Synthesis and biochemical characterization of EGF receptor in a water-soluble membrane model system

Abstract

ErbB (Erythroblastic Leukemia Viral Oncogene Homolog) receptor tyrosine kinases are critical for tissue development and maintenance, and frequently become oncogenic when mutated or overexpressed. In vitro analysis of ErbB receptor kinases can be difficult because of their large size and poor water solubility. Here we report improved production and assembly of the correctly folded full-length EGF receptor (EGFR) into nanolipoprotein particles (NLPs). NLPs are ~10 nm in diameter discoidal cell membrane mimics composed of apolipoproteins surrounding a lipid bilayer. NLPs containing EGFR were synthesized via incubation of baculovirus-produced recombinant EGFR with apolipoprotein and phosphoplipids under conditions that favor self-assembly. The resulting EGFR-NLPs were the correct size, formed dimers and multimers, had intrinsic autophosphorylation activity, and retained the ability to interact with EGFR-targeted ligands and inhibitors consistent with previously-published in vitro binding affinities. We anticipate rapid adoption of EGFR-NLPs for structural studies of full-length receptors and drug screening, as well as for the in vitro characterization of ErbB heterodimers and disease-relevant mutants.

Fig 1. Baculovirus-expressing EGFR production.

A, Map of the EGFR baculovirus using the flashBACULTRA system with a FLAG epitope placed between the leader sequence peptide (Lrig1 SS) and the full-length mammalian EGFR sequence (EGFR). B, Sf9 insect cells were infected with a MOI of 10 of control baculovirus (C) or EGFR baculovirus (E) for 48 hours followed by collection of total cell lysate and Western Blotting to detect EGFR expression. A set of these cells was treated with 10 ng/ml EGF in order to detect ligand-induced receptor phosphorylation. C, Sf9 insect cells were seeded on glass coverslips and infected with an MOI of 10 of control (C) or EGFR (E) baculovirus for 48 hours. Cells were fixed in 4% paraformaldehyde followed by immunostaining with anti-EGFR antibody (in red) with (+P) or without (-P) permeabilization in 100% methanol. The brightfield image shows cell morphology. Cells incubated with secondary antibody in the absence of EGFR antibody (no primary) were used as a negative control. D, FLAG-EGFR is readily purified from Sf9 insect cells as shown by a representative image of anti-EGFR immunoblot of FLAG EGFR purification and E, a Coomassie-stained SDS-PAGE gel. Bovine serum albumin (BSA) was loaded as a protein of known concentration. Total insect cell lysate starting material (SM), purification washes 1, 3 (W1, W3), and elution (E) were loaded to show efficiency of the FLAG-purification.

Fig 2. Characterization of EGFR-NLP size and shape.

A, Following NLP self-assembly, the specified NLPs were separated by a Yarra 3u SEC-4000 column and typical traces are shown. Empty NLPs elute between 9.5–10.5 minutes (P2) with EGFR-NLPs eluting earlier at 5.5–9.5 minutes (P1). Numbers above indicate molecular mass (in kDa) of known protein standards. B, Quantification of area under the curve of the P1 and P2 SEC peaks in μV*sec from a typical SEC trace. C, Empty and EGFR-NLPs were analyzed by DLS to determine NLP diameter.

Fig 3. EGFR incorporation in self-assembled NLPs.

A, P1 and P2 SEC fractions were collected and spotted onto nitrocellulose membrane followed by immunoblotting with anti-ApoA1 and anti-EGFR antibodies showing that the majority of EGFR is incorporated into the higher molecular weight NLPs of the P1 fraction. B, Indicated NLPs were separated by denaturing SDS-PAGE and analyzed with anti-pEGFR, anti-EGFR, and anti-ApoA1 antibodies. Numbers indicate band intensity relative to EGFR-NLP normalized to ApoA1 signal. C, Purified Empty and EGFR-NLPs were separated by NativePAGE gels and immunostained with anti-pY20 and anti-EGFR antibodies showing EGFR expression in the higher molecular weight ranges above 480 kDa. D, Bar graph of determination of EGFR insertion into NLP. Total amount of EGFR in the NLP assembly mixture and purified EGFR-NLPs were determined by ELISA and EGFR insertion rate indicated. The values represent the mean ± standard error of the mean (SEM) for two technical replicates. E, EGFR-NLPs were extracted with PBS (control) or sodium carbonate (Na₂CO₃), centrifuged to separate the supernatant containing NLPs (S) and pellet containing insoluble free protein (P), separated by denaturing SDS-PAGE, and analyzed with anti-EGFR antibody showing that EGFR is not susceptible to carbonate extraction.

Fig 4. NLP-associated EGFR is capable of binding its ligand, EGF.

A, Schematic of the OctetRed assay used to determine EGFR ligand binding. Streptavidin biosensors were loaded with biotinylated EGF and EGFR-NLP association and dissociation with the sensor was quantitated via BLI. B, Association (0–2000 s) and dissociation (2000–4000 s) curves of EGFR-NLPs binding to biotinylated EGF on the sensors. EGFR-NLP concentrations were titrated by a 1.5-fold dilution to the concentrations displayed to the right of the graph. Curves are normalized by subtracting the 0 μM EGFR-NLP curve. C, Log plot of EGFR-NLP binding response (nm) to biotinylated EGF. Plot shows mean of two technical duplicates and are representative of results from three biological repeats.

Fig 5. NLP assemblies contain EGFR monomers, dimers, and multimers.

Empty and EGFR-NLPs were separated by SDS-PAGE with (+β-ME/heat) and without denaturing conditions (-β-ME/heat) to observe EGFR multimers and analyzed by immunoblotting with anti-pY4G10, anti-EGFR, and anti-ApoA1 antibodies. Similar results were observed in each of three biological replicates. Numbers indicate band intensity relative to EGFR-NLPs normalized to ApoA1 signal.

Fig 6. NLP-associated EGFR interacts with EGFR-targeted therapeutic agents.

A, NLP incorporated EGFR was dephosphorylated via phosphatase treatment (CIP) then allowed to autophosphorylate in the absence or presence of indicated tyrosine kinase inhibitors (M, mubritinib; A, afatinib; C, canertinib; L, lapatinib) followed by immunoblotting with anti-pY4G10 and anti-EGFR antibodies. Similar results were observed in each of three biological replicates. *Autophosphorylation is observed without additional EGF due to EGF being present in the fetal bovine serum and in the initial assembly mixture. B, Association and dissociation of 100 nM of empty or EGFR-NLP to biotinylated EGF on the sensors with and without 1 μM unlabeled EGF or cetuximab. Data points indicate the mean of two technical duplicates and are representative of results from two biological replicates. C, Association and dissociation curves of 100 nM EGFR-NLPs binding to biotinylated EGF on the sensors in the presence of increasing levels of Cetuximab. Cetuximab concentrations were titrated by a 1.5-fold dilution to the concentrations displayed to the right of the graph. Curves were normalized by subtracting the 0 μM Cetuximab curve. D, Log plot of cetuximab inhibiting the binding of EGFR-NLPs to biotinylated EGF on the sensors. Data points indicate the mean of two technical duplicates and are representative of results from two biological replicates.