The giant axolotl genome uncovers the evolution, scaling, and transcriptional control of complex gene loci

Abstract

Vertebrates harbor recognizably orthologous gene complements but vary 100-fold in genome size. How chromosomal organization scales with genome expansion is unclear, and how acute changes in gene regulation, as during axolotl limb regeneration, occur in the context of a vast genome has remained a riddle. Here, we describe the chromosome-scale assembly of the giant, 32 Gb axolotl genome. Hi-C contact data revealed the scaling properties of interphase and mitotic chromosome organization. Analysis of the assembly yielded understanding of the evolution of large, syntenic multigene clusters, including the Major Histocompatibility Complex (MHC) and the functional regulatory landscape of the Fibroblast Growth Factor 8 (Axfgf8) region. The axolotl serves as a primary model for studying successful regeneration.

Keywords: Topological Associating Domains; axolotl; genome assembly; regeneration.

Fig. 1.

Hi-C–mediated scaffolding of the giant axolotl genome. (A) Procedure for scaffolding the large and repeat-rich genome. Seed clusters were created based on the linkage map data from ref. . Unassigned contigs were assigned to the clusters based on contact frequency. Subsequent scaffolding of the clusters and visual inspection of contact maps allowed their splitting and merging. This was repeated until no more changes could be made to the clusters. The scaffolding itself creates a graph structure from the contact data and contigs contained in a cluster turning contigs into nodes and contacts into edges. Based on the edges, the most likely left and right neighbor of each node are computed, and continuous chains of nodes having each other as their most likely neighbor merged into paths (effectively a new node) subsuming their constituent nodes. This process is repeated until no more paths can be constructed. (B) Hi-C heatmap of d/d contact data of all 28 scaffolded chromosome arms. Denser areas of red signal off-diagonal represent interactions between the arms of the same chromosome. (C) Summary of assembly characteristics. For chromosome arm lengths, see SI Appendix, Fig. S1B.

Fig. 2.

Macrosynteny of axolotl assembly in comparison to gar genome and CLGs. (Upper) Oxford plots of 11,677 1:1 axolotl–gar orthologs showing relative arrangement on chromosomes. (Lower) Oxford plots of 4,938 axolotl–CLG orthologs showing relative arrangement on chromosomes. Clear syntenic clustering of orthologs is visible.

Fig. 3.

Evolution and expansion of AxMHC. (A) Representation of the human MHC, with colored circles indicating when an ortholog was found in the axolotl MHC locus. The colored lines next to the genes represent groups of genes that are neighboring genes in axolotl. Genes coming from MHC class I, II, and III in human are colored in red, blue, and green, respectively. Adapted from ref. . (B–E) Pairwise comparison of gene placement between. (B) axolotl and human without the highly amplified gene families HLA and TRIM. (C) Axolotl and human only showing highly amplified gene families HLA and TRIM. (D) Axolotl and X. laevis. (E) X. laevis and human.

Fig. 4.

Expanded regulatory architecture of Axfgf8. (A and B) Human FGF8 regulatory locus. FGF8 is marked with a blue arrowhead and a dotted line. (A) High magnification at 5 Kb resolution. White circles display CNEs surrounding the hsFGF8, 27 of which show FGF8 enhancer function. (B) Human Hi-C map at 10 Kb resolution (chr10: 101 Mb to 105 Mb). (C and D) Axolotl Fgf8 regulatory unit with Hi-C contact map. Axfgf8 is marked with a blue arrowhead and a line dotted. (C) High magnification at 100 Kb resolution. The linear sequence of genes is given below the heatmap. White circles display 43 CNEs surrounding Axfgf8. (D) Hi-C map at 200 Kb resolution (chr8q: 608 Mb to 696 Mb). (E) Characterization of the axolotl CNE80. (Top) mVISTA (40) plot showing the aligment of a 589 bp AxCNE80 candidate region against the human and mouse CNE80 sequences (mVISTA SLAGAN alignment program: Criteria 70%, 100 bp). The track is shown with a 100 bp window. The candidate region contains a highly conserved 133 bp sequence. (Bottom) Schema of transgenesis construct used to test AxCNE80 function. (F) Expression of Axfgf8 mRNA in stage 44 (St. 44) (41) axolotl limb bud revealed using fluorescent in situ hybridization chain reaction (HCR). Axfgf8 mRNA is expressed in the distal–anterior mesenchyme. The limb bud is outlined by dotted lines. (G) Axolotl CNE80 drives expression in the Axfgf8 expression domain. GFP expression is in the limb bud of an axCNE80 > EGFP transgenic animal at day 12 and 29 after injection (St. 44 to 47). (Top) GFP images. (Bottom) Overlay with widefield. Strong GFP expression is detected in the distal–anterior mesenchyme, similar to Axfgf8 mRNA.

Fig. 5.

Expanded features of axolotl interphase chromosomes. (A) TAD length comparison. Ratio of orthologous TADs between axolotl and human. TADs surrounding gene orthologs were identified genome-wide. On average, axolotl TADs are 7× longer than their human counterparts. The TAD that encompasses Axfgf8 (green bar and arrowhead) is 49× longer than its orthologous interval in human. (B) Contact probability plots for axolotl (blue) and chicken (orange). Genome-wide contact probability plots are presented versus genomic distance between loci normalized to the contact probability at 10 Kb. In the interphase data, the inflection point (dashed lines) in chicken is at 40 Kb, while in axolotl, it is at 300 Kb, suggesting that large loops accumulate in axolotl interphase chromosomes. (C) Axolotl intra-TAD contact probability as a function of distance in conserved TADs, classified into different size categories. Graphs show the same slope indicating same contact frequencies within TADs of different sizes. Upturn at end of each line reflects increased contact frequency found near TAD boundaries.

Fig. 6.

Expanded features of axolotl mitotic chromosomes. (A) Mitotic chromosome heatmap. Hi-C map of chromosome arm 6q during mitosis. The inset demonstrates the presence of the secondary diagonals, which is characteristic for mitotic Hi-C heatmaps (scale bar 100 Mb). (B) Contact probability plots. Genome-wide contact probability plots are presented versus genomic distance between loci normalized to the contact probability at 10 Kb. In mitotic chromosomes (Bottom), the peak at 35 Mb (dashed line) corresponds to one turn of the helix. The peak in the first derivative (right) at 600 Kb corresponds to the length of a single loop within that turn. (C) Proposed model of helical organization of mitotic chromosomes. During mitosis, the DNA is packaged in ∼600 Kb loops, which themselves are arranged into helical turns of the polymer. A single turn comprises ∼35 Mb DNA, which agrees well with the values observed in the contact probability plot.