Mutation-specific peripheral and ER quality control of hERG channel cell-surface expression

Impaired functional plasma membrane (PM) expression of the hERG K+-channel is associated with Long-QT syndrome type-2 (LQT2) and increased risk of cardiac arrhythmia. Reduced PM-expression is primarily attributed to retention and degradation of misfolded channels by endoplasmic reticulum (ER) protein quality control (QC) systems. However, as the molecular pathogenesis of LQT2 was defined using severely-misfolded hERG variants with limited PM-expression, the potential contribution of post-ER (peripheral) QC pathways to the disease phenotype remains poorly established. Here, we investigate the cellular processing of mildly-misfolded Per-Arnt-Sim (PAS)-domain mutant hERGs, which display incomplete ER-retention and PM-expression defects at physiological temperature. We show that the attenuated PM-expression of hERG is dictated by mutation-specific contributions from both the ER and peripheral QC systems. At the ER, PAS-mutants experience inefficient conformational maturation coupled with rapid ubiquitin-dependent proteasomal degradation. In post-ER compartments, they are rapidly endocytosed from the PM via a ubiquitin-independent mechanism and rapidly targeted for lysosomal degradation. Conformational destabilization underlies aberrant cellular processing at both ER- and post-ER compartments, since conformational correction by a hERG-specific pharmacochaperone or low-temperatures can restore WT-like trafficking. Our results demonstrate that the post-ER QC alone or jointly with the ER QC determines the loss-of-PM-expression phenotype of a subset of LQT2 mutations.

resistances of 1.5 -3MΩ were backfilled with a pipette solution containing 135mM KCl, 5mM EGTA, 1mM MgCl2, and 10mM HEPES (pH 7.2 with KOH, ~285mOsm). The liquid junction potential (LJP) between the extracellular solution and pipette solution (1.5mV) was corrected offline using the formula Vmembrane = VPipette -VLJP as described previously 3 . All experiments were performed at room temperature (~21°C), and all cells were perfused with extracellular solution for 10 minutes prior to experimentation to ensure complete replacement of cellular media.
Command pulses were generated by a Digidata 1440A (Axon Instruments) via pClamp 10.4 software or by a Digidata 1322A digitizer (Axon Instruments) via pClamp 10.2 software. Data were acquired at 20kHz and low pass filtered at 2kHz or 3kHz.
Upon the formation of a GΩ seal and prior to membrane rupture, currents were corrected for pipette (fast) capacitance. Once ruptured, cell capacitance (picofarad; pF) was determined using a 30ms, 10mV depolarizing pulse from a holding potential of -80mV, at 2Hz. Currents were corrected for whole-cell capacitance and series resistance compensated to 80% (Axopatch) or ~100% (Alembic). All presented cells have access resistances below 15MΩ, membrane capacitances greater than 10pF, and reversal potentials between -70mV and -90mV (determined offline). Cells that did not express hERG, or were characterized as "low expressers" (i.e. 5% of or less of mean current), were excluded. Representative traces are presented in pA, current-voltage (I-V) relationships in pA/pF, and time-constants (τ) in ms.

Voltage protocols and analysis
The steady-state I-V relationship and steady-state activation curve were determined using a two-step protocol. Cells were held at -80mV, stepped in +10mV depolarizing pulses from -60mV to +50mV for 4 seconds (P1 pulse), and subsequently stepped to -50mV for 4.5 seconds (P2 pulse). The full series of currents obtained using this protocol for HA-hERG-HbH and HA-hERG control are shown in Supplementary Fig. S9. The steady-state I-V relationship was obtained by plotting the peak current at the end of the P1 pulse against the P1 pulse voltage. The steady-state activation curve was obtained by plotting the peak tail currents generated at the onset of the P2 pulse against the previous P1 pulse voltage. These values were normalized values and fit with the Boltzmann sigmoidal equation. Tables   Table T1: Characteristics of selected PAS-domain mutations Properties of LQT-associated PAS-domain mutations used in this study. Mutations categorized based on their location either at the PAS-CNBD interface or elsewhere in the PAS domain (Fig. 1). Deactivation kinetic data previously described 4 and validated for a subset of mutants (R56Q, C64Y and M124R, data not shown).

Mutation
Location Deactivation  Table T2: hERG cellular processing defects and their contribution to PM-expression Summary of empirically determined hERG PM-expression and processing defects (left) and estimated contribution of ER and peripheral QC systems to the disease phenotype (right). PMexpression determined by ELISA (Fig. 1) and expressed as fraction relative to WT following normalization for mRNA content. Maturation efficiency determined by metabolic pulse-chase ( Fig.  2) and expressed as a fraction relative to WT. It is assumed that maturation during the 3h chase is proportional to biosynthetic flux. PM-turnover measured by PM-ELISA; rate-constants of degradation determined by curve-fitting as in Fig. 3 and are expressed as fold-increase relative to WT. The ER QC contribution to overall loss-of-expression is assumed to be proportional to the ERmaturation efficiency defect; the estimated peripheral QC contribution was calculated as the difference between the ER-maturation defect and total expression defect as in Fig. 8

sequences in cytosolic regions
Tyrosine-based sorting signals and KFERQ-related motif present in the hERG cytosolic domains. The peptide motif is underlined and shown in the context of the five-adjacent flanking amino acids. Polypeptide motifs in the hERG protein sequence were identified using the ScanProSite online tool (Swiss Institute of Bioinformatics via expasy.org).   Densitometric quantification of ER-stress marker expression. Mature (complex-glycosylated) and immature (core-glycosylated) hERG indicated by solid and empty arrow, respectively. Asterisk (*) indicated non-specific band. (C) PAS-domain mutations do not increase hERG aggregation propensity. HeLa cells stably expressing HA-tagged hERG solubilized in detergent (1% Triton X-100). Detergent-soluble (S) and insoluble (pellet, P) protein fractions isolated by centrifugation at 18,000g and evaluated by immunoblotting. Equal fractions of soluble (S) and pellet (P) were loaded. GAPDH: soluble protein control. (D) Densitometric quantification of mature, immature, and total (mature + immature) hERG isolated from detergent-insoluble pellet. Detergent-insoluble hERG expressed as a fraction of total (S+P) hERG. Positive control: mutant (G601S) hERGexpressing HeLa cells treated overnight with MG132 (1µM, purple). (E-F) Similar results were obtained when whole-cell lysates were subject to high-speed centrifugation (100,000g vs. 18,000g). To aid densitometric quantification, only 1/10 th of soluble fraction was loaded relative to 18,000g and 100,000g pellets. * P < 0.05, ** P < 0.01, *** P < 0.001, n.s. indicates no significant difference (See Methods for explanation of statistical analysis). Representative immunoblots shown (uncropped images in Supplemental Fig. S13). Solid line indicates different parts of the same gel. White space indicates separate gels. Supplementary Fig. S4: Additional microscopy images of hERG lysosomal delivery Post-endocytic lysosomal delivery of additional PAS-mutant hERG (F29L, R56Q and T65P) evaluated by confocal immunofluorescence microscopy. Endocytic hERG pool labelled by Ab capture (15min at 37°C) and remaining cell-surface hERG blocked with unconjugated secondary F'ab (1h on ice). Cells then chased at 37°C for 3h prior to fixation. Lysosomal compartments labelled with LAMP1 pAb. hERG (green) and LAMP1 (magenta) staining visualized by laser confocal microscopy. Whole-cell (scale bar: 10µM, left) and high-magnification (scale bar: 5µM, right) images shown. Magnified area indicated by white box. Representative control (WT-hERG) images previously shown in Fig. 4.  Supplementary Fig. S7: Inhibition of ubiquitination and clathrin-dependent internalization (A) Domain structure of CD4-chimeric constructs. Native CD4 is a single-span transmembrane protein with an endogenous dileucine signaling motif (LL). CD4tl: endogenous cytosolic tail replaced by flexible linker. CD4tl-ub: CD4-tl construct with C-terminal fused ubiquitin (ub) which supports polyubiquitination. CD4-cc-ubΔR: coiled-coil (cc) tetramerization motif and C-terminal fused lysine-free ubiquitin (ubΔR). Lys-less ubiquitin does not support polyubiquitination, but tetramerized construct mimics multi-mono and poly-ubiquitination 1,2 . (B) Inhibition of clathrindependent internalization prevents internalization of CD4-ubiquitin chimeras. Clathrin-dependent endocytosis inhibited by hypertonic shock in media supplemented with 300mM sucrose (15min at 37°C followed by 45min at 4°C). Amount of CD4 internalized during 5-minutes measured using cell-surface ELISA. (C) Overexpression of dominant-negative ubiquitin suppresses polyubiquitination of CD4-tl-ubi. CD4 constructs coexpressed with excess empty vector, wild-type ubiquitin (ub-WT) or lys-free ubiquitin unable to support formation of linked chains (ub-DN). Overexpression of ub-DN but not the WT ubiquitin prevented the internalization of CD4-tl-ub, presumably by inhibiting formation of linked chains. Internalization of CD4-cc-ubΔR, which mimics polyubiquitinated cargo yet is unsusceptible to linked chain polyubiquitination remains unaffected. * P < 0.05, ** P < 0.01, *** P < 0.001, n.s. indicates no significant difference (See Methods for explanation of statistical analysis).

a)
…ccg ggc agt ggc gcg cca gga gga ggt ggg tct gga ggt gga gga tcc ggt ggg ggt ggg tct cat cat cac cac cat cat gct gga aag gcc ggt gaa ggt gaa atc cct gcc cct ctt gct ggt acc gtt tct aag ata ctg gta aaa gaa ggt gac act gtt aaa gct gt caa aca gtt ctg gtg ctg gag gct atg aaa atg gag aca gaa att aac gct cct act gac gga aaa gtt gaa aag gtg tta gtt aag gaa aga gat gct gtt caa ggt ggt caa ggt cta atc aag atc ggc gtt cat cat cac cac cat cat taa tga  WT-hERG (left) and WT-hERG-HBH (right) evoked from a holding potential of -80 mV, stepped in +10 mV depolarizing pulses from -60 mV to +50 mV for 4 seconds (P1 pulse), and subsequently stepped to -50 mV for 4.5 seconds (P2 pulse). (F) Steady state activation curve obtained from a plot of normalized P2 peak tail currents (at -50 mV) against P1 voltages for WT-hERG (squares) and WT-hERG-HBH (circles). Normalized plots (mean ± SEM) were fit with a Boltzmann function. (G) Steady state current voltage relationship obtained from a plot of normalized P1 peak currents against P1 voltages for WT-hERG (squares) and WT-hERG-HBH (circles). Normalized plots are presented as mean ± SEM. No statistically-significant difference was found in data shown in (F) and (G), as determined by a paired two-tailed T-test. In addition, the curves describing the kinetics of hERG current activation and inactivation were found to be indistinguishable (data not shown). Representative immunoblots shown ( Supplementary Table T3. (B) PAS mutant hERG do not appear to aggregate at the cell-surface. Cell-surface hERG labelled using HA antibody on-ice prior to fixation and permeabilization. PAS-mutant (F29L and T65P) hERG imaged with higher detector gain in order to visualize cell-surface distribution. Cell-surface staining of PAS-mutant hERG was notably weaker than WT (Fig. 1) but the distribution pattern was not appreciably different. Whole-cell (scale bar: 10µM) and high-magnification (scale bar: 2µM) images shown. Magnified area indicated by white box. (C) Overview of hERG mutation-specific proteostatic processing described in this study. The location of PAS-domain mutations (blue), severe ER-retained mutations (G601S and F805C, red) and drug-induced disruption of WT-hERG by ouabain (ouab) and desipramine (des) indicated on the hERG1a domain structure. The observed PM-expression and stability and ER-processing phenotype are noted.