Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2018 Mar 1;555(7694):61-66.
doi: 10.1038/nature25762. Epub 2018 Feb 21.

Extreme disorder in an ultrahigh-affinity protein complex

Alessandro Borgia  1 , Madeleine B Borgia  1 , Katrine Bugge  2 , Vera M Kissling  1 , Pétur O Heidarsson  1 , Catarina B Fernandes  2 , Andrea Sottini  1 , Andrea Soranno  1   3 , Karin J Buholzer  1 , Daniel Nettels  1 , Birthe B Kragelund  2 , Robert B Best  4 , Benjamin Schuler  1   5
Affiliations

Extreme disorder in an ultrahigh-affinity protein complex

Alessandro Borgia et al. Nature. .

Abstract

Molecular communication in biology is mediated by protein interactions. According to the current paradigm, the specificity and affinity required for these interactions are encoded in the precise complementarity of binding interfaces. Even proteins that are disordered under physiological conditions or that contain large unstructured regions commonly interact with well-structured binding sites on other biomolecules. Here we demonstrate the existence of an unexpected interaction mechanism: the two intrinsically disordered human proteins histone H1 and its nuclear chaperone prothymosin-α associate in a complex with picomolar affinity, but fully retain their structural disorder, long-range flexibility and highly dynamic character. On the basis of closely integrated experiments and molecular simulations, we show that the interaction can be explained by the large opposite net charge of the two proteins, without requiring defined binding sites or interactions between specific individual residues. Proteome-wide sequence analysis suggests that this interaction mechanism may be abundant in eukaryotes.

PubMed Disclaimer

Figures

Extended Data Figure 1.
Extended Data Figure 1.. Titrations of ProTα and Globular Domain (GD).
(a) Titration of 15N-ProTα with 0- to 7-fold molar addition of GD followed by 1H,15N-HSQC spectra. (b) Peak intensity ratios for assigned residues of ProTα relative to the free state induced by 0- to 1.7-fold molar addition of GD. (c) CSPs per residue of ProTα induced by 0- to 7-fold molar addition of GD. For comparison, CSPs of ProTα upon 1-fold molar addition of H1 are shown in grey. Panels a-c follow color key 1; light grey stars indicate prolines and unassigned residues. (d) ProTα CSPs plotted against concentration and times excess of GD relative to the free state for residues 46-106 upon 0- to 7-fold molar addition of GD. Colors used for discriminability. (e) Far-UV CD spectrum of GD. (f) Thermal denaturation of GD followed by the change in ellipticity at 222 nm (Tm = 320.5 ± 0.3 K, ΔHm = −44 ± 2 kcal mol−1). Insert: Fraction unfolded GD (fu) as a function of temperature. (g) Titration of 100 μM 13C,15N-GD with 0- to 7-fold molar addition of ProTα followed by 1H,15N-HSQC spectra (color key 2). Peak intensities gradually decrease during the titration. At 3.5×- and 7× excess ProTα, natural abundance peaks of free ProTα appear (1H,15N-HSQC spectrum of 15NProTα shown in grey for comparison). (h) CSPs of GD plotted against concentration and times excess of ProTα relative to the free state upon 0- to 7-fold molar addition of ProTα. A total of 66 (unassigned) amide backbone peaks were followed and grouped according to the standard deviation (STD) of the CSPs (1 STD = 0.0254 ppm). Of these, 55% had CSPs larger than 1 STD. Colors used for discriminability.
Extended Data Figure 2.
Extended Data Figure 2.. Titration of 15N-ProTα with H1.
(a) 1H,15N-HSQC spectrum of 11 μM free 15NProTα with residue labels (left) and titrated with 0- to 4-fold molar addition of H1 (right) (see color key). (b) Weighted backbone amide chemical shift perturbations (CSPs) of ProTα (residues 46-106) relative to the free state upon 0- to 4-fold molar addition of H1, plotted against concentration and times excess of H1. Colors used for discriminability. (c) CSPs and (d) peak intensity ratios for assigned residues of ProTα induced by 0- to 4-fold molar addition of H1 (for bar colors, see key). (e) Longitudinal 15N relaxation times (T1) of free (red) and H1-bound (purple) 15N-ProTα. ⟨T1⟩ is 610 ms (free) and 636 ms (complex). (f) Transverse 15N relaxation times (T2) of free (red) and H1-bound (purple) 15N-ProTα. ⟨T2⟩ is 302 ms (free) and 217 ms (complex). In c-f, light grey stars indicate prolines and unassigned residues, dark grey stars overlap and/or insufficient data quality.
Extended Data Figure 3.
Extended Data Figure 3.. Titration of 13C,15N-H1 with ProTα.
(a) 1H,15N-HSQC spectra of free 13C,15NGD (globular domain, dark green) and free 13C,15N-H1 (orange). The majority of the amide peaks of the GD overlap with the more dispersed peaks from full-length H1, indicating the similarity in structure of the GD in isolation and within H1. (b) Titration followed by 1H,15N-HSQC spectra of 13C,15N-H1 with 0- to 4-fold molar addition of ProTα. Data acquired on His6-tagged H1. (c) CSPs relative to free H1 of eleven traceable H1 amide backbone peaks from the intrinsically disordered region (based on overlay with 1H,15N-HSQC spectra of GD (a)) upon 0 to 4-fold molar addition of ProTα plotted against concentration and times excess. Colors used for discriminability. (d) CSPs plotted against peak intensity ratios relative to the free state of H1 of the eleven H1 amides at 1× excess of ProTα. Colors as in (c). (e) Overlay of Cα,Hα region from 1H,13C-HSQC spectra of free 13C,15N-H1 (blue) and 13C,15N-GD (green). The H1 1H,13C-HSQC is dominated by intense clusters of peaks not present in the GD spectrum, consistent with the large fraction of repeats in the H1 disordered regions. (f) Cα,Hα region of 13C,15N-H1 upon titration with ProTα. The lack of detectable changes in Cα,Hα resonances is consistent with the absence of secondary structure induction in the disordered regions of H1 upon binding.
Extended Data Figure 4.
Extended Data Figure 4.. Hydrodynamic radii and stoichiometry of the H1-ProTα complex.
(a) Hydrodynamic radii, RH, of free and bound 15N-ProTα (100 μM) determined with pulsed-field gradient NMR at 283 K. The signal decays of free 15N-ProTα (red), with H1 at a 1:1 molar ratio (purple), and with H1 GD at a 1:7 molar ratio (green) as a function of gradient strength, together with corresponding fits and a table of the diffusion coefficients and resulting RH values. (b) RH measured by 2f-FCS at 295 K. Lines show the RH of H1 -1C (blue) and ProTα D2C (red) labeled with Alexa 594 in the absence of binding partner. Symbols show labeled ProTα (5 nM) in the presence of equimolar concentrations of unlabeled ProTα and unlabeled H1, with s.d.s indicated by error bars or shaded bands. (c) Stoichiometry ratio versus transfer efficiency plots from intermolecular single-molecule FRET experiments with singly labeled protein variants as indicated in the panels. A stoichiometry ratio of 0.5 indicates a 1:1 complex. (d,e) Transfer efficiency changes at large excess of unlabeled binding partner for FRET-labeled ProTα C56C110 (d) and H1 C104C194 (e).
Extended Data Figure 5.
Extended Data Figure 5.. Fluorescence lifetime analysis.
Plots of the fluorescence lifetimes of donor (Alexa 488),τDD, and acceptor (Alexa 594),τDA, normalized by the intrinsic donor lifetime, τD0, versus the ratiometric transfer efficiency, E (calculated from the number of donor and acceptor photon counts), as a diagnostic for the presence of a broad distance distribution rapidly sampled during the time of a fluorescence burst , , . If fluctuations in transfer efficiency occur on a timescale between the donor fluorescence lifetime (~4 ns) and the burst duration (~1 ms), the normalized donor lifetimes cluster above, and the acceptor lifetimes below the solid diagonal line expected for a single fixed distance, as previously observed for intrinsically disordered proteins , . The large deviation from the diagonal observed for both unbound and bound ProTα and H1 supports the presence of broad, rapidly sampled distance distributions.
Extended Data Figure 6.
Extended Data Figure 6.. Simulation results.
(a) Decision graph using the Rodriguez-Laio clustering algorithm70, showing only a single density maximum distant from other density maxima, i.e. a single distinct cluster. (b) Free energy for association of ProTα and H1 from simulation, yielding a KD of 7 fM at RPH = 0 (black curve). Blue and red curves are the free energies for addition of a second H1 or a second ProTα, respectively, to an existing H1-ProTα complex. (c) Principal component (PC) vectors shown as contact maps. Colors indicate the increase or decrease in each pair distance for that PC, relative to the other distances. ProTα and H1 residue numbers are indicated in red and blue, respectively. Each PC describes a feature of the chain arrangement: PC1, e.g., captures the presence or absence of interactions between the ProTα N-terminus and H1. (d) Intramolecular (top row) and intermolecular (rows 2 to 4) distributions of distances corresponding to FRET labeling sites, for the ProTα-H1 complex (labels PX-HY refer to residues X and Y in ProTα and H1, respectively). Filled distributions: simulations without explicit chromophores; green lines: simulations with explicit chromophores.
Extended Data Figure 7.
Extended Data Figure 7.. Kinetics of H1-ProTα binding measured by stopped flow.
FRET-labeled ProTα 56-110 is mixed rapidly with unlabeled H1 in TBS buffer, and the resulting increase in acceptor fluorescence is monitored (inset, measured at 10 nM H1 with single-exponential fit and residuals above, see Methods for details). Decay rates were obtained from single-exponential fits, assuming an instrument dead time of 3 ms. Standard errors for each H1 concentration were obtained via bootstrapping. The observed rates, kobs, are shown as a function of H1 concentration (cH1); for H1 concentrations between 10 and 100 nM, where pseudo-first order conditions apply (ProTα concentration after mixing was 2 nM), they were fit with kobs =koncH1 +koff =koncH1 +konkD, using the independently determined KD of 2.1 pM (Extended Data Table 2). The fit yields a bimolecular association rate coefficient of kon = (3.1 ± 0.1)·109 M−1 s−1 and an apparent dissociation rate coefficient of koff = (6.5 ± 3.1)·10−3 s−1. The gray area represents the 95% confidence band.
Extended Data Figure 8.
Extended Data Figure 8.. Example of the quality of the H1 preparation.
Electrospray ionization mass spectrum of H1 T161C labeled with Alexa 488 (calculated mass 21,800 Da) and reversed-phase HPLC (Vydac C4) chromatogram (inset) showing absorption at 280 nm (red) and 488 nm (blue) and the elution gradient from solvent A (5% acetonitrile in H2O + 0.1% TFA) to solvent B (100% acetonitrile) (black), illustrating the high purity of the sample. The peak at ~5.5 min corresponds to free Alexa 488, the peak at ~16.8 min to H1 T161C labeled with Alexa 488.
Figure 1.
Figure 1.. ProTα and H1 remain unstructured upon binding.
Extended configurations of H1 (a) and ProTα (b), net charges, and surface electrostatic potentials with color scale (units in kBT/e). For the globular domain of H1, only residues with a solvent-accessible surface area (SASA) > 0.5 nm2 are included (cf. Extended Data Table 1). (c) Far-UV CD spectra of ProTα (red), H1 (blue), the ProTα-H1 mixture (purple), and their calculated sum (black) at 5 μM for each protein. (d) Cα secondary chemical shifts (SCScα) of ProTα free (red), in complex with H1 (purple), and their differences (black). (e) 1H,15N-HSQC spectra of 15N-ProTα in the absence (red) and presence (purple) of unlabeled H1 and (f) 15N-H1 in the absence (blue) and presence (purple) of unlabeled ProTα with zooms (①,②). (g) H1 spectra from (f) at lower contour level.
Figure 2.
Figure 2.. ProTα and H1 form an electrostatically driven high-affinity complex.
(a)Single-molecule transfer efficiency histograms of FRET-labeled ProTα (positions 56 and 110) without (top) and with increasing concentrations of unlabeled H1 as indicated in the panels, fitted with two peaks, unbound (red) and bound (purple). (b) Binding isotherms based on transfer efficiency histograms for full-length H1 (, KD=2.1−0.8+1.1pM), N- (, KD=173−28+29nM) and C-terminal (, KD=40−4+6pM) regions, andthe globular domain of HI (, KD=1.9−0.3+0.3μM) at 165 mM ionic strength (see Extended Data Table 1 for details). (c) KD of H1-ProTα complex as a function of ionic strength with fit (purple line) and 95% confidence interval (shaded). See Methods for details of data analysis.
Figure 3.
Figure 3.. Dynamics, interactions, and distances in the complex.
(a-d) Examples of nsFCS probing long-range dynamics based on intra- and intermolecular FRET (see Extended Data Table 2 for details). (e) Average number of contacts of each ProTα residue with H1 based on the simulations (Fig. 4b). (f) Ratios of NMR resonance intensities of ProTα in the presence (I) and absence (I0) of H1. (g) Weighted backbone amide chemical shift perturbations (CSPs) of ProTα induced by equimolar H1 binding (see Extended Data Fig. 2 for other stoichiometries). In (f-g), the grey horizontal lines represent the average of three unassigned but traceable Glu residues in the range 62-67 with error bars from their standard deviation (see Methods for details). (h) Ratios of longitudinal (T1) and transverse (T2) 15N relaxation times of ProTα in the free (red) and bound (purple) states (see Extended Data Fig. 2 for details). Light grey stars indicate prolines and unassigned residues, dark grey stars resonance overlap and/or insufficient data quality. The dashed box indicates the sequence range with the largest changes. (i) Transfer efficiency (E) histograms from intermolecular single-molecule FRET experiments between different positions in ProTα and H1, fitted with a single peak (purple, E values shown). The signal at E ≈ 0 originates from molecules without FRET acceptor.
Figure 4.
Figure 4.. Architecture of the complex from simulations.
(a) Comparison of experimental (filled squares) and simulated transfer efficiencies (empty symbols) in the H1-ProTα complex for the pairs of dye positions indicated below (triangles and circles: simulations with and without explicit chromophores, respectively). (b) Intra- and intermolecular average distance maps of H1 and ProTα from the simulations, separately and in the complex. The white dashed square indicates the globular domain (only surface-exposed residues shown, see Extended Data Table 1). (c) Examples of conformations of H1 (blue) and ProTα (red) in the complex; the N-termini are indicated by small spheres. The structures are projected onto the first three principal components (PC) of the distance map, with projections of the full ensemble shown as gray scatter plots (units of Å, see also Extended Data Fig. 6). Numbers indicate the positions of the structures in the PC projections.
See this image and copyright information in PMC

Comment in

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Wright PE & Dyson HJ Linking folding and binding. Curr. Opin. Struct. Biol 19, 31–38 (2009). - PMC - PubMed
    1. Habchi J, Tompa P, Longhi S & Uversky VN Introducing protein intrinsic disorder. Chem Rev 114, 6561–6588 (2014). - PubMed
    1. Tompa P & Fuxreiter M Fuzzy complexes: polymorphism and structural disorder in protein-protein interactions. Trends in biochemical sciences 33, 2–8 (2008). - PubMed
    1. Baker JM et al. CFTR regulatory region interacts with NBD1 predominantly via multiple transient helices. Nat. Struct. Mol. Biol 14, 738–745 (2007). - PMC - PubMed
    1. Mittag T et al. Dynamic equilibrium engagement of a polyvalent ligand with a single-site receptor. Proc. Natl. Acad. Sci. U. S. A 105, 17772–17777 (2008). - PMC - PubMed

Publication types