PCS-based structure determination of protein–protein complexes
- First Online:
- Cite this article as:
- Saio, T., Yokochi, M., Kumeta, H. et al. J Biomol NMR (2010) 46: 271. doi:10.1007/s10858-010-9401-4
- 818 Downloads
A simple and fast nuclear magnetic resonance method for docking proteins using pseudo-contact shift (PCS) and 1HN/15N chemical shift perturbation is presented. PCS is induced by a paramagnetic lanthanide ion that is attached to a target protein using a lanthanide binding peptide tag anchored at two points. PCS provides long-range (~40 Å) distance and angular restraints between the lanthanide ion and the observed nuclei, while the 1HN/15N chemical shift perturbation data provide loose contact-surface information. The usefulness of this method was demonstrated through the structure determination of the p62 PB1-PB1 complex, which forms a front-to-back 20 kDa homo-oligomer. As p62 PB1 does not intrinsically bind metal ions, the lanthanide binding peptide tag was attached to one subunit of the dimer at two anchoring points. Each monomer was treated as a rigid body and was docked based on the backbone PCS and backbone chemical shift perturbation data. Unlike NOE-based structural determination, this method only requires resonance assignments of the backbone 1HN/15N signals and the PCS data obtained from several sets of two-dimensional 15N-heteronuclear single quantum coherence spectra, thus facilitating rapid structure determination of the protein–protein complex.
KeywordsLanthanide-binding peptide tag Two-point anchoring Paramagnetic NMR Pseudo-contact shift Rigid-body docking Autophagy
The structure determination of a protein–protein complex is an important step in revealing the interaction mechanism; however, the application of X-ray crystallography or NMR spectroscopy to this end is not straightforward. In crystallography, complexes are often difficult to crystallize and the possibility of crystal artifacts must always be taken into consideration. In NMR spectroscopy, the protein structure is generally determined on the basis of the short distance restraints derived from nuclear Overhauser effects (NOEs), and it is often difficult to collect a sufficient number of distance restraints for precise structure determination.
Paramagnetic lanthanide ions induce several effects in observed nuclei, such as a pseudo-contact shift (PCS) and residual dipolar coupling (RDC) due to the anisotropy of the magnetic susceptibility tensor (Δχ-tensor). PCS provides long-range distance and angular information between the lanthanide ion and the observed nuclei situated up to ~40 Å apart from the lanthanide ion (Allegrozzi et al.2000). Accordingly, the paramagnetic lanthanide ion can be used as a powerful probe for solution structure determination, especially for larger molecular weight proteins, multidomain proteins, and protein complexes. For metalloproteins, metal ions such as Ca2+ and Mg2+ can be replaced by the paramagnetic lanthanide ions, and paramagnetic lanthanide probes have been successfully applied to metalloproteins (Bertini et al.2001, 2004, 2007; Barbieri et al.2002; Pintacuda et al.2006, 2007; Allegrozzi et al.2000).
However, the application of paramagnetic lanthanide probes to non-metalloproteins requires a general method that attaches the lanthanide ions to the protein at a fixed position. Recently, the lanthanide binding peptide tag (LBT), which can be attached to the target protein by two anchoring points, a disulfide bridge and an N-terminal fusion, has been reported (Saio et al.2009a). Several other lanthanide-binding tags have been also reported, including lanthanide-chelating reagents attached via disulfide bonds (Dvoretsky et al.2002; Haberz et al.2006; Pintacuda et al.2004; Prudêncio et al. 2004; Ikegami et al.2004; Leonov et al.2005; Gaponenko et al.2002, 2004; Vlasie et al.2007; Keizers et al. 2007, 2008; Su et al.2008b), and lanthanide-binding peptides attached through N- or C-terminal fusion (Gaponenko et al.2000; Wöhnert et al.2003; Martin et al.2007; Ma and Opella 2000; Zhuang et al.2008) or a disulfide bond (Su et al.2006, 2008a). However, single-anchored tags tend to be mobile, while symmetrically designed, double-anchored synthetic chelators often suffer from peak doubling caused by enantiomeric conformers. Furthermore, most of the synthetic tags reported to date are not commercially available. Among these, the lanthanide-attaching method using a two-point anchored peptide tag has a number of advantages in terms of chiral purity, rigidity, and ready availability for protein NMR researchers (Saio et al.2009a).
We applied this lanthanide tagging method, which introduces the lanthanide ion using two-point anchored peptide tag, to the structure determination of the protein–protein complex of p62 PB1. P62 is a multi-module adaptor protein that plays an important role in autophagy and the NF-κB signaling pathway. In autophagy, p62 interacts with ubiquitinated proteins via its UBA domain, and self-assembles through its PB1 domain to form large protein aggregates (Bjørkøy et al.2005). The aggregates are then transported to the autophagosome through interaction with LC3 (Noda et al.2008). The p62 PB1 domain forms a homo-oligomer in a front-to-back manner using its conserved interaction motifs, the OPCA motif and the conserved Lys motif (Saio et al.2009b). In order to avoid the homo-oligomerization of the p62 PB1-PB1 complex, we introduced site-directed mutations into the interaction motifs and prepared two mutants that only limited 1:1 dimer formation. A monomer structure of the p62 PB1 mutant that abrogates homo-oligomerization has been already solved by NMR spectroscopy (Saio et al.2009b), but the structure of p62 PB1-PB1 complex has not yet been solved. By attaching the lanthanide binding peptide tag to the one subunit of the dimer, we fixed the lanthanide ion on the protein and obtained inter-subunit structural information from the PCS. Here, we demonstrate a simple and fast method for the structure determination of protein–protein complexes in which the monomer structures are docked based on 1H/15N PCS and 1H/15N chemical shift perturbation data.
Wild-type p62 PB1 forms a homo-oligomer in front-to-back manner, thus making NMR analysis more difficult. On the basis of our previous study (Saio et al. 2009b), we therefore prepared two p62 PB1 mutants, hereafter referred to as DR and KE, that have mutations in the conserved interaction surfaces, the OPCA motif and the conserved Lys motif, respectively, and thus form a 1:1 dimer. For DR, site-directed mutations were introduced into the conserved acidic residues on the OPCA motif to form a D67A/D69R double mutation. To attach the lanthanide ion to DR, a lanthanide binding sequence comprised of 16 amino acids, CYVDTNNDGAYEGDEL (LBT) (Nitz et al. 2003, 2004; Su et al. 2006, 2008a), was attached to the N-terminus of DR, according to our previous report (Saio et al.2009a), to which is hereafter referred as LBT-DR. LBT-DR was subcloned, together with a GST tag and a tobacco etch virus (TEV) protease cleavage site, into a pGSTV vector derived from the pET-21 plasmid (Novagen, USA). As a binding partner for LBT-DR, we prepared the KE mutant in which the two basic residues on the conserved basic surface, Lys7 and Arg94, were mutated to Glu and Ala, respectively. The KE mutant was subcloned, with a GST tag and HRV3C protease cleavage site, into a pGSPS vector derived from the pET-21 plasmid (Novagen).
P62 PB1 has two cysteine residues, Cys 26 and Cys42. We changed Cys42 on LBT-DR, and Cys26 and Cys42 on KE to serine in order to guarantee proper S–S formation between LBT and the Cys26 on DR.
Proteins were expressed in E. coli strain BL21 (DE3) cells. For the unlabeled samples, cells were grown in Luria–Bertani media. For the uniformly 15N- or 13C/15N-labeled samples, cells were grown in M9 media containing 15NH4Cl (1 g/l), Celtone-N powder (0.2 g/l) (Cambridge Isotope Laboratories, USA) and unlabeled glucose (10 g/l), or 15NH4Cl (1 g/l), Celtone-CN powder (0.2 g/l) (Cambridge Isotope Laboratories, USA) and [U-13C] glucose (2 g/l), respectively. The uniformly 15N/2H-labeled sample was prepared by culturing cells in 100% 2H2O M9 medium using 15NH4Cl and [U-2H] glucose as the sole nitrogen and carbon sources. Cells were grown at 37°C to A600 of 0.8, and protein expression was induced by the application of Isopropyl β-D-1-thiogalactopyranoside to a final concentration of 0.5 mM for 16 h at 25°C. For the preparation of amino acid selectively 15N-labeled samples, the cells were grown at 37°C in 1 l of minimal media supplemented with 1 g 14NH4Cl and 200 mg of 19 unlabeled amino acids, respectively. Protein expression was induced at A600 of 0.8 by the addition of isopropyl β-D-1-thiogalactopyranoside to a final concentration of 0.5 mM and was cultured for 8 h at 25°C. Fifty mg of specific 15N-labeled amino acid was added to the medium 15 min before induction.
For the preparation of LBT-DR, the disrupted cells were centrifuged and the supernatant was applied to glutathione-Sepharose 4B resin (GE Healthcare, UK) for affinity purification. The GST tag was removed by incubation for 4 h at room temperature with TEV protease. The isolated protein was further purified by gel filtration chromatography on a Superdex 75 column (GE Healthcare). LBT-DR was also expressed in the inclusion body, and retrieved by high-pressure refolding (Schoner et al.2005; Qoronfleh et al. 2007). Details of the refolding process will be published elsewhere. KE was prepared from the soluble fraction, according to the procedure described previously (Saio et al.2009b).
After the gel filtration, LBT-DR was incubated with 1 mM 5, 5′-ditiobis(2-nitrobenzoic acid) (DTNB) for 2 h at room temperature, which linked the N-terminal Cys of LBT and the C26 on DR via an intramolecular disulfide bond (Saio et al.2009a). The oxidized LBT-DR was then mixed with KE, followed by gel filtration chromatography on a Superdex 75 column.
For NMR measurements, the samples were prepared in 20 mM MES buffer (pH 6.5) with 50 mM NaCl. All NMR experiments were run on Inova 800, 600 or 500 MHz NMR spectrometers (Varian, USA) at 25°C. Spectra were processed using the NMRPipe program (Delaglio et al.1995) and data analysis was performed with the help of the Olivia program developed in our laboratory (Yokochi et al. http://fermi.pharm.hokudai.ac.jp/olivia/). Intermolecular NOEs were obtained from a 3D 15N-edited NOESY experiment with a mixing time of 200 ms on 15N/2H-labeled LBT-DR complexed with unlabeled KE.
PCS-based rigid body docking was carried out using the Xplor-NIH program (Schwieters et al.2003, 2006), equipped with PARA restraints for Xplor-NIH (Banci et al.2004). The coordinates of LBT-DR (including the metal) were held fixed, whereas KE was treated as a rigid body. As the starting structure of DR, conformer 1 of the family of NMR structures of DR (PDB code: 2KKC) was used, with the exception that Cys42 was replaced with Ser using the PyMOL program (http://www.pymol.org/). The structure of the KE mutant was built based on the structure of DR, with the six surface residues of DR (Lys7, Cys 26, Cys 42, Ala67, Arg69, and Arg94) changed to Glu, Ser, Ser, Asp, Asp, and Ala, respectively, in accordance with the amino acid sequence of KE.
At the start of the docking calculation, the relative orientation and position of KE were randomized to generate 100 starting structures that were located within 100 Å from the DR mutant. The coordinates of DR and the metal, on the other hand, were fixed, with the position of the metal determined by tensor-fits from PCSs observed for LBT-DR. Next, the rigid body docking calculation was performed based on the PCS and contact-surface restraints. During the calculation, the coordinates of DR and the metal were fixed, whereas those of KE were freely rotated and translated. For the PCS restraints, a pseudo atom representing the tensor axis was introduced. The atom representing the origin of the axis was restrained within 0.3 Å of the metal, while the coordinates of the tensor were freely rotated around the origin. The target function was calculated based on three terms: a square-well quadratic term for ambiguous distance restraints (ENOE; Clore and Schwieters 2003), the least square energy penalty for PCS restrains (EPCS; Banci et al.2004), and a quartic van der Waals repulsion term (Erepel). Ambiguous distance restraints were set with upper-limit of 5 Å. During the minimization process, the force constant for ENOE and EPCS were held constant at 0.01 kcal mol−1 Å−2 and 0.8 kcal mol−1 ppm−2, respectively. The force constraint for Erepel was geometrically increased over 14 cycles from 0.004 to 1 kcal mol−1 Å−4. The van der Waals radius scale factor was decreased from 1.0 to 0.78. The Xplor-NIH script for the docking calculation is provided as Supporting information.
Results and discussion
Construct design for LBT attachment
Resonance assignment and PCS measurement
For the backbone amide resonance assignment of the LBT-DR/KE complex, a standard set of triple resonance NMR spectra was measured using the 13C15N-labeled LBT-DR/unlabelled KE and unlabelled LBT-DR/13C15N-labeled KE complexes, both of which contain 1 eq diamagnetic lanthanide Lu3+. Resonance assignment was accomplished with reference to those of free-state DR (Saio et al.2009b). The assignment rate of the backbone amide signals of LBT-DR and KE was 91 and 98%, respectively.
Determination of the metal position
Δχ-tensor parameters for lanthanide ions in complex with LBT-DR/KE, determined on the basis of the monomer structure of DR and the PCS values obtained from LBT-DR signals
40.8 ± 1.1
−27.2 ± 1.3
28.6 ± 1.5
−10.4 ± 0.3
20.7 ± 0.9
−18.9 ± 1.0
21.7 ± 1.0
−9.1 ± 0.2
The contact-surface restraints
The DR and KE mutants were docked based on the PCS and contact-surface restraints. The contact surface restraints were generated from the 1HN/15N backbone chemical shift differences between the free and bound states of KE (Fig. 3d). The chemical shift of backbone 1HN/15N is sensitive to the chemical environment of the two nuclei, which is very useful for the identification of the interaction surface on proteins. Unlike NOE-based analysis, the chemical shift perturbation of the backbone signals can easily be obtained without time-consuming side-chain assignment. Combined with the backbone PCS restraints, backbone chemical shift perturbation mapping ensures fast and reliable structure determination of protein–protein complexes. However, chemical shift perturbations can result either from a direct ligand interaction or from a conformational rearrangement around the observed nuclei, and it is possible that the signals of the residue on the opposite side of the interaction surface induce sizable perturbations, as a consequence of a change in the local structure. Thus, we selected interfacial residues according to the three criteria proposed by Clore and Schwieters (2003): (A) significant chemical shift perturbation is observed upon complex formation, (B) at least one or two atoms of the residue are exposed on the surface of the protein, and (C) the selected residue is involved in a cluster of residues on a contiguous, single binding surface. On binding with LBT-DR, several 1H-15N HSQC signals for KE indicated significant chemical shift perturbations (Fig. 3d, e). Asp67, Glu68, Asp69, Asp71, Val73, Phe75, Ser76, Ser77, and Asp90 all indicated large chemical shift perturbations. The eight residues other than Asp90 fulfilled the above-mentioned criteria, whereas Asp90, which indicated a sizable chemical shift difference and whose atoms are exposed on the surface of the protein, failed to comply with criterion C in that Asp90 is located on the opposite side of the continuous cluster comprised of the other eight residues. Thus, we concluded that Asp67, Glu68, Asp69, Asp71, Val73, Phe75, Ser76, and Ser77 are involved in the binding surface, and we converted the chemical shift perturbation information into the contact-surface restraints. It should be noted that the chemical shift perturbation mapping of LBT-DR, on binding with KE, indicated well-defined contact-surface area (Supporting information Figure S4).
The docking calculation was carried out using the Xplor-NIH program (Schwieters et al.2003, 2006) with a rigid body minimization protocol (Clore 2000; Tang and Clore 2006). For the calculation, a total of 459 backbone 1H and 15N PCS restraints derived from Tb3+ and Tm3+ as well as contact-surface restraints based on the chemical shift perturbation were used. The contact-surface restraints were added only to restrict the binding surface of KE. Details of the calculation are described in the “Method” section. A total of 100 structures were calculated, each of which started from the randomly arranged KE coordinates around DR. During the calculation, DR and the metal were held in a fixed position, while KE was freely rotated and translated as a rigid body. In the present calculation, we excluded Asp90 from the contact-surface restraints as it failed to satisfy all inclusion criteria (see above). However, test calculations showed that the inclusion of Asp90 in the contact-surface restraints had no effects on the results (data not shown). This may be due to the ambiguity of the distance restraints defined using the r−6 averaging option.
The PCS-isosurface observed with a paramagnetic lanthanide ion is symmetric, thus a PCS data set derived from one lanthanide ion causes four degenerate solutions obtained by rotation around the x, y, and z axes of the principal axis of the Δχ-tensor. In principle, the degeneracy can be overcome by adding a second PCS data set from another lanthanide ion, as the direction of the principal axis of the Δχ-tensor of a second lanthanide ion would be different from that of the first (Pintacuda et al.2006). In our test calculations using only PCS restraints, the combined use of multiple PCS data sets couldn’t overcome the degeneracy (data not shown), presumably due to the minute difference in the orientation of the principle axes of the Δχ-tensors (Table 1). However, only one of the four degenerate solutions satisfied the contact surface restraints, thus a combination of the contact surface and PCS restraints allows the identification of a proper solution from among the four degenerate solutions.
Validation of the calculated structure
Representative examples of intermolecular NOEs between LBT-DR and KE
Proton group in LBT-DR
Proton group on KE
Δχ-tensor parameters for lanthanide ions in complex with LBT-DR/KE, determined on the basis of the docking structure of DR/KE and the PCS values obtained both from LBT-DR and KE signals
36.4 ± 1.5
−23.4 ± 1.3
21.9 ± 0.8
−10.0 ± 0.4
23.8 ± 0.7
−20.1 ± 0.5
20.4 ± 0.7
−9.9 ± 0.3
A paramagnetic lanthanide ion provides valuable information for NMR protein structural analysis as PCS contains both long-range distance and angular information, which cannot be replaced by other probes, such as spin labels, NOEs, or paramagnetic metal ions (e.g., Cu2+, Mn2+, or Gd3+), that only yield distance dependent information. Bertini et al. (2009) demonstrated accurate solution structure determinations of multi-domain metalloproteins utilizing paramagnetic lanthanide probes. Despite the advantages associated with the use of lanthanide probes, the application of this approach has been limited to certain metal-binding proteins. To apply this method to non metal-binding proteins, a wide variety of lanthanide ion anchoring tags have been developed, including lanthanide binding peptide tags and synthetic chelating reagents (Su and Otting 2009). However, lanthanide tagging has not yet been applied to protein structural analysis apart from a limited number of studies (Gaponenko et al.2002, 2004; Zhuang et al.2008; Xu et al.2009). Recently, Feng et al. (2007) reported the structural analysis on the homo-oligomeric domain, Par-3 NTD, and utilized the lanthanide tagging method to obtain structural information for the complex. However, the high mobility of the lanthanide ion prevented the quantitative analysis of PCS. In general, the flexibility of the lanthanide binding tag prevents the wider application of lanthanide probes.
Recently, we reported a two-point anchoring method for a lanthanide binding peptide tag that fulfills both the need for ready availability for protein NMR researchers and higher rigidity (Saio et al. 2009a). Using this lanthanide tagging method, we here determined the protein–protein complex structure of the p62 PB1 homo-dimer, based on the distance and angular restraints from backbone 1H/15N PCSs and the contact-surface restraints derived from backbone chemical shift perturbations. These two kinds of restraints can easily be obtained by the measurement of 1H-15N HSQC spectra, as long as the backbone assignment of the target is available. On the other hand, NOE-based methods require experiments, far less sensitive than 1H-15N HSQC spectra, for side chain assignment and NOE collection, which is time-consuming and often difficult, especially for larger molecular weight targets. Using the two-point anchored peptide tag, which is readily available and holds a lanthanide ion in a fixed position, the application of the paramagnetic lanthanide probe will become more widely used.
This article is distributed under the terms of the Creative Commons Attribution Noncommercial License which permits any noncommercial use, distribution, and reproduction in any medium, provided the original author(s) and source are credited.