Study design.

a List of mutant alleles examined in this study. b Schematic representation of genetic interaction analysis. The fitness (as expressed in particular phenotypic characteristics) of genotypes determines whether genetic interaction is alleviating (when the phenotype is less severe than expected from the phenotypes of single mutants) or synthetic (when the phenotype is more severe than expected). When the phenotype conforms to expectation, then a non-interacting situation is recorded (see Methods for details of the multiplicative model used for the present analysis).

Characterization of a zebrafish pole1 hypomorphic allele.

a Diagnostic whole-mount RNA in situ hybridization using rag1 (thymus; purple circle), and gh (hypophysis; orange circle) of zebrafish embryos at 5 days after fertilization (dpf) (dorsal view; representative of 9 wild-type/mutant pairs; all wild-type fish exhibit strong rag1 signals, whereas 5 mutant fish lack a rag1 signal; 4 mutant fish have weak rag1 signal; see also Fig. 9). b Thymic lymphopenia revealed by a ikzf1:EGFP reporter (lateral views; representative of 5 wild-type/mutant pairs; all wild-type fish exhibit strong signals, whereas 4 mutant fish have a minimal signal (as shown); 1 mutant fish lacks clearly discernible signal). c Genetic map for the region on chromosome 5 in the vicinity of the pole1 gene (not to scale; transcriptional orientations indicated). The number of recombination events observed between several genetic markers and the mutated locus in the number of meioses shown is indicated in brackets. d Identification of an I633K missense mutation in the pole1 gene of HG010 mutants. Partial protein sequences encompassing the region of the first of three catalytic aspartates (asterisk; orange font) and the universally conserved isoleucine (I) residue in family B DNA polymerases (red outline) are shown for several species (Genbank accession numbers): D. rerio Pole1, amino acids [aa] 623-643 (NP_001121995); M. musculus Pole1, aa 624-644 (AAD46482); N. vectensis DNA Pole, aa 610-620 (XP_001628405); D. melanogaster DNA Pole, aa 622-642 (BAB17608); S. cerevisiae POL2, aa 638-658 (CAA63235); E. coli polB, aa 417-437 (YP_851261); bacteriophage Rb69 DNA polymerase ChainA, aa 409-429 (1WAJ_A). e Phenocopy of impaired T cell development in pole morphants; in 19 embryos, the rag1 signal was specifically lost (the gh signal was unchanged), otherwise only slightly reduced or normal. f Phenotypic rescue by mouse Pole1; in 11 of 14 mutants, rag1 signals could be detected after BAC injection. g Normal embryonic haematopoiesis in pole1 mutants, as determined by whole-mount RNA in situ hybridization at 24 h after fertilization with the indicated gene-specific probes; representative of between 4 and 7 wild-type/mutant pairs. h Near normal craniofacial structures as determined by whole-mount RNA in situ hybridization with dlx2, at 3 days after fertilization (dpf) (representative of 5 wild-type/mutant pairs). i Analysis of craniofacial structures as determined by alcian blue staining to visualize cartilage structures at 4 dpf (representative of 3 wild-type/mutant pairs). j Normal shape of pharyngeal ectoderm indicated by gcm2 expression; representative of 5 wild-type/mutant pairs. k Normal pharyngeal endoderm indicated by foxn1 expression; representative of 5 wild-type/mutant pairs. l Impaired T cell differentiation indicated by expression of ikzf1, ccr9b and tcrb (representative of between 5 and 15 wild-type/mutant pairs). m Genetic interaction between pole and pold1 (representative embryos shown). First row: Uninjected wild-type (pole^+/+) embryos are shown for reference. The gh signal is marked in green, the rag1 signal in the thymus in purple (whole mount RNA in situ hybridization, left panel); the arrow points to the intersection of cartilaginous structures of the ethmoid plate of the neurocranium and the mandibular arch (alcian blue staining; right panel). The fraction of embryos with rag1-positive thymi is shown in the right-hand column. Second row: Injection of high concentrations of anti-pold1 oligonucleotides (1 mM) results in the loss of rag1 signals in the majority of embryos (15/24). The mandibular arch is shorter in 23 out of 27 morphants. Third row: Injection of low concentrations of anti-pold1 oligonucleotides (0.2 mM) results in milder phenotypes; 17 out of 20 morphants had a reduced rag1 signal in the thymus; the mandibular arch was normal in 2 and somewhat smaller in 27 out of 29 morphants. Fourth row: Injection of low concentrations of anti-pold1 oligonucleotides (0.2 mM) into embryos heterozygous for the pole^t20320 allele caused severe phenotypes. In 5 out of 9 morphants, no rag1 signal was observed, whereas it was present but reduced in four fish; the mandibular arch was severely affected in 2 (representative embryo shown), mildly affected in 6 and normal in 2 morphants. The effect of anti-pold1 oligonucleotides on rag1 expression in wild-type versus pole^+/t20320 heterozygotes is statistically significant (t- test; two-tailed). Scale bars, (a, e–m [1 mm]; b [0.25 mm]).

Global changes in DNA methylation patterns in single and double mutant zebrafish.

a Global methylation patterns of CG dinucleotides expressed as mean methylation ratios; median±m.a.d. b Quantification of data in (a) represented as median±m.a.d.; DMR, differentially methylated regions. The P values for pair-wise comparisons (t-test; two-sided) are indicated, corrected for multiple testing. c Methylation levels around transcriptional start sites (TSS). d Variable methylation levels of 222 differentially methylated regions (as assessed by CG methylation) present in dnmt1 mutants across different genotypes.

Expression analysis of genes associated with the DNA methylation process.

qPCR analysis was performed on 5 dpf embryos of wild-type and mutants (n = 3 for each genotype) for the four mutant lines; mean values are indicated. The P values (t-test; two-tailed) for pairwise comparisons of each wild-type (white columns)/mutant (red columns) group are indicated above the columns.

Alterations in non-CG methylation patterns in dnmt1 and pole1 mutants.

a Variable numbers of methylated sites for CG dinucleotides, and CHG and CHH tri-nucleotides. The number of sites in wild-type fish is set to 100 (for absolute numbers of total and methylated sites, see Supplementary Table 1). b Violin plots of methylation ratios; median±m.a.d. The significance levels of the differences between mean methylation ratios are indicated (t-test, two-tailed; corrected for multiple testing). c Correlation plot between the fractions of sites methylated and their mean methylation ratios. For b, c, color code for genotypes indicated in a.

Alterations in non-CG methylation patterns in mat2aa and pole1 mutants.

a Variable numbers of methylated sites for CG dinucleotides, and CHG and CHH tri-nucleotides. The number of sites in wild-type fish is set to 100 (for absolute numbers of total and methylated sites, see Supplementary Table 1). b Violin plots of methylation ratios; median±m.a.d. The significance levels of the differences between mean methylation ratios are indicated (t-test, two-tailed; corrected for multiple testing). c Correlation plot between the fractions of sites methylated and their mean methylation ratios. For b, c, color code for genotypes indicated in a.

Transcriptional landscapes in dnmt1 and pole1 mutants.

a Dynamic changes in transcriptomes of dnmt1 and pole1 single and double mutants, as indicated by principal component analysis. b Venn diagram indicating the overlap of differentially expressed genes for down-regulated and up-regulated genes in fish of the indicated genotypes. c Four clusters of co-regulated genes are identified in dnmt1 mutant fish; each column represents a biological replicate. d Co-regulation of groups of genes in (c) identifies 2 gene clusters with restored wild-type levels in dnmt1/pole1 double mutants; box plots (mean; box, upper/lower quartile; whiskers, minimal/maximal values).

Transcriptional landscapes in mat2aa and pole1 mutants.

a Dynamic changes in transcriptomes of mat2aa and pole1 single and double mutants, as indicated by principal component analysis. b Venn diagram indicating the overlap of differentially expressed genes for down-regulated and up-regulated genes in fish of the indicated genotypes. c Expression patterns of genes in the four clusters of co-regulated genes identified in dnmt1 mutant fish; each column represents a biological replicate. d No restoration of expression levels in mat2aa/pole1 double mutants for the genes in the four groups identified in dnmt1 mutant fish; box plots (mean; box, upper/lower quartile; whiskers, minimal/maximal values).

Epistasis analysis of dnmt1 and pole1 mutations.

a Whole-mount RNA in situ hybridization results of larvae of the indicated genotypes are shown (rag1 signal in thymus [indicated by purple circle]; gh signal in hypophysis [indicated by light green circle]). The lens and retina are outlined. Scale bar, 1 mm. b Representation of rag1/gh ratios of fish with the indicated genotypes (each symbol represents an individual fish); mean±s.e.m. c Quantitative analysis of data in (a) (mean±s.d.; t-test; two-tailed); alleviating interaction according to the multiplicative model (see Methods) is demonstrated. d Relative size of retinas of fish with the indicated genotypes (each symbol represents an individual fish; mean±s.e.m. e Quantitative analysis of data in (d) (mean±s.d.; t-test; two-tailed); alleviating interaction according to the multiplicative model (see Methods) is demonstrated.

Genetic interaction analysis.

a Summary of genetic interaction analyses for the indicated pairs using the rag/gh ratio as parameter; fold changes are shown relative to the default state (1; no interaction). The multiplicative model was used throughout (see Methods). b Global gene expression analysis in the indicated genotypes; the type of genetic interaction follows the nomenclature in⁸⁶. c Expression pattern of co-regulated genes in cluster #20 in fish of the indicated genotypes. The numbers of genes per cluster are indicated; box plots (mean; box, upper/lower quartile; whiskers, minimal/maximal values). d Pathway analysis of genes in cluster #20 of co-regulated genes (see Supplementary Fig. 23 for full STEM analysis).