Fig 2: OTUD4 interacts with RNA. (A) Poly(A)-RNA was purified with oligo(dT)-beads from HEK293T cells and co-purified OTUD4 protein was visualized by anti-OTUD4 western blotting (WB). RNase treatment (50 µg/ml) reduced the amount of OTUD4 collected on oligo(dT)-beads, while control treatment at 37°C did not. A representative result of four independent experiments is shown. (B) Recombinant GST–OTUD4 binds RNA. GST–OTUD4 protein or GST alone were coupled to glutathione–Sepharose and incubated with biotinylated RNA or without RNA. Binding was detected by incubation with fluorescently labeled streptavidin. The mean±s.e.m. of three independent experiments is shown. (C) RNA-immunoprecipitation (RNA-IP) leads to enrichment of RNA with OTUD4. Anti-FLAG IP was performed from lysates of control (vector)- or FLAG–OTUD4-transfected cells. The amount of isolated RNA with OTUD4 is shown in relation to RNA purified in the control samples. Shown is the mean±s.e.m. from n=6 experiments. (D) Detection of OTUD4 mRNA bound to OTUD4. RNA-IP was performed from FLAG–OTUD4-transfected HEK293T cells under stringent conditions. RNA was extracted from IgG-control beads or anti-FLAG beads, and equal RNA volumes were used for RT-PCR reactions. To avoid overamplification of transfected OTUD4 RNA, primers for the 3'UTR of OTUD4-mRNA were chosen, which is not part of the overexpression construct. RT-PCR was performed from two independent RNA-IPs. (E) Domain structure of OTUD4. Top panel: OTUD4 contains an ovarian tumor (OTU) domain that provides activity to cleave ubiquitin chains as well as a putative Tudor-like domain. PONDR-VSL2 (middle panel) and foldIndex (bottom panel) algorithms predict that large portions of OTUD4 are disordered or unfolded.

Fig 3: OTUD4 is required for correct stress granule formation. (A) Knockdown of OTUD4 decreases stress granule size and increases stress granule number. HeLa cells were transfected with two different siRNA oligonucleotides (oligo5 and oligo7) against OTUD4 or with control siRNA. OTUD4 and TIAR staining after 30 min arsenite treatment (0.5 mM) reveals differences in stress granule formation in the absence of OTUD4. White arrows indicate residual OTUD4 protein after knockdown, resulting in larger stress granules, resembling control siRNA-transfected cells. Scale bar: 10 µm. The experiment was repeated three times. (B) Quantification of the average number of stress granules per cell. Stress granules were scored in at least 200 cells per condition using CellProfiler software and the average number of granules per cell was depicted in a box plot. (C) Quantification of the average stress granule area. Granule area was determined with CellProfiler software as in B, and size distribution is presented as a box plot. (D) Re-introduction of siRNA-resistant OTUD4 rescues defects in stress granule formation. HeLa cells were transfected with FLAG–OTUD4 or its catalytic inactive mutant (C45A) 24 h after OTUD4 knockdown. Arsenite-treated cells (as in A) were stained with anti-OTUD4 and anti-TIAR antibodies. Scale bar: 20 µm. (E,F) Quantification of granule number and area in 55–100 FLAG–OTUD4-expressing cells per condition was performed as above. Wild-type and C45A-mutated OTUD4 reverse defects in granule formation. Cells with exogenous OTUD4 expression above endogenous levels were omitted for the analysis to exclude overexpression artifacts. In B,C,E,F, the box represents the 25–75th percentiles, and the median is indicated. The whiskers show the 1–99th percentiles and outliers are indicated. *P<0.05, **P<0.01, ***P<0.001 (B,E, quasi-Poissonian regression analysis; C,F, gamma regression analysis). (G) Analysis of changes in core stress granule proteins and apoptotic markers by western blotting. HeLa cells were transfected with OTUD4 siRNA or control siRNA with or without arsenite (30 min, 0.5 mM). Western blotting shows that G3BP1 and TIAR levels are unaffected by OTUD4 depletion. Depletion of OTUD4 leads to activation (cleavage) of caspase-3, as detected with anti-cleaved caspase-3 (Asp175) antibody and increased amounts of cleaved PARP, both indicating induction of apoptosis upon loss of OTUD4. The experiment was repeated three times. In D–G, oligo 7 was used for knockdown. (H) Knockdown of OTUD4 enhances the sensitivity of SH-SY5Y cells. SH-SY5Y cells were transfected with siRNA against OTUD4 or control siRNA as indicated. Cells were treated with arsenite (30 min, 0.5 mM) 24 h after siRNA transfection or left untreated. Cells were lysed 6 h later and levels of OTUD4, cleaved caspase-3 and actin were analyzed by western blotting. OTUD4-knockdown leads to increased caspase activation in response to arsenite treatment. The experiment was performed three times.

Fig 4: OTUD4 associates with polysomes and loss of OTUD4 impairs protein translation. (A) Loss of OTUD4 impairs protein translation. Control siRNA- or OTUD4 siRNA (oligo7, siOTUD4)-transfected HeLa cells were treated with caspase inhibitor z-VAD-fmk (24 h, 20 µM, starting 24 h after siRNA transfection) to control for caspase-related effects, or were left untreated. 10 µg/ml puromycin was added for 15 min to label newly synthesized proteins. Protein lysates were analyzed by immunoblotting for puromycylated proteins and levels of phosphorylated (P)-eIF2a, eIF2a, cleaved caspase-3, OTUD4 and actin. Anti-puromycin blot revealed lower levels of puromycin-labeled proteins in OTUD4 knockdown cells, indicating reduced translation in the absence of OTUD4. Phosphorylation of eIF2a was increased in OTUD4 knockdown cells. The observed effects were not blocked by caspase inhibition. A representative of at least three independent experiments is shown. (B) Puromycylation was performed for 10 or 15 min in control or siOTUD4 (oligo7)-transfected HeLa cells 48 h after transfection. Lane intensities were quantified using Fiji software and normalized to actin levels. The mean±s.e.m. of n=4 experiments is indicated. *P<0.05, **P<0.01 [paired t-test; P-values were 0.0219 (10 min puromycin) and 0.0036 (15 min puromycin)]. (C) The cellular mRNA content is not affected by OTUD4 knockdown. Poly(A)-RNA was extracted from HeLa cells transfected with control siRNA or OTUD4 siRNA. The mRNA content of siOTUD4-transfected cells is shown as fold level compared to that in control cells (set at 1). The mean±s.e.m. of n=3 experiments is shown. (D) OTUD4 is present in polysome-containing fractions. HeLa cells were either treated with arsenite (0.5 mM, 30 min), leading to polysome disassembly, or with cycloheximide (100 µg/ml, 15 min) to conserve mRNA–ribosome complexes. Lysates were subjected to sucrose gradient fractionation (15–50%) and individual fractions were divided for RNA preparation or for protein precipitation and subsequent immunoblot analysis. The approximate sedimentation behavior of polysomes and individual ribosomes, as judged by RNA profile and ribosomal protein S6 (RPS6) location, is indicated. Upon arsenite treatment, OTUD4 was found in fractions 5–9, while under cycloheximide conditions, it was also found in the putative polysome-containing fractions (2–4). For comparison, protein profiles of FMRP, known to associate with polysomes, RPS6 as ribosomal marker and GAPDH as a protein presumably unrelated to translation, are shown. A representative of three experiments is shown.

Fig 5: OTUD4 interactome analysis places OTUD4 in a network of stress granule-associated proteins and RBPs. (A) Identification of putative OTUD4-interacting proteins from mouse cerebellum and cortex lysate by mass spectrometry. Purified HA–OTUD4 or a HA-tagged control peptide were used as bait to pulldown interacting proteins from mouse brain lysates. Potential interactors were identified by mass spectrometry. An enrichment index for hits from at least two out of three experiments (per tissue) was calculated vs controls. Proteins with more than 30-fold enrichment over controls were considered as potential OTUD4 interactors. 290 proteins were identified in the cortex samples and 298 in the cerebellum samples with 133 proteins found in both tissues (Tables S1 and S2). 40% of these have been found in stress granules in previous studies (Jain et al., 2016; Markmiller et al., 2018) and 68% in mRNA interactomes (Beckmann et al., 2015). (B) Gene ontology analysis with the PANTHER GO slim tool (molecular function) revealed a strong enrichment of RNA-binding-related terms for the OTUD4-interacting proteins. Shown is the fold enrichment of terms compared to the mouse reference genome; depicted are all terms with an enrichment factor >5. The false discovery rate (FDR) for each of the terms is indicated in the respective bar. The analysis was performed for 298 potential interactors from cerebellum. (C–G) Co-IP experiments to check interactions of putative binding partners with OTUD4. (C) Endogenous OTUD4 was precipitated from HeLa cell lysates. Lysates were incubated at 37°C for 15 min in the absence or presence of 50 µg/ml RNase A. Co-precipitation of endogenous SMN1 was detected by western blotting. (D–G) HEK293T cells were transfected as indicated and lysates treated as above. Tagged OTUD4 was precipitated. Co-purification of exogenously expressed HuB (D) and IGF2BP3 (E) was confirmed, as well as of cellular G3BP1 (F) and TIAR (G). Treatment with RNase A (third lane of each panel) reduced or abolished the interaction with SMN1, HuB and G3BP1, demonstrating that these interactions are RNA dependent. In contrast, TIAR and IGF2BP3 seem to bind in an RNA-independent manner (* denotes cross-reaction of anti-TIAR-antibody with GFP). A representative of at least three independent experiments is shown for each IP.