Latent CSN-CRL complexes are crucial for curcumin-induced apoptosis and recruited during adipogenesis to lipid droplets via small GTPase RAB18

Abstract

The COP9 signalosome (CSN) and cullin-RING ubiquitin ligases (CRLs) form latent CSN-CRL complexes detectable in cells. We demonstrate that the CSN variants CSN^CSN7A and CSN^CSN7B preferentially bind to CRL3 or CRL4A and CRL4B, respectively. Interestingly, the interacting protein ubiquitin-specific protease 15 exclusively binds to latent CSN^CSN7A-CRL3, while p27^KIP attaches to latent CSN^CSN7B-CRL4A complex. Inhibition of deneddylation by CSN5i-3 or neddylation by MLN4924 do not impede the formation of latent complexes. Latent CSN^CSN7A-CRL3 and latent CSN^CSN7B-CRL4A/B particles are essential for specific cellular functions. We found that curcumin-induced cell death requires latent CSN^CSN7B-CRL4A. Knockout of CSN7B in HeLa cells leads to resistance against curcumin. Remarkably, the small GTPase RAB18 recruits latent CSN^CSN7A-CRL3 complex to lipid droplets (LDs), where CRL3 is activated by neddylation, an essential event for LD formation during adipogenesis. Knockdown of CSN7A or RAB18 or destabilization of latent complexes by cutting off CSN7A C-terminal 201-275 amino acids blocks adipogenesis.

Keywords: Cell biology; Cellular physiology; Functional aspects of cell biology.

Graphical abstract

Figure 1

Preferential binding of CSN7A to CUL3 and CSN7B to CUL4A and specific interactions of USP15, p27, RAB18, and CAV1 with CSN7A and CSN7B (A) HeLa, LiSa-2, and MRC5 cell lysates were immunoblotted (Input). Immunoprecipitates of CSN7A or CSN7B were analyzed by indicated antibodies. Control was performed with non-immunized serum from rabbit (Control). (B) AD293WT, AD293-FLAG-CSN7A, and AD293-FLAG-CSN7B cell lysates were immunoblotted (Input). Pull-downs of FLAG-CSN7A and FLAG-CSN7B were analyzed by indicated antibodies. Empty FLAG-vector stably transfected into AD293 cells was used as control. (C) LiSa-2 cell lysates were immunoblotted (Input). Immunoprecipitates of anti-CUL3 and anti-CUL4A antibodies were analyzed by indicated antibodies. Control was performed with non-immunized serum from rabbit (Control). (D) Blots of LiSa-2 cells shown in (A) or in (S1A) were quantified by densitometry to determine the relative amounts of CUL1, CUL3, and CUL4A which were divided by relative amounts of immunoprecipitated CSN7A or CSN7B. Data are expressed as ratios of relative amounts of CUL1, CUL3, and CUL4A divided by relative amounts of CSN7A or CSN7B plus/minus standard deviations (SD, n = 3–5). Unpaired Student’s t-test was used for statistical analysis. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. (E) LiSa-2 cells were incubated without or with CSN5i-3 (1 μM) or MLN4924 (1 μM) for 24 h, and then lysates were immunoblotted (Input). After incubation, immunoprecipitates of CSN7A or CSN7B were analyzed by indicated antibodies. Control was performed with non-immunized serum from rabbit (Control). (F) Blots of LiSa-2 cells as shown in (E) were quantified by densitometry to determine the relative amounts of USP15, p27, RAB18, and CAV1 in the precipitates which were divided by the relative amounts of immunoprecipitated CSN7A or CSN7B. Data are expressed as ratios of relative amounts of USP15, p27, RAB18, and CAV1 divided by relative amounts of CSN7A or CSN7B plus/minus standard deviations (SD, n = 3). Unpaired Student’s t-test was used for statistical analysis. ∗p < 0.05, ∗∗p < 0.01.

Figure 2

Main CSN7A and CSN7B domains and relevance of their C-terminal 201-stop amino acids for selected protein-protein interactions (A) Amino acid structure of human CSN7A/COPS7A (NM_001164093.1) and CSN7B/COPS7B (NM_001282950.3). The unique N-terminal 1–32 amino acids (blue) and C-terminal 201-stop amino acids (orange) and the homologous PCI domain (88–177 amino acids, red) are marked. FLAG-CSN7A^1-200 and FLAG-CSN7B^1-200 were used in further experiments. (B) LiSa-2-FLAG-vector, LiSa-2-FLAG-CSN7A, LiSa-2-FLAG-CSN7B, LiSa-2-FLAG-CSN7A^1-200, and LiSa-2-FLAG-CSN7B^1-200 cell lysates were immunoblotted and analyzed with indicated antibodies. Additional bands seen in CSN7A and CSN7B are wild-type subunits (Input). FLAG pull-downs were performed, and FLAG-eluates were probed with the same antibodies. Empty FLAG-vector stably transfected into LiSa-2 cells was used as control. Since anti-CSN7B antibody has its epitope in the 201–273 amino acid region, it does not recognize FLAG-CSN7B^1-200. (C) In HeLa cells, HA-USP15 was co-transfected with FLAG-CSN5, FLAG-CSN7A, or FLAG-CSN7A^1-200. After 16 h, immunoblots were analyzed with indicated antibodies (Input). At the same time, immunoprecipitations were performed with anti-HA antibody (IP: HA). Precipitates were evaluated with the same antibodies. (D) In HeLa cells, FLAG-CSN7B and EGFP-RAB18 or FLAG-CSN5 and EGFP-RAB18 as well as FLAG-CSN7B and FLAG-p27 or FLAG-CSN5 and FLAG-p27 were co-transfected (Input). After 16 h, immunoblots were analyzed with indicated antibodies (Input). In parallel, immunoprecipitations were performed with anti-RAB18 or -p27 antibodies. Precipitates were evaluated with the same antibodies (IP: RAB18 and IP: p27).

Figure 3

Curcumin-induced apoptotic cell death in HeLa cells requires latent CSN^CSN7B-CUL4A complex (A) HeLa-WT, HeLa-CSN7A-KO, and HeLa-CSN7B-KO cells were treated with 0, 12.5, 25, and 50 μM of curcumin for 12 h. Lysates were analyzed by indicated antibodies. (B) Blots in (A) were quantified by densitometry for p53 (n = 3), normalized against γ-tubulin, and means of relative p53 amounts plus/minus SD were plotted against curcumin concentration. (C) Blots in (A) were quantified by densitometry for p21 (n = 3), normalized against γ-tubulin, and means of relative p21 amounts plus/minus SD were plotted against curcumin concentration. (D) Blots in (A) were quantified by densitometry for p27 (n = 3), normalized against γ-tubulin, and means of relative p27 amounts plus/minus SD were plotted against curcumin concentration. (E) HeLa-WT, HeLa-CSN7A-KO, and HeLa-CSN7B-KO cells were treated with curcumin (50 μM). After 18 h of drug exposure, annexin V staining was analyzed by flow cytometry. Staurosporin (1 μM) was used as reference. (F) Apoptosis in percent of three independent experiments as shown in (E) (n = 3) was calculated in HeLa-WT, HeLa-WT + curcumin (50 μM) (HeLa-Cur), HeLa-WT + staurosporin (1 μM) (HeLa-Stau), HeLa-CSN7A-KO, HeLa-CSN7A-KO + curcumin (50 μM) (HeLa-CSN7A-KO + Cur), HeLa-CSN7B-KO, and HeLa-CSN7B-KO + curcumin (50 μM) (HeLa-CSN7B-KO + Cur). Data are expressed as % apoptosis. Error bars indicate SDs. Unpaired Student’s t test was used for statistical analysis. Significant differences are indicated. ∗∗p < 0.01. (G) Lysates of HeLa-WT, HeLa-CSN7A-KO, and HeLa-CSN7B-KO cells as well as LiSa-2-FLAG-CSN7A, LiSa-2-FLAG-CSN7B, LiSa-2-FLAG-CSN7A^1-200, and LiSa-2-FLAG-CSN7B^1-200 were immunoblotted and analyzed with indicated antibodies. Blots are representative of three independent experiments.

Figure 4

RAB18 recruits CSN^CSN7A-CRL3 complexes to LD membranes during adipogenesis in LiSa-2 cells (A) LiSa-2-WT, LiSa-2-CSN7A-KD, and LiSa-2-CSN7B-KO cell lysates were analyzed with indicated antibodies after 1, 8, and 15 days of adipogenic differentiation. (B) LiSa-2-WT, LiSa-2-CSN7A-KD, and LiSa-2-CSN7B-KO cells were stained with ORO and quantified by photometry. Data obtained from three different experiments are expressed as means of relative amounts plus/minus SD. Student’s t-test was used for statistical analysis. ∗p < 0.1. (C) LiSa-2-WT, LiSa-2-CSN7A-KD, and LiSa-2-CSN7B-KO cells were analyzed by light microscopy (LM) and confocal fluorescence microscopy. After 15 days of adipogenic differentiation, cells were stained with specific antibodies as indicated and analyzed by confocal microscopy. Cell nuclei were stained with DAPI. Regions of interest such as LDs are shown with yellow arrow. The yellow bar in Merge channel corresponds to 15 μm. (D) LiSa-2-WT cells were stably transfected with GFP-CSN7A. After 15 days of adipogenic differentiation, CUL3 and RAB18 were stained with specific primary antibodies as indicated and with species-appropriate fluorescence secondary antibodies, Alexa Flour 647 (red) and Cy3 (magenta), respectively, and analyzed by confocal microscopy. Cell nuclei were stained with DAPI. Regions of interest such as LDs are shown with yellow arrow. The yellow bar channel corresponds to 15 μm.

Figure 5

Recruited CUL3 is neddylated on LD membrane during adipogenesis in LiSa-2 cells (A) Control siRNA (siGFP) and specific siRNA against RAB18 (siRAB18) were transfected into LiSa-2 cells. After 48 h transfected cells were incubated for 1 and 8 days of adipogenesis. After 8 days samples are presented as doublet: 2 x siGFP and 2 x siRAB18. (B) Blots in (A) were quantified by densitometry for neddylated and deneddylated CUL3 (n = 4), normalized against γ-tubulin and expressed as means of relative amounts of neddylated CUL3 divided by deneddylated CUL3 plus/minus SD. Unpaired Student’s t-test was used for statistical analysis. ∗p < 0.05. (C) Blots in (A) were quantified by densitometry for neddylated and deneddylated CUL4A (n = 4), normalized against γ-tubulin and expressed as means of relative amounts of neddylated CUL4A divided by deneddylated CUL4A plus/minus SD. (D) LiSa-2 cells were preincubated with control (siGFP) and with siRAB18 for 48 h. After 8 days of adipogenesis control cells (siGFP) and siRAB18 cells were stained with ORO and analyzed with light microscopy. Bars represent 50 μm. (E) Stains shown in (D) were quantified by photometry measuring ORO with OD_520nm. Data obtained from three different experiments are expressed as means of relative amounts plus/minus SD. Student’s t-test was used for statistical analysis. ∗∗∗p < 0.1.

Figure 6

Models of curcumin-induced cell death in HeLa cells and LD formation during adipogenesis in LiSa-2 cells (A) Curcumin blocks the activity of serine-threonine kinase Akt and nuclear factor kappa of activated B cells (NF-κB) signaling which leads to increase of p53, p21, p27, and caspase signaling. The presence of latent CSN^CSN7B-CRL4A complex prevents ubiquitylation and degradation of p21 and p27. The consequence is an arrest of cell cycle and apoptosis. As a result of CSN7B knockout, the CRL4A activity lost its natural brake, the CSN^CSN7B variant, which causes rapid ubiquitylation/degradation of p21 and p27. Therefore, HeLa-CSN7B-KO cells have no cell-cycle arrest and reduced apoptosis. They are characterized by high resistance against curcumin-induced apoptosis. (B) Mutations in the 201–273 amino acids region of CSN7B were obtained from https://cancer.sanger.ac.uk/. For instance, mutation of Glu210 to Val (E210V) causing a missense substitution in the CSN7B C terminus is published in Shi et al. (C) As a prerequisite of LD formation, RAB18, presumably in its phosphorylated GTP-RAB18 form, recruits CSN^CSN7A-CRL3 to LD membranes. At the outer LD membrane CSN^CSN7A-CRL3 is activated by neddylation, whereas CSN^CSN7A deneddylation is probably inhibited by RAB18 binding to CSN5. In case of CSN7A knockdown, not enough CSN^CSN7A-CRL3 complex can be recruited to LDs by RAB18. Under this condition, LD formation is blocked. Likewise, cutting off the C-terminal ends of CSN7A, which results in disintegration of the latent CSN^CSN7A-CRL3 complex, also leads to a blockade of LD formation and of adipogenesis. (D) Mutations in the C-terminal 201–275 amino acid region of CSN7A were obtained from https://cancer.sanger.ac.uk/. An example is the substitution of Ala246 by Val (A246V) published in Berndt et al.