A platform for rapid exploration of aging and diseases in a naturally short-lived vertebrate

Abstract

Aging is a complex process that affects multiple organs. Modeling aging and age-related diseases in the lab is challenging because classical vertebrate models have relatively long lifespans. Here, we develop the first platform for rapid exploration of age-dependent traits and diseases in vertebrates, using the naturally short-lived African turquoise killifish. We provide an integrative genomic and genome-editing toolkit in this organism using our de-novo-assembled genome and the CRISPR/Cas9 technology. We mutate many genes encompassing the hallmarks of aging, and for a subset, we produce stable lines within 2-3 months. As a proof of principle, we show that fish deficient for the protein subunit of telomerase exhibit the fastest onset of telomere-related pathologies among vertebrates. We further demonstrate the feasibility of creating specific genetic variants. This genome-to-phenotype platform represents a unique resource for studying vertebrate aging and disease in a high-throughput manner and for investigating candidates arising from human genome-wide studies.

Figure 1. A versatile platform for rapid exploration of aging and longevity genes in the naturally short-lived turquoise killifish

(A) Lifespan of non-vertebrate and vertebrate model systems widely used for aging and disease research (top panel), when compared to the lifespan of the turquoise killifish (bottom panel). The turquoise killifish originates from ephemeral water ponds in Zimbabwe and Mozambique (bottom panel). (B) Examples of genes encompassing the hallmarks of human aging (modified with permission from (Lopez-Otin et al., 2013)). (C) Genomic pipeline to generate CRISPR/Cas9 guide RNAs (gRNAs) in a new model organism using our newly created genomic tools (de novo assembled turquoise killifish genome, epigenome, and transcriptome). Gene models and gRNA selection are available via CHOPCHOP. (D) CRISPR/Cas9 genome-editing pipeline to generate stable mutant fish lines in the turquoise killifish. Overall, the total time for generating a stable mutant line in the lab (i.e. steps 1–4) is about 2–3 months.

Figure 2. Example of rapid genome-editing of TERT, the protein component of telomerase, in the turquoise killifish

(A) The telomerase complex, and gene model prediction for TERT and TERC using genomic and epigenomic profiling. (B) Conservation of TERT protein domains between human (hTERT) and the turquoise killifish (kTERT). (C) TERT protein sequence divergence predicts evolutionary tree. (D) Relative expression of TERT mRNA in brain, liver, testis and tail using RNA-seq. FPKM: fragments per kilobase of exon per million fragments mapped. (E) Successful editing of the turquoise killifish TERT gene. The wild-type (WT) sequence as well as the length of deletions (Δ) is indicated relative to the protospacer adjacent motif (PAM, in gray) and the guide RNA sequence (gRNA, in red). The deletions that gave rise to stable lines (Δ3 and Δ8) are indicated (in yellow with black outline). (F) Top panel: location of the gRNA successfully targeting TERT exon 2 (red line), which is upstream of the exons encoding TERT catalytic domains. TERTΔ8 allele is predicted to generate a protein with a premature stop codon. Bottom panel: the TERT Δ8 allele is successfully transcribed to RNA, as measured by RT-PCR followed by cDNA sequencing. RT: reverse-transcriptase.

Figure 3. TERT^Δ8/Δ8 fish show no telomerase activity and exhibit a progressive loss of fertility in the first generation

(A) Intercrossing of TERT^Δ8/+ heterozygous (het) fish to generate generation 1 (G1) TERT^Δ8/Δ8 fish. G1 TERT^Δ8/Δ8 fish are observed at the expected Mendelian ratios (no difference between expected and observed frequencies, p = 0.8809, χ² test). (B) G1 TERT^Δ8/Δ8 embryos (left panels) and adults (right panels) are outwardly normal. (C) Schematic for the Telomere Repeat Amplification Protocol (TRAP). Telomerase enzymatic activity in liver is evaluated by the ability of tissue extract to add telomeric repeats to radio-labeled artificial telomeres in vitro. (D) Telomerase enzymatic activity as measured by the TRAP assay in TERT^+/+ and G1 TERT^Δ8/Δ8 fish liver samples. IC: TRAP internal control product. Representative of 3 independent experiments. (E) Experimental design to assess male fertility. TERT^Δ8/+ (control) and G1 TERT^Δ8/Δ8 (mutant) males, at two different age groups (2 and 4 months), were mated with young (2 months) wild-type (WT) females. Fertilized eggs (gray) were counted after 1 week. (F) Ratio of fertilized eggs per week of egg-lay in TERT^Δ8/+ (control) and G1 TERT^Δ8/Δ8 (mutant). Mean±SD of >70 eggs, generated from 4–5 crosses per age group. Wilcoxon signed-rank test, *p<0.05, **p<0.01. For the comparison between age groups, standardized values to age-matched controls were used. (G) Histological sections of testis from TERT^+/+ (control) and G1 TERT^Δ8/Δ8 fish at 4–5 months (4m, full size image) and 2 months (2m, insert). Sz: spermatozoa (mature sperm); St: spermatids. Scale bar: 50μm. Representative of n≥6 individuals from each genotype (4–5 months) and n=2 individuals from each genotype (2 months).

Figure 4. TERT-deficient turquoise killifish exhibit genetic anticipation

(A) Experimental design. G1 TERT^Δ8/Δ8 (left) or TERT^+/+ (right) fish were intercrossed to generate generation 2 (G2) TERT^Δ8/Δ8 or TERT^+/+ fish, respectively. The development of embryos was assessed until hatching. (B) Representative images of TERT^+/+ and G2 TERT^Δ8/Δ8 embryos at an equivalent developmental stage. Scale bar: 300μm. (C) Ratio of successful hatching per week of egg-lay for the indicated genotypes. Mean+SD of >70 embryos for each parental genotype (TERT^Δ8/+ versus G1 TERT^Δ8/Δ8). (D) Telomere length measurement using telomere restriction fragment (TRF) Southern-blot. Left panel: TERT^+/+ and G2 TERT^Δ8/Δ8 embryos. Representative of 3 experiments. LC: loading control for genomic DNA. Right panel: G1 TERT^Δ8/Δ8 and G2 TERT^Δ8/Δ8 embryos. Expanded version is in Figure S3. White asterisk: non-specific probe binding.

Figure 5. Precise generation of human disease mutation in TERT and insertion of a short sequence in POLG

(A) Genome-editing pipeline for specific point mutations and insertions. ssDNA: single-strand DNA template. NHEJ: non-homologous end joining. HDR: homology-directed repair. (B) Top panel: Disease-associated variants in hTERT. Conservation of the disease-causing residues between human TERT and turquoise killifish TERT is color-coded (red: identical, pink: similar in turquoise killifish TERT). Bottom panel: K902 in human TERT is evolutionary conserved and corresponds to K836 in turquoise killifish TERT. (C) Top panels: Location of a selected gRNA (red line) in close proximity to K836 in exon 11 of the turquoise killifish TERT, and core sequence of the co-injected ssDNA template. Bottom panel: precise editing of specific codons leading to the nucleotide change (A to G) corresponding to the K836R mutation. An example chromatogram is shown at the bottom. The K836R mutation is highlighted with green background. fs: frame shift, del: deletion. (D) Location of the gRNA targeting exon 2 of the turquoise killifish POLG, and core sequence of the co-injected ssDNA template to introduce an exogenous NdeI site. Bottom panels: precise insertion of the NdeI restriction sequence, as shown by direct sequencing or restriction digestion. Representative of 2 independent experiments.

Figure 6. A toolkit for vertebrate aging and age-related disease research

(A) Genes that were successfully edited in the 9 hallmarks of aging. Genes and pathways for which we chose to generate stable lines are indicated in yellow with a black outline. (B) Detailed stages of editing completion in specific genes, color-coded as indicated. Presence of orthologs in different species is indicated in gray. (C) Selected examples of targeted genes, depicting detailed genomic, epigenomic and expression information (upper box), relative expression in tissues (lower left box), and types of observed indels and substitutions (lower right box). Germline-transmitted alleles assessed in pooled F1 embryos are in yellow. Stable lines are in yellow with a black outline. Whenever assessed, the targeting efficiency in eggs was indicated as a percentage. For ASH2L, the Δ6 stable line was generated by a separate pair of founders and was not part of the efficiency calculation. Example of a sequencing chromatogram showing the substitution in p15INK4B.