Paramagnetic encoding of molecules

Abstract

Contactless digital tags are increasingly penetrating into many areas of human activities. Digitalization of our environment requires an ever growing number of objects to be identified and tracked with machine-readable labels. Molecules offer immense potential to serve for this purpose, but our ability to write, read, and communicate molecular code with current technology remains limited. Here we show that magnetic patterns can be synthetically encoded into stable molecular scaffolds with paramagnetic lanthanide ions to write digital code into molecules and their mixtures. Owing to the directional character of magnetic susceptibility tensors, each sequence of lanthanides built into one molecule produces a unique magnetic outcome. Multiplexing of the encoded molecules provides a high number of codes that grows double-exponentially with the number of available paramagnetic ions. The codes are readable by nuclear magnetic resonance in the radiofrequency (RF) spectrum, analogously to the macroscopic technology of RF identification. A prototype molecular system capable of 16-bit (65,535 codes) encoding is presented. Future optimized systems can conceivably provide 64-bit (~10^19 codes) or higher encoding to cover the labelling needs in drug discovery, anti-counterfeiting and other areas.

Fig. 1. The principle of paramagnetic encoding presented in this work.

A sequence of two lanthanide ions (M¹M²) is chemically written into a tripeptide molecule M¹M²-TP1, which incorporates two stereochemically well-defined amino acid/chelator building blocks. The paramagnetic effects of the lanthanides combine to influence a reporter CF₃ group. Each M¹M² permutation of the elements (codon) results in a unique ¹⁹F NMR shift (δ_F) that is given by the sum of individual paramagnetic shift contributions from each M¹ and M² (δ_p¹ and δ_p², respectively) and a constant diamagnetic shift (δ_d). The encoded molecules are further multiplexed to produce codes readable by NMR or MRI. A small number of elements allows the writing of a very high number of codes. The exact chemical structures of the amino acid/chelator building blocks used in this work are provided in Figs. 2 and 3.

Fig. 2. Chemical structures of amino acid/chelator building blocks for incorporation of Ln³⁺ ions into peptides discussed in this work.

Structures were redrawn without protective groups and metal ions from references: Cryptand-K, DOTA-K^,, DOTA-F, and DOTAla. The DO3A-Hyp family of building blocks presented in this work comprises two isomers that differ in the configuration of chiral centers in the Hyp (hydroxyproline) moiety: isomer 2R,3S,4S (L¹, green shading) and isomer 2S,3R,4S (L², violet shading).

Fig. 3. Synthetic scheme for building blocks L¹, L², and their protected variants.

Conditions: (i) BnBr (benzyl bromide), TEA (triethylamine), THF (tetrahydrofurane); (ii) PPh₃, THF, 0 °C followed by DIAD (diisopropyl azodicarboxylate), MeI, 0 °C → RT; (iii) (PhSe)₂, EtOH, NaBH₄, 0 °C; (iv) H₂O₂, THF, 0 °C → RT; (v) MCPBA (meta-chloroperoxybenzoic acid), CHCl₃, 85 °C; (vi) cyclen, t-BuOH, 105 °C; (vii) t-BuO₂CCH₂Br, K₂CO₃, MeCN; (viii) H₂, Pd@C, AcOH, MeOH; (ix) TFA; (x) FmocCl, aq. borate/NaOH buffer (pH 9.0), MeCN; (xi) MeO₂CCH₂Br, K₂CO₃, MeCN; (xii) Ac₂O, TEA, DMAP (4-dimethylaminopyridine), MeCN. Intermediates in brackets were not isolated. Colors define stereochemistry of the hydroxyproline moiety: 2R,3S,4S (green) and 2S,3R,4S (violet). Detailed synthetic procedures are provided in Supplementary Figs. 22–36.

Fig. 4. Molecular structures of [Dy(L¹)] and [Dy(L²)] in the solid state.

In both structures, the ligand coordinated to the central Dy^III cation with four ring nitrogen donors and four carboxylate donors. The view highlights the absolute configuration of the proline ring. Thermal ellipsoids were set at 50% probability. A [Dy(L¹)] contains 2R,3S,4S configuration of the proline ring (green shading) and the chelate adopts ∆λλλλ square antiprismatic (SA) conformation. For detailed information, see Supplementary Figs. 1–3. B [Dy(L²)] contains 2S,3R,4S configuration of the proline ring (violet shading) and the chelate adopts ∆δδδδ twisted square antiprismatic (TSA) conformation. For detailed information, see Supplementary Figs. 5–7.

Fig. 5. Synthesis of M¹M²-TP1 and M¹M²-TP2 compounds.

A Main synthetic pathway to well-defined M¹M²-TP1 products. B Post-synthetic complexation pathway to statistical mixtures of four major M¹M²-TP1 products. C Synthetic pathway to well-defined M¹M²-TP2 products (analogous to A). Conditions: (i) Fmoc-Phe{p-CF₃}-OH, PyAOP ((7-azabenzotriazol-1-yloxy)tripyrrolidinophosphonium hexafluorophosphate), DIPEA (N,N-diisopropylethylamine), DMSO (dimethyl sulfoxide) followed by DBU (1,8-diazabicyclo[5.4.0]undec-7-ene), DMF (dimethylformamide); (ii) Ac-Fmoc-Me₃L¹, PyAOP, DIPEA, DMSO followed by TFA (trifluoroacetic acid); (iii) M²Cl₃, aq. MOPS (3-(N-morpholino)propanesulfonic acid)/NaOH (pH 7.0) followed by LiOH, H₂O, MeOH; (iv) M¹Cl₃, aq. MOPS/NaOH (pH 7.0); (v) LiOH, H₂O, MeOH; (vi) M¹Cl₃, M²Cl₃, aq. MOPS/NaOH (pH 7.0). For the full structures of the building blocks, see Fig. 3. Detailed synthetic procedures including alternative synthetic pathways are provided in Supplementary Figs. 37–95.

Fig. 6. ¹⁹F NMR spectroscopy (470.4 MHz, T = 298.1 K) of molecules encoded with Dy³⁺/Ho³⁺ ions.

Spectra were measured in aq. MOPS/NaOH buffer (pH = 7) with external D₂O for frequency lock. TFA (trifluoroacetic acid) was used as internal NMR reference (−76.55 ppm). Charges of Ln³⁺ ions were omitted for clarity. A Statistical mixture resulting from a post-synthetic complexation of Dy³⁺/Ho³⁺ ions provided four major peaks (marked with triangles) with clearly distinguishable shifts representing all M¹M² permutations of the metal ions. The mixture was measured without purification. B M¹M²-TP1 compounds encoded with specific sequences of Dy³⁺/Ho³⁺ ions, prepared via controlled synthesis, each show one major peak matching with the statistical mixture in the panel above. C M¹M²-TP2 compounds encoded with the same sequences of Dy³⁺/Ho³⁺ ions, prepared via controlled synthesis, each show one major peak, but the shifts (both absolute and relative) are very different from the analogous M¹M²-TP1 compounds in the panels above. D M¹M²-TP3 control compounds encoded with the same sequences of Dy³⁺/Ho³⁺ ions fail to produce distinguishable peaks. E Legend for color-coding of M¹M² sequences.

Fig. 7. Molecular codes achievable with multiplexing.

A Methods that rely on direct detection of the elements (e.g., by luminescence or mass) provide a maximum of three unique codes from two elements. B Paramagnetic encoding with two elements provides four permutations (codons) with unique signals that are further multiplexed into 15 unique codes. C Physical realization of paramagnetic encoding and multiplexing with Ho³⁺/Dy³⁺ ions in M¹M²-TP1 system leading to 15 unique codes. The color-coding of the NMR peaks corresponds to specific M¹M² permutations as shown in the top middle of this panel. D Enumeration of codes achievable with methods of direct element reading vs. paramagnetic encoding for 2–4 elements and considering codon size of two or three elements for paramagnetic encoding. The null case (all zeros) is ignored, hence the –1 term in the equations.

Fig. 8. Password decoding from a single sample with z-resolved NMR.

A Physical realization of a single NMR sample containing 5 different mixtures of M¹M²-TP1 compounds encoded with Tb³⁺/Dy³⁺/Ho³⁺/Yb³⁺ ions. Aqueous solutions (in MOPS/NaOH buffer, pH = 7) of the mixtures (layers 1–5) are separated by layers of CCl₄. The capillary is submerged in D₂O. B Z-resolved ¹⁹F NMR spectrum (470 MHz, T = 298.1 K) of the sample from panel A shows signals of the M¹M²-TP1 compounds and internal standard (TFA, trifluoroacetic acid) resolved for the layers along the z-axis. Shifts from different layers are not perfectly aligned due to bulk magnetic susceptibility (BMS) effects that varied for each layer. C Evaluation of the signals from panel B by comparison with an independently obtained 1D NMR spectrum of a mixture containing all 16 M¹M²-TP1 compounds. Note that peaks of DyDy and YbTb are close but distinguished in the 1D spectrum. Black dots mark shifts from the z-resolved spectrum after correcting for BMS effects in each layer (referencing to TFA). Good match is achieved between expected shifts (vertical lines) and determined shifts (black dots). D Binary representation of the information from C. E Final table of the binary codes for 10 ASCII (American Standard Code for Information Interchange) characters encoded in the sample. Relative to D, the binary codes are rearranged to start at the position of DyDy (green shading), which is not used to avoid confusion with YbTb. F ASCII characters obtained by conversion from the binary codes in E with indication of their order in a password. G The final decoded 10-character password.

Fig. 9. Image encoding and reading with ¹⁹F MRI (4.7 T).

A Distribution pattern of M¹M²-TP1 compounds encoded with Dy³⁺/Ho³⁺ ions within a 7 × 5-well plate (final concentrations ∼0.36 mM each compound, well volumes completed to 75 μL with water). B ¹⁹F MR spectrum obtained from the entire well plate shows four distinct peaks of the four encoded compounds (marked with triangles, color-coding as explained in A). The spectrum was created by projecting the maximum signal intensity from each slice of (CSI) MRI data onto the frequency dimension. The spectral resolution was 19.2 Hz (0.10 ppm) per slice. C–F Selective display of HoDy-TP1 (C), DyDy-TP1 (D), HoHo-TP1 (E) and DyHo-TP1 (F) compounds reveals the patterns encoded with them into the well plate as legible letters (average of 15 adjacent slices for each compound peak). Displayed field of view: 50×50 mm, voxel size 0.78 × 0.78 × 10 mm. Image intensities were mapped to colors as shown in the legends. See also Supplementary Movie 1 for a fly-through view of the MRI data.