Selective metal coordination is central to the functions of metalloproteins:1,2 each metalloprotein must pair with its cognate metallocofactor to fulfil its biological role3. However, achieving metal selectivity solely through a three-dimensional protein structure is a great challenge, because there is a limited set of metal-coordinating amino acid functionalities and proteins are inherently flexible, which impedes steric selection of metals3,4. Metal-binding affinities of natural proteins are primarily dictated by the electronic properties of metal ions and follow the Irving–Williams series5 (Mn2+ < Fe2+ < Co2+ < Ni2+ < Cu2+ > Zn2+) with few exceptions6,7. Accordingly, metalloproteins overwhelmingly bind Cu2+ and Zn2+ in isolation, regardless of the nature of their active sites and their cognate metal ions1,3,8. This led organisms to evolve complex homeostatic machinery and non-equilibrium strategies to achieve correct metal speciation1,3,8,9,10. Here we report an artificial dimeric protein, (AB)2, that thermodynamically overcomes the Irving–Williams restrictions in vitro and in cells, favouring the binding of lower-Irving–Williams transition metals over Cu2+, the most dominant ion in the Irving–Williams series. Counter to the convention in molecular design of achieving specificity through structural preorganization, (AB)2 was deliberately designed to be flexible. This flexibility enabled (AB)2 to adopt mutually exclusive, metal-dependent conformational states, which led to the discovery of structurally coupled coordination sites that disfavour Cu2+ ions by enforcing an unfavourable coordination geometry. Aside from highlighting flexibility as a valuable element in protein design, our results illustrate design principles for constructing selective metal sequestration agents.
This is a preview of subscription content
Subscribe to Nature+
Get immediate online access to the entire Nature family of 50+ journals
Subscribe to Journal
Get full journal access for 1 year
only $3.90 per issue
All prices are NET prices.
VAT will be added later in the checkout.
Tax calculation will be finalised during checkout.
Get time limited or full article access on ReadCube.
All prices are NET prices.
The principal data supporting the findings of this work are available within the figures and the Supplementary Information. Additional data that support the findings of this study are available from the corresponding author on request. Crystallographic data for protein structures (coordinates and structure factors) have been deposited into the RCSB PDB under the following accession codes: 7MK4, 7LRV, 7LV1, 7N4G, 7N4F, 7LR5, 7LRA, 7LRB and 7LRR. The model structure used for molecular replacement is available in the PDB (2BC5).
Waldron, K. J., Rutherford, J. C., Ford, D. & Robinson, N. J. Metalloproteins and metal sensing. Nature 460, 823–830 (2009).
Gray, H. B., Stiefel, E. I., Valentine, J. S. & Bertini, I. Biological Inorganic Chemistry: Structure and Reactivity (University Science Books, 2007).
Waldron, K. J. & Robinson, N. J. How do bacterial cells ensure that metalloproteins get the correct metal? Nat. Rev. Microbiol. 7, 25–35 (2009).
Dudev, T. & Lim, C. Competition among metal ions for protein binding sites: determinants of metal ion selectivity in proteins. Chem. Rev. 114, 538–556 (2014).
Frausto da Silva, J. J. R. & Williams, R. J. P. The Biological Chemistry of the Elements (Oxford University Press, 2001).
Kisgeropoulos, E. C. et al. Key structural motifs balance metal binding and oxidative reactivity in a heterobimetallic Mn/Fe protein. J. Am. Chem. Soc. 142, 5338–5354 (2020).
Grāve, K., Griese, J. J., Berggren, G., Bennett, M. D. & Högbom, M. The Bacillus anthracis class Ib ribonucleotide reductase subunit NrdF intrinsically selects manganese over iron. J. Biol. Inorg. Chem. 25, 571–582 (2020).
Reyes-Caballero, H., Campanello, G. C. & Giedroc, D. P. Metalloregulatory proteins: metal selectivity and allosteric switching. Biophys. Chem. 156, 103–114 (2011).
O’Halloran, T. V. & Culotta, V. C. Metallochaperones, an intracellular shuttle service for metal ions. J. Biol. Chem. 275, 25057–25060 (2000).
Tottey, S. et al. Protein-folding location can regulate manganese-binding versus copper- or zinc-binding. Nature 455, 1138–1142 (2008).
Lombardi, A., Pirro, F., Maglio, O., Chino, M. & DeGrado, W. F. De novo design of four-helix bundle metalloproteins: one scaffold, diverse reactivities. Acc. Chem. Res. 52, 1148–1159 (2019).
Lu, Y., Yeung, N., Sieracki, N. & Marshall, N. M. Design of functional metalloproteins. Nature 460, 855–862 (2009).
Yu, F. et al. Protein design: toward functional metalloenzymes. Chem. Rev. 114, 3495–3578 (2014).
Schwizer, F. et al. Artificial metalloenzymes: reaction scope and optimization strategies. Chem. Rev. 118, 142–231 (2018).
Churchfield, L. A. & Tezcan, F. A. Design and construction of functional supramolecular metalloprotein assemblies. Acc. Chem. Res. 52, 345–355 (2019).
Faiella, M. et al. An artificial di-iron oxo-protein with phenol oxidase activity. Nat. Chem. Biol. 5, 882–884 (2009).
Zastrow, M. L., Peacock, F. A., Stuckey, J. A. & Pecoraro, V. L. Hydrolytic catalysis and structural stabilization in a designed metalloprotein. Nat. Chem. 4, 118–123 (2012).
Studer, S. et al. Evolution of a highly active and enantiospecific metalloenzyme from short peptides. Science 362, 1285–1288 (2018).
Khare, S. D. et al. Computational redesign of a mononuclear zinc metalloenzyme for organophosphate hydrolysis. Nat. Chem. Biol. 8, 294–300 (2012).
Yeung, N. et al. Rational design of a structural and functional nitric oxide reductase. Nature 462, 1079–1082 (2009).
Song, W. J. & Tezcan, F. A. A designed supramolecular protein assembly with in vivo enzymatic activity. Science 346, 1525–1528 (2014).
Churchfield, L. A., Medina-Morales, A., Brodin, J. D., Perez, A. & Tezcan, F. A. De novo design of an allosteric metalloprotein assembly with strained disulfide bonds. J. Am. Chem. Soc. 138, 13163–13166 (2016).
Zhou, L. et al. A protein engineered to bind uranyl selectively and with femtomolar affinity. Nat. Chem. 6, 236–241 (2014).
Wegner, S. V., Boyaci, H., Chen, H., Jensen, M. P. & He, C. Engineering a uranyl-specific binding protein from NikR. Angew. Chem. Int. Ed. Engl. 48, 2339–2341 (2009).
Brodin, J. D. et al. Evolution of metal selectivity in templated protein interfaces. J. Am. Chem. Soc. 132, 8610–8617 (2010).
Guffy, S. L., Der, B. S. & Kuhlman, B. Probing the minimal determinants of zinc binding with computational protein design. Protein Eng. Des. Sel. 29, 327–338 (2016).
Akcapinar, G. B. & Sezerman, O. U. Computational approaches for de novo design and redesign of metal-binding sites on proteins. Biosci. Rep. 37, BSR20160179 (2017).
Byrd, J. & Winge, D. R. Cooperative cluster formation in metallothionein. Arch. Biochem. Biophys. 250, 233–237 (1986).
Halling, D. B., Liebeskind, B. J., Hall, A. W. & Aldrich, R. W. Conserved properties of individual Ca2+-binding sites in calmodulin. Proc. Natl Acad. Sci. USA 113, E1216–E1225 (2016).
Zygiel, E. M. & Nolan, E. M. Transition metal sequestration by the host-defense protein calprotectin. Annu. Rev. Biochem. 87, 621–643 (2018).
Rittle, J., Field, M. J., Green, M. T. & Tezcan, F. A. An efficient, step-economical strategy for the design of functional metalloproteins. Nat. Chem. 11, 434–441 (2019).
Faraone-Mennella, J., Tezcan, F. A., Gray, H. B. & Winkler, J. R. Stability and folding kinetics of structurally characterized cytochrome c-b562. Biochemistry 45, 10504–10511 (2006).
Choi, T. S., Lee, H. J., Han, J. Y., Lim, M. H. & Kim, H. I. Molecular insights into human serum albumin as a receptor of amyloid-β in the extracellular region. J. Am. Chem. Soc. 139, 15437–15445 (2017).
Burgot, J.-L. Ionic Equilibria in Analytical Chemistry (Springer, 2012).
Osman, D. et al. Bacterial sensors define intracellular free energies for correct enzyme metalation. Nat. Chem. Biol. 15, 241–249 (2019).
Young, T. R. et al. Calculating metalation in cells reveals CobW acquires CoII for vitamin B12 biosynthesis while related proteins prefer ZnII. Nat. Commun. 12, 1195 (2021).
Jeschek, M. et al. Directed evolution of artificial metalloenzymes for in vivo metathesis. Nature 537, 661–665 (2016).
Thompson, A. N. et al. Mechanism of potassium-channel selectivity revealed by Na+ and Li+ binding sites within the KcsA pore. Nat. Struct. Mol. Biol. 16, 1317–1324 (2009).
Capdevila, D. A., Braymer, J. J., Edmonds, K. A., Wu, H. & Giedroc, D. P. Entropy redistribution controls allostery in a metalloregulatory protein. Proc. Natl Acad. Sci. USA 114, 4424–4429 (2017).
Tokuriki, N. & Tawfik, D. S. Protein dynamism and evolvability. Science 324, 203–207 (2009).
Motlagh, H. N., Wrabl, J. O., Li, J. & Hilser, V. J. The ensemble nature of allostery. Nature 508, 331–339 (2014).
Papaleo, E. et al. The role of protein loops and linkers in conformational dynamics and allostery. Chem. Rev. 116, 6391–6423 (2016).
Arslan, E., Schulz, H., Zufferey, R., Künzler, P. & Thöny-Meyer, L. Overproduction of the Bradyrhizobium japonicum c-type cytochrome subunits of the cbb3 oxidase in Escherichia coli. Biochem. Biophys. Res. Commun. 251, 744–747 (1998).
Bailey, J. B., Subramanian, R. H., Churchfield, L. A. & Tezcan, F. A. in Peptide, Protein and Enzyme Design: Methods in Enzymology Vol. 580 (ed. Pecoraro, V. L.) 223–250 (Academic Press, 2016).
Martel, A., Liu, P., Weiss, T. M., Niebuhr, M. & Tsuruta, H. An integrated high-throughput data acquisition system for biological solution X-ray scattering studies. J. Synchrotron Radiat. 19, 431–434 (2012).
Manalastas-Cantos, K. et al. ATSAS 3.0: expanded functionality and new tools for small-angle scattering data analysis. J. Appl. Crystallogr. 54, 343–355 (2021).
Svergun, D., Barberato, C. & Koch, M. H. J. CRYSOL—a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Crystallogr. 28, 768–773 (1995).
Collaborative Computational Project. The CCP4 suite: programs for protein crystallography. Acta Crystallogr. D 50, 760–763 (1994).
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. D 66, 486–501 (2010).
Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010).
The PyMOL Molecular Graphics System v.1.8 (Schrödinger, 2015).
Schuck, P. Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling. Biophys. J. 78, 1606–1619 (2000).
Manoil, C. & Beckwith, J. A genetic approach to analyzing membrane protein topology. Science 233, 1403–1408 (1986).
Dapprich, S., Komáromi, I., Byun, K. S., Morokuma, K. & Frisch, M. J. A new ONIOM implementation in Gaussian98. Part I. The calculation of energies, gradients, vibrational frequencies and electric field derivatives. J. Mol. Struct. THEOCHEM 461–462, 1–21 (1999).
Vreven, T., Morokuma, K., Farkas, Ö., Schlegel, H. B. & Frisch, M. J. Geometry optimization with QM/MM, ONIOM, and other combined methods. I. Microiterations and constraints. J. Comput. Chem. 24, 760–769 (2003).
Tao, P. et al. Matrix metalloproteinase 2 inhibition: combined quantum mechanics and molecular mechanics studies of the inhibition mechanism of (4-phenoxyphenylsulfonyl)methylthiirane and its oxirane analogue. Biochemistry 48, 9839–9847 (2009).
Becke, A. D. Density‐functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 98, 5648–5652 (1993).
Lee, C., Yang, W. & Parr, R. G. Development of the Colle–Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 37, 785–789 (1988).
Hariharan, P. C. & Pople, J. A. The effect of d-functions on molecular orbital energies for hydrocarbons. Chem. Phys. Lett. 16, 217–219 (1972).
Rassolov, V. A., Pople, J. A., Ratner, M. A. & Windus, T. L. 6-31G* basis set for atoms K through Zn. J. Chem. Phys. 109, 1223–1229 (1998).
Rassolov, V. A., Ratner, M. A., Pople, J. A., Redfern, P. C. & Curtiss, L. A. 6-31G* basis set for third-row atoms. J. Comput. Chem. 22, 976–984 (2001).
Freindorf, M., Shao, Y., Furlani, T. R. & Kong, J. Lennard–Jones parameters for the combined QM/MM method using the B3LYP/6-31G*/AMBER potential. J. Comput. Chem. 26, 1270–1278 (2005).
Case, D. A. et al. The Amber biomolecular simulation programs. J. Comput. Chem. 26, 1668–1688 (2005).
Bakowies, D. & Thiel, W. Hybrid models for combined quantum mechanical and molecular mechanical approaches. J. Phys. Chem. 100, 10580–10594 (1996).
Weiner, S. J., Singh, U. C. & Kollman, P. A. Simulation of formamide hydrolysis by hydroxide ion in the gas phase and in aqueous solution. J. Am. Chem. Soc. 107, 2219–2229 (1985).
Kakkis, A., Gagnon, D., Esselborn, J., Britt, R. D. & Tezcan, F. A. Metal-templated design of chemically switchable protein assemblies with high-affinity coordination sites. Angew. Chem. Int. Ed. Engl. 59, 21940–21944 (2020).
Kocyła, A., Pomorski, A. & Krężel, A. Molar absorption coefficients and stability constants of metal complexes of 4-(2-pyridylazo)resorcinol (PAR): revisiting common chelating probe for the study of metalloproteins. J. Inorg. Biochem. 152, 82–92 (2015).
Kuzmič, P. Program DYNAFIT for the analysis of enzyme kinetic data: application to HIV proteinase. Anal. Biochem. 237, 260–273 (1996).
Stoll, S. & Schweiger, A. EasySpin, a comprehensive software package for spectral simulation and analysis in EPR. J. Magn. Reson. 178, 42–55 (2006).
Smilgies, D.-M. & Folta-Stogniew, E. Molecular weight-gyration radius relation of globular proteins: a comparison of light scattering, small-angle X-ray scattering and structure-based data. J. Appl. Crystallogr. 48, 1604–1606 (2015).
We thank members of the Tezcan group, J. Rittle, H. Gray, S. Cohen and M. Green for discussions. We also thank A. Kakkis and C.-J. Yu for AUC and EPR measurements, respectively. This work was funded primarily by the National Institutes of Health (NIH; R01-GM138884) and by NASA (80NSSC18M0093; ENIGMA: Evolution of Nanomachines in Geospheres and Microbial Ancestors (NASA Astrobiology Institute Cycle 8)). Parts of this research were carried out at the Stanford Synchrotron Radiation Lightsource (supported by the DOE, Office of Basic Energy Sciences contract DE-AC02-76SF00515 and NIH P30-GM133894) and the Advanced Light Source (supported by the DOE, Office of Basic Energy Sciences contract DE-AC02-05CH11231and NIH P30-GM124169-01). EPR data was acquired on an instrument obtained through a Major Research Instrumentation fund supported by the National Science Foundation (NSF; CHE-2019066).
The authors declare no competing interests.
Peer review information
Nature thanks Jan Lipfert and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
a, Sedimentation velocity/AUC analysis of the oligomerization state of (AB)2 (black), (AC)2 (red), and (BC)2 (blue) at 25 μM dimer concentration. The inset shows a higher-order oligomer population of (BC)2. b–d,Guinier analysis of (b) AB (black), (c) AC (red), and (d) BC (blue) constructs measured by solution small-angle X-ray scattering (SAXS) at 0.8 mM dimer concentration. Larger I(0) values in (AC)2 and (BC)2 indicate higher oligomeric states compared to (AB)2. Radius of gyration (Rg) value of (AB)2 corresponds to approximately 25 kDa in Mw–Rg relationship of globular proteins70, indicating a dimeric state for (AB)2.
a, X-band EPR spectrum of CuII-(AB)2 (orange) and simulated pattern (black line), along with the fit parameters consistent with a rhombic coordination geometry. Conditions: 2.5 mM CuII, 20 mM MOPS (pH 7.4) and 150 mM NaCl, 298 K. b, Crystal structure of 2CoII-(AB)2. CoII ions are represented as magenta spheres. c, Solution characterization of 2CoII-(AB)2 complex by competitive Fura-2 titration (left), SAXS (middle), and ESI–MS (right). Circles (magenta) in ESI–MS spectrum represent the number of CoII ions bound to (AB)2. Experimental data points and error bars in the Fura-2 titration are presented as mean and standard deviation of three independent measurements. d, Competitive metal-binding titrations (AB)2 for MnII (pink) and ZnII (grey) binding. Mag-Fura-2 and Fura-2 were used for MnII and ZnII titrations, respectively. e, Investigation of MII-(AB)2 complexation using ESI–MS: MnII (top), FeII (middle), and ZnII (bottom). MnII- and FeII-(AB)2 complexes were not observed owing to the low metal-binding affinities. f, ESI–MS spectra of (AB)2 collected under CuII//ZnII and ZnII//CuII competition conditions. CuII outcompetes ZnII or forms heterometallic complexes of (AB)2. g, Theoretical SAXS scattering profiles of MII-(AB)2 generated by CRYSOL. h, i, Experimental SAXS profiles of (h) 2CoII- and 2NiII-(AB)2 complexes compared with theoretical scattering profile of 1CuII-(AB)2 (cyan), and (i) 1CuII-(AB)2 complex compared with theoretical scattering profiles of 2CoII- (magenta) and 2NiII-(AB)2 (green). j, k, Log-scale plots of Fig. 1e–f to compare (j) experimental scattering profiles between 2CoII/2NiII-(AB)2 and 1CuII-(AB)2 and (k) experimental scattering profiles of 2CoII- (left), 2NiII- (middle), and 1CuII-(AB)2 (right) with theoretical scattering profiles. l, m, Expanded low q-ranges of the scattering plots in (l) (panel j) and (m) (k).
Extended Data Fig. 3 Structural characterization and metal-binding analysis of MII-(AB)2 complexes in competitive conditions with CuII.
a, Crystal structures (ribbon models) of (AB)2 formed under CoII//CuII (PDB:7LRB), CuII//CoII (PDB:7LRR), NiII//CuII (PDB:7LR5), and CuII//NiII (PDB:7LRA) conditions. Root mean square deviation (RMSD) values were determined in comparison with the crystal structures of 2CoII-(AB)2 and 2NiII-(AB)2 (grey cartoon models). b, Experimental SAXS profile of (AB)2 (0.8 mM) in CoII//CuII, CuII//CoII, NiII//CuII, and CuII//NiII conditions ([MII] = 1.6 mM) compared with the theoretical SAXS profile of 1CuII-(AB)2. c, ESI–MS spectrum of (AB)2 (5 μM) without metal ions. d, ESI–MS spectrum of (AB)2 (5 μM) with sub-stoichiometric amounts of CuII (5 μM) and CoII or NiII (5 μM). Circles in ESI–MS spectra represent the number of CoII (magenta), NiII (green), and CuII (cyan) ions bound to (AB)2. The number of circles indicates the equivalents of bound metal ions. e, Competitive metal-binding titration of (AB)2 with CoII (magenta), NiII (green), and CuII (cyan) in 20 mM NH4HCO3 (pH 7.8). Mag-Fura-2 (5 μM) was used for competitive CoII and NiII titration, and Newport green DCF (5 μM) was used for competitive CuII titration. Experimental data points and error bars are presented as mean and standard deviation of three independent measurements. f, g, Log-scale plots (left) of Fig. 2c to compare experimental scattering profiles of (f) CoII//CuII and CuII//CoII, and (g) NiII//CuII and CuII//NiII with theoretical scattering profiles. Right panels present expanded low q-ranges of the scattering plots (left).
Extended Data Fig. 4 Peripheral 5His coordination site of 2CoII-(AB)2 and rotamer analysis of the H100 side chain.
a, CoII-coordination in the dimer interface of (AB)2 with the 2mFo-DFc electron density map (grey mesh) contoured at 5.0σ (metal) and 1.5σ (ligand). Coordination distances and angles between CoII and ligands are shown on the right. b, Possible rotamer orientations of H100 residue in the crystal structure of 2NiII-(AB)2. Rotamers were generated with the combination of χ1 (torsion angle of Cα-Cβ = 60–300°, 60° interval) and χ2 (torsion angle of Cβ-Cγ = 0–330°, 30° interval). c, Averaged number of atomic contacts of H100 rotamers as a function of van der waals (VdW) overlap. Error bars reflect the standard deviations of the atomic contacts in rotamers with fixed χ1 and variable χ2. Compared to the original conformation of H100, all possible rotamers show significant clashes with proximal residues. d, Location of proximal residues interacting with H100 rotamers.
a–c, QM/MM-optimized structures of (a) Model 1: CuII in the peripheral site of (AB)2 (b) Model 2: CuII in the peripheral site of H100A(AB)2, and (c) Model 3: CuII in the central site of (AB)2. Conformations of the opposite peripheral binding site (red) from the modelled CuII coordination site are shown in the panel on the right in a and b. d, RMSDs of Cα positions between QM/MM-optimized models and the crystal structure (His: Cα’s of 5His or 4His residues in the opposite peripheral site, All: all Cα’s in (AB)2 or H100A(AB)2, and Interface: Cα’s of residues in the dimer interface.
a, ESI–MS spectra of H100A(AB)2 (5 μM) complexed with metal ions. Circles in ESI–MS spectra represent the number of CoII (magenta), NiII (green), and CuII (cyan) ions bound to H100A(AB)2. The number of circles indicates the equivalents of bound metal ions. Metal concentrations were 10 μM. b, Fura-2 competitive metal titration assay of H100A(AB)2 with CuII. Changes in Fura-2 absorbance at 335 nm (cyan) are plotted with theoretical metal-binding isotherms in the absence (grey) and the presence (black) of (AB)2. Experimental data points and error bars in the Fura-2 titration are presented as mean and standard deviation of three independent measurements. c, d, Crystal structure of (c) CoII-H100A(AB)2 (PDB:7N4G) and (d) NiII-H100A(AB)2 (PDB:7N4F). CoII and NiII ions are represented as magenta and green spheres. Atomic details of each metal coordination site are shown in the right panels, with the 2mFo-DFc electron density map (grey mesh) contoured at 5.0σ (metal) and 1.5σ (ligand).
a, b, SDS–PAGE of periplasmic extracts and medium stained with (a) Coomassie blue for all proteinaceous contents and (b) o-dianisidine for haem-proteins. Regardless of supplemented metal ions, no significant amount of (AB)2 dimer and (AB) monomer was observed in the medium. Uncropped gel images are shown in Supplementary Fig. 2. Reproducibility of SDS–PAGE was tested with two independently extracted sample sets. c, ESI–MS spectra of (AB)2 extracted from cells grown without metal supplement (top) and with CoII (middle) or NiII (bottom) supplement (20 μM) in LB medium. d, e, ESI–MS spectra of (AB)2 extracted from cells grown with (d) 100 µM and (e) 5 µM metal ions in LB medium. Inlet spectra represent expanded m/z ranges for 2MII-(AB)2 complexes with magenta, green, and cyan lines corresponding to theoretical m/z values of 2CoII, 2NiII, and 2CuII-(AB)2 complexes. Supplemented metal ions are represented as magenta (CoII), green (NiII), and cyan (CuII) circles. The number of circles indicates the equivalents of metal ions bound to (AB)2. D and M in gel pictures refer to the protein dimer and monomer, respectively.
About this article
Cite this article
Choi, T.S., Tezcan, F.A. Overcoming universal restrictions on metal selectivity by protein design. Nature 603, 522–527 (2022). https://doi.org/10.1038/s41586-022-04469-8