Letter | Published:

Programmable design of orthogonal protein heterodimers


Specificity of interactions between two DNA strands, or between protein and DNA, is often achieved by varying bases or side chains coming off the DNA or protein backbone—for example, the bases participating in Watson–Crick pairing in the double helix, or the side chains contacting DNA in TALEN–DNA complexes. By contrast, specificity of protein–protein interactions usually involves backbone shape complementarity1, which is less modular and hence harder to generalize. Coiled-coil heterodimers are an exception, but the restricted geometry of interactions across the heterodimer interface (primarily at the heptad a and d positions2) limits the number of orthogonal pairs that can be created simply by varying side-chain interactions3,4. Here we show that protein–protein interaction specificity can be achieved using extensive and modular side-chain hydrogen-bond networks. We used the Crick generating equations5 to produce millions of four-helix backbones with varying degrees of supercoiling around a central axis, identified those accommodating extensive hydrogen-bond networks, and used Rosetta to connect pairs of helices with short loops and to optimize the remainder of the sequence. Of 97 such designs expressed in Escherichia coli, 65 formed constitutive heterodimers, and the crystal structures of four designs were in close agreement with the computational models and confirmed the designed hydrogen-bond networks. In cells, six heterodimers were fully orthogonal, and in vitro—following mixing of 32 chains from 16 heterodimer designs, denaturation in 5 M guanidine hydrochloride and reannealing—almost all of the interactions observed by native mass spectrometry were between the designed cognate pairs. The ability to design orthogonal protein heterodimers should enable sophisticated protein-based control logic for synthetic biology, and illustrates that nature has not fully explored the possibilities for programmable biomolecular interaction modalities.

Access optionsAccess options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Data availability

Coordinates and structure files have been deposited in the Protein Data Bank with accession codes: 6DMP (DHD13_XAAA), 6DKM (DHD131), 6DLC (DHD37_1:234), 6DLM (DHD127), 6DMA (DHD15 heterodimer) and 6DM9 (DHD15 heterotetramer). The native MS spectra generated and analysed during the current study are available at http://files.ipd.uw.edu/pub/de_novo_heterodimers_2018/180813_native_ms_raw.zip. Raw X-ray diffraction images have been deposited at https://proteindiffraction.org/. All source data are available upon request.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


  1. 1.

    Jones, S. & Thornton, J. M. Principles of protein–protein interactions. Proc. Natl Acad. Sci. USA 93, 13–20 (1996).

  2. 2.

    Harbury, P. B., Zhang, T., Kim, P. S. & Alber, T. A switch between two-, three-, and four-stranded coiled coils in GCN4 leucine zipper mutants. Science 262, 1401–1407 (1993).

  3. 3.

    Diss, M. L. & Kennan, A. J. Orthogonal recognition in dimeric coiled coils via buried polar-group modulation. J. Am. Chem. Soc. 130, 1321–1327 (2008).

  4. 4.

    Thomas, F., Boyle, A. L., Burton, A. J. & Woolfson, D. N. A set of de novo designed parallel heterodimeric coiled coils with quantified dissociation constants in the micromolar to sub-nanomolar regime. J. Am. Chem. Soc. 135, 5161–5166 (2013).

  5. 5.

    Crick, F. H. C. The Fourier transform of a coiled-coil. Acta Cryst. 6, 685–689 (1953).

  6. 6.

    Zarrinpar, A., Park, S.-H. & Lim, W. A. Optimization of specificity in a cellular protein interaction network by negative selection. Nature 426, 676–680 (2003).

  7. 7.

    Aakre, C. D. et al. Evolving new protein–protein interaction specificity through promiscuous intermediates. Cell 163, 594–606 (2015).

  8. 8.

    Joachimiak, L. A., Kortemme, T., Stoddard, B. L. & Baker, D. Computational design of a new hydrogen bond network and at least a 300-fold specificity switch at a protein–protein interface. J. Mol. Biol. 361, 195–208 (2006).

  9. 9.

    Skerker, J. M. et al. Rewiring the specificity of two-component signal transduction systems. Cell 133, 1043–1054 (2008).

  10. 10.

    Crooks, R. O., Baxter, D., Panek, A. S., Lubben, A. T. & Mason, J. M. Deriving heterospecific self-assembling protein–protein interactions using a computational interactome screen. J. Mol. Biol. 428, 385–398 (2016).

  11. 11.

    Gradišar, H. & Jerala, R. De novo design of orthogonal peptide pairs forming parallel coiled-coil heterodimers. J. Pept. Sci. 17, 100–106 (2011).

  12. 12.

    Thompson, K. E., Bashor, C. J., Lim, W. A. & Keating, A. E. SYNZIP protein interaction toolbox: in vitro and in vivo specifications of heterospecific coiled-coil interaction domains. ACS Synth. Biol. 1, 118–129 (2012).

  13. 13.

    Reinke, A. W., Grant, R. A. & Keating, A. E. A synthetic coiled-coil interactome provides heterospecific modules for molecular engineering. J. Am. Chem. Soc. 132, 6025–6031 (2010).

  14. 14.

    Acharya, A., Rishi, V. & Vinson, C. Stability of 100 homo and heterotypic coiled-coil a-a′ pairs for ten amino acids (A, L, I, V, N, K, S, T, E, and R). Biochemistry 45, 11324–11332 (2006).

  15. 15.

    Grigoryan, G. & Keating, A. E. Structure-based prediction of bZIP partnering specificity. J. Mol. Biol. 355, 1125–1142 (2006).

  16. 16.

    Gonzalez, L. Jr, Woolfson, D. N. & Alber, T. Buried polar residues and structural specificity in the GCN4 leucine zipper. Nat. Struct. Biol. 3, 1011–1018 (1996).

  17. 17.

    Lumb, K. J. & Kim, P. S. A buried polar interaction imparts structural uniqueness in a designed heterodimeric coiled coil. Biochemistry 34, 8642–8648 (1995).

  18. 18.

    Tatko, C. D., Nanda, V., Lear, J. D. & Degrado, W. F. Polar networks control oligomeric assembly in membranes. J. Am. Chem. Soc. 128, 4170–4171 (2006).

  19. 19.

    Grigoryan, G. & Degrado, W. F. Probing designability via a generalized model of helical bundle geometry. J. Mol. Biol. 405, 1079–1100 (2011).

  20. 20.

    Huang, P.-S. et al. High thermodynamic stability of parametrically designed helical bundles. Science 346, 481–485 (2014).

  21. 21.

    Boyken, S. E. et al. De novo design of protein homo-oligomers with modular hydrogen-bond network-mediated specificity. Science 352, 680–687 (2016).

  22. 22.

    Leaver-Fay, A. et al. ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods Enzymol. 487, 545–574 (2011).

  23. 23.

    Ruotolo, B. T. & Robinson, C. V. Aspects of native proteins are retained in vacuum. Curr. Opin. Chem. Biol. 10, 402–408 (2006).

  24. 24.

    Sahasrabuddhe, A. et al. Confirmation of intersubunit connectivity and topology of designed protein complexes by native MS. Proc. Natl Acad. Sci. USA 115, 1268–1273 (2018).

  25. 25.

    Zhou, M., Huang, C. & Wysocki, V. H. Surface-induced dissociation of ion mobility-separated noncovalent complexes in a quadrupole/time-of-flight mass spectrometer. Anal. Chem. 84, 6016–6023 (2012).

  26. 26.

    Zhou, M. & Wysocki, V. H. Surface induced dissociation: dissecting noncovalent protein complexes in the gas phase. Acc. Chem. Res. 47, 1010–1018 (2014).

  27. 27.

    Anderson, G. P., Shriver-Lake, L. C., Liu, J. L. & Goldman, E. R. Orthogonal synthetic zippers as protein scaffolds. ACS Omega 3, 4810–4815 (2018).

  28. 28.

    Rothemund, P. W. K. Folding DNA to create nanoscale shapes and patterns. Nature 440, 297–302 (2006).

  29. 29.

    Qian, L. & Winfree, E. Scaling up digital circuit computation with DNA strand displacement cascades. Science 332, 1196–1201 (2011).

  30. 30.

    Zhang, Y. & Skolnick, J. TM-align: a protein structure alignment algorithm based on the TM-score. Nucleic Acids Res. 33, 2302–2309 (2005).

  31. 31.

    Rocklin, G. J. et al. Global analysis of protein folding using massively parallel design, synthesis, and testing. Science 357, 168–175 (2017).

  32. 32.

    Schrödinger. The PyMOL Molecular Graphics System, Version 1.8. (2015).

  33. 33.

    Kabsch, W. XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125–132 (2010).

  34. 34.

    Otwinowski, Z. & Minor, W. Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326 (1997).

  35. 35.

    McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674 (2007).

  36. 36.

    Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D 66, 213–221 (2010).

  37. 37.

    Afonine, P. V. et al. Joint X-ray and neutron refinement with phenix.refine. Acta Crystallogr. D 66, 1153–1163 (2010).

  38. 38.

    Terwilliger, T. C. et al. Iterative model building, structure refinement and density modification with the PHENIX AutoBuild wizard. Acta Crystallogr. D 64, 61–69 (2008).

  39. 39.

    Emsley, P. & Cowtan, K. Coot: model-building tools for molecular graphics. Acta Crystallogr. D 60, 2126–2132 (2004).

  40. 40.

    Davis, I. W. et al. MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 35, W375–W383 (2007).

  41. 41.

    Dyer, K. N. et al. High-throughput SAXS for the characterization of biomolecules in solution: a practical approach. Methods Mol. Biol. 1091, 245–258 (2014).

  42. 42.

    Rambo, R. P. & Tainer, J. A. Characterizing flexible and intrinsically unstructured biological macromolecules by SAS using the Porod-Debye law. Biopolymers 95, 559–571 (2011).

  43. 43.

    Schneidman-Duhovny, D., Hammel, M. & Sali, A. FoXS: a web server for rapid computation and fitting of SAXS profiles. Nucleic Acids Res. 38, W540–W544 (2010).

  44. 44.

    Schneidman-Duhovny, D., Hammel, M., Tainer, J. A. & Sali, A. Accurate SAXS profile computation and its assessment by contrast variation experiments. Biophys. J. 105, 962–974 (2013).

  45. 45.

    Schiestl, R. H. & Gietz, R. D. High efficiency transformation of intact yeast cells using single stranded nucleic acids as a carrier. Curr. Genet. 16, 339–346 (1989).

  46. 46.

    Chien, C. T., Bartel, P. L., Sternglanz, R. & Fields, S. The two-hybrid system: a method to identify and clone genes for proteins that interact with a protein of interest. Proc. Natl Acad. Sci. USA 88, 9578–9582 (1991).

  47. 47.

    Bartel, P. L., Roecklein, J. A., SenGupta, D. & Fields, S. A protein linkage map of Escherichia coli bacteriophage T7. Nat. Genet. 12, 72–77 (1996).

  48. 48.

    Guzmán, C., Bagga, M., Kaur, A., Westermarck, J. & Abankwa, D. ColonyArea: an ImageJ plugin to automatically quantify colony formation in clonogenic assays. PLoS ONE 9, e92444 (2014).

  49. 49.

    Dyachenko, A. et al. Tandem native mass-spectrometry on antibody-drug conjugates and submillion Da antibody–antigen protein assemblies on an orbitrap EMR equipped with a high-mass quadrupole mass selector. Anal. Chem. 87, 6095–6102 (2015).

  50. 50.

    Waitt, G. M., Xu, R., Wisely, G. B. & Williams, J. D. Automated in-line gel filtration for native state mass spectrometry. J. Am. Soc. Mass Spectrom. 19, 239–245 (2008).

  51. 51.

    VanAernum, Z. et al. Surface-induced dissociation of noncovalent protein complexes in an extended mass range Orbitrap mass spectrometer. Preprint available at https://doi.org/10.26434/chemrxiv.7415603.v1 (2018)

  52. 52.

    Marty, M. T. et al. Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensembles. Anal. Chem. 87, 4370–4376 (2015).

  53. 53.

    Bern, M. et al. Parsimonious charge deconvolution for native mass spectrometry. J. Proteome Res. 17, 1216–1226 (2018).

  54. 54.

    Jones, D. T. Protein secondary structure prediction based on position-specific scoring matrices. J. Mol. Biol. 292, 195–202 (1999).

Download references


We thank Rosetta@Home volunteers for contributing computing resources; A. Kang for protein crystallization support; B. Sankaran for assistance with diffraction data collection; K. Lau and B. Groves for assistance with Y2H assays; S. Rettie for MS support; S. Ovchinnikov for help with TMalign; M. Marty, M. Bern and A. Norris for assistance with native MS; S. Pennington for making media for Y2H assays; the SIBYLS mail-in SAXS program, supported by the DOE BER IDAT grant (DE-AC02-05CH11231) and ALS-ENABLE (P30 GM124169) for SAXS; and A. Keating, G. Rocklin and N. Woodall for feedback on the manuscript. Additional funding and computing resources are listed in the Supplementary Information.

Reviewer information

Nature thanks G. Grigoryan, C. Robinson and the other anonymous reviewer(s) for their contribution to the peer review of this work.

Author information

Z.C., S.E.B. and D.B. designed the research. Z.C. and D.B. wrote the manuscript. M.J., F.B., Z.L.V., A.S. and V.H.W. performed native MS experiments and analysed data. L.P.C. prepared proteins for NMR experiments. D.F.-S. and N.G.S. performed NMR experiments. Z.C. wrote the heptad stacking code. S.E.B. improved the HBNet method. V.K.M. wrote the parametric backbone generation code. T.J.B. wrote the loop closure code. Z.C. and S.E.B. carried out design calculations, and R.A.L. and S.B. helped. Z.C., M.J.B., P.L. and F.D. solved crystal structures. All authors discussed results and commented on the manuscript.

Competing interests

Z.C., S.E.B., R.A.L., S.B. and D.B. are inventors on US provisional patent application no. 62755264 and patent application WO2017173356A1. D.B. and S.E.B. hold equity in Lyell Immunopharma.

Correspondence to David Baker.

Extended data figures and tables

Extended Data Fig. 1 Overview of different topologies designed.

ad, Overall topologies on the left and example HBNets on the right. a, A left-handed supercoiled backbone, with each monomer being helix hairpins. b, A backbone-permuted ‘3 + 1’ design; one monomer is a single helix and the other is a three-helix bundle. c, A left-handed supercoiled backbone, with each monomer being a three-helix bundle. d, A straight, untwisted backbone, with each monomer being a helix hairpin. e, Hydrogen-bond pairing in DNA bases. Top, A–T base pairing. Bottom, C–G base pairing. Green arrows point from hydrogen-bond donors to acceptors. f, Two examples of hydrogen-bond pairing in designed protein hydrogen-bond networks. g, Top-down view of antiparallel twisted (top) and parallel untwisted (bottom) backbones sampled in this study. h, Comparison of a designed protein heterodimer (right) with B-form DNA (left) on the same scale.

Extended Data Fig. 2 Example HBNets resulting from the systematic search.

a, Overlay of 50 backbones with different Crick parameters for each helix. b, Example hydrogen-bond networks from the systematic search, each involving at least four residues and contacting all four helices.

Extended Data Fig. 3 Thermal and chemical denaturation of DHDs.

a, b, CD spectra for thermal denaturation of DHD_15 and DHD_20, respectively. Top, wavelength scan at 25 °C, 75 °C, 95 °C and final 25 °C. Designs were α-helical and stable up to 95 °C. Bottom, CD temperature melts, monitoring absorption at 222 nm as temperature was increased from 25 °C to 95 °C. c, GdnHCl denaturation of DHD_127 measured by CD monitoring absorption at 222 nm. All CD experiments were performed once.

Extended Data Fig. 4 Backbone and hydrogen-bond network permutations.

a, On a 2 + 2 backbone (left), two loops were designed to connect the four helices into a single monomer in two different ways (middle), after which four different cut points were introduced to generate four possible backbone-permuted heterodimers of a single helix and a three helix bundle (3 + 1 heterodimers, right). For example, 2:134 refers to a heterodimer in which the original helix 2 is a single helix, and helices 1, 3 and 4 were connected into a three-helix bundle. b, Hydrogen-bond network permutation. Each unique network was assigned a letter (networks ‘A’ and ‘B’ in this case), with the hydrophobic packing assigned X. The backbone on the left reads ‘ABXB’; its first heptad accommodates network A, its second and fourth heptad accommodate network B, and its third heptad accommodates hydrophobic packing only (X).

Extended Data Fig. 5 Biophysical characterization of hydrogen-bond-network-permuted homodimers.

a, SEC traces of all six homodimer designs. bg, SAXS profiles of hydrogen-bond network-permuted homodimer designs. Black, experimental SAXS data; red, spectra computed from the designed backbones. Two (a) or one (bg) biologically independent repeats were performed.

Extended Data Fig. 6 SAXS profiles of all tested DHDs.

Black, experimental SAXS data; red, spectra computed from the designed backbones. a, SAXS profiles with χ values smaller than 6. b, SAXS profiles with χ values greater than 6. All tested designs showed close agreement to expected radius of gyration (Rg) and maximum distance (dmax).

Extended Data Fig. 7 Crystal structure of the domain-swapped DHD_15 and biophysical characterization of higher-order oligomers.

a, Crystal structure of DHD_15 at pH 6.5, with 2.25 Å resolution. b, Superposition of design models (in colour) onto both halves of the crystal structure (in white), with backbone r.m.s.d. of 1.83 Å. c, Native MS study of DHD_15 at different pH values indicates that heterodimers, rather than heterotetramers, are dominant in solution. dg, SEC traces of the induced dimerization DHD_9-13 fusion (d), DHD_15-37 fusion (e), DHD_13-37 fusion (f), and the scaffolding complex in Fig. 3d (g; the peak at around 15 ml corresponds to the fully assembled complex, followed by a peak representing an excess of individual components). h, CD thermal melt curves for the scaffolding complex in Fig. 3d. Wavelength scan was performed at 25 °C, 75 °C, 95 °C and final 25 °C. Design was α-helical and stable up to 95 °C. i, CD chemical denaturation profile of the scaffolding complex in Fig. 3d. Two (cg) or one (h, i) biologically independent repeats were performed.

Extended Data Fig. 8 Y2H all-against-all assay of 16 DHDs.

a, Y2H assay with cell growth on agar plates containing 100 mM 3-AT, lacking tryptophan, leucine and histidine. Plates were imaged on day 5. Yellow, no growth on agar plates; light blue, weak growth forming non-circular colonies; dark blue, strong growth. b, Y2H result by growing yeast culture in liquid medium containing 100 mM 3-AT, lacking tryptophan, leucine and histidine. OD600 values were measured on day 2 to evaluate cell growth. c, An additional set of DHDs tested by Y2H showing improved orthogonality. d, Distribution of OD600 values for non-cognate interactions in b. The majority of cells grew to OD600 < 0.4, indicating weak interactions for non-cognate binding. eg, Box plots of various properties for designs that assembled to off-target oligomeric states by native MS (failure) and that assembled into constitutive heterodimers (success). n = 88; 25th, 50th and 75th percentiles are shown in the box with the centre being median, extended to 1.5 × interquartile range (IQR) beyond the box. e, The number of buried bulky polar residues correlates strongly with design success. f, Successful designs tend to have a bigger polar interface surface area. g, Designs with better hydrophobic packing (as reported by the Rosetta filter value Average Degree on Ile, Leu and Val residues) tend to have a higher chance of being constitutive heterodimers as assessed by native MS. h, Contribution of bulky residues and hydrogen-bond networks to specific dimer formation. dSASA_polar measures interface hydrophilicity and correlates positively with the surface area of hydrogen-bond networks at the interface. Bulky polar residues in core counts the total number of buried bulky residues that participate in hydrogen-bond networks. Constitutive heterodimer formation (blue circles) or off-target oligomer formation (red circles) were determined with native MS. Filter cutoff values of dSASA_polar > 970 Å2 and more than one polar bulky residue buried in the core includes most of the successful designs and excludes most of the design failures. i, On the basis of the Y2H data in b, all 32 monomers from the 16 pairs were categorized as being specific (blue, has ≤1 non-cognate binding), or non-specific (red, has >1 non-cognate binding). With application of secondary structure prediction scores (PsiPred54) and Rosetta centroid energy score per residue as filters, designs with higher PsiPred values and lower Rosetta centroid score per residue are more specific (green box). Two independent experiments were performed (ac).

Extended Data Fig. 9 Hydrogen-bond network sequence motifs of the set of six orthogonal pairs in Y2H experiments.

Green patches mark the locations of hydrogen-bond network-forming residues on the backbones. Letters along the backbones indicate residue identities.

Extended Data Fig. 10 The workflow of native MS mixing experiments.

a, Protein samples were characterized using online desalting coupled to native MS and deconvoluted using UniDec software. Proteins showing expected masses were mixed in equimolar ratio, and the final mix was divided into two parts: in the experimental group (DN), proteins were denatured by 5 M GdnHCl at 75 °C and refoled into 150 mM AmAc; in the control mixing experiment (N), denaturation and refolding steps were omitted. Sample mixtures in each group were further equally divided into three parts that were individually injected on LC–MS with cation exchange and anion exchange, respectively, coupled with CID or SID. LC–MS analysis was performed for mixtures in full MS mode and MSMS mode with HCD and SID, respectively. Data were deconvoluted using Intact Mass. The deconvoluted mass lists from Intact Mass were searched against a theoretical mass list of all possible monomer, dimer, trimer and tetramer combinations. Dimers were identified using the full MS runs and MSMS runs with both subunits being detected at the same retention time. b, In the control mixing experiment (N), after mixing all 16 proteins in solution without the denaturation and renaturation steps, no exchange among proteins were observed. c, CD data for a mixture of purified DHDs in PBS (red) or 5 M GdnHCl and 75 °C (blue). Protein mixture was fully denatured under the latter conditions. d, A mixing experiment of DHD_37_ABXB and 15N-labelled DHD_37_ABXB with (red) or without (black) the denaturation and refolding steps. MS peaks merged after subunit exchange owing to the similarity in the masses of 15N-labelled and unlabelled subunits. Two biologically independent experiments were performed (bd).

Supplementary information

Supplementary Information

This file contains Supplementary Text, Supplementary Table legends 1-17 and Supplementary References.

Reporting Summary

Supplementary Tables

This zipped file contains Supplementary Tables 1-17.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark
Fig. 1: Modular heterodimer design.
Fig. 2: Structural characterization of designed heterodimers.
Fig. 3: New functionality from DHD combinations.
Fig. 4: All-against-all orthogonality assessment.
Extended Data Fig. 1: Overview of different topologies designed.
Extended Data Fig. 2: Example HBNets resulting from the systematic search.
Extended Data Fig. 3: Thermal and chemical denaturation of DHDs.
Extended Data Fig. 4: Backbone and hydrogen-bond network permutations.
Extended Data Fig. 5: Biophysical characterization of hydrogen-bond-network-permuted homodimers.
Extended Data Fig. 6: SAXS profiles of all tested DHDs.
Extended Data Fig. 7: Crystal structure of the domain-swapped DHD_15 and biophysical characterization of higher-order oligomers.
Extended Data Fig. 8: Y2H all-against-all assay of 16 DHDs.
Extended Data Fig. 9: Hydrogen-bond network sequence motifs of the set of six orthogonal pairs in Y2H experiments.
Extended Data Fig. 10: The workflow of native MS mixing experiments.


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.