henhaobio
Sign in

eLifeDevelopmental BiologyPeer-reviewed journal article

A high-resolution mRNA expression time course of embryonic development in zebrafish

Richard Joseph WhiteiD, John E CollinsiD, Ian M. SealyiD, Neha WaliiD, Christopher M. DooleyiD, Zsofia DigbyiD, Derek L. StempleiD, Daniel N. MurphyiD, Konstantinos BillisiD, Thibaut HourlieriD, Anja FüllgrabeiD, Matthew P. A. DavisiD, Anton James EnrightiD, Elisabeth M. Busch‐NentwichiD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

We have produced an mRNA expression time course of zebrafish development across 18 time points from 1-cell to 5 days post-fertilisation sampling individual and pools of embryos. Using poly(A) pulldown stranded RNA-seq and a 3' end transcript counting method we characterise the temporal expression profiles of 23,642 genes. We identify temporal and functional transcript co-variance that associates 5,024 unnamed genes with distinct developmental time points. Specifically, a class of over 100 previously uncharacterised zinc finger domain containing genes, located on the long arm of chromosome 4, is expressed in a sharp peak during zygotic genome activation. The data reveal complex and widespread differential use of exons and previously unidentified 3' ends across development, new primary microRNA transcripts and temporal divergence of gene paralogues generated in the teleost genome duplication. To make this dataset a useful baseline reference, the data are accessible to browse and download at Expression Atlas and Ensembl.

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 2 (20 Feb 2017) · CC BY 4.0

RNA-seq and 3′ end counting over 18 zebrafish stages (1-cell to day 5) profile 23,642 genes; over 100 chromosome 4 zinc finger genes peak at zygotic genome activation.

Zebrafish embryogenesis has rich morphological and phenotype data but lacked a matching systematic survey of molecular events, and earlier zebrafish baseline transcriptomes targeted specific questions (maternal-zygotic transition, transcript types, promoters, annotation) or sampled few time points and replicates. The dataset covers 18 stages with up to 29 biological replicates each and can be browsed in Expression Atlas and Ensembl without computational processing.

Key results

  1. ShownReplicates agree closely, expression changes smoothly from one stage to the next, and the number of detected genes rises over development.

    18 stages; RNA-seq on pools of 12 embryos (5 replicates), DeTCT on pools of 8 and on single embryos (11-12 replicates each). Mean reads per sample: 3.8 million (RNA-seq) and 9 million (DeTCT). Genes at ≥5 normalised counts rise from a mean of 10,395 to 18,225 (RNA-seq; 6,259 to 15,215 for DeTCT). Six principal components capture just over 90% of variance (58.2% for the first). · A high-resolution transcriptional profile of…

  2. ShownOver 100 zinc finger genes on the long arm of chromosome 4 are co-expressed in a sharp burst during zygotic genome activation. NLR family genes in the same region do not share this pattern.

    Binomial tests flagged four co-expression clusters (4, 22, 37, 84) with excess chromosome 4 genes (adjusted p < 0.05, log2 fold enrichment > 1), mostly zinc finger genes expressed in a burst spanning dome to 75% epiboly (4.3-8 hours post-fertilisation). Of 312 NLR family genes in Ensembl v85, 149 are expressed, with a different profile. · A burst of transcription of highly related zinc…; Discussion; Tables

  3. ShownGenes whose expression peaks at dome and gastrula stages are disproportionately uncharacterised. Zebrafish- or teleost-specific genes do not account for this excess.

    5,024 of 23,642 genes carry automatic names (zgc:*, si:ch*). They make up 29.2-39.4% of genes peaking at dome and gastrula stages vs 14.4-26.5% at other stages. The shares classed as Danio-only, Teleost or Vertebrate from Ensembl Compara orthologues stay broadly constant across all 18 stages. · GO enrichment; Discussion

  4. SuggestedParalogue pairs from the teleost-specific genome duplication tend to share temporal profiles, yet half correlate below 0.45. Most pairs with ZFIN anatomy annotation have largely non-overlapping annotated expression.

    Across 3144 expressed one-to-one paralogue pairs, Pearson correlations of normalised counts skew positive compared with random gene pairs, but 50% fall below 0.45. Of 683 pairs with ZFIN annotation for both genes, 83.7% (572 pairs) share less than 50% of their stage-anatomy annotations. · Conservation/Divergence of the expression…; Discussion

  5. SuggestedStrictly filtered 3′ end calls show alternative ends used at different stages. RNA-seq of drosha and dgcr8 mutants yields primary miRNA (pri-miRNA) transcripts for miRNA loci that Ensembl v87 leaves without one.

    From 423 DeTCT samples, 253,627 candidate regions fell to 37,724 after filtering and 8358 under stricter filters; 1551 genes have two or more ends (pdlim5b: one end early, three in segmentation). The assembly has 398 pri-miRNA transcripts over 283 loci, 94 lacking one in Ensembl, vs 313 over 220 in Ensembl v87 (a 42% gain in miRNAs with a primary transcript). · Alternative 3′ end use; Novel primary miRNA transcripts; Discussion

Methods, data and reproducibility

Zebrafish embryos at 18 stages (1-cell to 5 days post-fertilisation) were profiled by stranded poly(A) RNA-seq and DeTCT 3′ end counting. Genes were grouped by peak stage and by Markov Cluster Algorithm clustering of a correlation network, then tested for Gene Ontology and Zebrafish Anatomy Ontology enrichment. Primary miRNA transcripts were assembled from drosha and dgcr8 mutant RNA-seq.

  • Stranded poly(A) RNA-seq of pooled wild-type HLF strain embryos (12 per pool) at 18 stages, 1-cell to 5 days post-fertilisation · zebrafish (Danio rerio) · n = 5 replicate libraries per stage · ERP014517
  • DeTCT 3′ end sequencing of pools of 8 embryos and of individual embryos at the same 18 stages · zebrafish (Danio rerio) · n = 11-12 replicates per sample type per stage; 423 samples in total · ERP006948; ERP013756
  • Stranded RNA-seq of homozygous, heterozygous and wild-type embryos from drosha (sa191) and dgcr8 (sa223) mutant lines · zebrafish (Danio rerio) · n = 5 replicates per genotype per line · ERP013690
  • Processed FPKM expression profiles across the time course in Expression Atlas · zebrafish (Danio rerio) · n = 18 stages · E-ERAD-475
  • Processed DeTCT count data (figshare) · zebrafish (Danio rerio) · 10.6084/m9.figshare.4622311
  • BioLayout network layout file for the co-expression clustering (figshare) · zebrafish (Danio rerio) · 10.6084/m9.figshare.4622293
  • Ensembl teleost-specific one-to-one paralogue pairs and ZFIN stage/anatomy expression annotations · zebrafish (Danio rerio) · n = 3144 expressed pairs; 683 with ZFIN anatomy annotation for both genes
  • Ensembl Compara orthologues used to classify genes as Danio, Teleost or Vertebrate · zebrafish plus six other fish species, mouse, rat and human
Methods & tools
DeTCT, Illumina HiSeq 2500, TopHat v2.0.13, htseq-count v0.6.0, STAR v2.5.1a, Samtools v1.3.1, Picard v2.7.1, Cufflinks/Cuffmerge v2.2.1, bedtools v2.24.0, BioLayout Express^3D (Markov Cluster Algorithm), Ontologizer, topGO
Software & environment
GRCz10 genome; Ensembl v85 annotation for RNA-seq counts, v86 for 3′ end filtering and assembly, v87 for miRNA loci; TopHat v2.0.13, htseq-count v0.6.0, STAR v2.5.1a, Samtools v1.3.1, Picard v2.7.1, Cufflinks v2.2.1, bedtools v2.24.0; no container or requirements file stated.
Reproducibility gaps
DeTCT library preparation and analysis are described only by reference to earlier work; cluster assignments, enrichments and pri-miRNA BED files are Supplemental Data Files with no URLs in the text; no compute details given.
Limitations (6)
  • ZFIN expression annotation is far from complete, so non-overlapping paralogue annotations may only mean that one gene was examined at a stage the other was not; the 83.7% figure is an estimate. · stated by the authors
  • The data resolve stage but not tissue, so paralogues that diverged spatially but not temporally are missed by the correlation analysis. · stated by the authors
  • Some assembled pri-miRNA transcripts may be fragments; their 5′ and 3′ termini still need experimental confirmation, for example by 5′ rapid amplification of cDNA ends. · stated by the authors
  • Raw 3′ end calls include many false positives from poly(T) priming on genomic poly(A) tracts inside transcripts, so the alternative-end analysis relies on a strictly filtered subset (8358 of 253,627 regions). · stated by the authors
  • The proposed roles of the chromosome 4 zinc finger genes at zygotic genome activation (opening chromatin or repressing transposable elements) rest on expression timing alone; no binding, chromatin or loss-of-function data are shown. · noted in this summary
  • All embryos come from a single wild-type strain (HLF), and the functional links for unnamed genes rest on co-expression and annotation enrichment, with no in situ or knockdown follow-up. · noted in this summary

Read this if you need a stage-resolved reference for zebrafish polyadenylated transcripts, alternative 3′ ends or primary miRNA annotation from the 1-cell stage to 5 days post-fertilisation, or co-expression clues for uncharacterised genes.

Citations
362
FWCI
10.05
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISFang '26Rauwerda '22Fishman '23Kanchi '25Farnsworth '19Sur '23
73% semantic matchDevelopmental Biology
Fei Fang, William Poulos, Yifan Yue+4 more·bioRxiv·Mar 26, 2026