Short disordered termini and proline-rich domain are major regulators of UBQLN1/2/4 phase separation.

Highly homologous ubiquitin-binding shuttle proteins UBQLN1, UBQLN2, and UBQLN4 differ in both their specific protein quality control functions and their propensities to localize to stress-induced condensates, cellular aggregates, and aggresomes. We previously showed that UBQLN2 phase separates in vitro, and that the phase separation propensities of UBQLN2 deletion constructs correlate with their ability to form condensates in cells. Here, we demonstrated that full-length UBQLN1, UBQLN2, and UBQLN4 exhibit distinct phase behaviors in vitro. Strikingly, UBQLN4 phase separates at a much lower saturation concentration than UBQLN1. However, neither UBQLN1 nor UBQLN4 phase separates with a strong temperature dependence, unlike UBQLN2. We determined that the temperature-dependent phase behavior of UBQLN2 stems from its unique proline-rich region, which is absent in the other UBQLNs. We found that the short N-terminal disordered regions of UBQLN1, UBQLN2, and UBQLN4 inhibit UBQLN phase separation via electrostatics interactions. Charge variants of the N-terminal regions exhibit altered phase behaviors. Consistent with the sensitivity of UBQLN phase separation to the composition of the N-terminal regions, epitope tags placed on the N-termini of the UBQLNs tune phase separation. Overall, our in vitro results have important implications for studies of UBQLNs in cells, including the identification of phase separation as a potential mechanism to distinguish the cellular roles of UBQLNs and the need to apply caution when using epitope tags to prevent experimental artifacts.

Short N-terminal disordered regions and the proline-rich domain are major regulators of phase transitions for full-length UBQLN1, UBQLN2 and UBQLN4.

Highly homologous ubiquitin-binding shuttle proteins UBQLN1, UBQLN2 and UBQLN4 differ in both their specific protein quality control functions and their propensities to localize to stress-induced condensates, cellular aggregates and aggresomes. We previously showed that UBQLN2 phase separates in vitro, and that the phase separation propensities of UBQLN2 deletion constructs correlate with their ability to form condensates in cells. Here, we demonstrated that full-length UBQLN1, UBQLN2 and UBQLN4 exhibit distinct phase behaviors in vitro. Strikingly, UBQLN4 phase separates at a much lower saturation concentration than UBQLN1. However, neither UBQLN1 nor UBQLN4 phase separates with a strong temperature dependence, unlike UBQLN2. We determined that the temperature-dependent phase behavior of UBQLN2 stems from its unique proline-rich (Pxx) region, which is absent in the other UBQLNs. We found that the short N-terminal disordered regions of UBQLN1, UBQLN2 and UBQLN4 inhibit UBQLN phase separation via electrostatics interactions. Charge variants of the N-terminal regions exhibit altered phase behaviors. Consistent with the sensitivity of UBQLN phase separation to the composition of the N-terminal regions, epitope tags placed on the N-termini of the UBQLNs tune phase separation. Overall, our in vitro results have important implications for studies of UBQLNs in cells, including the identification of phase separation as a potential mechanism to distinguish the cellular roles of UBQLNs, and the need to apply caution when using epitope tags to prevent experimental artifacts.

Analyses of binding partners and functional domains for the developmentally essential protein Hmx3a/HMX3.

HMX3 is a homeodomain protein with essential roles in CNS and ear development. Homeodomains are DNA-binding domains and hence homeodomain-containing proteins are usually assumed to be transcription factors. However, intriguingly, our recent data suggest that zebrafish Hmx3a may not require its homeodomain to function, raising the important question of what molecular interactions mediate its effects. To investigate this, we performed a yeast two-hybrid screen and identified 539 potential binding partners of mouse HMX3. Using co-immunoprecipitation, we tested whether a prioritized subset of these interactions are conserved in zebrafish and found that Tle3b, Azin1b, Prmt2, Hmgb1a, and Hmgn3 bind Hmx3a. Next, we tested whether these proteins bind the products of four distinct hmx3a mutant alleles that all lack the homeodomain. Embryos homozygous for two of these alleles develop abnormally and die, whereas zebrafish homozygous for the other two alleles are viable. We found that all four mutations abrogate binding to Prmt2 and Tle3b, whereas Azin1b binding was preserved in all cases. Interestingly, Hmgb1a and Hmgn3 had more affinity for products of the viable mutant alleles. These data shed light on how HMX3/Hmx3a might function at a molecular level and identify new targets for future study in these vital developmental processes.