Discovery and functional assessment of a novel adipocyte population driven by intracellular Wnt/β-catenin signaling in mammals

  1. Zhi Liu
  2. Tian Chen
  3. Sicheng Zhang
  4. Tianfang Yang
  5. Yun Gong
  6. Hong-Wen Deng
  7. Ding Bai
  8. Weidong Tian  Is a corresponding author
  9. YiPing Chen  Is a corresponding author
  1. Department of Cell and Molecular Biology, Tulane University, United States
  2. State Key Laboratory of Oral Diseases, National Clinical Research Center for Oral Disease, West China Hospital of Stomatology, Sichuan University, China
  3. Tulane Center of Biomedical Informatics and Genomic, Deming Department of Medicine, School of Medicine, Tulane University, United States
6 figures and 6 additional files

Figures

Figure 1 with 3 supplements
The existence of adipocytes exhibiting active Wnt/β-catenin signaling.

(A and B) Immunofluorescent and microscopy images of active Wnt/β-catenin in vivo (A) and in vitro (B), indicated by GFP expression from the TCF/Lef:H2B-GFP allele, in adipocytes marked by Perilipin or BODIPY (red). Nuclei were stained with DAPI (blue). Scale bars, 50 μm; close-up scale bars, 20 μm. (C) Quantification of Wnt+ adipocytes among total adipocytes in various fat depots of adult male mice in (A). n = 6–9 mice. (D) Quantification of Wnt+ adipocytes among total adipocytes induced from SVF cells derived from adult male fat depots and BM stroma in (B). . n = 7–10 independent experiments. (E) Immunofluorescent staining of Pparγ (purple) and Adiponectin (red) in cultured GFP+ adipocytes derived from bone marrow in (B). n = 3 independent experiments. Scale bars, 20 μm. (F) Representative images of GFP+ (green arrows) and GFP- (white arrows) adipocytes induced from human bone marrow stromal cells infected with TCF/Lef:H2B-GFP reporter lentiviral virus. n = 2 independent experiments, 3 independent wells each. Scale bar, 20 μm. Data are mean ± s.e.m., ***p < 0.001, one-way ANOVA followed by Tukey’s test.

Figure 1—figure supplement 1
An inducible mouse model validates the existence of Wnt/β-catenin-positive adipocytes.

(A) Schematic overview of the TCF/LefCreERT2 allele. (B) Illustration of the validation experiments in Wnt/β-catenin-positive lineage reporter mice. (C) Microscopy images of freshly isolated mGFP-positive adipocytes from iBAT of TCF/LefCreERT2;Rosa26RmTmG mice. (D) GFP expression in the cell membrane and nucleus of an adipocyte in Tcf/LefCreERT2;Rosa26RmTmG;T/L-GFP mice. Scale bar, 10 μm. (E and F) Tcf/LefCreERT2;Rosa26RmTmG mice exhibited mGFP-positive adipocytes (white arrows) in iBAT (E) and iWAT (F) after tamoxifen treatment. n = 4 mice each. Scale bars, 50 μm; close-up scale bars, 20 μm. White arrows point to Wnt+ adipocytes.

Figure 1—figure supplement 2
Age- and sex-dependent dynamics of Wnt+ adipocytes.

(A) Immunofluorescent images of subcutaneous WAT adjacent to iBAT from E17.5 T/L-GFP mouse showing initial appearance of Wnt+ adipocytes. n = 3 embryos. Scale bar, 100 μm; close-up scale bar, 25 μm. (B) Immunofluorescent images of iBAT from E18.5 T/L-GFP mice showing initial appearance of Wnt+ adipocytes. n = 3 embryos. Scale bar, 100 μm; close-up scale bar, 25 μm. (C) Immunofluorescent images of iBAT and iWAT from adult male and female mice showing Wnt+ adipocytes. n = 4–7 mice. Scale bar, 50 μm. (D) Quantification of Wnt+ adipocytes among total adipocytes in (C). (E) Immunofluorescent images of various fat depots from T/L-GFP mice at different ages showing the presence of Wnt+ adipocytes. n = 3 mice each. Scale bar, 50 μm. (F) Quantification of Wnt+ adipocytes among total adipocytes in (E). (G) Immunofluorescent images of Wnt+ adipocytes induced from E13.5 MEFs in vitro. n = 2 independent experiments from 8 embryonic mice, 3 independent wells each. Scale bar, 100 μm; close-up scale bar, 20 μm. White arrows in all panels point to Wnt+ adipocytes. Data are mean ± s.e.m., **p < 0.01, ***p < 0.001, n.s., not significant, unpaired Student’s t-test (D) or one-way ANOVA followed by Tukey’s test (F).

Figure 1—figure supplement 3
Infection of lentiviral virus in hBMSCs study.

(A) Construct of the lentiviral shuttle vector. (B) Representative immunofluorescent images of infected hBMSCs in pro-osteogenic differentiation medium showing the expression of GFP in the nuclei of osteocytes, indicative of the activation of Wnt/β-catenin signaling. n = 2 independent experiments, 3–4 independent wells each. Scale bars, 50 μm.

Figure 2 with 2 supplements
Characterization of mouse Wnt+ adipocytes.

(A) Real-time monitoring microscopy images of Wnt+ adipocytes during adipogenesis, which were induced from BMSCs of T/L-GFP mice. n = 3 independent experiments. Scale bar, 50 μm. (B) Schematic of the experiments and microscopy images of separated Wnt+ and Wnt- adipocytes by FACS. n = 3 independent experiments. Scale bars, 100 μm. (C) Immunofluorescent images of iWAT and iBAT of AdipoqCre;Ctnnb1dm/flox;T/L-GFP mice showing complete absence of Wnt+ adipocytes. Muscle cells were included as positive controls for GFP expression. n = 3 mice. Scale bar, 50 μm. (D) Time-lapse microscopy images of induced Wnt+ adipocytes from BMSCs of T/L-GFP mice with Ctnnb1 and control siRNA-mediated knockdown. n = 2 independent experiments, 3 independent wells each. Scale bars, 50 μm. (E and F) Immunofluorescent images of Wnt+ adipocytes induced from two independent cell lines (GFPpos-1 and –2) with LF3 treatment (50 μM) for 1 (E) and 2 days (F), respectively. LF3 was added into the medium after 3-day pro-adipogenic induction. Note that by 1 day LF3 administration, GFP signals were significantly quenched in Wnt+ adipocytes, along with obviously reduced cell number. By 2-day LF3 administration, remarkable cell death of Wnt+ adipocytes was seen, compared to controls. Scale bar, 100 μm.

Figure 2—figure supplement 1
Wnt/β-catenin signaling in adipocytes is activated intracellularly.

(A) FACS analysis of freshly isolated SVFs from iWAT of T/L-GFP mice showing the lack of GFP+ cell population. Representative of at least five biological replicates. (B) Microscopy images of GFP- and GFP+ colonized adipocytes after 7-day differentiation induced from BMSCs of T/L-GFP mice. n = 4 independent experiments. Scale bar, 200 μm. (C and D) Immunofluorescent images of active β-catenin (C) and TCF7L2 (D) in Wnt+ adipocytes. n = 3 mice each. Scale bars, 20 μm (C), 25 μm (D). (E) Microscopy images and Oil Red O staining of Wnt+ adipocytes induced from iBAT-derived SVF cells of T/L-GFP mice with control and LF3 treatment in different doses. n = 5 independent experiments. Scale bars, 50 μm. (F and G) Immunofluorescent images of induced Wnt+ adipocytes from BMSCs of T/L-GFP mice treated with DKK1 (F) and IWP-2 (G). n = 4 independent experiments each. Scale bars, 50 μm. (H) Quantification of Wnt+ adipocytes among total adipocytes in (F) and (G). Data are mean ± s.e.m., n.s., not significant, one-way ANOVA followed by Tukey’s test. White arrows indicate Wnt+ adipocytes.

Figure 2—figure supplement 2
Wnt/β-catenin signaling plays a role in adipogenesis of Wnt- adipocytes.

(A and B) Immunofluorescent images of Wnt- adipocytes induced from two independent cell lines (GFPneg-1 and –2) with LF3 treatment (50 μM) for 1 (A) and 2 days (B), respectively. LF3 was added into the medium after 3-day pro-adipogenic induction. Note that by 1-day LF3 administration, Wnt- adipocytes began to show delayed adipogenic maturation. By 2-day LF3 administration, delayed adipogenic maturation in Wnt- adipocytes became clear, compared to controls. Scale bar, 100 μm. (C) Microscopy images of Oil Red O staining in adipocytes induced from GFPneg-1 and –2 cell lines showing reduced lipid accumulation by LF3 treatment for 2 days. n = 3 independent experiments. Scale bar, 50 μm. (D) Quantitative RT-PCR analysis of the expression of adipogenic genes in (C). The levels of mRNA expression are normalized to that of 36B4. (E) Schematic of rescue experiment and microscopy images of Oil Red O staining in adipocytes induced from GFPneg-1 cell line with LF3 treatment. n = 3 independent experiments. Scale bar, 50 μm. (F) Quantitative RT-PCR analysis of the expression of adipogenic genes in (E). The levels of mRNA expression are normalized to that of 36B4. Data are mean ± s.e.m., *p < 0.05, **p < 0.01, ***p < 0.001, n.s., not significant, one-way ANOVA followed by Tukey’s test (F) or unpaired Student’s t-test (D).

Figure 3 with 1 supplement
Distinct molecular and genomic signatures of Wnt+ and Wnt- adipocytes.

(A) Schematic of scRNA-seq and scATAC-seq experiments on SVF-induced adipocytes. (B) Uniform Manifold Approximation and Projection (UMAP) visualization of 2537 adipocytes from iBAT (1710 GFP+ and 827 GFP-) and 1345 adipocytes from iWAT (984 GFP+ and 361 GFP-) in scRNA-seq. (C) UMAP visualization of 4302 adipocytes from iBAT (1727 GFP+ and 2575 GFP-) and 1383 adipocytes from iWAT (562 GFP+ and 821 GFP-) in scATAC-seq. (D and E) Heat maps of expression of top 30 signature genes (Supplementary file 2) in iBAT- (B) and iWAT-derived (C) Wnt+ and Wnt- adipocytes in scRNA-seq. (F) Violin plots of induced Wnt+ and Wnt- adipocytes showing the distribution of normalized expression values of some representative genes in scRNA-seq. (G–J) Hallmark gene sets (G), KEGG pathway (H), Wikipathway (I), and GO Biological Processes ontology (J) analyses of DEGs enriched in iWAT-derived Wnt+ adipocytes in scRNA-seq (Supplementary file 3).

Figure 3—figure supplement 1
Wnt+ adipocytes are distinct from Wnt- ones in molecular signatures.

(A and B) UMAP visualization of unsupervised clustering in scRNA-seq (A) and scATAC-seq (B). (C and D) ScATAC-seq tracks of iBAT- (C) and iWAT-derived (D) Wnt+ and Wnt- adipocytes showing distinct chromatin accessibilities of representative genes. (E and F) Enriched DNA motifs of iBAT- (E) and iWAT-derived (F) Wnt+ (upper panels) and Wnt- (lower panels) adipocytes in scATAC-seq. (G and H) Immunofluorescent staining of Cidea in adipocytes induced from iBAT-derived SVFs (G) and Cyp2e1 in adipocytes induced from iWAT-derived SVFs (H). White arrows point to Wnt- adipocytes without Cidea or Cyp2e1 expression, green arrows point to Wnt+ adipocytes with positive staining of indicated protein. n = 3 independent experiments. Scale bars, 100 μm; close-up scale bars, 25 μm.

Figure 4 with 1 supplement
AKT/mTOR cascade is indispensable to trigger intracellular β-catenin signaling in Wnt+ adipocytes.

(A) Western blot analysis showing protein levels of AKT, phosphorylated AKT, and β-actin in adipocytes induced from mBaSVF, GFPpos-1, and GFPneg-1 cell lines. n = 3 independent experiments. (B) Western blot analysis showing protein levels of GSK-3β, phosphorylated GSK-3β, 4E-BP1, and phosphorylated 4E-BP1 in adipocytes induced from mBaSVF, GFPpos-1, and GFPneg-1 cell lines. n = 3 independent experiments. (C and D) Immunofluorescent images of Wnt+ and Wnt- adipocytes induced from two independent precursor cell lines (GFPpos-1 and –2) with LY294002 treatment (14 μM) for 1 (C) and 2 days (D), respectively. LY294002 was added into the medium after 3-day pro-adipogenic induction. LY294002 treatment diminished GFP signals prior to causing marked cell death in Wnt+ adipocytes. n = 5 independent experiments. Scale bar, 100 μm. (E) Immunofluorescent images of iWAT and iBAT of T/L-GFP mice treated with or without Temsirolimus (Tem). n = 5 mice. Scale bar, 100 μm. (F) Quantification of Wnt+ adipocytes among total adipocytes in (E). (G) Immunofluorescent images of iWAT and iBAT of male Ins2Akita;T/L-GFP mice at 8 weeks of age. n = 5 mice. Scale bar, 100 μm. (H) Quantification of Wnt+ adipocytes among total adipocytes in (G). Data are mean ± s.e.m., ***p < 0.001, unpaired Student’s t-test.

Figure 4—source data 1

Full-sized western blot images for Figure 4A and B.

Full-scans of western blots. Full-sized image of western blot from Figure 4A and B. Red box indicates area that was cropped and displayed in the indicated figure.

https://cdn.elifesciences.org/articles/77740/elife-77740-fig4-data1-v1.zip
Figure 4—figure supplement 1
Inhibition of Akt signaling by LY294002 yields similar results as LF3 treatment.

(A) Microscopy images of LY294002-treated adipocytes showing reduced number of Wnt+ adipocytes and delayed adipogenesis after 3- and 7-day treatment compared to controls. n = 4 independent experiments. Scale bar, 20 μm. (B) Quantification of Wnt+ adipocyte among total adipocytes after 7-day treatment in (A). (C and D) Immunofluorescent images of Wnt- adipocytes induced from two independent precursor cell lines (GFPneg-1 and –2) with LY294002 treatment (14 μM) for 1 (C) and 2 days (D), respectively. LY294002 was added into the medium after 3 day pro-adipogenic induction. LY294002 treatment led to slightly reduced lipid storage in Wnt- adipocytes. n = 5 independent experiments. Scale bar, 100 μm. (E) Quantification of the percentage of 7-AAD-positive cells by flow cytometry analyses showing comparable cell viability in immortalized GFPpos and GFPneg precursor cell lines treated with LF3, LY294002, or DMSO at the same doses. n = 3 independent experiments. (F) Schematic of bulk RNA-seq analyses on Wnt+ adipocytes with and without LF3 treatment. (G) KEGG pathway analysis of DEGs that were down-regulated in LF3-treated Wnt+ adipocytes in bulk RNA-seq (Supplementary file 5). Data are mean ± s.e.m., ***p < 0.001, n.s., not significant, one-way ANOVA followed by Tukey’s test.

Figure 5 with 3 supplements
Wnt+ adipocytes are essential for initiating adaptive thermogenesis.

(A) Immunofluorescent staining of mitochondrial membrane potentials in Wnt+ adipocytes induced from iBAT-derived SVF cells of T/L-GFP mice. n = 5 independent experiments. Scale bar, 100 μm; close-up scale bar, 20 μm. (B) Quantification of staining in (A). AOD, average optical density. Data are mean ± s.e.m., unpaired Student’s t-test; ***p < 0.001. (C) OCR plots of four groups of adipocytes differentiated from GFPpos-1, GFPpos-2, GFPneg-1, and mBaSVF cell lines, respectively. n = 3 independent experiments. (D) Immunofluorescent staining of iWAT from T/L-GFP mice with 4-day thermal challenge showing close association of Wnt+ adipocytes with UCP1+ beige adipocytes. n = 5 mice. Scale bar, 100 μm. (E) Immunofluorescent staining of iWAT from tamoxifen-treated Tcf/LefCreERT2;Rosa26RmTmG mice after 4-day cold exposure. n = 8 mice. Scale bar, 50 μm. (F) Immunofluorescent staining of iWAT from tamoxifen-treated T/L-DTA mice after 4-day cold exposure. Before cold challenge, mice were rested for 48 hr after final tamoxifen treatment. n = 7 mice. Scale bar, 50 μm. (G) Quantitative RT-PCR analysis of gene expression in iWAT from control (Fabp4-Flex-DTA) and T/L-DTA mice in (F). n = 6 and 7 mice. Levels of mRNA expression are normalized to that of Adipoq. (H) Western blot analysis showing protein levels of UCP1, OXPHOS complexes, and β-actin in iWAT from tamoxifen-treated controls and T/L-DTA mice under cold (6 °C) and ambient (22 °C) temperatures. n = 4 mice. (I) Immunofluorescent staining for UCP1 (green) and Perilipin (red) in the iWAT from control and Akita mice. Scale bars, 50 μm. Data are mean ± s.e.m., *p < 0.05, ** p < 0.01, ***p < 0.001, n.s., not significant, two-way repeated ANOVA followed by Bonferroni’s test (C) or unpaired Student’s t-test (G).

Figure 5—source data 1

Full-sized western blot images for Figure 5H.

Full-scans of western blots. Full-sized image of western blot from Figure 5H. Red box indicates area that was cropped and displayed in the indicated figure.

https://cdn.elifesciences.org/articles/77740/elife-77740-fig5-data1-v1.zip
Figure 5—figure supplement 1
Wnt+ adipocytes participate in adaptive thermogenesis in iWAT.

(A and B) Calculated basal (A) and maximal (B) respiration levels in OCR assay in Figure 5C. (C) Immunofluorescent images of iWAT from T/L-GFP mice after 2-day cold exposure showing emergence of beige fat cells surrounding to Wnt+ adipocytes (white arrows). Representative of five biological replicates. Scale bar, 50 μm; close-up scale bars, 100 μm. (D) Immunofluorescent images of iWAT from T/L-GFP mice after CL316,243 treatment (n = 6 mice). White arrows point to Wnt+ adipocytes. Scale bar, 100 μm. (E) Quantification of Wnt+ adipocytes among total adipocytes in iWAT from T/L-GFP mice after 2-day cold stress (n = 5 mice), CL316,243 treatment (n = 6 mice), and controls (room temperature, n = 4 mice). (F) Schematic overview of the Fabp4-Flex-DTA allele. (G) Quantification of Wnt+ adipocytes among total adipocytes of iWAT and iBAT from Fabp4-Flex-DTA;T/L-GFP and T/L-DTA;T/L-GFP mice 24 hours after 2 day tamoxifen treatment prior to cold challenge. n = 4 mice each. (H) Immunofluorescent staining of iWAT from tamoxifen-treated T/L-DTA mice after 2-week cold exposure. Before cold challenge, mice were rested for 48 hr after the second tamoxifen injection. n = 4 mice. Scale bar, 50 μm. (I) Quantitative RT-PCR analysis of Adipoq expression in iWAT from control (Fabp4-Flex-DTA) and T/L-DTA mice in Figure 5G, showing comparable results between two groups. n = 6 and 7 mice. The levels of mRNA expression are normalized to that of 36B4. Data are mean ± s.e.m., **p < 0.01; ***p < 0.001, one-way ANOVA followed by Tukey’s test (A), (B), and (E) or unpaired Student’s t-test (G and I).

Figure 5—figure supplement 2
Two additional mouse models validate the critical role of Wnt+ adipocytes in adaptive thermogenesis in iWAT.

(A) Schematic overview of the Tcf/Lef-rtTA allele. (B) Schematic of TRE-Cre;Tcf/Lef-rtTA;Rosa26RmTmG mice on a doxycycline diet (600 mg/kg) starting at the age of 3 weeks old. (C) Immunofluorescent stereo microscopic images showing mGFP-positive dermal cells of TRE-Cre;Tcf/Lef-rtTA;Rosa26RmTmG mice in (B) as positive controls of known canonical Wnt signaling activity. Scale bar, 20 μm. (D) TRE-Cre;Tcf/Lef-rtTA;Rosa26RmTmG mice exhibited mGFP-positive Wnt+ adipocytes (white arrows) in both iBAT and iWAT in (B), n = 4 mice. Scale bar, 50 μm. (E) GFP expression in the cell membrane and nucleus of an adipocyte in TRE-Cre;Tcf/Lef-rtTA;Rosa26RmTmG;T/L-GFP mice on a doxycycline diet. Scale bar, 10 μm. (F) Immunofluorescent staining of iWAT from doxycycline-treated rtTA-DTA mice after 1-week cold exposure. White arrowhead indicates lymph node (LN). n = 5 mice. Scale bar, 50 μm. (G) Immunofluorescent staining of iWAT from doxycycline-treated rtTA-PpargF/F mice after 1-week cold exposure. White arrowhead indicates lymph node (LN). n = 3 mice. Scale bar, 50 μm.

Figure 5—figure supplement 3
Akita mice manifest impaired adaptive thermogenic response.

(A) Quantitative RT-PCR analysis of gene expression in iWAT from Akita mice and littermate controls after 2-day cold challenge. n = 3 mice for each genotype. Levels of mRNA expression are normalized to that of 36B4. Data are mean ± s.e.m., *p < 0.05, ** p < 0.01, n.s., not significant, unpaired Student’s t-test. (B) Western blot analysis showing protein levels of UCP1, OXPHOS complexes, Pparγ, Adiponectin, and β-actin in iWAT from Akita mice and littermate controls after 2-day cold exposure. n = 3 mice for each genotype.

Figure 5—figure supplement 3—source data 1

Full-sized western blot images for Figure 5—figure supplement 3B.

Full-scans of western blots. Full-sized image of western blot from Figure 5—figure supplement 3B. Red box indicates area that was cropped and displayed in the indicated figure.

https://cdn.elifesciences.org/articles/77740/elife-77740-fig5-figsupp3-data1-v1.zip
Wnt+ adipocytes enhance systemic glucose homeostasis.

(A) Glucose tolerance test (GTT) with calculated area under the curve (AUC) in tamoxifen-treated control (Fabp4-Flex-DTA) and T/L-DTA mice on regular chow diet. n = 7 mice each. (B) Schematic of Wnt+ adipocyte gain-of-function studies by cell implantation. (C) Photographs of fat pad formed by implanted cells. Blue agarose beads were included to locate the Matrigel pad. Black arrow shows benign vascularization of fat pad within two weeks. Scale bar, 50 μm. (D) Immunofluorescent staining for Perilipin showing mature adipocytes and accompanied agarose beads (marked by B) in the ectopically formed fat pad in (C). Scale bar, 50 μm. (E) GTT with calculated AUC in mice that received implantation of committed pre-adipocytes/adipocytes from mBaSVF (n = 8 mice) or GFPpos-1 (n = 7 mice) cell lines for 2 weeks. Data are mean ± s.e.m., *p < 0.05, **p < 0.01, two-way repeated ANOVA followed by Bonferroni’s test. AUC was analyzed by two-tailed t-test, p = 0.0012 (A), 0.0017 (E).

Additional files

Supplementary file 1

Primer sequences used for qRT-PCR.

https://cdn.elifesciences.org/articles/77740/elife-77740-supp1-v1.docx
Supplementary file 2

Sheet 1: Enriched genes in scRNA-seq.

Sheet 2: Overrepresented DNA motifs in scATAC-seq.

https://cdn.elifesciences.org/articles/77740/elife-77740-supp2-v1.xlsx
Supplementary file 3

Pathway analyses results of DEGs enriched in iWAT-derived Wnt+ adipocytes in scRNA-seq.

https://cdn.elifesciences.org/articles/77740/elife-77740-supp3-v1.xlsx
Supplementary file 4

Core body temperature results of T/L-DTA and control mice.

Core body temperature below 34.5 °C was considered as hypothermia and such mice were subsequently euthanized.

https://cdn.elifesciences.org/articles/77740/elife-77740-supp4-v1.xlsx
Supplementary file 5

Sheet 1: Significantly down-regulated genes of LF3-treated Wnt+ adipocytes in bulk RNA-seq.

Sheet 2: KEGG pathway analysis results of DEGs down-regulated in LF3-treated Wnt+ adipocytes in bulk RNA-seq.

https://cdn.elifesciences.org/articles/77740/elife-77740-supp5-v1.xlsx
Transparent reporting form
https://cdn.elifesciences.org/articles/77740/elife-77740-transrepform1-v1.docx

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. Zhi Liu
  2. Tian Chen
  3. Sicheng Zhang
  4. Tianfang Yang
  5. Yun Gong
  6. Hong-Wen Deng
  7. Ding Bai
  8. Weidong Tian
  9. YiPing Chen
(2022)
Discovery and functional assessment of a novel adipocyte population driven by intracellular Wnt/β-catenin signaling in mammals
eLife 11:e77740.
https://doi.org/10.7554/eLife.77740