SUMO-SIM interactions regulate the activity of RGSZ2 proteins

Abstract

The RGSZ2 gene, a regulator of G protein signaling, has been implicated in cognition, Alzheimer's disease, panic disorder, schizophrenia and several human cancers. This 210 amino acid protein is a GTPase accelerating protein (GAP) on Gαi/o/z subunits, binds to the N terminal of neural nitric oxide synthase (nNOS) negatively regulating the production of nitric oxide, and binds to the histidine triad nucleotide-binding protein 1 at the C terminus of different G protein-coupled receptors (GPCRs). We now describe a novel regulatory mechanism of RGS GAP function through the covalent incorporation of Small Ubiquitin-like MOdifiers (SUMO) into RGSZ2 RGS box (RH) and the SUMO non covalent binding with SUMO-interacting motifs (SIM): one upstream of the RH and a second within this region. The covalent attachment of SUMO does not affect RGSZ2 binding to GPCR-activated GαGTP subunits but abolishes its GAP activity. By contrast, non-covalent binding of SUMO with RH SIM impedes RGSZ2 from interacting with GαGTP subunits. Binding of SUMO to the RGSZ2 SIM that lies outside the RH does not affect GαGTP binding or GAP activity, but it could lead to regulatory interactions with sumoylated proteins. Thus, sumoylation and SUMO-SIM interactions constitute a new regulatory mechanism of RGS GAP function and therefore of GPCR cell signaling as well.

Figure 1. RGSZ2 contains a SUMO-attachment site and SUMO-interacting motifs.

A, Protein domains of the murine RGSZ2 protein (NP_064342). The cysteine- rich domain (CRD) is underlined, the RH domain (RGS box) is shaded in gray and the α-helical residues of the secondary structures are indicated. The Small Ubiquitin-like Modifier (SUMO) consensus motif is in bold-italics, with lysine121 enlarged-underlined, SUMO-Interacting Motifs (SIMs) are in bold, indicating the clusters of negative amino acids surrounding the SIMs. The PDZ domain binding motifs (PDZ bm) are also indicated. Below, a diagram of the RGSZ2 protein indicates the position of the cysteine-rich domain, RH region, SUMO covalent modification K121 (LKKE), non-covalent SIM IQVL (64–67) and ISIL (141–144). B. The RGSZ2 and the RH region were cloned into E. coli BL21 (DE3) with or without the SUMO machinery . GST-free RGSZ2 proteins (TEV cleavage) were resolved by SDS-PAGE, and probed with anti-RGSZ2 and anti-SUMO1 antibodies. Lane 1: control (no RGSZ2 vector); lane 2: RH region; lane 3: whole RGSZ2 sequence. C. The K121R RGSZ2 mutant induced along with the SUMO machinery was not sumoylated. Assays were repeated at least twice and produced comparable results. Representative experiments are shown.

Figure 2. Sumo binds non covalently the recombinant RGSZ2 protein.

A. Recombinant RGSZ2 and its RH region were incubated with SUMO1- (lane 3), SUMO2- (lane 4) and SUMO3-agarose (lane 5). SUMO-agarose conjugates captured RGSZ2 and its RH domain. Lane 1, RGSZ2/RH added. Lane 2, RGSZ2/RH in presence of agarose without SUMO. B. The Gαi2 subunit shows no binding to SUMO1/2/3-agarose.

Figure 3. Influence of SUMO on Gαi subunit association with RGSZ2 proteins.

Binding of ³⁵S-Gαi to sumoylated and unsumoylated recombinant RGSZ2 proteins. In vitro-translated ³⁵S-labeled GαiGDP subunits (10 µl) were incorporated into the samples, alone or with 2 mM MgCl2 and 30 µM AlF4- (30 µM AlF3 + 30 mM NaF). Samples in lanes 1 and 2 received glutathione sepharose (GS) beads; lanes 3 and 4, GST-RGSZ2 protein bound to GS beads; lanes 5 and 6, sumoylated GST-RGSZ2 attached to GS beads. Lanes 2, 4 and 6, 2 mM MgCl2 and 30 µM AlF4- were added to the incubation mixture. Lane 7 shows the Gαi, which was added to the samples analyzed in lanes 1 to 6. At the end of 2 h incubation, GS beads were precipitated and washed; RGSZ2-Gαi association was determined by autoradiography.

Figure 4. Effect of covalent sumoylation of RGS box on the GAP activity of RGSZ2 proteins.

A. RGSZ2 GAP activity. In the single turnover assay, 100 nM Gαi was incubated with 30 or 200 nM of RGSZ2. The estimated k _cat (min⁻¹) for Gαi when alone was 2.2; while in the presence of 30nM and 100nM RGSZ2 it was 4.4 (k _gap = 2.2 min⁻¹) and 8.0 (k _gap = 5.8 min⁻¹), respectively. The maximal release of Pi in these experimental conditions was about 1.1 pmol. *Significantly different from the value for Gαi alone; P<0.05. B. The effect of covalent attachment of SUMO1 on RGSZ2 GAP activity. Sumoylated RGSZ2 (200 nM) shows no GAP activity on 100 nM Gαi subunits in the single turnover assay, k _gap = 0.1 min⁻¹. C. Sumoylated RGSZ2 blocks the liberation of Pi generated by Gαi GTPase in the steady-state. *Significantly different from the value for Gαi alone; P<0.05. For all assays, triplicate samples were collected at the intervals indicated and transferred to charcoal quenching solution. The values at time 0 were subtracted. Assays were repeated twice and the results were comparable.

Figure 5. Non-covalent interaction of free SUMO1 abolishes RGSZ2 GAP activity.

A. In single turnover and steady-state GTPase assays, the endogenous GTPase activity of 100 nM Gαi subunits was unaffected by 500 nM free SUMO1 in the medium, which nonetheless inhibited the GAP activity of 200 nM RGSZ2. The k _cat values (min⁻¹) were: Gαi alone  = 2.1, Gα1 + SUMO1 = 2.0, Gαi + RGSZ2 = 7.9, Gαi + RGSZ2 + SUMO1 = 2,4. Free SUMO1 also impairs the GAP activity of the RGSZ2 RH domain. In single turnover assays of GTP hydrolysis by Gαi, the RH region shows GAP activity on Gαi subunits. This activity was blocked by the addition of SUMO1 to the medium. The k _cat values (min⁻¹) were: Gαi alone  = 2.2, + RGSZ2 RH  = 4.0, + RGSZ2 RH and SUMO1  = 2.5. The RGSZ2 K121R mutant acts as a GAP on Gαi subunits, which is blocked by free SUMO1 in the medium. The k _cat values (min⁻¹) were: Gαi  = 2.0, + K121R RGSZ2  = 6.9, + K121R RGSZ2 + SUMO1  = 2.1. *Significantly different from the value for Gαi alone; P<0.05. B. Concentration-dependent inhibition of RGSZ2 GAP activity by free SUMO1 -single turnover assay. The Gαi subunits were loaded with GTPγ³²P alone or with RGSZ2 and increasing concentrations of SUMO1. The Pi release was determined 60 s after GTPase activation. *Significantly different from the value for Gα alone; P<0.05. Inset: At 60 s after initiation of the reaction, RGSZ2 was immunoprecipitated (IP; CT antibody) and the associated Gαi proteins were analyzed in Western blots probed with an anti-Gαi antibody.

Figure 6. The RH domain SIM (141–144) binds SUMO1 and blocks RGSZ2 binding to Gα subunits.

A. Disruption of the SIM (141–144) by mutation reverses free SUMO-mediated steric hindrance of RGSZ2-Gαi binding. The I141N, I143S and L144Q RGSZ2 mutants displayed GAP activity on Gαi, even in the presence of 1 µM free SUMO1. *Significantly different from the value for Gα alone; P<0.05. B. SUMO proteins bind the I143S RGSZ2 mutant but not the double V66D+I143S mutated RGSZ2 (C), indicating the presence of a second SIM upstream of the RGS box (see Fig. 1A).

Figure 7. RGSZ2 mRNA and protein expression in mouse brain and CHO cells.

A. Identification of RGSZ2 mRNA in mouse brain and CHO cells. One hundred nanograms of total RNA isolated from (1) mouse brain and (2) CHO cells were used for RT-PCR analysis. Arrows indicate the bp PCR product of RGSZ2 obtained using alternative primers pair PP1 & PP2, see Methods. DNA Ladders: 1550, 850, 400, 200; 1031, 900, 800, 700, 600, 500, 400, 300, 200 bp. The CHO 464 bp product (2) has been intensified. B. Light micrographs taken from coronal sections through the cerebral cortex (I) and PAG (II) illustrating the localization of the RGSZ2 immunoreactivity (IQ antibody). I, shows labeled neurons pyramidal in shape in layer V of the cerebral cortex. II, represents strong immunostaining labeling cells (arrows) and processes in the PAG. Aq: midbrain aqueduct. C. Reduction of RGSZ2 mRNA expression. Confluent CHO cells (80–90%) were transfected with control siRNA or siRNA. Total RNA was extracted, retrotranscripted to cDNA and mRNA expression of RGSZ2 gene analyzed by real-time PCR. GAPDH gene expression was used as internal standard. Bars are the mean + SEM of RGSZ2 mRNA expression as percentage of control cells without being transfected. *Significantly different from the control siRNA, P<0.05. D. Reduction of RGSZ2 mRNA expression brings about decreases in the levels of the encoded protein (IQ antibody). For western blot assays CHO cells were solubilized in Laemmli buffer with reducing agents. Immunocytochemistry of CHO endogenous RGSZ2. Nuclei stained with DAPI in blue. The diminishing effect of siRNA on CHO cells RGSZ2 expression was analyzed 48 h after transfection.

Figure 8. RGSZ2 GAP activity after SUMO removal.

A. Effect of SUMO removal by SENP1 or SENP2 on neural RGSZ2 protein. RGSZ2 from SDS-octylthioglucoside solubilized synaptosomal membranes was retained by the NHS-agarose-coupled RGSZ2 antibody IQ. Affinity-purified RGSZ2 proteins were incubated alone (lane 1) or with proteases that remove SUMO (lanes 2 & 3). SENP1 preferentially removes SUMO1 whereas SENP2 removes SUMO1, SUMO2 and SUMO3. The samples were then solubilized in Laemmli buffer without reducing agents. In its absence and without SENP treatment, RGSZ2 heterocomplexes hardly enter the gel (lane 1). The anti-RGSZ2 CT was used to detect those signals. B. Sumoylated neural RGSZ2 protein lacks GAP activity but this is restored by SENP2 treatment. The observed k _cat (min⁻¹) were: Gαi alone  = 2.1; in the presence of endogenous RGSZ2  = 1.8; and of SENP2-treated endogenous RGSZ2  = s.7. Data are the mean ± S.E.M. of triplicate samples collected at the intervals indicated. *Significantly different from the value for Gαi alone; P<0.05.