Physiological CRAC channel activation and pore properties require STIM1 binding to all six Orai1 subunits

Abstract

The binding of STIM1 to Orai1 controls the opening of store-operated CRAC channels as well as their extremely high Ca²⁺ selectivity. Although STIM1 dimers are known to bind directly to the cytosolic C termini of the six Orai1 subunits (SUs) that form the channel hexamer, the dependence of channel activation and selectivity on the number of occupied binding sites is not well understood. Here we address these questions using dimeric and hexameric Orai1 concatemers in which L273D mutations were introduced to inhibit STIM1 binding to specific Orai1 SUs. By measuring FRET between fluorescently labeled STIM1 and Orai1, we find that homomeric L273D mutant channels fail to bind STIM1 appreciably; however, the L273D SU does bind STIM1 and contribute to channel activation when located adjacent to a WT SU. These results suggest that STIM1 dimers can interact with pairs of neighboring Orai1 SUs. Surprisingly, a single L273D mutation within the Orai1 hexamer reduces channel open probability by ∼90%, triples the size of the single-channel current, weakens the Ca²⁺ binding affinity of the selectivity filter, and lowers the selectivity for Na⁺ over Cs⁺ in the absence of divalent cations. These findings reveal a surprisingly strong functional coupling between STIM1 binding and CRAC channel gating and pore properties. We conclude that under physiological conditions, all six Orai1 SUs of the native CRAC channel bind STIM1 to effectively open the pore and generate the signature properties of extremely low conductance and high ion selectivity.

Figure 1.

An L273D Orai1 SU can bind STIM1 only if adjacent to a WT SU. E-FRET at the plasma membrane was measured between CFP-2xOrai1 and STIM1, CAD, or SOAR concatemers expressed in HEK293 cells. (A) Fluorescence and E-FRET images of a cell coexpressing YFP-STIM1 and CFP-2xOrai1, after store depletion by TG treatment. FRET lookup table range, 0–0.35. (B) E-FRET between Orai1 dimer variants and STIM1. Data from individual cells are plotted with means ± SEM. P > 0.99 between WT-ΔCT and WT-PolyMut cells (NS), and P < 0.0001 for all other pairwise comparisons (all P values calculated using one-way ANOVA with Tukey’s correction for multiple comparisons). (C) E-FRET between Orai1 dimer variants and CAD. P > 0.99 between WT-ΔCT and WT-PolyMut cells; P = 0.0001 between WT-WT and WT-L273D, WT-PolyMut, or WT-ΔCT; and P = 0.02 between WT-L273D and WT-PolyMut or WT-ΔCT. (D) E-FRET between Orai1 dimer variants and the dimeric SOAR concatemer (S-S). Unlike STIM1 or CAD, S-S FRETs equally well with WT-PolyMut and WT-L273D. P = 0.75 between WT-L273D and WT-PolyMut cells (NS), and P < 0.0001 for all other pairwise combinations. (E) Patch-clamp recordings of currents in 20 mM Ca²⁺ from HEK293 cells expressing full-length mCherry-STIM1 with CFP- or GFP-tagged 2xOrai1 dimer variants after passive store depletion. Peak currents during steps to −100 mV from individual cells are shown with means ± SEM. P < 0.05 for all comparisons between WT-WT and mutant Orai dimers. Representative I-V relations are displayed in the boxed inset.

Figure 2.

The L273D Orai1 SU contributes to CRAC channel activation when adjacent to a WT SU. Whole-cell currents measured at −100 mV in 20 mM Ca²⁺ from HEK293 cells cotransfected with GFP-labeled Orai1 hexameric concatemers and mCherry-STIM1. (A) Current densities of individual cells expressing WT, 1xL273D, 1xΔCT, and 1xPolyMut hexamers are displayed with means ± SEM. WT and 1xL273D data are reproduced from a previous publication (Yen et al., 2016) for comparison, and L273D mutants in SUs 1, 3, and 6 were pooled. P = 0.04 between 1xL273D and 1xΔCT, and P > 0.05 between 1xPolyMut and 1xL273D or 1xΔCT (Mann-Whitney U test). (B) Representative I-V relations of individual cells expressing the indicated hexameric Orai1 variant.

Figure 3.

Relationship between current amplitude and the number of L273D and L286S mutations. Current densities of HEK293 cells coexpressing GFP-labeled Orai1 hexameric concatemers and mCherry-STIM1 were recorded as in Fig. 2, normalized to that of WT (0 mutations per hexamer, −15.6 pA/pF) and plotted as a function of mutations per channel. WT and 1xL273D values are reproduced from Yen et al. (2016). Each filled circle represents the mean ± SEM of 7–23 cells for the specified hexamer variant. The corresponding solid lines describe exponential fits given by I = kⁿ, where I is the normalized current amplitude, n is the number of mutations, and k is the fractional activity for a single mutation (0.36 for L273D and 0.56 for L286S). For L273D mutations, data were pooled for single mutants made in SUs 1, 3, and 6, double mutants in SU1+2, 2+3, and 1+3, and triple mutants in SU1+3+6. For L286S, the single mutation was placed in SU1, double mutation in SU1+3, and triple mutation in SU1+3+6.

Figure 4.

Noise analysis of Orai1 channels carrying a single L273D mutation. (A) A cell coexpressing 1xL273D Orai1 hexamer + mCherry-STIM1 was store-depleted and briefly exposed to DVF solutions containing differing free [Ca²⁺]_ext to cause graded block of CRAC channels (colored bars). Current was measured at −100 mV, and cells were returned to 20 mM Ca²⁺ between DVF exposures. (B) 200-ms sweeps collected at the times indicated by the corresponding colored dots in A. (C) Variance versus mean current plots in two cells expressing WT (black) or 1xL273D (red) Orai1 hexamers. The curves show the best fits of σ² = Ii – I²/N to the data, with N ∼7,300 and 540, and i = −81 and −321 fA for WT and 1xL273D channels, respectively. Data from the WT cell are reproduced from Yen et al. (2016). Dashed lines indicate i values corresponding to the variance/current slope at the limit of maximum block (P_o = 0). Maximum P_o values in DVF calculated as I/Ni are indicated for each cell. (D) Estimating unitary Na⁺ current through the 1xL273D Orai1 hexamer. Pooled data from four cells shows that σ²/mean current increases linearly with 1-P_o with a slope of −248 ± 22 fA (95% confidence limits, r² = 0.75) indicating the unitary current amplitude. (E) Ca²⁺ block of Na⁺ current in two cells expressing WT (black) or 1xL273D Orai1 hexamers (red). Lines represent fits to the Hill equation, $block=1/\left[1+{\left({\mathrm{K}}_{1/2}/\left[\mathrm{Ca}\right]\right)}^{{\mathrm{n}}_{\mathrm{H}}}\right]$. K_1/2 was 25 and 42 µM, with a Hill coefficient (n_H)of 1.2 and 0.7 for WT and 1xL273D channels, respectively.

Figure 5.

Ion selectivity of L273D currents. I-V relations of hexameric WT and 1xL273D channels in cells coexpressing mCherry-STIM1. Selectivity in physiological Ca²⁺ was assessed in 2 mM Ca²⁺ Ringer’s, in the presence of extracellular Na⁺ (black) or after replacement of Na⁺ by NMDG (red). Exemplar I-V relations of a cell coexpressing mCherry-STIM1 with WT (A) or 1xL273D (B) hexamer channels. No change was observed when extracellular Na⁺ was removed, indicating that Na⁺ does not permeate either WT or 1xL273D channels. (C) I-V relations in DVF Ringer’s from WT (black) or 1xL273D (red) channels were normalized to their peak current at −100 mV and averaged (n = 3–4 cells per variant). The SEM is plotted for every 10th point. E_rev was reduced in 1xL273D cells (25.9 ± 2.0 mV compared with 45.8 ± 1.2 mV in WT cells; see inset), indicating an increased Cs⁺ to Na⁺ permeability. The average I-V relation for WT Orai1 hexamer is reproduced from Yen et al. (2016) for comparison.