolved differences as small as weak axial versus equatorial coordination of a thioether side chain, and the spectral parameters of [CuL]⁺, [CuH₋₁L] and [CuH₋₂L]⁻ differed enough that stability constants and metal ion speciation could be calculated from the CD measurements themselves ^[15].

The sensitivity cuts both ways. For the pentapeptides acyl-GGGTH(92–96) and acyl-TNMKH(107–111), EPR and ¹H NMR suggested very similar coordination geometry at the two sites, yet their visible CD spectra were almost mirror images of one another ^[16]. A set of empirical rules derived from simple histidine-containing tripeptides now allows the appearance of a Cu²⁺ visible CD spectrum to be predicted for histidine and amide main-chain coordination ^[16]. Two CD spectra of opposite sign do not by themselves license the conclusion that the donor sets differ.

EPR

EPR reports on the paramagnetic species, and with a disordered peptide what it reports may be an average. EPR on His/Ala analogues of Aβ supported a dynamic view of the tetragonal Cu²⁺ complex, with axial as well as equatorial coordination of imidazole nitrogens creating an ensemble of coordination geometries in exchange with each other ^[10].

The larger risk is measuring the buffer. A systematic survey over pH 6.0–8.0 found that Tris coordinates Cu(II) across that range, HEPES binds Cu(II) weakly at near-neutral pH, and MOPS essentially does not chelate Cu(II) but precipitates it instead; these buffer-dependent complexes produce distinct EPR signatures that can interfere with spectral interpretation and spin quantification, and a histidine affinity tag generated its own Cu(II) signature ^[17]. In a bacterial multicopper oxidase, switching from HEPES to Tris eliminated adventitiously bound Cu(II) and revealed the true copper content and sites ^[17].

Mass spectrometry

Mass spectrometry establishes composition. It does not establish solution stability constants for Cu(II)–peptide complexes, and the failure is systematic rather than a tuning problem. ATCUN-family peptides with high and thoroughly determined Cu(II) binding constants were re-estimated by two ESI-MS techniques — standard conditions with serial dilution, and very mild conditions for competition experiments — and simulating the sample acidification that accompanies electrospray formation with a pH-jump stopped-flow experiment showed that the electrospray formation process compromises the entropic contribution to complex stability, yielding underestimated stability constants ^[18]. Where MS earns its place in this work is compositional: it is how the copper-binding contaminants in the HEPES matrix were identified ^[12].

Bound copper and iron oxidise the peptide you are measuring

Metal-catalysed oxidation of histidine to 2-oxo-histidine is an important pathway of oxidation in peptides and proteins both in vivo and in vitro, and predicting which histidine residues are susceptible requires understanding how sequence and conformation control the mechanism; the analysis is complicated further because aggregation and precipitation of the target can compromise it ^[19]. The damage is site-specific, which is precisely the problem: it falls on the residues that make up the binding site. In proteins exposed to metal-catalysed oxidation systems only one or a few amino acid residues are modified and relatively little peptide bond cleavage occurs, because the oxidation system reduces Fe(III) to Fe(II) and O₂ to H₂O₂, and those products react at metal-binding sites on the protein to generate active oxygen species that attack the side chains of residues at that site — methionyl to methionyl sulfoxide, cysteinyl to mixed disulfides, histidyl to asparagine or aspartate, and prolyl, arginyl and lysyl residues to carbonyl derivatives ^[20].

For copper the reactivity has the structural origin described earlier: because Cu(I) and Cu(II) impose different geometries on the same disordered peptide, an electron cannot simply move in place, and it is the in-between state that carries the chemistry ^[5]. The ATCUN case points the same way, the transient intermediate rather than the mature complex being the species reduced by glutathione ^[6].

Oxidative modification needs three things at once — O₂, Fe(III) or Cu(II), and an appropriate electron donor ^[20] — which converts into a short handling checklist. The electron donor is usually something already in the tube: ascorbate, dithiothreitol, glutathione, a residual reductant carried over from cleavage or refolding. An oxidised peptide is a different ligand with a different donor set, so any coordination chemistry measured afterwards belongs to a partly degraded sample. Screening for oxidation products by LC-MS before and after a binding experiment costs less than reinterpreting a stability constant measured on the wrong molecule.

Reporting a binding measurement

Enough of the above is systematic that a fixed reporting set removes most of the ambiguity from a quoted constant:

• Buffer identity, concentration and pH, together with the constant assumed for the buffer's own metal complex — and for Tris, tied to the actual buffer and metal concentrations and their ratio, because the apparent affinity moves with them ^[13].
• The competitor and its conditional constant on a stated scale, since a competition result is only as good as the standard behind it ^{[9, 10]}.
• Ionic strength, temperature, and the equilibration time between mixing and measurement, given that ATCUN-type sites pass through intermediates on a hundreds-of-milliseconds to seconds timescale ^[6].
• Whether the figure quoted is a conditional value at a stated pH or an overall constant from a full speciation model ^[4].
• Which techniques were combined, and which structural conclusion rests on which of them ^{[4, 15, 16]}.