Antibody RING-Mediated Destruction of Endogenous Proteins

Abstract

To understand gene function, the encoding DNA or mRNA transcript can be manipulated and the consequences observed. However, these approaches do not have a direct effect on the protein product of the gene, which is either permanently abrogated or depleted at a rate defined by the half-life of the protein. We therefore developed a single-component system that could induce the rapid degradation of the specific endogenous protein itself. A construct combining the RING domain of ubiquitin E3 ligase RNF4 with a protein-specific camelid nanobody mediates target destruction by the ubiquitin proteasome system, a process we describe as antibody RING-mediated destruction (ARMeD). The technique is highly specific because we observed no off-target protein destruction. Furthermore, bacterially produced nanobody-RING fusion proteins electroporated into cells induce degradation of target within minutes. With increasing availability of protein-specific nanobodies, this method will allow rapid and specific degradation of a wide range of endogenous proteins.

Keywords: ARMeD; E3 ligase; nanobody-RING fusion; proteasome; protein degradation; ubiquitin.

Graphical abstract

Figure 1

Antibody-RING Mediated Destruction (ARMeD)—Principle, Tool Development, and Degradation of EYFP-PARG (A) Schematic representation of the principle of ARMeD. SUMO recognition motifs (SIMs) of the SUMO-targeted ubiquitin ligase (STUbL) RNF4 are replaced with a nanobody targeting a protein substrate of interest. Expression of this fusion protein allows binding of substrate to nanobody and RING-mediated ubiquitination leading to proteasomal degradation. HeLa Flp-in/T.Rex cells engineered to inducibly express GNb-1xRING or GNb-2xRING were either untreated (−) or doxycycline-treated (+) for 24 h. (B) mRNA levels were analyzed by qRT-PCR with beta-2 microglobulin (B2M) as housekeeping control, and products at 24 cycles were separated on an agarose gel. (C) Quantitative expression data were obtained from three independent RNA preparations from each condition, normalized to B2M mRNA and uninduced control samples. Error bars represent mean ± SD from three independent replicates. (D) Protein levels were analyzed by western blotting using an anti-camelid antibody. (E and F) HeLa Flp-in/T.Rex cells engineered to inducibly express GNb-1xRING and stably express YFP-PARG were induced as above and analyzed by western blotting using an anti-GFP antibody (E), or cells were grown in 96-well plates fixed and visualized by high-content (HC) imaging using IN Cell analyzer 2000 (F). (G) HC data were obtained from 152,668 (uninduced) or 80,745 (induced) cells in 6 wells, and quantitation of intracellular YFP performed using the InCell Developer toolbox. YFP intensity data are plotted as the mean of 6 wells ± SD. (H and I) To establish the pathway of protein degradation, cells were incubated with autophagy inhibitor bafilomycin A1 (Baf, 100 nM) or proteasome inhibitors bortezomib (1 μM) or MG132 (10 μg/mL) for 1.5 h prior to 16 h doxycycline induction. The role of other E3 ligases in degradation of substrate was examined by subjecting cells to inhibitors without Dox induction. Western blotting (H) and HC analysis (I) were performed as above. HC YFP-PARG intensity data were obtained from 20,000–40,000 cells grown in 12 wells of a 96-well plate for each condition and plotted as the mean of 12 well replicates ± SD (I). (J and K) YFP-PARG degradation and production of the ARMeD fusion was assessed in a time course experiment by collecting cells at indicated times after doxycycline addition. Samples were analyzed by western blotting (J) or HC imaging (K). HC YFP intensity data were obtained from 115,000–250,000 cells grown in 8 wells of a 96-well plate for each time point, normalized to uninduced control cells, and plotted as the mean of 8 well replicates ± SD. Statistical analysis was performed by a two-tailed unpaired t test.

Figure 2

ARMeD of YFP-PML (A and B) HeLa Flp-in/T.Rex cells engineered to inducibly express GNb-2xRING and stably express YFP-PML were either untreated (−) or Doxycycline treated (+) for 24 h. Protein levels were analyzed by western blotting using an anti-GFP antibody (A) or analyzed by HC imaging using IN Cell analyzer 2000 (B). (C) HC data were obtained from 33,775 (uninduced) or 33,434 (induced) cells in 9 wells, and quantification of YFP fluorescence was performed using the InCell Developer toolbox. Data representing YFP-PML total area/cell are plotted as means of 9 wells ± SD. (D and E) To establish the pathway of protein degradation, cells were incubated with autophagy inhibitor bafilomycin A1 (Baf, 100 nM) or proteasome inhibitors bortezomib (1 μM) or MG132 (10 μg/mL) for 1.5 h prior to 16 h doxycycline induction. The role of E3 ligases other than the ARMeD fusion in degradation of substrate was examined by subjecting cells to inhibitors without Dox induction. Western blotting (D) and HC analysis (E) were performed as above. The HC YFP-PML data (total area/cell) were obtained from 20,000–50,000 cells grown in 12 wells of a 96-well plate for each condition and plotted as the mean of 12 well replicates ± SD (E). (F and G) YFP-PML degradation was assessed over time by collecting cells at indicated times after doxycycline addition and performing western blotting (F) or high-content analysis (G). The HC EYFP-PML data (total area/cell) were obtained from 20,000–25,000 cells grown in 8 wells of a 96-well plate for each time point, normalized to uninduced control cells, and plotted as the mean of 8 well replicates ± SD. Statistical analysis was performed by a two-tailed unpaired t test.

Figure 3

ARMeD of YFP-RNF146 and YFP-PEX10 (A and C) HeLa Flp-in/T.Rex cells engineered to inducibly express GNb-1xRING and stably express YFP-RNF146 (A) or PEX10 (C) were grown in 96-well plates and either untreated (−Dox) or doxycycline-treated (+Dox) for 24 h, fixed, and visualized by HC imaging using IN Cell analyzer 2000. (B and D) HC data were obtained from 45,278 (uninduced) or 45,749 (induced) cells for RNF146 (B) or from 128,710 (uninduced) or 110,743 (induced) cells for PEX10 (D), and quantitation of intracellular YFP was performed using InCell Developer toolbox. YFP intensity data are plotted as mean of 3 wells ± SD. Statistical analysis was performed by a two-tailed unpaired t test.

Figure 4

Degradation of Endogenous NEDD8 Protease NEDP1 with ARMeD Constructs (A) HeLa Flp-in/T.Rex cells were transfected with non-targeting (siNT, lane 1) or NEDP1 (siNEDP1, lane 2) siRNA, and cell extracts harvested 72 h after transfection. Lanes 3–10: HeLa Flp-in/T.Rex cells engineered to inducibly express NEDP1 specific nanobody-RING constructs were untreated (−) or doxycycline-treated (+) for 24 h. Protein levels were analyzed by western blotting using anti-NEDP1, anti-camelid, and anti-NEDD8 antibodies. α-Tubulin was used as loading control. NEDD8-cullins and NEDD8 monomers and dimers are indicated by arrows. (B and C) To establish the pathway of protein degradation by NNb2-1xRING (B) and NNb9-2xRING (C), cells were incubated with autophagy inhibitor bafilomycin A1 (Baf, 100 nM) or proteasome inhibitors bortezomib (1 μM) or MG132 (10 μg/mL) for 1.5 h prior to 16 h doxycycline induction. The role of other E3 ligases in degradation of substrate was examined by subjecting cells to inhibitors without Dox induction. (D) Induction of NEDP1 ARMeD fusions and rate of NEDP1 degradation after doxycycline addition was assessed by western blotting using anti-NEDP1 and anti-camelid antibodies. A non-specific (NS) band recognized by the NEDP1 antibody served as an additional loading control. (E) Parallel reaction monitoring to quantify NEDP1 depletion. Example MS2 chromatograms for fragment ions y3–y6 of the NEDP1 peptide LAFVEEK with and without doxycycline treatment. Dashed lines are peak boundaries as reported by Skyline. Mass errors for most prominent peaks are labeled as ppm. (F) Example MS2 peak areas used for quantification of NEDP1 ARMeD knockdown (NNb2-1xRING) and control strain (parental). (G) Median enrichment or depletion compared to PARENTAL-dox of the LAFVEEK, LEAFLGR, and QVAEKLEAFLGR peptides. Error bars are SE. Statistical analysis was performed by a two-tailed unpaired t test. See also Figure S1 and Table S2.

Figure 5

Total Proteome Consequences of Nanobody-RING Fusion Expression (A) Design of a SILAC experiment to identify protein abundance changes to cells during induction of the NEDP1 nanobody-RING by Dox. (B) Gel image of whole cell extracts from SILAC mixes as shown from (A). (C) Scatterplot showing the SILAC H/L ratio data for 4,907 proteins common to data derived from two SILAC mixes. Grey markers indicate proteins not identified as significantly different in both comparisons nor consistently responding to Dox. Red markers are proteins with significant abundance change after Dox treatment that is consistent across both mixes. x = -y line shown to highlight consistently responding proteins. (D) Rank versus average response to Dox across both SILAC mixes for the 4,907 proteins shown in (C) CDK6, cyclin-dependent kinase 6; RRM1, ribonucleoside-diphosphate reductase large subunit; MKI67, marker of proliferation Ki-67; PSAT1, phosphoserine aminotransferase 1; HSPA6, heat shock protein family A (Hsp70) member 6; HMOX1, heme oxygenase 1. NEDP1^∗ included for comparison; NEDP1 data derived from PRM experiment shown in Figure 4 and not from this SILAC experiment. (E) Slice-specific total protein intensity data for NEDD8 (left panel) and the NEDP1 (right panel) nanobody-RING construct. For each slice, the average intensity across both mixes is shown. See also Figure S2 and Table S1.

Figure 6

Acute and Rapid Degradation of Target Proteins by ARMeD Proteins (A) Coomassie-blue stained SDS-PAGE analysis of purified GFP nanobody-RING fusions, wild-type (WT) (GNb-1xRING and GNb-2xRING), and (GNb-1xmtRING and GNb-2xmtRING). (B) Nickel bead pull-down assays of recombinant 6His-GFP-SUMO1 with nanobody-RING fusions evaluated with SDS-PAGE and Coomassie staining (I, input; S, supernatant; P, pull-down). Fused RNF4 RING (2xRING) is used as negative control. Shuffle and Rosetta are bacterial strains used for expressing recombinant fusion proteins. (C) Lysine discharge assays with ubiquitin loaded Ubc5 (Ub-Ubc5) in presence of fused RNF4 RING (2xRING), GNb-2xRING, and GNb-2xmtRING. Samples were removed at indicated times (minutes) and analyzed by non-reducing SDS-PAGE. A sample reduced with DTT is indicated. (D) HeLa Flp-in/T.Rex cells stably expressing YFP-PMLIII were injected with a 1:1 mixture of GNb-2xRING and mCherry-SIM, and images were collected every 2 min. Injected cells (red arrow) and uninjected cells (yellow arrow) are indicated. The images shown were taken at 0, 10, 16, and 20 min following injection. (E) Injected cells were identified using mCherry fluorescence, and mean summed YFP intensity of injected cells was obtained following background subtraction and plotted for each time point ± SD. (F and G) Efficiency of (F) purified protein delivery to cells and (G) target protein degradation. HEK293 cells stably expressing YFP-SP100 were electroporated with a mix of mCherry-SIM protein and either GNb-2xRING or GNb-2xmtRING (either 0.375 pg or 1.5 pg of each purified protein/cell) and (F) mCherry or (G) YFP fluorescence analyzed by HC imaging using IN Cell analyzer 2000. HC data were obtained from 29,923/26,007 (0.375/1.5 pg GNb-2xRING/cell) or 21,901/32,866 (0.375/1.5 pg GNb-2xmtRING/cell) cells in 12 wells, and quantitation of each fluorescence signal was determined individually using InCell Developer toolbox. Percentage of cells with mCherry fluorescence above background are plotted as means of 12 wells ± SD (F); and YFP fluorescence representing PML/SP100 foci total area/cell are plotted as means of 12 wells ± SD (G). Statistical analysis was performed by a two-tailed unpaired t test. (H) Rate of degradation was assessed by electroporating HEK293 YFP-SP100 cells above with 12 μg of purified GNb-2xRING in a total volume of 100 μL and collecting samples at the indicated times after electroporation (lanes 2–7). To prolong the activity of the GNb-2xRING fusion, the cells were co-transfected with the same amount of the purified recombinant fusion and 500 fg/cell of an in vitro-generated GNb-2xRING mRNA combined with a similar amount of similarly generated mCherry mRNA as a transfection internal positive control (lanes 8–13). Samples were collected at the same time intervals as the GNb-2xRING fusion without mRNA transfections (lanes 2–7). Negative control cells (lane 1) were only transfected with 1 pg/cell of the mCherry mRNA and collected after 6 h. YFP-SP100 degradation was scored by western blotting, using an anti-GFP antibody. A non-specific band (NS) just below the target protein served as a loading control. GNb-2xRING was detected using an anti-camelid antibody and expression of electroporated mRNAs was monitored using an anti-mCherry antibody. See also Figures S3 and S4 and Videos S1 and S2.

Figure 7

Rapid Antibody RING-Mediated Destruction of Endogenous NEDP1 (A, D, and G) Nickel bead pull-down assays of recombinant 6His-NEDP1 with nanobody-RING fusions NNb2-1xRING and 2xRING (A), NNb9-1xRING and 2xRING (D), and NNb7-1xRING and 2xRING (G) were evaluated with SDS-PAGE and Coomassie staining (In, input; S, supernatant; P, pull-down). Fused RNF4 RING (2xRING) is used as negative control. (B, E, and H) The activity of the NNb2-1xRING and 2xRING (B), NNb9-1xRING and 2xRING (E), and NNb7-1xRING and 2xRING (H) was tested in lysine discharge assays with ubiquitin loaded Ubc5 (Ub-Ubc5). A fused RNF4 RING (2xRING) served as a positive control. Samples were removed at indicated times (minutes) and analyzed by non-reducing SDS-PAGE. (C, F, and I) To assess activity of purified recombinant NEDP1 ARMeD fusions in cells, HEK293 cells were electroporated with NNb2-1xRING and 2xRING (C), NNb9-1xRING and 2xRING (F), or NNb7-1xRING and 2xRING (I) and harvested at the indicated time point after electroporation. NEDP1 and NEDD8 were analyzed by western blotting. NEDP1, a non-specific band (NS), NEDD8-cullins, and NEDD8 monomers and dimers are indicated by arrows. See also Figures S5 and S6.