Engineering an orthogonal ubiquitin transfer cascade with RING E3 RNF38 by phage display to reveal its regulation of nuclear transport

TL;DR

Scientists engineered a system to identify proteins targeted by RNF38, a ubiquitin ligase, revealing its role in regulating nuclear transport and other cellular processes.

Contribution

A novel phage display-based method was developed to identify RNF38 substrates and their roles in cellular functions.

Findings

RNF38 targets proteins involved in nuclear transport, protein translation, and endosomal sorting.

RNF38 ubiquitination leads to degradation of substrates and inhibits nuclear translocation of E2F1 and p-STAT3.

Phage selection identified key residues for E2-E3 interactions, aiding in engineering orthogonal ubiquitin transfer systems.

Abstract

RING E3 ubiquitin (UB) ligases rely on signature RING domains for mediating UB transfer to substrate proteins. The large number of RING E3s and their weak association with substrates pose a significant challenge in identifying the substrates of individual E3s, thereby hindering the elucidation of their biological functions. Here, we utilized phage display to engineer an "orthogonal UB transfer" (OUT) cascade with RING E3 RNF38, enabling the exclusive transfer of an engineered UB (xUB) to its substrates in the cell. The OUT screen revealed RNF38 substrates regulating nucleocytoplasmic transport (Ran, RanGAP1, and KPNA2), protein translation (HuR and Rack1), and endosomal sorting (VPS35). Furthermore, RNF38-catalyzed ubiquitination was found to induce the degradation of the substrate proteins and negatively affect the translocation of transcription factors E2F1 and phosphorylated STAT3 (p-STAT3) into the nucleus. Phage selection of the RNF38 RING library also revealed hotspot residues for E2 interaction, which may guide the engineering of orthogonal E2-E3 pairs with other RING E3s. Overall, our work discovered new roles of RNF38 in regulating nuclear transport and established an anchoring point for expanding OUT cascades within the large family of RING E3s for revealing their UB transfer targets and cellular functions.

Figures

Validation of RNF38-catalyzed ubiquitination of substrate proteins identified by the OUT cascade in cells. RNF38 was overexpressed in HEK293 cells with an increasing transfection of the RNF38 expression plasmid. The cells were treated with proteasome inhibitor MG132 to accumulate ubiquitinated proteins, and the designated substrates were immunoprecipitated from the cell with specific antibodies for probing their ubiquitination levels on the western blots with an anti-UB antibody. Bands designated with a star on the western blots correspond to the size of the IgG heavy chain. *A*, increasing ex

Full text

E3 ubiquitin (UB) ligases recruit cellular targets for conjugation with the 76-residue UB protein and regulate a broad band of eukaryotic biology, from cell development, division, and proliferation to metabolism, stress response, and apoptosis (1, 2). E3s are supported by UB-activating enzymes (E1) and UB-conjugating enzymes (E2) that form thioester conjugates with UB for its delivery to the substrate proteins (3, 4, 5). Depending on the domain structures embedded in the E3 to engage the UB-E2 conjugate and catalyze substrate ubiquitination, E3s are categorized into the HECT, RING, RING-between-RING (RBR), and U-box subtypes, with the HECT and RBR E3s forming a thioester intermediate with UB to facilitate its transfer to the substrates, while the RING and U-box E3s bridge the direct transfer of UB from the E2∼UB conjugate to the substrate proteins (6, 7, 8, 9). The RING E3s constitute the largest E3 subclass, with more than 600 predicted members encoded in the human genome (9, 10). They may function as monomers, homodimers, heterodimers, or oligomers or be integrated into multimodular E3 complexes such as the cullin RING ligases (CRLs) and anaphase-promoting complex/cyclosome (APC/C) to catalyze substrate ubiquitination (1, 11, 12, 13, 14). The large number of E3s and their transient interactions with substrates present a significant challenge to identifying the substrates of a designated E3 (15). Additionally, since E3s cross-regulate each other's activity and stability, perturbing the expression or enzymatic activity of one E3 may impact ubiquitination pathways regulated by other E3s (16). Consequently, assigning the direct ubiquitination targets of an E3 has been technically challenging and so is the precise mapping of E3s onto signaling networks to elucidate their roles in cell regulation and disease development. Over the years, many methods have been developed for E3 substrate profiling based on affinity purification of substrate proteins with E3 as the bait (17, 18, 19), detecting the change in ubiquitination level or the stability of the substrates coupled with the perturbation of E3 expression or activities (20, 21, 22, 23, 24, 25, 26), constructing E3 fusions with UB, Nedd8 or UBA domains for trapping E3 substrates, and fusing the E3 with biotin ligase for labeling the UB-tagged substrates with biotin (27, 28, 29, 30, 31, 32). These methods were instrumental in revealing the fundamental roles of the E3s in cell regulation. Still, the diverse classes of E3s call for the expansion of the technical platforms to study E3 functions by first identifying their ubiquitination targets with precision and fidelity.

We recently developed a method known as "orthogonal UB transfer (OUT)" to identify E3 substrates by following the elusive delivery of an engineered UB (xUB) through an engineered xE1-xE2-xE3 cascade to the substrates of a specific E3 in the cell ("x" designates engineered UB or enzyme variants orthogonal to their native partners) (33). We previously generated OUT cascades with HECT E3s E6AP/UBE3A and Rsp5, U-box E3s CHIP and E4B, and RBR E3 Parkin to profile their substrate specificities (34, 35, 36, 37, 38, 39). Here, we report the engineering of RING E3 RNF38 to enable the specific transfer of xUB through the engineered xRNF38 to its substrates for their identification by proteomics. The establishment of the OUT cascade with RNF38 will guide the engineering of other RING E3s to expand the "OUT-reach" of the engineered UB transfer cascade for revealing the mechanisms of action of diverse classes of E3s in the cell.

The OUT cascade eliminates the complex cross-reactions among various E2s and E3s and assigns E3 substrates by directly tracing xUB transfer from the E3 to its target proteins (33). To guarantee the orthogonality between the OUT and the native cascade for UB transfer, we identified a UB mutant (R42E and R72E) that is rejected by the wild type (wt) E1 (Uba1 and Uba6) and used it as the xUB for labeling the substrates of a specific E3. To enable xUB transfer through the OUT cascade, we identified mutations in the UB-binding site of Uba1 (Q576R, D591R, and E594R) that would restore the binding and activation of xUB. We also introduced mutations to the E2-binding site in Uba1 (E1004K, D1014K, and E1016K) to reject the binding of wt E2s and used phage display to engineer the N-terminal helix of E2 enzymes UbcH5b and UbcH7 to match the mutations in Uba1 and restore the transfer of xUB. In this way, we generated xUba1-xUbcH5b and xUba1-xUbcH7 pairs for the exclusive delivery of xUB to a specific E2 (40). The N-terminal helix of E2 is a key element for recognition by different classes of E3s, and mutations in xE2 (K4E and K8E in xUbcH5b, K5E and K9E in xUbcH7) would interfere with the transfer of xUB to the wt E3s (37, 38). In this study, we engineered the RING domain of RNF38 by phage display to establish its interaction with the engineered xUbcH5b to construct the OUT cascade. We then used the OUT cascade of RNF38 to profile its ubiquitination targets in the cell. Due to the availability of the crystal structures of the RNF38 RING in complex with E2∼UB conjugates for guiding the engineering of the RING domain (41, 42), we chose RNF38 as a prototype to construct an OUT cascade with RING E3s.

RNF38 is an E3 UB ligase with a C-terminal RING domain, which binds the E2∼UB conjugate to mediate UB transfer to its substrates (43). RNF38 was found to be mainly located in the nucleus (43), and there has been limited study of its cellular functions besides its regulation of cell proliferation and association with cancer metastasis. It was found that RNF38 induces the ubiquitination and degradation of neuroblast differentiation-associated protein AHNAK to reverse its suppression of the TGF-β pathway and promote cell growth (44). RNF38 also ubiquitinates protein tyrosine phosphatase 1 (SHP-1) to activate signal transducer and activator of transcription 3 (STAT3) and cell division (45). The role of RNF38 in promoting cell proliferation aligns with its overexpression observed in several types of cancers, such as gastric cancer, hepatocellular carcinoma, and non-small cell lung cancers (44, 45, 46). On the other hand, RNF38 has been found to ubiquitinate α-actinin-4 (ACTN4), an actin-binding protein, and LIM domain-binding protein 1 (LDB1), a transcription factor, for suppressing cancer cell proliferation (47, 48). RNF38 has also been found to ubiquitinate transcription factors, including p53 and Runt-related transcription factor 1 (RUNX1) to inhibit gene activation (43, 49). Here, we engineered the OUT cascade of RNF38 to profile its substrates in HEK293 cells and identified its role in inducing the ubiquitination and degradation of numerous substrates, such as GTP-binding nuclear protein Ran, Ran GTPase-activating protein 1 (RanGAP1) and importin subunit alpha-1 (KPNA2) associated with nucleocytoplasmic transport, ELAV-like RNA-binding protein 1 (ELAVL1/HuR) and small ribosomal subunit protein Rack1 associated with protein translation, and vacuolar protein sorting-associated protein 35 (VPS35) in the endosome sorting and recycling pathways. Based on the newly established link of RNF38 with the components of the nuclear transport machinery, we confirmed the role of RNF38 in regulating the translocation of transcription factors, such as E2F1 and phosphorylated STAT3, into the nucleus. Moreover, our work establishes a new anchoring point for expanding the OUT cascade within the broad class of RING E3 ligases to elucidate their ubiquitination targets and biological functions.

Results

Design of a RING domain library of RNF38 for the engineering of an orthogonal xUbcH5b-xRNF38 pair

We first tested the reactivity of wt RNF38 and its RING domain in xUB transfer by combining them with the xUba1-xUbcH5b pair and found that neither the full-length RNF38 nor the RING domain was active in the self-ubiquitination reaction with xUB. In contrast, they showed strong activity with the wt UB transferred through the wt Uba1-UbcH5b pair (Fig. S1A). Also, while wt RNF38 could be supported by wt Uba1-UbcH5b to ubiquitinate p53 with wt UB, it could not transfer xUB to p53 by connecting with the xUba1-xUbcH5b pair (Fig. S1B). This suggested that the mutations in the N-terminal helix of xUbcH5b might block its interaction with the RING E3 to mediate xUB transfer. We thus needed to engineer the E2-binding interface of the RNF38 RING domain to restore its activity with xUbcH5b.

The crystal structure of the RING domain of RNF38 in complex with the UbcH5b∼UB conjugate suggests UbcH5b could be one of the E2 partners of RNF38 and reveals key interactions between loop1 of the RING domain and the N-terminal helix of UbcH5b that harbors the K4E and K8E mutations in the engineered xUbcH5b (Fig. 1, A and B) (41). K4 of UbcH5b is near C418 within the loop1 of the RING domain, and K8 of UbcH5b is surrounded by L412, M417, C418, and D419 in loop1 with a contacting distance (2.8 Å) between the amino group of K8 and the C418 thiol of the RING. Another structure of the RNF38 RING in complex with the UBE2K (E2-25K)∼UB conjugate also showed the loop1 of the RING with the sequence ^412^LCVVCMCD^419^ engaging the N-terminal helix of UBE2K with similar interactions (42). We thus constructed a RING domain library of RNF38 by randomizing loop1 residues L412, M417, C418, and D419 to select RING mutants with restored reactivity with the xUB-xUbcH5b conjugate for transferring xUB to RNF38 substrates. For constructing the RING domain library, we used the NNK ((N = A, C, G or T; K = G or T) codons to replace the original codons of the residues to be randomized and cloned the randomized RNF38 RING gene into the pComb3H vector for displaying the RING library on the surface of M13 phage (50, 51).Figure 1**Phage selection for engineering the RINGdomain of RNF38 to construct the OUT cascade.**A, crystal structure of RNF38 RING domain in complex with the UbcH5b∼UB conjugate (PDB ID 4V3L) (41) showing loop1 of the RING domain interacting with the N-terminal helix of UbcH5b. B, zoomed-in view of the RING-UbcH5b interaction shown in A with K4 and K8 that were mutated to Glu in xUbcH5b interacting with loop1 residues L412, M417, C418, and D419 of the RNF38 RING. C, selection of RING mutant based on the catalytic transfer of biotin-wt UB from xUbcH5b to the RING domain library displayed on phage. D, phage ELISA assay showed the labeling of wt RNF38 RING domain displayed phage by biotin-wt UB after reacting the phage with the wt Uba1-UbcH5b pair. The ELISA signals (OD 450 nm) were plotted in the panel on the right. R1, phage labeling reaction with biotin-wt UB, wt Uba1 and wt UbaH5b. C1, C2, and C3, control reactions missing either Uba1, UbcH5b, or biotin-wt UB. E, alignment of the RNF38 RING domain sequences after five rounds of phage selection, showing the convergence of the library clones. RING domain residues L412, M417, C418, and D419 that were randomized in the library were designated with red stars.

Phage selection of the RING domain library for constructing the OUT cascade with RNF38

We previously used phage display to engineer the U-box domain of E4B and CHIP to identify mutations at their E2-binding interface for restoring their interactions with the engineered xUbcH5b (38). We thus explored the display of the RNF38 RING domain on M13 phage for RING library selection. We cloned the wt RING domain into the pComb3H vector for its expression as an N-terminal fusion with a truncated pIII protein that would anchor the RING domain on the phage capsid (Fig. 1C). This allowed the transfer of biotin-labeled UB from xUbcH5b to the RING domain in a self-ubiquitination reaction. If there was productive interaction between xUbcH5b and any of the RING mutants, the RING domain would be labeled with biotin-UB and the corresponding phage particles selected by binding with a streptavidin-immobilized plate. The phage bound to the streptavidin plate could then be released by the TEV protease that would cleave a peptide linker harboring the TEV cleavage site between the RING domain and pIII in the fusion protein. We found the wt RING domain of RNF38 could be efficiently displayed on the M13 phage by probing the Flag-RING-pIII fusion with an anti-DDDDK antibody on the western blot (Fig. S1C). When we probed the phage reaction with HA-UB, Uba1 and UbcH5b with an anti-DDDDK antibody, we found the intensity of the Flag-RING-pIII band significantly decreased, while not detecting the UB conjugates of the fusion protein (Fig. S1C). This suggests that the Flag-RING-pIII fusion was reactive with the Uba1-UbcH5b pair in the ubiquitination reaction, but UB-conjugated fusion proteins could not be detected by the anti-DDDDK antibody, likely due to interference from the conjugated UB molecules with the antibody binding to the Flag tag. Nevertheless, the ELISA assay showed that the RING domain-displayed phage was labeled with biotin due to the transfer of biotin-wt UB to the RING domain through the wt Uba1-UbcH5b pair, suggesting the phage displayed RING domain was reactive in the UB transfer reaction (Fig. 1D).

To assay the selection efficiency for the phage displaying a catalytically active RING domain, we mixed phage displaying the RNF38 RING and the SV5V virus protein (52) with no E3 ligase activity at various ratios of 1/1, 1/10, and 1/100, and reacted the phage mixture with biotin-UB and the wt Uba1-UbcH5b pair. After the reaction, the phage mixture was selected by binding to a streptavidin plate. The selected phage population was released by TEV cleavage, and the phage identities were revealed by PCR reactions of individual E. coli colonies after phage infection. The colony PCR showed that nine out of ten colonies from the selection of the 1/1 and 1/10 mixed phage library were infected with RING domain-displayed phage, and three out of ten colonies from the selection of the 1/100 phage library were infected with RING domain-displayed phage (Fig. S2A). So approximately, each round of phage selection could enrich catalytically active RING domain phage by 30-fold.

We then carried out phage selection for the catalytically active RING domains that can pair with xUbcH5b to mediate the transfer of xUB. Since xUB has the R42E and R72E mutations that might complicate the selection for RING mutants to restore their interaction with xUbcH5b, we used a Uba1 mutant, xUba1(UFD), with a wt UB binding site but with an engineered E2 binding site in the UB-fold domain (UFD) to load biotinylated wt UB on xUbcH5b (40). This was followed by the addition of the RING domain library to the reaction mixture, enabling the transfer of biotin-wt UB to RING mutants displayed on phage with restored interactions with xUbcH5b. The phage reaction mixture was then selected by binding to the streptavidin plate, and the selected phage was eluted by TEV (Fig. 1C). A total of five rounds of phage selection were carried out with elevated stringency by decreasing the amount of phage input, the concentration of biotin-wt UB, xUba1(UFD), and xUbcH5b, and the reaction time in the following rounds. We observed an increasing titer of phage from the selection reaction compared to the controls with the elimination of either xUba1(UFD), xUbcH5b, or biotin-wt UB (Fig. S2B). This suggests the phage selection reaction depended on the catalytic transfer of UB to the RING domain anchored on the phage capsid. The selected phage clones were sequenced after each round of selection. Convergence of the RING domain sequence at the randomized sites was observed after the fifth round of selection, with the LZ1 mutant (L412M, M417L, C418K, and D419E) appearing eight times among the 20 sequenced clones, dominating the selected library (Fig. 1E). Other clones, such as LZ2-4, bore homologous mutations as LZ1 with L or M residues selected at L412, L at M417, R or K at C418, and D or E at D419. The crystal structure of the RNF38 RING in complex with UbcH5b ∼ UB showed C418 in loop1 of the RING is juxtaposed with the K8 residue in the N-terminal helix of UbcH5b (Fig. 1B). So, the K8E mutation in xUbcH5b would favor R or K of the opposite charge, replacing C418 of the RING domain to restore xUbcH5b-RING interaction. K4E is another mutation in the N-terminal helix of xUbcH5b and is also in the vicinity of C418, so mutating C418 to positively charged R or K may favor the interactions with the K4E mutation in xUbcH5b. The K8 residue of UbcH5b is also surrounded by L412, M417, and D419 that were randomized in the RING domain library, but the K8E mutation in xUbcH5b did not induce significant changes at these sites—the selected RING mutants mainly bore M/L at L412, L at M417 and D/E at D419 (Fig. 1E). Still, the selected residues converged at these sites, suggesting these residues still played a significant role in mediating the interactions between the N-terminal helix of E2 and loop1 of the RING domain. However, such interactions were not charge-based, as no positively charged residues were selected at these sites to match the charge-reversed K4E and K8E mutations in xUbcH5b. To verify the reactivity of individual RING mutants on the phage surface, we separately prepared phage displaying LZ1-4 and followed the phage selection conditions to react them with biotin-wt UB, xUba1(UFD), and xUbcH5b. Phage ELISA showed that all individual phage clones were labeled with biotin-wt UB, demonstrating the restored interaction between the RING mutants and xUbcH5b to mediate UB transfer. The phage clones of LZ1 and LZ3 showed the highest level of biotin-UB labeling, suggesting they were more active in pairing with xUbcH5b in the ubiquitination reaction (Fig. S2C).

Constructing the OUT cascade of RNF38 based on the phage-selected RING domain

We expressed the selected RING mutants LZ1-4 from E. coli with a GST tag and assayed their activities with xUbcH5b following the phage selection condition that used the xUba1(UFD)-xUbcH5b pair to transfer wt UB to the RING domain. We found all the mutants were active with xUbcH5b in the transfer of wt UB as suggested by the anti-GST western blot that detected the self-ubiquitination of the RING mutants (Fig. 2A). Among all the RING mutants assayed, LZ4 showed the weakest activity, suggesting the L412K and M417Y mutations in LZ4 were not so effective in restoring the interaction of the RING domain with xUbcH5b as the more converged L412M and M417L mutations selected in LZ1-3 (Fig. 1E). We then assayed the activity of the LZ1-4 RING mutants in mediating the transfer of xUB through the xUba1-xUbcH5b pair. LZ1 showed the strongest activity in the self-ubiquitination reaction with xUB supported by the xUba1-xUbcH5b pair compared with other RING mutants (Fig. 2B). This confirms the success of the phage selection reaction in engineering loop1 of the RING domain for E2-binding to restore its interaction with xUbcH5b for xUB transfer. Next, we incorporated the mutations selected for LZ1 into the full-length RNF38 to assay its capacity in the transfer of xUB. RNF38 with LZ1 mutations was able to efficiently transfer xUB in the self-ubiquitination reactions upon its assembly with the xUba1-xUbcH5b pair (Fig. 2C). Furthermore, the RNF38 mutant could transfer xUB to its known substrate p53 (43), suggesting its capacity as an xE3 in the OUT cascade for delivering xUB to RNF38 substrates (Fig. 2D). We could thus use the engineered RNF38 with LZ1 mutations as xRNF38 to assemble an OUT cascade to profile E3 substrates in the cell.Figure 2**Activities of engineered RINGdomains and xRNF38 in xUB transfer.**A, self-ubiquitination of phage-selected RING mutants LZ1-4 in reactions with biotin-wtUB, xUba1(UFD), and xUbcH5b following the conditions of phage selection. B, activities of RING mutants LZ1-4 in self-ubiquitination reactions with xUB transferred from the xUba1-xUbcH5b pair. C, the activity of full-length xRNF38 with engineered mutations from LZ1 in the self-ubiquitination reaction with xUB and the xUba1-xUbcH5b pair. D, xRNF38-catalyzed p53 ubiquitination reaction with xUB and the xUba1-xUbcH5b pair.

Profiling RNF38 substrates with the OUT cascade

The construction of the OUT cascade of RNF38 allowed us to profile its substrates by tracing the exclusive transfer of xUB from xRNF38 to the substrate proteins. We previously screened HEK293 cells stably expressing the xUba1-xUbcH5b pair (38). We thus transfected the cells with pLenti plasmids carrying the genes of engineered xRNF38 and xUB with the 6×His and biotin carboxyl carrier protein (BCCP) tags at the N-terminus (HBT-xUB) and confirmed the expression of HBT-xUB and all components of the OUT cascades in the cell (Fig. 3A, left panels). To assay the transfer of xUB to all the OUT components, we purified HBT-xUB conjugated proteins using the Ni-NTA resin and verified the presence of Flag-xUba1, V5-xUbcH5b, and GFP-xRNF38 among the purified proteins. This confirmed that the OUT cascade of RNF38 was capable of transferring HBT-xUB through the cascade enzymes (Fig. 3A, right panels). We thus designated the cells expressing the full OUT cascade of RNF38 as OUT cells for purifying and identifying RNF38 substrates. As a control, we transfected HBT-xUB expressing plasmids into the cells stably expressing the xUba1-xUbcH5b pair without xRNF38 to identify proteins with xUB modification in the absence of xRNF38. We purified xUB-conjugated proteins in parallel from OUT and control cells with tandem affinity purification by Ni-NTA and streptavidin resin and performed trypsin digestion on proteins bound to the streptavidin resin for their identification by proteomics (Fig. 3B). We carried out three repeats of affinity purification and proteomic identification from the OUT and control cells to acquire datasets to reveal potential RNF38 substrates with an average PSM ratio > 2 (Log_2_ [PSM ratio OUT/control] > 1) between the purification from the OUT and control cells and a p value < 0.1 (-Log_10_p > 1.0) (Figs. 3C and S3 and Table S1). We found 150 target proteins that fell into this category, as shown in the volcano plot, and assigned them as potential RNF38 substrates (Fig. 3C). Gene Ontology analysis revealed significant biological functions of potential RNF38 substrates from the OUT screen related to protein translation, intracellular protein transport, and nuclear RNA processing, and these proteins are associated with oncogenic pathways for the development of stomach and breast cancers, matching the known roles of RNF38 in cancer pathogenesis (44, 45, 47) (Figs. S4 and S5 and Tables S2 and S3). So far, RNF38 has been found to ubiquitinate a handful of substrates in various cell types, including AHNAK in hepatocellular carcinoma cells, SHP-1 and p53 in HEK293 cells, ACTN4 in nasopharyngeal carcinoma cells, LDB1 in colorectal cancer cells, and RUNX1 in human leukemia K562 cells (43, 44, 45, 47, 49). Among the reported substrates of RNF38, the OUT screen identified ACTN4 as a high-fidelity RNF38 substrate in the HEK293 cells (Fig. 3C and Table S1). The other reported substrates did not appear in the OUT screen, and this might be due to the difference in cell culture conditions or proteome composition between various cell types.Figure 3**Identifying the substrates of RNF38 in HEK293 cells by the OUT cascade.**A, expression of the OUT-cascade components Flag-xUba1, V5-xUbcH5b, GFP-xRNF38 for the transfer of HBT-xUB in HEK293 cells. The expression of the cascade enzymes in OUT cells with the full xUba1-xUbcH5b-xRNF38 cascade and control cells excluding xRNF38 was verified by western blots of the cell lysates probed with antibodies specific for the tags on each component (left panels). The formation of HBT-xUB conjugates with xUba1, xUbcH7, and xRNF38 was confirmed by purifying the xUB-conjugates in the cell lysate by Ni-NTA column and detecting each component with specific tags (right panels). B, tandem purification of xUB-conjugated proteins from OUT and the control cells. Lane assignments: 1. cell lysates before Ni-NTA binding; 2. flow-through of the cell lysates after Ni-NTA binding; 3. wash of the Ni-NTA beads; 4. elution from the Ni-NTA beads; 5. flow-through from the streptavidin beads; 6. wash of the streptavidin beads; 7. proteins bound to the streptavidin beads. The western blot was probed with an anti-UB antibody to reveal ubiquitinated species enriched by tandem purification. C, Volcano plot of RNF38 substrates identified by the OUT screen. N = 3 independent biological replicates. Red dots designate proteins with Log_2_[PSM ratio OUT/control] >1 and -Log_10_p > 1. D, heatmap with normalized PSM values for a short list of potential RNF38 substrates from the OUT screen, including RanGAP1, ELAVL1/HuR, Ran, Rack1, VPS35, and KPNA2, that were further characterized in this study (names in red). P1-P3: 3 replicates from the OUT cells expressing the OUT cascade of RNF38; C1-C3, 3 replicates from the control cells expressing xUba1-xUbcH5b without xRNF38. E, protein–protein interaction network generated by STRING for RNF38 substrates associated with the intracellular and nuclear transport pathways.

Verification of RNF38 substrates by reconstituted ubiquitination assays in vitro and in the cell

Since the OUT screen revealed the association of RNF38 substrates with intracellular and nuclear transport pathways, we chose to verify RNF38-catalyzed ubiquitination of Ran and RanGAP1, two key nuclear transport proteins (53) with highly significant -Log_10_p values of 1.28 and 1.76, respectively (Fig. 3, D and E). KPNA2 has a -Log_10_p of 0.995, and it functions as a cargo adaptor for nuclear transport (54). We thus verified its ubiquitination by RNF38. The OUT screen also suggested the role of RNF38 in regulating protein translation. Among the potential substrates identified by the OUT screen, we picked HuR/ELAVL1 and Rack1 associated with protein translation pathways for verification (55, 56), and they have a −Log_10_p of 1.30 and 1.06, respectively. In addition, we assayed the RNF38-catalyzed ubiquitination of VPS35 (-Log_10_p = 1.04), a component of the retromer complex for protein sorting and recycling through the endosome (57). In vitro reactions of the substrates with wt Uba1, UbcH5b, and RNF38 for wt UB transfer confirmed the ubiquitination of the substrate proteins catalyzed by RNF38 with RanGAP1, KPNA2, HuR, and VPS35, showing polyubiquitinated bands of the substrates on the western blots probed with antibodies against Flag or GFP tags appended to the substrates (Fig. 4). Ran and Rack1 mainly formed monoubiquitinated species in RNF38-catalyzed ubiquitination reactions. We then evaluated the role of RNF38 ubiquitination of these substrates in HEK293 cells by purifying each substrate with specific antibodies and measuring their ubiquitination levels with an anti-UB antibody (Fig. 5). Cells transfected with plasmids for RNF38 and tagged substrates were treated with the proteasome inhibitor MG132 to inhibit the degradation of ubiquitinated proteins. We first assayed the ubiquitination of known RNF38 substrates ACTN4 (−Log_10_p = 1.04 from the OUT screen) and p53 (43, 47) and found they both showed enhanced ubiquitination with increasing levels of RNF38 expression (Fig. 5, B and C). We then assayed the newly identified substrates, including Ran, RanGAP1, KPNA2, HuR, Rack1, and VPS35, and found they all showed substantial increases in ubiquitination upon RNF38 expression (Fig. 5, D–I). Cumulatively, our results demonstrated that RNF38 recognized substrates identified by OUT as direct ubiquitination targets in the cell.Figure 4In vitro ubiquitination of RNF38 substrates in reconstituted reactions.A–F, reactions were set up with ATP and wt Uba1, UbcH5b, and RNF38 for the transfer of wt UB to RNF38 substrates with various tags for detection - Flag-Ran, Flag-RanGAP1, KPNA2, GFP-HuR, Rack1, and VPS35. In control reactions, each component of the UB transfer cascade of RNF38 was excluded. Lane assignments: 1, reaction of the substrate protein and the Uba1-UbcH5b-RNF38 cascade for the transfer of wt UB to the substrates; 2, reaction missing Uba1 as the E1; 3, reaction missing UbcH5b as the E2; 4, reaction missing RNF38 as the E3; 5, reaction missing wt UB.Figure 5Validation of RNF38-catalyzed ubiquitination of substrate proteins identified by the OUT cascade in cells. RNF38 was overexpressed in HEK293 cells with an increasing transfection of the RNF38 expression plasmid. The cells were treated with proteasome inhibitor MG132 to accumulate ubiquitinated proteins, and the designated substrates were immunoprecipitated from the cell with specific antibodies for probing their ubiquitination levels on the western blots with an anti-UB antibody. Bands designated with a star on the western blots correspond to the size of the IgG heavy chain. A, increasing expression of RNF38 in transfected HEK293 cells. B and C, increasing expression of RNF38 led to enhanced ubiquitination of known RNF38 substrates ACTN4 and p53. D–I, verifying the enhanced ubiquitination of OUT-identified RNF38 substrates, including Ran, RanGAP1, KPNA2, HuR, Rack1, and VPS35, with increasing levels of RNF38 expression.

Regulation of substrate degradation by RNF38 in the cell

After confirming RNF38-catalyzed substrate ubiquitination in vitro and in cells, we assessed the effect of RNF38 on the stability of the newly identified substrates. Overexpression of RNF38 in the cell decreased the steady-state levels of its substrates, including Ran, RanGAP1, HuR, KPNA2, Rack1, and VPS35, suggesting that RNF38 may regulate the stability of these proteins (Fig. 6A). We also monitored the stability of the substrate proteins after treating the cells with the protein synthesis inhibitor cycloheximide (CHX). Under these conditions, enhanced expression of RNF38 in the cell accelerated the degradation of the substrate proteins, with Ran, KPNA2, and Rack1 exhibiting a significant destabilization effect coupled with RNF38 expression and HuR, RanGAP1 and VPS35 showing a moderate effect (Figs. 6B and S6). So RNF38-catalyzed substrate ubiquitination may have varying effects on inducing the degradation of the substrate proteins. As a control, ACTN4 underwent accelerated degradation with enhanced expression of RNF38 (Fig. S6). We have thus confirmed new ubiquitination pathways regulating the stability of the cellular targets through the OUT screen of RNF38 substrates.Figure 6**Regulation of the substrate stability in the cell by RNF38.**A, enhanced expression of RNF38 in the cell suppressed the level of the substrate proteins identified by OUT. HEK293 cells were transfected with increasing amounts of RNF38 expression plasmid, and the levels of the substrate proteins in the cell were assayed by substrate-specific antibodies. The correlation of decreased substrate levels with increased RNF38 expression was shown in the companion plot. B, RNF38 expression in HEK293 cells accelerated the degradation of the substrate proteins in the cell. RNF38 expression plasmid was transfected into the HEK293 cells, and cells were treated with cycloheximide (CHX) to inhibit protein expression. The degradation of the RNF38 substrates was followed by substrate-specific antibodies in cells with enhanced expression of RNF38 and compared with background HEK293 cells. The degradation of Ran, KPNA2, HuR, and Ran at different chasing times after the addition of CHX was plotted in the panels on the right. The plots for the degradation of ACTN4, RanGAP1, and VPS35 were shown in Fig. S6.

Regulation of nuclear transport by RNF38

The verification of RNF38 substrates constituting nuclear transport machinery, including Ran, RanGAP1, and KPNA2, prompted us to assay the effect of RNF38 on protein trafficking into the nucleus. We expressed RNF38 in HeLa cells and found that elevated RNF38 expression suppressed the levels of Ran, RanGAP1, and KPNA2, as well as the cargo protein HuR, both in the cytoplasm and in the nucleus (Fig. S7). This matched the role of RNF38 in catalyzing the ubiquitination and inducing the degradation of these proteins, as was found in HEK293 cells (Fig. 6). Next, we assayed whether RNF38 would affect the nuclear transport of established KPNA2 cargo proteins that were not identified as RNF38 substrates by the OUT screen. The transcription factor E2F1 relies on KPNA2 for its translocation into the nucleus (58). In addition, once STAT3 is activated by phosphorylation, it is imported by KPNA2 into the nucleus to function as a transcription activator (59). Neither E2F1 nor STAT3 was in the substrate profile of RNF38 identified by OUT (Fig. S3 and Table S1). However, the expression of RNF38 in HeLa cells significantly reduced the levels of E2F1 and phosphorylated STAT3 (p-STAT3) in the nucleus, while cytoplasmic levels of these proteins were not significantly altered, as shown by the western blots following fractionation of the cytoplasm and nucleus proteins (Fig. 7, A and B). Furthermore, fluorescence imaging of the cells demonstrated that enhanced expression of RNF38 decreased the levels of E2F1 and p-STAT3 in the nucleus relative to their levels in the cytoplasm, suggesting a reduced import of these cargo proteins into the nucleus (Fig. 7, C–E). These results suggest a role for RNF38 in regulating nuclear protein trafficking by ubiquitinating components of the nuclear transport machinery and suppressing their activity by induced degradation in the cell.Figure 7**RNF38 regulates the translocation of E2F1 and p-STAT3 into the nucleus.**A. expression of RNF38 in HeLa cells did not change the levels of E2F1 or p-STAT3 in the cytoplasm but reduced the level of the two proteins in the nucleus. Cytoplasmic and nuclear fractions were isolated from cells transfected with the RNF38 plasmid for overexpression (OE) and blank control cells, and the levels of E2F1 and p-STAT3 in the fractionated samples were assayed by western blotting in three independent sets of samples probed with specific antibodies. β-actin and histone H3 were used as the loading control for the cytoplasmic and nuclear fractions, respectively. B. Bar charts of E2F1 and p-STAT3 levels in the cytoplasm and nucleus in control and RNF38-OE cells assayed by the western blotting shown in A. E2F1 and p-STAT3 levels were shown as ratios to β-actin in the cytoplasm and histone H3 in the nucleus, respectively. C, immunofluorescence images of E2F1 distribution in the cytoplasm and nucleus in blank HeLa cells (upper panels) and cells with the expression of GFP-RNF38 (bottom panels). Cells were stained with an antibody against E2F1 to visualize protein localization and with DAPI to reveal the nucleus. D, same as in (C) except that the cells were stained with an antibody against p-STAT3. E, ratio of E2F1 and p-STAT3 between the nuclear and cytoplasmic fractions in blank HeLa cells and cells with RNF38 OE. The levels of proteins in the cytoplasm and nucleus were quantified based on the fluorescence images shown in (D). In (B and E), the error bars represent SEM for protein quantification from three independent experiments (n = 3).

Discussion

The signature RING domains of the E3 engage the E2∼UB conjugate and activate UB transfer from the E2 to the substrate proteins bound to the E3s (1, 9). The RING domain binds to both E2 and UB connected by a thioester bond, and the three-way interaction between the RING and the E2∼UB conjugate promotes UB to adopt a "closed" conformation that would favor the Lys residues on the substrate to attack E2∼UB to complete the ubiquitination reaction (60, 61). The crystal structures of various E2∼UB conjugates in complexes with the RING domains of RNF4, Birc7, Cbl-b, RNF38, TRIM25, TRAF6, and Ark2C reveal that the RING domains organize an E2-binding interface with common structural elements, including loop1, the central α-helix and loop3, and the counterparts in E2 for RING binding are the N-terminal helix and two loop regions in E2 (42, 60, 62, 63, 64, 65, 66). Taking the RNF38 RING-UbcH5b ∼ UB complex for an example, besides the docking of the N-terminal helix of UbcH5b on loop1 of the RING, loops in E2 containing residues P61-F62 and P95-A96 are in contact distances with RING residues V415 in loop1, W443 in the central helix, and P451-I452 in loop3 (Fig. S8A) (41). In this study, we found the charge-reversed K4E and K8E mutations in the N-terminal helix of xUbcH5b were sufficient to block E2-RING interaction, while complementary mutations in loop1 of the RNF38 RING selected by phage display restored the interaction of the RING domain with xUbcH5b to enable the transfer of xUB to RNF38 substrates. LZ1, the dominant RING mutant from the phage selection, bore four mutations in loop1 (L412M, M417L, C418K, and D419E), with the C418K mutation complementing the charge-reversed K4E and K8E mutations in xUbcH5b. Since mutations at other sites are close to their original identities (Fig. 1E), the C418K mutation likely played an essential role in mediating xE2-xRING interaction in the orthogonal pair. Sequence alignment of the RING domains reveals that numerous RINGs have a negatively charged D/E occupying the equivalent position of C418 in the RNF38 RING, suggesting that positively charged K/R mutations at this position may enable the assembly of xE2-xE3 pairs with other RING E3s (Fig. S8B). So, phage selection has identified key residues in the RING for interacting with xE2, and such knowledge will guide the expansion of the OUT cascade in the RING family for profiling E3 substrates in the cell. In addition, our engineering of the xUbcH5b-xRNF38 pair for building the OUT cascade to screen RNF38 substrates may guide the engineering of other xE2-xRNF38 pairs to reveal the shift in the RNF38 substrate pool by pairing RNF38 with different E2s.

One unique feature of using the OUT cascade to identify E3 substrates is that OUT assigns E3-substrate relationships based on the direct transfer of xUB from the engineered xE3 constituting the OUT cascade to its ubiquitination targets in the cell. The mutations in xE3 are at the E2 binding interface to generate the xE2-xE3 pair, and these mutations are unlikely to affect the binding of the native substrates to xE3. So, xE3 would re-enact the role of wt E3 in recognizing its cellular targets while exclusively labeling them with HBT-xUB that is covalently attached to the substrate proteins for their enrichment by tandem purification. In this way, the low affinity of E3 with the substrates and their transient interaction would not affect the purification of E3 substrates from cell lysates for proteomic identification. Moreover, by directly following the conjugation of xUB to the substrate proteins catalyzed by xE3, OUT does not rely on reading the change of diGly modification or the expression level of the substrates in the cell for assigning E3 substrates since these features are more prone to be perturbed by the up or downregulation of E3 expression that would affect the level and catalytic activities of other E3s, deubiquitinating enzymes (DUBs), or the proteasomes in the cell. In addition, the engineered xE3 in the OUT cascade is free of fusion with non-native E3 components, such as UB, Nedd8, UBA, or biotin ligase, which may unintentionally distort the substrate profiles of E3s. Still, E3s adopt diverse structures and mechanisms to catalyze substrate ubiquitination; various methods may find their best matches of E3s for deconvoluting their biological functions in the cell, starting from substrate identification.

One limitation of our study is that we overexpressed HBT-xUB and xE3 in the cells stably transfected with the xE1-xE2 pair for labeling potential E3 targets with HBT-xUB. The overexpression of xE3 may cause the transfer of xUB to non-physiological targets. In our future studies, this can be avoided by editing the host cell genome to incorporate xE3 mutations into the native E3 gene, so the xE3 will be expressed at the native level for substrate labeling and identification. Another limitation of our work is that, as a proof-of-concept study, we used HEK293 cells to express the OUT cascade of RNF38 to profile its substrate specificity, while the known roles of RNF38 are most relevant to cancer development. It would be helpful to express the RNF38 OUT cascade in specific types of cancer models that overexpress RNF38, thereby increasing the likelihood of identifying RNF38 substrates in oncogenic pathways. Due to the low expression of RNF38 in HEK293 and HeLa cells used in this study, we were not able to knock down the expression of RNF38 in these cells to measure the effects of decreased RNF38 expression on substrate ubiquitination and stability. Future studies could use cancer cells with high endogenous expression of RNF38 to verify its regulation of the substrates by silencing RNF38 expression.

The RNF38 substrates identified by the OUT screen in this study are mainly associated with pathways regulating nucleocytoplasmic transport and mRNA translation, two tightly coupled cellular processes that involve the shuffling of transcription regulators and mRNAs in and out of the nucleus to control gene activation and protein synthesis. To establish the connection between RNF38 and the nuclear transport process, we verified several components of the nuclear transport machinery, including Ran, RanGAP1, and KPNA2, as RNF38 substrates and found that RNF38-catalyzed ubiquitination induced their degradation in the cell. Ran is a Ras-like GTPase that cycles between GTP and GDP-bound states due to the activation of its GTPase activity by RanGAP1 (67, 68, 69). The Ran cycle controls the assembly and disassembly of cargo proteins in complexes with importin and exportin proteins, driving the bidirectional trafficking of cargo proteins across the nuclear membrane (Fig. 8) (70). KPNA2, also known as importin α1 (Impα1), is a component of the importin complex that binds to Ran-GTP to trigger the release of cargo proteins in the nucleus (54, 71, 72). Here, we found RNF38 suppressed the trafficking of KPNA2 cargo proteins E2F1 and p-STAT3 into the nucleus, suggesting RNF38 may regulate nuclear transport in the cell by affecting the stability of integral parts of the nuclear transport machinery. We previously found that the viral oncoprotein E6 of the human papillomavirus (HPV) stimulates the host E3 E6AP to ubiquitinate nuclear transport adaptors KPNA1-3 and induce their degradation, leading to the suppressed transport of p-STAT1 into the nucleus and delayed apoptosis of cervical cancer cells (35). These results, combined, suggest that nuclear transport components are key targets of E3-catalyzed ubiquitination reactions for manifesting the roles of E3s in cell regulation.Figure 8A model for the regulation of nuclear transport pathways by RNF38. Cargo proteins, including HuR, E2F1, and p-STAT3, are assembled with the importin complex composed of the KPNA2-KPNB1 pair in the cytoplasm and transported across the nuclear pore complex (NPC) to enter the nucleus. The association of Ran-GTP with the cargo-importin complex results in the disassembly of the complex and the release of cargo proteins into the nucleus. Cargo proteins in the nucleus are also bound to exportin for their transport to the cytoplasm. RanGTP hydrolysis to RanGDP in the cytoplasm is activated by RanGAP1, triggering the disassociation of the cargo proteins from exportin. The OUT screen identified RNF38-catalyzed ubiquitination of nuclear transport components, including Ran, RanGAP1, KPNA2, and KPNB1, as well as the cargo protein HuR. We also verified the role of RNF38 in regulating the import of KPNA2 cargos E2F1 and p-STAT3 into the nucleus.

Besides triggering cargo release from the importin complex in the nucleus, RanGTP also facilitates the binding of cargo proteins with exportin to facilitate nuclear export. Once the trimeric RanGTP-exportin-cargo complex is transported to the cytoplasm, RanGAP1 activates the GTPase activity of Ran, leading to cargo release to the cytoplasm (Fig. 8) (73, 74). Therefore, RNF38-catalyzed ubiquitination of Ran and RanGAP1 may affect both the nuclear import and export processes. The OUT screen also suggested that RNF38 may ubiquitinate other proteins in the nuclear transport pathway, including KPNB1, the Impβ cargo receptor for nuclear transport, and NUTF2/NUF2, the adaptor protein binding to NPC (Fig. 3D and Table S1). KPNA2 and KPNB1 form an Impα-Impβ complex for cargo import into the nucleus (75, 76), and NUTF2 bridges the interaction of RanGDP with NPC to recycle RanGDP back to the nucleus (77, 78). Since multiple interactions between RNF38 and the components of the nuclear transport machinery have been identified, the effects of RNF38 on the bidirectional trafficking of cargo proteins across the nuclear envelope warrant further investigation.

HuR, also known as ELAVL1, is an RNA-binding protein that is released from the nucleus to stabilize a sub-pool of mRNA in the cytoplasm and promote their translation (55, 79). Interestingly, it relies on KPNA2 to be shuttled back to the nucleus (80). We verified RNF38-catalyzed ubiquitination of HuR as a high-fidelity target from the OUT screen, suggesting RNF38 may affect the stability of cargo proteins for nuclear transport (Fig. 8). The overexpression of HuR as cargo, as well as KPNA2, Ran, and RanGAP1 as components of the carrier system, has been identified as pro-oncogenic (81, 82, 83, 84, 85, 86). Our discovery of their ubiquitination and induced degradation by RNF38 suggests the E3 may function as a rheostat to fine-tune the material flow between the cytoplasm and nucleus and regulate cell proliferation.

In addition to proteins of the nuclear transport pathway, we also verified the RNF38-catalyzed ubiquitination of Rack1 and VPS35 based on the OUT-substrate profile. Rack1 is an adaptor protein that is homologous to the Gβ protein and interacts with a wide range of partner proteins, including ion channels, membrane receptors, G proteins, and kinases (87). Rack1 is also incorporated into the ribosome and can recruit protein kinase C and eukaryotic initiation factors to the ribosome to activate protein translation (56). Our identification of Rack1 as an RNF38 substrate suggests ubiquitination pathways mediated by RNF38 may affect diverse cell signaling processes and protein translation in the cell. VPS35 is a core component of the retromer that supports endosomal sorting and recycling of membrane cargos (88, 89). VPS35 and retromer activities are important for removing damaged mitochondria by mitophagy and sustaining the function of lysosomes (90, 91, 92). By controlling the ubiquitination and stability of VPS35, RNF38 may affect protein trafficking and mitochondrial and lysosome functions in the cell. Although RNF38 was found to be mainly in the nucleus, it has been found to target cytoplasmic proteins for ubiquitination, such as ACTN4, an actin-binding protein, and SHP-1, a SH2-containing protein tyrosine phosphatase (45, 47). Our verification of VPS35 as an RNF38 substrate suggests that RNF38 may also function outside of the nucleus.

In this work, we engineered an OUT cascade of RNF38 by phage display and used it to acquire a substrate profile of the RING E3, which suggests its role in regulating nuclear transport and protein translation. Guided by the OUT screen, we verified that RNF38 induces the ubiquitination and degradation of Ran, RanGAP1, and KPNA2, which are key components of the nuclear transport machinery, as well as HuR, Rack1, and VPS35, which underlie the potential roles of RNF38 in regulating protein translation and intracellular trafficking. Previous experimental and pathological data suggest that RNF38 may exert either oncogenic or tumor-suppressive functions, depending on the cell type and context. Further studies on the newly verified RNF38 substrates and other candidates in the substrate profile generated by OUT may clarify the role of RNF38 in managing cancer-related pathways and yield new mechanistic insights on E3-regulated cellular processes. Moreover, the establishment of the OUT cascade with RNF38 proves the feasibility of using OUT to profile RING E3 substrates and will guide the engineering of other RING E3s for incorporating them into the OUT platform to discover E3 functions.

Experimental procedures

Materials

pET-15b and pET-28a plasmids for protein expression from E coli were from Novagen. pGEX-4T1 vector was from Amersham Biosciences. pLenti4/V5-DEST-zeocin (K498000) was from Life Technologies. The pComb3H plasmid for the display of the RING domain library was previously reported (51). pLenti6-hygromycin-HBT-xUB plasmid was constructed in a previous study for expressing xUB with the R42E and R72E mutations with an N-terminal 6×His and biotin carboxyl carrier protein (BCCP) tag (37). The following plasmids were acquired from Addgene for the expression of RNF38 substrates: pEGFP-C2 RanGAP1 (Plasmid #13378), pET11d-RAN (Plasmid #108919), pCMVTNT-T7-KPNA2 (Plasmid #26678), pFRT_TO_eGFP_ELAVL1 (Plasmid #106105), pEGFP-N1-Rack1 (item #41088), and pLenti6/V5/DEST-VPS35 (item #21691). RNF38 plasmid was a kind gift from Dr C. Kenneth Kassenbrock of Colorado State University, Colorado, USA. pGEX-wtRNF38 RING plasmid for the expression of the RING domain was a gift from Dr Danny Huang of the Beatson Institute for Cancer Research. The plasmids for the expression of wt UB, Uba1, and UbcH5b, and engineered xUB, xUba1, and xUbcH5b in E coli and mammalian cells were from a previous study (38). E coli BL21 DE3 cells (200131) for protein expression and XL1 Blue cells (200228) and VCSM13 helper phage (200251) for phage preparation were from Agilent Technologies. HEK293T and HeLa cells were from the American Tissue Culture Collection (ATCC). HEK293T cells stably expressing Flag-xUba1 and V5-xUbcH5b were generated in a previous study (38). High-glucose Dulbecco's modified Eagles medium (DMEM) (Life Technologies, 11965092) with 10% (v/v) Fetal bovine serum (FBS) (Life Technologies, 11965092) was used as cell culture medium. Doxycycline, hygromycin, blasticidin, zeocin, and puromycin were from GiBCO/Invitrogen and RPI. The following antibodies were purchased from Santa Cruz Biotechnology: anti-ACTN4 antibody (sc-390205), anti-GFP antibody (sc-9996), anti-GST antibody (sc-138), anti-HuR/ELAVL1 antibody (sc-5261), anti-KPNA2 antibody (sc-55538), anti-Myc antibody (sc-40), anti-p53 antibody (sc-126), anti-Rack1 antibody (sc-17754), anti-Ran antibody (sc-271376), anti-RanGAP1 antibody (sc-28322), anti-ubiquitin antibody (sc-8017), anti-Histone H3 antibody (sc-517576), and anti-VPS35 antibody (sc-374372). The following antibodies were from ABclonal Technology: anti-DDDDK antibody (AE005), anti-RNF38 antibody (A17263), anti-Ran antibody (A4374), anti-E2F1 antibody (A2067), and anti-β-actin antibody (AC004). Protein A/G plus-agarose (sc-2003) was from Santa Cruz Biotechnology. The anti-Flag antibody was from Sigma Aldrich (F3165). The anti-p-STAT3 antibody (#9145) was purchased from Cell Signaling Technology (Danvers, MA). Alexa Fluor 594-labelled donkey antibody against rabbit IgG H&L (ab150064) was obtained from Abcam. The restriction enzymes for molecular cloning were from New England Biolabs. The DNA oligonucleotides for cloning were from Integrated DNA Technologies. DharmaFECT kb DNA transfection reagent was from Horizon Discovery Biosciences. Full-length wt RNF38 protein (H00152006) was purchased from Abnova Corporation. DAPI (#D523) was obtained from Dojindo Molecular Technologies.

Construction of plasmids

The RING domain of RNF38 was PCR-amplified from the pGEX-wtRNF38 RING plasmid, double digested by SacII and SpeI restriction enzymes, and ligated into the pComb3H vector for the display of the RING with an N-terminal Flag tag. After phage selection, the genes encoding the selected RING mutants in the pComb3H phagemid vector were amplified by PCR, digested with the BamHI and NotI restriction enzymes, and cloned into the pGEX vector for expressing the RING mutants as GST fusions. The full-length gene of xRNF38 with mutated RING domain (LZ1) was cloned into the pEBG vector and expressed as a fusion with an N-terminal GFP tag. The genes of RNF38 substrates verified in this study, including Ran, RanGAP1, KPNA2, HuR/ELAVL1, Rack1, and VPS35, were cloned into the pET vector for expression in E coli cells.

Protein expression

For expressing the RING domain and mutants of RNF38, the pGEX plasmids containing the wt RNF38 RING or mutant genes were transformed into BL21(DE3) cells and expressed and purified by glutathione affinity resin following the vendor's protocol. Briefly, cells harboring the pGEX-RING plasmids were cultured in 1 L 2× YT broth supplemented with kanamycin (70 mg/ml) at 37 °C until the cell culture reached an optical density of 0.6 ∼ 0.8 (600 nm). At this point, isopropyl-β-d-thiogalactopyranoside (IPTG) was added to the cell culture to reach a concentration of 1 mM. The culture was switched to 15 °C for shaking overnight. The cells were then harvested by centrifugation at 5000 rpm (4225g) for 30 min and the cell pellets were resuspended in 10 ml of lysis buffer (20 mM Tris pH 7.5, 0.1% Triton X-100, 1 mg/ml lysozyme, and 1 μg/ml DNase). The cell suspension was treated with 2 mg/ml lysozyme (Alfa Aesar) and incubated on ice for 30 min. Then, the cells were lysed by either French press or sonication. The cell debris was pelleted by centrifugation at 10,000 rpm (11,952g) at 4 °C for 25 min, and the cleared cell lysate was transferred to a conical tube and incubated with 1 ml Pierce glutathione agarose (Thermo, Cat.16102BID) preequilibrated in the binding buffer (20 mM Tris pH 7.5, containing 0.1% Triton X-100). The tube was incubated at 4 °C with rotation for 14∼18 h. Then, the glutathione agarose beads were washed with the binding buffer, and the protein was eluted with 6 ml elution buffer (50 mM Tris-HCl, 10 mM reduced glutathione, pH 8.0). The eluted protein was dialyzed overnight at 4 °C in the dialysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 1 mM DTT) and concentrated. The concentration of the protein was measured by the Bradford assay (Bio-Rad).

Ubiquitination assays

The self-ubiquitination assays with the RING domain or full-length RNF38 contains 1 μM Uba1, 5 μM Ubch5b, 5 μM GST-tagged RING or RNF38 protein, and 30 μM HA-tagged UB in 50 μl reaction buffer (50 mM Tris pH 7.5, 10 mM MgCl_2_, 2.5 mM ATP, and 1 mM DTT). Control reactions were also set up with the exclusion of Uba1, UbcH5b, RNF38, or UB. The reactions were incubated at 37 °C for 1 h and analyzed by SDS-PAGE and western blotting probed with an anti-GST antibody. To assay the ubiquitination of RNF38 substrates, 5 to 10 μM substrate proteins purified from E coli cells were incubated with 1 μM wt Uba1, 5 μM wt UbcH5b, 2 μM full-length RNF38, and 30 μM wt UB in TBS supplemented with 10 mM MgCl_2_ and 5 mM ATP. After 2-h reaction at room temperature, the ubiquitination of substrates was analyzed by western blotting probed with either substrate-specific antibodies or antibodies against the tags fused to the substrates.

Preparation of the phage displaying the RING domain of RNF38

The pComb-RNF38 RING phagemid was transformed into XL1-blue cells by electroporation and plated on LB-agar plates containing 100 μg/ml ampicillin and 2% glucose. After overnight incubation at 37 °C, a single colony of the transformed cells was picked to inoculate 5 ml LB medium supplemented with 2% glucose, 10 μg/ml tetracycline, and 100 μg/ml ampicillin, and the culture was grown overnight at 37 °C. The overnight culture was added to 20 ml of 2×YT medium containing 2% glucose, 10 μg/ml tetracycline, and 50 μg/ml ampicillin to reach a starting optical density (OD) of 0.1 at 600 nm, and the culture was shaken at 37 °C until the OD reached 0.6. The approximate number of cells in the culture was calculated assuming 1 OD equaled 8 × 10^8^ cells/ml. The VCSM13 helper phage was then added to the culture at a multiplicity of infection of 10. The culture was mixed well with helper phage and stood still in a 37 °C incubator, allowing phage attachment to the cells. After 1 h of incubation, the cells were pelleted by centrifugation at 3700 rpm (3210 g) for 10 min and resuspended in 200 ml of fresh 2xYT media with 100 μg/ml ampicillin and 50 mg/ml kanamycin; the culture was then grown overnight at 30 °C with shaking. The next morning, the cells in the culture were precipitated by centrifugation at 5000 rpm (4225 g) for 20 min, and the supernatant containing the phage was decanted to a clean centrifuge bottle. For phage precipitation from the culture media, one-fifth volume of filter-sterilized solution of PEG 8000 (200 g/L) and NaCl (2.5 M) was added to the media. The PEG/NaCl solution was thoroughly mixed with the supernatant by shaking the centrifuge bottle, and the bottle was incubated on ice for 1 h. The bottle was centrifuged for 30 min at 8000 rpm (10,816 g) to precipitate the phage that would appear on the sidewall of the bottle as a layer of white deposit. The supernatant was removed by decanting, and the remaining liquid was removed by pipetting. The pellet was then washed off the wall of the centrifuge bottle with 1 ml of sterilized TBS buffer (20 mM TrisHCl, 150 mM NaCl, pH 7.5) and thoroughly resuspended in the solution by repeated pipetting. The phage solution was added to a microcentrifuge tube and centrifuged for 10 min at 10,000 rpm (6121g) to precipitate the cell debris left in the solution. The cleared phage solution was filtered with a 0.2 μm sterilized syringe filter to remove the E coli cells left in the solution. The phage concentration was titered, and the concentration of the phage should be > 10^12^ cfu/ml. The RING domain display on phage was confirmed by SDS-PAGE analysis of the phage particles after lysing the phage in the gel loading buffer with boiling. The western blotting of the gel was probed with an antibody against the Flag tag fused to the N-terminal of the RING-pIII fusion.

Ubiquitination assays of the RING domain displayed on the phage surface

Approximately 1 × 10^11^ phage particles displaying the wt RING domain of RNF38 were incubated with 1 μM Uba1, 5 μM UbcH5b, and 20 μM N-terminally tagged HA-UB in 50 μl reaction containing 50 mM Tris pH 7.5, 10 mM MgCl_2_, 2.5 mM ATP, and 1 mM DTT. The reaction was incubated for 1 h at 37 °C with gentle shaking. The reaction was then quenched by boiling in the gel loading buffer containing 100 mM DTT and analyzed by SDS-PAGE and immunoblotting probed with an anti-Flag antibody.

Alternatively, UB loading on the RING domain displayed on the M13 phage was assayed by phage ELISA. Ubiquitination reactions with the phage particles displaying the RING domain were set up as above with 1 μM Uba1, 5 μM UbcH5b, and 0.5 μM N-terminal labeled biotin-UB (Boston Biochem). In parallel, control reactions were also set up, excluding Uba1, UbcH5b, or biotin-wt UB. The reactions were allowed to proceed for 1 h, and 10 μl of the reaction mixture with the phage was added to 90 μl 1% BSA solution in TBS in a 96-well streptavidin plate pre-blocked with 3% BSA in TBS for 1 h. The phage solution was mixed well in the plate, and 10× serial dilution was carried out across the wells in the plate. The phage solution was incubated with the plate for 1 h to allow the binding of biotin-labeled phage with the streptavidin surface. The phage solution was removed from the plate, and the plate wells were washed five times with TBST buffer (20 mM TrisHCl, 150 mM NaCl, 0.1% Tween 20, pH 7.5) and five times with TBS (200 μl buffer in each well each time). A solution of the anti-M13 antibody-horseradish peroxidase diluted in 3% BSA-TBS (1:10,000) was added to the plate to detect the phage bound to the plate. After washing the plate, a solution of 3,3′,5,5′-tetramethylbenzidine (TMB, 70 pg/ml) was added, and the absorbance at 450 nm was quantified.

After rounds of phage selection, the activity of converged phage mutants displaying the RING variants of LZ1, LZ2, LZ3, and LZ4 were also verified by ELISA. The phage displaying each RING mutant was individually prepared, and approximately 2 × 10^10^ phage particles were reacted with 0.5 μM xUba1(UFD), 5 μM xUbch5b, and 0.5 μM biotin-wt UB in 50 μl reaction containing 50 mM Tris pH 7.5, 10 mM MgCl_2_, 5 mM ATP and 0.1% BSA. The reaction lasted for 1 h, and phage ELISA was performed using the same procedure to detect the binding of biotin-UB-loaded phage with the streptavidin surface.

Model selection of RNF38 RING domain phage

Phage particles displaying the RING domain of RNF38 were mixed with the phage particles displaying the viral protein SV5V with a ratio of 1:1, 1:10, and 1:100. The phage mixture containing 1 × 10^11^ particles was reacted with 1 μM Uba1, 5 μM UbcH5b and 0.5 μM N-terminal labeled biotin-UB in a 100 μl reaction, following the same reaction conditions for phage ELISA. After the reaction, the phage mixtures were selected by binding to the streptavidin plate using the same procedure as phage ELISA. The phage particles bound to the streptavidin plate were eluted by proteolytic cleavage with the TEV protease (5 units in 100 μl ProTEV buffer containing 1 mM DTT for each well). The eluted phage particles were added to log-phase XL1-blue cells for infection, and the cells were plated on the LB-agar plates supplemented with 50 μg/ml ampicillin and 2% glucose. After overnight incubation of the plate at 37 °C), the colonies on the plate were analyzed by colony PCR using primers binding to the phagemid backbone to amplify the RING domain or the SV5V genes of different sizes. The PCR products were analyzed by agarose gel electrophoresis to identify the E coli colonies infected with either the RNF38 RING domain phage or the SV5V phage.

Construction of the pComb3H-RNF38 RING library

The RING domain gene of RNF38 with randomized codons at L412, M417, C418, and D419 was ordered from Genscript. The library gene was amplified by PCR, digested with SacII and SpeI restriction enzymes and ligated into the pComb3H vector to construct the phagemid library. After the ligation, the library DNA was transformed into XL1-blue cells by electroporation. The cells were plated on LB-agar plates containing 100 μg/ml ampicillin and 2% glucose and incubated overnight at 37 °C. The next day, the cells were scraped from the agar plates, and the library DNA was extracted from the cells using a DNA Maxiprep kit (Qiagen).

Preparation of the RNGF38 RING library phage for the selection reaction

1 μl of pComb3H-RNF38 RING library phagemid (∼0.5 mg/ml) was used to transform 100 μl electrocompetent XL1-blue cells. After electroporation, the cells were incubated in 1 ml SOC medium with shaking for 1 h at 37 °C. Then 20 ml 2×YT broth containing 50 μg/ml ampicillin and 10 μg/ml tetracycline was added to the cell culture, and the cells continued to grow for 2 ∼ 4 h at 37 °C until the OD at 600 nm reached 0.6 ∼ 0.8. Then, VCSM13 helper phage was added to infect the cells at 10-fold multiplicity of infection. The helper phage and the cells were mixed well by rotating the culture tube, and the tube was left standing still for 1 h at 37 °C to facilitate phage attachment to the E coli cells. After infection, the cell culture was centrifuged to pellet the cells and the same procedure for preparing the RNF38 RING domain phage was followed for preparing the phage library. The display of the RING domain on the phage particles in the library was confirmed by SDS-PAGE analysis followed by western blotting probed with an anti-Flag antibody.

Selection of the RING domain phage library based on the catalytic transfer of biotin-UB

For the first round of selection, a UB transfer reaction was set up with 1 × 10^11^ phage particles of the RNF38 RING library in the presence of 1 μM xUba1(UFD), 5 μM xUbcH5b and 0.5 μM biotin-wt UB in 100 μl reaction buffer containing 50 mM Tris pH 7.5, 10 mM MgCl_2_, and 3 mM ATP. xUba1(UFD) has the mutations of E1004K, D1014K, and E1016K in the UFD domain for mediating the transfer of wt UB to xUbcH5b (40). The reaction was incubated for 1 h at room temperature in parallel with control reactions with the same amount of phage input but eliminating xUba1(UFD), xUbcH5b, or biotin-UB from the reaction. After incubation, the reactions were diluted 10-fold into 3% BSA-TBST and loaded into 10 separate wells of the streptavidin-coated plate to allow the binding of biotin-labeled phage with streptavidin for 1 h. Then, the supernatant of the phage solution in each well was removed and saved for phage titering. The plates were then thoroughly washed 30 times with TBST and 30 times with TBS, each time using 200 μl of wash buffer for each well. After washing, 100 μl ProTEV buffer containing TEV protease (5 units) and DTT (1 mM) was added to each well, and the plate was incubated at 30 °C for 30 to 45 min to release the phage from the streptavidin surface by cleaving the linker peptide between the RING domain and the pIII protein of the phage particles. The solution containing eluted phage was added to a mid-log culture of XL1Blue cells in LB at a ratio of 1/10 (v/v) with gentle shaking at 37 °C for 1 h to allow phage infection of the cell. The cells were then plated out on LB-agar plates containing 100 μg/ml ampicillin and 2% glucose and allowed to grow overnight at 37 °C. The next day, the cells were scraped from the plate, and the phagemid DNA was extracted from the cell using a DNA maxiprep kit. The phagemid DNA can then be used for the transformation of XL1Blue cells to prepare the phage for the next round of selection. Both the input phage number in the supernatant for streptavidin binding and the output phage number in the eluted phage solution were titered to measure phage enrichment from each round of selection. In parallel, the phage in the control reactions excluding xUba1(UFD), xUbcH5b, or biotin-wt UB was also bound to the plate, and the input phage and output phage numbers were measured in the same way after phage binding to streptavidin plate and elution by TEV cleavage.

DNA transfection into the HEK293 cells

HEK293 cells were seeded in a 75 cm^2^ cell culture flask and allowed to grow at 30 °C for 24 h in the DMEM medium to reach a confluency of 70%. Plasmid DNA for transfection was diluted in 0.5 ml serum-free DMEM and mixed with the DharmaFECT kb transfection reagent by gentle pipetting at a ratio suggested by the manufacturer. The mixture was incubated at room temperature for 20 min to allow the packaging of DNA into the transfection reagent. The cell culture medium was aspirated from the flask, followed by the immediate dispensing of the DNA mixture into the flask. The flask was gently rocked and incubated at 37 °C in a CO_2_ incubator. The cells were collected in a time frame of 24 to 72 h after the transfection to check the expression of the protein from the plasmid.

Tandem affinity purification of xUB-conjugated proteins

A previously developed procedure for purifying HBT-xUB conjugated proteins was followed to enrich RNF38 substrates from cells expressing the RNF38 OUT cascade (38). Briefly, ten 75 cm^2^-flasks of HEK293T cells stably expressing the xUba1-xUbcH5b pair were acutely co-transfected with xRNF38 and HBT-xUB for 48 h. Cells were treated with 10 μM MG132 for 4 h at 37 °C before harvesting to inhibit proteasome activity. The culture medium was then removed from the flasks, and the cells were washed twice with ice-cold 1×PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na_2_HPO_4_, and 1.8 mM KH_2_PO_4_, pH 7.4) and harvested by a cell scraper with buffer A (8 M urea, 300 mM NaCl, 50 mM tris, 50 mM NaH_2_PO_4_, 0.5% NP-40, 1 mM phenylmethylsulfonyl fluoride, and benzonase (125 U/ml), pH 8.0). For Ni-NTA purification, cell lysates were centrifuged at 15,000g for 30 min at room temperature. 35 μl of Ni-NTA sepharose beads (GE Healthcare) for each 1 mg protein in the lysate were added to the clarified supernatant. The mixture was rocked overnight at room temperature in buffer A with 10 mM imidazole. Ni-NTA beads were then pelleted by centrifugation at 100g for 3 min and washed sequentially with a 20-bead volume of buffer A (pH 8.0), buffer A (pH 6.3), and buffer A (pH 6.3) with 10 mM imidazole. After the washing was done, proteins were eluted from the Ni-NTA beads twice with a 5-bead volume of buffer B (8 M urea, 200 mM NaCl, 50 mM Na_2_HPO_4_, 2% SDS, 10 mM EDTA, 100 mM Tris, and 250 mM imidazole, pH 4.3). For streptavidin purification, the pH of the elution solution from Ni-NTA beads was adjusted to 8.0. 50 microliters of streptavidin sepharose beads (Thermo Fisher Scientific) were added to the solution to bind HBT-xUB-conjugated proteins. The solution mixtures were incubated on a rocking platform overnight at room temperature. Then, the streptavidin beads were pelleted and washed sequentially with 1.5 ml buffer C (8 M urea, 200 mM NaCl, 2% SDS, and 100 mM Tris, pH 8.0), buffer D (8 M urea, 1.2 M NaCl, 0.2% SDS, 100 mM Tris, 10% EtOH, and 10% isopropanol, pH 8.0), and buffer E (8 M urea and 100 mM NH_4_HCO_3_, pH 8). After washing, the streptavidin beads were spun down, the residual urea solution was removed, and the streptavidin beads bound with HBT-xUB-conjugated proteins were submitted to the Emory Integrated Proteomics Core Facility for identification by proteomics.

Sample digestion and LC-MS/MS analysis

After tandem purification, the streptavidin beads were washed to remove the residual urea, and HBT-xUB conjugated proteins bound to the beads were digested by LysC and trypsin as previously reported (37, 38). The supernatant of the digestion reaction with minimal amount of streptavidin beads was desalted and cleaned using HLB microelution plates or columns. The digested peptide fragments were identified by liquid chromatography coupled to tandem MS (LC-MS/MS) on an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific) at the Emory Integrated Proteomics Core. The spectra collected were compared to the human UniProt database using Proteome Discoverer 2.0. (90,300 target sequences). Fully tryptic restriction and a parent ion mass tolerance of ±20 parts per million were used as search criteria. Methionine oxidation (+15.99492 Da), asparagine and glutamine deamidation (+0.98402 Da), lysine ubiquitination (+114.04293 Da), and protein N-terminal acetylation (+42.03670 Da) were all variable modifications (up to three per peptide were permitted); cysteine received a fixed carbamidomethyl modification (+57.021465 Da). The PSMs were filtered using a percolator to achieve a 1% false discovery rate. False discovery rate is based on the target decoy search approach in Proteome Discover's Sequest module. The reverse sequence Decoy database is automatically generated by the software and automatically concatenated to the target database. Percolator within Proteome Discoverer was used to filter each experiment.

In-solution digestion was performed on bead, and supernatant (with minimal beads) was desalted and cleaned using HLB microelution plates.

Bioinformatics analysis

IPA software (www.ingenuity.com) was used to map and identify the biological networks and molecular pathways with a significant proportion of genes having RNF38 ubiquitination targets. Fisher's exact test in IPA software was used to calculate the p values for pathways and networks. The level of statistical significance was set at p < 0.05. IPA was also used to visualize the identified biological networks.

Lysis of the cells and assaying substrate protein ubiquitination in the cell

HEK293T cells were transfected with increasing amounts of pEBG-GFP-RNF38 plasmid (0–4 μg) for 24 h. The cells were lysed in RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, and 1% Triton-X) supplemented with 1 mM PMSF. The cell lysates were sonicated, and the cell debris was pelleted by centrifugation at 10,000 rpm (6121 g) for 10 min in a microcentrifuge at 4 °C to give a clear cell lysate. The cell lysate was first precleared for non-specific antibody binding by adding 0.25 μg of the control IgG (corresponding to the host species of the primary antibody) to approximately 1 ml of whole cell lysate, followed by the addition of 20 μl (25% v/v) suspended protein A/G agarose, and the mixture was incubated with gentle rocking for 30 min at 4 °C. The tubes were centrifuged at 3000 rpm (551 g) for 30 s at 4 °C to precipitate the agarose beads. The precleared cell lysate was then transferred to a new microcentrifuge tube. 10 μg of primary antibody specific for individual substrate proteins was added to 1 ml of precleared cell lysate that contained 100 to 1000 μg cell lysate protein. The tubes were incubated at 4 °C for 1 to 2 h with mixing, and then 20 μl protein A/G agarose suspension was added. The tubes were rotated for several hours or overnight at 4 °C before they were centrifuged at 3000 rpm (551 g) for 30 s at 4 °C to precipitate the agarose beads. The supernatant in the tubes was removed by aspiration, and the agarose was resuspended and rinsed four times with the PBS buffer. Finally, the SDS-PAGE gel loading buffer was added to the agarose beads with the target protein bound, and the tubes were boiled for 5 min before loading the samples on the SDS-PAGE gel for western blotting analysis probed with antibodies targeting specific RNF38 substrate proteins or UB.

Protein degradation assays

HEK293 cells were transfected with increasing amounts of pEBG-GFP-RNF38 and collected at 48 h post-transfection, and the amount of substrate proteins in the cell lysate was assayed by immunoblotting with substrate-specific antibodies. For cycloheximide (CHX) chase assays, HEK293 cells were transiently transfected with 2 μg empty pEBG or pEBG-GFP-RNF38 plasmids. Cells were then cultured for 48 h before being treated with 100 μg/ml CHX to block protein synthesis. The cells were then collected after 0-, 2-, 4-, and 6-h treatment with CHX, and the amount of substrate proteins in the cell was assayed by immunoblotting with antibodies specific to the substrate. Protein levels were normalized to the level of β-actin.

Assay of the substrate levels in the cytoplasmic and nuclear portions of the cells

HeLa cells were cultured to 70 to 80% confluence, and the pEBG-GFP-RNF38 plasmid was transfected into cells with Lipofectamine 3000 transfection reagent following the manufacturer's protocol. The pEBG plasmid expressed wt RNF38 with an N-terminal GFP tag. The cells were then lysed in RIPA buffer containing protease inhibitors to harvest total proteins. Nuclear and cytoplasmic proteins were extracted using a commercially available nuclear and cytoplasmic extraction kit (#78833, Thermo Fisher Scientific, Logan, UT) following the manufacturer's protocol. Proteins of the cytoplasmic and nuclear fractions were separated by SDS PAGE and transferred to PVDF membrane for western blotting analysis. The membrane was first incubated with TBST buffer containing 5% skim milk for 1 h at room temperature, followed by incubation with appropriate primary antibodies overnight at 4 °C. The next day, the membrane was rinsed three times with TBST (5-min incubation with the TBST buffer for each rinse) and incubated with a 1:20,000 dilution of horseradish peroxidase-coupled secondary antibody for 1 h at room temperature. The membrane was then incubated with enhanced chemiluminescence reagents to visualize the target proteins. The Image J software was used to quantify the amount of proteins on the western blots.

Immunofluorescence staining of the nuclear transport cargo proteins in the cell

HeLa cells were transfected with the pEBG-GFP-RNF38 plasmid for 48 h for the expression of GFP-RNF38 fusion. The cells were then fixed with 4% paraformaldehyde for 15 min and permeabilized with 0.1% Triton X-100. After triple washes with PBS for 5 min each, the cells were blocked in PBS with 1% BSA for 1 h at room temperature, followed by overnight incubation at 4 °C with antibodies against E2F1 or p-STAT3 at a dilution of 1:200. Subsequently, the cells were washed and incubated with a donkey antibody against rabbit IgG conjugated with Alexa Fluor 594 for 1 h at room temperature. The cells were then washed with PBS for three times and incubated with a 1:200 dilution of DAPI for 15 min, followed by analysis using a laser confocal microscope. The immunofluorescent intensity of each target protein was calculated using the software Image J.

Data availability

The data that support the findings of this study are included in the article and are available from the corresponding author. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (93) partner repository with the dataset identifier PXD072090. The processed proteomics data for assigning RNF38 substrates are listed in Table S1.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they do not have any conflicts of interest with the content of this article.