Subunits of an E3 Ligase Complex as Degrons for Efficient Degradation of Cytosolic, Nuclear, and Membrane Proteins

Protein degradation is a highly regulated cellular process crucial to enable the high dynamic range of the response to external and internal stimuli and to balance protein biosynthesis to maintain cell homeostasis. Within mammalian cells, hundreds of E3 ubiquitin ligases target specific protein substrates and could be repurposed for synthetic biology. Here, we present a systematic analysis of the four protein subunits of the multiprotein E3 ligase complex as scaffolds for the designed degrons. While all of them were functional, the fusion of a fragment of Skp1 with the target protein enabled the most effective degradation. Combination with heterodimerizing peptides, protease substrate sites, and chemically inducible dimerizers enabled the regulation of protein degradation. While the investigated subunits of E3 ligases showed variable degradation efficiency of the membrane and cytosolic and nuclear proteins, the bipartite SSD (SOCSbox-Skp1(ΔC111)) degron enabled fast degradation of protein targets in all tested cellular compartments, including the nucleus and plasma membrane, in different cell lines and could be chemically regulated. These subunits could be employed for research as well as for diverse applications, as demonstrated in the regulation of Cas9 and chimeric antigen receptor proteins.


Note S1: Design of degrons based on E3 ligase
The design of degrons from on SCF-Skp2 E3 ligase was based on previously published data on structure of Cul1-Rbx1-Skp1-F box Skp2 complex and interactions between its constituent proteins.[1,2] Design of SOCSbox degron is based on previously published data on structure of SOCS2-ElonginC-ElonginB E3 ligase complex.[3] Interaction data is summarized in figures S1-S6.
Skp2 F-box is based on a 69 aminoacid long F-box motif from Skp2.It consists of aminoacid residues 101-169 of human Skp2 between and including L1 and H4 domains.Sequence and interaction data is presented in figure S1. [2] With Skp1, we tested the full-length 162 aminoacid long human Skp1 protein.Skp1 binds Skp2 with C-terminal H5, H6, H7 and H8 helices.[2] Skp1 mutants were designed to not include these helices to prevent binding to cellular F-box proteins.H8 and H7 were removed in Skp1(ΔC131), H8, H7 and H6 in Skp1(ΔC111) and H8, H7,H6, and H5 in Skp1(ΔC94).Number in the mutant name represents the amino acid location in the Skp1 where the C-terminal deletion is made.Sequence and interaction data is presented in figure S2.
Cul1(ΔN501) is based on 275 C-terminal aminoacids of human Cul1 protein, which contain Rbx1 binding motifs.[1] Sequence and interaction data is presented in figure S3.
With Rbx1, we tested the whole 107 aminoacid long human Rbx1 protein.The Rbx1(ΔN36) mutant is missing 36 N-terminal aminoacids, that contain Cul1 binding motif.Rbx1(ΔN36) still contains Cterminal RING motif, which binds cellular E2 proteins.Such mutant would not be able to bind cellular Cul1 proteins but still be able to bind E2 proteins and potentially act as a degron.Sequence and interaction data is presented in figure S4.
SOCSbox consists of 41 C-terminal aminoacids of human SOCS2 protein containing H1, H2 and H3 alfa helices, which interact with ElonginB and ElonginC from the SOCS2-ElonginC-ElonginB E3 ligase complex.[3] Sequence and interaction data is presented in figure S5.S1.Gel image band intensity was quantified using ImageJ software.Band intensity values are normalized to the intensity of beta-actin band.

Figure S7 -Western blot of coiled-coil mediated degradation of luciferase substrate.
HA-tagged firefly luciferase substrate protein fused with the P3 CC-forming peptide was cotransfected with individual degrons fused with P4 partner CC-forming peptide.Transfection and immunoblotting is described in the main body of the paper, transfection plasmid mixtures are listed in Table S1.Gel image band intensity was quantified using ImageJ software.All band intensity values are normalized to the intensity of beta-actin band, ratios of anti-HA/beta-actin are normalized to the anti-HA/beta-actin ratio of firefly luciferase without cotransfected degrons.a Inhibition of coiled-coil mediated substrate degradation by MG132 proteasome inhibitor.Plasmids expressing luciferase substrate in fusion with P3 coil-forming peptide and best performing degrons in fusion with P4 coil-forming peptide were cotransfected into HEK293T cells.24h after transfection MG132 proteasome inhibitor in DMSO at final concentration of 3 µM or DMSO at the same v/v ratio were added in treated wells.Luciferase activity was measured 48h post-transfection.Values represent the mean ± SD of three cell cultures experiments and are normalized to the expression of luciferase without cotransfected degron constructs and untreated with MG132 or DMSO.b Western blot analysis of expression of luciferase substrate in fusion with P4 coil-forming peptide in cotransfection with best performing degrons in fusion with P4 coil-forming peptide in the presence and absence of MG132.Plasmids expressing the described proteins were transfected into HEK293T cells.24h post transfection MG132 proteasome inhibitor in DMSO at final concentration of 3 µM were added in treated wells.Immunoblotting is described in the main body of the paper, transfection plasmid mixtures are listed in Table S1.Gel image band intensity was quantified using ImageJ software.Band intensity values are normalized to the intensity of beta-actin band.Rapamycin or DMSO might effect the expression of luciferase substrates used in this study with a mechanism other than degradation by dimerization of substrate with degron, as we proposed.Here we titrated rapamycin and DMSO (at the same v/v ratio as rapamycin) to HEK293T cells transfected with all luciferase based substrate reporter proteins used in this study: FKBP:fLuc and fLuc:FKBP (a), FKBP:fLuc:NLS (b) and firefly luciferase reporter gene under minimal promotor in cotransfection with dCas and sgRNA (c).Plasmids expressing the described proteins were cotransfected into HEK293T cells.Rapamycin or DMSO were added to wells 24h post-transfection and luciferase activity was measured 48h post-transfection.

Figure S1 -Figure S2 -Figure S3 -Figure S4 -Figure S5 -
Figure S1 -Skp2 aminoacid sequence and interaction dataAmino acid sequence of human Skp2.Amino acid residues in red bind Skp1.Data on interacting amino acid residues is based on previously published reports.Aminoacid positions are annotated to human Skp2, with sequence before aminoacid 101 not shown.[2]

Figure S9 -
Figure S9 -Heat map data on the effect of linker length between CC dimer and both the substrate and degron.Values in Fig. 2c in the main body of the paper represent mean values of values presented here.Statistical analysis was conducted by performing two-sided unpaired t-test and is presented for each individual degron for selected data, indicated on the right, comparing both CCdegron and CC-substrate degradation.S -linker between CC and substrate, D -linker between CC and degron.

Figure S10 -
Figure S10 -Effect of rapamycin and DMSO on the expression of luciferase reporter substrates.

Figure S11 -
Figure S11-CAR proteins translocate to the plasma membrane in HEK293T cells.HEK293T cells were transfected with CAR:mCit:FKBP or CAR:mCit:FRB and BFP.Images were acquired 48 hours after transfection according to the methods described in the main body of this paper.mCitrine emission (530-550 nm) is here colored in cyan and BFP emission (420-460 nm) in magenta.As expected BFP is localized in the cytosol while CAR constructs localize to the plasma membrane.

Table S1 -Plasmid transfection mixtures performed in this study Figure Description of plasmid transfection mixture
S8b 50 ng of P3:fluc:HA construct, 500 ng of each P4-degrons fusion protein construct, empty pcDNA to a final amount of 1600 ng plasmid DNA per well S10 10 ng of each construct described in the figure, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well S11 50 ng of CAR:mCit:FKBP or CAR:mCit:FRB construct, 100 ng of BFP plasmid, empty pcDNA to a final amount of 400 ng plasmid DNA per well each luciferaze and degron fusion protein construct, amounts of TEVp are stated in the figure, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well 3c 10 ng of luciferaze-FKBP fusion construct, 100 ng of each FRB-degron fusion protein construct, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well 4b 50 ng of CAR-mCit-FRB construct, 400 ng of degron-FKBP construct, 100 ng of BFP, empty pcDNA to a final amount of 800 ng plasmid DNA per well 4c 10 ng of luciferaze-FKBP fusion construct, 100 ng of each FRB-degron fusion protein construct, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well 4d 50 ng of 1A-pMin-fLuc2CP reporter plasmid, 5 ng of FKBP:dCas9:VPR, 25 ng of sgRNA A, 100 ng of each FRB-degron fusion protein construct, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well 5b Left and right graph -10 ng of luciferaze-FKBP fusion construct, 100 ng of each FRBdegron fusion protein construct, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well Middle graph -50 ng of CAR-mCit-FRB construct, 400 ng of degron-FKBP construct, 100 ng of BFP, empty pcDNA to a final amount of 800 ng plasmid DNA per well 5c 10 ng of FKBP:fLuc fusion construct, 100 ng of each FRB:SSD fusion protein construct, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well 5d 10 ng of FKBP:fLuc fusion construct, 100 ng of each FRB:SSD fusion protein construct, 10 ng Renilla plasmid, empty pcDNA to a final amount of P3:fluc:HA construct, 500 ng of each P4-degrons fusion protein construct, empty pcDNA to a final amount of 1600 ng plasmid DNA per well S8a 10 ng of P3:fluc, 100 ng of each degron construct described in the figure, 10 ng Renilla plasmid, empty pcDNA to a final amount of 200 ng plasmid DNA per well