Endocytosis of HERG is clathrin-independent and involves arf6

Abstract

The hERG potassium channel is critical for repolarisation of the cardiac action potential. Reduced expression of hERG at the plasma membrane, whether caused by hereditary mutations or drugs, results in long QT syndrome and increases the risk of ventricular arrhythmias. Thus, it is of fundamental importance to understand how the density of this channel at the plasma membrane is regulated. We used antibodies to an extracellular native or engineered epitope, in conjunction with immunofluorescence and ELISA, to investigate the mechanism of hERG endocytosis in recombinant cells and validated the findings in rat neonatal cardiac myocytes. The data reveal that this channel undergoes rapid internalisation, which is inhibited by neither dynasore, an inhibitor of dynamin, nor a dominant negative construct of Rab5a, into endosomes that are largely devoid of the transferrin receptor. These results support a clathrin-independent mechanism of endocytosis and exclude involvement of dynamin-dependent caveolin and RhoA mechanisms. In agreement, internalised hERG displayed marked overlap with glycosylphosphatidylinositol-anchored GFP, a clathrin-independent cargo. Endocytosis was significantly affected by cholesterol extraction with methyl-β-cyclodextrin and inhibition of Arf6 function with dominant negative Arf6-T27N-eGFP. Taken together, we conclude that hERG undergoes clathrin-independent endocytosis via a mechanism involving Arf6.

Figure 1. Surface density of HA-hERG containing channels.

(A) Schematic of the transmembrane topology of the hERG subunit illustrating the insertion site of the HA sequence. (B) HEK-MSR cells were transfected with various ratios of HA-hERG and hERG expression vectors. Surface (non-permeabilised) and total (permeabilised) levels of HA-hERG_hERG were quantified using TMB substrate (n ≥ 4).

Figure 2. Internalised hERG channels and transferrin predominantly localise to distinct endosomes.

HeLa cells transfected with HA-hERG_hERG were incubated with anti-HA prior to washing, fixing, permeabilisation and staining with Cy3 conjugated secondary antibody. Alexa Fluor^® 488-conjugated Tfn was used as marker of the Tfn receptor. Different incubation times (indicated in minutes in bottom left corner of images) at 37°C were used to compare protein distributions following 5 minutes internalisation (A) or after allowing sufficient time to label recycling pathways (C). Incubation at 16°C (B) inhibits trafficking of the Tfn receptor beyond EE, facilitating assessment of early trafficking events. Amplified boxes highlight points of co-localisation and bars = 10 µm.

Figure 3. Internalised hERG channels co-localise with CIE cargo.

(A) HEK-hERG cells transfected with HA-hERG and MHCI or GFP-GPI were incubated with anti-HA for 2 hours at 37°C; anti-HLA (MHCI) or anti-GFP respectively were included for the last hour. Cells were stained with Alexa Fluor^® 488 (MHCI) or Alexa Fluor^® 633 (GFP-GPI) -conjugated antibodies (both pseudo-coloured green), along with Cy3 conjugated secondary antibody (HA-hERG). (B) Internalised GFP-GPI distribution in H9c2 cells co-expressing hERG and GFP-GPI was determined as described for (A). Total hERG was stained after fixation with anti-Kv11.1 and Cy3-conjugated secondary antibody. Arrows and amplified boxes highlight points of co-localisation and bars = 10 µm.

Figure 4. Clathrin-independent hERG internalisation.

(A) The distribution of HA-hERG_hERG or HA-K_ATP transiently expressed in HeLa cells was compared after 30 minutes incubation with anti-HA in the presence or absence of 80 µM dynasore, 5 mM MβCD, 5 mM αCD or 15 µM/50 µM SecinH3. Cells were pre-incubated with each drug for 30 minutes. After fixation total HA-hERG_hERG or HA-K_ATP were stained with anti-Kv11.1 (recognising a C-terminal epitope) or anti-SUR1 respectively. Bar = 20 µm. (B-C) Internalisation of HA-hERG_hERG was quantified in HEK-MSR cells. After 30 minutes at 37°C with anti-HA cells were fixed, incubated with HRP-conjugated secondary antibodies and exposed to TMB substrate. Internalisation was considered as a ratio of surface (non-permeabilised cells) to surface + internalised (permeabilised cells). Increases in this ratio, indicative of decreased internalisation, were probed for after treatment with 80 µM dynasore and 15 µM SecinH3 (B) (n = 6) or co-transfection with dominant negative Rab5 and Arf6 expression vectors (C) (n = 4).

Figure 5. Arf6-Q67L-eGFP positive vacuoles contain hERG channels.

hERG channels expressed in HeLa (A) or H9c2 cells (B&C) were stained with anti-Kv11.1 and Cy3-conjugated secondary antibodies and their distribution compared with co-expressed Arf6-Q67L-eGFP (A&C). In H9c2 cells Arf6-Q67L-eGFP (C) causes an increased central localisation of hERG compared with control (B). Amplified boxes highlight points of co-localisation and bars = 10 µm.

Figure 6. Arf6-T27N increases hERG current density.

Assessment of dominant negative Arf6, cdc42 and µ2 constructs on hERG current density. (A) Current-voltage relationships of hERG channels co-expressed with Arf6-T27N-eGFP, eGFP-cdc42-T17N or µ2-D176A/W421A in HEK-MSR cells; representative current families are shown. (B) Current-voltage relationships of tail currents determined as in (A) (n = 4-9). (C) Current densities measured from tail currents at -60 mV after a +60 mV prepulse (data from (B)) (n = 4-9).

Figure 7. Clathrin-independent internalisation in myocytes.

Effects of 80 µM dynasore, 3 mM MβCD, 3 mM αCD and 50 µM SecinH3 on internalisation of native ERG in NRVCM at 37°C. Cells were incubated for 1 hour with anti-Kv11.1 (extracellular epitope), fixed and stained with Cy3-conjugated secondary antibodies. In the bottom row, cells were washed with acidic buffer prior to fixation to remove surface bound antibodies. Staining of nuclei (inset) confirms the presence of cells within the field of view for 4°C and MβCD. Bar = 10 µm.