Increased CSN5 expression enhances the sensitivity to lenalidomide in multiple myeloma cells

Abstract

Lenalidomide (LEN) is commonly used as an effective therapeutic agent for multiple myeloma (MM). However, in some patients, primary resistance to LEN is observed, the mechanisms of which remain poorly understood. In this study, we combined a LEN sensitivity assay with proteomics data from 15 MM cell lines to identify protein expression profiles associated with primary LEN resistance. Our findings revealed that CSN5 expression is lower in LEN-resistant cell lines than in LEN-sensitive lines. Moreover, we established that CSN5 is degraded via the cullin-RING ubiquitin ligase (CRL)-mediated ubiquitin-proteasome pathway through ubiquitination at lysine 194. Our data suggest that reduced CSN5 expression leads to abnormalities in the ubiquitination cycle of CRL4A, resulting in the inhibition of LEN-mediated degradation of IKZF1 and IKZF3. These findings delineate an additional mechanism of LEN resistance in MM cells and may contribute to the development of alternative therapeutic strategies to overcome LEN resistance.

Keywords: Cancer; Cell biology; Molecular biology; Proteomics.

Graphical abstract

Figure 1

Identification of the proteins involved in LEN resistance (A) Workflow used for the identification of proteins involved in LEN resistance. Based on LEN sensitivity assay results, the 15 MM cell lines were classified into sensitive and resistant groups. The proteomics data of each cell line were used for OPLS-DA in SIMCA 17.0 to differentiate between the LEN-sensitive and -resistant groups, yielding regression coefficients for all proteins. (B) Regression coefficients for the proteins identified. (C) Expression profiles of the top 10% and bottom 10% proteins based on regression coefficients. Red and blue indicate high and low protein expression, respectively. (D) GO enrichment analysis of the top 10% and bottom 10% proteins based on the regression coefficients using Metascape (https://metascape.org/).

Figure 2

COP9 signalosome protein expression in LEN-sensitive and -resistant groups (A) Comparison of the expression levels of COP9 signalosome proteins identified through proteomics. Student's t test. (B) Comparison of the proportions of neddylated- and non-neddylated-CUL4 in MM.1S and RPMI8226 cells, which were the most LEN-sensitive and -resistant cells, respectively, among the 15 cell lines assessed in this study. Comparisons were performed in triplicate. Band densities were quantified using iBright Analysis Software, and were averaged across the data obtained for triplicate assessments. Error bar represents standard deviation. Student's t test.

Figure 3

Overexpression of CSN5 induces LEN sensitivity (A) Induction of LEN sensitivity in RPMI8226, KMS11, KMS12BM, and MM.1S cells overexpressing FLAG-CSN5. Empty vector (lentiEF1-FLAG-P2A-Blast) was used as a control. We used 100 μM and 1 μM LEN for LEN-resistant and -sensitive cells, respectively. Data represent the mean ± standard deviation (n = 5). Student's t test. (B) Relationship between LEN sensitivity and the endogenous CSN5 expression levels in the 15 cell lines. Correlations between protein levels and cell viability were evaluated using simple linear regression in GraphPad Prism 8. Statistical significance was set at p < 0.05. (C) Western blots representing the expression of CUL4A and CSN5 in RPMI8226 cells overexpressing WT FLAG-CSN5 or FLAG-CSN5-H138Q. β-actin was used as a loading control. The larger proportion of neddylated CUL4A suggested that CSN5-H138Q decreased deneddylation activity. (D) LEN sensitivity in RPMI8226 cells overexpressing FLAG-CSN5-H138Q. The lentiEF1-FLAG-P2A-Blast vector harboring FLAG-tagged CSN5 or FLAG-tagged CSN5 with H138Q mutation was used for lentiviral infection with RPMI8226. We used 100 μM LEN in this assay. Data represent the mean ± standard deviation (n = 5). Student's t test. (E) LEN response pathway (CRBN axis) involved in the antitumor mechanism of LEN. (F) Western blots of IKZF1, IKFZ3, IRF4, CSN5 expression levels after treatment with the indicated dose of LEN in FLAG-CSN5- or empty vector-overexpressing RPMI8226 cells. β-Αctin was used as the loading control. The protein bands in western blotting of IKZF1, IKZF3 and IRF4 were quantified using iBright Analysis Software. Band densities of IKZF1, IKZF3 and IRF4 were normalized to that of β-actin. Data represent the mean (n = 2).

Figure 4

Mechanism underlying the low expression of CSN5 in LEN-resistant cells (A) CSN5 protein level was compared in MM.1S and RPMI8226 cells. Data represent the mean (n = 2). (B) CSN5 mRNA level was compared in MM.1S and RPMI8226 cells. Data represent the mean ± standard deviation (n = 3). Student's t test. (C) Comparison of CSN5 degradation time between MM.1S and RPMI8226 cells using a CHX chase assay. The cells were incubated with 10 μM CHX. Data represent the mean (n = 2). (D) Western blots of CSN5, LC3, and ubiquitin levels after treatment with the proteasome inhibitor MG132 or the autophagy inhibitor bafilomycin A1 in RPMI8226 cells. β-Αctin was used as the loading control. Data represent the mean (n = 2). (E) Identification of ubiquitination sites on CSN5 in MM.1S and RPMI8226 cells by proteomics. The experiment was conducted in technical duplicate. (F) Comparison of CSN5 degradation time in RPMI8226 cells overexpressing FLAG-CSN5 or CSN5-K194R using a CHX chase assay. The cells were incubated with 10 μM CHX. Band densities were quantified using iBright Analysis Software. Data represent the mean (n = 2).

Figure 5

E3 ubiquitin ligase is responsible for CSN5 ubiquitination (A) Western blots representing CSN5, CUL1, CUL2, CUL3, CUL4A, CUL4B, and CUL5 expression levels in MM.1S and RPMI8226 treated with DMSO or 10 μM MLN4924 for 12 h. This experiment was performed in triplicate. β-Actin was used as a loading control. Band densities were quantified using iBright Analysis Software. Data are represented as mean ± standard deviation. Student's t test. (B) Proximity proteomics was performed using RPMI8226 cells overexpressing APEX2-CSN5 or TurboID-CSN5. We identified proteins that were highly enriched (2-fold or more, p < 0.01) in biotin-tyramide- or biotin-treated samples as proximal proteins of CSN5. We selected ubiquitin-related proteins from the enriched proteins and performed an STRING analysis (https://string-db.org/). A protein interaction network generated using STRING is shown. (C) Comparison of the levels of DDB1 and DCN1 expression quantified in proteomics analysis of LEN-sensitive and resistant cell groups among the 15 MM cell lines. Student's t test. (D) Comparison of the protein levels between CSN5 and DDB1 quantified in proteomics analysis of 15 MM cell lines.

Figure 6

Hypotheses of CSN5 reduction in LEN-resistant cell lines and mechanisms by which low CSN5 expression of induces LEN resistance (A) Hypothesis of CSN5 degradation in LEN-resistant MM cells. K194 of CSN5 is ubiquitinated by CRL4B containing DDB1 and degraded by the proteasome. Increased expression of DDB1 in LEN-resistant cells is presumed to promote the ubiquitination of CSN5 and reduce the CSN5 level. (B) Hypothesis of the LEN resistance mechanism in LEN-resistant cells with low CSN5 expression. The low level of CSN5 in LEN-resistant cells induces an abnormal ubiquitination cycle of CRL4A, which inhibits receptor exchanges in the CRL complex. Consequently, the incorporation efficiency of CRBN into CRL4A is decreased, resulting in inhibition of the ubiquitination of IKZF1 and IKZF3 upon LEN treatment.