ARL15 modulates magnesium homeostasis through N-glycosylation of CNNMs

Abstract

Cyclin M (CNNM1-4) proteins maintain cellular and body magnesium (Mg²⁺) homeostasis. Using various biochemical approaches, we have identified members of the CNNM family as direct interacting partners of ADP-ribosylation factor-like GTPase 15 (ARL15), a small GTP-binding protein. ARL15 interacts with CNNMs at their carboxyl-terminal conserved cystathionine-β-synthase (CBS) domains. In silico modeling of the interaction between CNNM2 and ARL15 supports that the small GTPase specifically binds the CBS1 and CNBH domains. Immunocytochemical experiments demonstrate that CNNM2 and ARL15 co-localize in the kidney, with both proteins showing subcellular localization in the endoplasmic reticulum, Golgi apparatus and the plasma membrane. Most importantly, we found that ARL15 is required for forming complex N-glycosylation of CNNMs. Overexpression of ARL15 promotes complex N-glycosylation of CNNM3. Mg²⁺ uptake experiments with a stable isotope demonstrate that there is a significant increase of ²⁵Mg²⁺ uptake upon knockdown of ARL15 in multiple kidney cancer cell lines. Altogether, our results establish ARL15 as a novel negative regulator of Mg²⁺ transport by promoting the complex N-glycosylation of CNNMs.

Keywords: CNNM2; CNNM3; Glycosylation; Magnesium transport; Protein-protein interaction.

Fig. 1

ARL15 interacts with CNNMs. a Immunoblot of GST-mCNNM2 c-tail pGex and empty pGex was used as control. The C-tail of mCNNM2 is purified and bound to glutathione beads. The DCT-enriched fraction from 8 parvalbumin GFP mice were added to the beads. b List of binding proteins to mCnnm2 C-tail (emPAI value). c Representative pathways related to metal ion transport and Golgi-trafficking identified using gene ontology biological process overrepresentation analysis with ARL15 interacting partners. d Endogenous CNNM3 was immunoprecipitated from HEK293 lysates and endogenous ARL15 co-immunoprecipitated with it. IgG antibody of the same subclass as CNNM3 antibody was used as negative control. e Anti-FLAG beads were used to immunoprecipitate overexpressed FLAG-tagged CNNM1-4 proteins. The blot shows that ARL15 interacts with CNNM1-4, as well as endogenous PRL-1 and PRL-2. emPAI Exponentially modified protein abundance index, TCL Total cell lysate, IP Immunoprecipitation, GST Glutathione-S-transferase, ARL15 ADP ribosylation factor-like GTPase 15, CNNM Cyclin M

Fig. 2

ARL15 interacts with CNNM2 cytoplasmic region. a Schematic overview of CNNM2 constructs including the predicted protein domains and truncations. b HEK293 cells were co-transfected with truncated CNNM2-HA and ARL15-Flag. The upper two blots show the detection of the Flag-tagged proteins in anti-HA precipitated cell lysates. The lower two blots show input controls of HA-tagged and Flag-tagged proteins, respectively. c Quantification of ARL15/CNNM2 binding between the different truncated CNNM2 proteins. Results are the mean ± SEM of 3 independent experiments. *indicates significant differences compared to CNNM2 + ARL15 transfected cells (P < 0.05). SP, Signal peptide; TM, Transmembrane; CBS, Cytosolic cystathionine β-synthase; CNBH, Cyclic nucleotide monophosphate-binding homology domain; ARL15, ADP ribosylation factor-like GTPase 15; CNNM, Cyclin M. d SDS-polyacrylamide gel run with the isolated fractions of the CNNM2_BAT·ARL15 complex. Bands correspond to the isolated peak (marked with black asterisk in E) of the chromatographic run of the CNNM2_BAT·ARL15 complex (M_{th(ARL15monomer)} = 19.57 kDa and M_{th(CNNM2BATmonomer)} = 17.72 kDa, where M_th indicates the theoretical molecular weight of each subunit). e Gel filtration of the CNNM2_BAT.ARL15 complex and of its individual components. Bands correspond to the peak (marked with black asterisk) of the chromatographic run of the complex CNNM2_BAT·ARL15 (M_{th(ARL15monomer)} = 19.57 kDa and M_{th(CNNM2BATmonomer)} = 17.72 kDa, where M_th indicates the theoretical molecular weight of each subunit). f The calibration trendline calculated from protein standards is y = − 0.2451x + 1.5726 (y = K_av; x = logM).

Fig. 3

Computational structural model of the interaction between the CNNM2 cytosolic domains and ARL15. a Richardson (ribbon) diagram of the cytosolic CNNM2 dimer domains. Each monomer is in a different color, and the model of flexible linker joining the Bateman domains to the CNBH ones are in light green. b Electrostatic potentials at the surface of CNNM2 cytosolic domains. Electrostatic grids were generated with APBS 3.0 (Baker et al. 2001) at 150 mM ionic strength. c Best solution of ARL15 (red ribbons) docking to CNNM2 computations. d Dissection of the interface between ARL15 and CNNM2 (middle) illustrating both the surface and charge complementarity between the two partners.

Fig. 4

ARL15 and CNNMs co-localize in the Golgi system. a Immunocytochemistry of HEK293 overexpressing ARL15, CNNM2 and mCherry-tagged Golgi-apparatus marker, B4GALT1. Nuclei are stained with DAPI. b Immunocytochemistry of HEK293 overexpressing ARL15, CNNM3 and mCherry-tagged Golgi-apparatus marker, B4GALT1. Nuclei are stained with DAPI. c Mouse kidneys were permeabilized and co-immunostained with anti-ARL15 and anti-CNNM2. Merged pictures stained for ARL15 in green, CNNM2 in red and DAPI (nuclei) in blue. Bar in figure A and B represents 5 μm, bar in figure C represents 50 μm. Representative image presented of three independent experiments with three replicates. G, Glomerulus, PT, Proximal tubule, DCT, Distal convoluted tubule. d SK-RC-39 cells were co-transfected with ARL15-mClover3, CNNM3-mCherry (WT or N73A N-glycosylation mutant) and pmTurquoise2-ER or -Golgi. Plasma membrane localization is indicated with arrows, co-localization in the Golgi with triangles and co-localization in the ER with asterisks.

Fig. 5

CNNM3 N-glycosylation is modulated by ARL15 and Mg²⁺ a Wild-type and N-glycosylation mutant CNNM3 were treated with PNGase F and Endo H glycosidases to assess the presence of different glycoforms of CNNM3 and to confirm asparagine 73 as the site of glycosylation. b Schematic representation of different types of glycans. c Lectin gel-shift assay of CNNM3 indicates that N73 is the only site of N-glycosylation d Overexpression of ARL15 increases complex CNNM3 N-glycosylation in kidney cancer cells. e SK-RC-39 cells were grown in the presence of absence of magnesium in the media and the status of CNNM3 glycoforms was assessed using western blotting. CTRL Control, sg Guide RNA, Scr scramble, ARL15 ADP ribosylation factor-like GTPase 15, CNNM cyclin M, PNGase F peptide:N-glycosidase F, Endo H endoglycosidase H.

Fig. 6

ARL15 affects Mg²⁺ flux and ATP production. a ²⁵Mg²⁺ uptake in stably overexpressed and knockdown ARL15 ACHN, Caki-1, RCC4, and SK-RC-39 cells. These various types of renal carcinoma cell lines were incubated with ²⁵Mg²⁺ for 15 m and results were normalized to 0 m. Each data represents the mean of 3 independent experiments ± SEM. *indicates significant difference compared to control cells. b ATP production in stably overexpressed and knockdown ARL15 SK-RC-39 and RCC4 cells. Results are the mean ± SEM of 3 independent experiments. Sg Guide RNA, Scr scramble, OE overexpression, ARL15 ADP ribosylation factor-like GTPase 15, CNNM cyclin M.

Fig. 7

Summary of the effect of ARL15 CNNM complex on Mg²⁺ flux. In the presence of ARL15, it interacts with CNNMs in the ER and Golgi, resulting in their complex N-glycosylation, which in turn decreases Mg²⁺ uptake. In the absence of ARL15, CNNMs are found in less complex glycoforms and that results in increased Mg²⁺ uptake. ER endoplasmic reticulum, ARL15 ADP ribosylation factor-like GTPase 15, CNNM cyclin M.