Voltage-gated K⁺ (K_V) channels facilitate the movement of K⁺ ions across the lipid bilayer in response to membrane depolarization and are essential for signaling in electrically excitable tissues (Jan and Jan, 2012). Among K_V channels, KCNQ1 (K_V7.1, K_VLQT1) is special because of its wide range of physiological behaviors (Abbott, 2014). This versatility enables KCNQ1 to function distinctly in both excitable cells such as cardiomyocytes and in non-excitable cells such as epithelia (Abbott, 2014). The functional diversity of KCNQ1 is a consequence of its ability to form channel complexes with any one of five tissue-specific KCNE auxiliary proteins (KCNE1-5) (Barhanin et al., 1996; Sanguinetti et al., 1996; Schroeder et al., 2000; Tinel et al., 2000; Teng et al., 2003; Angelo et al., 2002).

KCNQ1 channels consist of four identical pore-forming subunits, each containing six membrane-spanning segments (S1-S6) and a pore loop (P loop) (Figure 1A). The centrally located pore domain (PD, S5-P-S6) forms the ion permeation pathway and is surrounded by four voltage-sensing domains (VSDs, S1-S4). The VSD S4 helix carries positively charged residues that trigger S4 movement upon membrane depolarization (Bezanilla, 2000), leading to three detectable VSD conformational states in KCNQ1 (resting, intermediate, and activated) (Panaghie and Abbott, 2007; Wu et al., 2010; Zaydman et al., 2014; Taylor et al., 2020). S4 movement is thought to exert a lateral pull on the S4-S5 linker (S4-S5L), which triggers opening of the helical S6 gate making the channel conductive (Long et al., 2005). KCNEs serve as β-subunits of KCNQ1 and contain a single transmembrane-spanning domain (TMD) in addition to sizeable extra- and intracellular domains (McCrossan and Abbott, 2004; Figure 1A).

Figure 1

Download asset Open asset

Co-assembly of KCNQ1 with KCNE1 generates a channel complex that exhibits slow activation that occurs at more positive potentials and with higher conductance relative to KCNQ1 alone (Barhanin et al., 1996; Sanguinetti et al., 1996). KCNQ1-KCNE1 channels generate the slow delayed rectifier K⁺ current (I_Ks) in the heart that contributes to the repolarization phase of the cardiac action potential. Heritable mutations in KCNQ1 and KCNE1 predispose individuals to life-threatening ventricular arrhythmia and cause type 1 and type 5 long QT syndrome (LQTS) (Bohnen et al., 2017), respectively. By contrast, pairing of KCNQ1 with another KCNE subunit, KCNE3, produces channels that are constitutively active over the full physiological voltage range (Schroeder et al., 2000).

Different mechanisms have been proposed to explain how KCNE1 modulates KCNQ1 function including alteration of S4 movement (Nakajo and Kubo, 2007; Rocheleau and Kobertz, 2008; Osteen et al., 2010; Ruscic et al., 2013; Barro-Soria et al., 2014), perturbation of gate opening (Tapper and George, 2001; Melman et al., 2004; Panaghie et al., 2006), changes in VSD-PD coupling (Zaydman et al., 2014; Westhoff et al., 2019), or a combination of these effects (Nakajo and Kubo, 2014; Barro-Soria et al., 2017). However, a clear structural explanation is lacking owing to the absence of a high-resolution structure for the KCNQ1-KCNE1 complex. Previously, low-resolution spatial restraints for the KCNQ1-KCNE1 interaction were derived from disulfide crosslinking (Chung et al., 2009; Wang et al., 2011; Chan et al., 2012; Wang et al., 2012), metal ion bridging (Tapper and George, 2001), and site-directed mutagenesis data (Strutz-Seebohm et al., 2011; Li et al., 2015). In conjunction with computational modeling (Strutz-Seebohm et al., 2011; Kang et al., 2008; Gofman et al., 2012; Xu et al., 2013), these restraints have provided initial insight into the KCNQ1-KCNE1 channel architecture. Those models suggested that KCNE1 binds in a cleft surrounded by two VSDs and the PD, and is therefore in a location where it can simultaneously modulate S4 and the channel gate (Kang et al., 2008; Gofman et al., 2012; Xu et al., 2013). Additional studies have determined sites in KCNE1 that are crucial for its functional modulation of KCNQ1. Specifically, a three amino acid motif (F57-T58-L59, FTL) in the middle of the KCNE1 TMD (Figure 1B) was found to be necessary for induction of slow activation of KCNQ1 (Melman et al., 2001; Melman et al., 2002). Replacement of the corresponding segment in KCNE3 (T71-V72-G73, TVG) with FTL confers KCNE1-like gating properties onto the KCNQ1-KCNE3 channel (Barro-Soria et al., 2017; Melman et al., 2001; Melman et al., 2002). Likewise, mutation of FTL to TVG renders the KCNQ1-KCNE1 channel similar to KCNQ1-KCNE3, in that faster activation at more negative potentials is observed (Barro-Soria et al., 2017; Melman et al., 2001). How this so-called ‘activation motif’ determines the distinct gating properties of KCNQ1-KCNE channels is unclear. Mutations of residues in S6 (S338, F339, F340) alter the effect of mutations in the activation motif and vice versa, which was interpreted as a consequence of direct physical interaction between KCNE1 and S6 (Panaghie et al., 2006). However, this idea has been challenged recently, because these S6 residues reside deep within the PD in the KCNQ1 structure (Sun and MacKinnon, 2017; Sun and MacKinnon, 2020) and cysteine exchange experiments failed to confirm disulfide bond formation with any of the residues in the activation motif (Xu et al., 2013).

Here, we combined computational protein-protein docking, molecular dynamics, and electrophysiology to develop refined molecular models for the KCNQ1-KCNE1 complex to address the question of how the KCNE1 TMD modulates activation gating. Our results suggest that the KCNE1 FTL motif interacts with sites in S1, S4, and S5 in KCNQ1, and affects the channel gate by an allosteric network involving the S5-S6 interface. Comparison of independently constructed KCNQ1-KCNE1 models with the recently determined structures of the KCNQ1-KCNE3 complex (Sun and MacKinnon, 2020) shows a conserved TMD-binding mode for KCNE1 and KCNE3, but reveals specific differences in the interaction of the activation motif with the channel, consistent with the different effects of these KCNE proteins on channel gating (Barro-Soria et al., 2017). Our results provide more precise information on the state-specific structural requirements and specificity of KCNE subunit interactions with KCNQ1.

How KCNE subunits modify KCNQ1 function in such profoundly different ways is a longstanding topic of investigation. Our KCNQ1-KCNE1 models and subsequent comparative analysis with structures of the KCNQ1-KCNE3 channel aim to address two questions: How does the KCNE1 FTL motif interact with KCNQ1 to control KCNQ1 activation gating? And, what can be concluded about the TMD-binding mode for both KCNE1 and KCNE3 and the mechanism underlying the different impact on channel activation by KCNE3?

We found that KCNE1 FTL binds in a cleft between the KCNQ1 S1 (F127, F130), S4 (V241), and S5 (I263, L266, Y267 and F270) in both the RC and AO conformation (Figure 8D, Figure 9A and B). This binding mode places the activation motif in proximity to the conserved PAG (P343-A344-G345) motif in S6 – a segment that undergoes critical conformational changes during channel gating. Gating occurs as a consequence of S6 bending at the PAG motif, which causes S6 to swing away from the channel axis, opening the cytosolic gate. While the FTL residues are not in direct contact with S6, our structural model suggests that they can affect the nearby PAG motif through the mediation of S5 (Figure 9D, Video 3). Alanine mutational scanning of S5 previously showed that the V_1/2 shift of KCNQ1 activation by KCNE1 is reduced by mutations Y267A and F270A, and to a smaller extent by L271A, I274A, and F275A (Strutz-Seebohm et al., 2011). Mutation F270A also altered the volume-dependency of V_1/2 changes resulting from substitution of KCNE1 T58 with amino acids of different size (Strutz-Seebohm et al., 2011). Furthermore, KCNE1 FTL was responsible for the suppression of constitutive currents in KCNQ1 channels with mutation I268A, even in cases of functionally decoupled voltage sensor and pore domains (Barro-Soria et al., 2017). These data are consistent with binding of FTL to this region on S5 leading to triggering of changes in the activation gate of KCNQ1. Indeed, dynamical network analysis of our MD simulations shows that S5 connects KCNE1 T58 with KCNQ1 S6 through several short pathways in a residue interaction network (Figure 9—figure supplement 1A and B). These allosteric interactions could alter the conformational dynamics of S6 around the PAG motif and influence gate opening.

Figure 8 with 1 supplement see all

Download asset Open asset

Figure 8—source data 1

Excel file with MD H-bond time series data used for Figure 8 panel F.

https://cdn.elifesciences.org/articles/57680/elife-57680-fig8-data1-v2.xlsx

Download elife-57680-fig8-data1-v2.xlsx

Figure 9 with 2 supplements see all

Download asset Open asset

There are several lines of experimental evidence that support this model. KCNQ1 residues F339 and F340, which are one helix turn before the PAG motif, were shown to be energetically coupled to KCNE1 T58 based on double mutant cycle analysis (Strutz-Seebohm et al., 2011; Li et al., 2015). This is in accord with our dynamical network analysis (Figure 9—figure supplement 1A and B) and with the observation that mutations at either F339, F340, or T58 produce similar functional outcomes (Melman et al., 2004). Furthermore, the phenotype of some mutations in S6 can be altered or rescued by KCNE1: non-functional KCNQ1 A341V mutant channels can be rendered functional when co-expressed with KCNE1 (Mikuni et al., 2011), and constitutively active F340W channels can be suppressed by mutation at KCNE1 T58 (Panaghie et al., 2006), which points to an effect of KCNE1 on the channel gate. Previously, these data were taken as evidence for a direct physical contact between T58 and S6, leading to a model in which KCNE1 lies close to or forms a part of the KCNQ1 PD (Tapper and George, 2001; Melman et al., 2004; Strutz-Seebohm et al., 2011; Li et al., 2015). However, our KCNQ1-KCNE1 models, in accord with structures of the related KCNQ1-KCNE3 complex (Sun and MacKinnon, 2020), indicate that KCNE1 is too far away to directly contact the middle region of S6. Instead, we propose an allosteric mechanism for control of KCNQ1 gating, in which binding of the FTL motif influences the conformational state of the S6 PAG motif by interacting with S5, which then directly impacts the key gating residues in S6 (Figure 9D, Video 3).

Comparison of the binding modes of KCNE1 and KCNE3 also suggests how the activation motifs of these proteins could trigger different effects on the KCNQ1 gate through mediation of S5. We observed different interactions are made by the KCNE1 FTL versus the KCNE3 TVG (Figure 9B+C, Video 4). KCNE1 F57 and L59 are much bulkier than the corresponding KCNE3 residues, T71 and G73, possibly leading to different steric effects on neighboring residues in KCNQ1 S1 and S5. Furthermore, the model predicts an H-bond between KCNE1 T58 and KCNQ1 Y267, which is absent for KCNE3 V72. Replacing V72 with a hydroxylated amino acid could restore this H-bond interaction, which offers an explanation why the KCNE3 V72T mutant leads to channel properties akin to KCNQ1-KCNE1 (Melman et al., 2002). Scoring with the Rosetta energy function suggests a larger binding free energy for FTL compared to TVG (Figure 9D, Figure 9—figure supplement 1C) and a decrease of binding by Ala mutations of residues in S5 (Figure 9—figure supplement 1D). Together these observations support the notion that FTL and TVG induce distinct interactions of varying strength with S5 that determine the effect of the KCNE subunits on the S6 gate.

Video 4

Download asset

The outlined mechanistic model agrees qualitatively with observations from Barro-Soria et al., 2017, who showed that KCNE1 and KCNE3 have different effects on the VSD and PD. While KCNE1 acts both on the VSD and PD, shifting S4 movement to more negative potentials and gate opening to more positive potentials, KCNE3 mainly affects the VSD (Barro-Soria et al., 2017), effectively eliminating gating by stabilizing the intermediate and fully activated VSD states (Sun and MacKinnon, 2020; Kroncke et al., 2016). The FTL and TVG motifs were found to determine whether the KCNE subunits affect the gate and the second S4 movement that is seen in voltage clamp fluorometry studies of KCNQ1-KCNE1 and correlates with the opening of KCNQ1-KCNE1 channels (Barro-Soria et al., 2017). This can be explained by the differences in the interaction modes between FTL and TVG seen in our models. Furthermore, significantly less contacts are observed for FTL or TVG with S4 (only KCNQ1 V241 is within van der Waals distance of either T58 or V72) (Figure 8D), which is consistent with the notion that these sites have minimal impact on S4 movement. This is in contrast to other residues in KCNE1 and KCNE3, one to two helix turns C-terminal to FTL and TVG, which make direct contacts with S4, explaining how both KCNE proteins affect S4 movement (Panaghie and Abbott, 2007; Nakajo and Kubo, 2007; Rocheleau and Kobertz, 2008; Osteen et al., 2010; Ruscic et al., 2013; Barro-Soria et al., 2014; Barro-Soria et al., 2015).

Despite different interactions of KCNQ1 with the activation motifs from KCNE1 and KCNE3, we observed impressive similarity in their global TMD-binding modes. This observation led us to wonder if this feature is conserved for other KCNEs. Some support for this hypothesis comes from the overall high degree of sequence homology of the TMD within the KCNE family (11–60% identity, 45–78% similarity) (Figure 9—figure supplement 2). It is noteworthy that Y46 in KCNE1, which we identified as a high impact mutation site, is conserved in all KCNE proteins. Perhaps, this site is part of a channel docking/anchoring mechanism shared by all KCNE family members. A region with considerably less sequence homology exists in the middle of the TMD and coincides with the FTL motif from KCNE1. We speculate that different KCNEs might have evolved different amino acid sequences for this motif in order to elicit different functional responses in KCNQ1 and other ion channels. This sequence motif may therefore represent a common regulatory element in KCNEs.

There is relatively little structure-function data available for other KCNEs to inform on the degree of conservation of their TMD-binding modes. Using cysteine scanning mutagenesis and analyzing the pattern of functional perturbations in KCNE2, Wang et al., 2012 developed a schematic model for the orientation of the KCNE2 TMD within the KCNQ1 channel complex. KCNQ1 C331 was observed to form disulfide bonds to different positions in KCNE1 (F54) and KCNE2 (M59, corresponding to F53 in KCNE1), which trapped the channel in open and closed states, respectively. While this result could imply a rotation of the KCNE subunit during channel open-to-close transition, it could also indicate a different TMD orientation. In contrast to this hypothesis, our KCNQ1-KCNE1 models and the KCNQ1-KCNE3 structures fail to reveal a significant rotation for the KCNE subunit between open and closed channel conformations. Furthermore, in these structures, F53 and F54 (L67 and F68 in KCNE3) have approximately the same distance to C331 (KCNE1: F53 – 10.1 Å, F54 – 13.2 Å; KCNE3: L67 – 9.4 Å, F68 – 12.5 Å). We suggest that the proposed KCNE1 binding pose of this work could potentially satisfy both disulfide bond restraints, and that experimental KCNE-specific differences in disulfide bond formation could be the result of local structural differences in and around the activation triplet between KCNE1 and KCNE2. Another study by Li et al., 2015 identified functional interactions between the KCNQ1 PD (F275, F340) and the middle residue in the KCNE2 activation triplet (I64). This is consistent with our proposed allosteric mechanism between those corresponding sites in KCNE1 and suggests a similar mechanism may exist in KCNE2.

In summary, our results imply that all KCNE subunits interact with KCNQ1 by adopting a common binding mode within their TMD. Moreover, our experimental/computational models of the KCNQ1-KCNE1 channel complex provide an explanation for how the KCNE1 TMD interacts with the KCNQ1 channel and controls KCNQ1 activation through its intramembrane FTL motif. Binding of the FTL motif appears to affect the channel gate through allosteric interactions with S6 that are mediated by S5. Differences in their interactions with S5 appear to explain the different effects of KCNE1 and KCNE3 on the gate in KCNQ1.

The structural models developed in this work have been deposited in PDB-Dev under accession code PDBDEV_00000042 and are included as supplementary files 2 and 3 to this manuscript. The experimental restraints and starting model coordinates used for docking, as well as representative snapshots from the MD simulations have been deposited under: https://doi.org/10.5281/zenodo.3598943. All electrophysiology data generated and analyzed during this study are included in the manuscript or the supplementary source data files. All data needed to evaluate the conclusions in the paper are present in the paper and/or in the figure supplements and supplementary material.

1. 1. Kuenze G
   2. Vanoye CG
   3. Desai RR
   4. Adusumilli S
   5. Brewer KR
   6. Woods H
   7. McDonald EF
   8. Sanders CR
   9. George AL
   10. Meiler J

   (2020) Zenodo

   Allosteric Mechanism for KCNE1 Modulation of KCNQ1 Potassium Channel Activation.

   https://doi.org/10.5281/zenodo.3598943

1. 1. Abbott GW

   (2014) Biology of the KCNQ1 potassium channel

   New Journal of Science 2014:1–26.

   https://doi.org/10.1155/2014/237431

   • Google Scholar
2. 1. Angelo K
   2. Jespersen T
   3. Grunnet M
   4. Nielsen MS
   5. Klaerke DA
   6. Olesen SP

   (2002) KCNE5 induces time- and voltage-dependent modulation of the KCNQ1 current

   Biophysical Journal 83:1997–2006.

   https://doi.org/10.1016/S0006-3495(02)73961-1

   • PubMed
   • Google Scholar
3. 1. Arnestad M
   2. Crotti L
   3. Rognum TO
   4. Insolia R
   5. Pedrazzini M
   6. Ferrandi C
   7. Vege A
   8. Wang DW
   9. Rhodes TE
   10. George AL
   11. Schwartz PJ

   (2007) Prevalence of long-QT syndrome gene variants in sudden infant death syndrome

   Circulation 115:361–367.

   https://doi.org/10.1161/CIRCULATIONAHA.106.658021

   • PubMed
   • Google Scholar
4. 1. Barhanin J
   2. Lesage F
   3. Guillemare E
   4. Fink M
   5. Lazdunski M
   6. Romey G

   (1996) K(V)LQT1 and lsK (minK) proteins associate to form the I(Ks) cardiac potassium current

   Nature 384:78–80.

   https://doi.org/10.1038/384078a0

   • PubMed
   • Google Scholar
5. 1. Barlow KA
   2. Ó Conchúir S
   3. Thompson S
   4. Suresh P
   5. Lucas JE
   6. Heinonen M
   7. Kortemme T

   (2018) Flex ddG: Rosetta ensemble-based estimation of changes in protein-protein binding affinity upon mutation

   The Journal of Physical Chemistry B 122:5389–5399.

   https://doi.org/10.1021/acs.jpcb.7b11367

   • PubMed
   • Google Scholar
6. 1. Barro-Soria R
   2. Rebolledo S
   3. Liin SI
   4. Perez ME
   5. Sampson KJ
   6. Kass RS
   7. Larsson HP

   (2014) KCNE1 divides the voltage sensor movement in KCNQ1/KCNE1 channels into two steps

   Nature Communications 5:3750.

   https://doi.org/10.1038/ncomms4750

   • PubMed
   • Google Scholar
7. 1. Barro-Soria R
   2. Perez ME
   3. Larsson HP

   (2015) KCNE3 acts by promoting voltage sensor activation in KCNQ1

   PNAS 112:E7286–E7292.

   https://doi.org/10.1073/pnas.1516238112

   • PubMed
   • Google Scholar
8. 1. Barro-Soria R
   2. Ramentol R
   3. Liin SI
   4. Perez ME
   5. Kass RS
   6. Larsson HP

   (2017) KCNE1 and KCNE3 modulate KCNQ1 channels by affecting different gating transitions

   PNAS 114:E7367–E7376.

   https://doi.org/10.1073/pnas.1710335114

   • PubMed
   • Google Scholar
9. 1. Barth P
   2. Schonbrun J
   3. Baker D

   (2007) Toward high-resolution prediction and design of transmembrane helical protein structures

   PNAS 104:15682–15687.

   https://doi.org/10.1073/pnas.0702515104

   • PubMed
   • Google Scholar
10. 1. Bellocq C
    2. van Ginneken AC
    3. Bezzina CR
    4. Alders M
    5. Escande D
    6. Mannens MM
    7. Baró I
    8. Wilde AA

    (2004) Mutation in the KCNQ1 gene leading to the short QT-Interval syndrome

    Circulation 109:2394–2397.

    https://doi.org/10.1161/01.CIR.0000130409.72142.FE

    • PubMed
    • Google Scholar
11. 1. Bezanilla F

    (2000) The voltage sensor in voltage-dependent ion channels

    Physiological Reviews 80:555–592.

    https://doi.org/10.1152/physrev.2000.80.2.555

    • PubMed
    • Google Scholar
12. 1. Bianchi L
    2. Priori SG
    3. Napolitano C
    4. Surewicz KA
    5. Dennis AT
    6. Memmi M
    7. Schwartz PJ
    8. Brown AM

    (2000) Mechanisms of I(Ks) suppression in LQT1 mutants

    American Journal of Physiology. Heart and Circulatory Physiology 279:H3003–H3011.

    https://doi.org/10.1152/ajpheart.2000.279.6.H3003

    • PubMed
    • Google Scholar
13. 1. Bohnen MS
    2. Peng G
    3. Robey SH
    4. Terrenoire C
    5. Iyer V
    6. Sampson KJ
    7. Kass RS

    (2017) Molecular pathophysiology of congenital long QT syndrome

    Physiological Reviews 97:89–134.

    https://doi.org/10.1152/physrev.00008.2016

    • PubMed
    • Google Scholar
14. Report

    1. Case DA
    2. Betz RM
    3. Cerutti DS
    4. Cheatham Iii TE
    5. Darden TA
    6. Duke RE
    7. Giese TJ
    8. Gohlke H
    9. Goetz AW
    10. Homeyer N
    11. Izadi S
    12. Janowski P
    13. Klaus J
    14. Kovalenko A
    15. Lee TS
    16. LeGrand S
    17. Li P
    18. Lin C
    19. Luchko T
    20. Luo R
    21. Madej BD
    22. Mermelstein D
    23. Merz KM
    24. Monard G
    25. Nguyen H
    26. Nguyen HT
    27. Omelyan I
    28. Onufriev A
    29. Roe DR
    30. Roitberg A
    31. Sagui C
    32. Simmerling CL
    33. Botello-Smith WM
    34. Swails J
    35. Walker RC
    36. Wang J
    37. Wolf RM
    38. Wu X
    39. Xiao L
    40. Kollman P A

    (2016) Amber 16

    University of California.

    https://doi.org/10.13140/RG.2.2.27958.70729

    • Google Scholar
15. 1. Chan PJ
    2. Osteen JD
    3. Xiong D
    4. Bohnen MS
    5. Doshi D
    6. Sampson KJ
    7. Marx SO
    8. Karlin A
    9. Kass RS

    (2012) Characterization of KCNQ1 atrial fibrillation mutations reveals distinct dependence on KCNE1

    The Journal of General Physiology 139:135–144.

    https://doi.org/10.1085/jgp.201110672

    • PubMed
    • Google Scholar
16. 1. Chen YH
    2. Xu SJ
    3. Bendahhou S
    4. Wang XL
    5. Wang Y
    6. Xu WY
    7. Jin HW
    8. Sun H
    9. Su XY
    10. Zhuang QN
    11. Yang YQ
    12. Li YB
    13. Liu Y
    14. Xu HJ
    15. Li XF
    16. Ma N
    17. Mou CP
    18. Chen Z
    19. Barhanin J
    20. Huang W

    (2003a) KCNQ1 gain-of-function mutation in familial atrial fibrillation

    Science 299:251–254.

    https://doi.org/10.1126/science.1077771

    • PubMed
    • Google Scholar
17. 1. Chen H
    2. Kim LA
    3. Rajan S
    4. Xu S
    5. Goldstein SA

    (2003b) Charybdotoxin binding in the I(Ks) pore demonstrates two MinK subunits in each channel complex

    Neuron 40:15–23.

    https://doi.org/10.1016/S0896-6273(03)00570-1

    • PubMed
    • Google Scholar
18. 1. Chen J
    2. Liu Z
    3. Creagh J
    4. Zheng R
    5. McDonald TV

    (2020) Physical and functional interaction sites in cytoplasmic domains of KCNQ1 and KCNE1 channel subunits

    American Journal of Physiology-Heart and Circulatory Physiology 318:H212–H222.

    https://doi.org/10.1152/ajpheart.00459.2019

    • PubMed
    • Google Scholar
19. 1. Chen H
    2. Goldstein SA

    (2007) Serial perturbation of MinK in IKs implies an alpha-helical transmembrane span traversing the channel corpus

    Biophysical Journal 93:2332–2340.

    https://doi.org/10.1529/biophysj.107.109702

    • PubMed
    • Google Scholar
20. 1. Chung DY
    2. Chan PJ
    3. Bankston JR
    4. Yang L
    5. Liu G
    6. Marx SO
    7. Karlin A
    8. Kass RS

    (2009) Location of KCNE1 relative to KCNQ1 in the I(KS) potassium channel by disulfide cross-linking of substituted cysteines

    PNAS 106:743–748.

    https://doi.org/10.1073/pnas.0811897106

    • PubMed
    • Google Scholar
21. 1. Conway P
    2. Tyka MD
    3. DiMaio F
    4. Konerding DE
    5. Baker D

    (2014) Relaxation of backbone bond geometry improves protein energy landscape modeling

    Protein Science 23:47–55.

    https://doi.org/10.1002/pro.2389

    • PubMed
    • Google Scholar
22. 1. Darden T
    2. York D
    3. Pedersen L

    (1993) Particle mesh Ewald - an n.log(N) Method for Ewald sums in large systems

    Journal of Chemical Physics 98:10089–10092.

    https://doi.org/10.1063/1.464397

    • Google Scholar
23. 1. Davis IW
    2. Leaver-Fay A
    3. Chen VB
    4. Block JN
    5. Kapral GJ
    6. Wang X
    7. Murray LW
    8. Arendall WB
    9. Snoeyink J
    10. Richardson JS
    11. Richardson DC

    (2007) MolProbity: all-atom contacts and structure validation for proteins and nucleic acids

    Nucleic Acids Research 35:W375–W383.

    https://doi.org/10.1093/nar/gkm216

    • PubMed
    • Google Scholar
24. 1. DiMaio F
    2. Leaver-Fay A
    3. Bradley P
    4. Baker D
    5. André I

    (2011) Modeling symmetric macromolecular structures in Rosetta3

    PLOS ONE 6:e20450.

    https://doi.org/10.1371/journal.pone.0020450

    • PubMed
    • Google Scholar
25. 1. Eargle J
    2. Luthey-Schulten Z

    (2012) NetworkView: 3D display and analysis of protein-RNA interaction networks

    Bioinformatics 28:3000–3001.

    https://doi.org/10.1093/bioinformatics/bts546

    • PubMed
    • Google Scholar
26. 1. Girvan M
    2. Newman ME

    (2002) Community structure in social and biological networks

    PNAS 99:7821–7826.

    https://doi.org/10.1073/pnas.122653799

    • PubMed
    • Google Scholar
27. 1. Glykos NM

    (2006) Software news and updates. Carma: a molecular dynamics analysis program

    Journal of Computational Chemistry 27:1765–1768.

    https://doi.org/10.1002/jcc.20482

    • PubMed
    • Google Scholar
28. 1. Gofman Y
    2. Shats S
    3. Attali B
    4. Haliloglu T
    5. Ben-Tal N

    (2012) How does KCNE1 regulate the Kv7.1 potassium channel? Model-structure, mutations, and dynamics of the Kv7.1-KCNE1 complex

    Structure 20:1343–1352.

    https://doi.org/10.1016/j.str.2012.05.016

    • PubMed
    • Google Scholar
29. 1. Gray JJ
    2. Moughon S
    3. Wang C
    4. Schueler-Furman O
    5. Kuhlman B
    6. Rohl CA
    7. Baker D

    (2003) Protein-protein docking with simultaneous optimization of rigid-body displacement and side-chain conformations

    Journal of Molecular Biology 331:281–299.

    https://doi.org/10.1016/S0022-2836(03)00670-3

    • PubMed
    • Google Scholar
30. 1. Gray JJ

    (2006) High-resolution protein-protein docking

    Current Opinion in Structural Biology 16:183–193.

    https://doi.org/10.1016/j.sbi.2006.03.003

    • PubMed
    • Google Scholar
31. 1. Hamill OP
    2. Marty A
    3. Neher E
    4. Sakmann B
    5. Sigworth FJ

    (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches

    Pflügers Archiv - European Journal of Physiology 391:85–100.

    https://doi.org/10.1007/BF00656997

    • Google Scholar
32. 1. Homeyer N
    2. Horn AH
    3. Lanig H
    4. Sticht H

    (2006) AMBER force-field parameters for phosphorylated amino acids in different protonation states: phosphoserine, phosphothreonine, phosphotyrosine, and phosphohistidine

    Journal of Molecular Modeling 12:281–289.

    https://doi.org/10.1007/s00894-005-0028-4

    • PubMed
    • Google Scholar
33. 1. Hong K
    2. Piper DR
    3. Diaz-Valdecantos A
    4. Brugada J
    5. Oliva A
    6. Burashnikov E
    7. Santos-de-Soto J
    8. Grueso-Montero J
    9. Diaz-Enfante E
    10. Brugada P
    11. Sachse F
    12. Sanguinetti MC
    13. Brugada R

    (2005) De novo KCNQ1 mutation responsible for atrial fibrillation and short QT syndrome in utero

    Cardiovascular Research 68:433–440.

    https://doi.org/10.1016/j.cardiores.2005.06.023

    • PubMed
    • Google Scholar
34. Software

    1. Hubbard S
    2. Thornton J

    (1993)

    NACCESS, Computer Program

    Department of Biochemistry Molecular Biology, University College London.
35. 1. Humphrey W
    2. Dalke A
    3. Schulten K

    (1996) VMD: visual molecular dynamics

    Journal of Molecular Graphics 14:33–38.

    https://doi.org/10.1016/0263-7855(96)00018-5

    • PubMed
    • Google Scholar
36. 1. Jan LY
    2. Jan YN

    (2012) Voltage-gated potassium channels and the diversity of electrical signalling

    The Journal of Physiology 590:2591–2599.

    https://doi.org/10.1113/jphysiol.2011.224212

    • PubMed
    • Google Scholar
37. 1. Kang C
    2. Tian C
    3. Sönnichsen FD
    4. Smith JA
    5. Meiler J
    6. George AL
    7. Vanoye CG
    8. Kim HJ
    9. Sanders CR

    (2008) Structure of KCNE1 and implications for how it modulates the KCNQ1 potassium channel

    Biochemistry 47:7999–8006.

    https://doi.org/10.1021/bi800875q

    • PubMed
    • Google Scholar
38. 1. Karplus M
    2. Kushick JN

    (1981) Method for estimating the configurational entropy of macromolecules

    Macromolecules 14:325–332.

    https://doi.org/10.1021/ma50003a019

    • Google Scholar
39. 1. Kollman PA
    2. Massova I
    3. Reyes C
    4. Kuhn B
    5. Huo S
    6. Chong L
    7. Lee M
    8. Lee T
    9. Duan Y
    10. Wang W
    11. Donini O
    12. Cieplak P
    13. Srinivasan J
    14. Case DA
    15. Cheatham TE

    (2000) Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models

    Accounts of Chemical Research 33:889–897.

    https://doi.org/10.1021/ar000033j

    • PubMed
    • Google Scholar
40. 1. Kroncke BM
    2. Van Horn WD
    3. Smith J
    4. Kang C
    5. Welch RC
    6. Song Y
    7. Nannemann DP
    8. Taylor KC
    9. Sisco NJ
    10. George AL
    11. Meiler J
    12. Vanoye CG
    13. Sanders CR

    (2016) Structural basis for KCNE3 modulation of potassium recycling in epithelia

    Science Advances 2:e1501228.

    https://doi.org/10.1126/sciadv.1501228

    • PubMed
    • Google Scholar
41. 1. Kuenze G
    2. Duran AM
    3. Woods H
    4. Brewer KR
    5. McDonald EF
    6. Vanoye CG
    7. George AL
    8. Sanders CR
    9. Meiler J

    (2019) Upgraded molecular models of the human KCNQ1 potassium channel

    PLOS ONE 14:e0220415.

    https://doi.org/10.1371/journal.pone.0220415

    • PubMed
    • Google Scholar
42. 1. Leaver-Fay A
    2. Tyka M
    3. Lewis SM
    4. Lange OF
    5. Thompson J
    6. Jacak R
    7. Kaufman K
    8. Renfrew PD
    9. Smith CA
    10. Sheffler W
    11. Davis IW
    12. Cooper S
    13. Treuille A
    14. Mandell DJ
    15. Richter F
    16. Ban YE
    17. Fleishman SJ
    18. Corn JE
    19. Kim DE
    20. Lyskov S
    21. Berrondo M
    22. Mentzer S
    23. Popović Z
    24. Havranek JJ
    25. Karanicolas J
    26. Das R
    27. Meiler J
    28. Kortemme T
    29. Gray JJ
    30. Kuhlman B
    31. Baker D
    32. Bradley P

    (2011) ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules

    Methods in Enzymology 487:545–574.

    https://doi.org/10.1016/B978-0-12-381270-4.00019-6

    • PubMed
    • Google Scholar
43. 1. Li P
    2. Liu H
    3. Lai C
    4. Sun P
    5. Zeng W
    6. Wu F
    7. Zhang L
    8. Wang S
    9. Tian C
    10. Ding J

    (2015) Differential modulations of KCNQ1 by auxiliary proteins KCNE1 and KCNE2

    Scientific Reports 4:4973.

    https://doi.org/10.1038/srep04973

    • Google Scholar
44. 1. Lindau M
    2. Neher E

    (1988) Patch-clamp techniques for time-resolved capacitance measurements in single cells

    Pflügers Archiv European Journal of Physiology 411:137–146.

    https://doi.org/10.1007/BF00582306

    • Google Scholar
45. 1. Lomize MA
    2. Pogozheva ID
    3. Joo H
    4. Mosberg HI
    5. Lomize AL

    (2012) OPM database and PPM web server: resources for positioning of proteins in membranes

    Nucleic Acids Research 40:D370–D376.

    https://doi.org/10.1093/nar/gkr703

    • PubMed
    • Google Scholar
46. 1. Long SB
    2. Campbell EB
    3. Mackinnon R

    (2005) Voltage sensor of Kv1.2: structural basis of electromechanical coupling

    Science 309:903–908.

    https://doi.org/10.1126/science.1116270

    • PubMed
    • Google Scholar
47. 1. Long SB
    2. Tao X
    3. Campbell EB
    4. MacKinnon R

    (2007) Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment

    Nature 450:376–382.

    https://doi.org/10.1038/nature06265

    • PubMed
    • Google Scholar
48. 1. Lundquist AL
    2. Manderfield LJ
    3. Vanoye CG
    4. Rogers CS
    5. Donahue BS
    6. Chang PA
    7. Drinkwater DC
    8. Murray KT
    9. George AL

    (2005) Expression of multiple KCNE genes in human heart may enable variable modulation of I(Ks)

    Journal of Molecular and Cellular Cardiology 38:277–287.

    https://doi.org/10.1016/j.yjmcc.2004.11.012

    • PubMed
    • Google Scholar
49. 1. Lvov A
    2. Gage SD
    3. Berrios VM
    4. Kobertz WR

    (2010) Identification of a protein-protein interaction between KCNE1 and the activation gate machinery of KCNQ1

    The Journal of General Physiology 135:607–618.

    https://doi.org/10.1085/jgp.200910386

    • PubMed
    • Google Scholar
50. 1. Maier JA
    2. Martinez C
    3. Kasavajhala K
    4. Wickstrom L
    5. Hauser KE
    6. Simmerling C

    (2015) ff14SB: improving the accuracy of protein side chain and backbone parameters from ff99SB

    Journal of Chemical Theory and Computation 11:3696–3713.

    https://doi.org/10.1021/acs.jctc.5b00255

    • PubMed
    • Google Scholar
51. 1. Manderfield LJ
    2. George AL

    (2008) KCNE4 can co-associate with the I(Ks) (KCNQ1-KCNE1) channel complex

    FEBS Journal 275:1336–1349.

    https://doi.org/10.1111/j.1742-4658.2008.06294.x

    • PubMed
    • Google Scholar
52. 1. McCrossan ZA
    2. Abbott GW

    (2004) The MinK-related peptides

    Neuropharmacology 47:787–821.

    https://doi.org/10.1016/j.neuropharm.2004.06.018

    • PubMed
    • Google Scholar
53. 1. Melman YF
    2. Domènech A
    3. de la Luna S
    4. McDonald TV

    (2001) Structural determinants of KvLQT1 control by the KCNE family of proteins

    Journal of Biological Chemistry 276:6439–6444.

    https://doi.org/10.1074/jbc.M010713200

    • PubMed
    • Google Scholar
54. 1. Melman YF
    2. Krumerman A
    3. McDonald TV

    (2002) A single transmembrane site in the KCNE-encoded proteins controls the specificity of KvLQT1 channel gating

    The Journal of Biological Chemistry 277:25187–25194.

    https://doi.org/10.1074/jbc.M200564200

    • PubMed
    • Google Scholar
55. 1. Melman YF
    2. Um SY
    3. Krumerman A
    4. Kagan A
    5. McDonald TV

    (2004) KCNE1 binds to the KCNQ1 pore to regulate potassium channel activity

    Neuron 42:927–937.

    https://doi.org/10.1016/j.neuron.2004.06.001

    • PubMed
    • Google Scholar
56. 1. Mikuni I
    2. Torres CG
    3. Bienengraeber MW
    4. Kwok W-M

    (2011) Partial restoration of the long QT syndrome associated KCNQ1 A341V mutant by the KCNE1 β-subunit

    Biochimica Et Biophysica Acta (BBA) - General Subjects 1810:1285–1293.

    https://doi.org/10.1016/j.bbagen.2011.07.018

    • Google Scholar
57. 1. Miller BR
    2. McGee TD
    3. Swails JM
    4. Homeyer N
    5. Gohlke H
    6. Roitberg AE

    (2012) MMPBSA.py: an efficient program for End-State free energy calculations

    Journal of Chemical Theory and Computation 8:3314–3321.

    https://doi.org/10.1021/ct300418h

    • PubMed
    • Google Scholar
58. 1. Morin TJ
    2. Kobertz WR

    (2008) Counting membrane-embedded KCNE beta-subunits in functioning K+ channel complexes

    PNAS 105:1478–1482.

    https://doi.org/10.1073/pnas.0710366105

    • PubMed
    • Google Scholar
59. 1. Murray CI
    2. Westhoff M
    3. Eldstrom J
    4. Thompson E
    5. Emes R
    6. Fedida D

    (2016) Unnatural amino acid photo-crosslinking of the IKs channel complex demonstrates a KCNE1:KCNQ1 stoichiometry of up to 4:4

    eLife 5:e11815.

    https://doi.org/10.7554/eLife.11815

    • PubMed
    • Google Scholar
60. 1. Nakajo K
    2. Ulbrich MH
    3. Kubo Y
    4. Isacoff EY

    (2010) Stoichiometry of the KCNQ1 - KCNE1 ion channel complex

    PNAS 107:18862–18867.

    https://doi.org/10.1073/pnas.1010354107

    • PubMed
    • Google Scholar
61. 1. Nakajo K
    2. Nishino A
    3. Okamura Y
    4. Kubo Y

    (2011) KCNQ1 subdomains involved in KCNE modulation revealed by an invertebrate KCNQ1 orthologue

    The Journal of General Physiology 138:521–535.

    https://doi.org/10.1085/jgp.201110677

    • Google Scholar
62. 1. Nakajo K
    2. Kubo Y

    (2007) KCNE1 and KCNE3 stabilize and/or slow voltage sensing S4 segment of KCNQ1 channel

    Journal of General Physiology 130:269–281.

    https://doi.org/10.1085/jgp.200709805

    • PubMed
    • Google Scholar
63. 1. Nakajo K
    2. Kubo Y

    (2014) Steric hindrance between S4 and S5 of the KCNQ1/KCNE1 channel hampers pore opening

    Nature Communications 5:4100.

    https://doi.org/10.1038/ncomms5100

    • Google Scholar
64. 1. Osteen JD
    2. Gonzalez C
    3. Sampson KJ
    4. Iyer V
    5. Rebolledo S
    6. Larsson HP
    7. Kass RS

    (2010) KCNE1 alters the voltage sensor movements necessary to open the KCNQ1 channel gate

    PNAS 107:22710–22715.

    https://doi.org/10.1073/pnas.1016300108

    • PubMed
    • Google Scholar
65. 1. Panaghie G
    2. Tai KK
    3. Abbott GW

    (2006) Interaction of KCNE subunits with the KCNQ1 K⁺ channel pore

    The Journal of Physiology 570:455–467.

    https://doi.org/10.1113/jphysiol.2005.100644

    • PubMed
    • Google Scholar
66. 1. Panaghie G
    2. Abbott GW

    (2007) The role of S4 charges in voltage-dependent and voltage-independent KCNQ1 potassium channel complexes

    Journal of General Physiology 129:121–133.

    https://doi.org/10.1085/jgp.200609612

    • PubMed
    • Google Scholar
67. 1. Peng G
    2. Barro-Soria R
    3. Sampson KJ
    4. Larsson HP
    5. Kass RS

    (2017) Gating mechanisms underlying deactivation slowing by two KCNQ1 atrial fibrillation mutations

    Scientific Reports 7:45911.

    https://doi.org/10.1038/srep45911

    • PubMed
    • Google Scholar
68. 1. Plant LD
    2. Xiong D
    3. Dai H
    4. Goldstein SA

    (2014) Individual IKs channels at the surface of mammalian cells contain two KCNE1 accessory subunits

    PNAS 111:E1438–E1446.

    https://doi.org/10.1073/pnas.1323548111

    • PubMed
    • Google Scholar
69. 1. Rocheleau JM
    2. Kobertz WR

    (2008) KCNE peptides differently affect voltage sensor equilibrium and equilibration rates in KCNQ1 K+ channels

    Journal of General Physiology 131:59–68.

    https://doi.org/10.1085/jgp.200709816

    • PubMed
    • Google Scholar
70. 1. Roe DR
    2. Cheatham TE

    (2013) PTRAJ and CPPTRAJ: software for processing and analysis of molecular dynamics trajectory data

    Journal of Chemical Theory and Computation 9:3084–3095.

    https://doi.org/10.1021/ct400341p

    • PubMed
    • Google Scholar
71. 1. Ruscic KJ
    2. Miceli F
    3. Villalba-Galea CA
    4. Dai H
    5. Mishina Y
    6. Bezanilla F
    7. Goldstein SA

    (2013) IKs channels open slowly because KCNE1 accessory subunits slow the movement of S4 voltage sensors in KCNQ1 pore-forming subunits

    PNAS 110:E559–E566.

    https://doi.org/10.1073/pnas.1222616110

    • PubMed
    • Google Scholar
72. 1. Ryckaert J-P
    2. Ciccotti G
    3. Berendsen HJC

    (1977) Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes

    Journal of Computational Physics 23:327–341.

    https://doi.org/10.1016/0021-9991(77)90098-5

    • Google Scholar
73. 1. Sahu ID
    2. Kroncke BM
    3. Zhang R
    4. Dunagan MM
    5. Smith HJ
    6. Craig A
    7. McCarrick RM
    8. Sanders CR
    9. Lorigan GA

    (2014) Structural investigation of the transmembrane domain of KCNE1 in proteoliposomes

    Biochemistry 53:6392–6401.

    https://doi.org/10.1021/bi500943p

    • Google Scholar
74. 1. Sanguinetti MC
    2. Curran ME
    3. Zou A
    4. Shen J
    5. Spector PS
    6. Atkinson DL
    7. Keating MT

    (1996) Coassembly of K(V)LQT1 and minK (IsK) proteins to form cardiac I(Ks) potassium channel

    Nature 384:80–83.

    https://doi.org/10.1038/384080a0

    • PubMed
    • Google Scholar
75. 1. Schroeder BC
    2. Waldegger S
    3. Fehr S
    4. Bleich M
    5. Warth R
    6. Greger R
    7. Jentsch TJ

    (2000) A constitutively open potassium channel formed by KCNQ1 and KCNE3

    Nature 403:196–199.

    https://doi.org/10.1038/35003200

    • PubMed
    • Google Scholar
76. 1. Smart OS
    2. Neduvelil JG
    3. Wang X
    4. Wallace BA
    5. Sansom MSP

    (1996) HOLE: a program for the analysis of the pore dimensions of ion channel structural models

    Journal of Molecular Graphics 14:354–360.

    https://doi.org/10.1016/S0263-7855(97)00009-X

    • Google Scholar
77. 1. Strutz-Seebohm N
    2. Pusch M
    3. Wolf S
    4. Stoll R
    5. Tapken D
    6. Gerwert K
    7. Attali B
    8. Seebohm G

    (2011) Structural basis of slow activation gating in the cardiac IKs channel complex

    Cellular Physiology and Biochemistry : International Journal of Experimental Cellular Physiology, Biochemistry, and Pharmacology 27:443–452.

    https://doi.org/10.1159/000329965

    • PubMed
    • Google Scholar
78. 1. Sun J
    2. MacKinnon R

    (2017) Cryo-EM structure of a KCNQ1/CaM complex reveals insights into congenital long QT syndrome

    Cell 169:1042–1050.

    https://doi.org/10.1016/j.cell.2017.05.019

    • PubMed
    • Google Scholar
79. 1. Sun J
    2. MacKinnon R

    (2020) Structural basis of human KCNQ1 modulation and gating

    Cell 180:340–347.

    https://doi.org/10.1016/j.cell.2019.12.003

    • Google Scholar
80. 1. Tapper AR
    2. George AL

    (2001) Location and orientation of minK within the I(Ks) potassium channel complex

    The Journal of Biological Chemistry 276:38249–38254.

    https://doi.org/10.1074/jbc.M103956200

    • PubMed
    • Google Scholar
81. 1. Taylor KC
    2. Kang PW
    3. Hou P
    4. Yang N-D
    5. Kuenze G
    6. Smith JA
    7. Shi J
    8. Huang H
    9. White KM
    10. Peng D
    11. George AL
    12. Meiler J
    13. McFeeters RL
    14. Cui J
    15. Sanders CR

    (2020) Structure and physiological function of the human KCNQ1 channel voltage sensor intermediate state

    eLife 9:e53901.

    https://doi.org/10.7554/eLife.53901

    • Google Scholar
82. 1. Teng S
    2. Ma L
    3. Zhen Y
    4. Lin C
    5. Bähring R
    6. Vardanyan V
    7. Pongs O
    8. Hui R

    (2003) Novel gene hKCNE4 slows the activation of the KCNQ1 channel

    Biochemical and Biophysical Research Communications 303:808–813.

    https://doi.org/10.1016/S0006-291X(03)00433-9

    • PubMed
    • Google Scholar
83. 1. Tinel N
    2. Diochot S
    3. Borsotto M
    4. Lazdunski M
    5. Barhanin J

    (2000) KCNE2 confers background current characteristics to the cardiac KCNQ1 potassium channel

    The EMBO Journal 19:6326–6330.

    https://doi.org/10.1093/emboj/19.23.6326

    • PubMed
    • Google Scholar
84. 1. Vallat B
    2. Webb B
    3. Westbrook JD
    4. Sali A
    5. Berman HM

    (2018) Development of a prototype system for archiving integrative/hybrid structure models of biological macromolecules

    Structure 26:894–904.

    https://doi.org/10.1016/j.str.2018.03.011

    • PubMed
    • Google Scholar
85. 1. Vanoye CG
    2. Welch RC
    3. Daniels MA
    4. Manderfield LJ
    5. Tapper AR
    6. Sanders CR
    7. George AL

    (2009) Distinct subdomains of the KCNQ1 S6 segment determine channel modulation by different KCNE subunits

    The Journal of General Physiology 134:207–217.

    https://doi.org/10.1085/jgp.200910234

    • PubMed
    • Google Scholar
86. 1. Vanoye CG
    2. Desai RR
    3. Fabre KL
    4. Gallagher SL
    5. Potet F
    6. DeKeyser JM
    7. Macaya D
    8. Meiler J
    9. Sanders CR
    10. George AL

    (2018) High-Throughput Functional Evaluation of KCNQ1 Decrypts Variants of Unknown Significance

    Circulation. Genomic and Precision Medicine 11:e002345.

    https://doi.org/10.1161/CIRCGEN.118.002345

    • PubMed
    • Google Scholar
87. 1. Wang J
    2. Wang W
    3. Kollman PA
    4. Case DA

    (2006) Automatic atom type and bond type perception in molecular mechanical calculations

    Journal of Molecular Graphics and Modelling 25:247–260.

    https://doi.org/10.1016/j.jmgm.2005.12.005

    • PubMed
    • Google Scholar
88. 1. Wang YH
    2. Jiang M
    3. Xu XL
    4. Hsu KL
    5. Zhang M
    6. Tseng GN

    (2011) Gating-related molecular motions in the extracellular domain of the IKs channel: implications for IKs channelopathy

    The Journal of Membrane Biology 239:137–156.

    https://doi.org/10.1007/s00232-010-9333-7

    • PubMed
    • Google Scholar
89. 1. Wang Y
    2. Zhang M
    3. Xu Y
    4. Jiang M
    5. Zankov DP
    6. Cui M
    7. Tseng GN

    (2012) Probing the structural basis for differential KCNQ1 modulation by KCNE1 and KCNE2

    Journal of General Physiology 140:653–669.

    https://doi.org/10.1085/jgp.201210847

    • PubMed
    • Google Scholar
90. 1. Wang RY
    2. Song Y
    3. Barad BA
    4. Cheng Y
    5. Fraser JS
    6. DiMaio F

    (2016) Automated structure refinement of macromolecular assemblies from cryo-EM maps using Rosetta

    eLife 5:e17219.

    https://doi.org/10.7554/eLife.17219

    • PubMed
    • Google Scholar
91. 1. Wang Y
    2. Eldstrom J
    3. Fedida D

    (2020) The IKs Ion Channel Activator Mefenamic Acid Requires KCNE1 and Modulates Channel Gating in a Subunit-Dependent Manner

    Molecular Pharmacology 97:132–144.

    https://doi.org/10.1124/mol.119.117952

    • PubMed
    • Google Scholar
92. 1. Westhoff M
    2. Eldstrom J
    3. Murray CI
    4. Thompson E
    5. Fedida D

    (2019) IKs ion-channel pore conductance can result from individual voltage sensor movements

    PNAS 116:7879–7888.

    https://doi.org/10.1073/pnas.1811623116

    • PubMed
    • Google Scholar
93. 1. Wu D
    2. Delaloye K
    3. Zaydman MA
    4. Nekouzadeh A
    5. Rudy Y
    6. Cui J

    (2010) State-dependent electrostatic interactions of S4 arginines with E1 in S2 during Kv7.1 activation

    The Journal of General Physiology 135:595–606.

    https://doi.org/10.1085/jgp.201010408

    • PubMed
    • Google Scholar
94. 1. Wu EL
    2. Cheng X
    3. Jo S
    4. Rui H
    5. Song KC
    6. Dávila-Contreras EM
    7. Qi Y
    8. Lee J
    9. Monje-Galvan V
    10. Venable RM
    11. Klauda JB
    12. Im W

    (2014) CHARMM-GUI membrane builder toward realistic biological membrane simulations

    Journal of Computational Chemistry 35:1997–2004.

    https://doi.org/10.1002/jcc.23702

    • PubMed
    • Google Scholar
95. 1. Xiao L
    2. Diao J
    3. Greene D
    4. Wang J
    5. Luo R

    (2017) A continuum Poisson-Boltzmann model for membrane channel proteins

    Journal of Chemical Theory and Computation 13:3398–3412.

    https://doi.org/10.1021/acs.jctc.7b00382

    • PubMed
    • Google Scholar
96. 1. Xu Y
    2. Wang Y
    3. Meng XY
    4. Zhang M
    5. Jiang M
    6. Cui M
    7. Tseng GN

    (2013) Building KCNQ1/KCNE1 channel models and probing their interactions by molecular-dynamics simulations

    Biophysical Journal 105:2461–2473.

    https://doi.org/10.1016/j.bpj.2013.09.058

    • PubMed
    • Google Scholar
97. 1. Yarov-Yarovoy V
    2. Schonbrun J
    3. Baker D

    (2006) Multipass membrane protein structure prediction using Rosetta

    Proteins: Structure, Function, and Bioinformatics 62:1010–1025.

    https://doi.org/10.1002/prot.20817

    • Google Scholar
98. 1. Zaydman MA
    2. Kasimova MA
    3. McFarland K
    4. Beller Z
    5. Hou P
    6. Kinser HE
    7. Liang H
    8. Zhang G
    9. Shi J
    10. Tarek M
    11. Cui J

    (2014) Domain-domain interactions determine the gating, permeation, pharmacology, and subunit modulation of the IKs ion channel

    eLife 3:e03606.

    https://doi.org/10.7554/eLife.03606

    • PubMed
    • Google Scholar

Acceptance summary:

This paper presents an experimentally validated model of the IKs channel, KCNQ1+E1 in two states, Activated/Open and Resting /Closed, based in large part on an extensive review of experimentally-derived constraints. The models reported, deposited in PDB-dev will be of great value to the community. By comparing to the cryo-EM structure of KCNQ1+E3, the structural origin of profound functional differences with important physiological consequences is revealed.

Decision letter after peer review:

Thank you for submitting your article "Allosteric Mechanism for KCNE1 Modulation of KCNQ1 Potassium Channel Activation" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Lucie Delemotte as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Kenton Swartz as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Leon D Islas (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

KCNQ1 channels are a principal component of the slow-activating repolarization current in the heart. The slow activation character is produced by an interaction with the KCNE1 regulatory subunit. Although a well-characterized interaction, the molecular details of this are still controversial. The paper by Kuenze et al. investigates the molecular details of the association between the KCNQ1 channel and its regulatory subunit, KCNE1. Through molecular modeling and protein-protein docking, the authors arrive at a proposed mode of interaction between KCNE1 and models of the closed and open KCNQ1 channel. These detailed molecular interactions then are validated via a varied number of experiments, including mutagenesis, electrophysiological analysis of the effects of mutations and cysteine-cysteine crosslinking. Contrary to previous belief, the FTL motif in KCNE1 is not directly interacting with S6. Instead, the authors propose that FTL affects S6 via a allosteric network via S5. Comparing their KCNQ1/KCNE1 models with the KCNQ1/KCNE3 structure, they found that KCNE1 and KCNE3 share a very similar binding site. But different interactions of FTL in KCNE1 and TVD in KCNE3 with KCNQ1 are proposed to account for the different functional modulation of KCNQ1 by different KCNE subunits.

The reviewers agreed that the models deposited in PDB-dev would be of great value to the community, and that the review of experimental restraints used to build the models are in and of itself of great value. They also judged that the experimental validation of the proposed models was thorough and increased the confidence in the proposed modes of interaction between the two proteins. The data were thus convincing, but the conclusions about interactions were sometimes overstated and a more extensive discussion about the mechanism of KCNE1 was needed. Additionally there were some concerns about the way the cross-linking experiments were carried out. A detailed review is included below.

Experiments:

1) Related to Figure 2 and all other experiments using crosslinking: the authors show that under control, reducing or oxidizing conditions the magnitude of the currents are different. The way the data are presented and described led us to understand that each treatment, control, DTT or Cu-Phenantroline was applied to different groups of cells. If this is the case, this is a major experimental problem. Since there is variability in expression level and possibly changes in open probability and/or single channel conductance in each of the mutants that the authors are not accounting for, a simple comparison of the current density magnitude between different treatments is meaningless. These experiments should be carried out in the same cell. Typically disulfide bond formation (crosslinking) is induced by Cu-Phe after basal current recording and then reversed by DTT. There is a vast literature making use of this technique and this is the accepted way of performing this experiment. The authors should clarify if this is indeed what they did and show an example of such an experiment.

Conversely, if each treatment was applied to different groups of cells, the authors should show that the introduced cysteine residue produces no changes in open probability and single-channel conductance.

2) The recordings shown by the authors were obtained in a limited voltage range, up to 60 mV. The tail currents show that the activation is far from saturation at this voltage and in consequence the estimation of the V1/2 from these curves incurs in an error, which could be very large. The authors should comment on these limitations or alternatively estimate this parameter from currents obtained at a wider range of voltages.

Modeling:

1) To strengthen confidence in the modeling procedure, the following controls should be carried out: the docking of E1 should be done on the recent human structures of the activated/open state hKCNQ1 instead of only on the author's previous model. Conversely, docking of E3 to their AO model (and to the experimental structure mentioned above) should be performed to demonstrate that the procedure allows to recapitulate the experimental binding pose.

2) Figure 3—figure supplement 1 shows a comparison of the models obtained in this work with the cryo-EM structure. The authors show that the backbone RMSD is fairly low between models and structure, but it would be more informative to show the degree of deviation of the position of side chains, especially in the regions in the models that were used for docking the KCNE1.

3) Another aspect that is not even mentioned by the authors is the fact that the stoichiometry of KCNQ1-KCNE1 complexes appears to be variable. There is no experimental consensus on how many KCNE1 subunits are present per tetrameric KCNQ1 and the evidence seems to support the view that the number of KCNE1 subunits associated with a KCNQ1 tetramer depends on the expression level of the former. The authors seem to have chosen a 2:1 KCNE1:tetramer assembly for their simulations. Although a different stoichiometry might or might not alter the details of the interaction, it has been show that the variable stoichiometry can alter the biophysical effect (degree of voltage shift among others) of KCNE1, indeed suggesting that the mode of interaction might also not be unique. Did you try docking four KCNE1s per tetramer? Would this different stoichiometry alter the mode of interaction between KCNE1s and KCNQ1? At the very least, the authors should discuss this variable affects and their choice of stoichiometry.

Contextualization and Interpretation of data:

1) Abstract: "FTL… affects the channel gate via an allosteric network." This is not shown, but only a proposed mechanism. Please state as suggestion, not fact.

2) “ Likewise, mutation of FTL to TVG renders the KCNQ1-KCNE1 channel constitutively active, similar to KCNQ1-KCNE3” and other similar statements. Previous studies have shown that KCNQ1/KCNE1-TVG is not a constitutively open KCNQ1/KCNE3-like channel (Panaghie et al., 2006; Barro-Soria et al., 2017), in contrast to the work by Melman. Maybe good to tone down this conclusion? Or more evidence showing that TVG converts KCNQ1-KCNE1 into constitutively conductive KCNQ1-KCNE3 would be good.

3) In Figure 2, why was the focus of the experiments on the extracellular portion of the (TM) region of E1? From the SI tables, there seems to be fewer constraints from interactions of the intracellular portion.

4) “ Both mutations resulted in smaller peak current amplitudes and significantly depolarized activation V_1/2 (ΔV_1/2,_K362A = 14.7 mV, ΔV_1/2,N365A = 14.7 mV) (Figure 5D+E, Supplementary file 1—table 5), indicating that these mutant channels required more energy to open, possibly due to a loss of stabilizing interactions with KCNE1.”. The conclusion is a little too strong. The fact that mutations at K362 and N365 make the channel activate at more depolarized voltages doesn't mean they affect the interaction with KCNE1, even though they seem close to KCNE1 in the model. K362 and N365 are in S6 and maybe just important for pore opening. Crosslinking or double mutant cycle analysis for the C ends of S6 and E1 (say K362 and S68), or KCNE1-dependent effects of K362 and N365 mutations would be needed to claim interactions.

5) Subsection “Experimental validation of KCNE1 residues interacting with KCNQ1” paragraph four. Mutations at W323 and Y46 share similar phenotypes of GV relations, but authors didn't show any experimental evidence to confirm that they are in direct contact (In the model they seem to be close together). Also, Y46 mutant had faster activation kinetics while W323 had wt-like activation kinetics, why is that if they are indeed functionally interacting?

6) Discussion paragraph two. "FTL can affect the nearly PAG motif through the mediation of S5", what residues in S5 is responsible for the mediation? From Figure 9 the authors labeled 263, 266, 267 and 270 and mentioned that mutations of them change the channel activation. Please clarify these mutational effects and discuss how they would differ for the FTL motif and the TVG motif.

7) Discussion paragraph three. This whole paragraph provides some experimental studies to support the model that FTL motif affects PAG motif via S5 involvement. But how this model induces the slower activation gating of KCNQ1 is not clear. To answer this question that the authors mentioned in the beginning of discussion, we would like more evidence and discussion concerning how FTL motif makes the KCNQ1 channel slow to open.

8) Discussion paragraph five. In Barro-Soria et al., 2017, they concluded that FTL also affects the S4 movement.

9) “ This mechanism may account for at least some of the gating differences between KCNE1 and KCNE3” and “ This proposed mechanism can account for the different effects of KCNE1 and KCNE3 on KCNQ1 gating”. This is too strong a statement. Similar to comment 7 above, how does the FTL motif makes Q1 channel open slowly while TVG motif makes Q1 constitutively leaky using this proposed model. It seems like the authors have more ideas about how KCNE1 affects KCNQ1 than how KCNE3 affects KCNQ1? Maybe stick to the more obvious effects seen/proposed?

10) Figure 5. Why does the amplitude go down for mutants? Is this really a good parameter to measure interactions? What is the rationale behind this conclusion?

11) Figure 8. Y267-T58 interaction seems very important for the proposed mechanism, but very little data is shown for this interaction (just simulations, except Y267F reducing expression, but consider concern above). Any more data for this interactions?

12) Figure 7. Y65A looks as good as Y65 or Y65F. So is the aromatic ring at Y65 really important?

13) In several places, data that is used to impose constraints in the model is then used to corroborate the relevance of the model. That should be avoided, or at least clarified.

14) The role of the interaction between Y267 and residues on S4 (M238 and D242) is intriguing, and its disruption when E1 is bound but not E3. Are those state-dependent and can they be related to the stabilization of the specific VSD state?

15) The 274-45 pair that are suggested to crosslink are not that close in either model (15A) and seem to point in different directions.

Presentation of results:

1) In Figure 5, the authors focus on residue W323, Y267 and K362/N365 because they had the largest number of contacts with E1 in a specific segment. Where is the data supporting this claim? The way the contacts are presented in this figure was confusing to several reviewers.

2) Figure 2—figure supplement 2. This is not the correct figure. wt KCNE1 should be shown, not as now with mutant KCNE1. This is a crucial control experiment. Also, the very important controls for wt KCNQ1 coexpressed with KCNE1 cys mutants are missing.

3) Figure 7 B-C. What are the authors trying to convey with these data? The reviewers did not see any obvious pattern. Is the correlation they propose really significant?

The reviewers agreed that the models deposited in PDB-dev would be of great value to the community, and that the review of experimental restraints used to build the models are in and of itself of great value. They also judged that the experimental validation of the proposed models was thorough and increased the confidence in the proposed modes of interaction between the two proteins. The data were thus convincing, but the conclusions about interactions were sometimes overstated and a more extensive discussion about the mechanism of KCNE1 was needed. Additionally there were some concerns about the way the cross-linking experiments were carried out. A detailed review is included below.

Experiments:

1) Related to Figure 2 and all other experiments using crosslinking: the authors show that under control, reducing or oxidizing conditions the magnitude of the currents are different. The way the data are presented and described led us to understand that each treatment, control, DTT or Cu-Phenantroline was applied to different groups of cells. If this is the case, this is a major experimental problem. Since there is variability in expression level and possibly changes in open probability and/or single channel conductance in each of the mutants that the authors are not accounting for, a simple comparison of the current density magnitude between different treatments is meaningless. These experiments should be carried out in the same cell. Typically disulfide bond formation (crosslinking) is induced by Cu-Phe after basal current recording and then reversed by DTT. There is a vast literature making use of this technique and this is the accepted way of performing this experiment. The authors should clarify if this is indeed what they did and show an example of such an experiment.

Conversely, if each treatment was applied to different groups of cells, the authors should show that the introduced cysteine residue produces no changes in open probability and single-channel conductance.

We agree with the reviewer that ideally the treatments would be carried out on the same cell. However, the I_Ks current when analyzed by whole-cell voltage clamp in mammalian cells exhibits rundown (see Vanoye et al., 2018, Supplemental Figure 2), which varies in both rate and magnitude, thus confounding experiments in which there is a time lapse between conditions or treatments. To avoid this problem, we recorded from multiple cells for each treatment (Control, +DTT, +Cu-phen.) to average the cell-to-cell variability. In addition, we recorded cells for all treatments from a specific cell transfection batch, and we used at least 3 transfections for each KCNQ1-KCNE1 combination. This information was previously omitted from Materials and methods, it has now been added to the end of subsection “Electrophysiology”.

We also now show that introduction of cysteine residues to KCNQ1 positions 141 and 274 do not affect channel activity behavior alone or in the presence of wild-type KCNE1 under control, +DTT or +Cu-phen (Figure 2—figure supplement 1 and Figure 2—figure supplement 2, respectively). These results support our conclusion that KCNQ1 residues V141 and I274 are in close proximity to KCNE1 residues L48 and L45, respectively (Figure 2).

2) The recordings shown by the authors were obtained in a limited voltage range, up to 60 mV. The tail currents show that the activation is far from saturation at this voltage and in consequence the estimation of the V1/2 from these curves incurs in an error, which could be very large. The authors should comment on these limitations or alternatively estimate this parameter from currents obtained at a wider range of voltages.

Normalized KCNQ1-KCNE1 tail currents do not saturate at the potentials tested nor would they saturate at more depolarized potentials. This is a widely observed phenomenon for the I_KS channel complex and is observed in both mammalian cells and Xenopus oocytes (e.g., Wang et al., 2020, Figure 3C; Wang et al., 2012, Figure 3C). We acknowledge this by referring to the V_1/2 value that we determine by curve fitting as the “apparent” activation V_1/2. We have changed V_1/2 to V_1/2app throughout the text.

Modeling:

1) To strengthen confidence in the modeling procedure, the following controls should be carried out: the docking of E1 should be done on the recent human structures of the activated/open state hKCNQ1 instead of only on the author's previous model. Conversely, docking of E3 to their AO model (and to the experimental structure mentioned above) should be performed to demonstrate that the procedure allows to recapitulate the experimental binding pose.

We thank the reviewers for this suggestion. We have performed these important control experiments and docked the NMR structures of KCNE1 and KCNE3 to the new cryo-EM structure of human KCNQ1 in the AO state and to the Rosetta AO model. The results of these experiments are presented in the paper in Figure 3—figure supplement 4 and in the text:

“To assure the robustness of our structure prediction protocol, control docking calculations were performed with the cryo-EM-determined AO state structure of human KCNQ1 (Sun and MacKinnon, 2020), which became available only after our KCNQ1-KCNE1 models were completed. These control calculations arrived at a model that was very similar to the one developed by docking KCNE1 to the Rosetta homology model of KCNQ1 (interface RMSD (I-RMSD) = 3.5 Å, Figure 3—figure supplement 4A). Additional control calculations were carried out with KCNE3, starting either with the cryo-EM-determined or Rosetta-predicted KCNQ1 model. Guided by a set of published experimental restraints for the KCNQ1-KCNE3 complex (Kroncke et al., 2016), this procedure was able to reproduce the experimental KCNE3 binding pose with an accuracy of I-RMSD = 2.5 Å or 4.2 Å, respectively, for these two structures (Figure 3—figure supplement 4B+C).”

2) Figure 3—figure supplement 1 shows a comparison of the models obtained in this work with the cryo-EM structure. The authors show that the backbone RMSD is fairly low between models and structure, but it would be more informative to show the degree of deviation of the position of side chains, especially in the regions in the models that were used for docking the KCNE1.

We have compared our models of human KCNQ1 used for docking with the cryo-EM structures of the AC and AO state in the region where KCNE1 was docked, i.e. between helices S1+S4 of one subunit and helices S5+S6+P from two neighboring subunits. We have calculated the sidechain RMSD and rotamer similarity of surface-exposed residues in these segments. The results of this analysis have been added to Figure 3—figure supplement 1 and are described in the paper: “In the putative KCNE1 binding region used for docking, the homology models agree well with the cryo-EM structures; surface-exposed residues have a sidechain RMSD less than 2.5Å and 4.0Å in the RC and AO model, respectively (Figure 3—figure supplement 1).”

3) Another aspect that is not even mentioned by the authors is the fact that the stoichiometry of KCNQ1-KCNE1 complexes appears to be variable. There is no experimental consensus on how many KCNE1 subunits are present per tetrameric KCNQ1 and the evidence seems to support the view that the number of KCNE1 subunits associated with a KCNQ1 tetramer depends on the expression level of the former. The authors seem to have chosen a 2:1 KCNE1:tetramer assembly for their simulations. Although a different stoichiometry might or might not alter the details of the interaction, it has been show that the variable stoichiometry can alter the biophysical effect (degree of voltage shift among others) of KCNE1, indeed suggesting that the mode of interaction might also not be unique. Did you try docking four KCNE1s per tetramer? Would this different stoichiometry alter the mode of interaction between KCNE1s and KCNQ1? At the very least, the authors should discuss this variable affects and their choice of stoichiometry.

We thank the reviewers for this thoughtful comment. In addition to describing the preparation of KCNQ1-KCNE1 models with 4:2 stoichiometry in Materials and methods, we have now added more discussion about our choice of the KCNQ1:KCNE1 stoichiometry and the variable effects on KCNQ1 function that can be caused by different numbers of KCNE1 subunits.

“For MD simulations, KCNQ1-KCNE1 models were prepared with a stoichiometry of 4:2 KCNQ1:KCNE1 subunits. […] In our final MD analysis, we focused on the 4:2 stoichiometry and observed no significant changes in the interaction mode between the two KCNE1 subunits and with respect to the model obtained by Rosetta docking.”

Contextualization and Interpretation of data:

1) Abstract: "FTL… affects the channel gate via an allosteric network." This is not shown, but only a proposed mechanism. Please state as suggestion, not fact.

We have rephrased this sentence as a suggestion: “The FTL motif binds at a cleft between the voltage-sensing and pore domains and appears to affect the channel gate by an allosteric mechanism.”

2) “ Likewise, mutation of FTL to TVG renders the KCNQ1-KCNE1 channel constitutively active, similar to KCNQ1-KCNE3” and other similar statements. Previous studies have shown that KCNQ1/KCNE1-TVG is not a constitutively open KCNQ1/KCNE3-like channel (Panaghie et al., 2006; Barro-Soria et al., 2017), in contrast to the work by Melman. Maybe good to tone down this conclusion? Or more evidence showing that TVG converts KCNQ1-KCNE1 into constitutively conductive KCNQ1-KCNE3 would be good.

We have corrected these statements to clarify that TVG does not completely convert KCNQ1-KCNE1 into a constitutively open channel.

“Mutation of FTL to TVG renders the KCNQ1-KCNE1 channel similar to KCNQ1-KCNE3, in that faster activation at more negative potentials is observed (Melman et al., 2001; Barro-Soria et al., 2017).”

“Replacement of FTL with TVG from KCNE3 shifts the G(V) curve of KCNQ1-KCNE1 channels towards that of KCNQ1-KCNE3 and removes KCNE1-specific effects on the gate and S4 movement (Barro-Soria et al., 2017).”

3) In Figure 2, why was the focus of the experiments on the extracellular portion of the (TM) region of E1? From the SI tables, there seems to be fewer constraints from interactions of the intracellular portion.

The previous KCNQ1-KCNE1 model by Kang et al., 2008, suggested that the extracellular region of the KCNE1 TMD is in close distance to two gain-of-function mutation sites, V141 and I274, in KCNQ1, and we pursued to test this hypothesis. It has also been our experience that crosslinking of intracellular sites of KCNQ1+KCNE1 expressed in mammalian cells is technically challenging due to the limited accessibility of these sites to the oxidizing/reducing reagents.

We have now added an additional paragraph to the manuscript comparing our models with published interaction data for the intracellular C-terminal domains of KCNQ1 and KCNE1 (see response to the next comment). These restraints were initially not used in model building because the interaction sites are located in flexible linker regions of KCNE1 and inclusion of flexible linkers in docking proved computationally intractable. However, the experimentally observed interaction sites are close to the last KCNE1 residue in our models. This supports the model-predicted proximity between the intracellular portions of KCNE1 and KCNQ1 S6.

4) “ Both mutations resulted in smaller peak current amplitudes and significantly depolarized activation V_1/2 (ΔV_1/2,_K362A = 14.7 mV, ΔV_1/2,N365A = 14.7 mV) (Figure 5D+E, Supplementary file 1—table 5), indicating that these mutant channels required more energy to open, possibly due to a loss of stabilizing interactions with KCNE1.”. The conclusion is a little too strong. The fact that mutations at K362 and N365 make the channel activate at more depolarized voltages doesn't mean they affect the interaction with KCNE1, even though they seem close to KCNE1 in the model. K362 and N365 are in S6 and maybe just important for pore opening. Crosslinking or double mutant cycle analysis for the C ends of S6 and E1 (say K362 and S68), or KCNE1-dependent effects of K362 and N365 mutations would be needed to claim interactions.

We have removed statements from the manuscript which suggested that the effects seen with mutations at K362 and N365 could indicate that these residues interact with KCNE1. Instead we have added a paragraph comparing the structural models with published experimental interaction data for site H363.

“A proximity between the C-terminal ends of the KCNE1 TMD and S6 is supported by the results of cysteine-crosslinking experiments (Lvov et al., 2010), which showed that H363C in KCNQ1 formed disulfide bonds with H73C, S74C, and D76C in KCNE1. […] This result can also be explained by our structural models (Figure 5—figure supplement 3).”

5) Subsection “Experimental validation of KCNE1 residues interacting with KCNQ1” paragraph four. Mutations at W323 and Y46 share similar phenotypes of GV relations, but authors didn't show any experimental evidence to confirm that they are in direct contact (In the model they seem to be close together). Also, Y46 mutant had faster activation kinetics while W323 had wt-like activation kinetics, why is that if they are indeed functionally interacting?

To test if W323 and Y46 are functionally interacting, we have designed a double mutant cycle experiment to measure the free energy changes of channel activation (ΔG) for KCNQ1-KCNE1 channels carrying Ala mutations at either KCNQ1 W323, KCNE1 Y46, or at both sites simultaneously. Unfortunately, channels formed by KCNQ1 W323A and KCNE1 Y46A generated very low currents (see Author response image 1), which prohibited determination of ΔG for this channel and calculation of the coupling energy between W323 and Y46 (paragraph three subsection “Experimental validation of KCNE1 residues interacting with KCNQ1”).

Author response image 1

Download asset Open asset

Nevertheless, the prediction of our model that W323 and Y46 are interacting is consistent with the observation that mutations at both sites show similar G(V) relations and that the homologous residue in KCNE3 (Y60) forms a direct contact with W323 in the KCNQ1-KCNE3 channel structure. Furthermore, we observed that mutations at both positions, W323 and Y46, led to faster activation compared to the WT KCNQ1-KCNE1 channel (see Author response image 2), an effect that was more pronounced for mutations to smaller amino acid (Ala, Leu) than for mutations to Phe.

For easier comparison of the kinetics of channels formed by different KCNQ1 and KCNE1 mutants in this work, we have now plotted the activation and deactivation time constants and show these results in Figure 7 and Figure 5—figure supplement 2 in the manuscript.

Author response image 2

Download asset Open asset

6) Discussion paragraph two. "FTL can affect the nearly PAG motif through the mediation of S5", what residues in S5 is responsible for the mediation? From Figure 9 the authors labeled 263, 266, 267 and 270 and mentioned that mutations of them change the channel activation. Please clarify these mutational effects and discuss how they would differ for the FTL motif and the TVG motif.

We have clarified the mutational effects seen for residues in S5, which suggest an interaction with KCNE1 FTL.

“Alanine mutational scanning of S5 previously showed that the V_1/2 shift of KCNQ1 activation by KCNE1 is reduced by mutations Y267A and F270A, and to a smaller extent by L271A, I274A, and F275A (Strutz-Seebohm el al., 2011). […] These data are consistent with a binding of FTL to this region on S5 leading to triggering of changes in the activation gate of KCNQ1.”

We describe differences between the interaction modes of FTL and TVG, and, using Rosetta energy calculations and in-silico alanine mutagenesis, compare possible changes in their binding strength with S5.

“We observed different interactions are made by the KCNE1 FTL versus the KCNE3 TVG (Figure 9B+C, Video 4). […] Scoring with the Rosetta energy function suggests a larger binding free energy for FTL compared to TVG (Figure 9D, Figure 9—figure supplement 1C) and a decrease of binding by Ala mutations of residues in S5 (Figure 9—figure supplement 1D). […]”

7) Discussion paragraph three. This whole paragraph provides some experimental studies to support the model that FTL motif affects PAG motif via S5 involvement. But how this model induces the slower activation gating of KCNQ1 is not clear. To answer this question that the authors mentioned in the beginning of discussion, we would like more evidence and discussion concerning how FTL motif makes the KCNQ1 channel slow to open.

We agree with the reviewer that our structural models cannot fully explain how the KCNE1 FTL motif causes slow activation gating of KCNQ1.Therefore, we have revised our discussion and focus on questions which are answered more directly by the models developed in this work.

“How does the KCNE1 FTL motif interact with KCNQ1 to control KCNQ1 activation gating? And, what can be concluded about the TMD binding mode for both KCNE1 and KCNE3 and the mechanism underlying the different impact on channel activation by KCNE3?”

As in our previous manuscript version, we describe the basis for an allosteric mechanism by which KCNE1 appears to affect the KCNQ1 gate, and provide experimental evidences for this mechanism.

We then discuss how differences between the binding modes of KCNE1 FTL and KCNE3 TVG lead to different molecular forces acting on S5, and propose that different force transductions determine the effect of the KCNE proteins on the KCNQ1 gate via involvement of S5.

“Comparison of the binding modes of KCNE1 and KCNE3 also suggests how the activation motifs of these proteins could trigger different effects on the KCNQ1 gate through mediation of S5. […] Together these observations support the notion that FTL and TVG induce distinct interactions of varying strength with S5 that determine the effect of the KCNE subunits on the S6 gate.”

As in our previous manuscript submission, we conclude the discussion with thoughts about whether the TMD binding mode seen for KCNE1 and KCNE3 could be conserved in other KCNE proteins.

8) Discussion paragraph five. In Barro-Soria et al., 2017, they concluded that FTL also affects the S4 movement.

We have corrected this statement: “The FTL and TVG motifs were found to determine whether the KCNE subunits affect the gate and the second S4 movement that is seen in voltage clamp fluorometry studies of KCNQ1-KCNE1 and correlates with the opening of KCNQ1-KCNE1 channels.”

9) “ This mechanism may account for at least some of the gating differences between KCNE1 and KCNE3” and “ This proposed mechanism can account for the different effects of KCNE1 and KCNE3 on KCNQ1 gating”. This is too strong a statement. Similar to comment 7 above, how does the FTL motif makes Q1 channel open slowly while TVG motif makes Q1 constitutively leaky using this proposed model. It seems like the authors have more ideas about how KCNE1 affects KCNQ1 than how KCNE3 affects KCNQ1? Maybe stick to the more obvious effects seen/proposed?

As suggested by the reviewer, we have revised our discussion and stick to the more obvious effects proposed by the structural models (see our response to comment 7). We have adjusted our claims and removed the first statement and rephrased the second statement as a suggestion: “Differences in their interactions with S5 appear to explain the different effects of KCNE1 and KCNE3 on the gate in KCNQ1.”

10) Figure 5. Why does the amplitude go down for mutants? Is this really a good parameter to measure interactions? What is the rationale behind this conclusion?

We agree with the reviewers that current amplitude is not a good parameter to measure KCNQ1-KCNE1 interaction. In fact, in our manuscript, conclusions about the KCNQ1-KCNE1 interaction were not derived from peak current data. We reported these data in order to fully describe the functional phenotype of the KCNQ1- KCNE1 mutants tested. To avoid confusion, we have removed the unnormalized I-V curves from Figure 5E and Figure 7A. Instead, we show the activation and deactivation times for KCNE1 Y46, F57, and Y65 mutants in Figure 7A. We include the current density data in Supplementary file 1—table 6, but have removed statements comparing current density between WT and mutant conditions from the manuscript text.

11) Figure 8. Y267-T58 interaction seems very important for the proposed mechanism, but very little data is shown for this interaction (just simulations, except Y267F reducing expression, but consider concern above). Any more data for this interactions?

To test if KCNQ1 Y267 and KCNE1 T58 are interacting, we have performed a double mutant cycle experiment by substituting Y267 with Phe and T58 with Val, either separately or in combination, and determined the changes in the free energy of activation of the resulting channel complexes.

“The energy changes for KCNQ1 Y267F–KCNE1 (ΔG = 0.66 kcal/mol), KCNQ1–KCNE1 T58V (ΔG = -0.36 kcal/mol) and KCNQ1 Y267F–KCNE1 T58V (ΔG = 0.90 kcal/mol) were not additive, however, the net energy change (|ΔΔG| = 0.60 kcal/mol) was smaller than 1.0 kcal/mol, which is commonly used as lower cutoff to identify two residues as interacting. Thus, we were not able to experimentally confirm an interaction between Y267 and T58. However, we note that the free energy changes at T58 were previously observed to have a pronounced sidechain volume dependency (Strutz-Seebohm el al., 2011) and that substitutions to amino acids involving a more drastic change in sidechain size could reveal a stronger energetic coupling between Y267 and T58 than determined in this work.”

12) Figure 7. Y65A looks as good as Y65 or Y65F. So is the aromatic ring at Y65 really important?

Our data suggest that both sidechain size and OH functionality at KCNE1 Y65 contribute to in the interaction with KCNQ1, because mutations to Phe, which preserves the aromatic ring but lacks the OH group, as well as substitutions with Leu or Ala led to significant changes in channel activation properties as described:

“Mutations at Y65 led to a shift of V_1/2app to more positive voltages (ΔV_1/2app,Y65A = 4.3 mV, ΔV_1/2app,Y65L = 12.3 mV, ΔV_1/2app,Y65F = 9.4 mV), and Y65L and Y65A showed faster activation and deactivation kinetics (Figure 7A, right panel). This is consistent with the observation of Y65 forming sidechain packing and hydrogen bond interactions with multiple KCNQ1 residues in the MD simulations.”

13) In several places, data that is used to impose constraints in the model is then used to corroborate the relevance of the model. That should be avoided, or at least clarified.

We apologize for this mistake. We have removed the sentence about mutation W323C from the manuscript. Our intention was to compare the volume effects of different mutations at position 323. We now compare our results to mutational effects at other residues in S6 and helix P that are close to W323.

“Mutations of other residues in S6 (V324, V334) and in the nearby P helix (A300, V307) also caused channel opening at more negative voltages, likely via destabilization of the closed state. This region has been implicated with the positive G(V) shift by KCNE1 (Kakajo et al., 2011).”

14) The role of the interaction between Y267 and residues on S4 (M238 and D242) is intriguing, and its disruption when E1 is bound but not E3. Are those state-dependent and can they be related to the stabilization of the specific VSD state?

In our structural modeling work of the KCNQ1 VSD in different activation states we have observed that Y267 can make interactions with different S4 residues in the resting, intermediate, and activated state, respectively (see Author response image 3). It is possible that the type and strength of these state-dependent interactions are affected differently by the presence of KCNE1 or KCNE3 as basis for stabilizing specific VSD states. To fully and quantitatively investigate this hypothesis, we are considering additional computational and functional experiments.

Author response image 3

Download asset Open asset

15) The 274-45 pair that are suggested to crosslink are not that close in either model (15A) and seem to point in different directions.

In the model of the KCNQ1-KCNE1 closed state, which appears to be the state that is promoted by I274C-L45C crosslinking, L45 and I274 are oriented at an angle of about 90° and have a CA-CA distance of ~15 Å (see Author response image 4). In MD simulations, which allowed for exploration of additional protein flexibility, L45 and I274 came within a CA-CA distance of 13.5 Å which is still larger than the expected maximal CA-CA cysteine-crosslinking distance of ~8 Å. We speculate that this experimentally observed crosslink may have induced a moderate perturbation of the KCNQ1-KCNE1 structure which brought L45C and I274C closer together than in the WT model inducing a low-conductance closed-like state.

We have added a sentence to the manuscript providing a possible explanation for this restraint violation:

“The restraint with the medium violation involved residues KCNE1 L45 and KCNQ1 I274. It is possible that crosslinking between these cysteine-substituted sites slightly perturbed the KCNQ1-KCNE1 structure leading to a low-conductance closed-like state, which could explain why in the WT channel model these residue sidechains have suboptimal geometry for disulfide bond formation.”

Author response image 4

Download asset Open asset

Presentation of results:

1) In Figure 5, the authors focus on residue W323, Y267 and K362/N365 because they had the largest number of contacts with E1 in a specific segment. Where is the data supporting this claim? The way the contacts are presented in this figure was confusing to several reviewers.

We apologize for the confusion and have refined Figure 5. The number of contacts that is made by a KCNQ1 residue with KCNE1 is displayed in the surface plots in panels A and B and colored from white (contact number ≤ 1) to red (contact number > 20). For improved readability, we have added the contact numbers in parentheses next to the residue labels.

2) Figure 2—figure supplement 2. This is not the correct figure. wt KCNE1 should be shown, not as now with mutant KCNE1. This is a crucial control experiment. Also, the very important controls for wt KCNQ1 coexpressed with KCNE1 cys mutants are missing.

We apologize for this mistake. When submitting the manuscript, Figure 2—figure supplement 2 got accidently changed to a different figure. We now provide the correct version of Figure 2—figure supplement 2 showing control data for KCNQ1 cysteine mutants co-expressed with KCNE1 WT.

In addition, in Figure 2—figure supplement 3 we show that cysteine-substituted KCNQ1-KCNE1 residue pairs V141C-V47C, I274C-V47C, and I274C-L48C led to no significant changes in peak current in the presence of oxidizing or reducing conditions, respectively. Our negative results with those three KCNQ1-KCNE1 residue pairs indicated to us that the effects we observed in the presence of DTT or Cu-phenanthroline with KCNQ1_V141C-KCNE1_L48C or KCNQ1_I274C-KCNE_L45C are residue pair specific. In addition, earlier experiments carried out in Xenopus oocytes (see Author response image 5) showed that the KCNQ1_V141C-KCNE1_L45C pair was insensitive to oxidizing conditions. Together, these results indicated to us that crosslinking was residue-pair specific and we did not test the KCNE1 L45C, V47C and L48C substitutions against wild type KCNQ1.

Author response image 5

Download asset Open asset

3) Figure 7 B-C. What are the authors trying to convey with these data? The reviewers did not see any obvious pattern. Is the correlation they propose really significant?

We validate the model-predicted binding mode for KCNE1 by comparing the location of KCNQ1-contacting residues in KCNE1, indicated in Figure 7B by changes in SASA >20% between KCNE1 alone and KCNE1+KCNQ1, with changes in activation V_1/2 owing to mutagenesis in KCNE1 displayed in Figure 7C. KCNE1 residues with |ΔV_1/2| > |ΔV_1/2|_threshold (20 mV for KCNE1 expressed in oocytes in previous studies (Chen et al., 2007; Wang et al., 2012), 10 mV for KCNE1 expressed in CHO-K1 cells in this study) were considered high-impact mutation sites. For ease of comparison, we have refined Figure 7 and highlight high-impact mutation sites and positions with ΔSASA > 20% in blue color. In addition, we have computed the Spearman rank correlation between ΔSASA and |ΔV_1/2| and performed a χ² test of independence of the observed frequencies of KCNQ1-contact sites and high-impact mutation sites in KCNE1. We compare the significance of this correlation between the model-predicted KCNE1 binding mode and three other hypothetical binding orientations where the KCNE1 TM helix is rotated 90° or 180° in the clockwise or anti-clockwise direction. The results of this analysis are shown in Figure 7—figure supplement 1 and summarized in the text:

“Thus, there is a clear dependence between KCNQ1-contact sites and high impact mutation sites in KCNE1 for the orientation proposed by the structural models (P < 0.05, χ²-test). However, no significant correlation was found when we simulated other hypothetical orientations for KCNE1 by rotation around its helical screw axis (Figure 7—figure supplement 1).”

Author details

1. Georg Kuenze

   1. Center for Structural Biology, Vanderbilt University, Nashville, United States
   2. Department of Chemistry, Vanderbilt University, Nashville, United States
   3. Institute for Drug Discovery, Leipzig University, Leipzig, Germany

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   For correspondence

   georg.kuenze@gmail.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-1799-346X

2. Carlos G Vanoye

   Department of Pharmacology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

3. Reshma R Desai

   Department of Pharmacology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   Data curation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

4. Sneha Adusumilli

   Department of Pharmacology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   Data curation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

5. Kathryn R Brewer

   1. Center for Structural Biology, Vanderbilt University, Nashville, United States
   2. Department of Biochemistry, Vanderbilt University, Nashville, United States

   Contribution

   Data curation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

6. Hope Woods

   1. Center for Structural Biology, Vanderbilt University, Nashville, United States
   2. Department of Chemistry, Vanderbilt University, Nashville, United States

   Contribution

   Data curation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

7. Eli F McDonald

   1. Center for Structural Biology, Vanderbilt University, Nashville, United States
   2. Department of Chemistry, Vanderbilt University, Nashville, United States

   Contribution

   Data curation, Investigation, Methodology, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0572-330X

8. Charles R Sanders

   1. Center for Structural Biology, Vanderbilt University, Nashville, United States
   2. Department of Biochemistry, Vanderbilt University, Nashville, United States

   Contribution

   Conceptualization, Resources, Data curation, Supervision, Funding acquisition, Validation, Project administration, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2046-2862

9. Alfred L George Jr

   Department of Pharmacology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Validation, Project administration, Writing - review and editing

   Competing interests

   No competing interests declared

10. Jens Meiler

    1. Center for Structural Biology, Vanderbilt University, Nashville, United States
    2. Department of Chemistry, Vanderbilt University, Nashville, United States
    3. Institute for Drug Discovery, Leipzig University, Leipzig, Germany
    4. Department of Pharmacology, Vanderbilt University, Nashville, United States

    Contribution

    Conceptualization, Resources, Supervision, Funding acquisition, Validation, Project administration, Writing - review and editing

    Competing interests

    No competing interests declared

• 1,786

  Page views

• 191

  Downloads

• 7

  Citations

Article citation count generated by polling the highest count across the following sources: PubMed Central, Crossref, Scopus.