Skip to main content
PMC home page
eLife logoLink to eLife
. 2022 Jun 28;11:e78170. doi: 10.7554/eLife.78170

Gill developmental program in the teleost mandibular arch

Mathi Thiruppathy 1, Peter Fabian 1, J Andrew Gillis 2,3, J Gage Crump 1,✉
Editors: Carole LaBonne4, Marianne E Bronner5
PMCID: PMC9239679  PMID: 35762575

Abstract

Whereas no known living vertebrate possesses gills derived from the jaw-forming mandibular arch, it has been proposed that the jaw arose through modifications of an ancestral mandibular gill. Here, we show that the zebrafish pseudobranch, which regulates blood pressure in the eye, develops from mandibular arch mesenchyme and first pouch epithelia and shares gene expression, enhancer utilization, and developmental gata3 dependence with the gills. Combined with work in chondrichthyans, our findings in a teleost fish point to the presence of a mandibular pseudobranch with serial homology to gills in the last common ancestor of jawed vertebrates, consistent with a gill origin of vertebrate jaws.

Research organism: Zebrafish

Introduction

Gills are the major sites of respiration in fishes. They are composed of a highly branched system of primary and secondary filaments, housing blood vessels, a distinct type of cellular filament cartilage, pillar cells (specialized endothelial cells), and epithelial cells maintaining ionic balance. In teleost gills, two rows of filaments are anchored to a prominent gill bar skeleton. Both the filaments and supportive gill bars develop from the embryonic pharyngeal arches that consist of mesenchyme of neural crest and mesoderm origin and epithelia of endodermal and ectodermal origin (Fabian et al., 2022; Mongera et al., 2013). The third and more posterior arches generate gills in most fishes. The second (hyoid) arch also forms a hemibranch (one row of gill filaments) in the jawless lamprey fish (Dohrn, 1882; Gaskell, 1908), in cartilaginous and various non-teleost fishes (e.g. coelacanth, lungfishes, sturgeon, and gar), but not in teleost fishes (Goodrich, 1930; Jollie, 1962). A classical theory for the origin of jaws posits that an ancestral gill support skeleton in the mandibular arch was repurposed for jaw function (Gegenbaur et al., 1878). However, extant agnathans (the cyclostomes lamprey and hagfish) lack a mandibular gill (Cole, 1905; Mallatt, 1996), and fossil evidence for ancestral vertebrates with a mandibular gill is scant. Whereas exceptional soft tissue preservation of Metaspriggina walcotti from the Cambrian Burgess Shale had suggested a dorsoventrally segmented cartilaginous gill bar in the presumptive mandibular arch, gill filaments were not observed (Morris and Caron, 2014).

The pseudobranch is an epithelial structure located just behind the eye that has been proposed to regulate ocular blood pressure and/or have an endocrine function (Jollie, 1962). While it shares an anatomical resemblance to gill filaments and is found in many jawed fishes (Dohrn, 1882), its embryonic arch origins remain debated (Miyashita, 2016). In parallel work in little skate, we identify a mandibular arch origin of the pseudobranch in chondrichthyans (Hirschberger and Gillis, 2022), which form one branch of jawed vertebrates. Whether the mandibular origin of the pseudobranch is conserved across vertebrates, including bony fishes, remained unknown. Another important question is whether the pseudobranch and gills can be considered serially homologous, i.e., representing morphologically related structures that arise through shared developmental and genetic mechanisms. Through lineage tracing and genetic analyses in zebrafish here, and lineage analysis in skate (Hirschberger and Gillis, 2022), we infer that the pseudobranch is a mandibular arch-derived serial homolog of the gills that was present in at least the last common ancestor of jawed vertebrates.

Results

In zebrafish, the pseudobranch is located anterior to the gill filaments and connected to the eye via the ophthalmic artery (Figure 1a, c), as described for other fishes (Laurent and Dunel-Erb, 1984). The pseudobranch appears in histological sections as a small bud behind the eye at 4 days post-fertilization (dpf) (Figure 1b). Examination of Sox10:Cre; acta2:loxP-BFP-Stop-loxP-dsRed zebrafish shows this bud to be composed of a core of Cre-converted dsRed+ neural crest-derived cells ensheathed by unconverted BFP+ epithelia (Figure 1—figure supplement 1a). The position of this bud corresponds to kdrl:mCherry labeling of a branch of the first aortic arch that likely gives rise to the ophthalmic artery (Figure 1—figure supplement 1b). At 17 dpf, the pseudobranch is composed of five distinct filaments that resemble the primary gill filaments, with the five filaments merging to form a single pseudobranch by adult stages (90 dpf) (Figure 1b). Alcian Blue staining reveals that the adult pseudobranch contains five cartilage rods, reflecting the five fused filaments, with this cartilage resembling the specialized filament cartilage seen in the gills (Figure 1d; Fabian et al., 2022).

Figure 1. The zebrafish pseudobranch derives from mandibular arch mesenchyme and first pouch epithelia.

(a), Schematic showing the pseudobranch (arrows), gill filaments (branched green structures) connected to gill bars (blue), teeth (purple), vasculature (pink), and jaw and jaw-support skeleton (gray). (b) Hematoxylin and Eosin-stained sections show emergence of the pseudobranch bud at 4 dpf (adapted from https://bio-atlas.psu.edu/zf/view.php?atlas=5&s=41), five filaments at 17 dpf (adapted from https://bio-atlas.psu.edu/zf/view.php?atlas=65&s=1738), and the fused pseudobranch at 90 dpf (adapted from https://bio-atlas.psu.edu/zf/view.php?atlas=29&s=312). (c) Dissected adult pseudobranch shows the ophthalmic artery connecting it to the eye. (d) Alcian staining shows five cartilage rods in the pseudobranch and similar cartilage in gill primary filaments. (e) Photoconverted kikGR-expressing mesenchyme (red) from the dorsal first arch (numbered) at 1.5 dpf contributes to the palatoquadrate cartilage (pq) and pseudobranch mesenchyme (arrow) at 3.5 dpf. Photoconverted dorsal second arch cells do not contribute to the pseudobranch. In green, fli1a:GFP labels the vasculature and neural crest-derived mesenchyme, with mesenchyme also labeled by unconverted sox10:kikGR. (f) In fgf10:nEOS embryos, photoconversion of first pouch endoderm (numbered) at 1.5 dpf labels the pseudobranch epithelium (arrow) at 5 dpf. n numbers denote experimental replicates in which similar contributions were observed. Scale bars, 50 µm.

Figure 1.

Figure 1—figure supplement 1. Development of zebrafish pseudobranch and lineage analysis of gill filament epithelia.

Figure 1—figure supplement 1.

(a) In Sox10:Cre; acta2:loxP-BFP-Stop-loxP-dsRed fish at 5 dpf, the developing pseudobranch (white arrow) and gill buds (yellow arrow) consist of Cre-converted dsRed+ neural crest-derived mesenchyme (magenta) and unconverted BFP+ epithelia (gray). (b) At 6 dpf, kdrl:mCherry labeling of vasculature reveals a branch of the first aortic arch in the position of the pseudobranch (white arrow), and branches of the posterior aortic arches in the positions of the gills (yellow arrow). (c) Endoderm is labeled in red by adding 4OH-tamoxifen to sox17:CreERT2; ubb:loxP-Stop-loxP-mCherry fish at 6.5 hr post-fertilization to induce Cre recombination that removes the Stop cassette and allows mCherry expression (mCherry channel alone shown in inset). Co-localization shows fgf10b:nEOS expression (green) in endodermal pouches. (d), Endoderm labeling by addition of 4OH-tamoxifen to sox17:CreERT2; ubb:loxP-Stop-loxP-mCherry fish at 6.5 hr post-fertilization results in contribution to cdh1:mlanYFP+ pseudobranch epithelium at 5 dpf. (e,f) In fgf10:nEOS embryos, photoconversion of first pouch endoderm (and some more ventral mandibular cells) at 1.5 dpf labels pseudobranch (white arrow) but not gill epithelia (yellow arrow) at 5 dpf in (e), and conversion of third pouch cells labels the first gill filament epithelium (yellow arrow, boxed region magnified to right and shown in merged and red-only channels) in (f). Scale bars, 50 µM.
Open in a new tab

To determine from which arch the pseudobranch arises, we performed short-term lineage tracing using a photoconvertible sox10:kikGR reporter expressed in neural crest-derived mesenchyme. Photoconversion of dorsal first arch mesenchyme at 1.5 dpf labeled the pseudobranch mesenchymal bud at 3.5 dpf, as well as the palatoquadrate cartilage, a known first arch derivative; photoconversion of dorsal second arch mesenchyme did not label the pseudobranch (Figure 1e). To trace the epithelial origins of the pseudobranch, we performed short-term lineage tracing using fgf10b:nEOS, in which the photoconvertible nuclear-EOS protein is expressed in endodermal pouch epithelia (Figure 1—figure supplement 1c). Photoconversion of first pouch endoderm at 1.5 dpf labeled pseudobranch epithelia at 5 dpf (Figure 1f; Figure 1—figure supplement 1e), similar to labeling of first gill filament epithelia after photoconversion of third pouch endoderm (Figure 1—figure supplement 1f). We also confirmed endodermal origin of cdh1:mlanYFP+ pseudobranch epithelia by 4OH-tamoxifen-mediated conversion of early endoderm in sox17:CreERT2; ubb:loxP-Stop-loxP-mCherry zebrafish (Figure 1—figure supplement 1d). The pseudobranch therefore arises from mandibular arch neural crest-derived mesenchyme and first pouch endodermal epithelia.

In skate, the pseudobranch and gills share expression of foxl2, shh, gata3, and gcm2 (Hirschberger and Gillis, 2022). To test whether this reflects shared gene regulatory mechanisms indicative of serial homology, we examined activity of several gill-specific enhancers (Fabian et al., 2022). At 5 dpf, the gata3-p1 enhancer drives GFP expression in the growing tips of both the pseudobranch and gill buds (Figure 2a). At 14 dpf, the ucmaa-p1 enhancer, active in gill filament but not hyaline cartilage in the face, drives GFP expression in both pseudobranch and gill filament cartilage (Figure 2b), as seen for endogenous expression of ucmaa (Figure 3—figure supplement 1a). In our single-cell chromatin accessibility analysis of neural crest-derived cells (Fabian et al., 2022), we also identified an irx5a proximal enhancer selectively accessible in pillar cells, a specialized type of endothelial cell in the gill secondary filaments (Figure 3—figure supplement 2a). At 13, 20, and 60 dpf and one-year-old adult fish, the irx5a-p1 enhancer drives GFP expression in pillar cells of the pseudobranch and gills (Figure 2c; Figure 3—figure supplement 2b, c). These findings show that cells with similar gene expression and cis-regulatory architecture are present in both the pseudobranch and gills of zebrafish.

Figure 2. Shared regulatory program for pseudobranch and gill development.

Figure 2.

Open in a new tab

(a-c) In the pseudobranch (white arrows) and gill filaments (yellow arrows), gata3-p1:GFP labels growing buds, ucmaa-p1:GFP labels cellular cartilage (distinct from hyaline cartilage, arrowhead), and irx5a-p1:GFP labels pillar cells. sox10:dsRed labels cartilage for reference. Images in (b) and (c) are confocal projections, with magnified regions shown below in single sections for gata3-p1:GFP and ucmaa-p1:GFP. Scale bars, 50 µM.

Zebrafish mutant for gata3 fail to form gill buds (Sheehan-Rooney et al., 2013), and single-cell chromatin accessibility analysis of neural crest-derived cells had implicated gata3 and gata2a in development of gill filament cell type differentiation (Fabian et al., 2022). We find that gata3 and gata2a are prominently expressed in both the developing pseudobranch and gill buds at 3 and 5 dpf (Figure 3a; Figure 3—figure supplement 1b, c). The pseudobranch is also much reduced in gata3 mutants at 5 dpf, with fewer neural crest-derived cells labeled by Sox10:Cre; acta2:loxP-BFP-Stop-loxP-dsRed or gata3-p1:GFP contributing to both the pseudobranch and gills (Figure 3b and c). Similar genetic dependency of the pseudobranch and gills further supports serial homology.

Figure 3. Pseudobranch and gill development requires gata3 function.

(a) Similar expression of gata3 and gata2a in developing pseudobranch (white arrows) and gill regions (yellow arrows). (b) Sox10:Cre; acta2:loxP-BFP-Stop-loxP-dsRed labels Cre-converted dsRed+ neural crest-derived mesenchyme (magenta) and unconverted BFP+ epithelia (gray). (c) gata3-p1:GFP labels pseudobranch and gill filament buds, and sox10:dsRed labels cartilage. For both (b) and (c), 3/3 gata3 mutants displayed reduced formation of the pseudobranch (white arrows) and gill filaments (yellow arrows), compared to 3 controls each. Scale bars, 50 µM.

Figure 3.

Figure 3—figure supplement 1. Pseudobranch shares gene expression with gill filaments.

Figure 3—figure supplement 1.

(a) Expression of ucmaa in pseudobranch cartilage from one-year-old fish. (b,c) Developing pseudobranch (white arrows) and gill buds (yellow arrows) express gata3 and gata2a at 5 dpf. DAPI labels nuclei in blue (a) or white (b,c). Scale bars, 50 µM.

Figure 3—figure supplement 2. The pseudobranch shares irx5a-p1 pillar cell enhancer activity with gill filaments.

Figure 3—figure supplement 2.

(a) An intergenic irx5a-p1 region displays accessible chromatin specifically in pillar cells at 60 dpf. (b) irx5a-p1 drives GFP expression in pillar cells of the developing gills (yellow arrows) and pseudobranch (insets). (c,d) Dissected pseudobranch from a 60 dpf irx5a-p1:GFP adult shows GFP-positive pillar cells adjacent to a core of filament cartilage. The boxed region magnified below depicts an individual pillar cell (white arrow) flanked by characteristic lacunae (yellow arrows). Cartilage is labeled by sox10:dsRed in (b) and (c). Scale bars, 50 µM.
Open in a new tab

Discussion

Our findings that both a cartilaginous and teleost fish have a mandibular gill-like pseudobranch suggest that the last common ancestor of jawed vertebrates did so as well, thus providing plausibility to the model that jaws evolved from a gill-bearing mandibular arch. The absence of a pseudobranch in extant agnathans (i.e. lamprey and hagfish) (Cole, 1905; Mallatt, 1996) suggests that either the pseudobranch arose along the gnathostome stem (i.e. prior to the divergence of cartilaginous and bony fishes), or that it was an ancestral feature of vertebrates that has been lost in cyclostomes. The latter would be analogous to loss of the hyoid hemibranch gill during teleost fish evolution (Goodrich, 1930; Jollie, 1962), consistent with our failure to observe gill filament gene expression or transgene activity in the hyoid arch of zebrafish.

Whereas our data clearly point to the filament systems of the pseudobranch and gills being serially homologous, the major skeletal bars supporting the jaws and gills (not to be confused with the gill filament cartilage) appear to develop largely independently from the filaments. Unlike the gill filaments, the zebrafish pseudobranch is not attached to a major skeletal bar. Conversely, the skeletal bars derived from the seventh arch of zebrafish lack gill filaments and instead anchor pharyngeal teeth, and no gill filaments were observed with the fossilized rostral-most gill bar of M. walcotti (Morris and Caron, 2014). In addition, gata3 loss affects the pseudobranch and gill filaments but not the gill bars (Sheehan-Rooney et al., 2013). It is therefore possible that, rather than the pseudobranch evolving from an ancestral mandibular gill whose gill bar was transformed into the jaw skeleton, the pseudobranch arose independently after appearance of the jaw by co-option of a gill filament developmental program. While we demonstrate gill-like developmental potential of the mandibular arch in extant vertebrates, whether an ancestral mandibular gill bar gave rise to vertebrate jaws awaits more definitive fossil evidence.

Materials and methods

Key resources table.

Reagent type (species) or resource Designation Source or reference Identifiers Additional information
Gene
(Danio rerio)		 ucmaa   Ensembl: ENSDARG00000027799  
Gene
(Danio rerio)		 gata3   Ensembl: ENSDARG00000016526  
Gene
(Danio rerio)		 gata2a   Ensembl: ENSDARG00000059327  
Gene
(Danio rerio)		 irx5a   Ensembl: ENSDARG00000034043  
Genetic reagent (Danio rerio) Tübingen ZIRC RRID:ZIRC_ZL57 Wildtype strain of zebrafish
Genetic reagent (Danio rerio) Tg(fli1a:eGFP)y1 Lawson and Weinstein, 2002
Genetic reagent (Danio rerio) Tg(sox10:kikGR)el2 Balczerski et al., 2012    
Genetic reagent (Danio rerio) Tg(ucmaa_p1:GFP, cryaa:Cerulean)el851 Fabian et al., 2022    
Genetic reagent (Danio rerio) Tg(gata3_p1:GFP, cryaa:Cerulean)el858 Fabian et al., 2022    
Genetic reagent (Danio rerio) Tg(fgf10b:nEOS)el865 Fabian et al., 2022    
Genetic reagent (Danio rerio) Tg(–3.5ubb:loxP-STOP-loxP-mCherry)el818 Fabian et al., 2020    
Genetic reagent (Danio rerio) Tg(Mmu.Sox10-Mmu.Fos:Cre)zf384 Kague et al., 2012    
Genetic reagent (Danio rerio) Tg(actab2:loxP-BFP-STOP-loxP-dsRed)sd27 Kobayashi et al., 2014
Genetic reagent (Danio rerio) Tg(−6.5kdrl:mCherry)ci5 Proulx et al., 2010    
Genetic reagent (Danio rerio) Tg(–5.0sox17:Cre-ERT2,myl7:DsRed)sid1Tg Hockman et al., 2017    
Genetic reagent (Danio rerio) Tg(cdh1:mlanYFP)xt17Tg Cronan and Tobin, 2019    
Genetic reagent (Danio rerio) gata3b1075 Sheehan-Rooney et al., 2013    
Genetic reagent (Danio rerio) Tg(irx5a-p1:GFP, cryaa:Cerulean)el859 This paper   See Materials and Methods, Section Zebrafish Lines
Recombinant DNA reagent PCS2FA-transposase Tol2Kit PUBMED: 17937395
396.pCS2-transposase		  
Recombinant DNA reagent pDestTol2AB2-irx5a-p1-E1B:GFP_pA This paper   See Materials and Methods, Section Zebrafish Lines
Sequence-based reagent ucmaa RNAScope probe (Danio rerio); Channel 1 ACD Bio
Sequence-based reagent gata2a RNAScope probe (Danio rerio); Channel 1 ACD Bio    
Sequence-based reagent gata3 RNAScope probe (Danio rerio); Channel 2 ACD Bio    
Commercial assay or kit In-Fusion HD Cloning Plus Takara Takara:638,910
Commercial assay or kit RNAScope Multiplex Fluorescent v2 Assay ACD Bio ACD Bio:323,100  
Other Draq5 nuclear dye Abcam Abcam:Ab108410 See Materials and Methods, Section Imaging
Open in a new tab

Zebrafish lines

The Institutional Animal Care and Use Committee of the University of Southern California approved all animal experiments (Protocol 20771). Zebrafish lines include Tg(fli1a:eGFP)y1 (Lawson and Weinstein, 2002); Tg(–4.9sox10:kikGR)el2 (Balczerski et al., 2012); Tg(ucmaa_p1:GFP, cryaa:Cerulean)el851, Tg(gata3_p1:GFP, cryaa:Cerulean)el858, and Tg(fgf10b:nEOS)el865 (Fabian et al., 2022); Tg(–5.0sox17:Cre-ERT2,myl7:DsRed)sid1Tg (Hockman et al., 2017); Tg(cdh1:mlanYFP)xt17Tg (Cronan and Tobin, 2019); Tg(–3.5ubb:loxP-STOP-loxP-mCherry)el818 (Fabian et al., 2020); Tg(Mmu.Sox10-Mmu.Fos:Cre)zf384 (Kague et al., 2012); Tg(actab2:loxP-BFP-STOP-loxP-dsRed)sd27 (Kobayashi et al., 2014); Tg(−6.5kdrl:mCherry)ci5 (Proulx et al., 2010); and gata3b1075 (Sheehan-Rooney et al., 2013). To generate Tg(irx5a-p1:GFP, cryaa:Cerulean)el859, we synthesized the intergenic peak associated with irx5a (chr7:35838071–35838577) using iDT gBlocks, cloned it into a modified pDestTol2AB2 construct containing the E1b minimal promoter, GFP, polyA, and the lens-specific cryaa:Cerulean marker using in-Fusion cloning (Takara Bio). We injected plasmid and Tol2 transposase RNA (5–10 ng/µL each) into one-cell stage zebrafish embryos and screened for founders at adulthood based on lens CFP expression in progeny. Two independent germline founders were identified that showed activity in gill pillar cells.

Histology

Adult fish were fixed in 4% paraformaldehyde for 1 hr at 25°C followed by dissection of the gills and further fixation in 4% paraformaldehyde for 1 hr at 25°C. For pseudobranch dissection, adults were fixed in 4% paraformaldehyde at 4°C for 3 days prior to dissection. Alcian Blue staining was performed on whole tissue as previously described (Paul et al., 2016). Samples were imaged using a Leica DM2500 microscope. Image levels were adjusted in Adobe Illustrator.

Photoconversion-based lineage tracing

To photoconvert mesenchyme in sox10:kikGR; fli1a:GFP fish at 1.5 dpf, we used the ROI function in ZEN software on a Zeiss LSM800 confocal microscope to expose dorsal first or second arch mesenchyme to UV light for 20 s. Imaging confirmed successful and specific photoconversion of kikGR from green to red fluorescence in the intended region. At 3.5 dpf, confocal imaging was used to assess contribution of photoconverted cells to pseudobranch mesenchyme. We included fli1a:GFP to help in identification of the pseudobranch bud. For fgf10b:nEOS photoconversion, we used the ROI function to expose nEOS-high expressing cells in the first or third pharyngeal pouches to UV light for 60 s, with immediate confocal imaging confirming intended photoconversion of nEOS from green to red fluorescence. At 5 dpf, confocal imaging was used to assess contribution of photoconverted cells to pseudobranch and gill epithelia. To confirm that nEOS-high expressing cells in the fgf10b:nEOS line were of endodermal original, we crossed these onto the sox17:CreERT2; ubb:loxP-Stop-loxP-mCherry transgenic background and treated embryos with 4-hydroxytamoxifen (Sigma) at 6.5 hpf to induce Cre recombination. We then imaged on the confocal microscope at 1.5 dpf to visualize co-localization of nEOS and mCherry. All results were independently confirmed in at least three animals.

In situ hybridization

We performed in situ hybridization on whole embryos at 3 and 5 dpf and on paraffin sections from adult zebrafish heads using RNAscope probes synthesized by Advanced Cell Diagnostics in channel 1 (ucmaa, gata2a) and channel 2 (gata3). Samples were prepared by fixation in 4% paraformaldehyde overnight. Embryos were dehydrated in methanol and stored overnight before proceeding with the RNAScope Assay for Whole Zebrafish Embryos as described in the manufacturer’s protocols. Following fixation, the pseudobranch was dissected and mounted in 0.2% agarose in molds. Once solidified, agarose chips containing the pseudobranch were cut out of the mold, dehydrated, embedded in paraffin, and 5 µm sections were collected using a Shandon Finesse Me+ microtome (cat. no. 77500102). Paraformaldehyde-fixed paraffin-embedded sections were deparaffinized, and the RNAscope Fluorescent Multiplex V2 Assay was performed according to manufacturer’s protocols using an ACD HybEZ Hybridization oven. In situ patterns were confirmed in at least three independent animals, with exception of the ucmaa in situ that was performed on three separate sections of the same animal.

Imaging

Images of whole-mount or section fluorescent in situ hybridizations and live transgenic fish were captured on a Zeiss LSM800 confocal microscope using ZEN software. For adult imaging of the irx5a-p1:GFP reporter, whole animals were euthanized and the pseudobranch and gills dissected out. The tissue was stained with Draq5 nuclear dye (Abcam) for 20 min to help identify pillar cells. Reported expression patterns for enhancer lines were confirmed in at least five animals.

Acknowledgements

We thank Megan Matsutani for fish care, Johann Eberhart and Mary Swartz for the gata3 mutant fish, and Keith Cheng, Jean Copper, and Daniel Vanselow for providing the original images related to Figure 1b. We used https://biorender.com/ to create Figure 1a.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

J Gage Crump, Email: gcrump@usc.edu.

Carole LaBonne, Northwestern University, United States.

Marianne E Bronner, California Institute of Technology, United States.

Funding Information

This paper was supported by the following grants:

  • National Institute of Dental and Craniofacial Research 5R35DE027550 to J Gage Crump.

  • National Institute of Dental and Craniofacial Research 5K99DE029858 to Peter Fabian.

  • National Institute of Dental and Craniofacial Research 1F31DE030706 to Mathi Thiruppathy.

  • Royal Society UF130182 to J Andrew Gillis.

  • Royal Society URF\R\191007 to J Andrew Gillis.

Additional information

Competing interests

No competing interests declared.

No competing interests declared.

Author contributions

Conceptualization, Formal analysis, Investigation, Methodology, Writing - original draft, Writing – review and editing.

Investigation.

Conceptualization, Writing – review and editing.

Formal analysis, Funding acquisition, Supervision, Writing - original draft, Writing – review and editing.

Ethics

This study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The Institutional Animal Care and Use Committee of the University of Southern California approved all animal experiments (Protocol 20771).

Additional files

MDAR checklist

Data availability

The current manuscript contains solely images, so no data have been generated for this manuscript. The n number for each image type is clearly stated in the manuscript.

The following previously published dataset was used:

Crump G, Fabian P, Tseng K, Chen H. 2021. Single-cell profiling of cranial neural crest diversification across a vertebrate lifetime. NCBI Gene Expression Omnibus. GSE178969

References

  1. Balczerski B, Matsutani M, Castillo P, Osborne N, Stainier DYR, Crump JG. Analysis of sphingosine-1-phosphate signaling mutants reveals endodermal requirements for the growth but not dorsoventral patterning of jaw skeletal precursors. Developmental Biology. 2012;362:230–241. doi: 10.1016/j.ydbio.2011.12.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Cole FJ. A Monograph on the General Morphology of the Myxinoid Fishes Based on A Study of Myxine. Edinburgh: Robert Grant; 1905. [Google Scholar]
  3. Cronan MR, Tobin DM. Endogenous Tagging at the cdh1 Locus for Live Visualization of E-Cadherin Dynamics. Zebrafish. 2019;16:324–325. doi: 10.1089/zeb.2019.1746. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Dohrn A. Studien Zur Urgeschichte Des Wirbelthierkörpers. Neapel: s.n; 1882. [Google Scholar]
  5. Fabian P, Tseng KC, Smeeton J, Lancman JJ, Dong PDS, Cerny R, Crump JG. Lineage analysis reveals an endodermal contribution to the vertebrate pituitary. Science. 2020;370:463–467. doi: 10.1126/science.aba4767. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Fabian P, Tseng KC, Thiruppathy M, Arata C, Chen HJ, Smeeton J, Nelson N, Crump JG. Lifelong single-cell profiling of cranial neural crest diversification in zebrafish. Nature Communications. 2022;13:13. doi: 10.1038/s41467-021-27594-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Gaskell WH. The Origin of Vertebrates. London: Longmans, Green, and co; 1908. [DOI] [Google Scholar]
  8. Gegenbaur C, Bell FJ, Lankester ER. Elements of Comparative Anatomy. London: Macmillan and Co; 1878. [DOI] [Google Scholar]
  9. Goodrich ES. Studies on the Structure & Development of Vertebrates. London: Macmillan and co., limited; 1930. [DOI] [Google Scholar]
  10. Hirschberger C, Gillis JA. The pseudobranch of jawed vertebrates is a mandibular arch-derived gill. Development (Cambridge, England) 2022 doi: 10.1242/dev.200184. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Hockman D, Burns AJ, Schlosser G, Gates KP, Jevans B, Mongera A, Fisher S, Unlu G, Knapik EW, Kaufman CK, Mosimann C, Zon LI, Lancman JJ, Dong PDS, Lickert H, Tucker AS, Baker CV. Evolution of the hypoxia-sensitive cells involved in amniote respiratory reflexes. eLife. 2017;6:e21231. doi: 10.7554/eLife.21231. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Jollie M. Chordate Morphology. New York: Reinhold; 1962. [DOI] [Google Scholar]
  13. Kague E, Gallagher M, Burke S, Parsons M, Franz-Odendaal T, Fisher S. Skeletogenic fate of zebrafish cranial and trunk neural crest. PLOS ONE. 2012;7:e47394. doi: 10.1371/journal.pone.0047394. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kobayashi I, Kobayashi-Sun J, Kim AD, Pouget C, Fujita N, Suda T, Traver D. Jam1a-Jam2a interactions regulate haematopoietic stem cell fate through Notch signalling. Nature. 2014;512:319–323. doi: 10.1038/nature13623. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Laurent P, Dunel-Erb S. The Pseudobranch: Morphology and Function. Fish Physiology. 1984;10:285–323. doi: 10.1016/S1546-5098(08)60188-0. [DOI] [Google Scholar]
  16. Lawson ND, Weinstein BM. In vivo imaging of embryonic vascular development using transgenic zebrafish. Developmental Biology. 2002;248:307–318. doi: 10.1006/dbio.2002.0711. [DOI] [PubMed] [Google Scholar]
  17. Mallatt J. Ventilation and the origin of jawed vertebrates: a new mouth. Zoological Journal of the Linnean Society. 1996;117:329–404. doi: 10.1111/j.1096-3642.1996.tb01658.x. [DOI] [Google Scholar]
  18. Miyashita T. Fishing for jaws in early vertebrate evolution: a new hypothesis of mandibular confinement. Biological Reviews of the Cambridge Philosophical Society. 2016;91:611–657. doi: 10.1111/brv.12187. [DOI] [PubMed] [Google Scholar]
  19. Mongera A, Singh AP, Levesque MP, Chen YY, Konstantinidis P, Nüsslein-Volhard C. Genetic lineage labeling in zebrafish uncovers novel neural crest contributions to the head, including gill pillar cells. Development. 2013;140:916–925. doi: 10.1242/dev.091066. [DOI] [PubMed] [Google Scholar]
  20. Morris SC, Caron JB. A primitive fish from the Cambrian of North America. Nature. 2014;512:419–422. doi: 10.1038/nature13414. [DOI] [PubMed] [Google Scholar]
  21. Paul S, Schindler S, Giovannone D, de Millo Terrazzani A, Mariani FV, Crump JG. Ihha induces hybrid cartilage-bone cells during zebrafish jawbone regeneration. Development. 2016;143:2066–2076. doi: 10.1242/dev.131292. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Proulx K, Lu A, Sumanas S. Cranial vasculature in zebrafish forms by angioblast cluster-derived angiogenesis. Developmental Biology. 2010;348:34–46. doi: 10.1016/j.ydbio.2010.08.036. [DOI] [PubMed] [Google Scholar]
  23. Sheehan-Rooney K, Swartz ME, Zhao F, Liu D, Eberhart JK. Ahsa1 and Hsp90 activity confers more severe craniofacial phenotypes in a zebrafish model of hypoparathyroidism, sensorineural deafness and renal dysplasia (HDR. Disease Models & Mechanisms. 2013;6:1285–1291. doi: 10.1242/dmm.011965. [DOI] [PMC free article] [PubMed] [Google Scholar]

Editor's evaluation

Carole LaBonne 1

This is an interesting and important paper that investigates pseudobranch development in zebrafish in the context of seeking evidence for a proposed gill arch origin for the vertebrate jaw. It provides data that supports that the pseudobranch is derived from the mandibular arch and that the pseudobranch is a segmental homolog of the gills providing strong support for the classic gills-to-jaws hypothesis.

Decision letter

Editor: Carole LaBonne1
Reviewed by: Craig T Miller2

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Gill developmental program in the teleost mandibular arch" for consideration by eLife. Your article has been reviewed by 2 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Marianne Bronner as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Craig T Miller (Reviewer #2).

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

In this manuscript, the authors investigate pseudobranch development in zebrafish in the context of seeking evidence for a proposed gill arch origin for the vertebrate jaw. They provide data that supports that the pseudobranch is derived from the mandibular arch and that the pseudobranch is a segmental homolog of the gills. The reviewers agree that this work is potentially appropriate for publication in eLife if certain concerns can be addressed.

1. The first arch kikGR fate mapping example in Figure 1e is of poor resolution and it's difficult to see the cellular composition of the forming pseudobranch. (by contrast, the Sox10Cre:b-actin:BFP)

2. Similarly for Figure 3S2b (the 13 dpf irx5a-p1 figure) is that the best representative figure? This is a weak data point with only a single blip of GFP positive signal which may or may not even be a cell.

3. The fgf10b:nEOS does not appear to be truly endodermal specific. Thus, it's not clear whether the magenta label in Figure 1f came from the first pouch as claimed, or alternatively from non-endodermal fgf10b expressing cells. Do the authors have data showing sox17:CreER; ubb:stop-mCherry tamoxifen labeled magenta cells in the pseudobranch? Alternatively, can the authors show an example of photoconverting nEOS in the fgf10b:nEOS line that is more localized onto the first pharyngeal pouch (the example in Figure 1F appears to include photoconverted cells both posterior but especially more ventral than the first pharyngeal pouch).

4. It seems that in this group's recent 2022 Nat. Comm. scRNA-seq paper, pseudobranch cells might also have been sequenced. Can the authors ask whether pseudobranch cells in their existing scRNA-seq data cluster with different gill cell clusters? E.g. perhaps gill cells are hox+ (and gata3+, ucmaa+, or irx5a+) while pseudobranch cells are hox- (and gata3+, ucmaa+, or irx5a+)? Perhaps the sequencing depth is not deep enough to answer this question, or perhaps the authors have no markers that distinguish pseudobranch cells from gill cells and this approach would not work. But it seems possible that the scRNA-seq data could provide additional evidence for serial homology of the pseudobranch and gills.

5. Are there gill-like structures derived from the second arch in teleosts? The authors mention the hemibranch derived from the hyoid arch in cartilaginous fishes. Is there a similar structure in any teleost? If there are no arch 2 gill-like structures, how this fits into the gill origin of jaw hypothesis should be discussed. Also, there is no discussion of jawless fish and how the considerable literature from species such as lamprey relates to the findings presented here. Do agnathans develop a pseudobranch? Addressing both of these points would help a general audience.

6. Better care should be taken with the word homology, and the distinction between serial homology vs historical or functional homology should be made clear. It would be helpful to define for a general audience what is meant by serial homology.

7. The first sentence of the discussion is not fully supported by the data. More species need to be included to make this claim. This statement should be tempered.

Other points:

In the figures, the font size of text and figure panels should be increased throughout for clarity.

In Figure 1A, the jaw and hyoid skeleton is cartooned in gray, not just the hyoid, so in the Figure 1A legend, "jaw support skeleton" should be more accurately changed to "jaw and jaw support skeleton". The uninterrupted medial gray bar should be removed from this diagram as it's neither anatomically accurate nor relevant for the images/data in this paper. Likewise, the gray blob in front of the pterygoid process should either be removed or extended posteriorly to more accurately represent the neurocranium.

In the Figure 1D legend, one of the two "in green" should be removed from this sentence: "In green, fli1a:GFP labels the vasculature and neural crest-derived mesenchyme for reference in green, and unconverted sox10:kikGR cells also labels mesenchyme."

The Figure 1F legend is mislabeled as "e".

The Figure 3A legend should define the white arrow and the yellow arrow separately.

The Figure 1S1D legend indicates that pouches are numbered, but they are not numbered in the figure. Either numbers should be added or the legend corrected.

In Figure 3S1, labeling the gill buds with yellow arrows as in the other figures would be helpful and make the presentation more consistent.

p. 4, Reference(s) should be provided to support this claim: "The pseudobranch is an epithelial structure located just behind the eye of most fishes that regulates ocular blood pressure."

p. 7 "Gata3 and Gata2a" to "gata3 and gata2a (italicized)"?

Reviewer #2 (Recommendations for the authors):

Suggestions to strengthen the science are listed below:

1. The first arch kikGR fate mapping example in Figure 1e is of lowish resolution and difficult to see the cellular composition of the forming pseudobranch. It is listed that an n of 6 was done here. In contrast, the example shown in Figure 1S1A with the Sox10Cre:b-actin:BFP

2. The fgf10b:nEOS does not appear to be truly endodermal specific. Thus, it's not clear whether the magenta label in Figure 1f came from the first pouch as claimed, or alternatively from non-endodermal fgf10b expressing cells. Do the authors have data showing sox17:CreER; ubb:stop-mCherry tamoxifen labeled magenta cells in the pseudobranch? Alternatively, can the authors show an example of photoconverting nEOS in the fgf10b:nEOS line that is more localized onto the first pharyngeal pouch (the example in Figure 1F appears to include photoconverted cells both posterior but especially more ventral than the first pharyngeal pouch).

3. It seems that in this group's recent 2022 Nat. Comm. scRNA-seq paper, pseudobranch cells might also have been sequenced. Can the authors ask whether pseudobranch cells in their existing scRNA-seq data cluster with different gill cell clusters? E.g. perhaps gill cells are hox+ (and gata3+, ucmaa+, or irx5a+) while pseudobranch cells are hox- (and gata3+, ucmaa+, or irx5a+)? Perhaps the sequencing depth is not deep enough to answer this question, or perhaps the authors have no markers that distinguish pseudobranch cells from gill cells and this approach would not work. But it seems possible that the scRNA-seq data could provide additional evidence for serial homology of the pseudobranch and gills.

Suggestions to improve the manuscript:

4. In the figures, the font size of text and figure panels should be increased throughout for clarity.

5. In Figure 1A, the jaw and hyoid skeleton is cartooned in gray, not just the hyoid, so in the Figure 1A legend, "jaw support skeleton" should be more accurately changed to "jaw and jaw support skeleton". The uninterrupted medial gray bar should be removed from this diagram as it's neither anatomically accurate nor relevant for the images/data in this paper. Likewise, the gray blob in front of the pterygoid process should either be removed or extended posteriorly to more accurately represent the neurocranium.

6. In the Figure 1D legend, one of the two "in green" should be removed from this sentence: "In green, fli1a:GFP labels the vasculature and neural crest-derived mesenchyme for reference in green, and unconverted sox10:kikGR cells also labels mesenchyme."

7. The Figure 1F legend is mislabeled as "e".

8. The Figure 3A legend should define the white arrow and the yellow arrow separately.

9. The Figure 1S1D legend indicates that pouches are numbered, but they are not numbered in the figure. Either numbers should be added or the legend corrected.

10. In Figure 3S1, labeling the gill buds with yellow arrows as in the other figures would be helpful and make the presentation more consistent.

11. p. 4, Reference(s) should be provided to support this claim: "The pseudobranch is an epithelial structure located just behind the eye of most fishes that regulates ocular blood pressure."

12. p. 6, "Zebrafish mutant for gata3 fails to form gill buds": fails to fail, as zebrafish is plural here?

13. p. 7 "Gata3 and Gata2a" to "gata3 and gata2a (italicized)"?

14. p. 7 "…jaws evolved from a mandibular gill." Change to "…from a mandibular gill support", "…from a mandibular segment containing a gill", "…from a gill-bearing mandibular arch"? I don't think of a gill as containing the skeletal support element, so the wording as-is will be confusing to some.

No issues with the availability of data or reagents were identified.

eLife. 2022 Jun 28;11:e78170. doi: 10.7554/eLife.78170.sa2

Author response

1. The first arch kikGR fate mapping example in Figure 1e is of poor resolution and it's difficult to see the cellular composition of the forming pseudobranch. (by contrast, the Sox10Cre:b-actin:BFP)

We have repeated the sox10:kikGR imaging at better resolution and now show clearer data in Figure 1e. This has also led to an increase in n number for both first and second arch conversion.

2. Similarly for Figure 3S2b (the 13 dpf irx5a-p1 figure) is that the best representative figure? This is a weak data point with only a single blip of GFP positive signal which may or may not even be a cell.

The purpose of including an image of the irx5a-p1:GFP transgenic line at 13 dpf is to highlight the very earliest timepoint at which we see signal in the pseudobranch. While it is true that there are very few GFP+ cells at this stage, this is consistent across multiple animals. We now include a second example from this stage with slightly stronger contribution in Figure 3—figure supplement 2b. As can be seen in Figure 2c and Figure 3—figure supplement 2c, the number of irx5a-p1:GFP+ cells in the pseudobranch greatly increases by late juvenile and adult stages.

3. The fgf10b:nEOS does not appear to be truly endodermal specific. Thus, it's not clear whether the magenta label in Figure 1f came from the first pouch as claimed, or alternatively from non-endodermal fgf10b expressing cells. Do the authors have data showing sox17:CreER; ubb:stop-mCherry tamoxifen labeled magenta cells in the pseudobranch? Alternatively, can the authors show an example of photoconverting nEOS in the fgf10b:nEOS line that is more localized onto the first pharyngeal pouch (the example in Figure 1F appears to include photoconverted cells both posterior but especially more ventral than the first pharyngeal pouch).

We performed a lineage trace of sox17:CreERT2-labelled endoderm and show contribution to both gill and pseudobranch cdh1:mlanYFP+ epithelia (Figure 1—figure supplement 1d). We also performed more specific photoconversion of fgf10b:nEOS in just the first pouch and show similar contribution to pseudobranch epithelia (new Figure 1f, old exampled moved to Figure 1figure supplement 1d). These new experiments now better support contribution of first pouch endoderm to the pseudobranch epithelium.

4. It seems that in this group's recent 2022 Nat. Comm. scRNA-seq paper, pseudobranch cells might also have been sequenced. Can the authors ask whether pseudobranch cells in their existing scRNA-seq data cluster with different gill cell clusters? E.g. perhaps gill cells are hox+ (and gata3+, ucmaa+, or irx5a+) while pseudobranch cells are hox- (and gata3+, ucmaa+, or irx5a+)? Perhaps the sequencing depth is not deep enough to answer this question, or perhaps the authors have no markers that distinguish pseudobranch cells from gill cells and this approach would not work. But it seems possible that the scRNA-seq data could provide additional evidence for serial homology of the pseudobranch and gills.

We thank the reviewer for this excellent suggestion. Unfortunately, we have analyzed our scRNAseq data at 60 dpf but find no clear segregation of hox+ and hox- cells within gill cell types (ncam3+ pillar cells, ucmaa+ gill filament chondrocytes, fgf10a+ gill progenitors). The only Hox genes we detect at this stage are hoxa3a and hoxb3a, yet these do not co-localize in any clear subset of gill cells. It may be that Hox expression has largely been extinguished during these later differentiation stages, and/or that gill cells greatly outnumber pseudobranch cells, thus making it difficult to detect a distinct hox- pseudobranch cluster.

Author response image 1.

Author response image 1.

Open in a new tab

5. Are there gill-like structures derived from the second arch in teleosts? The authors mention the hemibranch derived from the hyoid arch in cartilaginous fishes. Is there a similar structure in any teleost? If there are no arch 2 gill-like structures, how this fits into the gill origin of jaw hypothesis should be discussed. Also, there is no discussion of jawless fish and how the considerable literature from species such as lamprey relates to the findings presented here. Do agnathans develop a pseudobranch? Addressing both of these points would help a general audience.

We now cite references that a hyoid hemibranch is present in non-teleost but not teleost fishes, suggesting it was lost along the teleost lineage. This is consistent with our observed lack of gill filament gene expression and transgene activity in the zebrafish hyoid arch. We also now cite references for a hyoid hemibranch and more posterior gills in lamprey, yet a lack of pseudobranch in either lamprey or hagfish.

pp. 2-3: “The second (hyoid) arch also forms a hemibranch (one row of gill filaments) in the jawless lamprey fish (Dohrn, 1882; Gaskell, 1908), in cartilaginous and various non-teleost fishes (e.g. coelacanth, lungfishes, sturgeon, gar), but not in teleost fishes (Goodrich, 1930; Jollie, 1962).”

pp. 3: “However, extant agnathans (the cyclostomes lamprey and hagfish) lack a mandibular gill (Cole, 1905; Mallatt, 1996)…”

pp. 6-7: “The absence of a pseudobranch in extant agnathans (i.e. lamprey and hagfish) (Cole, 1905; Mallatt, 1996) suggests that either the pseudobranch arose later along the gnathostome lineage, or that it was lost along the cyclostome lineage. The latter would be analogous to loss of the hyoid hemibranch gill during teleost fish evolution (Goodrich, 1930; Jollie, 1962), consistent with our failure to observe gill filament gene expression or transgene activity in the hyoid arch of zebrafish.”

6. Better care should be taken with the word homology, and the distinction between serial homology vs historical or functional homology should be made clear. It would be helpful to define for a general audience what is meant by serial homology.

We now define serial homology in the Introduction and are consistent with use of “serial homology” throughout, including in the Discussion as suggested.

p.3: “Another important question is whether the pseudobranch and gills can be considered serially homologous, i.e. representing morphologically related structures that arise through shared developmental and genetic mechanisms.”

7. The first sentence of the discussion is not fully supported by the data. More species need to be included to make this claim. This statement should be tempered.

We have modified this sentence to stress that our conclusions are based on close examination of only two species.

P 6: “Our findings that both a cartilaginous and teleost fish have a mandibular gill-like pseudobranch suggest that the last common ancestor of jawed vertebrates did so as well, thus providing plausibility to the model that jaws evolved from a mandibular gill.”

Other points:

In the figures, the font size of text and figure panels should be increased throughout for clarity.

We have increased font size throughout the figures.

In Figure 1A, the jaw and hyoid skeleton is cartooned in gray, not just the hyoid, so in the Figure 1A legend, "jaw support skeleton" should be more accurately changed to "jaw and jaw support skeleton". The uninterrupted medial gray bar should be removed from this diagram as it's neither anatomically accurate nor relevant for the images/data in this paper. Likewise, the gray blob in front of the pterygoid process should either be removed or extended posteriorly to more accurately represent the neurocranium.

The Figure 1a schematic and legend have been changed as suggested.

In the Figure 1D legend, one of the two "in green" should be removed from this sentence: "In green, fli1a:GFP labels the vasculature and neural crest-derived mesenchyme for reference in green, and unconverted sox10:kikGR cells also labels mesenchyme."

Corrected.

The Figure 1F legend is mislabeled as "e".

Corrected.

The Figure 3A legend should define the white arrow and the yellow arrow separately.

Corrected.

The Figure 1S1D legend indicates that pouches are numbered, but they are not numbered in the figure. Either numbers should be added or the legend corrected.

We have added pouch numbers to the figures.

In Figure 3S1, labeling the gill buds with yellow arrows as in the other figures would be helpful and make the presentation more consistent.

Figures S2 and S3 have been modified to include yellow arrows indicating gill buds as suggested. We have also double-checked that each figure contains white arrows for pseudobranches and yellow arrows for gills.

p. 4, Reference(s) should be provided to support this claim: "The pseudobranch is an epithelial structure located just behind the eye of most fishes that regulates ocular blood pressure."

By examining Jollie, 1962 and other anatomical literature, it is apparent that the function of the pseudobranch is not completely resolved. We therefore edit the text to reflect the two current theories behind its function.

“The pseudobranch is an epithelial structure located just behind the eye that has been proposed to regulate ocular blood pressure and/or have an endocrine function (Jollie, 1962).”

p. 7 "Gata3 and Gata2a" to "gata3 and gata2a (italicized)"?

Corrected.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Crump G, Fabian P, Tseng K, Chen H. 2021. Single-cell profiling of cranial neural crest diversification across a vertebrate lifetime. NCBI Gene Expression Omnibus. GSE178969

    Supplementary Materials

    MDAR checklist
    elife-78170-mdarchecklist1.docx (100.8KB, docx)

    Data Availability Statement

    The current manuscript contains solely images, so no data have been generated for this manuscript. The n number for each image type is clearly stated in the manuscript.

    The following previously published dataset was used:

    Crump G, Fabian P, Tseng K, Chen H. 2021. Single-cell profiling of cranial neural crest diversification across a vertebrate lifetime. NCBI Gene Expression Omnibus. GSE178969

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES