Norton G.
West†
a,
Rania S.
Seoudi†
a,
Anders J.
Barlow
a,
Dongchen
Qi
ab,
Ljiljana
Puskar
c,
Mark P.
Del Borgo
d,
Ketav
Kulkarni
d,
Christopher G.
Adda
a,
Jisheng
Pan
e,
Marie-Isabel
Aguilar
d,
Patrick
Perlmutter
a and
Adam
Mechler
*a
aDepartment of Chemistry and Physics, La Trobe Institute for Molecular Science, La Trobe University, Bundoora, Victoria 3086, Australia. E-mail: a.mechler@latrobe.edu.au
bSchool of Chemistry, Physics and Mechanical Engineering, Queensland University of Technology, Brisbane, Queensland 4001, Australia
cMonash Biomedicine Discovery Institute & Department of Biochemistry and Molecular Biology, Monash University, Clayton, Victoria 3800, Australia
dDepartment Locally Sensitive and Time-Resolved Spectroscopy, Helmholtz-Zentrum Berlin für Materialien und Energie, 12498 Berlin, Germany
eInstitute of Materials Research and Engineering, A*STAR (Agency for Science, Technology and Research), Singapore 138634, Singapore
First published on 2nd July 2020
Spontaneous formation of nanostructured materials of defined structure and morphology is a crucial milestone toward realizing true bottom-up nanofabrication. Supramolecular recognition offers unparalleled specificity, selectivity and geometric flexibility to design hierarchical nanostructures. However, competition between similar binding motifs and the dynamic nature of the attachment imposes a severe limitation on the complexity of the achievable structures. Here we outline a design based on two distinct binding motifs in a supramolecular fibrous assembly to realize a metallosupramolecular framework (MSF). Controlled geometries were achieved by one-dimensional supramolecular assembly of substituted oligoamide units. The assembly of the monomers yields nanorods of sub-nanometer diameter and lengths in the 100 μm range. Addition of Cu2+ led to the formation of well aligned two-dimensional arrays on mica surface. Vibrational spectroscopy confirmed that the backbone amide moieties are not affected by metal addition. XPS and NEXAFS results suggest that Cu(II) is reduced in the process to a mixture of Cu(I) and Cu(0), likely in an interaction with the amine moiety of the imidazole side chain. Our results indicate that the two dimensional superstructure is based on the formation of polynuclear metal complexes between the oligoamide nanorods, thus the structure is confirmed to be a metallosupramolecular framework.
The helical head-to-tail assembly of N-acyl β3 oligoamides yields one-dimensional nanomaterials that can form fibrous hierarchical structures via bundling in different ways, controlled by the second order interactions between the core nanorods.14,15 In order to create more regular two- and three-dimensional networks from these nanorods, an independent secondary self-assembly motif is necessary. Metal coordination is widely used to create nanostructured materials of controlled properties, linking multidentate organic ligands into metal organic frameworks (MOFs); the contemporary use of the term refers to exceptionally low-density structures.16 MOFs are usually regarded as infinite structures, whereas metallosupramolecular coordination complexes are discrete structures of complex metal–ligand networks.17 Metallosupramolecular materials are usually based on multi-ligand building units or complex ligands.17,18
While there were attempts to use peptides for the creation of MOF-like materials by metal coordination between carboxyl groups,19 the design of such structures is problematic as the nitrogen of the amide moiety is a good Lewis base and thus it can also coordinate to the metal, leading to loss of structure.20 In case of transition metals, in particular Ni(II) and Cu(II), extensive data exist to show that coordination to peptides frequently proceeds by deprotonation of the amide and direct nitrogen–metal bond formation as opposed to charge solvation to the carbonyl oxygens, although the latter is not entirely unprecedented.21–23 It should be noted that peptide–metal interaction is commonly characterized in the gas phase, and the presence of a solvent may affect the sterically accessible sites. Indeed, metal coordination is used in protein purification, where the imidazole sidechains of a polyhistidine tag coordinate to a metal, typically Ni(II), immobilized in a carrier molecule such as nitrilotriacetic acid. The proteins are cleaved from the tags by imidazole solution, which is only possible if backbone amides do not participate in the coordination. Therefore it is plausible to design metallosupramolecular assemblies based on histidine residues as ligands,22,24 provided that a structural/steric factor protects the peptide backbone.
The aim of this work is to use the well-defined one dimensional supramolecular self-assembly of N-acetylated 14-helical β3-oligoamides12–15 as a core structure in combination with a secondary binding motif based on metal coordination of histidine sidechains. The supramolecular three point hydrogen bonding motif that yields the head-to-tail self-assembly of helical β3-oligoamide units is very strong, hence fibres are observed in water and even in DMSO,25 without a noticeable monomeric/small oligomeric population.26,27 For this reason, traditional characterization methods such as NMR or CD are not feasible to perform, and therefore structure can only be inferred from microscopy imaging and indirect characterization methods.
A further challenge is that the synthesis of β3 histidine is problematic; however in principle it should be possible to combine natural α-amino acids with unnatural β-oligoamides to overcome this limitation. Hybridizing natural α-peptides with unnatural β-oligoamides has been regularly used to enhance the diversity of folding conformations.28–30 Hence, in this work a β3,α hybrid oligoamide Ac-β3A β3V β3S-αHαH-β3Aβ3V β3A (2H – Fig. 1a) was designed to create metallosupramolecular superstructures (metallosupramolecular frameworks, or MSFs). MSF is defined as a periodical an infinite metallosupramolecular system.27
Fig. 1 (a) 2H oligoamide structure, (b) AFM image and (c) TEM image of the spontaneously assembled oligoamides without metal. The height scale is 45.1 nm for (b). |
Incubating the oligoamides in 100 μM CuCl2 solution changed the appearance of the deposit. In TEM imaging, the mesh of fibres was replaced by a folded ribbon/sheet structure (Fig. 2a) in which parallel-aligned nanorods are barely discernible. For AFM imaging the sample was deposited on mica surface, which exhibits a higher surface energy than the carbon gossamer used for TEM, and therefore interacts with the fibres more strongly. In this case the incubation with Cu(II) yielded a single continuous layer (Fig. 2b) which was homogeneous over tens of micrometers, albeit with a few defects. Zooming in to one of the defect sites (Fig. 2c) reveals a corrugated pattern with a modulation depth of 0.1–0.2 nm. A single, looped fibre is also visible on the surface. A line profile across the defect site (Fig. 2d) shows that the thickness of the layer is ∼0.8 nm. The corrugation exhibits an irregular repeat distance of 10–20 nm implying a wave pattern in the 2D sheet formed by closely crosslinked nanorods, and not by the alignment of individual nanorods that, according to published crystallography results, should have a diameter of 0.4–0.5 nm.12 The 0.8 nm thickness of the structure is consistent with two layers of nanorods.
The regularity of the structure suggests that it is formed by coordination crosslinking of the nanorods without affecting the supramolecular self-assembly, that is, via the histidine (or another) side chain. Out of the side chains of the oligoamide residues, alanine and valine are both non-coordinating, while the –OH of the serine is a weak ligand only above pH 8–9.34 In this work, the solution was not buffered and hence the pH was defined by the dissociation equilibrium of the C-terminal carboxylic acid. Accordingly, the serine side chain is not expected to participate in the metal coordination. Thus, only the histidine side chains, and potentially the terminal carboxyl moieties provide suitable ligands. It was previously suggested that multiple histidine residues may form macrochelates through complexation of two to four imidazole rings from different peptides to the same metal ion.35 However it was also reported that the amide groups of the backbone are easily deprotonated during the metal coordination of monomeric, unstructured peptides.21 Therefore to confirm metal coordination to the imidazole rings, spectroscopic experiments were performed.
The IR spectra hold further clues to the structure of the coordination complex. The peaks in the 1250–1100 cm−1 range can be assigned to C–N modes of the amine moiety of the histidine side chain. Both of the bands at 1202 cm−1 and 1132 cm−1 can be assigned to ν(CN) with coupling to δ(NH) and δ(CH), respectively.
The higher frequency mode is split between the two tautomers of the protonated imidazole (4:1 between N1 and N3, respectively39), thus the shoulder at 1181 cm−1 (the green shaded region in Fig. 3) can be identified as the ν(CN), δ(NH) vibration of the protonated N3 tautomer.40 This shoulder shifts slightly upon hydration, probably due to hydrogen bonding to water. After the metal was added, the disappearance of this peak indicates that a hydrogen is lost from the imidazole ring in N3 position. The simultaneous shift in the lower frequency peak to 1142 cm−1 confirms a changed chemical environment of what was a composite peak of the ν(CN), δ(CH) modes of two tautomers. These observations further confirm that the metal coordination happens to the side chain without the involvement of the oligoamide backbone, which is a marked difference to literature reports that describe disruption of the alpha helix upon copper coordination.20,41 We should note that the coordination appears to prefer only one of the tautomers and thus it is expected that in the bulk material a fraction of the oligoamides does not participate in the coordination framework.
Nitrogen environment | Position (eV) | % area |
---|---|---|
Before Cu(II) addition | ||
Amine N3 | 402.8 | 3.16 |
Amine N1 | 400.7 | 12.51 |
Amide | 400.0 | 66.74 |
Imine N3 | 399.2 | 13.15 |
Imine small N1 | 398.5 | 4.44 |
Aged with Cu(II) | ||
1 | 401.9 | 1.94 |
2 | 400.7 | 8.74 |
Amide | 399.7 | 68.94 |
3 | 399.3 | 3.41 |
4 | 398.9 | 16.98 |
In the sample aged with Cu(II) in solution phase (Fig. 4b), the 402.7 eV component shifted to lower binding energy (peak 1), consistent with deprotonation of the H–N3 tautomer as identified by IR spectroscopy. The lack of any noticeable shift in the overall peak position, dominated by the amide chemical environment, confirms that amide moieties are not affected by the metal, even after aging the sample for weeks. Clear assignments of peaks 2–4 is not possible given the small contribution they make to the overall N 1s peak. However it appears that the amine mode of the N1-protonated tautomer does not shift, suggesting that deprotonation only happens at the N3 position; this is also consistent with the IR results. There is a slight shift in peak positions in the imine region, suggesting that the N3 environment of the N1-protonated tautomer may also participate in coordination.
Fig. 4c shows the O 1s peak of the oligoamide before the addition of the metal. The peak fitting reproduces the oxygen environments of the molecule: the amide, carboxylate, and alcohol oxygens in a stoichiometric ratio of 8:2:1 respectively (Table 2). The addition of copper brings significant change to the O 1s spectrum as shown in Fig. 4d. The carboxylate O disappears, whereas a new component appears located at 532.5 eV with a similar intensity; this suggests that the C-terminal carboxylate moiety is also strongly involved in the metal coordination. In contrast, the other two O components only shift slightly. The backbone amide oxygen peak shifted with 0.3–0.4 eV, indicating a relaxation of the core helix due to the added stability of the coordination crosslinking.
Oxygen environment | Position (eV) | % area |
---|---|---|
Without copper addition | ||
Amide | 531.7 | 80.0 |
Carboxylate | 536.2 | 12.2 |
Alcohol | 533.5 | 7.8 |
Aged with copper | ||
Amide | 531.4 | 72.8 |
Carboxylate | 532.5 | 18.2 |
Alcohol | 533.7 | 9.1 |
Change due to coordination | ||
Amide | 0.3 | |
Carboxylate | 3.7 | |
Alcohol | −0.2 |
The effect on the carboxylate moiety is notably missing in the IR measurements, and therefore it is the result of a much slower process than the in situ coordination that was studied with IR, and potentially linked to an oxidation state change of copper.
It should be noted that there is a potential overlap of the carboxylate O 1s peak and Na LMM; however the absence of any noticeable peak in the area in Fig. 4d confirms that the peak in Fig. 4c is originated from the oxygen species. The thickness of the sample makes any contribution from the SiO2 (substrate) O 1s signals negligible. Furthermore, the expected position of O 1s for SiO2 is 532.9 eV which was not observed in the fitted spectra.
Fig. 5a shows the Cu 2p core-levels of the coordinated material with a spin–orbit splitting of 19.8 eV between Cu 2p3/2 and Cu 2p1/2. While the main Cu 2p3/2 peak energy is very similar for all copper oxidation states, Cu(II) species can be identified from the presence of shake-up satellite peaks.42 Conversely the absence of the shake-up peaks, as observed in Fig. 5a, indicates that Cu(II) was reduced in the coordination process to Cu(I) and/or Cu(0).42 In contrast, for materials coordinated in solution the presence of a higher binding energy component in Cu 2p3/2 and the associated satellite peaks around 943 eV are characteristic of Cu(II) (Fig. 5b), suggesting incomplete coordination. This is most likely due to steric factors, as the nanorods could not align into parallel fibrils in the solution unlike in case of the surface templated coordination.
The copper LMM Auger spectrum (Fig. 5c) of the surface templated material shows two distinct peaks. Kinetic energies are summarized in Table 3. These loosely correlate to the expected kinetic energies of Cu(I) and Cu(0), consistent with the analysis of Cu 2p spectra. There has been both a relaxation and increase of the valence orbital energies, as there is a 2.2 eV reduction in the Cu(I) Auger peak and also a 3.5 eV increase in the position of the Cu(0) LMM peak.42 These assignments were made on the basis that a smaller peak position shift would be more likely (i.e. the peaks would not cross over). Therefore the LMM spectra suggest the presence of comparable amounts of Cu(I) and Cu(0) in the coordination complex. The distinct appearance of the Cu LMM peaks suggests well defined, and fundamentally different chemical environment for the two species.
Fig. 5d shows results of a NEXAFS measurement. While the CuCl2 salt (Cu-MI) shows an intense peak corresponding to the 2p → 3d electronic transition for Cu(II), the Cu L-edge spectrum for the complexes is characteristic of Cu(0)/Cu(I) with higher absorption threshold energy (950 eV is shifted to 955 eV) and broad features that are associated with transitions into the more broad 4s orbitals. Hence NEXAFS also supports the conclusions drawn from the Cu 2p XPS and Auger LMM spectra.
As per the spectroscopic results, it is clear that the coordination involves the histidine side chains and it leads to deprotonation of the amine nitrogen of the imidazole ring. Cu(II) is reduced in the process to Cu(I) and Cu(0). The prevalence and comparable amounts of the Cu(I) and Cu(0) oxidation states suggests two chemically distinct coordination sites per oligoamide monomer. It is important to note that the terminal carboxyl residue of the oligoamide monomer does not participate in the hydrogen bonding network forming the helix and/or the core nanorod, it is instead tilted away from the helix similarly to a peptide side chain12 and is therefore accessible to a metal. Carboxyl moieties are used to design copper-based MOFs,43 hence it is feasible to assume that one of the distinct chemical environments is provided by carboxyl ligands, whereas the other one is defined by the histidine (imidazole) ligands as established above. Consistently the two dimensional structure seen in Fig. 2 is defined by (i) the core nanorods formed by the supramolecular head-to-tail assembly of the oligoamide monomers, and (ii) arrangement of the histidine side chains and the carboxyl moieties, and hence the core nanorods, around the metal centers following their preferred coordination geometry. Cu(I) is known to form stable complexes with 2–3 ligands, although higher coordination numbers are also possible.13,44 Cu(0) is not common among copper complexes and hence not much is known about its preferred coordination geometries. By considering the available packing geometries for the nanorods, for the complex formed with histidine side chains as ligands the square planar coordination geometry is the most likely as it requires the least distortion of the core nanorods, allowing the formation of a two dimensional ordered structure with an array of metal cores (Fig. 6). The carboxylate site is less defined, with only one carboxyl moiety per monomer; the most likely arrangement is two carboxyl moieties from diagonally opposite nanorod coordinating a metal, where the carboxylates are likely providing bidentate ligands.
Fig. 6 Schematics of the supramolecular nanorod assembly and the coordination-driven secondary assembly into sheets. |
It is important to note the proximity of the copper nuclei. The repeat distance between the histidine residues along the helix can be approximated by 2 turns of the helix (Fig. 6). Considering that the rise of the 14-helix is 4–5 Å6,12 with some variation depending on side chains, the distance between the histidine sites is thus ∼8–9 Å; the terminal carboxyl residues are situated at approximately halfway between the histidine sites, placing the metal nuclei approximately 4–4.5 Å apart. The lattice constant of metallic copper is ≈3.6 Å.45 Therefore, the distance between the copper nuclei can potentially allow the formation of direct Cu–Cu bonds. This would result in an unprecedented, practically infinite polynuclear metal complex, yielding a one-dimensional copper wire surrounded by the oligoamide nanorods.
MATLAB2019a was used to plot the data. Inkscape 0.92.4 was used to assemble vector images into the two by two grid arrangement.
Footnote |
† These two authors contributed equally. |
This journal is © The Royal Society of Chemistry 2020 |