Peptide-Controlled Assembly of Macroscopic Calcium Oxalate Nanosheets

Abstract

The fabrication of two-dimensional (2D) biomineral nanosheets is of high interest owing to their promise for applications in electronics, filtration, catalysis, and chemical sensing. Using a facile approach inspired by biomineralization in nature, we fabricate laterally macroscopic calcium oxalate nanosheets using β-folded peptides. The template peptides are composed of repetitive glutamic acid and leucine amino acids, self-organized at the air–water interface. Surface-specific sum frequency generation spectroscopy and molecular dynamics simulations reveal that the formation of oxalate nanosheets relies on the peptide–Ca²⁺ ion interaction at the interface, which not only restructures the peptides but also templates Ca²⁺ ions into a calcium oxalate dihydrate lattice. Combined, this enables the formation of a critical structural intermediate in the assembly pathway toward the oxalate sheet formation. These insights into peptide–ion interfacial interaction are important for designing novel inorganic 2D materials.

The design and synthesis of 2D nanomaterials have opened up a new field of ultrathin, functional materials with atomic precision.¹ Organic sheet materials such as graphene have received much attention recently. Inorganic 2D nanomaterials composed of metal oxides, ceramics, and minerals also hold great promise as materials with new physical and chemical properties, which can be chemically tuned and self-assembled using effective, low-cost, bottom-up fabrication methods.² Synthesized from relatively cheap building blocks, such materials offer great flexibility in tailoring surface morphology, porosity, chemical functionality, and electronic properties. One particularly interesting strategy to obtain functional 2D inorganic materials has been their templated assembly via organic precursor structures. Here, the chemical and polymorph structure can be encoded in the chemistry and structure of the organic templating layer. Nature uses this mechanism with proteins, which “engineer” intricately mineralized tissue by steering the growth of minerals and thereby achieving polymorph and shape control in, e.g., bone and teeth³ and exoskeletons of mollusks or diatoms.⁴ Over the past years, researchers have therefore been taking clues from Nature in an attempt to mimic the control that specialized biomolecules involved in hard tissue biogenesis exert over the formation of inorganic layered materials. Despite the great efforts, biomimetic fabrication of 2D inorganic materials still largely relies on trial and error and, therefore, typically requires substantial investments of time and funds.

Surfactants,⁵ lipids,⁶ peptoids,^7,8 and polymers⁹ have been reported to grow 2D organic sheets. Such structures can potentially steer the growth of biomimetic mineral nanosheets at interfaces, such as the air–water¹⁰ or oil–water interfaces,⁸ into desired polymorphs and morphologies. Solid-supported layers have also been used for the synthesis of layered materials.¹ Oxide and ceramic materials are particularly interesting in this context for the assembly of electronic circuitry, semiconductors, photonic materials, as well as nanometer thin membranes for desalination, filtration, chemical separation, catalytic applications, and sensors. For biomedical or nanofabrication applications, biomineral nanosheets have been explored recently.^2,10,11 In particular, calcium carbonate and calcium oxalate (CaC₂O₄) are versatile biocompatible materials, which are chemically and structurally extremely stable.

The synthesis of extended calcium carbonate sheets using biomolecules as organic templates has recently been reported.¹⁰ In contrast, CaC₂O₄, the dominant part of kidney stones^12,13 and notoriously stable, has largely remained unexplored as a functional material. Despite the extraordinary stability and inertness of CaC₂O₄, for, e.g., coatings to protect marble and archeological monuments from environmental degradation,¹⁴ CaC₂O₄ precipitation studies have remained limited to particles and bulk solids; applications for 2D nanomaterials have thus far not been exploited.

To investigate the potential of CaC₂O₄ biominerals as a 2D material, we report here the synthesis of CaC₂O₄ nanosheets at the air–water interface using biomimetic peptides (Figure 1A). Proteins that can effectively bind to a mineral surface are generally also effective nucleators of that mineral. Our choice of peptide for this experiment was inspired by the highly acidic human urinary protein osteopontin; it can specifically bind and passivate (CaC₂O₄) kidney stones and thereby avoid the growth of larger stones.^12,13 The peptide sequence NH₂–E(LE)₉–COOH (abbreviated as LE10, Figure 1A) used in this study mimics the high glutamic acid content of osteopontin and has been shown to organize into strand-like structures at the air–water interface,¹⁰ reminiscent of binding sites in osteopontin. In a recent study, we found that β-folded leucine–glutamic acid (LE) peptides can effectively assemble vaterite calcium carbonate nanosheets by exploiting their structural flexibility at the air–water interface,¹⁰ which prompts us to explore this designed LE10 sequence to fabricate a novel 2D nanomaterial based on calcium oxalate.

Figure 1.

(A) LE10 peptide-mediated surface nucleation of CaC₂O₄ nanosheets. The sheets can be lifted off from the water surface with a TEM grid. (B) SEM image of a LE10 templated CaC₂O₄ sheet collected from the air–water interface using a TEM copper grid. Red arrows indicate the broken part of the sheet; defect-free sheets over 50 × 50 μm² can routinely be achieved.

Indeed, CaC₂O₄ nanosheets can be achieved by mineralizing CaC₂O₄ using the template of assembled LE10 peptides at the air–water interface. Figure 1A illustrates the formation process of the CaC₂O₄ nanosheet: Firstly, the LE10 peptides were assembled at the air–water interface of a 20 mL trough filled with a 0.1 mg/mL solution of LE10 peptides. Upon surface adsorption, the subphase was diluted by a factor of 8 to reduce the bulk peptide concentration. Subsequently, calcium cations (Ca²⁺) and oxalate anions (C₂O₄^2–) were injected into the subphase, this procedure led to CaC₂O₄ mineralization by LE10 peptides at the air–water interface. A thin layer of biomineral, consisting of mineral and peptides, could be lifted off of the interface with a transmission electron microscopy (TEM) grid. The scanning electron microscopy (SEM) image in Figure 1B shows that the CaC₂O₄ sheet extends over the entire grid and is without visible cracks or fractures over an area of 50 × 50 μm². The sheet is intact and freestanding over most of the grid holes where the film is without any support. The sheets shown in Figure 1B extend over several square millimeters across the TEM grid with comparable quality. It is challenging to retrieve the atomically resolved CaC₂O₄ crystal structure within the nanometer sheet from high-resolution TEM measurement, in particular, taking into account the fact that the CaC₂O₄ crystals are typically decomposed into the amorphous phase by interaction with high-energy electrons.¹⁵ However, the observable diffuse rings from electron diffraction (Figure S3) suggests that the CaC₂O₄ crystals in the sheet are rather arranged into ordered nuclei prior to crystal growth, which is also in agreement with the nonclassic views of CaC₂O₄ mineralization.¹⁵

X-ray photoelectron spectroscopy (XPS) analysis indicates that the nanosheets have a thickness of ∼2 nm, and the elemental composition of the sheets (Table S1) reveals that each Ca²⁺ ion is shared by ∼2 glutamic acid side chains of peptides. No contaminants or unexpected elements were detected within the sheets. The XP spectra (Figure S1) are in agreement with those of hybrid CaC₂O₄/peptide sheets forming directly at the peptide interface. Combined, the SEM and XPS results clearly show that LE10 peptides grows 2D CaC₂O₄ nanosheets.

To elucidate the molecular mechanism underlying CaC₂O₄ nanosheet formation, we applied surface-specific sum frequency generation (SFG) spectroscopy. SFG is a second-order nonlinear spectroscopy, sensitive to ordered species exclusively at the interface.¹⁶⁻¹⁹ This technique allows us to follow the molecular details of peptide arrangement, at the interface and in situ, during nanosheet formation. In our SFG measurement in the amide I region, D₂O was used as the solvent instead of H₂O to avoid spectral interference from H₂O bending modes. Figure 2A shows SFG spectra using an SPS (S-polarized SFG, P-polarized vis, S-polarized IR) polarization combination for LE10 peptides assembled at the air–water interface, at different times after Ca²⁺ ion (CaCl₂) injection, and after CaC₂O₄ mineralization (Na₂C₂O₄ injection). The chosen SPS polarization combination is sensitive to both backbone and Glu side chain modes.¹⁰ The spectra of the pristine peptides exhibit two bands at ∼1576 and ∼1612 cm^–1, corresponding to the asymmetric COO^– stretching vibration from the deprotonated Glu side chains^20,21 and the B2 mode of the β-sheet backbone, respectively.^22,23 The presence of two distinct bands indicates that both peptide backbone and side chains are well ordered at the buffer–air interface. Both bands increase in intensity upon Ca²⁺ ion interaction: the side chain band increases slightly, while the backbone amide I peak increases drastically, to saturate after ∼3 h. The gradual amide I band increase indicates that the β-strands refold into more compact, ordered states. A weak feature at ∼1690 cm^–1 appears after ∼2 h of Ca²⁺ ion interaction; this high-frequency band can be assigned to β-turn motifs in less folded peptide aggregates.²⁴ A detailed spectral analysis (Figure S6) reveals a slight frequency shift of the central amide I band from 1605 to 1612 cm^–1 with time, possibly related to the couplings between the local modes of the turn motifs.

Figure 2.

(A,B) Experimental SFG spectra in the amide I region for LE10 peptides at the air–water interface (gray), with Ca²⁺ ion interaction at different times after CaCl₂ injection (red), and following CaC₂O₄ mineralization (blue). The spectra were acquired under (A) achiral SPS and (B) chiral PSP polarization combinations. The fits (black) are superimposed onto the experimental data. (C,D) Calculated amide I SFG spectra for peptides before and after Ca²⁺ ion interaction agree well with experimental spectra.

The backbone folding structure can be resolved in more detail using SFG spectra under chiral PSP polarization combinations (Figure 2B) that exclusively probe the amide bonds from β-folded peptides without spectral interference from achiral side chain modes.^25,26 As expected, those spectra show only the predominant chiral amide I peak centered at ∼1620 cm^–1, which originates from the β-sheet backbone,²⁶ while the side chain band lacking chirality is fully suppressed. In accordance with the SPS spectra in Figure 2A, the interaction with Ca²⁺ ions enhances the amide I intensity and gives rise to a shoulder at ∼1690 cm^–1, testifying to the peptide backbones refolding into more ordered, compact β-turns.

The SFG results provide direct experimental proof of peptide refolding upon interaction with Ca²⁺ ions—a key step for CaC₂O₄ sheet formation. For a molecular picture visualizing the critical assembled intermediate of peptides and Ca²⁺ ions, we performed molecular dynamics (MD) simulations for LE10 peptides in the absence and presence of Ca²⁺ ions. Snapshots in Figure 3A show the strand-like assembly of LE10 in the absence of Ca²⁺ ions. This secondary structure is also expected based on folding of the analogue positively charged leucine–lysine (LK) peptides at interfaces.²⁷ For one unbiased result, Ca ions were randomly placed into the simulation box, with the initial snapshots shown in Figure S8. After simulation, Ca²⁺ ions were highly coordinated into interfacial peptides due to their strong interaction. A closer look (inset in Figure 3B) shows that each Ca²⁺ ion was chelated by two glutamic acid sites, which is in good agreement with the Calcium complexation ratio within the CaC₂O₄ sheet as determined from our XPS analysis. Here, the chelated Ca²⁺ ions, acting as the salt bridge, drive the backbones to “curl” and fold back on themselves, leading to more turn-like secondary structures, which is in agreement with our SFG band assignment.

Figure 3.

(A) Snapshots of MD simulations after 100 ns without and with Ca²⁺ ions. (B) Detailed view of glutamic acid side chains chelating Ca ions. Color scheme for rendered images: Ca ions, blue; peptide backbone, green; oxygen, red. (C) Radial distribution function G(r) of the Ca–Ca distance within the simulations (red). G(r_Ca–Ca) from crystal structure data for oxalate structures COM (black) and COD (blue) are also shown (see the main text for details).

We calculated theoretical SFG spectra of the amide I band over 200 frames from 100 to 500 ns in 2 ns steps for the four simulations, according to the formalism described in ref (28) (see the SI for more details). In calculations, the hydrogen bond-induced local-mode frequency shifts are modeled via the correlation found by Cho et al. among factors including the strength of the accepted and/or donated hydrogen bond(s), the C=O length in the amide groups, and the local-mode amide I frequency.²⁹ Specifically, we take the C=O length from the coordinates and apply the equation ν = ν₀ + αδr_CO, where ν₀ is the base frequency, α is the proportionality constant, and δr_CO is the deviation of the carbon–oxygen bond from its equilibrium value, which was 1.229 Å for amide groups with secondary amines. The optimal match with the experimental spectra was found with α = 1000 and ν₀ = 1622 cm^–1. The calculated amide I spectra, as presented in Figure 2C,D, capture the band positions, spectral shape, and intensity increase observed for the amide I band in the experiments. The gradual appearance of a small band at ∼1690 cm^–1 in Figure 2A,B is not visible in the calculated spectra; this high-frequency mode is related to the aggregated β-turns from many peptide monomers, which can not be formed by the limited peptides as placed in the simulation box. However, the generally good agreement between calculation and experiment lends credence to simulation and prompted us to perform further analysis. Non-classical crystallization theories suggest that the Ca²⁺ ions coordinated by peptide side chains may template the CaC₂O₄ lattice and steer its growth within the sheet.^15,30 We calculate the radial distribution function (RDF) of Ca²⁺ ions (Figure 3C), which shows a dominant peak near 4 Å. Comparing this number with the calcium spacing within the Calcium oxalate mono- (COM) and dihydrate (COD) polymorphs, it becomes clear that, while the calcium spacing is smaller in COM, there is a remarkable match of the Ca²⁺ ion spacing within the peptide layer with the calcium spacing in COD. The MD simulations therefore suggest that LE10 peptides can effectively stabilize COD. Indeed, SEM images of the sheets collected after extended assembly times show the octahedral crystal shapes expected for COD (Figure S2). While X-ray diffraction (XRD) measurements of the sheets have not yielded a diffraction signal, we have analyzed the crystal structure of CaC₂O₄ precipitated by LE10 in bulk solution. the XRD pattern further supports COD polymorph production by LE10 peptides (Figures S4 and S5).

In conclusion, using a facile approach, we have successfully fabricated CaC₂O₄ nanosheets by employing biomimetic β-sheet LE10 peptides assembled at the air–water interface. The obtained nanosheets are freestanding with defect-free areas exceeding 50 × 50 μm². Combined SFG and molecular dynamics simulation studies indicate that the nanosheet formation is enabled by the key interplay between peptides and Ca²⁺ ions: binding of the Ca²⁺ ions refines the peptide’s structural motif, which in turn templates the Ca²⁺ ions to nucleate the CaC₂O₄ nanosheet by matching the COD lattice. The study shows that the inclusion of ion interactions in the design of functional peptides will be important for the design and fabrication of 2D inorganic biomimetic mineral nanosheets. This strategy could potentially be used to design mineral sheets using related templating molecules such as peptoids and block copolymers.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpclett.9b00684.

• Description of experimental procedures, more characterization on XRD, SEM, TEM, and XPS, description of SFG analysis and fitting parameters, description of MD simulations and SFG spectra calculation (PDF)

The authors declare no competing financial interest.