SUMOylation silences heterodimeric TASK potassium channels containing K2P1 subunits in cerebellar granule neurons

Abstract

The standing outward K(+) current (IKso) governs the response of cerebellar granule neurons to natural and medicinal stimuli including volatile anesthetics. We showed that SUMOylation silenced half of IKso at the surface of cerebellar granule neurons because the underlying channels were heterodimeric assemblies of K2P1, a subunit subject to SUMOylation, and the TASK (two-P domain, acid-sensitive K(+)) channel subunits K2P3 or K2P9. The heterodimeric channels comprised the acid-sensitive portion of IKso and mediated its response to halothane. We anticipate that SUMOylation also influences sensation and homeostatic mechanisms in mammals through TASK channels formed with K2P1.

Figure 1. Interaction of native proteins at the neuronal surface: FRET between SUMO1 and K2P1, SUMO1 and K2P3, SUMO1 and K2P9, but not SUMO1 and K2P2

FRET in CGN was performed using antibodies to K2P1, K2P2, K2P3, or K2P9 to tag them with the donor fluorophore (green) and antibodies to label SUMO1 with the acceptor (red). Images are representative of 5–6 fields of view per K2P channel. Each row shows four small images of: channel (top left), SUMO1 (top right), SUMO1 after photobleaching (BLEACH), and increased donor fluorescence after photobleaching (FRET); the large panel in each row is an overlay of the channel and the resulting FRET. Images are pseudo-colored to indicate FRET efficiencies calculated for each pixel (calibration bar).

Figure 2. SUMO1 modulates IKso and K2P1 channels but not homodimeric K2P2, K2P3 or K2P9 channels

A. Rat CGN were studied in whole-cell mode with only solution B (control) in the pipette or with SUMO1 or SENP1 added (n = 10 cells). Each cell was perfused with solution A at pH 6.4 (open circle), 7.4 (gray circle) and 8.4 (closed circle). Top, IKso plotted against voltage. Example traces are shown at the top of each row. Bottom, Mean IKso is measured at −20 mV ± SEM (left). Vm is plotted as mean ± SEM (middle). R_IN is plotted as mean ± SEM (right). Excitability parameters (IKso, Vm and R_IN) are reported in Table S1. B. Human K2P1 or K2P3 channels heterologously-expressed in CHO cells and studied in inside-out, off-cell patches at 50 mV (normalized mean ± S.E.M, n = 7 patches). Top left, K2P1 channels, negligible basal currents were activated by SENP1 and suppressed by SUMO1. Reapplication of SENP1 restored channel activity despite extended washing (Wash). Top right, K2P3 channels were insensitive to application of SUMO1 and SENP1 (412 ± 36 pA). Bottom, current families and current-voltage relationships at times indicated by symbols. C. Histograms for normalized current (mean ± SEM) for recordings of K2P1 channels and K2P3 channels in panel B and for K2P9 channels and K2P2 channels when studied as in (B) (n = 7 patches). Unlike K2P1, the other homomeric channels were insensitive to SUMO1 and SENP1 (K2P9 channels, basal 345 ± 32 pA; K2P2 channels, basal 441 ± 43 pA).

Figure 3. FRET and protein purification both show that K2P1 assembles with K2P3 or K2P9

FRET was studied in live CHO cells. Western blots are representative of 3 experiments. A. CFP-tagged K2P1, K2P2, K2P3 or K2P9 subunits reach the cell surface (left). Exemplar photobleaching studies (right) show the decay of fluorescence intensity for single cells expressing CFP-K2P1 and YFP-K2P1 (○) or CFP-K2P1 and free YFP (●) fit by an exponential to give τ. Scale bar is 10 μm. B. FRET shows assembly of CFP-tagged K2P1, K2P3 or K2P9 with YFP-tagged K2P1, K2P3 or K2P9 but not with YFP-K2P2 or free YFP. CFP-K2P1 also did not interact with YFP-Kv2.1. CFP-K2P2 does assemble with YFP-K2P2 and YFP-K2P23Δ. Dotted lines show FRET between CFP and YFP-tagged subunits in homodimeric channels (K2P1, K2P3, K2P9 or K2P2). Significant changes in τ compared to free YFP are indicated (*, P<0.001) (Table S2). C. Assembly of K2P1 and K2P3 (or K2P1 and K2P9) was demonstrated by purification of unlabeled K2P1 (lane 3, ~37 kDa) by 1d4-mediated immunopurification when K2P3-1d4 (top panel) or K2P9-1d4 (bottom panel) was co-expressed; Western blot staining was with antibodies directed to K2P1. Controls: K2P1 was not isolated without K2P3-1d4 or K2P9-1d4 (lane 1); IgG antibodies did not isolate K2P1 (lane 4); staining with anti-1d4 antibodies confirmed K2P3-1d4 or K2P9-1d4 expression (lanes 7 and 8; ~44 and ~42 kDa respectively); starting material (SM) was shown to contain the expressed subunits (lanes 5 and 10). Black lines indicate where interceding lanes have been removed.

Figure 4. FRET shows that K2P1 recruits SUMO1 into heterodimeric channels with K2P3 or K2P9

Live CHO cells studied as in Figure 3. Fluorescence decay parameters (mean τ ± S.E.M.) are listed in Table S3. A. Assembly of CFP-K2P3 and wild-type K2P1 (WT) allows FRET with YFP-SUMO1 whereas K2P1-K274Q (Q) precludes both sumoylation and FRET. B. K2P1 is required for FRET between CFP-K2P3 (or CFP-K2P9) and YFP-SUMO. FRET is observed in homodimeric channels between subunits tagged with CFP and YFP (dotted lines); CFP-K2P3 or CFP-K2P9 show FRET with YFP-SUMO1 only when wild-type K2P1 (WT) is also expressed (NS,, indicates no significant change compared to homodimeric channel FRET signal). Controls: FRET was not observed with a linkage-deficient SUMO, YFP-SUMO1₉₅ (95), or a sumoylation-deficient mutant, K2P1-K274Q (Q).

Figure 5. One SUMO moiety is necessary and sufficient to silence heterodimeric tandem channels with K2P1 and K2P3 or K2P9

Channel subunits were heterologously-expressed in CHO cells and studied either as in Figure 2 or using TIRF microscopy to visualize and count GFP-SUMO1 monomers. A. Top left, WT-K2P3 is a tandem subunit formed by linking wild-type K2P1 (WT) and K2P3 (K2P1-Lys²⁷⁴ and the connection point between subunits, open circle, are indicated). Top right, normalized current for WT-K2P3 channels (mean ± S.E.M., n = 5 patches) at 50 mV measured in inside-out, off-cell patches as in Figure 2B. Basal currents (12 ± 4 pA) were activated by SENP1 (305 ± 11 pA) and suppressed by SUMO1. Reapplication of SENP1 restored channel activity despite extended washing (Wash). Bottom left, current families at times indicated by symbols. Bottom right, histograms showing normalized basal current and after application of SUMO1 and SENP1; Q-K2P3 channels are formed with K2P1-K274Q and K2P3 and do not respond to either SUMO1 or SENP1. B. Left, WT-K2P9 channels in off-cell patches excised from CHO cells are silent at baseline (10 ± 3 pA) and activated by SENP1 (355 ± 28 pA) when studied as in (A). Right, histograms showing normalized current, basal and on application of SUMO1 and SENP1; Q-K2P9 channels are formed with K2P1-K274Q and K2P9 and do not respond to SUMO1 or SENP1. C. GFP-tagged SUMO1 was studied at the CHO cell surface by TIRF. Left, a representative single, fluorescent GFP-K2P1 particle showing two photobleaching steps. Right, histogram showing that 90% of 90 GFP-K2P1 particles demonstrated two bleaching steps. Particles in cells expressing WT-K2P3 channels with GFP-SUMO1 or WT-K2P9 channels and GFP-SUMO1 showed only a single bleaching step (n = 100 particles).

Figure 6. K2P1-Y231F suppression and halothane augmentation of K2P channel activity

Subunits heterologously-expressed in CHO cells were studied in whole-cell mode. Current-voltage relationships and histograms are mean current ± S.E.M. for 8–12 cells per group. A. Exemplar current families are shown for SENP1-activated K2P1, K2P3, K2P9 and K2P2 expressed without or with (+) K2P1-Y231F. K2P1-Y231F suppressed current passed by K2P1, K2P3 and K2P9 but not K2P2, consistent with assembly of K2P1-Y231F with all but the last subunit type. Current inhibition at 50 mV was 89%, 88% and 76% for SENP1-activated K2P1, K2P3 and K2P9, respectively. Current-voltage relationships are shown; K2P1 basal current is indicated by the open squares. B. Histograms of current at 50 mV in CHO cells expressing the indicated channels before (solid bars) and on application of halothane (hashed bars). K2P3 and K2P9 were studied with control solution in the pipette. Halothane was applied to K2P1, WT-K2P3 and WT-K2P9 after activation by SENP1 in the pipette; as in patches (Figure 5), these channels pass negligible currents without activation by SENP (control). Halothane augmented channel current in the following order: WT–K2P9 ≫ K2P9 > K2P3 > WT–K2P3 or K2P1.

Figure 7. Channels that can incorporate K2P1 mediate the acid-sensitive portion of IKso and are those responsive to halothane

FRET analysis was performed on live CGN that heterologously expressed subunits tagged with CFP or YFP as in Figure 4. Fluorescence decay parameters (mean τ ± S.E.M.) are listed in Table S3. IKso was studied in whole-cell mode, as in Figure 2, with only solution B in the pipette (control), added SUMO1, or added SENP1 (n = 10 cells). Significant changes in IKso magnitude, compared to naive CGN studied with control solution in the pipette, are indicated (*, P<0.001). A. FRET between CFP-K2P3 (or CFP-K2P9) and YFP-SUMO1 requires K2P1. FRET is observed in homodimeric channels between subunits tagged with CFP and YFP (dotted lines); CFP-K2P3 or CFP-K2P9 show FRET with YFP-SUMO1 when wild type K2P1 (WT) is also expressed (NS, indicates no significant change compared to homodimeric channel FRET signal). Controls: FRET was not observed with heterologous-expression of subunits incapable of sumoylation, YFP-SUMO1₉₅ (95) and K2P1-K274Q (Q). B. Exemplar IKso traces and mean current-density histograms. Left, as shown in Figure 2A (Table S1), SENP1 augments IKso. Middle, heterologous expression of K2P1-Y231F subunits ablates augmentation by SENP1 and diminishes IKso by 60% (to 100 ± 9 pA/pF). SUMO1 does not alter IKso (245 ± 22 pA/pF). In contrast, SENP1 augments IKso by ~200%. C. Exemplar IKso traces and mean current-density histograms. Application of halothane (gray traces) augments IKso ~200% but has no effect on IKso in cells expressing K2P1-Y231F (105 ± 14 pA/pF). Adding SUMO1 into the pipette does not alter the effect of halothane (510 ± 44 pA/pF). With SENP1 in the pipette, halothane augments IKso ~345% to 860 pA/pF.