Zinc is a key regulator of the sperm-specific K+ channel (Slo3) function

  1. Rizki Tsari Andriani
  2. Tanadet Pipatpolkai
  3. Haruhiko Miyata
  4. Masahito Ikawa
  5. Yasushi Okamura
  6. Takafumi Kawai  Is a corresponding author
  1. Integrative Physiology, Graduate School of Medicine, The University of Osaka, Japan
  2. Graduate School of Medicine, The University of Osaka, JSPS International Research Fellow, Japan
  3. School of Physical and Mathematical Science, Nanyang Technological University, Singapore
  4. Research Institute for Microbial Diseases, The University of Osaka, Japan
  5. Graduate School of Frontier Bioscience, The University of Osaka, Japan
  6. Department of Molecular and Cellular Physiology, Graduate School of Medicine, Ehime University, Japan
12 figures, 1 table and 1 additional file

Figures

Figure 1 with 2 supplements
Intracellular zinc is exported from sperm during capacitation and inhibits the alkalinization-induced hyperpolarization in sperm.

(A–C) Intracellular zinc is exported from sperm during capacitation. (A) Representative image of FluoZin3 fluorescence observed in spermatozoa (head and flagellum, as indicated). (B) Time course of normalized FluoZin3 fluorescence response upon the application of capacitation medium. Both head and flagellum showed a decreased fluorescence response, indicating that intracellular zinc was exported from sperm in response to the application of capacitation medium. (C) Comparison of the fluorescence intensity from flagellum (908.6 ± 508.2; n = 12) and head (5409 ± 1648; n = 12) after capacitation. Unpaired t-test, p < 0.0001. All error bars are ± SD centered on the mean. (D–H) Intracellular zinc inhibits the alkalinization-induced hyperpolarization in sperm. (D–G) Membrane potential (Vm) recording by perforated patch-clamp was performed in mouse spermatozoa as depicted in (D). (E) Representative trace of Vm recording for the control group. The application of NH4Cl increased the intracellular pH (pHi) which then hyperpolarized the membrane, the process called alkalinization-induced hyperpolarization. Application of NH4Cl alone hyperpolarized the membrane by –17.37 ± 7.09 mV (n = 6). (F) Representative trace of Vm recording in the presence of zinc ionophore (zinc pyrithione, ZnPy). When 100 µM ZnPy was applied on top of NH4Cl application, no hyperpolarization was observed (∆Vm = 0.99 ± 6.17 mV; n = 3). All the dotted lines indicated 0 mV level. (G) Comparison of ∆Vm recording between control (n = 6) and in the presence of 100 µM ZnPy (n = 3), p = 0.0067, unpaired t-test. (H–K) Vm recording by whole-cell patch-clamp configuration in mouse spermatozoa as depicted in (H). (I) Representative trace of Vm recording for the control group. The application of NH4Cl increased the [pHi] which then hyperpolarized the membrane (∆Vm = –24.12 ± 13.16 mV; n = 5). (J) Representative trace of Vm recording in the presence of 100 µM ZnCl2. When zinc was applied to the intracellular part of spermatozoa on top of NH4Cl application, alkalinization-induced hyperpolarization was inhibited (∆Vm = –6.80 ± 2.35 mV; n = 4). All the dotted lines indicated 0 mV level. (K) Comparison of ∆Vm recording between control (n = 5) and in the presence of 100 µM ZnCl2 (n = 4), p = 0.0373, unpaired t-test. (L–N) Slo3 current recording using perforated patch-clamp in sperm. (L) Comparison of Slo3 current amplitude at +100 mV before NH4Cl application between control (152.4 ± 25.05 pA; n = 6) and in the presence of 100 µM ZnPy (187.5 ± 79.53 pA; n = 4), p = 0.4862, Welch’s t-test. (M) Comparison of Slo3 current amplitude at +100 mV after NH4Cl application between control (427.6 ± 88.28 pA; n = 4) and in the presence of 100 µM ZnPy (249.2 ± 130.6 pA; n = 4), p = 0.0703, Welch’s t-test. (N) Comparison of the fold change in Slo3 current amplitude at +100 mV after NH4Cl application between control (2.39 ± 0.55; n = 4) and in the presence of 100 µM ZnPy (1.4 ± 0.56; n = 4), p = 0.0465, Welch’s t-test. All error bars are ± SD centered on the mean.

Figure 1—figure supplement 1
Effects of intracellular zinc on sperm motility before and after capacitation.

Comparison of VAP, VCL, and VSL show that intracellular zinc affects these kinds of motions only after the sperm underwent capacitation. (A–E) Quantitation of sperm motility parameters at: 10 min (control); 10 min in the presence of zinc chelator TPEN; 2 hr after TYH incubation, a well-established capacitation-inducing medium (control); and 2 hr after THY incubation in the presence of TPEN. Bar graph shows the comparison of (A) VAP (average path velocity); (B) VCL (curvilinear velocity); (C) VSL (straight-line velocity); (D) LIN (linearity); and (E) STR (straightness) as depicted in (F). All error bars are ± SD centered on the mean.

Figure 1—figure supplement 2
The effects of zinc on Slo3 current in sperm cells.

(A) Comparison of normalized Slo3 current at +100 mV before NH4Cl application between control (50.67 ± 20.67 pA/pF; n = 6) and in the presence of 100 µM ZnPy (59.95 ± 30.24 pA/pF; n = 4), p = 0.6156, Welch’s t-test. (B) Comparison of normalized Slo3 current at +100 mV after NH4Cl application between control (144.8 ± 40.36 pA/pF) and in the presence of 100 µM ZnPy (77.59 ± 44.97 pA/pF; n = 4), p = 0.0684, Welch’s t-test. All error bars are ± SD centered on the mean.

Figure 2 with 1 supplement
Zinc inhibits mouse Slo3 (mSlo3) current.

(A–C) Inside-out patch-clamp recording of mSlo3 with pHi = 8.0 by applying ramp pulse from –100 to +100 mV with the holding potential of 0 mV. (A) Representative current traces upon the application of pH 8.0 (black); 100 µM of zinc in pH 8.0 (yellow-orange); pH 6.0 (blue); and washout by pH 8.0 (gray) after pH 6.0. (B) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, pH 6.0, and pH 8.0 as indicated (white: pH 8.0; yellow orange: 100 µM zinc in pH 8.0; solid black: pH 6.0). (C) Dose–response curve of zinc inhibition in mSlo3 current (pH 8.0). % current inhibition was plotted assuming that the dose–response curve has a standard slope, equal to a Hill slope of –1.0. IC50 = 145.2 µM; n = 2–6. (D–F) Inside-out patch-clamp recording of mSlo3 in Xenopus oocyte with pHi = 7.5 by applying ramp pulse from –100 to +100 mV with the holding potential of 0 mV. (D) Representative current traces upon the application of pH 7.5 (black); µM of zinc (yellow); and pH 6.0 (blue). (E) Time course of the change in current amplitude at +100 mV in response to pH 7.5, 100 µM zinc, and pH 6.0 as indicated (white: pH 7.5; yellow: 100 µM zinc in pH 7.5; solid black: pH 6.0). (F) Current amplitude at +100 mV upon the application of pH 7.5, 100 µM zinc in pH 7.5, and pH 6.0 from individual cells, as indicated. (G, H) Inside-out patch-clamp recording of mSlo3 with pHi = 8.0 by applying step pulses from –100 to +180 mV with the holding potential of –60 mV. Zinc inhibition is voltage-dependent. (G) Representative current traces upon the application of pH 8.0 (black) and 100 µM of zinc in pH 8.0 (yellow orange). (H) I–V relationship of control (pH 8.0; black) and zinc (100 µM; yellow orange). Normalization was performed by dividing the absolute current of mSlo3 at pH 8.0 of each voltage by the absolute current at the pre-determined highest voltage that still produced a stable mSlo3 current (i.e., good patch, good clamp). In this analysis, +140 mV was chosen as the highest voltage for normalization, since in a few cells the patch was lost at +160 and +180 mV. Similar to the control condition, the absolute current of mSlo3 in the presence of 100 µM zinc was normalized to the absolute current of the control at +140 mV. I–V curve was fitted by the Boltzmann equation for control and followed by the Woodhull equation for zinc inhibition (V50 Ctrl = 85.24; Slope Ctrl = 35.51; Vblock Zn = 188.8; Slopeblock Zn = –115.3; n = 13). (I–K) Inside-out patch-clamp recording of mSlo3 with the application of 5 mM EGTA in pH = 8.0 by applying ramp pulses from –100 to +100 mV with the holding potential of 0 mV. Zinc has a long-lasting inhibitory effect on mSlo3, and it could be rescued by EGTA (I) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, and pH 8.0 as indicated (white: pH 8.0; yellow orange: 100 µM zinc in pH 8.0; pink: pH 8.0 for wash-out). (J) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, and 5 mM EGTA in pH 8.0 as indicated (white: pH 8.0; yellow orange: 100 µM zinc in pH 8.0; light lilac: 5 mM EGTA in pH 8.0 for wash-out). (K) Comparison of percentage of current recovery between pH 8.0 only (5.62 ± 24.40%; n = 7) and 5 mM EGTA in pH 8.0 (49.22 ± 37.98%; n = 5), p = 0.0351, unpaired t-test. All error bars are ± SD centered on the mean.

Figure 2—figure supplement 1
Zinc inhibits mSlo3 current.

(A) Time course of Slo3 current at +100 mV upon the application of 0.1 µM zinc. (B) Relationship between relative current (Izinc/Icontrol) with membrane voltage at pH 8.0 shows that zinc inhibits mSlo3 current in a voltage-dependent manner. (C) G–V relationship comparison of mSlo3 current between control (pH 8.0; black filled circle) and upon the application of 100 µM zinc in pH 8.0 (yellow orange circle) (V50 control = 80.67 ± 2.32; V50 zinc = 49.677 ± 1.38; n = 13). Normalization was done based on the maximum conductance in control. (D) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 5 mM EGTA in pH 8.0, 100 µM zinc, pH 8.0, and 5 mM EGTA in pH 8.0 as indicated (white: pH 8.0; yellow orange: 100 µM zinc in pH 8.0; pale lilac: 5 mM EGTA in pH 8.0). The baseline current showed a slight increase upon perfusing with 5 mM EGTA, indicating that endogenous zinc is already present in mSlo3, and it is affecting the baseline current. (E) Comparison of percentage of mSlo3 current evoked by pH 8.0 and 5 mM EGTA in pH 8.0 (p = 0.0392). Statistical analysis was done by paired t-test, n = 3. All error bars are ± SD centered on the mean.

Figure 3 with 1 supplement
Zinc inhibits mSlo3 current when co-expressed with gamma subunit Lrrc52.

(A, B) Inside-out patch-clamp recording of mSlo3 co-expressed with mLrrc52 with pHi = 8.0 by applying ramp pulse from –100 to +100 mV with the holding potential of 0 mV. (A) Representative current traces upon the application of pH 8.0 (black); 100 µM of zinc in pH 8.0 (yellow orange); pH 6.0 (blue); and wash-out by pH 8.0 after pH 6.0 (gray). (B) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, pH 6.0, and pH 8.0 as indicated (white: pH 8.0; yellow orange: 100 µM zinc in pH 8.0; solid black: pH 6.0). (C) Dose–response curve of zinc inhibition in mSlo3 co-expressed with mLrrc52 current (pH 8.0). % current inhibition was plotted assuming that the dose response curve has a standard slope, equal to a Hill slope of –1.0. IC50=15.2 µM; n = 3–8. (D, E) Inside-out patch-clamp recording of mSlo3 in Xenopus oocyte co-expressed with mouse Lrrc52 (mLrrc52) with pHi = 7.5 by applying ramp pulse from –100 to +100 mV with the holding potential of 0 mV. (D) Representative current traces upon the application of pH 7.5 (black); µM of zinc (yellow); pH 7.5 for wash-out (grey); and pH 6.0 (blue). (E) Time course of the change in current amplitude at +100 mV in response to pH 7.5, 100 µM zinc, pH 6.0, and pH 7.5 for wash-out as indicated (white: pH 7.5; yellow: 100 µM zinc in pH 7.5; orange: pH 6.0). (F) Comparison of the percentage of current inhibition by 100 µM zinc between pH 7.5 (32.51 ± 8.62%; n = 3) and pH 8.0 (52.61 ± 8.23%; n = 8), p = 0.006; unpaired t-test. All error bars are ± SD centered on the mean.

Figure 3—figure supplement 1
Zinc inhibits mSlo3 current when coexpressed with auxiliary Lrrc52.

(A–C) Inside-out patch-clamp recording of mSlo3 co-expressed with mLrrc52 with pHi = 8.0 by applying step pulses from –100 to + 180 mV with the holding potential of –60 mV. Zinc inhibition is voltage-dependent. (A) Representative current traces upon the application of pH 8.0 (black) and 100 µM of zinc in pH 8.0 (yellow-orange). (B) I–V relationship of control (pH 8.0; black) and zinc (100 µM; yellow-orange). I–V curve was fitted by Boltzmann equation for control and followed by Woodhull equation for zinc inhibition (V50 Ctrl = 74.01; SlopeCtrl = 36.47; Vblock Zn = 185.5; Slopeblock Zn = –130.4; n = 8). (C) G–V relationship comparison of mSlo3 current co-expressed with mLrrc52 between control (pH 8.0; black filled circle) and upon the application of 100 µM zinc in pH 8.0 (yellow-orange circle) (n = 13). V50 control = 72.93 ± 2.06 V50 zinc = 56.22 ± 2.75. Normalization was done based on the maximum conductance in control. (D) Comparison of the percentage of Slo3 current inhibition at pH 7.5 between Slo3 only (–0.13 ± 0.08%; n = 3) and Slo3 + Lrrc52 (32.51 ± 8.62%; n = 3), p = 0.0028, unpaired t-test. All error bars are ± SD centered on the mean.

Figure 4 with 1 supplement
Computational studies show that the Zn-binding site on mSlo3 is located near E169 and E205.

(A) AlphaFold3 prediction of the Zn-binding site on the transmembrane segment of the Slo3 channel. Zn is shown as gray spheres, and Slo3 is shown using an orange cartoon. (B) The system is set up for flooding simulations. The phosphorus atoms on the POPC lipid are shown in orange. All Zn ions in the system are shown as gray spheres. The Slo3 channel is shown in the yellow cartoon. All water, KCl, and other atoms within POPC molecules are omitted for clarity. (C) The density of Zn ions within 4Å of the protein molecules was calculated using VolMap averaged across 200 ns flooding simulations (n = 3). Key acidic contacting residues are shown as red sticks. (D) Root mean square (RMSD) calculation of the 12 Zn ions fitted to the C alpha atoms of the protein on the first frame for 200 ns. Given that there are 4 Zn in each subunit per repeat, this leads to 12 data sets shown in different colored plots. The red dotted line marked the cutoff at 2.5 Å. (E) The fractions of frames where the RMSD of zinc ions are less than 2.5 Å in WT channel, E169A, and E205A mutants (p < 0.0001 and p = 0.0188, respectively). Statistical analysis was done by one-way ANOVA with Dunnett’s post hoc test compared to the WT group for each mutant (n = 12). All error bars are ± SD centered on the mean.

Figure 4—figure supplement 1
Computational studies and scanning mutagenesis of zinc-binding site in mSlo3 channel.

(A) The full-length Slo3 tetramer structure predicted by AlphaFold3, with RCK1 and RCK2 represented in blue and yellow cartoons, respectively, and 50 Zn2+ ions depicted as red spheres. (B) Comparison of the percent current inhibition upon the application of 100 µM zinc between wildtype and all the histidine mutants scanned (histidine residues located in RCK1, RCK1/RCK2 linker, RCK2, and the chimera in which the RCK2 of mSlo3 is replaced with the RCK2 from mSlo1). Dotted black line indicates mean of control (WT). (C) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, and pH 8.0 for wash-out as indicated for all the mutants as stated. All error bars are ± SD centered on the mean.

Figure 5 with 1 supplement
E169 and E205 located within voltage-sensing domain (VSD) are important for the sustained zinc inhibition of mSlo3 current.

(A) Tetrameric mSlo3 structure representation. Inset shows rotated view of selected region which indicates predicted zinc-binding site from MD simulations located within the voltage-sensor domain: D162 (S2–S3 linker), E169 (S3 domain), and E205 (S4 domain) which have been mutated to elucidate the zinc-binding site in mSlo3. mSlo3 structure representation in (A) was based on structure prediction using AlphaFold3 (Abramson et al., 2024), mSlo3 sequence was retrieved from NCBI NP_032458.3. Some portions of the S2 and S3 domains have been rendered highly transparent to enhance clarity. (B) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, and pH 8.0 for wash-out as indicated for: WT, D162S, E169A, and E205A. The dotted line indicated the level of activated mSlo3 current at pH 8.0. Black arrows indicated the current recovery observed upon wash-out. (C) Comparison of the percent current inhibition upon the application of 100 µM zinc between WT, D162S, E169A, and E205A (p = 0.6499, p = 0.2746, and p = 0.9629, respectively). Statistical analysis was done by one-way ANOVA with Dunnett’s post hoc test compared to the control group (WT), n = 3–7. (D) Comparison of percentage of current recovery upon wash-out using pH 8.0 between WT, D162S, E169A, and E205A (p = 0.8135, p = 0.0314, and p = 0.0112, respectively). Statistical analysis was done by unpaired t-test compared to the WT group for each mutant, n = 3–7. All error bars are ± SD centered on the mean.

Figure 5—figure supplement 1
Computational studies and mutagenesis of zinc-binding site in mSlo3 channel.

(A) The density of Zn ions within 4 Å of the protein molecules was calculated using VolMap averaged across 200 ns flooding simulations (n = 3). Key acidic contacting residues are shown as red sticks. (B) Tetrameric mSlo3 structure representation of selected region which indicates D162 (S2–S3 linker), E169 (S3 domain), E205 (S4 domain), and E326 (intracellular linker between TM and RCK1) which have been mutated to elucidate the zinc-binding site in mSlo3. (C) Time course of the change in current amplitude at +100 mV in response to pH 8.0, 100 µM zinc, and pH 8.0 for wash-out as indicated for E326A. (D) Comparison of the percent current inhibition upon the application of 100 µM zinc between wildtype and E326A (p = 0.8232, unpaired t-test, n = 2–7). (E) Comparison of percentage of current recovery upon wash-out using pH 8.0 between WT and E326A (p = 0.6388, unpaired t-test, n = 2–7). All error bars are ± SD centered on the mean. (F) Sequence alignment of mSlo3 (NP_032458.3), mSlo1 (NP_001240293.1), and mSlo2.2 (NP_001138875.1) using ClustalO. E169 and E205 residues (mSlo3 numbering) are highlighted in red text with a yellow box.

Figure 6 with 1 supplement
Zinc influenced the motion of the voltage-sensing domain (VSD) of mSlo3 channel.

(A) Representative current traces and fluorescence signals of mSlo3 L193C coexpressed with Lrrc52 in the absence (green) and presence (yellow orange) of zinc. The inset illustrates a schematic timeline of direct zinc injection using a glass needle and positive pressure. Voltage steps from –160 to + 160 mV were applied with a holding potential of –60 mV. Zinc was injected into the same cell after the initial voltage-clamp fluorometry (VCF) recording (green), followed by a second VCF recording 60 s post-injection (yellow orange). ∆F/FControl = 2.84 ± 1.25% (n = 7); ∆F/FZinc = 2.53 ± 1.37% (n = 7). The green and yellow orange traces represent the current and fluorescence signals at in +160 and –160 mV, in the absence (green) and presence (yellow orange) of zinc. (B) Comparison of the F–V relationship between control (green) and zinc (yellow orange). The F–V curve was fitted using the Boltzmann equation (V50 control = −21.68 ± 3.04; V50 zinc = 27.04 ± 1.05; n=7). Normalization was done based on the maximum ∆F/F for each control and zinc-injected conditions. (C) I–V relationship showing inhibition of mSlo3 L193C+Lrrc52 currents by direct zinc injection. Traces compare control (green) and zinc (yellow orange). (D) G–V relationship comparison of mSlo3 L193C current co-expressed with mLrrc52 between control (green circle) and zinc-injected (yellow orange circle) conditions (V50 control = 21.11 ± 4.8; V50 zinc = 38.62 ± 28.8; n = 7). Normalization was done based on the maximum conductance in control. (E) Comparison of ∆F/F in the absence and presence of zinc at +160 mV from the same cell (p = 0.1869, paired t-test, n = 7). (F) Comparison of ∆F/F in the absence and presence of zinc at 0 mV from the same cell (p = 0.0028, paired t-test, n = 7). All error bars are ± SD centered on the mean.

Figure 6—figure supplement 1
Scanning mutagenesis of mSlo3 channel voltage-clamp fluorometry (VCF) construct.

(A) Structural representation of the VCF scanning regions in the mSlo3 channel, highlighting the top of S3 (yellow), the S3–S4 linker (violet), and the top of S4 (mud green). (B–H) Representative current traces and fluorescence signals recorded upon voltage steps from –80 to +160 mV with a holding potential of –60 mV for L188C, K189C, S190C, N191C, L193C (+mLrrc52), G194C (+mLrrc52), and L195C (+mLrrc52), respectively (n = 4–7; ∆F/F mSlo3 L193C + Lrrc52:=1.15 ± 0.48% [n = 4]).

Author response image 1
AlphaFold3 prediction of the Zn binding site on IC side of Slo3.
Author response image 2
MD simulations of histidines residues located in IC of Slo3.
Author response image 3
AlphaFold3 model of Slo3 channel with Mg2+.
Author response image 4
VCF control experimentes: vehicle injection.
Author response image 5
Zinc addition.
Author response image 6
Δ comparition between control and TPEN.

Tables

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
Gene (Mus musculus)Mus musculus Slo3Schreiber et al., 1998
Gene (Mus musculus)Mus musculus Lrrc52Yang et al., 2011
Strain, strain background (Escherichia coli)DH5αTakara BioCat. #9057Competent cells
Biological sample (Mus musculus)SpermJapan SLC, IncFreshly isolated from Mus musculus cauda epididymis
Commercial assay or kitPrimeSTAR Mutagenesis Basal KitTakara BioCat. #R045A
Commercial assay or kitIn-Fusion HD CloningTakara BioCat. #639649
Commercial assay or kitmMESSAGE SP6 RNA transcription kitThermo Fisher ScientificAM1340
Chemical compound, drugZinc pyrithioneFujifilm Wako267-01232
Chemical compound, drugTPENCayman ChemicalCAS: 16858-02-9
Cat. #CAY13340
Chemical compound, drugGramicidinMerckCat. #G5002
Chemical compound, drugFluoZin3-AMInvitrogenCat. #F24195
Chemical compound, drug0.15% TricaineSigma-Aldrich and Tokyo Chemical IndustryE10521 and CAS 886-86-2
Chemical compound, drugCollagenase type 1Sigma-AldrichSCR103
Chemical compound, drugCollagenase-PRocheCOLLP-RO
Chemical compound, drugTMRM-5InvitrogenT6027
Chemical compound, drugZnCl2Fujifilm Wako260-01021
Software, algorithmIgor Pro 6.22AWavemetricsRRID:SCR_000325
Software, algorithmPyMOL Molecular Graphics SystemSchrodinger LLCRRID:SCR_014212
Software, algorithmGraphPad Prism 11GraphPad Software, IncRRID:SCR_002798
Software, algorithmAlphaFold 3Google DeepMindRRID:SCR_025885
Software, algorithmCHARMM-GUIWu et al., 2014RRID:SCR_01489
Software, algorithmCHARMM36mHuang et al., 2017
Software, algorithmGROMACS 2023.3Abraham et al., 2023RRID:SCR_014565

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Rizki Tsari Andriani
  2. Tanadet Pipatpolkai
  3. Haruhiko Miyata
  4. Masahito Ikawa
  5. Yasushi Okamura
  6. Takafumi Kawai
(2026)
Zinc is a key regulator of the sperm-specific K+ channel (Slo3) function
eLife 14:RP105450.
https://doi.org/10.7554/eLife.105450.3