Structural insights into the disruption of TNF-TNFR1 signalling by small molecules stabilising a distorted TNF

Abstract

Tumour necrosis factor (TNF) is a trimeric protein which signals through two membrane receptors, TNFR1 and TNFR2. Previously, we identified small molecules that inhibit human TNF by stabilising a distorted trimer and reduce the number of receptors bound to TNF from three to two. Here we present a biochemical and structural characterisation of the small molecule-stabilised TNF-TNFR1 complex, providing insights into how a distorted TNF trimer can alter signalling function. We demonstrate that the inhibitors reduce the binding affinity of TNF to the third TNFR1 molecule. In support of this, we show by X-ray crystallography that the inhibitor-bound, distorted, TNF trimer forms a complex with a dimer of TNFR1 molecules. This observation, along with data from a solution-based network assembly assay, leads us to suggest a model for TNF signalling based on TNF-TNFR1 clusters, which are disrupted by small molecule inhibitors.

Fig. 1. Effect of compounds on TNF–TNFR1 interaction.

a IMS-MS of hTNF with 5-fold excess hTNFR1 (left panel), hTNF plus UCB-0595 (10-fold excess) and 5-fold excess hTNFR1 (right panel). Circled signals and corresponding peaks show three ionisation states for each receptor-bound form. b Quantitative analysis of IMS-MS data generated using hTNF and hTNF plus UCB-0595 (10-fold excess) (left and right panels, respectively) over a range of hTNFR1 concentrations (x axis). Traces (calc—calculated) indicating the percentage of each species; 0 receptor, 1 receptor, 2 receptor and 3 receptor-bound are shown (yellow, red, green and blue traces, respectively). Symbols (obs—observed) represent experimentally measured molar fractions of the different species in equilibrium (n = 1 independent experiments, a similar reduction in affinity of the third receptor binding was observed for multiple compounds—Supplementary Table 1). Source data are provided as a Source Data file. c SPR of hTNF (left panel) and hTNF plus UCB-0595 (10-fold excess) (right panel) binding to immobilised hTNFR1 (first injection) followed by injection of hTNFR1 at a range of concentrations. Detail of the TNFR1 binding response (third receptor binding) is highlighted (dashed box) (n = 1 independent experiments).

Fig. 2. Comparison of mouse TNF-humanTNFR1 complex with and without compound.

Top view (a) and side view (b) of trimeric mTNF (green ribbons, monomers assigned A, B and C) with three copies of monomeric receptor-bound (pink surface rendered) (PDB: 7KP7 [https://www.rcsb.org/structure/unreleased/7KP7]). Top view (c) and side view (d) of trimeric mTNF (green ribbons) with UCB-4433 bound (orange space-fill) and two copies of dimeric hTNFR1 bound (pink and purple surface rendered) (PDB: 7KP8 [https://www.rcsb.org/structure/unreleased/7KP8]). e Side view of mTNF (green surface rendered) without compound—one copy of the hTNFR1 has been removed to reveal the non-distorted a–c receptor binding site (selected residues involved in hTNFR1 binding highlighted in red). f Side view of mTNF (green surface rendered) with UCB-4433 bound (not visible) revealing the distorted A–C receptor binding site (selected residues involved in receptor binding highlighted in red).

Fig. 3. Extended TNF–TNFR1 chain.

Four copies of trimeric mTNF (green ribbons) linked by hTNFR1 dimers (pink surface rendered) as present in the crystal lattice (PDB: 7KP8 [https://www.rcsb.org/structure/unreleased/7KP8]). Membrane proximal ends of each receptor are highlighted in red.

Fig. 4. In vitro TNF–TNFR1 network assembly.

a Aggregation of hTNF (3 mg/mL) +hTNFR1(41–201) (orange trace) and hTNF (3 mg/mL) +hTNFR1(41–184) (blue trace) over time (ratio of TNF:TNFR = 1:2.16). b Aggregation of hTNF +hTNFR1(41–201) over a range of concentrations and at two ratios of hTNF:hTNFR1 (1:1.2 blue bars and 1:2.16 red bars). c Aggregation state of hTNF (2 mg/mL) +hTNFR1 at equilibrium over a range of ratios. d Aggregation of hTNF (2 mg/mL) +hTNFR1 at five ratios (1:1.7 blue trace, 1:2 orange trace, 1:2.4 grey trace, 1:3 yellow trace, 1:3.5 green trace) over time. e Aggregation of hTNF (2 mg/mL) + hTNFR1 over a range of ratios with UCB-9260 (10-fold excess over hTNF trimer) (grey bars) and DMSO only (orange bars) (n = 1 independent experiments). f Aggregation of hTNF +hTNFR1 at a fixed ratio of TNF:TNFR = 1:3.2 over a range of total protein concentrations with UCB-9260 (5-fold excess over TNF) (blue trace) (n = 1 independent experiments, a similar rightward shift in the point of aggregation was observed for multiple compounds—Supplementary Fig. 3) and DMSO control (red trace) (n = 2 independent experiments). Source data are provided as a Source Data file.

Fig. 5. Inhibition of TNF signalling in cells.

Representative concentration–response curves showing the effect of UCB-9260 on signalling in an NFκβ reporter cell assay driven by 10 pM hTNF (blue trace), 100pM hTNF (green trace) and anti-TNFR1 antibody at 300 ng/mL (red trace). Source data are provided as a Source Data file.

Fig. 6. Model of TNF–TNFR1 signalling network.

a Schematic representation of TNFR1 dimers (pink and purple) pre-clustered, with a portion bound to TNF (green) (two copies of TNFR1 dimer bound per TNF trimer). Addition of more TNF (green) results in linking of the TNF-two-receptor-bound units into larger TNF-three-receptor-bound clusters finally assembling into a large network. b Same scenario as a in the presence of compound (yellow dots), resulting in incomplete network assembly.

Fig. 7. Model of TNF with long-form TNFR1 (41–201) dimers bound.

Side and bottom view of surface rendered trimeric TNF (green) with three copies of long-form TNFR1 (41–201) dimers bound (pink). The fourth cysteine-rich domain of each receptor is highlighted in red.