Background
Immunomodulatory drugs (IMiDs), such as lenalidomide, are therapeutically active compounds widely used in the treatment of multiple myeloma (MM) [
1]. Treatment with IMiDs results in significant effects on: immunomodulatory activities; anti-angiogenic activities; anti-inflammatory activities; anti-proliferation; pro-apoptotic effects; cell-cycle arrest; and inhibition of cell migration and metastasis [
2]. Although significant remissions in patients with MM have been induced with IMiDs, the molecular mechanism of IMiDs’ action has only recently unraveled.
Using immobilized thalidomide, Ito et al. identified cereblon (CRBN) and DNA damage-binding protein 1 (DDB1) as binding proteins and further demonstrated that CRBN was the primary target of thalidomide-induced teratogenicity [
3]. We subsequently found that CRBN expression was required for the anti-MM activity of IMiDs [
4].
CRBN has been found to be an E3 ubiquitin ligase substrate recruiter [
5‐
7], but the full functional role of CRBN in this complex is still not well known. In fact CRBN also binds to BK
Ca [
8,
9], ClC-2 [
10], AMPK [
11], PSMB4 [
12], ikaros (IKZF1) and aiolos (IKZF3) [
13‐
15] and MEIS2 [
16], thus it is possible that CRBN might function as a substrate-recruiter to bind each of these proteins for ubiquitination by the E3 ubiquitin ligase machinery and other binding partners with clinically relevant function may also exist.
Indeed, in this report, we have identified argonaute 2 (AGO2), also termed eukaryotic translation initiation factor 2 subunit C2 (EIF2C2), as a CRBN-downstream binding factor. AGO2 plays a pivotal role in microRNA (miRNA) maturation, stability and function [
17‐
19]. We show that the treatment of IMiD-sensitive MM cells with lenalidomide significantly increased CRBN, subsequently decreasing both AGO2 protein and its target miRNAs and inducing apoptosis. Furthermore, directly reducing cellular AGO2 levels produced cellular cytotoxicity regardless of whether they are IMiD-sensitive or -resistant MM cells. Therefore, the expression of CRBN-downstream binding protein AGO2, by regulating miRNA levels, plays an important role for MM cell growth and survival.
Discussion and Conclusions
We have identified AGO2 as a CRBN-downstream binding protein. This conclusion is based on: 1) AGO2 was pulled down with His-tagged CRBN (Table
1 and Additional file
2: Table S2); 2) CRBN was co-IPed with 42.4-tagged AGO2 (Fig.
2b); 3) 42.4-tagged AGO2 was co-IPed with CRBN (Fig.
2c); 4) endogenous AGO2 was co-IPed with wild-type CRBN (Fig.
2d and
e); 5) the steady-state levels of AGO2 in CRBN-high MM cells are significantly lower than the corresponding CRBN-low MM cells (Figs.
2a and
3a); and 6) treatment of MM cells with lenalidomide affects the steady-state levels of AGO2 (Fig.
3c,
d,
e and
f) and miRNAs (Fig.
7b and d).
AGO2 is considered as a master regulator of miRNA maturation and function [
17‐
19,
23‐
25] and miRNAs regulate up to 90 % of human genes via a silencing process mediated by miRNA-induced silencing complexes (miRISCs) [
23]. Dysregulation of miRNAs is associated with cancer initiation and progression [
26,
27]. It has been found that: 1) miR-125b induced myeloid leukemia by enhancing myeloid progenitor output from stem cells as well as inducing immortality, self-renewal and tumorigenesis in myeloid progenitors [
28]; 2) high-risk myeloma is associated with global elevation of miRNAs and over-expression of AGO2 [
29]; and 3) over-expression of AGO2 resulted in increased miRNA accumulation [
17,
30]. However, the mechanism of AGO2 regulation is largely un-known. Now we have found that AGO2 is a CRBN-downstream binding factor that is tightly regulated by the effective CRBN (Fig.
4) at the post-translational level. In addition, we have found that the steady-state levels of AGO2 in CRBN-high MM cells are significantly lower than the corresponding CRBN-low MM cells. Therefore, dysregulation of CRBN in cancer cells is responsible for malfunctions of AGO2 and miRNAs.
It has been reported that IMiDs decreased the expression of vascular endothelial growth factor and basic fibroblast growth factor [
31], thereby inhibiting new blood vessel formation and decreasing the tumor growth. Indeed, microvessel growth in the IMiDs treated samples was significantly less than in the corresponding controls [
32‐
37]. However, the molecular mechanism of IMiD-induced anti-angiogenic effects is not well documented. Recent finding indicated that over-expression of AGO2 increased angiogenesis, via regulation of miRNA levels, whereas silencing of AGO2 inhibited angiogenesis [
38]. We have found, in this report, that AGO2 is a CRBN-downstream binding protein and the treatment of MM cells with lenalidomide for 5 days resulted in decreased both AGO2 (Fig.
3) and miRNAs (Fig.
7), indicating that IMiD-induced anti-angiogenic effects may go through CRBN-AGO2-miRNA pathway.
We conclude that AGO2 plays an important role in regulating MM cell growth and survival. This conclusion is based on our finding that silencing AGO2 expression halted MM cell growth (Fig.
5a and
b). In other words, MM cell growth requires higher levels of AGO2. Based on this conclusion, we predicted that the growth rate of My5.LV cells, which have significant higher levels of AGO2 than My5.CRBN cells (Fig.
3a and
b), should be higher than that of My5.CRBN cells. Indeed, the growth rates of pLKO.1- and pCDH-treated My5.LV cells were significantly higher than the corresponding treated My5.CRBN cells (Additional file
1: Figure S4A and S4B). In addition, the growth rate of AGO2-cDNA-treated My5.LV cells was significantly higher than the corresponding treated My5.CRBN cells (Additional file
1: Figure S4C), implying that the steady-state levels of AGO2 might be controlled by CRBN. Furthermore, the cell survival of the AGO2-high My5.LV cells, upon treatment with AGO2-sh72, is significantly higher than the AGO2-low My5.CRBN cells treated with the same shRNA (Additional file
1: Figure S4D), suggesting that it may take longer time to decrease the endogenous AGO2 to a critical point to inhibit cell growth in AGO2-high My5.LV cells than in AGO2-low My5.CRBN cells.
One puzzle we had in the past is that the time of IMiD-induced IKZF1 or IKZF3 degradation (within hours) and IMiD-induced cytotoxicity (days) is not temporally consistent. IKZFs are zinc finger transcription factors that play important roles in lymphocyte differentiation [
39,
40]. Treatment with IMiDs or IKZF-shRNA not only induced fast-degradation of IKZFs [
13‐
15,
20], but also modulate, within relatively short time, the expression of their downstream factors, such as interferon regulatory factor 4 (IRF4) [
13,
15] or interleukin-2 (IL-2) [
20]. However, the time required for IKZF-shRNA-induced cytotoxicity was significantly longer than the time of IMiD-induced cell death [
13,
14], implying that there might be other CRBN binding factors that participate in IMiD-induced cytotoxicity. We have found that AGO2 is a CRBN binding factor that is modulated by lenalidomide. AGO2 is a critical component of miRISCs that modulate a wide variety of protein syntheses at translational level. Interestingly, the time of lenalidomide-induced degradation of AGO2 (Fig.
3f) is consistent with lenalidomide-induced alteration of miRNAs (Fig.
7c and
d). In addition, the time of lenalidomide-induced alteration of miRNAs (Fig.
7c and
d) is also more or less consistent with the time of lenalidomide induced cell death (Fig.
1c). Therefore, although detailed mechanism of IMiD-induced cell death is not clear yet, part of the IMiD-induced cell death may go through CRBN-AGO2-miRNAs pathway.
The conclusion made in previous paragraph elicited a question of whether AGO2 is the sole factor responsible for the early onset of IMiD-induced cytotoxicity. We have found that: 1) the steady-state levels of AGO2 in IMiD-resistant My5.LV cells, upon treatment with lenalidomide, were not significantly affected (Fig.
3e and
g); 2) the treatment of IMiD-resistant My5.LV cells with 10 μM lenalidomide for 3 days did not significantly alter the steady-state levels of miRNAs (Fig.
7a). Therefore, these results are consistent with IMiD-resistance of My5.LV cells (Fig.
1c). However, we have also found that, upon treatment of My5.LV cells with 10 μM lenalidomide for 5 days, the steady-state levels of some miRNAs were significantly down-regulated (Fig.
7b). This fact is more or less consistent with lenalidomide-induced moderate down-regulation of AGO2 (Fig.
3e) and mild cytotoxicity (Additional file
1: Figure S1). One possible explanation for this fact is that lenalidomide treatment for 5 days did not decrease the levels of critical miRNAs to a threshold to massively trigger apoptosis in those MM cells expressing low levels of CRBN. However, in considering the following observations: 1) the miRNAs altered by the treatment with AGO2-sh72 are not exactly the same as the treatment with lenalidomide (Fig.
6b versus Fig.
7b or Fig.
6d versus Fig.
7d or Additional file
1: Figure S5 versus S6); 2) the proteins altered by the treatment with AGO2-sh72 are not exactly the same as the treatment with lenalidomide [For example, IKZF1 or IKZF3 in My5.CRBN cells was significantly increased upon treatment with AGO2-sh72, whereas these proteins were significantly decreased upon treatment with lenalidomide (data not shown)]; 3) slightly different response of My5.LV and My5.CRBN cells to AGO2-shRNAs was observed (Additional file
1: Figure S4D); we cannot rule out the possibility that some other un-identified CRBN-downstream binding factors may also participate in IMiD-induced cytotoxicity. In addition, the effects occurred within the first two or three days of lenalidomide treatment may not be directly associated with AGO2 (Figs.
1c and
5). Furthermore, although the sensitivity of MM cells to lenalidomide is associated with: 1) the intracellular levels of CRBN (Fig.
1); 2) the differential regulation of miRNAs, upon treatment with lenalidomide, in MM cells with variant levels of CRBN (Additional file
1: Figure S6R and S6S), the detailed molecular mechanism of lenalidomide induced cell death is not clear yet.
It is clear that IMiD binding to CRBN enhanced accumulation of CRBN, especially in IMiD-sensitive cells (Fig.
3f), and eventually resulted in decreased levels of AGO2 (Fig.
3). The decreased levels of AGO2 (Fig.
3 and 5) resulted in down-regulation of miRNAs (Fig.
6 and 7) that further affected protein syntheses. It is not clear, however, which protein synthesis is affected by the decreased levels of miRNAs. It has been reported that silencing of AGO2 with its shRNA enhanced protein expression of cyclin-dependent kinase (CDK) inhibitors p21Waf1/Cip1 and p27Kip1 [
29]. Since these proteins are CDK inhibitors, enhanced expression of these proteins will inhibit CDKs and result in enhanced cell cycle arrest [
29]. Enhanced expression of p21Waf1/Cip1 might be regulated by decreased level of miR-106 [
29]. Indeed, the steady-state levels of miR-106a-3p and miR-106b-3p were significantly reduced upon treatment of the MM cells with lenalidomide for 5 days (Additional file
1: Figure S6C, S6D, S6G, S6K, S6L, and S6O). In fact, many miRNAs were either up-regulated or down-regulated at least 4 fold upon treatment of the MM cells with either AGO2-shRNA or lenalidomide (Additional file
2: Tables S3 and S4). Although we still don’t know how many proteins were up- or down-regulated upon these treatments, the treatment of IMiD-resistant or IMiD-sensitive MM cells (Fig.
1c) with AGO2-shRNA strongly inhibited cell growth and induced cell death (Figs.
5 and
8). Therefore, AGO2 could be considered as a novel therapeutic target for overcoming IMiD-resistance in MM cells.
Methods
Cell culture
Human MM cell lines were maintained in RPMI-1640 medium (Thermo Scientific) supplemented with 10 % heat-inactivated fetal bovine serum (Thermo Scientific) and 1 % penicillin/streptomycin. BHK cells were cultured in DMEM/F12 medium (Thermo Scientific) supplemented with 5 % fetal bovine serum and 1 % penicillin/streptomycin.
Pull-down His-tagged CRBN and its associated proteins
10 histidine residues were introduced into the C-terminus of human CRBN cDNA by polymerase chain reaction (PCR) and then cloned into pCDH-CMV-MCS-EF1-copGFP (green fluorescent protein) expression vector (System Bioscience, Mountain view, CA), named as pCDH.GFP.CRBN.His. Lentiviral particles harboring the cloning vector and pCDH.GFP.CRBN.His were prepared, according to the method described [
41], and used to infect human myeloma OCI-My5 (My5) cells to generate control cell line My5.LV and CRBN expressing cell line My5.CRBN.His. The infected MM cells were sorted for GFP expression at day 14 after infection.
My5.LV and My5.CRBN.His cells were grown up in RPMI-1640 medium at 37 °C, treated with 2 mM sodium butyrate for overnight before harvesting, harvested by centrifugation and stored at - 80 °C. The cell pellets were re-suspended in 1 × binding buffer (20 mM Tris/HCl, pH7.9; 500 mM NaCl; and 10 % glycerol) with protease inhibitors and then equilibrated on ice for 30 minutes at 800 p.s.i. in a Parr nitrogen cavitation bomb [
22]. After releasing the pressure, the cell lysates were centrifuged at 33,000 × g for 30 minutes at 4 °C to collect supernatant. The pellet was re-suspended in 1 × binding buffer, sonicated for 20 pulses on ice and centrifuged at 33,000 × g for 30 minutes at 4 °C to collect supernatant. The two supernatants were combined and the protein concentration of the supernatants was determined. Equal amount of proteins from My5.LV supernatant or My5.CRBN.His supernatant were mixed, in two separated tubes, with similar amount of nickel-charged NTA (Ni-NTA) agarose beads, adjusted to have 5 mM imidazole, and gently rotated on a rocker overnight at 4 °C. The loaded Ni-NTA agarose beads were packed into a column (in two separated columns) and washed with 1 × binding buffer containing 10 mM imidazole (1
st wash), 20 mM imidazole (2
nd wash) and 40 mM imidazole (3
rd wash) [
42]. The bound proteins were eluted with 1 ×binding buffer containing 100 mM EDTA.
Tandem mass spectrometry (MS/MS) analysis
For the proteomics experiments, the proteins pulled-down with nickel-charged NTA agarose beads were resolved by a 10 % sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE). The gel bands (10 gel bands per lane) were excised and destained, digested with trypsin, and desalted using ZipTips (Millipore, Billerica, MA) as previously described [
43]. HPLC-ESI-MS/MS analysis was performed on a Thermo Scientific Orbitrap Elite Velos Pro fitted with an EASY-Spray Ion Source (Thermo Scientfic, Waltham, MA). On-line HPLC was performed using a Thermo Scientific Dionex Ultimate 3000 Series Nano/Cap System NCS-3500RS nanoLC with an Acclaim PepMap100 trap column (ThermoScientific, 75 μm ID ×2 cm, 3 μm C18, 100 Å) and an Acclaim PepMap RSLC analytical column (ThermoScientific, 75 μm ID ×15 cm, 2 μm C18, 100 Å); loading was performed at flow rate of 6 nl/min, and elution was performed at a gradient of 2 to 35 % Buffer B for 65 minutes, followed by a step to 90 % Buffer B and a hold for 5 minutes, followed by a return to 4 % Buffer B (Buffer A: 0.1 % formic acid in water, Buffer B: 80 % acetonitrile in 0.1 % formic acid and water); analytical flow rate, 300 nl/min. A “top-10” data-dependent MS/MS analysis was performed (acquisition of a full scan spectrum followed by collision-induced dissociation mass spectra of the 10 most abundant ions in the survey scan). The fragment mass spectra were then searched against the human SwissProt_2013_02 database (539,165 entries) using Mascot (Matrix Science, London, UK; version 2.4). The search variables used were: 1) 10 ppm mass tolerance for precursor ion masses and 0.5 Da for product ion masses; 2) digestion with trypsin; 3) a maximum of two missed tryptic cleavages; 4) variable modifications of oxidation of methionine and phosphorylation of serine, threonine, and tyrosine. Cross-correlation of Mascot search results with X! Tandem was accomplished with Scaffold (version Scaffold_4.2.1; Proteome Software, Portland, OR, USA). Probability of peptide assignments and protein identifications were made through the use of Scaffold. Only peptides with ≥ 95 % probability were considered.
Validation of AGO2 as a CRBN downstream binding protein
In order to validate AGO2 as a CRBN binding protein, the epitope of 42.4 antibody, a multidrug resistance-associated protein 1 (MRP1) antibody [
22], was introduced into the C-terminus of human AGO2 cDNA (from DNASU, Tempe, AZ) by PCR and then cloned into the pCDH-CMV-MCS-EF1-copGFP expression vector that provides GFP expression, named as pCDH.GFP.AGO2.42.4. In the meantime, the 10 His-tagged human CRBN cDNA was shifted from pCDH.GFP.CRBN.His to pNUT expression vector [
44,
45], named as pNUT.CRBN.His.
Three cell lines had been established by introducing, via calcium phosphate transfection method [
42], the following pairs of DNAs: 1) pCDH.GFP.AGO2.42.4 + pNUT (named as AGO2/pNUT); 2) pCDH-CMV-MCS-EF1-copGFP + pNUT.CRBN.His (named as pCDH/CRBN); 3) pCDH.GFP.AGO2.42.4 + pNUT.CRBN.His (named as AGO2/CRBN), into BHK cells [
22]. Permanent cell lines were established by selection with 100 μM MTX.
For co-immunoprecipitation (Co-IP) of CRBN with AGO2, the aforementioned three BHK cell lines were lysed with NP40 lysis buffer (0.1 % NP40; 150 mM NaCl; 10 mM NaMoO4• 2H20; and 50 mM Tris–HCl, pH7.6), supplemented with 1 ×protease inhibitor cocktail containing: Aprotonin, 2 μg/mL; Benzamide, 121 μg/mL; E64, 3.5 μg/mL; Leupeptin, 1 μg/mL; and Pefabloc, 50 μg/mL. Primary antibody, after pre-cleaning with protein G beads (Invitrogen), was added to the cell lysates and gently rotated overnight at 4 °C. Protein G beads were added to the antibody treated cell lysates and washed with NP40 cell lysis buffer three times. The bound proteins were eluted with 45 μL of 2 ×SDS-PAGE loading buffer at 90 °C for 10 minutes.
To Co-IP of AGO2 with CRBN, the cell lysates prepared from My5.LV and JJN3 MM cell lines were IPed with our recently made anti-human CRBN monoclonal antibody 2F11G5.
Immunoblotting
Western blot was performed according to the routine protocol. Briefly, equal amounts of proteins were subjected to SDS-PAGE, followed by transferring the proteins to nitrocellulose membranes, probed with primary antibody overnight at 4 °C, washed with phosphate buffered saline containing 0.1 % Tween-20 and then incubated with appropriate horseradish peroxidase-conjugated secondary antibody. Chemiluminescent film detection was performed according to the manufacturer’s recommendation. Anti-human CRBN mouse monoclonal antibodies, 2B11G10 and 2F11G5, were made by ourselves (through GenScript). Rabbit-anti-CRBN antibody was purchased from Sigma, whereas rabbit-anti-AGO2 antibody or other antibodies were from Cell Signaling Technology.
MTT assay
Cell viability was measured by employing 3–(4,5-dimethylthiazol-2-yl)–2,5-diphenyltetrazolium bromide (MTT) dye (Sigma-Aldrich) performed according to the method described [
42,
46].
Modulating the expression of human AGO2 in MM cells
The 42.4-tagged AGO2 cDNA was shifted from pCDH.GFP.AGO2.42.4 to pCDH-CMV-MCS-EF1-Puro, named as pCDH.puro.AGO2.42.4, and used to make lentiviral particles, according to the routine method. AGO2-shRNA constructs were purchased from Sigma-Aldrich (The sequences of the shRNAs are shown in Additional file
2: Table S1) and used to make lentiviral particles. In the meantime, pLKO.1, the vector which harbored AGO2-shRNAs, and pCDH-CMV-MCS-EF1-Puro, the vector which harbored 42.4-tagged AGO2 cDNA, were also used to make lentiviral particles. These viral particles were used to infect My5.LV or My5.CRBN.His cells and selected with 1 μM puromycin, performed according to the routine method.
MicroRNA array (miRNA array) and quantitative polymerase chain reaction (qPCR)
For miRNA array, total RNA was isolated from My5 cells with miRNeasy Mini kit (QIAGEN). Reverse transcription was performed with miScript II RT kit (QIAGEN), according to the protocol provided by QIAGEN. Quantitative analysis of miRNA was performed with miScript SYBR Green PCR Kit (QIAGEN) and miScript miRNA PCR Arrays (QIAGEN), performed with our Applied Biosystems 7900HT Fast Real-Time PCR System, according to the protocol provided by the manufacturer. Data analyses were performed with QIAGEN online software. The miRNA array data presented in this manuscript have been deposited in the NCBI GEO database (accession no. GSE61693).
For qPCR, total RNA was used to do qPCR with either CRBN primers (forward: 5’-CAGTCTGCCGACATCACATAC; reverse: 5’-GCACCATACTGACTTCTTGAGGG) or AGO2 primers (forward: 5’-TCCACCTAGACCCGACTTTGG; reverse: 5’-GTGTTCCACGATTTCCCTGTT).
Apoptosis analysis
The Alexa Fluor 647 Annexin V kit (BioLegend) was used for detecting early apoptotic cells, late stage apoptotic cells, necrotic cells and live cells, performed according to the manufacture’s instructions.
Statistical analysis
Mean values and two-tailed P values were calculated based on the unpaired t test from GraphPad Software Quick Calcs. By conventional criteria, if P value is less than 0.05, the difference between two samples is considered to be statistically significant.
Availability of data and materials
The datasets supporting the conclusions of this article are included within the article and its additional files.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
XBC, AKS and QX designed the experiments; QX, YXH and XBC performed most of the experiments; PL, ML and LJM participated in mass spectrometry analysis; PE, JZ, CXS, YXZ and YX participated in some experiments. All authors participated in discussion and contributed to interpretation of data; XBC and QX drafted the manuscript. All authors have read and approved the final manuscript.