Although it is well known that ion channels conduct ions across biomembranes, whether ions are conducted by some non-membrane proteins is not known because of the lack of a detection method. Calsequestrin-2 (CSQ2) is a sarcoplasmic reticulum (SR) Ca2+-binding protein suppling Ca2+ for the ryanodine receptor Ca2+ release during the excitation–contraction coupling in cardiomyocytes. CSQ2 mutations, even in some heterozygous occasions, causes catecholaminergic polymorphic ventricular tachycardia (CPVT2), suggesting that CSQ2 may function beyond a Ca2+ buffer. Here, we identify a non-transmembrane channel in Ca2+-enriched CSQ2 dimers, which facilitates fast Ca2+ mobilization. Using crystallography, we solved the high-resolution structure of Ca2+-bound CSQ2 and discovered that the negatively charged residues at the dimer interface encompassed a tubular channel-like structure, dubbed “tunnel,” in which ∼15 Ca2+ ions aligned across the ∼5 nm tunnel path. To determine the potential tunnel conductance, we developed a graphene-based nanoelectronic technology to connect a CSQ2 dimer into a nanocircuit. In the Tyrode solution containing 1 mM Ca2+, a CSQ2 dimer exhibited a conductance one order of magnitude higher than the background level. This conductance was Ca2+ dependent, and was largely suppressed by the single-residue mutation D309N at the bottleneck region of the tunnel path, indicating that the tunnel conducted Ca2+ fluxes. When the D309N mutant CSQ2 was expressed in wild-type rat cardiomyocytes by adenoviral vectors, isoproterenol treatment induced chaotic Ca2+ waves, delayed after-depolarizations and trigged activities resembling those occurring in CPVT2 models. This dominant negative effect of CSQ2 mutation agreed well with our structural observation that CSQ2 tunnels were interconnected to form a tunnel network. Taken together, these results revealed that CSQ2 builds a nano-highway network for energy-efficient Ca2+ mobilization in the SR. Factors that block the Ca2+ highway may lead to arrhythmogenesis.

This work was supported by the State R&D Program (2016YFA0500401).

Author notes


X.-X. Fan, Z. Liu, and H. Yu contributed equally to this abstract.

This article is available under a Creative Commons License (Attribution–Noncommercial–Share Alike 4.0 International license, as described at