Alexandra M.
Webster
* and
Anna F. A.
Peacock
*
School of Chemistry, University of Birmingham, Edgbaston, B15 2TT, UK. E-mail: alexamwebster@gmail.com; a.f.a.peacock@bham.ac.uk
First published on 15th June 2021
For much of their history, lanthanides were thought to be biologically inert. However, the last decade has seen the discovery and development of the field of native lanthanide biochemistry. Lanthanides exhibit a variety of interesting photophysical properties from which many useful applications derive. The development of effective functional lanthanide complexes requires control of their coordination sphere; something proteins manage very effectively through their 3D metal-binding sites. α-Helical coiled coil peptides are miniature scaffolds which can be designed de novo and can retain the favourable properties of larger proteins within a much simplified system. Metal binding sites, including those which bind lanthanides can be engineered into the coiled coil sequence. This review will highlight the opportunities presented by the use of coiled coil peptides as scaffolds for lanthanide binding and the potential to control the coordination environment by simple modifications to peptide sequence. Designed lanthanide coiled coils offer opportunities to gain greater insight into native lanthanide biochemistry as well as to develop new functional complexes, including imaging agents.
Metal-binding proteins are ubiquitous in nature; 30–40% of proteins are thought to require a metal to carry out their function.8 Proteins act as sophisticated ligands which are able to precisely control metal coordination chemistry, thereby tuning protein function.9–13 Thus harnessing the power of proteins as ligands for metal ions is an attractive prospect and an active area of research.14–19 One strategy is the design of structured peptides de novo, whereby miniaturized scaffolds with metal-binding sites that replicate the environment found within more complex proteins are engineered. Although various peptide folds have been successfully designed de novo, including β-hairpins20 and γ-turns,21 the majority of work has focussed on the design of α-helical assemblies, including coiled coils, which can be more predictably designed.
Herein, we will provide a brief overview of the photophysical, spectroscopic and biochemical properties of the lanthanides before discussing the development of lanthanide-binding coiled coils and, importantly, how this new class of ligands can control the lanthanide coordination chemistry and, in turn, the physical properties.
Whilst the lanthanide excited states are not significantly quenched by oxygen, they are very efficiently quenched by water, specifically the O–H stretching vibration.24,27 This means that in order to optimise lanthanide luminescence in aqueous solutions, metal-bound water molecules must be avoided (or minimised). This is made more difficult by the high affinity of the Ln3+ ions for water; the magnitude of their ΔHhydr values are high (ca. −3300 to −3700 kJ mol−1).28 As a result, lanthanide complexes with low denticity ligands rapidly dissociate in aqueous solutions to form aqua ions [Ln(H2O)n]3+, where n = 9 for La3+ to Gd3+, and then decreases to 8 moving along the series to Lu3+.29 These aqua ions are also prone to hydrolysis to their insoluble hydroxides, and a slightly acidic pH is needed to inhibit this process.30 In order to optimise lanthanide luminescence under aqueous conditions, complexes must possess both high kinetic and thermodynamic stability. To achieve this, polydentate ligands containing hard oxygen and nitrogen donors are used to improve thermodynamic stability via the chelate effect, and rigid ligands, in particular macrocyclic ligands, improve kinetic stability.
A related application is in MRI (magnetic resonance imaging). MRI is an imaging technique based on NMR which is widely used in clinical medicine to provide detailed information about different tissues. Contrast agents are used to enhance images generated by MRI by shortening the relaxation times of water protons. Contrast agents aim to alleviate two problems associated with MRI techniques. First, by increasing relaxivity and thereby decreasing the relaxation time of water protons, scans may be acquired more quickly. Second, accumulation of the contrast agent in particular tissues enables differentiation between tissues that would otherwise be difficult to distinguish from one another. The development of molecular imaging has led to the generation of ‘targeted’ and ‘smart’ contrast agents.37 Targeted contrast agents are designed to accumulate inside specific types of tissue whereas smart contrast agents respond to biological processes. For example, a targeted contrast agent may be conjugated to a ligand which binds to receptors found on the surface of particular cell types and thus reduce relaxation times of nearby protons, while a smart contrast agent will have no effect on proton relaxation times until interaction with species generated by certain biological processes.
Ln3+ complexes, especially Gd3+ may be used as contrast agents. Gd3+ has seven unpaired electrons, the most of any known elemental species, and may decrease T1 relaxation times by a factor of up to 106, making it the most commonly used species in MRI contrast agents.38 However, consideration of toxicity of free Gd3+ is necessary. Ln3+ ions are a similar size to Ca2+, and like Ca2+, prefer hard oxygen ligands. This means that lanthanides can interfere with calcium-dependent processes such as muscular contraction and neurotransmission. Free Ln3+ ions are also excreted slowly and accumulate in liver, spleen, kidney and bone tissue.39 Hence, gadolinium contrast agents must be stable to dissociation under physiological conditions, and as such, multidentate, macrocyclic ligands are prevalent in current clinically used gadolinium contrast agents.38–40
Whilst the binding affinity of Gd3+ for small molecule macrocyclic ligands is high (logK ∼ 23), relaxivity is suboptimal. Only one water molecule tends to be directly coordinated to Gd3+ and the rate of water exchange is slower than the Larmor frequency for machines used in the clinic. Additionally, the rapid molecular tumbling of small molecules in solution has a detrimental effect on relaxivity. The result is that in order to be effective contrast agents, gram quantities must be administered,41 and as such there is a desire for the development of more effective contrast agents. An increase in the hydration state of Gd3+, optimising water exchange rate and molecular tumbling rate are all approaches that can be adopted to improve relaxivity. Peptide-based contrast agents where coordination environment can be carefully controlled could therefore prove to be promising. Moreover, designs featuring peptides have shown promise as targeted contrast agents.42
Given that lanthanides are now recognised as biologically relevant metal ions, there is a need to expand our understanding of lanthanide biochemistry, and given the opportunities afforded by their chemical properties, there is an opportunity to develop new lanthanide-based biological agents. Harnessing control of the local lanthanide environment is paramount if either of these goals are to be achieved, and several strategies have been attempted as a means to do this in peptide-based systems.
![]() | ||
Fig. 1 Models generated using CCP4 software,134,135 oxygen atoms are red, nitrogen atoms are blue, Ca2+ is a pink sphere and Gd3+ is a grey sphere. (A) X-ray crystal structure of the 52 kDa domain of human cardiac troponin, residues 104–117, showing coordination of the Ca2+ binding site of one of the EF hand motifs of human cardiac troponin with coordinating side chains. Coordinating backbone oxygen omitted for clarity, pdb reference 1J1E.136 (B) X-ray crystal structure of a lanthanide binding tag showing the coordination environment of bound Gd3+ and coordinating side chains, coordinating backbone oxygen omitted for clarity, pdb reference 3VDZ.60 |
More recently it has been found that, rather than simply being capable of binding La3+ in a laboratory environment, EF hand motifs are present in proteins that natively bind La3+. Lanmodulin is a recently discovered bacterial protein containing four EF hand motifs, three of which bind lanthanides with picomolar affinities (the fourth with micromolar affinities), and shows a hundred million fold selectivity for La3+ over Ca2+.69 In 2019, Cotruvo and co-workers designed a fluorescent sensor which used lanmodulin to link two proteins as a Förster Resonance Energy Transfer (FRET) pair. The sensor was found to bind lanthanides with picomolar dissociation constants and only ∼2 fold weaker than the native lanmodulin.70
LBTs are not the only peptide or protein-based lanthanide binding agents. Recently, a TIM barrel protein which consisted of several de novo designed domains was shown to bind lanthanides with femtomolar dissociation constants.71 Other peptide lanthanide binding strategies include the use of cyclic peptides,72,73 the modification of zinc finger proteins,74–76 phosphorylated peptides77,78 and the use of non-natural amino acids.79 However a full discussion of these is beyond the scope of this perspective unless relevant to the peptide coiled coil scaffolds discussed.
One advantage of coiled coil ligands is the capacity to control both the primary and secondary coordination spheres of bound metals by small changes in the amino acid sequence. Control of the secondary coordination sphere can be difficult to regulate in a predictable manner in small molecule systems, but can have a significant influence on a metal's physical and chemical properties. For example, it can alter the catalytic activity of metal complexes by controlling selectivity and/or reactivity.9,96–98 This can occur through preorganisation of first coordination sphere ligands,99 steric encumbrance,100 and better interaction with biological targets.101 Specific to lanthanide complexes, secondary coordination sphere effects in small molecules complexes have been shown to alter their stability,102 luminescent101 and magnetic properties.104 Thus it is clear that control of the secondary coordination sphere is highly desirable, and something which is achievable with coiled coil peptide systems.
Peptide α-helical coiled coil design is based on a system of repeating heptads, denoted as (abcdefg)n where a–g are amino acid residues and n is the number of heptads (generally n ≥ 3 to promote folding). Generally, if the amino acids in the a and d positions are hydrophobic, this creates a hydrophobic face. Assembly of multiple α-helices into a coiled coil is driven by burying these residues, making a hydrophobic core (Fig. 2). Each heptad therefore contains two turns, but since this is not exact – there are 3.6 residues per turn – the hydrophobic face migrates around the α-helix in the opposite direction to the helix backbone leading to supercoiling.
![]() | ||
Fig. 2 Helical wheel diagrams of a (A) two- and (C) three-stranded coiled coil. Hydrophobic residues are black and salt bridge interactions between g and e residues are shown (red). X-ray crystal structures of a de novo designed homo- (B) dimer (pdb 1ZII) and (D) trimer (pdb 1ZIJ).137 Models generated using CCP4 software.134,135 |
Packing of the hydrophobic a and d residues determines the number of α-helices making up the coiled coil. Generally, a = isoleucine (Ile) and d = leucine (Leu) promotes the formation of dimeric coiled coils, a = d = Ile or Leu promotes the formation of coiled coil trimers and a = Leu and d = Ile promotes the formation of tetrameric coiled coils.108 Valine (Val) or alanine (Ala) may also be used.
As illustrated for a parallel three stranded coiled coil in Fig. 2, charged residues in the g position of one α-helix interact with complementary residues in the e position of another in an i – i′ + 5, or manner, forming interhelical salt bridges. Residues in the b and c positions tend to be water solubilising groups and/or helix promoters.109 Finally, the f position is least critical to coiled coil formation and as such provides a site where other functionality can be introduced. Examples have included unnatural amino acids that provide an opportunity for labelling (for example, azides and alkynes for copper catalysed azide–alkyne click reactions),110 or groups that aid coiled coil characterisation, such as 4-iodo-L-phenylalanine for X-ray crystallography.106
Metal-binding sites can be engineered externally or more commonly within the hydrophobic core of the coiled coil, and, through a process of peptide design, the coordination chemistry of the bound metal tuned. Given the recent recognition of the importance of understanding lanthanide biochemistry, and the established work demonstrating that engineered lanthanide sites are extremely useful spectroscopic tools, our group and others have explored lanthanide coordination to miniature de novo designed protein scaffolds, including the coiled coil.
a ΔΔGu is the change in free energy of unfolding on addition of LaCl3. LaCl3 concentration for E2 and E3 peptides was 50 mM, and for Gla2Nx was 5 mM. A positive ΔΔGu indicates an increase in coiled coil stability. |
---|
![]() |
Further work by Hodges and co-workers showed that the stabilising effect of La3+ addition on their coiled coils was highly dependent on the position of the Gln to Glu substitutions.112 Three peptides were made, (E2(15,20), E2(13,22) and E2(20,22), Table 1), with varying distances between the two repulsive Glu e and g residues. Addition of LaCl3 had a greater stabilising effect on E2(15,20) coiled coil formation than for E2(13,22), reflecting the reduction in the magnitude of Glu–Glu repulsion as the distance between them increases. Whilst, of the three peptides, the Glu residues are physically closest in the E2(20,22) coiled coil, addition of LaCl3 is instead destabilising due to the presence of a d position leucine (Leu). In the two stranded coiled coil, this bulky Leu side chain protrudes into the interhelical interface between the two Glu residues of the two peptide chains, interfering with the repulsive interaction between the two Glu residues, thus reducing relief provided by La3+ binding.
Substitution of a third Glu in place of a Gln residue in the E2(15,20) peptide to give E(13,15,20) and E(15,20,22) (Table 1) resulted in enhanced coiled coil stability on addition of trivalent La3+ regardless of whether the third Glu was in position 13 or 22. The resulting charge neutrality provides additional stabilisation, even if direct lanthanide coordination to all three Glu residues is not possible.
Lanthanide binding can be further enhanced by replacement of Glu residues by the uncommon amino acid γ-carboxyglutamic acid (Gla).113 The Gla side chain contains two carboxylic acid functional groups (compared to the one carboxylate group of the Glu side chain). Under physiological pH, Gla therefore carries a charge of −2 and twice as many potential O-donor atoms. Free Gla had previously been shown to bind Tb3+via the side chain carboxylate groups in a Ln(III):
Gla = 1
:
2 ratio.114
As might be expected, the increased magnitude of the negative charge as a result of Glu → Gla substitution results in greater destabilisation of the coiled coil structure. When the valine (Val) residue in position 23 was substituted for asparagine (Asn) in the Glu → Gla analogue of peptide E2(15,20) to give Gla2Nx (Table 1), the peptide adopted a random coil formation. Such is the affinity of La3+ for the Gla ligand, that LaCl3 addition was capable of recovering coiled coil formation to 100% and increasing coiled coil stability by up to 5.1 kcal mol−1.
Gla2Nx was also the first example of a de novo designed coiled coil peptide scaffold that was reported to bind lanthanides other than La3+; in spite of its smaller ionic radius, Yb3+ was shown to have a similar affinity for the Gla ligand as La3+. These peptides were shown to bind preferentially to lanthanides over other biologically relevant metals (Mg2+, Ca2+ and Zn2+). Binding preference for La3+ over Ca2+ is particularly notable given the similarity of lanthanide coordination chemistry with that of Ca2+ and is accounted for by the greater charge density of the trivalent lanthanide ion.
These early studies demonstrated that lanthanide binding sites could be engineered at the interface of complex helical peptides, and that lanthanide binding can even induce formation of higher order structures. Lanthanide binding can be achieved using readily available amino acids as ligands, with Gla being a particularly high affinity ligand. Importantly, these external binding sites show measurable selectivity for lanthanides over biologically relevant metals (Mg2+, Ca2+ and Zn2+).
These peptides featured the Gla for lanthanide binding whilst Pep4 also contains a tryptophan (Trp) as a sensitiser for lanthanide luminescence in position a of heptad 2 (i.e. in the layer above the Gla site), and Pep5 contained an Ala as the complementary ‘hole’ to accommodate the bulk of the Trp. Addition of either Eu3+, Tb3+ or Ce3+ induced formation of a trimeric coiled coil Ln(Pep4)(Pep5)2 and was accompanied by Trp sensitised Ln luminescence.
This work demonstrated for the first time that lanthanide binding sites could be engineered within the hydrophobic core of a coiled coil. However, the focus was predominantly on the effect of lanthanide ions on peptide folding and further exploration of the lanthanide coordination chemistry or their photophysical properties was not reported.
Peptide design was based on a repeating heptad of Ac-G(IAAIEQK)xG-NH2 which favours three-stranded coiled coil formation. The lanthanide binding site was rationally designed based around the coordination of Asp and Asn side chains. Asp was preferred to Glu because initial molecular dynamics modelling of the binding site suggested appropriate Ln–O bond lengths for the Asp side chain. Coordination of a Ln3+ ion to the Asp side chains in the core of a trimeric coiled coil will therefore lead to charge neutrality. The a site Asp (heptad 3) would also provide up to six oxygen donor atoms at roughly the desired distance for Ln–O bond formation. The Asn residue was located directly above the Asp layer (the d position of the second heptad) to provide an additional layer of oxygen donor atoms, thereby providing the Ln3+ with up to nine oxygen donor atoms. Finally, a Trp residue, a known sensitiser for some Ln3+ ions, was introduced adjacent to the designed binding site in an f position (MB1, see Table 3). Modelling of the design predicted binding of Ln3+ ions via three Asp side chains and three carbonyl oxygens of the Asn residues as well as the presence of a directly bound water molecule (Fig. 3). As was previously observed for lanthanide coiled coils, in the absence of any coordinating metal, the negatively charged coordinating side chains repel each other and destabilise the coiled coil structure. However, titration of Ln3+ ions into a solution of MB1 peptide at pH 7.0 induced folding reaching maximum folding at one equivalent of Ln3+ per peptide trimer, consistent with the design. Similar behaviour was observed from titrating a range of different lanthanide ions, (Ce3+, Nd3+, Eu3+, Dy3+, Er3+ and Yb3+) including the first report of a Gd3+ coiled coil complex.
![]() | ||
Fig. 3 (A) Structure of Gd(MB1)3 after 10.0 ns of molecular dynamics simulations and close up, (B) side-on and (C) top-down views of the Gd3+ coordination site. Shown are the main chain atoms represented as helical ribbons (green) and the Asn and Asp side chains in stick form (oxygen in red and nitrogen in blue), a water molecule in ball-and-stick form and the Gd3+ ion as a sphere in grey. Reproduced from ref. 116 (https://pubs.acs.org/doi/10.1021/ja408741h) with permission from the American Chemical Society. Permission for further use of this figure must be sought from the American Chemical Society. |
As with previously reported lanthanide-binding coiled coils, the binding site was specific for Ln3+ ions over Ca2+ ions. This is most likely a result of the complementary charge within a trimeric coiled coil. However, contrary to the model's prediction, investigation into the peptide-bound Tb3+ hydration state showed no water molecules were directly coordinated. Surprisingly, despite the lack of inner sphere water, MRI properties of Gd(MB1)3 were superior to those of Gd(DOTA) which is in clinical use as an MRI contrast agent under the trade name Dotarem. Gd(MB1)3 showed >three-fold enhanced transverse relaxivity, r2, at 7 T in comparison to Gd(DOTA).
The superior MRI relaxivities of this gadolinium coiled coil, achieved despite the lack of inner sphere water, which is normally considered a prerequisite for gadolinium MRI contrast agents, were the motivation to explore more fully the opportunities afforded by this new class of ligands for the lanthanide ions, and gadolinium in particular.
The following sections detail our efforts to tune the chemistry of lanthanide ions coordinated to coiled coil ligands, through modifications to both primary and secondary coordination sphere ligands. Furthermore, the size of these coiled coil ligands provides an opportunity for the inclusion of multiple distinct metal binding sites within a single design.
To investigate this, the binding site, including the adjacent Trp sensitiser, was systematically translated along the coiled coil (Fig. 4 and Table 3).117 The Asn3Asp3 metal binding site in MB1 is situated between the second and third heptad. However, there are three other sites in which this binding site could reside, one a heptad above and two in lower heptads. These peptides were denoted MB1-x where x = 1–4 and refers to the heptad containing the Asn residue, with the Asp residue in the a site of the following heptad. The original MB1 peptide is therefore renamed MB1-2.
![]() | ||
Fig. 4 Relationship between linear location of Asn3Asp3 Ln3+ binding site and coordination chemistry. Shown from left to right are cartoons of (A) MB1-1, (B) MB1-2, (C) MB1-3 and (D) MB1-4. Shown are the main chain atoms represented as helical ribbons (green), the Asn and Asp side chains in stick form (oxygen in red and nitrogen in blue) and the Gd3+ ion as a sphere (pink). Reproduced from ref. 117 with permission from the Royal Society of Chemistry. |
Core heptads have been reported to be up to three times more stabilising with respect to coiled coil formation than terminal heptads.118 Consistent with this observation, peptides with lanthanide binding sites located in core heptads were considerably less folded than peptides containing these in terminal heptads. Of the two peptides containing terminal binding sites, MB1-1 at the N-terminus and MB1-4 at the C-terminus, MB1-1 was both better folded and more stable. This is because the coiled coil is not symmetrical and the MB1-1 and MB1-4 sites are not identical. The binding site in MB1-1 is closer to the extremity of the coiled coil, whereas it is more buried in MB1-4. A similar trend was seen in the folding of the Gd-peptides, Gd(MB1-1)3 > Gd(MB1-4)3 > Gd(MB1-2)3 ∼ Gd(MB1-3)3.
Regardless of binding site location, the emission profile of bound Eu3+ in Eu(MB1-X)3 indicated a symmetric Eu3+ site, consistent with a single Eu3+ bound within a symmetric coiled coil. In contrast, the hydration state of bound Tb3+ was found to be highly dependent on the binding site location. As in the previous study, there was no evidence of inner sphere water when Tb3+ is bound centrally along the coiled coil (MB1-2 and MB1-3). However, water was able to directly coordinate to Tb3+ when bound in terminal binding sites, with evidence of three and two inner sphere waters for MB1-1 and MB1-4, respectively. Not surprisingly, the increase in inner sphere water correlates with an increase in MRI relaxivity for the Gd3+ analogues.
Gd(MB1-2)3 and Gd(MB1-3)3, with no evidence of inner sphere water, display comparable transverse and longitudinal relaxivity. Increasing the hydration of the bound metal to two (Gd(MB1-4)3) and three (Gd(MB1-1)3) water molecules increases the relaxivity accordingly. Consequently the 1 nm linear translation of an otherwise identical ligand presenting binding site along the coiled coil has altered the water access and lanthanide coordination chemistry to such an extent that we observe a four-fold increase in MRI transverse relaxivity of the Gd3+ complex.
The lack of inner sphere water for lanthanides coordinated to the centrally located binding sites is consistent with a coordinatively saturated environment where the peptide scaffold provides all of the coordinating donor atoms. However, the lanthanide hydration states at the terminal binding sites suggest that not all of the Asp/Asn residues are fully engaged in metal binding. In the case of MB1-1, it was proposed that the Asn layer does not coordinate to the bound lanthanide. To test this hypothesis, the analogous peptide which lacked the Asn layer, CS1-1, (Table 3), was prepared. This design was still able to bind lanthanide ions, and had a similar experimentally observed hydration state to MB1-1, suggesting the Asn residue was not essential for metal binding at the C-terminus.
CS1-1 was found to display optimal Tb3+ binding between pH 6–7. At low pH, it was speculated that protonation of Asp side chains precludes lanthanide binding, whilst at high pH, competing formation of lanthanide hydroxide species causes dissociation of Tb3+. Similar observations were made for the MB1-2 peptide, although the pH range through which Tb3+ was bound was considerably larger (pH 4–7).119
The starting point was the binding site within MB1-1, located towards the N-terminus of the coiled coil, which generates a highly hydrated lanthanide site, Tb(OH2)3(MB1-1)3. MB1-1, contains a non-coordinating terminal isoleucine (Ile) layer directly above the Asn/Asp binding site residues (in the a position of the first heptad) and it was the identity of this residue that was systematically altered to Ala, phenylalanine (Phe), tyrosine (Tyr) and Trp (MB1-1(2X), Fig. 5 and Table 3).
![]() | ||
Fig. 5 Pymol models of the MB1-1(2X) peptide illustrating mutation of the residue in position two (2X) through a space filling representation and their influence on lanthanide hydration state. The main chain atoms are represented as helical ribbons (green), Asn and Asp coordinating residues along with the modified residue in position two (2X) are shown in stick form (nitrogen in blue and oxygen in red) and the Ln3+ ion as a grey sphere. Top down view of the proposed coodinatively saturated lanthanide binding site. Reproduced from ref. 125 with permission from the Royal Society of Chemistry. |
Ala is sterically less bulky than Ile, and so might have been expected to increase the hydration state by allowing greater water access. However, luminescence lifetime decay experiments featuring Tb3+ as the bound lanthanide showed that the hydration state remained three, indicating that for the MB1-1 Asn/Asp binding site, this is the maximum hydration state accessible. In contrast, Phe which is bulkier than Ile leads to a less hydrated site with two water molecules bound to the Tb3+. As the steric bulk of the terminal second sphere residue increased further, so did the reduction in hydration state. Introduction of Tyr resulted in Tb3+ bound to the Asn/Asp site with one exogenous water molecule, and, with the bulkier Trp, water binding was prevented completely. Introduction of this second coordination sphere Trp therefore converts the terminal binding site into one more closely resembling a buried binding site located centrally within the coiled coil (MB1-2 or MB1-3), and this is mirrored in the chemistry of the sites, including near identical MRI relaxivity data for the Gd3+ complexes of MB1-1(2W) and MB1-2. Altering the identity of a single, second coordination sphere terminal residue in MB1-1, from Ile to the bulkier Trp, results in a four-fold reduction in transverse relaxivity for the Gd3+ complex.
Only one example of a lanthanide-containing hetero-bimetallic coiled coil has thus far been reported.119 The coiled coil features a lanthanide binding Asp3 site as well as a thiolate Cys3 site, suitable for binding softer metal ions including mercury.90,131 The soft thiol ligands contrast with the hard oxygen donors associated with lanthanide-binding, and provide a mechanism by which selectivity can be readily achieved based on hard-soft acid base theory. The presence of polar metal binding residues in the hydrophobic core of a coiled coil is destabilising and so both binding sites were located at opposite termini so as to minimise destabilisation.117 The Cys3 mercury binding site was located in the a position of the fifth, a terminal, heptad; a site that has been previously reported and well-studied.90,131 The least disruptive of the lanthanide binding sites studied to date, CS1-1, was located at the N-terminus and contains a single Asp layer in the a position of the second heptad. Thus the designed hetero-bimetallic coiled coil contained an Asp3 layer for lanthanide binding towards the N-terminus and a Cys3 site for mercury binding ∼4 nm away towards the C-terminus, to generate CS2-1,4 (Fig. 6).
![]() | ||
Fig. 6 Cartoon representation showing designed hetero bimetallic coiled coil, CS2-1,4, featuring LnAsp3 and HgCys3 sites towards the N- and C-termini, respectively. Shown are the main chain atoms represented as helical ribbons (green), the Asp, Trp and Cys side chains in stick form (oxygen in red, nitrogen in blue and sulfur in orange), and the metal ions as spheres (lanthanide in grey and mercury in cyan). Reproduced from ref. 119 with permission from the Royal Society of Chemistry. |
Mercury binding to the Cys3 site is known to be pH dependent, with binding as HgCys2 at neutral pH changing to HgCys3 above pH 7.6.132 The same behaviour was seen in the CS2-1,4 peptide, demonstrating that Hg2+ binding was selective for the soft Cys3 site and was insensitive to the presence of the hard Asp3 binding site. Luminescence experiments also showed that, at neutral pH, Tb3+ binding to CS2-1,4 was similar to Tb3+ binding to peptides lacking the Cys3 mercury binding site. Again, binding is selective for the intended, in this case hard Ln3+, metal binding site with no interference due to the presence of a second Cys3 metal binding site.
In this hetero-bimetallic system both d- and f-block metal binding sites were successfully introduced within a single coiled coil, demonstrating that it should be possible to exploit their complementary attractive properties. These could include the photophysical and magnetic properties of lanthanide sites, coupled with the breadth of electronic, catalytic or spectroscopic properties afforded by the d-block metals. Secondly the d- and f-block metal binding sites operated independently from each other, showing essentially the same binding behaviour as in the corresponding mononuclear peptides. As such, future design can select from an extensive toolbox of designed monometallic sites, in a “plug-and-play” type approach to achieve functional complex designs.
As with other metals, the coordination environment has been shown to play an essential role in the activity and behaviour of lanthanide complexes.102–104 Thus, if the true potential of lanthanide–protein assemblies is to be realised, the ability to rationally design lanthanide binding sites with predictable chemistry is essential. Coiled coils are a class of ligands which benefit from many of the advantages afforded by protein ligands without much of the complexity. Rational design principles can be used to tune the lanthanide coordination chemistry.
One challenge that remains is to improve the binding affinities of lanthanides to the coiled coil scaffolds. Table 4 provides a summary of lanthanide-binding coiled coils discussed in this review, with their principle metal-binding residues and logK values (where data was available). For comparison, Gd-DOTA, a small molecule MRI contrast agent in clinical use, has a log
K of ∼26, whereas lanthanide binding tags have log
K values of ∼8–9, and the lanthanide coiled coils typically have log
K values of ∼5. To use coiled coil-based systems for in vivo applications, significant improvements to binding affinity would be necessary. However, for in vitro, assay-based diagnostics these binding affinities may be adequate.
Peptide | Principle metal binding residues | Lanthanides investigated | Binding affinity (log![]() |
---|---|---|---|
a log![]() ![]() ![]() |
|||
Ln(E2(15,20))2 | Glu | La3+ | ∼1.9 (La3+)a |
Ln(E3(13,15,20))2 | Glu | La3+ | ∼2.5 (La3+)a |
Ln(E3(15,20,22))2 | Glu | La3+ | N/A |
Ln(Gla2Nx)2 | Gla | La3+, Yb3+ | 6.2 ± 0.2 (La3+), 6.4 ± 0.2 (Yb3+)b |
Ln(Pep3)3 | Gla | Ce3+, Nd3+, Eu3+, Dy3+, Er3+ and Yb3+ | N/Ac |
Ln(Pep4)(Pep5)2 | Gla | Ce3+, Eu3+, Tb3+ | N/Ac |
Ln(MB1-1)3 | Asp, Asn | Eu3+, Gd3+, Tb3+ | 5.30 ± 0.15 (Tb3+)d |
Ln(MB1-2)3 | Asp, Asn | Ce3+, Nd3+, Eu3+, Gd3+, Tb3+, Dy3+, Er3+ and Yb3+ | 5.48 ± 0.20 (Tb3+)d |
Ln(MB1-3)3 | Asp, Asn | Eu3+, Gd3+, Tb3+ | 5.16 ± 0.26 (Tb3+)d |
Ln(MB1-4)3 | Asp, Asn | Eu3+, Gd3+, Tb3+ | 5.26 ± 0.36 (Tb3+)d |
Ln(CS1-1)3 | Asp | Tb3+ | 4.57 ± 0.07 (Tb3+)d |
The potential of the coiled coil scaffold lies in the demonstrated ability to control coordination chemistry by making simple, rational changes to peptide sequence. For example, the lanthanide coordination environments for MRI contrast agents versus luminescent probes have different requirements. To optimise lanthanide luminescence, a sensitising ‘antenna’ (usually an aromatic organic group) must be incorporated, and water must be excluded from the coordination sphere to prevent quenching. Conversely, for MRI contrast agents, increasing the number of coordinating water molecules is desirable as this correlates with increased relaxivity. The ability to fine tune the lanthanide coordination environment to predictably either bind or exclude water means that the coiled coil is a versatile scaffold potentially capable of sensitising lanthanide luminescence and also acting as an MRI contrast agent, depending on the design of the lanthanide binding site. As such, this new class of lanthanide complexes provide valuable insights into the relationship between coordination environment and lanthanide complex behaviour and are therefore an important addition to the toolbox for scientists seeking to design new functional lanthanide-protein assemblies.
This journal is © The Royal Society of Chemistry 2021 |