USP7 Acts as a Molecular Rheostat to Promote WASH-Dependent Endosomal Protein Recycling and Is Mutated in a Human Neurodevelopmental Disorder

Abstract

Endosomal protein recycling is a fundamental cellular process important for cellular homeostasis, signaling, and fate determination that is implicated in several diseases. WASH is an actin-nucleating protein essential for this process, and its activity is controlled through K63-linked ubiquitination by the MAGE-L2-TRIM27 ubiquitin ligase. Here, we show that the USP7 deubiquitinating enzyme is an integral component of the MAGE-L2-TRIM27 ligase and is essential for WASH-mediated endosomal actin assembly and protein recycling. Mechanistically, USP7 acts as a molecular rheostat to precisely fine-tune endosomal F-actin levels by counteracting TRIM27 auto-ubiquitination/degradation and preventing overactivation of WASH through directly deubiquitinating it. Importantly, we identify de novo heterozygous loss-of-function mutations of USP7 in individuals with a neurodevelopmental disorder, featuring intellectual disability and autism spectrum disorder. These results provide unanticipated insights into endosomal trafficking, illuminate the cooperativity between an ubiquitin ligase and a deubiquitinating enzyme, and establish a role for USP7 in human neurodevelopmental disease.

Figure 1. USP7 is an integral component of the MAGE-L2-TRIM27 ubiquitin ligase complex and is required for WASH-mediated endosomal protein recycling

(A) Tandem affinity purification of MAGE-L2 followed by LC-MS/MS identified USP7 as a strong binding partner of MAGE-L2. Spectral counts normalized to control pulldown are shown. (B) Co-expressed Myc-MAGE-L2 and USP7-FLAG co-immunoprecipitate. Cells were transfected with indicated constructs for 48 hrs before anti-Myc (MAGE-L2) immunoprecipitation and immunoblotting. (C) Endogenous TRIM27 and USP7 co-immunoprecipitate. Cell lysates were immunoprecipitated with the indicated antibodies followed by immunoblotting. (D) GFP-tagged USP7 was expressed in U2OS cells and co-localization was determined with mCherry-tagged TRIM27 or anti-FAM21. Co-localization was quantitated by Pearson's correlation (Rr) and is shown. Scale bar represent 20 μm. (E) Both recombinant MAGE-L2 and TRIM27 bind purified GST-USP7, but not GST, in vitro. GST pulldown assays were performed with the indicated proteins followed by immunoblotting. (F) Summary of results determining the binding and architecture of the MAGE-L2-TRIM27-USP7 complex. See also Figure S1.

Figure 2. USP7 is required for WASH-mediated endosomal protein recycling

(A and B) Knockdown of USP7 disrupts CI-M6PR recycling. U2OS cells were treated with the indicated siRNAs for 72 hrs followed by immunostaining (A) and quantification (B). (C) Trafficking of cell surface labeled CI-M6PR is impaired upon knockdown of USP7. U2OS cells were treated with indicated siRNAs for 72 hrs before cell surface CI-M6PR was labeled with anti-CI-M6PR for 30 mins. Cells were fixed at the indicated time points after labeling, the internalized CI-M6PR was detected, and the percentage of cells with juxtanuclear internalized CI-M6PR was determined. (D and E) Depletion of USP7 impairs cholera toxin subunit B (CTxB), but not transferrin (Tf) trafficking. U2OS cells were treated with the indicated siRNAs for 72 hrs before fluorescently labeled CTxB and Tf were added to cells for 15 mins. Cells were fixed at the indicated time points post-labeling and the percentage of cells with CTxB and Tf localized with p97-golgin was determined by counting (D) or Pearson correlation (E). (F-I) Knockdown of USP7 decreases endosomal F-actin and Arp2/3 levels. U2OS cells were treated with the indicated siRNAs for 72 hrs followed by staining for F-actin (Phalloidin) and endosomal marker FAM21 (F) and quantification (G) or ArpC5 and endosomal marker VPS35 (H) and quantitation (I). (J) USP7-RNAi induced defects in CI-M6PR recycling are rescued by expression of endosomal-targeted WASH^VCA. Cells were transfected with the indicated siRNAs for 24 hrs followed by expression of the indicated plasmids for an additional 48 hrs before immunostaining and quantitation for CI-M6PR localization in GFP-positive transfected cells. (K) MAGE-L2 SA mutant has reduced USP7 affinity. HeLa cells were treated siMAGE-L2 for 24 hrs followed by expression of the indicated constructs for 48 hrs before anti-Myc (MAGE-L2) immunoprecipitation and immunoblotting. (L) USP7 interaction with MAGE-L2 is critical for CI-M6PR endosomal protein recycling. HeLa cells were treated with the indicated siRNAs for 24 hrs followed by expression of the indicated constructs for 48 hrs before anti-CI-M6PR immunostaining. Quantitation was performed only on transfected cells. (M and N) HeLa cells expressing MAGE-L2 SA mutant have reduced ArpC5 (M) and F-actin (N) on FAM21-positive endosomes. HeLa cells were treated with the indicated siRNAs for 24 hrs followed by expression of the indicated constructs for 48 hrs before anti-ArpC5 immunostaining (K) or F-actin staining (Phalloidin; L). Quantitation was performed only on transfected cells. Results are representative of at least three replicate experiments. More than 100 cells were quantitated. Values shown are mean + SD. Asterisks indicates p<0.05. Scale bars represent 20 μm. See also Figure S2.

Figure 3. USP7 protects TRIM27 from auto-ubiquitination-induced degradation

(A) Knockdown of USP7 dramatically reduces TRIM27 levels, without affecting WASH and retromer complex levels. HeLa cells were treated with the indicted siRNAs for 72 hrs before immunoblotting for the indicted proteins. Asterisks indicate non-specific bands. (B) USP7 knockout (USP7^−/−) cells have reduced TRIM27 protein levels. Cell lysates from the indicated HCT116 cells were immunoblotted. (C and D) Proteasome inhibitor (MG132) rescues TRIM27 protein levels upon USP7 knockdown. p53-proficient U2OS (C) and p53-deficient HeLa (D) cells were treated with the indicated siRNAs for 72 hrs, incubated with DMSO (vehicle) or MG132 for 4 hrs, and immunoblotted. (E) USP7 deubiquitinating enzymatic activity is critical for maintenance of TRIM27 protein levels. HeLa cells were treated with the indicated siRNAs for 24 hrs before expression of the indicated constructs for 48 hrs followed by immunoblotting for the indicated proteins. (F) USP7 deubiquitinates TRIM27 auto-ubiquitination in vitro. Purified proteins were used in in vitro ubiquitination reactions before samples were separated by SDS-PAGE and anti-TRIM27 immunoblotting was performed. Results are representative of at least three replicate experiments. See also Figure S3.

Figure 4. USP7 buffers MAGE-L2-TRIM27-induced WASH ubiquitination by deubiquitinating WASH

(A) TRIM27 3KR mutant is stable in the absence of USP7. HeLa cells were treated with the indicated siRNAs for 24 hrs before the indicated constructs were expressed for 48 hrs followed by immunoblotting. (B) TRIM27 3KR binds MAGE-L2. The indicated proteins were expressed in cells for 48 hrs before anti-Myc immunoprecipitation was performed and co-purifying proteins were detected by immunoblotting. (C) TRIM27 3KR mutant is still functional and can rescue TRIM27-RNAi-induced CI-M6PR mistrafficking. HeLa cells were treated with the indicated siRNAs for 24 hrs before expression of the indicated constructs for 48 hrs followed by anti-CI-M6PR immunostaining and quantification. (D) Knockdown of USP7 in cells reconstituted with TRIM27 3KR still abrogates CI-M6PR recycling, suggesting another critical activity of USP7. HeLa cells were treated with the indicated siRNAs for 24 hrs before expression of the indicated constructs for 48 hrs followed by anti-CI-M6PR immunostaining and quantification. Quantitation was performed only on transfected cells as marked by anti-HA (TRIM27) staining. (E and F) USP7 limits WASH-mediated endosomal Arp2/3 localization (E) and endosomal F-actin (F). HeLa cells were treated with the indicated siRNAs for 24 hrs before expression of the indicated constructs for 48 hrs followed by immunostaining and quantification. Quantitation was performed only on transfected cells as marked by anti-HA (TRIM27) staining. Endosomes were marked with anti-FAM21. (G) Knockdown of USP7 in TRIM27 3KR cells increases WASH ubiquitination levels. HeLa cells were treated with the indicated siRNAs for 24 hrs before expression of the indicated constructs for 48 hrs followed by anti-Myc (ubiquitin) immunoprecipitation and immunoblotting. (H) USP7 wild-type, but not inactive USP7 C233S, limits MAGE-L2-TRIM27-induced WASH ubiquitination (and TRIM27 auto-ubiquitination) in vitro. In vitro ubiquitination reactions were performed as indicated. Samples were separated by SDS-PAGE and immunoblotted for TRIM27 (left) or WASH (right). (I) Model for dual activities of USP7 in preventing TRIM27 from auto-ubiquitination-induced degradation, but also limiting WASH ubiquitination levels and activity through its deubiquitination. (J and K) U2OS cells were transiently transfected with the indicated constructs for 48 hrs before immunostaining and quantification of F-actin on FAM21-positive endosomes in GFP-positive cells (J) or quantification of CI-M6PR localization in GFP-positive cells (K). Results are representative of at least three replicate experiments. More than 100 cells were quantitated. Values shown are mean + SD. Asterisks indicates p<0.05. n.s. indicates not significant (p>0.05). See also Figure S4.

Figure 5. Mutation or deletion of USP7 impairs WASH-mediated protein recycling and occurs in children with a neurodevelopmental disorder

(A) Schematic illustrating USP7 genomic locus deletions or mutations in seven subjects with intellectual disability, autism spectrum disorder, and other neurodevelopmental phenotypes (see Table 1). (B) Partial knockdown of USP7 decreases TRIM27 protein levels (bottom) and impairs CI-M6PR recycling (top). HeLa cells were treated with the indicated siRNA concentrations for 72 hrs before anti-CI-M6PR immunostaining and quantification (top) or cell lysates were immunoblotted (bottom). Asterisks indicates non-specific band (bottom). (C and D) USP7^+/− HCT116 cells have defects in CI-M6PR recycling. The indicated HCT116 cells were subjected to anti-CI-M6PR immunostaining (C) and quantification (D). (E and F) USP7^+/− cells have decreased Arp2/3 (E) and F-actin (F) on VPS35-positive endosomes. The indicated HCT116 cells were subjected to immunostaining and quantification. (G) Other activities of USP7 are not significantly altered in USP7^+/− cells, such as alteration in p53, p21, or Bub3 levels. Cell lysates from the indicated HCT116 cells were separated by SDS-PAGE and immunoblotted. (H-J) Expression of subject #6 USP7 Y143X in HCT116 USP7^+/− cells does not rescue defects in CI-M6PR recycling (H) or accumulation of Arp2/3 (I) or F-actin (J) on FAM21-positive endosomes. The indicated HCT116 cells were transfected with GFP and wild-type (WT) or Y143X USP7 for 72 hrs before immunostaining and quantitation of GFP-positive cells. Results are representative of at least three replicate experiments. More than 100 cells were quantitated. Values shown are mean + SD. Asterisks indicates p<0.05. n.s. indicates not significant (p>0.05). Scale bar represents 20 μm. See also Figure S5.

Figure 6. MAGE-L2-TRIM27-USP7 plays critical role in the hypothalamus

(A and B) RT-QPCR analysis reveals that MAGE-L2 is highly enriched in the hypothalamus of both C57BL/6 (A) and BALB/c (B) mice. The indicated tissues were collected from six mice, pooled, and RT-QPCR analysis of MAGE-L2 expression was performed and normalized to 18S rRNA. (C) Expression of human MAGE-L2 is enriched in the brain. RNA from the indicated tissues were obtained and RT-QPCR analysis of MAGE-L2 expression was performed and normalized to 18S rRNA. (D) Knockdown of USP7 in hypothalamic neurons significantly decreases TRIM27 protein levels. Gt1-7 hypothalamic neurons were treated with the indicated siRNAs for 72 hrs before cell lysates were immunoblotted for the indicated proteins. (E-G) Knockdown of MAGE-L2 or USP7 in hypothalamic neurons impairs F-actin accumulation (E and F) and Arp2/3 endosomal localization (E and G). Gt1-7 hypothalamic neurons were treated with the indicated siRNAs for 72 hrs before immunostaining (E) and quantification (F and G). (H-J) CRISPR/Cas9 knockout of USP7 decreases TRIM27 protein levels (H) and levels of F-actin on FAM21-positive endosomes (I-J) levels. Gt1-7 hypothalamic neurons were transfected with GFP-Cas9 and the indicated gRNAs for 48 hrs before GFP-positive cells were collected by flow cytometry and analyzed by immunoblotting (H) and immunostaining (I-J) after culturing. Results are representative of at least three replicate experiments. More than 100 cells were quantitated. Values shown are mean + SD. Asterisks indicates p<0.05. Scale bar represents 20 μm. See also Figure S6.