UCH-L1 bypasses mTOR to promote protein biosynthesis and is required for MYC-driven lymphomagenesis in mice

Inhibitors of the mechanistic target of rapamycin (mTOR) are widely used as immunosuppressive and antineoplastic therapies. A serine-threonine kinase, mTOR assembles into 2 complexes with distinct biochemical effects.¹  When combined with the defining subunit raptor, the resulting mTOR complex 1 (mTORC1) phosphorylates downstream targets that in aggregate promote nucleotide, lipid, and protein synthesis. The targets p70S6 kinase (S6K) and the eIF4E binding protein 1 (4EBP1) both inhibit protein synthesis when in the unphosphorylated state. When mTOR binds to rictor to form mTORC2, it phosphorylates substrates, including AKT, that promote cellular survival and proliferation. The activity of both complexes is essential in normal physiology, with loss of either complex-defining subunit resulting in embryonic lethality in mice.² 

MYC is one of the most deregulated oncogenes in humans and regulates thousands of genes through its direct and indirect activity as a transcription factor.^3,4  Residing both upstream and downstream of the initiation step of mRNA translation, the oncogenic activity of MYC is closely related to its ability to increase protein biosynthesis. MYC directly stimulates protein synthesis in part by increasing the expression of translation initiation factors including eIF4E, eIF4G, and eIF4A, which assemble to form the eIF4F translation initiation complex.^5,6  The importance of this mechanism is evidenced by the acceleration of MYC-driven lymphomagenesis by the overexpression of eIF4E,⁷  and by the impaired lymphomagenesis seen in Eμ-myc mice when eIF4F components are suppressed by shRNA or small molecule inhibitors.^5,8  These strategies may provide important therapeutic insights for diseases with high MYC activity.

Through an unbiased screen, we found the expression of ubiquitin carboxy-terminal hydrolase L1 (UCH-L1) to be deregulated in B-cell lymphoma.⁹  Encoding a small (25 kDa) de-ubiquitinating enzyme, transgenic expression of Uchl1 is sufficient to drive spontaneous B-cell lymphomagenesis in mice and accelerate disease in the Eμ-myc model.^10,11  We uncovered a previously unrecognized function of UCH-L1 in reorganizing mTOR-AKT signaling both in cell lines and in mice.¹²  UCH-L1 disrupts the ubiquitination of the mTORC1 subunit raptor that is required for its stable association with mTOR, leading to reduced levels of mTORC1.^12,13  The direct result of UCH-L1 activity is therefore reduced phosphorylation of mTORC1 substrates, including S6K and 4EBP1.^1,14  The mTOR liberated by dissociation from raptor binds to rictor, forming mTORC2, and leading to increased phosphorylation of AKT.¹²  Although increased survival and proliferative signaling through mTORC2 has an obvious advantage in cancer, the inhibition of mTORC1 by UCH-L1 has been enigmatic, given the importance of protein biosynthesis in cancer in general, and MYC-driven lymphomagenesis in particular.

To better assess the functions of UCH-L1, we used proximity-based proteomics. We identified a novel association of UCH-L1 with the eIF4F translation initiation complex and find that it promotes eIF4F assembly and protein synthesis, despite inhibiting mTORC1. By using mice that lack UCH-L1, or those carrying catalytically active and inactive transgenes, we find that UCH-L1 and its catalytic activity are required for efficient MYC-driven lymphomagenesis in mice. These results identify a novel mechanism by which UCH-L1 bypasses 4EBP1 to drive protein biosynthesis and demonstrate an important functional relationship between UCH-L1 and MYC in lymphomagenesis.

Uchl1^Tg10, Uchl1^−/− (Uchl1^Dgen),¹⁵  and Eμ-myc¹⁶  (obtained from Jackson Laboratories, Bar Harbor, ME) mice were previously described. All strains were maintained on the C57BL/6 background for more than 10 generations. The DMBA carcinogenesis assay was performed as described.^17,18 Uchl1^TgC90A mice were generated similarly to the Uchl1^Tg (Hussain et al¹⁰  and in supplemental Methods, available on the Blood Web site). B-lymphocytes were purified, using magnetic separation methods as described.¹¹  Experimental procedures involving the use of these laboratory mice were reviewed and approved by the Institutional Animal Care and Use Committee of the Mayo Clinic.

Human UCH-L1 was cloned into the TSIN lentivirus vector,¹⁰  with BioID2¹⁹  fused to its N-terminus, using standard techniques. For proteomic analysis, the BioID2-UCHL1 construct was stably transduced into KMS11 cells that express endogenous UCH-L1. Large-scale biotinylation and mass spectrometric protein identification was performed as described.²⁰  BioID pulldowns were performed as described.²⁰  Briefly, cells were pulse labeled for 16 hours with biotin (50 μM), and were then lysed in urea lysis buffer (8 M urea, 50 mM Tris at pH 7.4, 1 mM dithiothreitol, and 1× protease inhibitor), and biotin-modified proteins were retrieved using streptavidin-coupled DynaBeads (MyOne Streptavidin C1; Thermo Fisher Scientific). Additional details are available in the supplemental Methods.

The list of 244 proteins enriched in the BioID2-UCH-L1 proximity proteome was analyzed using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) web tool (https://david.ncifcrf.gov), using the default DAVID annotation categories, with the classification stringency set to high. The terms within each retrieved annotation cluster were summarized, and a mean P-value was calculated. Analysis using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING; https://string-db.org/) was performed by limiting the active interaction sources to “experiments” and setting the “minimum required interaction score” to medium (0.4).

Cell lines used include HeLa (American Type Culture Collection, Manassas, VA), KMS11 and KMS12 (multiple myeloma, kindly provided by Takemi Otsuki), WSU-DLCL2, and SU-DHL6 (germinal center diffuse large B-cell lymphoma; both from American Type Culture Collection) were cultured under standard conditions.^10,12  The introduction of UCH-L1 and shRNA was performed using the TSiN or pTRIPZ lentiviral systems, as reported previously.^10,12  The activity of UCH-L1 and derivatives was determined using ubiquitin vinylmethylester, as previously described.²¹  Immunofluorescence imaging was performed as described.¹²  Polysome profiling was performed as described.²²  Immobilized m⁷GTP (AC-155) was from Jena Bioscience (Jena, Germany). m⁷GTP pulldowns were performed as described.²³  Assays for protein synthesis using the SUnSET technique were performed as described.²⁴ 

RNA extracted from lymphoma samples from Uchl1^TgWT mice was analyzed on the Affymetrix Mouse 430 2.0 array by the Mayo Clinic microarray core, and analyzed together with GSE35219 and GSE26408 in conjunction with the Mayo Clinic Department of Bioinformatics. Mate-pair sequencing was performed as described and mapped to the reference genome: GRCm38.p3.^25,26  The meta-analysis of UCHL1 and MYC transcript levels was performed using the 2D distribution function (R2: Genomics Analysis and Visualization Platform; http://r2.amc.nl) for all available samples and data sets, using the u133p2 platform. The parametric analysis of gene set enrichment analysis was performed with the indicated gene sets, using R2.

To better understand its molecular activities, we used proximity-based proteomics to identify the protein complexes that include UCH-L1. We used a second-generation BirA-based biotin ligase system (BioID2)¹⁹  expressed by itself, or as an N-terminal fusion with UCH-L1 in the multiple myeloma (plasma cell) line KMS-11. We verified the integrity of the UCH-L1 fusion by assessing its catalytic activity, its ability to promote AKT phosphorylation and suppress mTORC1 activity, and its subcellular localization (Figure 1A-D).

To identify proteins that are potential substrates and interaction partners, we performed 2 replicate purifications from biotin-pulsed KMS-11 cells expressing the BioID2 alone or the BioID2-UCH-L1 fusion. Taking advantage of the high affinity of streptavidin toward biotin, we performed these purifications on urea-denatured extracts to minimize the recovery of peripherally associated proteins. The retrieved proteins were identified by mass spectrometry, and the relative recovery of proteins between BioID2 and the BioID2-UCH-L1 fusion was used to identify putative binding partners. A total of 244 proteins were enriched at least 3-fold in the BioID2-UCH-L1 fusion compared with BioID2 alone, with 130 proteins present uniquely in pulldowns performed with the UCH-L1 fusion and not in those from cells expressing BioID2 alone (Table 1; supplemental Table 1). We analyzed the list of biotinylated proteins retrieved in cells expressing BioID2-UCH-L1, using DAVID (https://david.ncifcrf.gov). There were 6 functional clusters with significant enrichment when accounting for multiple comparisons and false-discovery rates (supplemental Table 2). Combining those with substantial overlap, we find that clusters including cell-cell adhesion, RNA-binding, translation initiation, and DEAD-box helicase are highly enriched among the proteins associating with BioID2-UCH-L1 (Table 2). Similar results were obtained analyzing this list using STRING (https://string-db.org/; supplemental Figure 1). Given the previously demonstrated role of UCH-L1 in the mTOR pathway and the importance of MYC in protein biosynthesis, we were drawn to a cluster of proteins involved in mRNA translation, particularly the process of translation initiation that is a highly regulated step (Table 3). As the assembly of the translation initiation eIF4F complex (consisting of eIF4E, eIF4G, and eIF4G) is inhibited by the mTOR substrate 4EBP1, we reasoned that this may shed more light on the mechanism of UCH-L1 in this pathway.

We previously showed that UCH-L1 destabilizes mTORC1 and reduces 4EBP1 phosphorylation, events that should reduce cap-dependent translation (Hussain et al¹² ; Figures 1C and 3D-E; supplemental Figure 4). As protein biosynthesis is essential to support the proliferation of cancer cells, this effect seems paradoxical with the oncogenic functions of UCH-L1. We therefore used the SUnSET technique to determine the rate of new protein translation in cells with or without UCH-L1 by measuring the incorporation of the aminonucleoside antibiotic puromycin into growing polypeptides.²⁴  We consistently observed increased synthesis of new proteins in cells expressing UCH-L1 compared with control (Figure 2A). This suggests that despite reduced mTORC1 phosphorylation of 4EBP1, global translation rates are not suppressed, and are in fact enhanced in the presence of UCH-L1.

After confirming the co-precipitation of UCH-L1 and one of the eIF4F subunits, eIF4A (Figure 2B), we analyzed the association of UCH-L1 with the translation machinery through polysome analysis. We observed that UCH-L1 cofractionated with components of the eIF4F preinitiation complex, but not with polysomes (Figure 2C). We did not observe a significant change in the polysome profiles in cells with or without UCH-L1 (supplemental Figure 2). As we identified eIF4G and eIF4A (2 of the 3 eIF4F subunits) in our proximity proteomic data, we hypothesized that association of UCH-L1 with this complex may affect its assembly. The UCH-L1-induced repression of 4EBP1 phosphorylation would be expected to impair eIF4F assembly and protein biosynthesis. We therefore examined eIF4F assembly through pulldowns with beads coupled with m⁷GTP that make up the 5′ cap on mRNA. As eIF4E directly binds to m⁷GTP, we normalized the recovered complexes to the level of this protein and examined amounts of eIF4G and eIF4A bound. Compared with control, expression of UCH-L1 increased the pulldown of eIF4G and eIF4A bound to eIF4E in 4 different cell lines (Figure 2D; supplemental Figure 3A-C). UCH-L1 was also retrieved with m⁷GTP beads, with less retrieved from cells expressing the catalytic mutant enzyme (C90A). We found no effect of UCH-L1 on the total level of eIF4F subunits in cell lines and mouse tissues, indicating that it is not affecting the degradation rate of these proteins (supplemental Figure 4A-D). As assembly and activity of eIF4F is also influenced by the activity of MNK1 on eIF4E^27,28  and S6 kinase, AKT, and MEK on eIF4B,²⁹  we examined the effect of UCH-L1 on these modifications, but found no such effect (supplemental Figure 4A-C). These data indicate that although UCH-L1 inhibits the phosphorylation of the eIF4F inhibitor 4EBP1, it results in increased levels of the translation initiation complex and increased protein translation.

Recent reports have indicated a key link between the oncogenic activity of MYC and the eIF4F complex including enhancing lymphomagenesis^5,7,30,31  and affecting chemoresistance.^7,32  We therefore hypothesized that UCH-L1 may play a role in MYC-driven lymphomagenesis. We previously found that transgenic UCH-L1 accelerated the development of lymphomas in Eμ-myc mice.¹⁰  Whether UCH-L1 catalytic activity is required for this acceleration, and whether endogenous UCH-L1 contributes to Myc-driven lymphomagenesis is not known.

To address the role of its catalytic activity in vivo, we generated novel transgenic mice carrying catalytically inactive UCH-L1 (Figure 3A; Uchl1^TgC90A) using an identical approach to that used to generate previously reported single-integration Uchl1^TgWT mice.¹⁰  After activation of the Uchl1^TgC90A transgene, relative expression of the WT and C90A transgenes was similar in various tissues analyzed, notably the skin, spleen, and purified splenic B cells (Figure 3B-C). As we previously showed that UCH-L1 regulates mTOR complex assembly,¹²  we examined the phosphorylation of mTOR substrates in splenocytes and purified B-cells from Uchl1^TgWT and Uchl1^TgC90A mice. As expected, we found that expression of the WT but not the C90A transgene increased the phosphorylation of AKT (mTORC2 substrate) and suppressed the phosphorylation of 4EBP1 and S6K (mTORC1 substrates) in cell lines and primary tissues (Figure 3D-E; supplemental Figure 5A-C). Surprisingly, the phosphorylation of 2 other mTORC2 substrates (PKC and SGK1) was not similarly affected (Figure 3E; supplemental Figure 5). This suggests that although UCH-L1 increases the overall level of mTORC2, the effect of this kinase is selectively directed to AKT over other substrates. The mechanisms that underlie this are unknown. These data support the use of these mice to dissect the role of UCH-L1 catalytic activity in vivo.

To assess the general oncogenic potential of wild-type and catalytic mutant UCH-L1, we assessed the tendency of wild-type and Uchl1 transgenic mice to develop tumors in response to the carcinogen 7,12-dimethylbenz[a]anthracene (DMBA). Newborn pups (day of life, 1-3) were exposed to a single topical dose of DMBA, and the tumor incidence was scored at 5 months of life. Compared with nontransgenic littermates, there was a significant increase in the incidence of skin tumors in Uchl1^TgWT mice, but not in Uchl1^TgC90A mice (Figure 4A-B). In mice that developed tumors, there was no difference in tumor size or the number of tumors per mouse across genotypes (data not shown). There was no difference in the incidence of lung tumors across the 3 cohorts (data not shown). We conclude that UCH-L1 catalytic activity is required for its ability to enhance DMBA induced skin tumors. To determine the role of its catalytic activity in lymphomagenesis, we crossed Uchl1^TgWT and Uchl1^TgC90A mice with the Eμ-myc model of lymphoma. As we observed previously, transgenic expression of wild-type UCH-L1 significantly accelerated the development of lymphoma in the Eμ-myc strain (Figure 4C). There was no change, however, in the lymphoma-free survival of double-mutant Eμ-myc/Uchl1^TgC90A mice. These data, together with the results of the DMBA carcinogenesis assay, indicate that catalytic activity is essential for the oncogenic effects of UCH-L1 in mice.

Although we have shown that UCH-L1 accelerates the development of B-cell lymphoma driven by Myc, we do not know whether its catalytic activity is required for the development of lymphomas. To address this, mice carrying a deletion in the Uchl1 gene (Uchl1^Dgen) were crossed with the Eμ-myc strain. As with the 2-other independent Uchl1 null strains,^33,34 Uchl1^Dgen mice develop a neurodegenerative state that ultimately is lethal.¹⁵  We therefore used relatively large cohorts to ensure adequate survival estimation, using the Kaplan-Meier method. Compared with heterozygous null mice on the Eμ-myc background, those with homozygous loss of Uchl1 had a significantly reduced incidence of lymphoma (Figure 4D). Notably, the rate of lymphoma onset was similar between the cohorts in animals less than approximately 20 weeks, after which no lymphomas were observed in Eμ-myc/Uchl1^Dgen/Dgen mice, with 52 mice surviving from between 20 and 106 weeks. These data lead us to conclude that UCH-L1 not only accelerates Myc induced lymphomas but is required for the full oncogenic potential of Myc in this model.

The accelerated development of Myc-driven lymphoma led us to hypothesize that UCH-L1 may affect the stability or function of MYC. However, we were unable to detect a direct interaction between UCH-L1 and MYC by co-immunoprecipitation in lymphoma cells (supplemental Figure 6A). Further, depletion or overexpression of UCH-L1 did not affect MYC (supplemental Figure 6B). To better understand their relationship, we performed a meta-analysis of UCHL1 and MYC mRNA levels across 324 gene expression studies involving more than 35 000 individual samples, using the R2 genomics analysis and visualization platform (http://r2.amc.nl). Although we had hypothesized a positive relationship, we instead found higher MYC values associated with lower UCHL1 (Figure 5A). We next looked at 5 primary mature B-cell lymphomas data sets containing at least 100 samples each. In 3 of the 5, there was no correlation, and a significant though weak positive correlation was observed in the other 2 (Table 4). To identify gene expression patterns within the cohorts of lymphomas with high UCHL1, we examined the expression of Hallmark gene sets, using parametric analysis of gene set enrichment. Surprisingly, this revealed a highly significant skewing of MYC target genes in all 5 data sets, with higher levels of “MYC target v1” consistently seen in UCHL1 high samples (Table 4; Figure 5B-C). This pattern was seen regardless of the presence or absence of a correlation between UCHL1 and MYC. Further, when cases with MYC rearrangements or those with high MYC were excluded, MYC target v1 genes continued to be enriched in the UCHL1 HI cases. These data led us to wonder whether UCH-L1 may, in fact, be a downstream effector of MYC, but not itself a MYC target.

To better understand the nature of the lymphomas arising in this model, we performed RNA microarray gene expression profiling on 5 unique spontaneous lymphomas from Uchl1^TgWT mice. We compared the profiles from these tumors with those obtained from purified normal mouse B-cell populations and lymphomas developing in other lymphoma models, including MYC-P110*,³⁵  Iμ-HABCL6,³⁶  and Lig4/p53 double-knockout mice.³⁷  Using a set of 3000 probes that efficiently discriminate normal B-cell subsets and lymphomas, we found that the lymphomas from genetically engineered models clustered together in a distinct pattern compared with normal B-cell populations (Figure 6A-B). Normal germinal center (GC) B cells did not cluster with any of the lymphoma samples from any model. We observed a striking similarity between the gene expression profiles of Uchl1^TgWT tumors and those from MYC-P110* mice that have concurrent enforced expression of C-Myc and a constitutively active form of the PI3K subunit P110.³⁵  Clustering reflected the visual similarity as the lymphomas from Uchl1^TgWT and MYC-P110* mice grouped separately from those from Iμ-HABCL6 and Lig4^−/− mice (Figure 6B-C). We performed mate-pair DNA sequencing of these same samples and found that in all but 1 sample, the Myc locus was intact (supplemental Figure 7, and data not shown). One sample contained a complex rearrangement of the C-Myc locus that juxtaposed the immunoglobulin heavy-chain with Myc. We used quantitative reverse transcription polymerase chain reaction to examine the level of Myc, and found that the samples had between 3- and 37-fold increased Myc compared with GC B cells purified from immunized wild-type mice (Figure 6D). The highest expression was seen in the sample carrying the Myc rearrangement (supplemental Figure 7). These data suggest that deregulated expression of UCH-L1 produces lymphomas with gene expression similar to those with genetically driven excessive MYC and PI3K signaling independent of Myc genomic alterations.

Despite being one of the first discovered de-ubiquitination enzymes, the molecular functions of UCH-L1 have been enigmatic. Here we use an unbiased proteomics approach to identify novel pathways in which UCH-L1 may play a role, and have found an unexpected function in directing the assembly of the eIF4F translation initiation complex. These findings address an important question regarding the oncogenic activity of UCH-L1. We previously showed that UCH-L1 disrupts the assembly of mTORC1 in vitro and in vivo,¹²  the activity of which is strongly associated with malignancy because of its roles in macromolecule biosynthesis. How then can UCH-L1 both inhibit mTORC1 and promote cancer? Our finding that UCH-L1 associates with and promotes the assembly of eIF4F provides a mechanism by which it bypasses mTORC1 and the negative effects of its suppression on protein biosynthesis. Interestingly, although we show here that protein synthesis is increased in response to UCH-L1 expression, we previously found that UCH-L1 expression leads to a reduction in cell size similar to the effect seen with rapamycin.¹²  This indicates that although UCH-L1 may bypass the role of mTORC1 on protein synthesis, it does not affect the global effect of mTORC1 reductions on metabolism. This is not unexpected, given that the phosphorylation of 4EBP1 and S6K have independent effects on cell size and that the latter has other downstream effects that affect cell growth such as lipid and nucleotide biosynthesis.^38,39 

The activity through the PI3K-mTOR-AKT pathways plays an important role in the biology of B-cell lymphomagenesis. In MYC/P110* mice, the co-activation of MYC and PI3K signaling in murine germinal center B cells leads to the development of B-cell lymphomas, with a striking resemblance to human Burkitt lymphoma.³⁵  The similar gene expression pattern we observe in the lymphomas from Uchl1^TgWT and MYC/P110* mice may partially reflect the enhanced mTOR-AKT signaling seen in both models. We provide further evidence, derived from gene expression profiles in human lymphomas, that high levels of UCH-L1 may also promote a MYC-like signature despite the lack of association with MYC expression. The mechanisms that may underlie this effect are unknown. UCH-L1 was recently shown to associate with chromatin in prostate cancer cells and HEK293T cells.⁴⁰  Chromatin immunoprecipitation and sequencing found that most of the regions occupied by UCH-L1 relate to either telomeric or intergenic regions, although it was also found to reside in the proximity of a small number of genes that clustered in the ρ signaling pathway. The UCH-L1 proximity proteome identified several proteins with roles in transcription that may also play a role in these events.

There is a growing body of evidence linking the oncogenic activity of MYC with the regulation of protein biosynthesis. Overexpression of the eIF4E cap binding protein is sufficient to strongly accelerate lymphomagenesis in the Eμ-myc model.⁷  A feed-forward loop also exists in which MYC increases the expression of eIF4F subunits, and the translation of MYC itself is further enhanced by increasing levels of the translation initiation complex.⁶  Reinforcing the importance of protein synthesis in MYC’s oncogenic activity, depletion of eIF4E using shRNA, or pharmacological suppression of eIF4F activity using the drug silvestrol, suppresses Eμ-myc lymphomagenesis.^5,6  Our data linking UCH-L1 with enhanced assembly of eIF4F and its requirement for MYC-driven lymphoma in this model are further consistent with this important connection.

The authors thank Jan van Deursen and the members of the Mayo Clinic Transgenic and Knockout Core Facility for their help with generating mice. P.J.G. is a former ASH Basic Research Scholar, a past recipient of the Howard Hughes Medical Institute Physician Scientist Early Career Award, and a former Harriet H. Samuelsson Foundation Pediatric Cancer Research Scientist.

The work was funded by the National Institutes of Health, National Cancer Institute (CA151351) (P.J.G.), the Multiple Myeloma Research Foundation (P.J.G.), Gabrielle's Angel Foundation for Cancer Research (P.J.G.), the Hyundai Hope on Wheels Foundation (P.J.G.), the T. Denny Sanford Pediatric Collaborative Research Fund Award (P.J.G. and K.J.R.), and the National Institutes of Health, National Institute for General Medical Sciences (P20GM103620, Protein Biochemistry Core) (K.J.R.).

Contribution: S.H., T.B., Q.L., D.G.M., and H.J. performed experiments; W.H., S.H.J., G.V., K.J.R., and P.J.G. designed the experiments, supervised their completion, and analyzed the data; and P.J.G. designed the study and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Paul J. Galardy, Mayo Clinic, 200 First Street SW, Guggenheim 15, Rochester, MN 55905; e-mail: galardy.paul@mayo.edu.