Structural and dynamic characterization of the C-terminal tail of ErbB2: Disordered but not random

Abstract

ErbB2 (or HER2) is a receptor tyrosine kinase overexpressed in some breast cancers and associated with poor prognosis. Treatments targeting the receptor extracellular and kinase domains have greatly improved disease outcome in the last 20 years. In parallel, the structures of these domains have been described, enabling better mechanistic understanding of the receptor function and targeted inhibition. However, the ErbB2 disordered C-terminal cytoplasmic tail (CtErbB2) remains very poorly characterized in terms of structure, dynamics, and detailed functional mechanism. Yet, it is where signal transduction is triggered via phosphorylation of tyrosine residues and carried out via interaction with adaptor proteins. Here, we report the first description, to our knowledge, of the ErbB2 disordered tail at atomic resolution using NMR, complemented by small-angle x-ray scattering. We show that although no part of CtErbB2 has any fully populated secondary or tertiary structure, it contains several transient α-helices and numerous transient polyproline II helices, populated up to 20 and 40%, respectively, and low but significant compaction. The presence of some structural elements suggests, along the lines of the results obtained for EGFR (ErbB1), that they may have a functional role in ErbB2's autoregulation processes. In addition, the transient formation of polyproline II helices is compliant with previously suggested interactions with SH3 domains. All in all, our in-depth structural study opens perspectives in the mechanistic understanding of ErbB2.

Figure 1

(A) Schematic representation of the human ErbB2 receptor domains and sequence of the CtErbB2 construct. CtErbB2, C-terminal tail; ECD, extracellular domain; TKD, tyrosine kinase domain; TMH, transmembrane helix. Our CtErbB2 construct starts at valine 988 in the full-length receptor, with four additional N-terminal residues. Asterisks indicate tyrosines that are phosphorylated by ErbB kinases in dimers of full receptors (autophosphorylation). (B) CD spectrum (molar ellipticity per residue) of CtErbB2. (C) Experimental SAXS intensity (black dots, the error bars represent the standard deviations computed from propagated Poisson distribution) and fit with the Sharp-Bloomfield equation (red line). (Insets) Top: distance distribution. Bottom: Kratky plot. (D) ¹H-¹⁵N HSQC spectrum of CtErbB2 at 950 MHz. To see this figure in color, go online.

Figure 2

CtErbB2 secondary structure and long-range contacts identification. Green box, α-helix; blue box, PPII helix. Gray area: zones of tertiary contact. For a Figure360 author presentation of this figure, see https://doi.org/10.1016/j.bpj.2021.03.005. (A) Top: SSP scores (54), reproduced from (34). Vertical bars represent the position of prolines. Blue bars: trans prolines (with small (<10%) or no cis conformation detected); gray bars: prolines with a major population of cis isomer detected; black bars: prolines of unknown conformation. Middle: N-H residual dipolar couplings compared with Flexible Meccano (52) calculations. Black: experimental, with error bars corresponding to the variation between two independent experiments; orange: calculated from a random coil Flexible Meccano ensemble (10,000 structures); red: calculated from a random-coil Flexible Meccano ensemble with the transient secondary structures presented above. Bottom: ³J_HNHA. Black: experimental, with error bars calculated from the noise level of the spectrum in CCPNMR; orange: calculated for a random coil (55). (B) Upper panel: experimental (points, with error bars from the CCPNMR fit) and recalculated (lines) ¹⁵N relaxation parameters of CtErbB2 at three magnetic fields using the DLM model described in the Materials and methods. χ²-values were calculated for experimental versus recalculated data $\left({\chi }_{exp}^2\right)$ and for 95% lower χ² Monte Carlo simulations versus recalculated data $\left({\chi }_{0.95}^2\right)$. Lower panel: dynamics parameters of CtErbB2 determined from fitting ¹⁵N R₁, ¹⁵N R₂, and heteronuclear nOes data at three magnetic fields to the DLM polymer model (see Materials and methods). Missing data correspond to prolines and weak, unassigned, or overlapping peaks. (C) PRE measurements for mutants C1133S, C1032S, (C1032S, C1133S, S1214C), and (C1032S, C1133S, S1235C) of CtErbB2. The paramagnetic probe, indicated by an asterisk, was attached to cysteine 1032, 1133, 1214, and 1235, respectively. The red line is the PRE values predicted for the ensemble generated by Flexible Meccano taking into account secondary structures. To see this figure in color, go online.

Figure 3

Model of CtErbB2 structural features with some highlighted functional sites. Along the sequence: the transient secondary structures revealed by SSP scores, RDCs, and scalar couplings are indicated, as well as the main fragments exhibiting transient long-range contacts observed in the PRE experiments. In addition to the main known interaction sites and corresponding partners (Shc and Grb2 (25,26), PTP-2c (26), and MEMO (27) for interaction with phosphotyrosines, FynSH3 (87) for interaction with a RxPxxP motif), the PxxP motifs located in PPII helices are indicated as the best-defined putative interaction sites of the proline-rich regions. The putative interaction of the N-terminal helix with the kinase domain is also indicated. Samples of a conformational ensemble: superimposition of 100 structures randomly chosen from the ensemble of 10,000 structures generated by Flexible Meccano with transient secondary structures as input. The structures were aligned on the first 20 N-terminal residues of the construct. To see this figure in color, go online.