Doa10 facilitates membrane release of Ubc6

Our results so far establish a minimal system for polyubiquitination and membrane extraction of Ubc6. However, it remains unclear if Doa10 acts as a retrotranslocase that provides a specific conduit for the movement of a substrate out of the membrane and, thus, reduces the energy required for Cdc48-mediated extraction. We hypothesized that retrotranslocase activity would result in some spontaneous disengagement of Ubc6 from the membrane that would normally remain undetected because of reinsertion. We reasoned that factors that stabilize released Ubc6, but do not provide a direct pulling force like Cdc48, should prevent reinsertion and thus drive the equilibrium towards the released state.

We used the chaperone Get3 to test if Ubc6 is spontaneously released from Doa10 containing liposomes. Get3 is involved in membrane targeting of tail-anchored mem-brane proteins and wraps around the hydrophobic TM anchor (Mateja et al., 2015).

In a turbidity assay, we showed that Get3 prevents aggregation of Ubc6 when diluted into detergent-free buffer confirming that Get3 chaperones Ubc6 (Figure A4A). We then incubated liposomes containing Doa10 and Ubc6 with Get3, immobilized lipo-somes to beads (Figure A4B), and determined the amount of Ubc6 in the unbound fraction. When Ubc6 liposomes were fused to Doa10 liposomes, 43 ± 5% of Ubc6 was detected in the unbound fraction (Figure 3.4A, 3.4B and A4C), corresponding to an almost quantitative release of correctly oriented Ubc6. When fusion was inhibited, or when we used Syb liposomes lacking Doa10, only 7-9% were found in the supernatant, probably corresponding to the fraction of improperly reconstituted protein (Figure A1J and A1K). In the absence of Get3, or when we used a Get3 mutant defective in tail-anchored membrane protein binding (Get3I193D, (Mateja et al., 2015)), we observed no or drastically reduced release, respectively. We observed no difference between WT Get3 and the ATPase-deficient mutant Get3_D57N, as previously shown for the holdase activity of Get3 (Voth et al., 2014). Thus, Ubc6 is released in a Doa10-dependent man-ner. Doa10 mediates spontaneous release of its substrate Ubc6. A chaperone that traps a non-membrane bound state is sufficient to drive spontaneous release. Importantly, this activity of Doa10 does not require the substrate to be ubiquitinated.

To test for spontaneous release in the absence of a trap, we used SUMO-Ubc6_DL680, an amino-terminal fusion of SUMO to Ubc6, in which we additionally introduced a TEV protease cleavage site between the luminal carboxyl terminus of Ubc6 and the fluores-cent dye (Figure 3.4C). Addition of Ulp1 readily clips off the SUMO-tag resulting in Ubc6_DL680, and thus identifies correctly-oriented Ubc6. We fused SUMO-Ubc6_DL680 li-posomes with Doa10-lili-posomes and then added Ulp1 to cleave correctly oriented Ubc6.

We then added TEV protease and compared cleavage of correctly and wrongly oriented Ubc6 over time. About 70% of Ubc6_DL680 is cleaved within 30 min (Figure 3.4D and 3.4E). In contrast, in liposomes lacking Doa10, only 20% of Ubc6_DL680 is accessible to TEV protease, corresponding to the not properly reconstituted Ubc6 (Figure A1J and A1K). Wrong-side out SUMO-Ubc6_DL680, which exposes the TEV cleavage site on the outside of liposomes, is completely accessible to TEV protease and cleaved with faster kinetics, indicating that protease accessibility over time is not due to slow protease ac-tion. This experiment confirms that the carboxyl terminus of Ubc6 is retrotranslocated in a Doa10-dependent manner even in the absence of a trap.

35

Released Ubc6 (Sup / In)

+ Doa10 - Doa10

1. Ulp1 2. TEV Protease

0 10 20 30 40 50 60

Fraction not cleaved by TEV protease

-Doa10, Ubc6DL680

Figure 3.4: Release of Ubc6 from Doa10 containing liposomes. (A) Release of Ubc6 in the presence of different Get3 variants. Ubc6_DL680 liposomes were incubated with liposomes with or without Doa10, in the absence (Fused) or presence of Sybsol (Inhibited), immobilized to streptavidin magnetic beads via biotinylated lipids, and incubated for 16 h with either buffer or 10 µM of either WT Get3 (WT), an ATPase deficient mutant of Get3 (D57N) or a Get3 mutant that cannot efficiently interact with Ubc6 (I193D) (f.c. of 100 nM Ubc6, 40 nM Doa10). Input and supernatant samples were analyzed by SDS-PAGE and fluorescence scanning. (B) Quantification (mean and SD) of three experiments as in (A). (C) Schematic depiction of the experiment shown in (D) and (E).

Ubc6 was fused at the N-terminus to a SUMO tag, and a TEV protease cleavage site inserted between the C-terminus of Ubc6 and the fluorescent dye (DL680, indicated as a star). Ulp1 protease cleaves off the SUMO moiety (black arrow head), TEV protease the fluorescent dye (red arrow head). Liposomes were first incubated with Ulp1 to identify correctly oriented Ubc6, followed by incubation with TEV protease. (continued)

Figure 3.4 (continued): (D) Accessibility of TEV cleavage site in Ubc6.

SUMO-Ubc6_DL680 liposomes were fused with liposomes containing or lacking Doa10, followed by incubation with either buffer or Ulp1 in the presence or absence of solubilizing amounts of detergent (det) (lanes 1-3). Ulp1-treated liposomes were then incubated with buffer (lanes 4 and 5), TEV protease (lanes 6-10) or TEV protease and detergent (lane 11), and aliquots were taken at the indicated times.

Samples were analyzed by SDS-PAGE and fluorescence scanning. (E) Quantifica-tion (mean and SD) of three experiments as in (D). Band intensities from samples treated with TEV protease were normalized to the corresponding band intensities of samples without TEV protease (–). (F) Schematic representation of antibody quenching experiment. A fluorescence quenching anti-AlexaFluor488 antibody quenches fluorescence of dye molecules attached to wrong-side out Ubc6A488 or Ubc6A488that has disengaged from the membrane (grey), whereas dye attached to correctly oriented Ubc6_A488 (green) is shielded from antibody. (G) Time-course of fluorescence quenching experiments as depicted in (F). Liposomes containing Ubc6A488 were incubated with liposomes without (dashed lines) or with (solid lines) Doa10, co-reconstituted with either Syb (red) or the fusion-deficient Syb∆84 mutant (blue). Addition of the quenching antibody or of solubilizing amount of detergent (Triton X-100) indicated by arrows. (H)Quantification (mean and SD) of four experiments as in (G). Released fraction is defined as F_–Doa10(30 min) – F_+Doa10(30 min). Additionally, quantification of experiments as in (G), but in the presence of Syb_sol is also shown.

We employed a third experimental system to show retrotranslocation, which is based on quenching of an AlexaFluor488 fluorophore (A488) by an antibody. We hypothesized that binding of the antibody to a carboxyl-terminal dye should prevent reinsertion and thus act as a trap similar to Get3 (Figure 3.4F). When we mixed Ubc6_A488- and Doa10-liposomes, but inhibited fusion with either Syb_sol or Syb∆84, and then added the anti-A488 antibody, we observed a sudden decrease in fluorescence by 50%, corresponding to the fraction of wrong-side out protein that exposes its C-terminus to the outside of liposomes. Upon solubilization of liposomes with detergent, the antibody quenches the fluorescence of all A488 epitopes (Figure 3.4G). When liposomes were allowed to fuse, and we then added anti-A488 antibody, again, a sudden decrease in fluorescence was seen, but this time followed by a slower decrease in fluorescence to about 10%

of the original fluorescence signal within 30 min. Thus, in the presence of Doa10, the luminally-encapsulated part of Ubc6 becomes accessible to the antibody. The antibody not only acts as a reporter system for release, but – similar to Get3 – drives release to completion by stabilizing the released state and preventing re-insertion. Together, Get3 capture assay, protease protection assay, and the antibody accessibility assay show that Doa10 facilitates movement of the Ubc6 TM from the membrane to the aqueous phase, and thus reveal its retrotranslocase activity.

3.2.7 Structural elements in Doa10 required for retrotranslocation and