Proteomic analysis of ubiquitin ligase KEAP1 reveals associated proteins that inhibit NRF2 ubiquitination

Abstract

Somatic mutations in the KEAP1 ubiquitin ligase or its substrate NRF2 (NFE2L2) commonly occur in human cancer, resulting in constitutive NRF2-mediated transcription of cytoprotective genes. However, many tumors display high NRF2 activity in the absence of mutation, supporting the hypothesis that alternative mechanisms of pathway activation exist. Previously, we and others discovered that via a competitive binding mechanism, the proteins WTX (AMER1), PALB2, and SQSTM1 bind KEAP1 to activate NRF2. Proteomic analysis of the KEAP1 protein interaction network revealed a significant enrichment of associated proteins containing an ETGE amino acid motif, which matches the KEAP1 interaction motif found in NRF2. Like WTX, PALB2, and SQSTM1, we found that the dipeptidyl peptidase 3 (DPP3) protein binds KEAP1 via an "ETGE" motif to displace NRF2, thus inhibiting NRF2 ubiquitination and driving NRF2-dependent transcription. Comparing the spectrum of KEAP1-interacting proteins with the genomic profile of 178 squamous cell lung carcinomas characterized by The Cancer Genome Atlas revealed amplification and mRNA overexpression of the DPP3 gene in tumors with high NRF2 activity but lacking NRF2 stabilizing mutations. We further show that tumor-derived mutations in KEAP1 are hypomorphic with respect to NRF2 inhibition and that DPP3 overexpression in the presence of these mutants further promotes NRF2 activation. Collectively, our findings further support the competition model of NRF2 activation and suggest that "ETGE"-containing proteins such as DPP3 contribute to NRF2 activity in cancer.

Figure 1. The KEAP1 protein interaction network is enriched for ETGE-containing proteins

(A) Cartoon schematic of NRF2 ubiquitination by KEAP1. KEAP1 inactivation is shown through cysteine modification and the competitive association of ETGE-containing proteins. (B) Schematic representation of the KEAP1 protein interaction network as defined by affinity purification and mass spectrometry. Nodes and edges were sized and colored according to probabilistic scoring approach, sequence and source of data. Circular nodes were sized according to SPOTLITE probability (10% FDR). Triangular nodes represent borderline SPOTLITE scored interactions that were observed across multiple APMS runs. See also Table S1.

Figure 2. The ETGE motif is required for KEAP1 association

(A) Streptavidin affinity purified KEAP1 protein complexes from HEK293T cells were analyzed by Western blot for the indicated FLAG-tagged proteins (SBP, streptavidin binding peptide; HA, hemagglutinin). (B) Streptavidin affinity purified protein complexes from HEK293T cells stably expressing the indicated SBPHA-tagged proteins were analysed by Western blot for the indicated proteins. (C) HEK293T cells stably expressing FLAG-tagged WDR1 or TSC22D4 were transfected with SBPHA-tagged KEAP1, KEAP1 BTB, or KEAP1 KELCH before FLAG affinity purification and Western blot (D) FLAG-tagged wild-type or ET/SGE-deletion constructs were expressed in HEK293T cells stably expressing SBPHA-KEAP1 before streptavidin affinity purification and Western blot. (E) HEK293T cells stably expressing FLAG-tagged KEAP1 were transfected with the indicated expression constructs; associated proteins were revealed by FLAG immunoprecipitation and Western blot. (F) HEK293T cells were transfected with the indicated ET/SGE-containing protein, constitutively expressed Renilla luciferase and the NRF2-driven Firefly luciferase (NQO1-ARE). Error bars represent standard deviation from the mean from 3 biological replicates (* p<0.05; Students T-test).

Figure 3. DPP3 is a KEAP1 interacting protein

(A) Schematic protein interaction network for DPP3 and KEAP1. Node and edge coloring and sizing are consistent with Figure 1. (B) HEK293T cells stably expressing KEAP1 and DPP3 were lysed and subjected to two sequential rounds of affinity purification before mass spectrometry. Data shown represent biological duplicate experiments, wherein the order of affinity purifications was reversed. (C) Protein complexes from HEK293T cells stably expressing FLAG-KEAP1 were affinity purified and analyzed by Western blot. (D) Protein complexes from HEK293T cells stably expressing SBPHA-DPP3 or SBPHA-DPP3-Y318F were streptavidin affinity purified and analyzed by Western blot. (E) Endogenous DPP3 from HEK293T cells was immunopurified and analyzed by Western blot for the indicated endogenous proteins. (F) Protein complexes were FLAG affinity purified from the lung adenocarcinoma cell line H2228 stably expressing FLAG-KEAP1 and analyzed by Western blot. (G) HEK293 cells were transfected with venus-KEAP1 and the indicated mCherry-fused DPP3 expression construct. (H) HEK293T cells transfected with FLAG-DPP3 and stained for FLAG and endogenous KEAP1. Scale =20 μm. (I) HEK293T cells were transfected with NQO1-ARE-luciferase, constitutively active Renilla luciferase and the indicated expression plasmid before lysis and luciferase quantification (* P<0.05 across three biological replicate experiments).

Figure 4. DPP3 interacts with the KELCH domain of KEAP1 via its ETGE motif

(A) Protein complexes from HEK293T cells stably expressing SBPHA-KEAP1, SBPHA-BTB, and SPBHA-KELCH were Streptavidin affinity purified and analyzed by Western blot. (B) Cells stably expressing FLAG-KEAP1 or the FLAG-KELCH domains of KEAP1 were transfected with the indicated SBPHA-DPP3 construct before affinity purification and Western blot. (C) Protein complexes from HEK293T cells stably expressing the indicated fusion protein were Streptavidin affinity purified and analyzed by Western blot. (D) HEK293T cells were transiently co-transfected with FLAG-GFP, FLAG-DPP3-WT, or FLAG-DPP3-AAGE (alaΔ) before FLAG-affinity purification and Western blot. (E) The KELCH domain of Keap1 (PDB ID 1X2R) adopts a six-bladed β-propeller structure (cyan). The ETGE motif of NRF2 (orange) binds near the central pore of the β-propeller. (F) The structure of human DPP3 (PDB ID 3FVY, blue) reveals an ETGE motif (residues 480–483, green) in an unstructured surface loop. (G) KEAP1 binding to the ETGE peptide (orange sticks) is stabilized by both hydrogen bonds (to serine and asparagine residues, cyan sticks) and electrostatic interactions (to arginine residues, cyan sticks) with KEAP1. (H) Superposition of the NRF2 ETGE motif bound to KEAP1 with the ETGE motif of DPP3 reveals similar conformations.

Figure 5. DPP3 competes with NRF2 for KEAP1 binding

(A) Schematic representation of the sequential affinity purification approach employed to purify a KEAP1-KEAP1 homodimer or KEAP1-BTB domain pseudo-monomer. (B) HEK293T cells stably expressing FLAG-KEAP1 and SBPHA-KEAP1 or FLAG-BTB and SBPHA- KEAP1 were transfected with VSV-DPP3-WT or VSV-WTX^1–211 truncation mutant before sequential streptavidin and FLAG affinity purification and Western blot. (C) HEK293T cells stably expressing FLAG-BTB and SBPHA-KEAP1 were co-transfected with VSV-WTX^(1–211), VSV-DPP3 or VSV-DPP3^ΔETGE. Protein complexes were affinity purified with Streptavidin then FLAG and analyzed by Western blot. (D and E) HEK293T cells stably expressing SBPHA-KEAP1 were co-transfected with VSV-UB1, FLAG-NRF2, and either SBPHA-DPP3-WT, SBPHA-DPP3^ΔETGE or negative control Venus-NPM1. NRF2 was FLAG affinity purified and analyzed by Western blot.

Figure 6. DPP3 is an activator of NRF2-mediated transcription

(A) HEK293T cells were transfected with the indicated plasmid along with constitutively expressed Renilla luciferase and the NQO-1 promoter driving Firefly luciferase (NQO1-ARE). Cells were lysed and luciferase values were normalized to Renilla. Error bars represent standard deviation from the mean over 3 biological replicates, * p<0.05; Students T-test as compared to NPM1. (B) HEK293T cells were transfected with 10nM of the indicated siRNA. Protein lysate was analyzed by Western blot for the indicated endogenous protein. (C) HEK293T cells stably expressing the ARE reporter and Renilla luciferase were transfected with the indicated siRNA. Error bars represent standard deviation from the mean from 3 biological replicates. * p<0.05; Students T-test as compared to CNT. (D and E) HEK293T cells were transfected with siRNAs against the indicated mRNAs before mRNA isolation and qPCR for the indicated endogenous target genes. Relative expression was calculated based on expression of endogenous target gene transcript normalized to GAPDH. Error bars represent standard deviation from the mean from 3 biological replicates. * p<0.05; Students T-test as compared to CNT. (F and G) HEK293T cells were transfected with the indicated siRNAs or plasmids. Cells were treated with 50μM tBHQ 18 h priors to lysis. Error bars represent standard deviation from the mean from 3 biological replicates. * p<0.05; Students T-test.

Figure 7. DPP3 expression positively associates with NRF2 activity in squamous cell lung cancer

(A) DPP3 mRNA abundance in 175 squamous cell lung carcinomas and 27 matched normal tissues (p=4.601e-14; Kruskal-Wallis test). Normal or lung SQCC subtype is indicated by color. With respect to tumor subtype, DPP3 is over-expression is enriched within the primitive subtype (p=0.03334; Kruskal-Wallis test; see also Figure S2). (B) DPP3 mRNA expression positively correlates with DPP3 genomic copy number (Spearman rank correlation). (C) Correlation of DPP3 mRNA expression with NRF2 mutational status (p=0.00141; Kruskal-Wallis test). (D) Correlation of DPP3 mRNA expression with KEAP1 mutational status (p=0.03718; Kruskal-Wallis test). (E and F) HEK293T cells were transiently transfected with NQO1-ARE-luciferase reporter, constitutively active Renilla reporter and the indicated expression plasmids (* P<0.05 across at least three biological triplicate experiments) (G) DPP3 mRNA expression positively associates with NRF2 target gene expression (Spearman rank correlation test). The NRF2 gene signature consists of 15 genes (3). Triangles represent tumors with NRF2 mutations. (See also Figure S2 and S3).