A novel effect of thalidomide and its analogs: suppression of cereblon ubiquitination enhances ubiquitin ligase function

Abstract

The immunomodulatory drug (IMiD) thalidomide and its structural analogs lenalidomide and pomalidomide are highly effective in treating clinical indications. Thalidomide binds to cereblon (CRBN), a substrate receptor of the cullin-4 really interesting new gene (RING) E3 ligase complex. Here, we examine the effect of thalidomide and its analogs on CRBN ubiquitination and its functions in human cell lines. We find that the ubiquitin modification of CRBN includes K48-linked polyubiquitin chains and that thalidomide blocks the formation of CRBN-ubiquitin conjugates. Furthermore, we show that ubiquitinated CRBN is targeted for proteasomal degradation. Treatment of human myeloma cell lines such as MM1.S, OPM2, and U266 with thalidomide (100 μM) and its structural analog lenalidomide (10 μM) results in stabilization of CRBN and elevation of CRBN protein levels. This in turn leads to the reduced level of CRBN target proteins and enhances the sensitivity of human multiple myeloma cells to IMiDs. Our results reveal a novel mechanism by which thalidomide and its analogs modulate the CRBN function in cells. Through inhibition of CRBN ubiquitination, thalidomide and its analogs allow CRBN to accumulate, leading to the increased cullin-4 RING E3 ligase-mediated degradation of target proteins.—Liu, Y., Huang, X., He, X., Zhou, Y., Jiang, X., Chen-Kiang, S., Jaffrey, S. R., Xu, G. A novel effect of thalidomide and its analogs: suppression of cereblon ubiquitination enhances ubiquitin ligase function.

The immunomodulatory drug (IMiD) thalidomide was originally known as a teratogen but is now used as a treatment for several clinical indications, including multiple myeloma (1). However, the mechanism by which the drugs affect cellular function has only begun to unfold with the identification of cereblon (CRBN) as a primary target of thalidomide (2). CRBN mRNA is highly expressed in cerebellum (3, 4) as well as many cancerous cells, such as lymphoma and multiple myeloma (5). This suggests that CRBN may play important roles in regulating cellular function in both the normal and disease state.

CRBN is a component of a cullin-4 really interesting new gene (RING) E3 ligase (CRL4) complex. This complex comprises regulator of cullins-1 (ROC1), cullin-4A (Cul4A) or cullin-4B, damage-specific DNA-binding protein 1 (DDB1), and CRBN (2). Previously, CRL4 complexes containing ROC1, Cul4A/B, DDB1, and diverse DDB1-binding substrate receptors such as DNA damage-binding protein 2 (DDB2) (6) and the simian virus 5 V protein (7, 8) were described. CRBN seems to also function as a substrate receptor (7). The finding that thalidomide binds to CRBN suggested that thalidomide likely affects the E3 ligase activity of the CRBN-containing CRL4 (CRL4-CRBN) complex. Indeed, thalidomide and its analogs inhibit the ubiquitination of CRBN (2, 9). Based on these results, it was suggested that thalidomide is an inhibitor of the CRL4-CRBN E3 ligase complex (2).

In addition to ubiquitinating their substrates, E3 ligases can also undergo autoubiquitination (10–16). Ubiquitination of the substrate receptor with a K48-linked polyubiquitin chain leads to its degradation, providing an important mechanism by which E3 ligases can trigger the replacement of their substrate receptors (14, 17). Substitution of substrate receptors on CRL4 complexes is therefore enhanced by ubiquitination and degradation of the substrate receptor, allowing a different substrate receptor to bind the adaptor protein and to form different E3 ligases (14). It is not known if CRBN is subjected to K48-linked ubiquitination and if this is the type of ubiquitination affected by thalidomide and its analogs.

Recent studies have identified ubiquitination targets of CRL4-CRBN complex. Two transcription factors, IKZF1 (Ikaros) and IKZF3 (Aiolos), were shown to be ubiquitinated by CRL4-CRBN (18, 19). Binding of these proteins to CRBN and their ubiquitination is induced by lenalidomide treatment, suggesting that thalidomide and its analogs can alter the specificity of the CRBN substrate adaptor so that it binds to these otherwise poor targets of CRBN. In this case, thalidomide and its analogs are associated with activation rather than inhibition of CRL4-CRBN E3 ligase activity (18, 19). Enhanced ubiquitination of these two essential transcription factors by this E3 ligase and their subsequent degradation by the proteasome lead to the death of myeloma cells (18, 19). This is a recently discovered mechanism by which IMiDs such as thalidomide and lenalidomide are used to treat myeloma patients (18, 19). Another target of this E3 ligase is homeobox protein MEIS2 (20). MEIS2 is displaced from CRBN due to thalidomide and its analogs binding to the same site on CRBN. As a result, MEIS2 is less ubiquitinated in response to thalidomide and its analogs (20). Calcium-activated potassium channel subunit α-1 (BK_Ca) and 5′-AMP-activated protein kinase catalytic subunit α-1 (AMPKα1), which are known CRBN-binding proteins, were recently shown to be ubiquitinated in a CRBN-dependent manner (21). However, how they are regulated by thalidomide and its analogs are not known.

Although thalidomide clearly influences the ubiquitination of various target proteins of the CRL4-CRBN E3 ligase complex, the significance of the thalidomide-induced blockade of CRBN ubiquitination has only recently been explored. A recent study proposed that IMiDs lenalidomide and pomalidomide induce a slight reduction of CRBN levels in several myeloma cell lines, which raises the possibility that the effects of lenalidomide and pomalidomide relate to the change of CRBN protein levels (22).

Here we explore the effect of thalidomide and its analogs on CRBN. We show that thalidomide and its analogs prevent CRBN from accumulating K48-linked polyubiquitin chains. These chains target CRBN for proteasomal degradation. In contrast to previous studies, we show that inhibition of K48-linked polyubiquitin chains on CRBN by thalidomide results in CRBN accumulation. This increased CRBN level is associated with increased CRL4-CRBN E3 ligase activity. Furthermore, we show that thalidomide and its analogs directly regulate CRBN protein levels in myeloma cell lines. This increased CRBN level potentiates the drug sensitivity of multiple myeloma cells. Taken together, these data identify an unappreciated mechanism by which thalidomide and its analogs can influence the CRL4-CRBN E3 ligase. By increasing CRBN levels, thalidomide and its analogs increase CRL4-CRBN activity in cells. These data suggest that thalidomide and its analogs constitute a novel class of E3 ligase stimulators that function by blocking the ubiquitination and degradation of an E3 ligase substrate adaptor and suggest new strategies for developing compounds that enhance E3 ligase activity.

MATERIALS AND METHODS

Materials

Human embryonic kidney (HEK) 293T cell line, human cervical cancer cell line HeLa, mouse neuroblastoma cell line N2a, myeloma cell line MM1.S, and U266 were from American Type Culture Collection (Manassas, VA, USA), and OPM2 myeloma cells were from DSMZ (Leibniz Institute DSMZ–German Collection of Microorganisms and Cell Cultures, Braunschweig, Germany). The ubiquitin and β-actin antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA); K48- and K63-linkage-specific ubiquitin antibodies from Millipore (Billerica, MA, USA); DDB1, Cul4A, and ROC1 antibodies from Abcam (Cambridge, MA, USA); FLAG M2 antibody from Sigma-Aldrich (St. Louis, MO, USA); BK_Ca antibody from University of California, Davis/National Institutes of Health NeuroMab facility (Davis, CA, USA); CRBN polyclonal antibody from Novus Biologicals (Littleton, CO, USA); glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody from HuaAn Biotechnology (Hangzhou, China); and secondary antibodies from Beyotime Biotechnology (Haimen, Jiangsu, China). Thalidomide, lenalidomide, and pomalidomide were from Selleck Chemicals (Houston, TX, USA), MG132 from Santa Cruz Biotechnology, proteasome inhibitor cocktail tablets from Roche (Indianapolis, IN, USA), and cycloheximide (CHX) from Sigma-Aldrich.

cDNA cloning

Total RNA was isolated from HEK293T (human) and N2a (mouse) cells with TRIzol reagent (Life Technologies, Carlsbad, CA, USA) according to the manufacturer’s instructions. The cloning of human FLAG-Strep tagged CRBN (FS-CRBN), Strep-CRBN, and mouse FS-mCRBN was carried out according to a method described previously (21). The wild-type (WT) Cul4A plasmid and dominant-negative Cul4A mutant plasmid, which contains the N-terminal domain for DDB1 binding but not the C-terminal domain for binding with ROC1 and the ubiquitin conjugating enzyme E2, were gifts from Pengbo Zhou’s laboratory (Weill Medical College, Cornell University). All positive clones were confirmed by sequencing.

Inhibition of CRBN ubiquitination with thalidomide and its analogs

HEK293T cells were grown in high-glucose DMEM supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 µg/ml streptomycin, and transfected with FS-CRBN or FS-mCRBN. Cells were treated with DMSO, thalidomide (100 µM), lenalidomide (10 µM), or pomalidomide (2 µM) for 1 h followed by the addition of MG132 (10 µM) for 3 h prior to cell lysis. CRBN was purified under the denaturing conditions in the presence of SDS, Triton X-100, and proteasome inhibitor cocktail, and the nonspecific binding proteins were removed. Western blotting was carried out according to a method described previously (21).

Dual purification of CRBN associated with the CRL4 E3 ligase complex

FLAG-Cul4A and Strep-CRBN plasmids were cotransfected into HEK293T cells using the calcium phosphate transfection method. The cells were treated with DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for 4 h, with MG132 (10 μM) added to the cells for the last 3 h. Cells were collected, washed with ice-cold PBS, and lysed. FLAG-Cul4A and its associated proteins were immunoprecipitated with the FLAG M2 antibody according to a previously described method (23). The immunoprecipitates were further purified with Strep-Tactin agarose beads (IBA GmbH, Goettingen, Germany) in the denaturing condition to obtain CRBN that was associated with the CRL4 E3 ligase complex. The dual purified samples were blotted with ubiquitin and CRBN antibodies.

Immunoprecipitation of endogenous CRBN in multiple myeloma cells

U266 cells were first treated with DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for 24 h, and then with MG132 (10 μM) for 12 h. Cells were lysed in modified RIPA buffer (0.1 M Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, and freshly added proteasome inhibitor cocktail) on ice. Cell lysates were incubated with protein A/G beads for 4 h to remove nonspecific interacting proteins and then incubated with a rabbit anti-CRBN polyclonal antibody (0.5 μg for about 1 mg of total proteins) overnight at 4°C. The samples were further incubated with 20 μl of protein A/G agarose beads for 4 h. The beads were washed 4 times with modified RIPA buffer and proteins were eluted with Laemmli sample buffer at 98°C. The eluate was combined and separated on an SDS-PAGE and blotted with an anti-ubiquitin antibody. Cell lysates were blotted with CRBN and GAPDH antibodies to show protein expression levels.

IMiD treatment in vitro and in cells

For the in vitro IMiD treatment of cell lysates, three 10 cm Petri culture dishes of HEK293T cells expressing FS-CRBN were lysed on ice for 15 min in a lysis buffer containing 0.1 M Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, and freshly prepared proteasome inhibitor cocktail, and sonicated with a Branson sonifier (Emerson, Danbury, CT, USA) for 10 s on ice. Cell lysates were centrifuged at 17,900 g for 10 min at 4°C. The supernatant was collected and equally divided into 3 Eppendorf tubes. Equal volumes of DMSO, thalidomide (0.5 mM), and lenalidomide (0.1 mM), respectively, were added to each tube, followed by incubation at 4°C for 1 h. For the in-cell treatment, HEK293T cells expressing FS-CRBN were treated with DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for 4 h and lysed in the lysis buffer. Purification of FS-CRBN and its interacting proteins from in vitro and in-cell treatment was carried out under the native condition in the absence of detergent to maintain protein interacting complex according to a previous publication (21). For the reciprocal immunoprecipitation, HEK293T cells were transfected with FLAG-Cul4A and treated with DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for 4 h. Cells were lysed and FLAG-Cul4A was immunoprecipitated with FLAG M2 antibody. The immunoprecipitates were blotted for endogenous CRBN and FLAG-Cul4A. The cell lysates were blotted for FLAG-Cul4A, CRBN, and GAPDH.

Multiple myeloma cells were grown in RPMI 1640 medium supplemented with 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin. Cells were treated with thalidomide, lenalidomide, and pomalidomide at the indicated concentration for desired length, and the whole-cell lysate was used for Western blotting analysis or flow cytometry experiments.

CHX treatment and CRBN stability measurement

For the CHX treatment, HEK293T cells were transfected with FS-CRBN and equally divided into 6 wells in a 6-well plate 24 h after transfection. Cells were treated with CHX (10 μg/ml) along with DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for different time. The whole-cell lysates were blotted for FLAG and GAPDH.

For the CRBN stability measurement, HEK293T cells were transfected with FLAG-Strep-CRBN WT or Y384A/W386A mutant (YW/AA) and then equally divided to each well of a 6-well plate 24 h after transfection. The cells were treated with thalidomide (100 μM) or lenalidomide (10 μM) for different times. The whole-cell lysates were blotted for FLAG and GAPDH.

siRNA knockdown of CRBN

CRBN was knocked down in HeLa cells by siRNA using Lipo RNAiMax (Life Technologies) according to the manufacturer’s instructions. The siRNA sequences for the control and CRBN were 5′-UUCUCCGAACGUGUCACGUTT-3′ and 5′-AAGUGUCUCAUGCACAUAUCCAUGA-3′, respectively. The cells were collected 48 h after transfection and cell lysates were immunoblotted for endogenous proteins.

Generation of MM1.S cell lines expressing green fluorescent protein or CRBN

MM1.S cells stably expressing green fluorescent protein (GFP) or CRBN were established by infecting MM1.S cells with the pLX304-GFP (RNAi Consortium at Broad Institute, Cambridge, MA, USA) or pLX304-CRBN (DNASU plasmid repository, Tempe, AZ, USA) lentiviral particles. The infected cells were grown under the selection with blasticidin (8 μg/ml) for 2 wk prior to conducting the experiments. About 84% of MM1.S cells were GFP-positive in the control sample at the completion of the experiment.

Measurement of the apoptotic cells and cell proliferation

MM1.S cells expressing GFP or CRBN were cultured in RPMI 1640 medium with 10% FBS in the presence of DMSO or pomalidomide at the indicated concentration for 4 d, stained with ToPro-3 (Life Technologies) according to the manufacturer’s instructions, and analyzed in a BD flow cytometer with a fluorescence excitation and emission of 650 and 660 nm, respectively. The data were analyzed with FlowJo software (Tree Star, Ashland, OR, USA). The cells undergoing apoptosis were calculated based on the ToPro-3 signal. MM1.S cells after DMSO or pomalidomide treatment were stained with trypan blue and added to a cell-counting plate. The live cells were counted under microscope. The relative viable cells upon drug treatment were normalized to the DMSO-treated samples. The paired 2-tailed Student’s t test was performed to calculate the P value.

Statistical analysis

The experiments were repeated 3 to 6 times as indicated in the figure legends. Densitometry of the corresponding bands was measured using ImageLab (Bio-Rad, Hercules, CA, USA). The relative densitometry of the protein bands was normalized to that of the corresponding bands from the loading control, GAPDH or β-actin, and the average of 3 to 6 replicates was determined. To compare the effects of IMiDs on the protein level, the value was further normalized to the DMSO-treated sample. Error bars in the graphs represent standard deviation; P values were evaluated using the paired 2-tailed Student’s t test.

RESULTS

Thalidomide and its analogs inhibit the ubiquitination of both human and mouse CRBN

It has long been a mystery why thalidomide does not cause birth defects in mice and rats, whereas it does in humans, zebrafish, and chickens (24, 25), even though the CRBN sequences between humans and mice are very similar (Supplemental Fig. 1). We considered the possibility that rodent CRBN is not affected by thalidomide. To test this idea, we overexpressed FLAG-Strep tagged human or mouse CRBN (FS-CRBN) in HEK293T cells and treated the cells with DMSO, thalidomide, lenalidomide, or pomalidomide followed by treatment with a proteasome inhibitor, MG132. FS-CRBN was isolated with Strep-Tactin agarose beads under denaturing conditions, which removes most nonspecific binding proteins. Ubiquitinated CRBN was detected by an anti-ubiquitin immunoblotting. Upon proteasome inhibition, human CRBN ubiquitination is significantly increased, and this effect is blocked by pretreatment with thalidomide, lenalidomide, or pomalidomide (Fig. 1A), consistent with previous studies (2, 9). Similarly, the ubiquitination of mouse CRBN is also inhibited by the drug treatment (Fig. 1B). The immunoblotting of the whole-cell lysates was used to show the similar protein level in different samples. The result showed that after the treatment of transfected HEK293T cells with proteasome inhibitors or IMiDs, CRBN protein levels were either similar to or slightly higher than that of the DMSO-treated samples.

Figure 1.

IMiDs thalidomide, lenalidomide, and pomalidomide inhibit CRBN ubiquitination. Thalidomide and its structural analogs inhibit human (A) and mouse (B) CRBN ubiquitination. HEK293T cells (in 10 cm dishes) were transfected with 10 μg of a plasmid expressing an FS-CRBN or FS-mCRBN by calcium phosphate transfection. Cells were treated with DMSO (−), thalidomide (Thal 100 μM), lenalidomide (Len 10 μM), or pomalidomide (Pom 2 μM) for 1 h prior to the addition of a proteasome inhibitor, MG132 (10 μM), for 3 h. Note that total treatment time for thalidomide, lenalidomide, or pomalidomide was 4 h. Cells were briefly washed with ice-cold PBS twice, collected, and lysed. FS-CRBN and FS-mCRBN were affinity-purified using Strep-Tactin agarose beads under denaturing conditions, separated in SDS-PAGE, and blotted with anti-ubiquitin or anti-FLAG antibodies. The cell lysates were blotted for FLAG (CRBN) and GAPDH to show the expression level of CRBN protein in the samples prior to affinity purification. IB, immunoblotting; Ub, ubiquitin.

Thalidomide and lenalidomide do not disrupt the CRL4-CRBN E3 ligase complex

We next sought to understand why thalidomide and its analogs inhibit CRBN ubiquitination. Because CRBN ubiquitination is likely mediated by transfer of ubiquitin from ubiquitin conjugating enzymes, E2s, interacting with the ROC1 component of the CRL4-CRBN E3 ligase complex, we considered the possibility that thalidomide and lenalidomide disrupt this complex, thereby preventing CRBN from ubiquitination. To test this, we incubated cell lysates from CRBN-overexpressing HEK293T cells with DMSO, thalidomide, or lenalidomide and isolated CRBN and its interacting partners under native conditions. Although we used high concentrations of thalidomide and lenalidomide during the in vitro incubation, immunoblotting of the pulled-down complex shows that there is almost no change in the relative intensity of each E3 ligase component for samples treated with or without thalidomide and lenalidomide (Fig. 2A). This indicates that thalidomide and lenalidomide do not dissociate the CRL4 E3 ligase complex. Similar results were obtained when cells expressing FS-CRBN were treated with thalidomide and lenalidomide before lysis (Fig. 2B).

Figure 2.

Effect of thalidomide (Thal) and lenalidomide (Len) on the CRL4-CRBN E3 ligase complex. A) Thalidomide and lenalidomide do not disrupt the CRL4-CRBN E3 ligase complex in vitro. HEK293T cells were transfected with an FS-CRBN plasmid and lysed 48 h after transfection. The lysate was treated with DMSO (−), thalidomide (0.5 mM), or lenalidomide (0.1 mM) for 1 h at 4°C and the E3 ligase complex was purified under native conditions. The samples were resolved in SDS-PAGE and proteins were immunoblotted. *Nonspecific band on the anti-Cul4A blot. B) Thalidomide and lenalidomide do not prevent the formation of the CRL4-CRBN E3 ligase complex in cells. HEK293T cells were transfected with a plasmid expressing FS-CRBN using the calcium phosphate transfection method. At 24 h posttransfection, cells were first treated with DMSO (−), thalidomide, or lenalidomide for 1 h, and then DMSO (−) or MG132 (10 μM final concentration) were added for 3 h. Cells were lysed and FS-CRBN was purified with Strep-Tactin agarose beads under the native condition. The purified samples were immunoblotted for FLAG, DDB1, and Cul4A. C) Thalidomide and lenalidomide do not affect the formation of the CRL4-CRBN E3 ligase complex in cells. HEK293T cells were transfected with 10 μg of FLAG-Cul4A plasmid, treated with DMSO (−), thalidomide (100 μM), or lenalidomide (10 μM) for 4 h. FLAG-Cul4A and its interacting proteins were immunoprecipitated with FLAG M2 antibody and immunoblotted for FLAG-Cul4A and endogenous CRBN. Cell lysates were also blotted with FLAG, CRBN, and GAPDH antibodies to show similar protein expression level in different samples. IB, immunoblotting; ROC1, regulator of cullins-1.

To further validate that the E3 ligase complex is not affected by IMiDs, we performed a reciprocal immunoprecipitation. In this experiment, we overexpressed FLAG-Cul4A in HEK293T cells, treated the cells with DMSO, thalidomide, and lenalidomide, and then immunoprecipitated Cul4A with FLAG M2 antibody. The immunoblotting of endogenous CRBN showed that thalidomide and lenalidomide did not influence the association of the E3 ligase complex in cells (Fig. 2C).

CRBN ubiquitination contains the K48-linked polyubiquitin chain

We next sought to determine the type of polyubiquitin chain linkage on CRBN that is inhibited by thalidomide and lenalidomide. We observed that ubiquitin conjugates were increased on CRBN after treatment with a proteasomal inhibitor (Fig. 1A, B, lane 2 and lane 3), indicating this type of polyubiquitin is most likely the K48-linked polyubiquitin, which is normally removed by the 26S proteasome (26). We next sought to directly test whether the ubiquitin that is lost in CRBN upon thalidomide treatment is composed of K48-linked polyubiquitin chains. To test this, we blotted affinity-purified human CRBN with a K48-linkage-specific ubiquitin antibody and found that the signal in the blot is significantly reduced after thalidomide treatment (Fig. 3A). However, immunoblotting with the K63-linkage-specific ubiquitin antibody did not detect any signal for the same set of samples (data not shown). The immunoblotting of the whole-cell lysates was also carried out to show protein expression level prior to affinity purification. It should be noted that polyubiquitin chains may contain mixed chain linkages. We cannot completely rule out the possibility that other types of ubiquitin chain linkage may exist on CRBN. However, this result indicates that K48-linked polyubiquitin chains are a prominent, and possibly predominant form of ubiquitin on CRBN, and thalidomide and its analogs block the appearance of this type of polyubiquitin chain.

Figure 3.

Thalidomide and lenalidomide inhibit the K48-linked polyubiquitin chains on CRBN. A) Thalidomide inhibits the formation of K48-linked polyubiquitin chains on CRBN. HEK293T cells were transfected with 10 μg of FS-CRBN plasmid, treated with DMSO (−) or thalidomide (10 and 100 μM) for 1 h prior to the addition of MG132 (10 μM) for 3 h. FS-CRBN was affinity-purified under denaturing conditions with Strep-Tactin agarose beads and the purified samples were blotted with anti-K48-ubiquitin and anti-FLAG antibodies. The cell lysates were blotted for FLAG (CRBN) and GAPDH. B) Thalidomide and lenalidomide inhibit the ubiquitination of CRBN associated with CRL4 E3 ligase. FLAG-Cul4A and Strep-CRBN were cotransfected to HEK293T cells and treated with DMSO (−), thalidomide (Thal 100 μM), lenalidomide (Len 10 μM), and MG132 (10 μM). FLAG-Cul4A and its interacting proteins were immunoprecipitated with FLAG M2 antibody and then the immunoprecipitates were further pulled down with Strep-Tactin agarose beads for Strep-CRBN under denaturing condition. The double affinity purified samples were immunoblotted with anti-ubiquitin and anti-CRBN antibodies. Blotting of the FLAG M2 immunoprecipitates and the cell lysates was also shown. IB, immunoblotting.

Thalidomide and lenalidomide inhibit the ubiquitination of CRBN associated with CRL4 E3 ligase complex

We next asked whether the ubiquitination of CRBN associated with the CRL4 E3 ligase is inhibited by IMiDs. To do so, we performed a dual purification experiment. FLAG-Cul4A was transfected alone or cotransfected with Strep-CRBN and then treated with DMSO, thalidomide, or lenalidomide and MG132. We first immunoprecipitated Cul4A and its associated E3 ligase complex with FLAG M2 antibody from cell lysates and then used Strep-Tactin agarose beads to further pull down CRBN from the first round immunoprecipitates under denaturing conditions. The samples after immunoprecipitation and Strep-Tactin pull-down were immunoblotted (Fig. 3B). The decreased ubiquitin signal suggests that thalidomide and lenalidomide inhibited the ubiquitination of CRBN that is associated with the E3 ligase complex. The cell lysates and the first round immunoprecipitates were also blotted for CRBN and Cul4A (Fig. 3B). In the cell lysates, after Cul4A transfection, the CRBN protein level was decreased and then slightly increased upon the addition of MG132. Treatment with thalidomide and lenalidomide further enhanced CRBN protein level. The blotting of immunoprecipitates shows that the CRL4-CRBN complex was not significantly affected by these two drugs, which is similar to the result shown in Fig. 2.

Thalidomide and lenalidomide increase the stability of exogenously expressed CRBN

The finding that thalidomide inhibits K48-linked ubiquitination suggests that treatment of cells with thalidomide and its analogs may lead to accumulation of CRBN due to reduced ubiquitination and subsequent enhanced CRBN stability. As a first test of this idea, we treated FS-CRBN–expressing HEK293T cells with DMSO, thalidomide, or lenalidomide in the presence of CHX to block synthesis of new proteins. We then blotted the whole-cell lysate with an anti-FLAG antibody to detect FS-CRBN. These experiments showed that the CRBN protein level was reduced rapidly when the samples were treated with CHX (Fig. 4A). However, when the cells were treated with thalidomide or lenalidomide along with CHX, the reduction of CRBN protein level was much slower (Fig. 4A). We further used thalidomide and lenalidomide to treat HEK293T cells expressing WT CRBN or a CRBN double mutant, Y384A/W386A (YW/AA). This mutant does not bind to thalidomide (2), and therefore we can use this mutant to study the effect of IMiDs on the CRBN stability. The results clearly showed that thalidomide and lenalidomide increase the protein level of the WT CRBN significantly but not the YW/AA double mutant (Fig. 4B). Together, these experiments indicate that CRBN is a labile protein, and blocking its ubiquitination using IMiDs stabilizes CRBN, consistent with the idea that IMiDs block ubiquitination-induced degradation of CRBN.

Figure 4.

Thalidomide (Thal) and lenalidomide (Len) increase CRBN stability. A) Thalidomide and lenalidomide reduce CRBN degradation. HEK293T cells were transfected with a FS-CRBN-expressing plasmid and then the transfected cells were equally divided into each well in 6-well plates. Cells were treated with CHX (10 μg/ml) in the presence of DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for the indicated time. The whole-cell lysates were blotted for FLAG and GAPDH. B) Thalidomide and lenalidomide increase the protein level of WT CRBN but not the thalidomide resistant Y384A/W386A (YW/AA) mutant. Plasmids expressing FS-CRBN WT or YW/AA mutant were transfected into HEK293T cells and then cells were equally divided into each well of a 6-well plate. The cells were treated with thalidomide (100 μM) or lenalidomide (10 μM) for indicated time. The whole-cell lysates were blotted for FLAG and GAPDH. IB, immunoblotting.

Thalidomide and lenalidomide increase CRBN protein levels in multiple myeloma cells

Because lenalidomide and pomalidomide are currently used for the treatment of myeloma, we next sought to examine whether these IMiDs can also inhibit CRBN ubiquitination and increase the endogenous CRBN level in multiple myeloma cell lines. To do so, we first tested the effect of IMiDs on the ubiquitination of endogenous CRBN in U266 myeloma cells. We immunoprecipitated the endogenous CRBN and blotted for ubiquitin. The immunoblotting showed that the ubiquitination of endogenous CRBN was reduced upon thalidomide and lenalidomide treatment and that the effect of lenalidomide on the ubiquitination of endogenous CRBN appeared to be more significant than thalidomide (Fig. 5A). To see whether the decreased ubiquitination can result in the increase of CRBN protein in myeloma cells, we treated 3 cell lines with IMiDs in the absence of proteasome inhibition, as this may reflect myeloma cells exposure to these compounds in a clinical setting. In these experiments, we first treated MM1.S and U266 myeloma cell lines with lenalidomide for 3 d and examined the endogenous CRBN levels at different time points with immunoblotting. These results showed that CRBN protein level was increased with the increase of lenalidomide treatment time, and this effect was very robust 72 h after the treatment (Fig. 5B). Therefore, in the following experiments, we treated multiple myeloma cell lines MM1.S, OPM2, and U266 with thalidomide or lenalidomide for 72 h and examined the endogenous CRBN level. We found that in these 3 cell lines, the CRBN protein levels were significantly increased after thalidomide or lenalidomide treatment and this effect was again more prominent for lenalidomide (Fig. 5C, D). The degree of the increase of the CRBN protein level upon IMiD treatment as marginally different in 3 myeloma cell lines. This likely reflects the rate at which CRBN becomes ubiquitinated, which may be slightly different in these cell lines. It should be noted that prolonged treatment with thalidomide and lenalidomide causes apoptosis of myeloma cells. This might also affect the magnitude of the increase in the protein level of CRBN. However, our experiments were carried out within 72 h of the treatment at which the cell proliferation was not significantly altered. Similar numbers of cells were used during the experiment and GAPDH was immunoblotted as a loading control in each cell line. These experimental conditions suggest that the increase of the protein level of CRBN is likely mainly induced by the reduction of its ubiquitination. To ascertain whether this increase is caused by inhibiting CRBN ubiquitination and not by increasing CRBN mRNA levels, we performed quantitative PCR experiments. Our results showed that none of the 3 multiple myeloma cell lines exhibited significant alterations in CRBN transcript levels upon lenalidomide treatment (Supplemental Fig. 2). These results support the idea that thalidomide and lenalidomide mediate the accumulation of endogenous CRBN protein in multiple myeloma cell lines by preventing CRBN from becoming ubiquitinated and degraded.

Figure 5.

Thalidomide (Thal) and lenalidomide (Len) treatment inhibits CRBN ubiquitination and increases CRBN levels in myeloma cells. A) Thalidomide and lenalidomide inhibit the ubiquitination of endogenous CRBN in myeloma cells. U266 cells were treated with DMSO, thalidomide, lenalidomide, and MG132, and endogenous CRBN was precipitated with an anti-rabbit CRBN antibody. The immunoprecipitates were blotted with an anti-ubiquitin antibody. The cell lysates were blotted with CRBN and GAPDH antibodies for loading control. IgG-HC is the heavy chain of the antibody. B) The effect of lenalidomide on the endogenous CRBN protein levels. Immunoblotting analysis of endogenous CRBN in multiple myeloma cells (MM1.S and U266) after lenalidomide (10 μM) treatment for different time. C) Endogenous CRBN levels are increased by thalidomide and lenalidomide treatment in multiple myeloma cell lines. Immunoblotting analysis of endogenous CRBN in multiple myeloma cells (MM1.S, OPM2, and U266) after DMSO, thalidomide (100 μM) and lenalidomide (10 μM) treatment. D) Quantification of immunoblotting was obtained from 6 replicated experiments shown in (C). **P < 0.01; ***P < 0.001. IB, immunoblotting; Ub, ubiquitin.

Increased CRBN protein levels result in enhanced CRL4-CRBN–mediated E3 ligase activity in multiple myeloma cells

Next we tested whether the increased CRBN level leads to increased functional CRL4-CRBN E3 ligase complexes in multiple myeloma cells. To do this, we measured the expression level of a CRBN target, BK_Ca. BK_Ca was originally identified as a CRBN-interacting protein in a yeast 2-hybrid screen (27). We subsequently showed that this protein is a target of the CRL4-CRBN E3 ligase (21), which is also consistent with another study (28). Treatment of MM1.S, U266, and OPM2 cells with thalidomide or lenalidomide for 72 h resulted in a marked reduction in endogenous BK_Ca levels (Fig. 6A, B). A time-dependent analysis showed that the decrease of BK_Ca occurred after the increase of CRBN (Supplemental Fig. 3). These data indicate that treatment with thalidomide and its analogs leads to increased CRL4-CRBN E3 ligase activity, which promotes the ubiquitination and degradation of its substrates, such as BK_Ca.

Figure 6.

Thalidomide (Thal) and lenalidomide (Len) enhance the degradation of a CRL4-CRBN substrate in multiple myeloma cells. A) Immunoblotting analysis of endogenous BK_Ca in multiple myeloma cells (MM1.S, OPM2, and U266) treated with DMSO, thalidomide (100 μM), or lenalidomide (10 μM) for 72 h. B) Quantification of relative BK_Ca levels from biological repeats. Experiments were repeated 3 times for MM1.S and U266 cells and repeated 4 times for OPM2. *P < 0.05; **P < 0.01; ***P < 0.001. C) Knockdown of CRBN (left panel) or transfection of dominant-negative Cul4A (right panel) increases the protein level of BK_Ca. Left panel: CRBN was knocked down in HeLa cells using siRNA. Right panel: DN-Cul4A was transfected to HEK293T cells by the calcium phosphate transfection method. Cells were lysed after 48 h and immunoblotted for BK_Ca, CRBN or Cul4A, and GAPDH. IB, immunoblotting.

To confirm that BK_Ca is indeed regulated by CRBN and its associated E3 ligase, we knocked down CRBN with siRNA in HeLa cells and immunoblotted for endogenous BK_Ca. These results showed that the BK_Ca level was significantly increased upon CRBN knockdown (Fig. 6C, left panel). In addition, expression of a dominant-negative Cul4A in HEK293T cells also increased endogenous BK_Ca (Fig. 6C, right panel). These results suggested that BK_Ca was indeed regulated by CRBN and its associated E3 ligase.

Our findings support the idea that the increased CRBN level contributes to the enhanced E3 ligase activity. It should be noted that an increase in CRBN may not always affect the protein level of CRBN-interacting proteins. Indeed, immunoblotting for endogenous AMPKα1, another known CRBN-interacting protein (29), did not show significant alteration in its protein level after thalidomide and lenalidomide treatment of myeloma cells (Supplemental Fig. 4). This result is consistent with a previous finding that the protein level of AMPKα1 is not changed upon expression of CRBN in HEK293FT cells (29). Thus, AMPKα1 levels in cells may not be regulated primarily via CRL4-CRBN complexes, at least in the cell lines that were tested.

Increased CRBN protein level improves the inhibition of the multiple myeloma cell growth upon pomalidomide treatment

To test whether the increased CRBN level has any effect on the apoptosis of multiple myeloma cells during drug treatment, we made 2 stable cell lines expressing GFP or CRBN, respectively. Western blotting showed that the CRBN-expressing stable cell line expressed substantially more CRBN than the control MM1.S cell line expressing GFP (Fig. 7A). Following treatment with pomalidomide at 3 different concentrations, CRBN-expressing MM1.S cells exhibited enhanced ToPro-3 positive signal, indicating the increase of apoptotic cells (Fig. 7B). The quantitative analysis of live cells showed that reduction in cell proliferation by drug treatment was greater in myeloma cells overexpressing CRBN than in those with GFP (Fig. 7C). This result demonstrated that the increase of CRBN level leads to enhanced sensitivity of MM1.S cells to pomalidomide.

Figure 7.

CRBN level modulates the sensitivity of multiple myeloma cells to pomalidomide (Pom). A) Western blotting analysis of CRBN protein level of MM1.S cells ectopically expressing exogenous GFP or CRBN. β-actin was used as a loading control. The relative CRBN level is indicated under the images. B) Flow cytometry analysis of the MM1.S cells in (A) treated with pomalidomide (0, 0.1, 1.0, and 10 μM) for 4 d and stained with ToPro-3. Data in each flow cytometry profile indicated the percentage of apoptotic and dead cells. C) The percentage of viable cells relative to those in the DMSO-treated samples (normalized percentage of viable cells) in (B) was determined by trypan blue exclusion and counting the live cells under a microscope. The P value is calculated using a paired 2-tailed Student’s t test. IB, immunoblotting.

DISCUSSION

Although considerable effort has been focused on identifying CRBN targets and how their ubiquitination is affected by thalidomide and its structural analogs, the significance of their regulation of CRBN ubiquitination has not previously been studied. Indeed, thalidomide and its structural analogs induce a prominent inhibitory effect on CRBN ubiquitination, suggesting that this process may contribute to their therapeutic effects in vivo. Our data, for the first time, show that CRBN acquires a K48-linked polyubiquitin chain that targets it for proteasome-mediated degradation. This effect is blocked by thalidomide, resulting in CRBN accumulation in cells, including multiple myeloma cells. The increase in CRBN levels leads to increased activity of CRL4-CRBN complex, facilitates the degradation of its targets, and increases the sensitivity of drug treatment.

Our results suggest that thalidomide and its analogs increase CRBN levels, which contrasts with an earlier study (22). This study examined various antimyeloma therapies, including bortezomib, dexamethasone, lenalidomide, and pomalidomide. These authors reported slightly reduced levels of CRBN after 5 d of treatment with lenalidomide and pomalidomide, which was measured indirectly, by immunoblotting CRBN from CRBN immunoprecipitates rather than immunoblotting the whole-cell lysates (22). However, earlier increased CRBN levels were evident in the same experiments (22) and in an earlier report (30). An additional basis for the discrepancy could relate to the cell death induced by the IMiDs after treatment for 5 d. This might account for the reduced levels of CRBN seen in their experiments. The data presented in our study clearly show that CRBN is subjected to turnover via a K48-linked ubiquitin modification, and IMiDs block this turnover, leading to increased CRBN levels under our experimental conditions.

Most small molecules that affect E3 ligase activity act to inhibit E3 ligase function. However, activators of E3 ligases have not been described. Thalidomide and its analogs appear to be the first class of compounds that function to enhance E3 ligase activity. This occurs by selectively blocking the ubiquitination of substrate receptors. Ubiquitination of substrate receptors is a mechanism by which different substrate receptors are removed from their tightly bound adaptor proteins, such as DDB1 (17). Thus, these drugs constitute a novel class of E3 ligase agonists that suppress the endogenous substrate receptor replacement, thereby resulting in increased levels of CRL4-CRBN complex and its E3 ligase activity.

The earliest data that suggested a molecular mechanism for thalidomide was the finding that these compounds block the ubiquitination of CRBN (2). The mechanism for this effect was not addressed, but it was inferred from these early studies that the function of thalidomide is to block the E3 ligase activity of CRL4-CRBN. Our data show that the activity of the complex is not inhibited following the treatment of thalidomide and its analogs. We find that the entire complex remains intact and is capable of degrading a target protein. This target protein BK_Ca is distinct from other targets, such as IKZF1 and IKZF3, in that BK_Ca readily binds CRBN in conditions that do not involve thalidomide and its analogs (27). Indeed, the proteomic screens that searched for proteins that were preferentially targeted by CRBN in a lenalidomide-dependent manner did not detect BK_Ca (18). Thus, this protein is a useful target for assessing the catalytic activity of CRL4-CRBN E3 ligase. Based on these findings, thalidomide and its analogs may not function as inhibitors of this ligase toward some targets.

However, it should be noted that the mechanism by which thalidomide and its analogs affect CRL4-CRBN-mediated ubiquitination is complex. Thalidomide and its analogs clearly inhibit CRBN ubiquitination (2, 21), but they also reduce the binding of CRBN to MEIS2, thus reducing MEIS2 ubiquitination (20). Thalidomide and its analogs also enhance CRBN binding to IKZF1 and IKZF3, thereby promoting their ubiquitination and degradation (18). In the case of BK_Ca, elevated CRBN level leads to enhanced degradation of this target. Thus, depending on the binding mechanism to CRBN, some targets will show increased ubiquitination and others show reduced ubiquitination following the treatment with thalidomide or lenalidomide.

What accounts for the loss of CRBN ubiquitination following the treatment of thalidomide and its analogs? Recent structural studies (20, 31) support a mechanism for the ability of thalidomide and its analogs to affect CRBN ubiquitination. Lenalidomide binds to a pocket in CRBN surrounded with several lysines, including Lys401, Lys406, Lys407, and Lys413. We recently showed that this C-terminal region is ubiquitinated in CRBN (21). Our current study also shows that ubiquitination of CRBN in complex with the CRL4-E3 ligase is inhibited by thalidomide and lenalidomide. Thus, binding of thalidomide and its analogs likely reduces the accessibility of these lysines to E2 for ubiquitination.

As described above, recent studies have shown that thalidomide and its analogs induce binding of CRBN to proteins that would otherwise not be substrates of CRL4-CRBN. In this manner, proteins such as IKZF1 and IKZF3 become targeted by CRBN in a lenalidomide-dependent manner. The mechanism that we describe is likely to synergize with this pathway. Thalidomide and its analogs block the ubiquitination and degradation of CRBN, thereby allowing more CRBN to accumulate and interact in a lenalidomide-dependent manner with IKZF1 and IKFZ3. Thus, stabilization of CRBN by thalidomide and its analogs is expected to enable more efficient degradation of IKZF1 and IKZF3.

Several studies of the activity of thalidomide and its analogs in patients and multiple myeloma cell lines provide further support for the idea that IMiDs do not inhibit CRBN function. One study showed that CRBN knockdown reduces lenalidomide sensitivity of a multiple myeloma cell line (32), and loss of CRBN is associated with clinical resistance to lenalidomide and pomalidomide (32, 33). If inhibition of CRL4-CRBN activity was the mechanism of lenalidomide, then CRBN knockdown would be expected to mimic lenalidomide treatment. Furthermore, a recent clinical study demonstrated that higher CRBN mRNA level is associated with better thalidomide response in patients with newly diagnosed multiple myeloma (34). Our experiments show that CRBN level is increased after thalidomide and lenalidomide treatment in the tested multiple myeloma cell lines, which increases the drug sensitivity of multiple myeloma cell lines. Thus, elevated CRBN levels and the corresponding increase in CRL4-CRBN activity likely mediate the effects of thalidomide and its analogs.

Interestingly, we noticed that lenalidomide has a greater inhibitory effect on CRBN ubiquitination and results in increased CRBN levels in cells than thalidomide. Lenalidomide appears to more clinically effective at a concentration much lower than that of thalidomide (35). Conceivably, the higher efficacy of lenalidomide reflects its improved ability to enhance CRBN accumulation, especially in cells that show low CRBN levels.

Taken together, our study suggests that thalidomide and its analogs may have an additional function to increase CRBN levels in cells, thereby potentiating its antimyeloma activity. Our study also shows that thalidomide and its analogs represent a novel class of compounds that lead to increased E3 ligase activity in cells and function through a previously undescribed mechanism of preventing the ubiquitination and degradation of a substrate receptor. Currently, there is a major interest in designing small molecules that interfere with ubiquitination in cells. It will be interesting to see if novel small molecules can be designed to enhance the activity of other ubiquitin ligases by preventing their ubiquitination.

Glossary

AMPKα1

5′-AMP-activated protein kinase catalytic subunit α-1

BK_Ca

calcium-activated potassium channel subunit α-1

CRBN

cereblon

CRL4

cullin-4 RING E3 ligase

CRL4-CRBN

cullin-4 RING cereblon E3 ligase

Cul4A

cullin-4A

CHX

cycloheximide

DDB1

damage-specific DNA-binding protein 1

DDB2

DNA damage-binding protein 2

FBS

fetal bovine serum

FS-CRBN

FLAG-Strep tagged human cereblon

FS-mCRBN

FLAG-Strep tagged mouse cereblon

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

GFP

green fluorescent protein

HEK

human embryonic kidney

IMiD

immunomodulatory drug

RING

really interesting new gene

ROC1

regulator of cullins-1

WT

wild-type