Golgi export of the Kir2.1 channel is driven by a trafficking signal located within its tertiary structure

Abstract

Mechanisms that are responsible for sorting newly synthesized proteins for traffic to the cell surface from the Golgi are poorly understood. Here, we show that the potassium channel Kir2.1, mutations in which are associated with Andersen-Tawil syndrome, is selected as cargo into Golgi export carriers in an unusual signal-dependent manner. Unlike conventional trafficking signals, which are typically comprised of short linear peptide sequences, Golgi exit of Kir2.1 is dictated by residues that are embedded within the confluence of two separate domains. This signal patch forms a recognition site for interaction with the AP1 adaptor complex, thereby marking Kir2.1 for incorporation into clathrin-coated vesicles at the trans-Golgi. The identification of a trafficking signal in the tertiary structure of Kir2.1 reveals a quality control step that couples protein conformation to Golgi export and provides molecular insight into how mutations in Kir2.1 arrest the channels at the Golgi.

Figure 1. The Kir2.1 N-terminal Golgi export structure juxtaposes the cytoplasmic C-terminus and an ATS1 mutation

A. A single Kir2.1 channel subunit with relevant residues color-coded. (N-terminus, green; ATS1 mutation (Δ314-315), red; NE319, blue; hydrophobic residues in the Beta J stand, orange; transmembrane domains (TM1 and TM2), grey). B. Atomic structure Kir2.1 cytoplasmic domains channel tetramer, forming the cytoplasmic pore are shown (from Pegan et al. PDB1u4f). Standard ribbon display of one subunit (subunit A) is highlighted against a surface rendering of the other three subunits. C. Magnified view of residues at the cytoplasmic domain interface. Residues affected by the Δ314-15 ATS1 mutation, SY_314-15(red), lie under the N-terminal trafficking determinant (green).

Figure 2. The Kir2.1 ATS1 mutant, Δ314-15, accumulates in the Golgi

A. Cell surface expression quantified by HA-antibody binding and luminometry in COS7 cells expressing external HA-tagged Kir2.1(wt), Kir2.1Δ314-15 or the ER-export defective mutant, Δ FCYENE. External myc-tagged Kir channels (myc-wt) were used as a negative control. (RLU, Relative Light Units.). B. Immunocytochemical analysis of external HA-tagged Kir2.1 channels in intact (red) and permeabilized (green) COS7 cells. C. Co-localization of Kir 2.1 Δ314-15 with the Golgi marker, GM130. D. Pearson's correlation of GM130 co-localization with HA-Kir2.1Δ314-315 (Mean ± S.E.M., n=49 cells from three transfections, * P < 0.001). E-G. Kir2.1Δ314-15 channels accumulate in Golgi structures of adult rat ventricular cardiomyocytes. Shown are 3-D rendered confocal serial sections of typical cardiomyocytes of either WT Kir2.1 or Kir2.1Δ314-15 with Golgi markers. Contrasting the localization of the WT Kir2.1 (E) at the t-tubule and sarcolemma, the mutant Δ314-15 channel co-localizes with (F) GM130 and (G) TGN-38 in the cis and trans-Golgi network (see magnified insets). H. Quantification of co-localization (mean ± S.E.M., n=20, *, P < 0.001). Scale bar =14 μm. (See also supplementary Fig.1).

Figure 3. The Δ314-15 mutation blocks forward trafficking at the Golgi

A. Pulse-chase, cell-surface immunoprecipitation: Newly synthesized channels (external HA-tagged, Wild-type, WT and Δ314-15) were captured at the plasmalemma by surface HA-immunoprecipitation at the indicated chase times (10 minutes after metabolic labeling), detected by autoradiography and then compared to the total cellular signal. B. Quantification of the autoradiography (N= 3). C-D. Movement of Kir2.1 at the Golgi (WT and Δ314-315) was monitored in live COS7 cells using the reversible photoactivable protein, Dronpa. C. Shown are representative cells, studied when most of the WT Dropna-Kir2.1 channel is still located in the ER and Golgi. After total cellular fluorescence (outline traced in white) was bleached to background levels (“Photobleach”), Dronpa- was specifically photoactivated within the Golgi (dotted white line, as detected with DS-red-Golgi marker), and time-lapsed images were acquired (“Post photoactivation”). D. Quantification of the Dronpa fluorescent signal at the Golgi after photoactivation. Results are shown relative to a stationary marker (mean ± S.E.M., n=8). Bar =10 μm (See also supplementary Fig.2)

Figure 4. Golgi export signal sequence elucidated by structure-guided mutagenesis

A. Cell surface expression and Golgi localization of external HA-tagged Kir2.1 channels as quantified by surface HA-antibody binding (mean ± S.E.M., n=6, *p<0.001, cell surface normalized to total protein abundance) and co-localization with the Golgi marker GM130, using Person's correlation (mean ± S.E.M., n=15, *p<0.001). Results of alanine-substitution mutations in the cytoplasmic N- terminal region in the cytoplasmic C- terminus are shown in the top and bottom panels, respectfully. B. Key residues in the Kir2.1 cytoplasmic domain structure (PDB ID 1u4f). Left, surface rendering of the structure, showing residues in the Golgi export signal whose side chains project toward the solvent exposed surface. Residues are shown in one subunit (highlighted in dark) of a tetramer. Right, key residues are shown in a magnified view of the N-C interface. Relevant residues are color-coded. (See also supplementary Fig.3)

Figure 5. Transplantable Golgi export signal requires cytoplasmic N- and C- terminal domains

A. Cartoon of the CD8-Kir2.1 chimeras. CD8-Kir2.1(C) contains the extracellular and transmembrane domains of CD8 tailless fused to the Kir2.1 C-terminal domain. CD8-Kir2.1(N) is comprised of CD8 tailless fused to the Kir2.1 N-terminal domain. CD8-Kir2.1(NC) is a fusion of CD8 tailless with the entire Kir2.1 N-C cytoplasmic domain. Chimeras containing wild-type sequences are indicated “wt”; those bearing the ATS1 mutation are indicated as “Δ314-315”. B. Immunolocalization of the indicated CD8 chimeras, detected with anti-CD8 antibodies (red) and the trans-Golgi marker, TGN46 (green). C. Cell surface expression was quantified by surface anti-CD8 binding and luminometry (mean ± S.E.M., n=8, *p<0.001, the cell surface normalized to total protein abundance of chimera). Golgi localization was quantified by Person's co-localization with TGN46. (mean ± S.E.M., n=20, *p<0.001) Scale bar = 25 μm

Figure 6. The AP1 clathrin adaptor interacts with the Kir2.1 Golgi export signal at the TGN

A. GST and GST-Kir2.1 fusion proteins (WT and the Δ314-15 mutant) following SDS-PAGE and Coomassie Brilliant Blue Staining (CBB). B. Proteins bound to the Kir2.1 cytoplasmic domain (CD) were detected by immunoblot (IB) with indicated antibodies and quantified by densitometry. The Δ314-15 mutation specifically reduced AP-1, γ binding without affecting interaction to the Lin-7/CASK complex and Filamin A (mean ± S.E.M., n=4, *p<0.01). C. Mutations in the Kir2.1 Golgi-export signal disrupt AP-1, γ interaction. AP-1, γ bound to each of the GST-Kir2.1 fusion proteins (CBB) was detected by immunoblot with γ specific antibodies and quantified (mean ± S.E.M., n=4, *p<0.01). D. The AP-1 interaction site for the Golgi export signal is harbored with in the γσ1 subunits. Recombinant AP1 subunits were prepared as the indicated hemicomplex forms in insect cells (β1μ-1 or γσ1, β1 and γ were tagged with the Flag or HA epitopes) and tested for direct interaction with GST alone or GST-Kir2.1CD (WT vs. Δ314-15). Bound AP1 subunit was detected in Western blots using anti-epitope tag antibodies. Shown is a representative experiment N=3 E. Immunoprecipitation (IP) of HA-tagged Kir2.1 (WT or Δ314-15 Kir2.1compared to untransfected controls) followed by immunoblotting (IB) with anti-AP1 γ antibodies in COS7 cells before and after trafficking at the trans-Golgi was released (37°C) from temperature-sensitive Golgi export blockade (19°C). F. Co-localization of EGFP-Kir2.1 with AP1 (Kir2.1, green; AP1γ, red; nucleus stained blue with DAP1). Upon release from block, Kir2.1 channels travel out of large cisternal structures, and then transiently co-localize with AP1 at the TGN before they exit in vesicles and tubules. Shown is a representative cell at 15 min into chase, when Kir2.1 is observed each of these structures. Arrowheads highlight several TGN structures, where Kir2.1 and AP1 co-localize, and where tubulo-vesicular carriers, containing the channel as cargo, are emerging. Boxed area, above, is shown below in individual color panels. Scale bar = 2.5 μm. E. Quantification of co-localization, before and after release from temperature block (mean ± S.E.M, N=8, * P < 0.001).

Figure 7. The AP1 clathrin adaptor targets the channel for Golgi export

A. Western blots of AP-1-γ-adaptin or actin in cells individually treated three different AP1γ siRNA probes (AP1 γ-siRNA1-3) or a negative control probe (siRNA-untargeted). B. As measured by surface antibody binding, AP-1-γ knockdown by each probe caused a significant reduction in Kir 2.1 (Black bars, *p<0.001, mean+S.E.M., n=5) or CD8-Kir2.1 (NC) (red bars, #p<0.001, n=5) but not influenza hemagglutinin (HA) (open bars, n=5) at the plasmalemma. Expression of RNAi resistant AP-1-γ (rescue) with the knockdown probe restored surface expression of Kir 2.1 and CD8-Kir2.1(NC) (##p<0.001, n= 5) C. Localization of HA-Kir2.1 (red, left pannels), or CD8-Kir2.1(NC) (red, right panels) and TGN46 (green), in cells transfected with the indicated RNAi probes. Golgi localization was quantified by Person's co-localization with TGN46. (mean ± S.E.M., n=10, *p<0.001)