Identification of a deubiquitinating enzyme as a novel AGS3-interacting protein

Abstract

Activator of G protein Signaling 3 (AGS3) is a receptor-independent G protein activator that has been implicated in multiple biological events such as brain development, neuroplasticity and addiction, cardiac function, Golgi structure/function, macroautophagy and metabolism. However, how AGS3 is regulated is little known. We demonstrate here that AGS3 interacts with a ubiquitin specific protease USP9x, and this interaction is at least partially mediated through the C-terminal G protein regulatory domain of AGS3. Knockdown of USP9x causes a moderate reduction in the level of AGS3. In contrast, overexpression of either USP9x or its deubiquitinating domain UCH increases the amount of AGS3, whereas expression of the mutant UCH domain that lacks deubiquitinating activity does not have the same effect. As previously observed in AGS3 knockdown cells, the localization of several marker proteins of the late Golgi compartments is disturbed in cells depleted of USP9x. Taken together, our study suggests that USP9x can modulate the level of a subpopulation of AGS3, and this modulation plays a role in regulating the structure of the late Golgi compartments. Finally, we have found that levels of AGS3 and USP9x are co-regulated in the prefrontal cortex of rats withdrawn from repeated cocaine treatment. In conjunction with the above data, this observation indicates a potential role of USP9X in the regulation of the AGS3 level during cocaine-induced neuroplasticity.

Figure 1. Identification of USP9x as an interacting protein of AGS3.

(A) The SDS-PAGE and mass spectrometric analysis of the immunoprecipitate of EGFP-AGS3 and its associated proteins from Flp-In CV1 stable cells (the top panel). The immunoprecipitate prepared from CV1 cells stably expressing the EGFP alone was included as a negative control. The nature of the doublet bands of EGFP-AGS3 is unclear but presumably caused by protein degradation. Stably expressed EGFP-AGS3, but not EGFP alone, was also able to co-immunoprecipitate Gαi3 as shown by western blotting (the bottom panel). (B) Characterization of the rabbit anti-USP9x antibody in western blotting. The antibody was raised using the N-terminal 500 a.a. of USP9x as an antigen and the crude sera was immunopurified through a peptide column. An equal amount of cell lysates made from HEK293 cells transfected with a non-targeting control siRNA or either of the two USP9x siRNAs (30 nM, 48 hrs) were loaded and probed with the immunopurified antibody (1 µg/ml). GAPDH was used as a loading control and detected with a monoclonal anti-GAPDH antibody (0.2 µg/ml). (C) Characterization of our rabbit anti-USP9x antibody in immunofluorescence. HEK293 Cells were treated with either control or one of the USP9x siRNAs as described above, and stained with the anti-USP9x antibody (1 µg/ml). (D) Co-immunoprecipitation of USP9x with AGS3 from HEK293 cell lysates. Lysates were incubated with an anti-AGS3 antibody (10 µg/ml) or normal rabbit IgG (10 µg/ml, a negative control) and the immunoprecipitates (IP) were probed with either anti-AGS3 (1 µg/ml) or anti-USP9x antibody (1 µg/ml). The heterogenous band pattern of AGS3 has been previously observed and is at least partially caused by the phosphorylation of AGS3 . (E) Co-immunoprecipitation of USP9x with AGS3 from the rat brain PFC lysate. The immunoprecipitation and western blot analysis were conducted as described in (D). (F) Co-immunoprecipitation of AGS3 with USP9x from the rat PFC lysate. The immunoprecipitation and western blot analysis were conducted as described in (D) except that the USP9x antibody (10 µg/ml) was used to immunoprecipitate USP9x and its associated proteins.

Figure 2. Mapping of the USP9x-interacting domain of AGS3.

(A) Schematic illustration of the regions covered by the different GST-AGS3 constructs used in the GST pull-down experiments. (B) and (C) Top panels: Coomassie blue gels showed the GST fusion proteins and their relative amounts used in the GST pull-down. The full-length fusions are indicated by asterisks except for GST-TPR whose expression was too low to be detected. The lower molecular weight bands are probably the products of degradation. Bottom panels: Equal amounts of HEK293 cell lysates were incubated with various GST fusion proteins bound to glutathione beads. After wash, the GST pull-down samples were eluted from the beads and probed with anti-USP9x (1 µg/ml) in western blot analysis. (D) The pull-down was performed as described in (C) except that the elutes were probed with an anti-Gαi3 antibody (1 µg/ml).

Figure 3. Effects of depleting or overexpressing USP9x on AGS3.

(A) HEK293 cells infected by the lentivirus expressing pLVX-shRNA1, USP9x-shRNA2, or USP9x-shRNA3 were lysed and the lysates were probed with anti-AGS3 (1 µg/ml), anti-GAPDH (0.2 µg/ml), and anti-β-Actin (0.1 µg/ml) antibodies, respectively. The values in the bar graph represent the averages from 6 independent western blot analyses quantified by Li-COR Odyssey Infrared Imaging System (error bars: standard deviations). A representative western blot image was shown. (B) Characterization of AGS3 antibody in immunofluorescence. Images of endogenous AGS3 in HEK293 cells transfected with a non-targeting control siRNA, or one of three different AGS3 siRNAs (siRNA1, 2 or 3; 30 nM, 48 hrs). Cells were then stained using the anti-AGS3 antibody (1 µg/ml). (C–E) Impact of overexpression of an HA-tagged USP9x (C), or the catalytic UCH domain of USP9x (D, E), on the intensity of AGS3 staining 24 hrs after transfection. For (C) and (D), cells overexpressing the relevant proteins are indicated by arrows, arrowheads and asterisks, and the same field is shown under both 20x and 63x magnification in (D). Cells were co-stained with an anti-HA antibody (1∶1000) and the anti-AGS3 (1 µg/ml) antibodies. For (E), the average intensity of AGS3 staining in 50 transfected cells expressing a comparable level of HA-UCH or the catalytically inactive UCH(C/A) mutant was further quantified as described in “Materials and Methods,” and the value was normalized to that of the non-transfected cells. Scale bar: 20 µm.

Figure 4. Impact of overexpression of the UCH domain of USP9x (A, B), or the catalytically inactive UCH(C/A) mutant (C) on the intensity of Gαi3, HSP90 or AGS3 staining 24 hrs after transfection.

Cells overexpressing the relevant proteins are indicated by arrows, arrowheads and asterisks. Cells were co-stained with an anti-HA antibody (1∶1000) (A–C) and anti-Gαi3 (1 µg/ml) (A), anti-HSP90 (5 µg/ml) or anti-AGS3 (1 µg/ml) (C) antibodies. The same field is shown under both 20x and 63x magnification. Scale bar: 20 µm.

Figure 5. Influence of USP9x knockdown on the distributions of marker proteins of ER, Golgi and lysosomes.

(A) HEK293 cells infected by the lentivirus expressing pLVX-shRNA1 (non-targeting), USP9x-shRNA2, or USP9x-shRNA3 were co-stained with anti-USP9x (1 µg/ml) and anti-β-GalT1 (1∶2500) (the top panels) or co-stained with anti-β-GalT1 (1∶2500) and anti-TGN46 (1∶2500) (the bottom panels). (B) Percentage of cells with condensed or dispersed TGN46 staining. The values on the bar graph represented the average from three independent experiments with more than 500 cells were counted in each experiment (error bars: standard deviation). (C–E) The same set of HEK293 cells were stained with anti-Calreticulin (1∶1000; C), p115 (1∶600; D) or Lamp1 (1∶3000; E), respectively. Scale bar: 20 µm.

Figure 6. Co-regulation of AGS3 and USP9x in the prefrontal cortex (PFC) of rats after prolonged cocaine withdrawal.

Experimental rats (n = 9) received once/daily i.p. injections of 15 mg/kg cocaine for 1 week followed by 3 weeks of withdrawal. Control rats (n = 9) were injected on the same schedule with equal volumes of i.p. saline. PFC homogenates were prepared, separated on SDS-PAGE gels (30 µg total protein per sample), and probed with anti-AGS3 (1 µg/ml), anti-USP9x (1 µg/ml), and anti-β-Actin (0.1 µg/ml) antibodies. Signal intensities of AGS3 and USP9x were first normalized to that of β-Actin for each sample, and the average normalized ratio of AGS3/Actin (A) or USP9x/Actin (B) for the saline-treated controls was arbitrarily set to 1. Cocaine-induced increases in AGS3 and USP9x are expressed as multiples of the saline-treated average. Both increases were determined to be statistically significant (AGS3 p = 0.0005, USP9x p<0.0001) using a two-tailed, equal-variance Student's t-test.