Selenoprotein N is an endoplasmic reticulum calcium sensor that links luminal calcium levels to a redox activity

Abstract

The endoplasmic reticulum (ER) is the reservoir for calcium in cells. Luminal calcium levels are determined by calcium-sensing proteins that trigger calcium dynamics in response to calcium fluctuations. Here we report that Selenoprotein N (SEPN1) is a type II transmembrane protein that senses ER calcium fluctuations by binding this ion through a luminal EF-hand domain. In vitro and in vivo experiments show that via this domain, SEPN1 responds to diminished luminal calcium levels, dynamically changing its oligomeric state and enhancing its redox-dependent interaction with cellular partners, including the ER calcium pump sarcoplasmic/endoplasmic reticulum calcium ATPase (SERCA). Importantly, single amino acid substitutions in the EF-hand domain of SEPN1 identified as clinical variations are shown to impair its calcium-binding and calcium-dependent structural changes, suggesting a key role of the EF-hand domain in SEPN1 function. In conclusion, SEPN1 is a ER calcium sensor that responds to luminal calcium depletion, changing its oligomeric state and acting as a reductase to refill ER calcium stores.

Keywords: SEPN1; calcium sensor; endoplasmic reticulum; stress of the endoplasmic reticulum.

Fig. 1.

SEPN1 is a type II ER membrane protein. (A) Schematic representation of the recombinant SEPN1 construct exploited in the experiments related to this panel and indicating the four glycosylation sites: Asn156, Asn449, Asn471, and Asn497. FLAG Immunoblot representing FLAG-SEPN1 from transfected cells from which proteins were extracted with lysis buffer, the buffer was exchanged by PD-10 desalting column and the proteins digested by EndoH. Ponceau stain served as a protein loading control. (B) Schematic representation of the two recombinant SEPN1 constructs used in the experiments related to this panel. Human SEPN1 was fused either to a His-6-tag at its C-terminal (SEPN1-HIS) or to a Strep-tag at its N-terminal end (STREP-SEPN1). The TM domain is the predicted transmembrane domain between amino acids 31 and 50, as predicted by TMHMM Server version 2.0; the U428C mutation is indicated. HeLa cells expressing SEPN1-HIS or STREP-SEPN1 or untransfected were subjected to anti-STRP, anti-HIS, and anti-SEPN1 antibody detection. Membrane-enriched protein fractions were treated with detergent Triton-X100, followed by limited trypsin digestion. Microsome-enriched fractions were subjected to trypsin digestion in the absence of the detergent, to maintain membrane integrity. Both anti-tags and anti-SEPN1 detection indicated the expression of an intact SEPN1 of ∼75 kDa. Trypsin treatment applied to the native membrane fractions removed the N-terminal Strep-tag but preserved the His-tag. Both tags were digested in the detergent-dissociated membrane fractions treated with trypsin, and short-degraded forms of SEPN1 were detected.

Fig. 2.

Ca²⁺ binding and Ca²⁺-dependent conformational change in SEPN1. (A) Alignment of SEPN1 predicted EF-hand sequence to the EF-hand HMM logo plot (obtained from PFAM; https://pfam.xfam.org/family/PF00036#tabview=tab4). The important conserved amino acid D80 within the SEPN1 sequence, depicted in red, was mutated to A. Pathogenic mutants identified within this sequence are highlighted by a star and correspond to the mutants M85V and Y86C. Residue numbering is based on UniProt entry Q9NZV5. (B) CD spectra of the indicated peptides at different calcium concentrations (0 to 5 mM). (C) CD (mdeg) values at 220 nm of the WT peptide at different calcium concentrations. (D) K_d^Ca2+ of WT calculated from CD values at 220 nm in four different experiments with different peptide preparations. (E) Analysis of the secondary structure of the WT by the K2D algorithm (dichroweb.cryst.bbk.ac.uk/).

Fig. 3.

Ca²⁺-dependent oligomeric change of SEPN1 in cells. (A) Scheme of FLAG-SEPN1 protein and its mutants. The EF-hand domain, the Trx domain, and all the amino acid mutants are noted. Below are nonreducing and reducing FLAG immunoblots of the indicated FLAG-SEPN1 and its mutants. The truncated, monomeric, and oligomeric SEPN1 are indicated. The asterisk indicates an additional band that is visible for the selenocysteine-containing form and whose origin was not analyzed. Membrane staining with Ponceau S is shown as protein-loading control. (B) Scheme of FLAG-SEPN1 and MYC-SEPN1 protein. Immunoblot of MYC-tagged SEPN1 immunopurified with FLAG-M2 antibody from lysate of HeLa cells that were untransfected or transfected with expression plasmids of the indicated proteins. The lower two panels represent 5% of the total input protein lysate immunopurified. The proteins were resolved on reducing SDS/PAGE. (C) Sucrose gradient analysis of FLAG-SEPN1 and FLAG-SEPN1^D80A in HeLa cells treated for 2 h with DMSO or Tg. Samples were analyzed on 4% to 20% sucrose gradients. Equal aliquots of the 19 fractions were analyzed by FLAG immunoblotting. The positions and the molecular weight of size markers (BSA, IgG, and catalase) are established from Coomassie blue-stained gel. The tables on the right indicate the percentages of the monomer and the oligomer. (D) Nonreducing FLAG immunoblotting of the peak fractions of BSA (of similar size as the SEPN1 monomer) and IgG (of similar size as the oligomer) indicating that SEPN1 protomers are kept together to form the oligomer not only by a disulfide bridge (running as an oligomer), but also by weak interactions (running as a monomer).

Fig. 4.

The attacking amino acid of SEPN1 in Trx domain is present in a reduced form. (A) Schematic representation of a SEPN1 construct exploited in the experiments related to this figure. (B) Coomassie blue-stained nonreducing SDS/PAGE of FLAG-immunopurified SEPN1 after treatment with Tg or DMSO indicating the bands of the monomeric (M) and oligomeric (O) form that were cut and the redox state of cysteines analyzed by nLC-ESI-MS/MS sequence analysis. Lane 1, transfected SEPN1^U428C and treated with DMSO; lane 2, empty; lane 3, transfected with SEPN1^U428C and treated with Tg. (C) Representative MS/MS spectrum for the peptide (423 to 432) derived from the monomer of SEPN1 and bearing NEM-alkylation of cysteines 427 and 428. (D) The redox state of cysteines 427 and 428 in SEPN1 after treatment with Tg or DMSO. Alkylation is reported as +NEM if the cysteines are alkylated by NEM and present in a reduced form or as +carbamidomethyl if the cysteines are alkylated by IAA and present in an oxidized form. The probability of different alkylation types (derived from MaxQuant analysis) is reported in parentheses.

Fig. 5.

The SEPN1 trapping mutant in conditions of low luminal calcium engages a higher number of interactors. (A) Schematic representation of SEPN1 constructs exploited in the experiments of this figure. FLAG and SERCA2 immunoblots of FLAG-tagged SEPN1 immunopurified with FLAG-M2 antibody from lysate of cells that were untransfected or transfected with expression plasmids of the indicated proteins and treated with Tg or DMSO. The lower two panels represent the 5% of the total input protein lysate immunopurified. The proteins were resolved on reducing SDS/PAGE. The graph on the right shows the relative levels of SERCA2 associated with its bait, FLAG-SEPN1 (set to 1 in no Tg) in arbitrary units in three different experiments (n = 3; P < 0.05, unpaired t test). (B) Coomassie blue-stained nonreducing SDS/PAGE of FLAG-immunopurified SEPN1 from cells after treatment with Tg or DMSO indicating the bands of the monomeric and oligomeric form. Each lane was divided into three slices (1 to 3) that were cut and analyzed for protein identification by mass spectrometry. Lane 1, transfected with empty expression vector; lane 2, empty; lane 3, transfected with SEPN1^{C427S, U428C} and treated with Tg; lane 4, empty; lane 5, transfected with SEPN1^{C427S, U428C} and treated with DMSO; lane 6, empty; lane 7, transfected with SEPN1^{C427S, U428S} and treated with Tg; lane 8, empty; lane 9, transfected with SEPN1^{C427S, U428S} and treated with DMSO. (C) Unsupervised hierarchical clustering heat maps comparing SEPN1^{C427S, U428C} with SEPN1^{C427S, U428S} and SEPN1^{C427S, U428C} with SEPN1^{C427S, U428S} treated with Tg. Numbers 1 to 3 indicate the interactors identified in the slices 1 to 3 of the Coomassie blue-stained nonreducing SDS/PAGE in B.

Fig. 6.

A reducing shift is lacking in SEPN1 KO cells after luminal calcium depletion. (A) Fluorescent photomicrographs of WT and SEPN1 KO cells transiently expressing an ER-localized roGFP2, immunostained for PDI as an ER marker. The merged panels with orthogonal views show an overlap of the roGFP2 signal with PDI (Scale bar: 20 μm.) (B) Traces of time-dependent changes in the fluorescence excitation ratio of roGFP2, reflecting the alterations in the redox state of roGFP2 localized in the ER of WT and SEPN1 KO cells. Cells were exposed to a DTT pulse of 20 min, followed by washout of the reductant. (C) Cells were exposed to the irreversible SERCA inhibitor Tg for 2 h. Each data point represents the mean ± SEM of the fluorescence excitation ratio of roGFP2. The experiment was reproduced five times, with similar results (P < 0.001, two-way ANOVA) (SI Appendix, Fig. S5B).

Fig. 7.

SEPN1 working model. Within the ER/SR, SEPN1 senses calcium levels by binding this ion through an EF-hand domain. When luminal calcium concentration drops (below 100–300 μM), the SEPN1 conformation changes and causes a shift toward a monomeric form, which is redox-active towards its partner (among others) SERCA. Activation of SERCA then leads to calcium entry into the ER and refilling of calcium stores.