EPO does not promote interaction between the erythropoietin and beta-common receptors

A direct interaction between the erythropoietin (EPOR) and the beta-common (βc) receptors to form an Innate Repair Receptor (IRR) is controversial. On one hand, studies have shown a functional link between EPOR and βc receptor in tissue protection while others have shown no involvement of the βc receptor in tissue repair. To date there is no biophysical evidence to confirm a direct association of the two receptors either in vitro or in vivo. We investigated the existence of an interaction between the extracellular regions of EPOR and the βc receptor in silico and in vitro (either in the presence or absence of EPO or EPO-derived peptide ARA290). Although a possible interaction between EPOR and βc was suggested by our computational and genomic studies, our in vitro biophysical analysis demonstrates that the extracellular regions of the two receptors do not specifically associate. We also explored the involvement of the βc receptor gene (Csf2rb) under anaemic stress conditions and found no requirement for the βc receptor in mice. In light of these studies, we conclude that the extracellular regions of the EPOR and the βc receptor do not directly interact and that the IRR is not involved in anaemic stress.


Purification of the extracellular region of EPOR
Bacmids and baculoviruses were generated using the Invitrogen Bac-to-Bac™ Baculovirus expression system. EPOR was expressed in High Five cell cultures (2.5x10 6 cells/mL) grown in Sf900 II media and incubated at 28°C, for 49 hr. The cell cultures were then centrifuged and the supernatant with the secreted protein filtered, buffer exchanged (10 mM HEPES, 500 mM NaCl, 15% glycerol pH 7.2) and concentrated using the Sartorius Stedim Sartojet tangential flow through system. The supernatant was then loaded onto a His Trap FF column (GE Healthcare) and eluted over a 30 min gradient with 10 mM HEPES, 500 mM NaCl, 15% glycerol, 500 mM imidazole, pH 7.2. Fractions were further purified using a Superdex 200 16/60 pg column (GE Healthcare) ( dataset.

Purification of the extracellular region of the βc receptor
Bacmids and baculoviruses were generated using the Invitrogen Bac-to-Bac™ Baculovirus expression system. Expression was carried out in attached Sf21 insect cells, with 10x10 6 cells in Sf900II media allowed to attach and grow in T-flasks for 72 hr at 27°C. The cells were then infected with the generated baculovirus and incubated with the virus for one hr, after which the media and virus were drained and replaced with fresh media before another 72 hr incubation for expression. The media with the secreted protein was then harvested from each T-flask and filtered, buffer exchanged (20 mM phosphate, 500 mM NaCl, 20 mM imidazole pH 7) and concentrated using the Sartorius Stedim Sartojet tangential flow through system. The supernatant was then loaded onto a His Trap HP column (GE Healthcare) and eluted (20 mM phosphate, 500 mM NaCl, 500 mM imidazole, pH 7).
The fractions were further purified using a Superdex 200 16/60 pg column (GE Healthcare). The purified protein was validated using the methods detailed above for EPOR, except for the CD study set 4 was used as reference dataset with the Dichroweb analysis.

In silico docking of EPOR to the βc receptor in the absence of EPO
The membrane proximal domain 2 (D2) of EPOR ( Fig. 1)   The ligands (i.e. EPOR) of the top 30 cluster centres were geometrically optimised using the CHARMM force field in the presence of the receptor (i.e. βc), to produce the final protein complex models. A biased dock of βc (D1/D4 only) and EPOR was also carried out, where the three crude EPOR:βc (D1/D4 only) models were used as input for RosettaDock. In the biased docking process, the initial guess of the EPOR:βc protein complex underwent a rigid body docking search followed by flexible side-chain optimisation. One thousand independent simulations were then performed and the resulting protein complex models were ranked based on interaction energy. For each of the three biased docking calculations, the top ten ranked models were retained and visually inspected.
D2 of EPOR and D4 of the βc receptor are the membrane proximal domains of each of these receptors ( Fig. 1) and they are connected to their respective transmembrane regions by a short juxtamembrane region. During the analysis of the docking results, the assumption was that D2 of EPOR must interact with D4 of βc to form Site 3 (i.e. formation of the IRR, Fig. 1) and any docking solutions not meeting this requirement were discarded. The interface(s) between the βc receptor and EPOR in the modelled IRR complexes were analysed using the Protein interfaces, surfaces and assemblies service (PISA)

In silico docking of the extracellular domains of EPOR and the bc receptor in the absence of EPO
In the absence of EPO, blind rigid body docking yielded only one solution (Model 1, Supplementary   Fig. S1a) in which the receptors were orientated with their membrane proximal domains (D2 of the EPOR and D4 of βc) in alignment. However, the arrangement of EPOR and βc in Model 1 is very different to that observed in the EPOR homodimer or the GM-CSFRα:βc portion of the GM-CSF ternary complex (Fig. 1) and there is little contact between the membrane proximal domains of the two receptors (referred to as Site 3, Supplementary Fig. S1). Thus, Model 1 is an unlikely representation of the extracellular domains of the pre-formed IRR heteroreceptor.
The interactions between residues at Site 3 for Models 2-4 (biased docking models) were analysed using PISA, followed by PredHS and KFC2. These analyses predicted residue E173 from EPOR, as well as N393 and S396 from the βc receptor, to be critical to the Site 3 interface (i.e. hotspot residues, Supplementary Tables S1 and S2). Of these, N393 and H396 from the βc receptor have been reported to mediate interactions between βc and the GM-CSFRα at Site 3 in the GM-CSF ternary complex (PDB ID: 4NKQ) 16 . In silico docking of the extracellular domains of EPOR and the βc receptor in the presence of

EPO
The models generated using RosettaDock in scenario one were analysed by PISA, followed by PredHS and KFC2. These analyses predicted residues E13 and E18 from EPO and S91, F93, V94, N116, H153, P203, S204 and F205 from EPOR to be hotspot residues involved at Site 1 (Models 5-9 Supplementary   Tables S3-S6). Some of these predicted EPOR hotspot residues have been reported to be key interacting residues in the EPO:EPOR homodimer structure; EPOR residues F93, V94, N116, H153 and S204 in Site 1 and F93, V94 and S204 in Site 2 2 . βc receptor residues R64 and V105 were also predicted to be hotspot residues at the EPO:βc interface (Site 2) in Models 5-9 (Supplementary Tables S3-S6).