Phase separation of polyubiquitinated proteins in UBQLN2 condensates controls substrate fate

Abstract

Ubiquitination is one of the most common posttranslational modifications in eukaryotic cells. Depending on the architecture of polyubiquitin chains, substrate proteins can meet different cellular fates, but our understanding of how chain linkage controls protein fate remains limited. UBL-UBA shuttle proteins, such as UBQLN2, bind to ubiquitinated proteins and to the proteasome or other protein quality control machinery elements and play a role in substrate fate determination. Under physiological conditions, UBQLN2 forms biomolecular condensates through phase separation, a physicochemical phenomenon in which multivalent interactions drive the formation of a macromolecule-rich dense phase. Ubiquitin and polyubiquitin chains modulate UBQLN2's phase separation in a linkage-dependent manner, suggesting a possible link to substrate fate determination, but polyubiquitinated substrates have not been examined directly. Using sedimentation assays and microscopy we show that polyubiquitinated substrates induce UBQLN2 phase separation and incorporate into the resulting condensates. This substrate effect is strongest with K63-linked substrates, intermediate with mixed-linkage substrates, and weakest with K48-linked substrates. Proteasomes can be recruited to these condensates, but proteasome activity toward K63-linked and mixed linkage substrates is inhibited in condensates. Substrates are also protected from deubiquitinases by UBQLN2-induced phase separation. Our results suggest that phase separation could regulate the fate of ubiquitinated substrates in a chain-linkage-dependent manner, thus serving as an interpreter of the ubiquitin code.

Keywords: Condensates; phase separation; proteasome; protein quality control; ubiquitin.

Fig. 1.

Ubiquitinated substrates phase-separate and sediment UBQLN2 in a linkage-dependent manner. (A) Fluorescence microscopy of 10 μM UBQLN2 (1% labeled with Alexa Fluor 647) incubated with or without 500 nM ubiquitinated (or not Ub’ed) substrate for 1 h at 30 °C. (B) Quantification of droplet size from images in B (K63: n = 125, Mixed: 161, K48: 102 droplets) and statistics applied with Welch’s two-tailed t test (****P < 0.0001). (C) Sedimentation assays use R-Neh2Dual-sGFP substrate ubiquitinated with K63-linked (Rsp5), K48-linked (Ubr1), or mixed linkage chains (Keap1/Cul3/Rbx1). After mixing substrate with UBQLN2 and incubating at 30 °C to allow phase separation to occur, condensates are pelleted (P) by centrifugation and separated from soluble (S) proteins before gel analysis. (D) 10 μM UBQLN2 (1% labeled with Alexa Fluor 647) was incubated ± 1 µM ubiquitinated substrate for 1 h; soluble (S) and pelleted (P) proteins were separated by centrifugation. (E) UBQLN2 sedimentation as a function of substrate concentration. (F) Quantification of replicate data (n = 4 to 6) from E.

Fig. 2.

Proteasome is recruited to phase-separated UBQLN2-substrate condensates. (A) 100 nM TagRFP-T-Rpn6 containing proteasome and 500 nM K48-linked or K63-linked or 250 nM mixed linkage substrate were incubated with 10 μM UBQLN2 (1% Alexa 647 labeled) and 100 μM proteasome inhibitor cocktail (epoxomicin, MG132, bortezomib) for 1 h as indicated. Individual channels are shown in SI Appendix, Fig. S4. Soluble (S) and pelleted (P) proteins were separated by centrifugation and imaged using fluorescence. (B and C) Quantification of replicate data from A. * indicates p ≤ 0.05 relative to proteasome alone (B) or UBQLN2 alone (C) (two-tailed Welch’s t test). Error bars are SEM from 3 to 4 measurements. (D) Fluorescence microscopy of 10 µM UBQLN2, 100 nM proteasome, 500 nM ubiquitinated sGFP substrates, proteasome inhibitor cocktail, 1× deg buffer, and 1 h at 30 °C (Scale bar, 10 µm.)

Fig. 3.

UBQLN2-induced phase separation inhibits proteasomal degradation in a polyUb-linkage-dependent manner. (A) Degradation of 250 nM ubiquitinated R-Neh2Dual-ACTR-DHFR substrate (Cy5- labeled N-terminal to DHFR) by 100 nM 26S proteasome with or without 10 μM UBQLN2 (1% DyLight-488 labeled) at 30 °C for 30′. ΔSTI1-II is a UBQLN2 variant that does not phase-separate on its own. (B) Quantification of replicate data (% substrate remaining, quantifying total amount of full-length and ubiquitinated substrate) from A. Error bars are SEM from 3 to 4 measurements. * indicates P ≤ 0.01 for proteasome + UBQLN2 relative to both proteasome alone and proteasome + UBQLN2ΔSTI1-II (two-tailed Welch’s t test). Quantified regions of example gels are shown in (SI Appendix, Fig. S6). (C) Sedimentation assays under the same condition as assays in A (but without proteasome) confirm reciprocal cosedimentation of ubiquitinated substrates with WT but not ΔSTI1-II UBQLN2.

Fig. 4.

Phase separation protects polyubiquitin and polyubiquitinated substrates from DUB activity. (A) Deubiquitination of 250 nM ubiquitinated (Ub+) R-Neh2Dual-ACTR-DHFR substrate (Cy5-labeled N-terminal to DHFR) by oTUB1, AMSH, or vOTU with or without 10 μM UBQLN2. Nonubiquitinated substrate (Ub−) is shown as a size reference. (B) Quantification of replicate data (% highly ubiquitinated substrate remaining) from A. Error bars are SEM from 3 to 4 measurements. * indicates P < 0.05 for DUB + UBQLN2 relative to both DUB alone and DUB + UBQLN2ΔSTI1-II (two-tailed welch’s t test). Quantified regions of example gels are shown in (SI Appendix, Fig. S9). (C) The linkage-independent DUB vOTU (100 nM AF488-labeled vOTU) colocalizes with 50 μM UBQLN2 in the presence of unlabeled 50 μM K63-Ub₄ (Scale bar, 10 μm). (D) 25 µM Ub (6.25 µM Ub₄) of K63-Ub₄ or K48-Ub₄ was incubated with 50 nM vOTU in the presence of 60 μM UBQLN2 or UBQLN2 ΔSTI1-II at 37 °C and time points were run on an SDS-PAGE gel. (E) Quantification of DUB assays from D. The Ub₄ band volumes were quantified as single-exponential fits (lines). Error bars are SD from n = 3 DUB assay experiments. (F) 100 nM each K48-linked R-Neh2Dual-ACTR-DHFR and K63-linked R-Neh2Dual-sGFP were incubated with 20 nM vOTU in the presence of 1 µM UBQLN2 or UBQLN2 ΔSTI1-II at 30 °C and time points were run on an SDS-PAGE gel and detected via fluorescence. Arrow indicates highly ubiquitinated K63-linked substrate that persists in the presence of UBQLN2.

Fig. 5.

Ubiquitinated substrates condense with UBQLN2 to form membraneless compartments capable of recruiting protein quality control machinery. Our data show that ubiquitinated substrates can induce condensate formation with UBQLN2 as well as also be recruited into UBQLN2-substrate condensates, with K63-linked substrates showing greater propensity for this behavior over K48-linked substrates at similar substrate concentrations. Proteasomes can be recruited into these condensates although a proportion remains soluble and excluded from the condensate population. K63-linked ubiquitinated substrates are protected from degradation while K48-linked ubiquitinated substrates can still be degraded.