Genomes are constantly exposed to exogenous and endogenous genotoxic insults. DNA-repair proficiency depends on cell type and cell cycle stage, and is particularly different in germ cells than in somatic cells. In germ cells, DNA repair is highly efficient to maintain genome integrity throughout generations; in somatic cells, DNA repair maintains genome integrity early in life, but operates inefficiently during later, post-reproductive stages1. Although germlines have mutation rates that are orders of magnitude lower than those in somatic tissues2,3,4, the high mutation rates in the soma increase in an age-dependent manner across species5. In C. elegans, somatic cells mostly terminally differentiate during embryogenesis and are entirely post-mitotic in the adult, whereas germ cells retain mitotic and meiotic activity. Germ cells survey their genome for helix-distorting lesions by global-genome nucleotide excision repair (GG-NER) and accurately repair DNA double-strand breaks (DSBs) through homologous recombination repair (HRR)6,7,8,9,10. In somatic cells, GG-NER11 and HRR8,12,13,14 are dispensable, as they instead use error-prone non-homologous end joining (NHEJ), and only actively expressed genes are surveilled by transcription-coupled NER (TC-NER)7,9. The resistance to DNA-damage-driven developmental growth impairment and functional deterioration during aging is thus limited by the restriction of somatic DNA-repair capacities. Also in mammals, the engagement of accurate DNA-repair systems depends on the cell cycle and differentiation state15. Not only HRR, but also additional repair pathways, are enhanced during replication, like single-strand annealing, microhomology end joining and long-patch base-excision repair (BER)16,17,18. By contrast, cell types that are either quiescent, terminally differentiated, or senescent have limited DNA-repair capacities.

Cellular quiescence and differentiation are controlled by the DREAM complex, formed by the Dp/Retinoblastoma(Rb)-like/E2F and the MuvB subcomplexes19. In C. elegans, the DREAM complex comprises subunits encoded by genes that were first discovered as synthetic multivulva (synMuv) class B genes, owing to their role in cellular differentiation in combination with mutant proteins encoded by other synMuv gene classes20,21,22. Single mutations in these genes were sufficient to promote germline-like characteristics in the soma, including misexpression of germline genes23,24,25,26. In humans, in addition to the highly conserved repressor function in quiescence, the components of the DREAM complex can associate with other proteins and instead function as a transcription activator during the cell cycle27,28,29. The specific assembly of the repressive DREAM complex in G0 is regulated by the DYRK1A protein kinase, which, through phosphorylating LIN52, can bind to p130 and form the complex30.

Here, we report that the DREAM complex represses a wide range of DNA-repair genes in the soma of C. elegans. Mutations in genes encoding DREAM components trigger DNA-damage resistance in somatic tissues against a broad range of genotoxic insults, including ultraviolet (UV) lesions, alkylations, interstrand crosslinks (ICLs) and DSBs. DREAM mutants showed accelerated lesion removal and suppressed the sensitivity to DNA damage of animals deficient in DNA-repair genes. In human cells, the DYRK1A inhibitors harmine and INDY31,32 triggered the induction of DREAM-targeted DNA-repair genes and conferred resistance to distinct DNA-damage types. In vivo, harmine treatment reduced DNA damage and apoptosis in the retinas of Ercc1−/− progeroid mice. We thus establish that pharmacologically targeting the DREAM complex could be applied to augment genome stability. We propose that the DREAM complex represses the expression of various DNA-repair mechanisms and thus limits DNA-repair capacities. Therefore, inhibition of DREAM could overcome the consequence of dysfunction in single DNA-repair systems and DNA-damage-driven aging.


DDR gene promoters carry the CDE-CHR DREAM-binding motif

To investigate the mechanisms underlying transcriptional regulation of DDR genes in C. elegans, we assessed whether specific transcription-factor-binding sites might be overrepresented in DDR gene promoters. An unbiased DNA-motif enrichment analysis of the 211 DDR genes (Supplementary Table 1) revealed a significant enrichment of the DPL-1-, EFL-1- and LIN-15B-binding motifs (Fig. 1a). DPL-1 and EFL-1 form the E2F–DP heterodimer, which directly contacts promoters by binding to the cycle-dependent element (CDE). E2F–DP is linked through the pocket protein LIN-35 to the MuvB subcomplex, which binds the cell cycle genes homology region (CHR) in promoters to then form the DREAM transcriptional repressor complex33,34,35. We identified the CDE-CHR motif in the promoters of 125 of the 211 DDR genes (Fig. 1b), suggesting that the DREAM complex is a regulator of DDR genes.

Fig. 1: Mutations in genes encoding components of the DREAM complex confer resistance to UV-induced DNA damage during development and adulthood.
figure 1

a, Sequences of the motifs found upon analysis of the promoters of the DDR genes using HOMER. BG, background; FC, fold change; E2F, E2F transcription factor; Zf, zinc finger domain. b, Result of the motif search for the CDE-CHR DREAM complex motif in the promoters of the DDR genes using HOMER. c,d, UV-irradiation assay during somatic development of WT, lin-52(n771), lin-35(n745), dpl-1(n2994), efl-1(se1) (c) and WT, lin-52(n771), dpl-1(n2994) and lin-52(n771); dpl-1(n2994) mutant worms (d). Y axis shows the percentage of the different larval stages, and x axis the UV dose applied in mJ/cm2. Representative graph showing n = 3 biological replicates from 1 of at least 3 independent experiments. Data are shown as mean ± s.d. of each larval stage (L1–L4-A). For statistical analysis, a two-tailed t-test between the fraction of each larval stage of a mutant compared with WT in the same treatment condition was used, except for lin-52(n771); dpl-1(n2994), which was compared with lin-52(n771). e, Lifespan assay upon exposure of WT and DREAM-complex-mutant worms to UV-B (0 and 400 mJ/cm2). A log-rank test was performed to compare the lifespan of the DREAM mutants and WT worms in the same conditions. Top graphs, n = 122 (without UV), 150 (with UV) for WT; 121, 152 for lin-52; and 128, 154 for dpl-1 mutants. Bottom graphs, n = 138, 218 for WT; 162, 215 for efl-1 and 144, 201 for lin-35 mutants. Bar graphs show the percentage by which mean lifespan decreased for each strain irradiated with UV-B compared with the mock-treated worms of the same strain. f, UV-irradiation assay for germline development of WT, xpc-1(tm3886), lin-52(n771) and the double mutant, lin-52(n771); xpc-1(tm3886). Representative graphs of n = 3 biological replicates, from 1 of 3 independent experiments. The mean ± s.d. of eggs laid or the percentage hatched is shown. Two-tailed t-tests were performed to compare the number of eggs laid and hatched between the different strains within the same condition. For a and b, the P value over the background was calculated with a hypergeometric test, and the q value shows the Benjamini–Hochberg-adjusted P values. For c and d, P > 0.05, not shown; *P < 0.05, **P < 0.01, ***P < 0.001, **** P < 0.0001. Remaining comparisons and detailed P values are provided in Supplementary Table 13, including results from Fisher’s exact test.

DREAM-complex mutants confer DNA-damage resistance

We next determined whether mutations in DREAM components influence DNA-damage sensitivity. We tested UV-induced DNA lesions because they impact developmental growth and the longevity of the animals9. UV-B irradiation induces formation of cyclobutane pyrimidine dimers (CPDs) and pyrimidine (6–4) pyrimidone photoproducts (6–4PPs) that are repaired by NER. Except for the two primordial germ cells, all 558 cells of the L1 larvae are somatic cells, of which 90% are terminally differentiated36. We exposed synchronized DREAM-mutant L1 larvae to UV and scored developmental growth 48 hours (h) later, counting the developmental stages for all the worms, from L1 to the consecutive L2, L3 and L4 larval and adult stages. Surprisingly, worms with loss-of-function mutations in lin-52, dpl-1, efl-1, lin-53 or lin-35, which encode DREAM-complex components, showed a significant improvement in somatic development compared with wild-type (WT) worms following UV exposure (Fig. 1c and Extended Data Fig. 1a). Even though lin-35 mutants showed a developmental delay in the absence of UV26, they proceeded through development more rapidly than did WT worms at high UV doses (Fig. 1c).

Double-mutant worms, with mutations in two genes (lin-52; dpl-1 and lin-52; efl-1), showed improved developmental growth, similar to that in lin-52 mutants, indicating that their encoded DREAM-complex subunits conferred UV resistance (Fig. 1d and Extended Data Fig. 1b). By contrast, worms with mutations in the synMuv B class of genes or a mutant component of the chromatin-remodeling NuRD complex that is not part of the DREAM complex did not have improved developmental growth upon UV exposure (Extended Data Fig. 1c). These data suggest that the specific function of LIN-52, DPL-1, EFL-1, LIN-53 and LIN-35 as subunits of the DREAM complex determine the animals’ ability to overcome DNA-damage-induced developmental delay.

To evaluate whether mutations in the DREAM-complex genes could affect DNA-damage-driven organismal aging, we UV-treated the DREAM-complex-mutant worms on day 1 of adulthood and assessed their lifespans (Fig. 1e). In humans, mutations in DNA-repair genes are sufficient to accelerate aging and lead to premature death37, but C. elegans worms cultured under laboratory conditions require exogenous DNA damage to shorten their lifespans9,38,39. Worms with mutations in lin-52, dpl-1, efl-1 or lin-35 significantly outlived WT worms upon DNA damage, despite the fact that some were short-lived without irradiation.

The reduced lifespans of dpl-1, elf-1 and lin-35 mutant animals under unperturbed conditions might be connected to their previously described roles in contributing to developmental processes, whereas lin-52(n771) might be a hypomorphic mutation that does not affect associations among the complex components40. In the absence of genotoxins, worms with the lin-52(n771) genotype showed only a slight reduction in egg-laying capacity and a mild sensitivity to starvation (Extended Data Fig. 2a,b). This suggests that a partial loss of function of DREAM’s function enhances the resistance to DNA damage without affecting other physiological processes.

The lifespan reduction under unperturbed conditions is consistent with the reported lifespan shortening of lin-9, lin-35 and lin-37 mutant animals41. Another report has found that lin-52, lin-37 or dpl-1 mutant worms treated with 5-fluoro-2′-deoxyuridine (FUdR)42, which is genotoxic, had extended lifespans, further supporting our findings that DREAM mutants alleviate DNA-damage-induced lifespan shortening.

Indicative of healthspan extension, lin-52 mutants animals retained more motility than did WT animals (Extended Data Fig. 3). Owing to the phenotypic specificity with mild adverse effects under unperturbed conditions and the strong UV-resistance phenotype, we decided to focus mainly on LIN-52 mutants to further investigate the role of the DREAM complex in regulating genome stability.

Considering that DREAM represses gene expression in non-dividing cells19, we hypothesized that the UV sensitivity in the germline would be unaffected. lin-52 mutant animals laid a comparable amount of eggs to WT worms upon UV exposure, with similar hatching rates, and the lin-52 mutation did not alleviate the germline hypersensitivity of worms with mutated xpc-1 (Fig. 1f). Therefore, DREAM mutations specifically augment DNA-damage resistance of the somatic tissues.

DREAM-complex mutants improve DNA repair

To test whether DREAM mutants enhance DNA repair, we measured the removal of the main UV-induced DNA lesion type, CPDs. Twenty-four hours following UV treatment of L1 larvae, we quantified CPDs using an anti-CPD antibody. The tested DREAM mutants showed significantly improved CPD repair compared with WT animals (Fig. 2a). We confirmed the improved repair by using an anti-6-4PP antibody, which also revealed a decrease in the amount of 6-4PPs in lin-52 mutant animals (Extended Data Fig. 4a).

Fig. 2: DREAM-complex mutants show enhanced repair of UV-induced DNA lesions and alleviate the UV sensitivity of csb-1 and xpc-1 mutant animals.
figure 2

a, DNA-repair capacity assay in WT, xpa-1(ok698), lin-52(n771) and dpl-1(n2994) L1 worms. A representative slot blot of three independent experiments is shown. Samples labelled as ‘UV’ were collected right after UV irradiation; samples ‘UV + 24 h’ were collected 24 h after UV irradiation. Graphs show the mean ± s.d. of the improved or decreased repair of the mutants compared with that of WT worms. n = 3 biological replicates. A two-tailed t-test was used to compare the mutants’ repair with that of WT worms. b, Representative images of a focal plane of the anterior region of adult worms irradiated and stained with antibodies to CPDs and DAPI, collected right after irradiation (0 h) or after incubation for 60 h. Scale bar, 25 µm. c, Quantification of CPD nuclei signal intensity in the heads of adult worms irradiated and collected immediately (0 h, blue dots) or 60 h after irradiation (orange dots). The number of nuclei quantified was, at 0 h and 60 h, respectively, n = 1,469 and 1,479 for WT, n = 1.454 and 1.321 for lin-52(n771), from 5–7 heads per condition. The y axis shows the log10-transformed intensity values of CPDs. Box midlines show the median, box limits show the top and bottom quartiles and whiskers extend to 1.5 × interquartile range (IQR). Two-way analysis of variance (ANOVA) between the strain (WT and lin-52) and the time component is shown. a.u., arbitrary units. df, UV-irradiation assay during somatic development of WT, lin-52(n771), csb-1(ok2335) and lin-52(n771); csb-1(ok2335) (d); WT, lin-52(n771), xpc-1(tm3886) and lin-52(n771); xpc-1(tm3886) (e); and WT, lin-52(n771), csb-1(ok2335); xpc-1(tm3886) and lin-52(n771); csb-1(ok2335); xpc-1(tm3886) (f). Y axis shows the percentage of the different larval stages, and x axis the UV dose applied in mJ/cm2. Graphs are representative of n = 3 biological replicates from 1 of 3 independent experiments. Data are shown as mean ± s.d. of each larval stage. Results from two-tailed t-tests between the fraction of each larval stage of lin-52-mutated worms compared with WT, and each larval stage of lin-52-mutated NER-deficient worms compared with the NER-deficient control, are shown for the same treatment conditions. P > 0.05, not shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. For df, detailed P values and comparisons against WT worms are in Supplementary Table 13, including results from Fisher’s exact test to analyze the overall distribution of the larval stages.

Source data

To exclude that enhanced lesion removal might be a consequence of damage dilution due to DNA replication, we performed an EdU-incorporation assay in L1 worms. Both lin-52 mutant and WT worms showed comparable DNA-replication events within 24 h of UV exposure, further confirming that the observed decrease in CPD and 6-4PP was due to repair (Extended Data Fig. 4b).

We next assessed whether the DREAM complex also regulates the DNA-repair capacity in adult animals. CPD quantification in the nuclei-dense heads of UV-treated worms at day 1 of adulthood revealed an improved removal of CPD lesions in lin-52 mutants compared with that in WT worms, indicating an augmented repair capacity (Fig. 2b,c). EdU incorporation in intestinal cells, which are among the most prone to hyperproliferate43 (Extended Data Fig. 4c) showed no replication events that could alter the response to DNA damage.

To determine whether the effect of lin-52 on DNA repair was specific to the somatic cells, we performed the same immunofluorescence DNA-repair analysis on germlines upon UV exposure. Both WT and lin-52 mutant worms had a similar, highly efficient repair capacity (Extended Data Fig. 5a). Therefore, the improved capacity to repair CPDs is specific to the somatic cells of lin-52 mutants.

NER is initiated either by the TC-NER protein CSB-1 in the transcribed strand, which is particularly relevant in somatic tissues, or by the GG-NER protein XPC-1, which recognizes damage throughout the whole genome and is crucial in the germline7. Both branches then recruit XPA-1 to assemble the NER core machinery. We evaluated whether mutations in the DREAM complex required these NER branches in order to confer resistance to UV-induced DNA damage. The lin-52 mutation alleviated the UV sensitivity of csb-1, csa-1 and xpc-1 mutants (Fig. 2d,e and Extended Data Fig. 5b) but had no effect on completely NER-deficient xpa-1 mutants or csb-1; xpc-1 double mutants (Fig. 2f and Extended Data Fig. 5c). Therefore, a mutation in the DREAM complex improved both GG-NER and TC-NER and could thus partially compensate for defects in either one of the NER-initiating systems, whereas the enhanced UV resistance depends on the presence of the NER machinery.

The DREAM complex represses multiple DNA-repair pathways

To address whether the DREAM complex could curb the repair capacity of somatic cells by directly repressing DNA-repair genes, we performed RNA sequencing (RNA-seq) of lin-52(n771) and WT L1 larvae (Fig. 3). The majority of differentially expressed genes were upregulated in lin-52 mutants, and these genes included a range of genes involved in DNA-repair mechanisms (Fig. 3a, Table 1 and Supplementary Table 2). Gene Ontology (GO) analysis of the significantly upregulated genes (adjusted P < 0.05) in lin-52 mutants compared with WT were strongly enriched for DDR-related terms (Fig. 3b and Supplementary Table 3). Similarly, proteomic analysis of lin-52 mutant and WT worms revealed that multiple DNA-repair proteins were upregulated (Fig. 3c) and multiple GO terms related to DNA repair were also enriched (Fig. 3d and Supplementary Table 4) in the mutants. The significantly regulated genes showed very similar regulation in both the transcriptome and proteome, with multiple DDR genes induced in both (Fig. 3e).

Fig. 3: The DREAM complex directly represses multiple DNA-damage-response genes that are normally enriched in the germline.
figure 3

a,c, Differentially expressed genes (a) or proteins (c) in lin-52(n771) mutants (adjusted P < 0.05). DDR genes (or their products) are shown in orange, and the significantly changed DDR genes in common between a and c are labeled. b,d, GO enrichment analysis for the upregulated genes (b) or proteins (d) in lin-52(n771) mutants compared with WT (adjusted P < 0.05, two-sided Fisher’s exact test with FDR). Highly overlapping terms were removed for simplicity. Terms related to DNA-damage responses are shown in red. The dashed lines mark an adjusted P value of 0.05. The full list is in Supplementary Table 3 and Supplementary Table 4. e, FC of genes that were significantly changed (adjusted P < 0.05) in both the proteome and transcriptome of lin-52 mutants. f, qPCR analysis of DDR genes in lin-52(n771), dpl-1(n2994) and efl-1(se1) mutants. Data are shown as mean ± s.d., n = 3 biological replicates. Results from two-tailed t-tests are in Supplementary Table 13. g, Overlap between the DDR genes upregulated in lin-52(n771) and the genes involved in the main DNA-repair pathways. Overlap between repair pathways is not shown. Pathway genes were obtained from the GO database released on 8 October 2019. h, Overlap between the DDR genes that were upregulated in lin-52 compared with WT worms and two published transcriptome datasets on lin-35(n745)44,45. i, GSEA of all DDR genes in the RNA-seq of lin-52 and the genes bound by DREAM, as described in ref. 35. Seventy-six out of the 211 DDR genes were found in both and were used for the analysis. DDR genes bound by DREAM that are upregulated in lin-52 mutants are shown in red; downregulated genes are shown in blue. NES, normalized enrichment score. j, Overlap between the upregulated DDR genes in lin-52 mutants and the genes that were found to be bound by the DREAM complex in the promoter area (41 in promoter area, 43 in total). This is a re-analysis of work in ref. 35. k,l, Overlap between all the DDR genes in C. elegans (k) and those upregulated in lin-52 mutants (l) and the genes that were enriched in the germline in ref. 46. m, GSEA of the RNA-seq of lin-52(n771) with the genes enriched in the germline46. Results from two-sided Fisher’s exact test are shown for overlap analyses. All GSEA statistics were done as described in ref. 77.

Table 1 The DREAM complex binds and represses DDR gene expression

We also determined by quantitative PCR (qPCR) that DDR genes were upregulated in other DREAM-complex mutants (Fig. 3f). A total of 53 DDR genes were significantly upregulated (adjusted P < 0.05) in lin-52 mutants compared with WT worms (Table 1). Among these were genes involved in NER, ICL repair, BER, HRR, mismatch repair and NHEJ, suggesting that the DREAM complex represses components of all main DNA-repair pathways (Fig. 3g).

We next searched for DNA-repair genes in other DREAM complex transcriptomic data. The induced DNA-repair genes in lin-52 mutants showed a remarkably consistent induction in two transcriptome datasets of lin-35(n745) mutants44,45. Out of the 53 DDR genes induced in lin-52 mutants, 35 were also found in lin-35 L1s, and 44 in lin-35 L3 worms (Fig. 3h and Table 1). This overlap in two independent studies further substantiates that the DREAM complex regulates an extensive amount of DDR genes.

To address whether the induced DDR genes might be directly repressed by the DREAM complex, we analyzed a published chromatin immunoprecipitation and sequencing (ChIP–seq) dataset on the DREAM complex in late embryos35. Gene Set Enrichment Analysis (GSEA) of all the DDR genes found in the lin-52 RNA-seq data revealed a significant enrichment of DDR genes bound by DREAM (Fig. 3i and Supplementary Table 5), with 80 of the 211 DDR genes being bound by DREAM (76 in the promoter, Supplementary Table 6). Forty-three out of the 53 DDR genes that were significantly upregulated in lin-52 mutants were bound by the DREAM complex (41 in the promoter area, 2 intergenic or intronic) by at least 6 out of 7 DREAM components tested35 (Fig. 3j and Table 1). We further confirmed the direct binding of DREAM to DDR genes by analyzing a ModENCODE ChIP–seq dataset for DREAM in L3-stage worms44. GSEA of all the DDR genes in the lin-52 RNA-seq data showed an enrichment of upregulated DDR genes in the list of genes bound by DREAM (Extended Data Fig. 6a and Supplementary Table 5). Sixty-three DDR genes were bound by at least one component of the DREAM complex (Supplementary Table 7). Furthermore, 33 of the 53 DDR-related genes that were significantly upregulated in lin-52 mutants were bound by at least one member of the DREAM complex, and 16 were bound by all of them (Extended Data Fig. 6b and Supplementary Table 8).

These analyses reveal that the DREAM complex directly binds and represses multiple genes involved in the DDR. The consistency of the induction of DDR genes in DREAM mutants, the direct binding of DREAM components to DDR gene promoters across different studies and the DDR protein upregulation observed indicate that DREAM constitutively represses DDR genes in somatic cells.

Germline-like expression signature of DNA-repair genes

Considering that the DREAM complex represses gene expression in somatic cells, we wondered whether the upregulation of DNA-repair genes in DREAM-complex mutants would resemble expression patterns of the germline46. We found that 111 of the 211 DDR genes were enriched in the germline (Fig. 3k and Supplementary Table 9), including 40 of the 53 DDR genes that were significantly upregulated in lin-52 mutants (Fig. 3l and Supplementary Table 9). Germline-specific genes were strongly enriched among the upregulated genes in lin-52 mutants (Fig. 3m and Supplementary Table 5); 271 out of 671 of them were germline-enriched genes, whereas only 26 out of 464 genes downregulated in lin-52 mutants were germline-enriched (Extended Data Fig. 6c and Supplementary Table 2).

Finally, we analyzed available data from microarrays of lin-54-mutant embryos and germline tissue47. Consistently, in lin-54 mutant embryos, DDR genes were upregulated (Extended Data Fig. 6d and Supplementary Tables 5 and 10), and no DDR genes were downregulated. The GO analysis of the upregulated genes showed a highly significant over-representation of ‘Cellular response to DNA damage’ (Supplementary Table 11). By contrast, the germline of lin-54 mutants did not present any enrichment of DDR genes compared with WT germlines (Extended Data Fig. 6e and Supplementary Table 5). These data indicate that DNA-repair genes whose expression in WT animals is restricted to the germline are particularly upregulated in somatic tissues of DREAM mutants.

In conclusion, our analysis of transcriptomic and ChIP–seq data indicates that the DREAM complex represses genes in the soma that are usually expressed in the germline. These genes are highly enriched in DDR genes, and thus a mutation in lin-52 leads to a germline-like upregulation of DDR genes in the soma.

DREAM mutants confer resistance to various types of DNA damage

On the basis of the range of DNA-repair pathways induced in the soma of lin-52 mutants, we hypothesized that DREAM mutants would show resistance to a wide variety of DNA-damaging insults that require different repair machineries. Somatic cells in the very early embryo are highly replicative and, upon ionizing radiation (IR), repair the DSBs through HRR, which is initiated by the BRC-1–BRD-1 complex13,14,48. The hatching of larvae from IR-treated eggs was evaluated for WT; the lin-52(n771) mutant; HRR-deficient brc-1(tm1145); brd-1(dw1) and lin-52; brc-1; brd-1 mutants; NHEJ-deficient cku-70(tm1524) and lin-52; cku-70 mutants; and the lin-52; brc-1; brd-1; cku-70 mutant, deficient for HRR and NHEJ (Fig. 4a). The proportion of egg hatching upon IR in the lin-52(n771) mutant was significantly higher than in WT worms. As early embryos predominantly employ HRR instead of NHEJ to repair DSBs13,14,48, brc-1; brd-1 HRR-deficient mutant eggs were highly IR sensitive, whereas NHEJ-deficient cku-70 mutants had similar IR sensitivity to that of WT worms. The IR resistance conferred by mutant lin-52 in early embryos suggests that the repression of DNA-repair genes is a property of a somatic function of DREAM and is not just associated with cellular quiescence or terminal differentiation.

Fig. 4: Mutations in the DREAM complex confer DNA-damage resistance against multiple damage types.
figure 4

a, IR sensitivity dependent on HRR was tested in WT, lin-52(n771), HRR-deficient brc-1(tm1145); brd-1(dw1), NHEJ-deficient cku-70(tm1524), lin-52(n771); brc-1(tm1145); brd-1(dw1), lin-52(n771); brc-1(tm1145); brd-1(dw1); cku-70(tm1524) and lin-52(n771); cku-70(tm1524) worms. Data are shown as mean ± s.e.m., n = 4 independent experiments (n = 3 for lin-52; cku-70). Each independent experiment had three biological replicates. Two-tailed t-tests were used for statistical comparisons, and the most relevant comparisons are shown. b, IR sensitivity assay dependent on NHEJ repair in WT, lin-52(n771), NHEJ-deficient cku-70(tm1524) and lin-52(n771); cku-70(tm1524) worms. The graph is representative of n = 3 biological replicates in 1 of 3 independent experiments, each of which had 3 biological replicates. Data are shown as mean ± s.d. of each larval stage. Two-tailed t-tests were used to compare the fraction of the larval stages of lin-52 compared with WT, and lin-52; cku-70 compared with cku-70 (non-significant). c, Alkylation-damage assay of WT, lin-52(n771), alkylation-damage-sensitive polh-1(lf31) and double-mutant lin-52(n771); polh-1(lf31) worms. The graph is representative of n = 3 biological replicates in 1 of 3 independent experiments. Data are shown as mean ± s.d. of each larval stage. Two-tailed t-tests were used to compare the fraction of the larval stages of lin-52 with WT, and lin-52; polh-1 with polh-1. d, ICL assay of WT and lin-52 mutants upon cisplatin treatment. Because cisplatin was diluted in DMF, worms were also given the maximum dose of DMF that was given for the cisplatin treatments as additional control. The graph is representative of n = 3 biological replicates in 1 of 3 independent experiments. Data are shown as mean ± s.d. of each larval stage. Two-tailed t-tests were used to compare the fraction of the larval stages of lin-52 compared with WT. P > 0.05, not shown. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Detailed P values and more comparisons can be found in Supplementary Table 13, including results from Fisher’s exact test, which was used to analyze the distribution of the larval stages.

Mutated lin-52 rescued the IR sensitivity of brc-1; brd-1 double mutants to levels similar to that of WT animals at high doses, suggesting that, in the absence of HRR, a mutation in lin-52 leads to the induction of alternative DSB-repair pathways. lin-52; cku-70 mutants also had a higher survival of embryos than did cku-70 worms. However, when lin-52 mutants were deficient in both NHEJ and HRR, we could no longer observe an improvement compared with brc-1; brd-1 mutants. Therefore, NHEJ is required for the rescue of the embryonic survival of brc-1; brd-1 by lin-52. These results suggest that lin-52 mutants have highly efficient HRR and NHEJ pathways that can compensate for the absence of either of these pathways (Fig. 4a).

Similar to lin-52 mutants, the increased survival upon IR treatment was also observed in efl-1(se1) mutant embryos (Extended Data Fig. 7a). However, this could not be tested in other DREAM mutant strains, such as dpl-1(n2994) and lin-35(n745) mutants, owing to the decreased embryonic survival in these strains without damage induction49,50.

The somatic cells switch from HRR in the early embryo (where most cell divisions occur) to NHEJ from the late embryo and onwards13,14. We determined whether a deficiency in the DREAM complex would also render the worms resistant to DSBs in an NHEJ-repair-dependent fashion. WT, lin-52(n771) and NHEJ-deficient cku-70(tm1524) and lin-52; cku-70 L1 larvae were exposed to IR, and their developmental growth was assessed 48 h later. IR-treated lin-52 mutants had significantly improved developmental growth compared with WT worms, and this was dependent on NHEJ (Fig. 4b). We next analyzed two other strains that are sensitive to IR51, xpa-1(ok698) and polh-1(lf31). The lin-52 mutation significantly rescued the IR sensitivity in both mutants (Extended Data Fig. 7b). These results suggest that mutations in lin-52 enhance NHEJ-dependent DSB repair, resulting in augmented IR resistance in IR-sensitive strains that have the canonical NHEJ pathway intact.

We wondered whether adult DREAM mutants might also be IR resistant. Adult C. elegans are extraordinarily resistant to IR treatment, necessitating very high doses to induce premature death52. Both lin-52(n771) and efl-1(se1) worms at day 1 of adulthood that were treated with IR showed a mild but significant lifespan extension compared with WT animals (Extended Data Fig. 7c). Thus, mutations in DREAM enhance the organismal resistance to DSBs in the soma during embryonic and larval development as well as during adulthood.

Next, we evaluated alkylation damage, a complex DNA insult repaired by several mechanisms involving DNA methyltransferases, AlkB enzymes and BER53. L1 worms were exposed to methyl methanesulfonate (MMS), and their development was assessed 48 h later. The lin-52 mutation led to improved development following MMS treatment and suppressed the MMS hypersensitivity of translesion synthesis DNA polymerase eta (polh-1) mutant animals to levels comparable to that of WT worms (Fig. 4c).

Finally, we assessed the response to cisplatin, a commonly used antitumor drug that causes intra- and ICLs54 that are repaired by a wide range of repair pathways, including Fanconi complex proteins, HRR and NER55. L1 larvae were treated with cisplatin and the development of WT and lin-52(n771) mutant worms was evaluated 48 h later (Fig. 4d). A mutation in the DREAM complex significantly alleviated the growth retardation following cisplatin-induced DNA damage for all doses tested.

Taken together, these data show that lin-52 mutants are resistant to a wide array of DNA-damage types, and the mutation alleviates the DNA-damage sensitivity of various mutants in single repair systems.

DREAM inhibition boosts DNA-damage resistance in human cells

We next wondered whether inhibition of the highly conserved DREAM complex could provide a pharmacological approach to augment DNA-repair capacities in human cells. We analyzed ChIP–seq data from quiescent human cells56 and searched for bound DNA-repair genes, following similar criteria as for the analyzed datasets from C. elegans57. We selected genes bound by at least LIN9, p130 and E2F4 simultaneously, where the binding occurred in 5′ between 0 and −1,000 bp of the TSS. Among the 328 gene promoters bound by DREAM (Supplementary Table 12), 67 genes are classified by GO as ‘DNA repair’ (Fig. 5a and Supplementary Table 12). Thus, the DREAM transcription repressor complex directly binds to DNA-repair gene promoters, indicating that the DREAM-mediated DNA-repair gene regulation is conserved in C. elegans and humans.

Fig. 5: Inhibition of DREAM using DYRK1A inhibitors confers DNA-damage resistance in human cells and decreases DNA damage and apoptosis in the retinas of Ercc1-deficient mice.
figure 5

a, Overlap between the DNA-repair genes in humans (GO database released 1 January 2021) and the genes bound by DREAM that were described in ref. 56. A two-tailed Fisher’s exact test was used for statistical analysis. b, FC comparison of DREAM target genes56 whose expression levels are significantly changed upon harmine or INDY treatment (FDR < 0.01 in at least one of the datasets). DNA-repair genes are shown in orange. The upper-right quadrant of derepressed genes has a 2.39× over-enrichment. Detailed statistics are in Supplementary Table 13. c,d,f,g, Representative density plots of biological triplicates of U2OS cells, labeled with annexin V and 7-AAD, that were mock treated or that received harmine hydrochloride and/or UV (c), INDY and/or UV (d), harmine hydrochloride and/or MMS (f) or INDY and/or MMS (g). e,h, Percentage of apoptotic annexin-V-positive U2OS cells upon harmine (left) or INDY (right) treatment and UV (e) or MMS (h) treatment. Data are shown as mean ± s.d., n = 3 biological replicates. Two-tailed t-tests were used for statistical analysis between the populations under the same irradiation conditions. i, Representative images of WT and Ercc1−/− retinas with TUNEL staining (green) and DAPI (blue). j, TUNEL-positive cells in the ONL from WT and Ercc1−/− mice upon treatment with harmine. n = 3 (2 male/1 female), 5 (3 male/2 female), 7 (4 male/3 female) and 7 (4 male/3 female) mice from left to right. Data are shown as mean ± s.e.m. For statistical analysis, two-tailed t-tests were used to compare groups that received or did not receive harmine treatment. k, Representative images of WT and Ercc1−/− retinas with γH2AX staining (red) and DAPI (blue). The intensity of the red channel has been equally increased in all images for visualization purposes. The INL of the retina is encircled by the dashed yellow line. l, γH2AX signal per nucleus in the INL of the retinas of WT and Ercc1−/− mice upon treatment with harmine. γH2AX signal per nucleus from an image stack per mouse is shown. n = 904, 889, 1,123 and 841 total nuclei, from left to right, from 5 (3 male/2 female) 5(3 male/2 female), 6 (3 male/3 female) and 5 (3 male/2 female) imaged mice, respectively. Box midlines show the median, box limits show the top and bottom quartiles, and whiskers to 1.5 × interquartile range (IQR). For statistical analysis, two-tailed t-tests were used to compare groups that received or did not receive harmine treatment.

In mammals, DREAM components not only form the DREAM repressor complex, but can also associate in other complexes that induce transcription27,28,29. We therefore used chemical inhibitors of the DYRK1A kinase, which phosphorylates LIN52, a modification required for the assembly of the DREAM complex, thus allowing its specific abrogation30. We employed two potent but distinct chemical inhibitors of the DYRK1A kinase: the beta-carboline alkaloid harmine, which has been widely used as specific DYRK1A inhibitor31, and the benzothiazole derivative INDY, which has been established as a highly selective DYRK1A inhibitor32. As the DREAM complex represses gene expression in G0 cells, we serum-starved confluent U2OS cells to obtain quiescent cell populations.

To confirm that the DYRK1A inhibitors abrogated DREAM-mediated gene repression, we performed RNA-seq analysis of quiescent cells treated with either harmine hydrochloride or INDY. Among the significantly upregulated genes (FDR-adjusted P < 0.01), all the known motifs bound by DREAM in human cells56 were significantly overrepresented (Extended Data Fig. 8), thus substantiating that INDY and harmine treatment resulted in gene upregulation by inhibiting the DREAM complex. We plotted all the genes bound by the DREAM complex (Fig. 5a and Supplementary Table 12) that were significantly up- or downregulated upon harmine or INDY treatment (Fig. 5b, statistics in Supplementary Table 13), most of which were upregulated upon both treatments. Of the 67 DNA-repair genes bound by DREAM (Fig. 5a), 58 were upregulated upon harmine and 46 upon INDY treatment, and 45 were upregulated upon both treatments (Fig. 5b and Supplementary Table 12). These results indicate that the pharmacological inhibition of DYRK1A with harmine or INDY results in upregulation of DREAM target genes, including the majority of DREAM targets encoding DNA-repair genes.

To directly assess whether DYRK1A-inhibitor treatment could augment DNA-damage resistance, we exposed harmine- or INDY-treated quiescent cells to UV irradiation or the alkylating agent MMS. We measured the apoptotic response to the DNA damage using annexin V and 7-AAD analysis by flow cytometry (Fig. 5c–h). Both DYRK1A-inhibitor treatments resulted in a highly significant reduction in DNA-damage-induced apoptosis compared with mock-treated cells. In conclusion, pharmacological inhibition of the DREAM complex kinase DYRK1A increased the expression of DREAM-targeted DNA-repair genes and conferred resistance to distinct types of DNA damage, suggesting a highly conserved function of the DREAM complex in regulating DNA-repair capacities.

Harmine treatment reduces retinal DNA damage

In humans, mutations in the NER pathway can lead to premature aging syndromes, such as Cockayne syndrome (CS), which is characterized by cachectic dwarfism, impaired development of the nervous system, pigmentary retinopathy (photoreceptor loss), cataracts, deafness and feeding difficulties58,59. In mice, mutations in Ercc1, which encodes an NER component, lead to a strong progeroid phenotype and premature death60,61,62. In order to test in vivo whether the inhibition of DYRK1A through harmine treatment could alleviate progeroid pathologies, we analyzed the loss of photoreceptor cells in the retina of Ercc1−/− mice, a hallmark of CS that can also be observed in NER-deficient mice, by performing TUNEL staining of the outer nuclear layer (ONL) of the retina63. Ercc1−/− and WT mice were harmine- or mock-treated intraperitoneally, and after 2 weeks of treatment, retinal degeneration was analyzed. Ercc1−/− mice showed increased TUNEL-positive apoptotic cells in the ONL compared with WT mice. The increased apoptosis levels in Ercc1−/− mice were significantly reduced upon harmine treatment (Fig. 5i,j).

Furthermore, we wondered whether the DNA damage present in the retinas of Ercc1−/− mice might be mitigated upon harmine treatment. Although the ONL shows very little γH2AX staining, even upon the induction of DNA damage, the cells in the inner nuclear layer (INL) typically show a stronger pan-nuclear γH2AX signal64,65. We quantified the γH2AX signal per nucleus in the INL in order to assess whether the treatment with harmine could reduce retinal DNA damage. The strong γH2AX signal in Ercc1−/− mice was significantly reduced upon harmine treatment (Fig. 5k,l).

In conclusion, treatment with harmine reduces the overall DNA-damage accumulation in Ercc1-deficient retinas and decreases the photoreceptor loss in this CS model.


Given the complexity of the repair mechanisms that respond to the distinct types of DNA lesions, it has remained elusive whether a mechanism exists that regulates the overall repair capacities of an organism. We uncovered that the C. elegans DREAM complex represses DNA-repair gene expression in somatic tissues, thus curbing their repair capacity and consequently limiting developmental growth, organismal health and lifespan upon DNA damage. Our data indicate that pharmacological targeting of the assembly of DREAM using DYRK1A inhibitors could be applied to augment DNA repair in human cells and in progeroid mice.

DNA repair in the germline is superior to that in the soma, with the germline preserving the genome, whereas somatic tissues accumulate mutations at increased rates with age2,3,4,5. Our data suggest that DREAM mutants confer enhanced germline-like DNA-repair capacities to somatic tissues. In C. elegans, it is not yet clear what mechanism keeps DREAM active in the soma and inactive in the germline. In mammals, the phosphorylation of LIN52 and the pocket proteins p130 and p107 is key to allowing the formation or disassembly of the complex, respectively. In C. elegans, the regulation might rely uniquely on the phosphorylation of the pocket protein LIN-35, which could disrupt the complex66. LIN-35 can be phosphorylated and inactivated by CDK-4 and the D-type cyclin CYD-1 (ref. 67), which might inactivate DREAM in the germline. Understanding the mechanisms of DREAM inactivation might allow modulation of its activity to specifically enhance DNA repair.

The overall elevated DNA-repair gene expression also revealed a compensatory role between the distinct repair pathways. Abrogation of the DREAM complex enhances the removal of UV-induced DNA lesions, and even suppresses defects in GG-NER and TC-NER, but requires the presence of the NER machinery. These data suggest that, in DREAM mutants, activity of each NER sub-pathway is elevated, and their respective compensation is enhanced. DREAM mutants even suppress HRR deficiency through elevated NHEJ. The suppression of polh-1 mutants’ sensitivity to ICLs and alkylating damage suggests that the consequence of DREAM abrogation might not necessarily be more error-prone repair. In the absence of DREAM, DNA-damage-driven developmental growth delay and aging are alleviated.

DREAM is a highly conserved transcription repressor complex that regulates the induction and maintenance of cellular quiescence by repressing cell cycle genes across multiple species19,68. Similar to the somatic cells in C. elegans, quiescent mammalian cells have limited DNA-repair capacities18. For example, quiescent hematopoietic stem cells (HSCs) and hair follicle stem cells use error-prone NHEJ69, which has been shown to be accountable for the increased mutagenesis during HSC aging70. We established that pharmacological targeting of the DREAM complex could be applied to augment DNA-damage resistance in quiescent human cells. Considering that the conserved DREAM complex is highly active in post-mitotic cells and that neurons are particularly susceptible to DNA-repair defects, such as TC-NER defects in Cockayne syndrome, DSB-repair deficiencies in ataxia telangiectasia, or impaired SSB repair in cerebellar ataxia71, targeting of the DREAM complex might provide new therapeutic avenues for a range of congenital DNA-repair deficiencies in humans. The alleviation of photoreceptor apoptosis upon harmine treatment in progeroid mice suggests that DREAM inhibition might indeed have therapeutic potential for DNA-damage-driven degenerative disorders. Intriguingly, the DYRK1A kinase is overexpressed in Down syndrome and involved in neurodegeneration in individuals with this condition, as well as in people with Alzheimer’s, Parkinson’s or Pick’s disease72,73,74. It will be interesting to explore whether DREAM-complex inhibition could prevent DNA-damage driven conditions, such as stem-cell exhaustion, neurodegeneration and premature aging.

Our data establish the DREAM complex as a master regulator of DNA-repair-gene expression in somatic tissues and quiescent cells. Abrogation of the DREAM-mediated repression of DNA-repair genes elevates somatic repair capacities and enhances resistance to DNA damage. We propose that the DREAM complex restricts somatic DNA repair, and removal of this restriction confers germline-like DNA-repair capacities to the soma. Given the central role of nuclear genome stability in the aging process75, the inhibition of the DREAM complex might provide a valuable intervention targeting DNA damage, which is a root cause of age-related diseases. Moreover, the suppression of various DNA-repair defects, such as GG- and TC-NER and HRR, suggests that the DREAM complex might also provide therapeutic opportunity in congenital DNA-repair deficiency syndromes that cause developmental growth failure and premature aging.


C. elegans strains

All strains were cultured under standard conditions78 and were always incubated at 20 °C during the experiments. The strains used were N2 (Bristol; WT):

DREAM: MT8839 lin-52(n771) III, MT10430 lin-35(n745) I, MT15107 lin-53(n3368) I/hT2 [bli-4(e937) let-?(q782) qIs48] (I;III), MT8879 dpl-1(n2994) II, MT11147 dpl-1(n3643) II, JJ1549 efl-1(se1) V, BJS634 dpl-1(n2994) II; lin-52(n771) III.

SynMuv B and NuRD: RB951 lin-13(ok838) III, RB1789 met-2(ok2307) III, PFR40 hpl-2(tm1489), MT8189 lin-15(n765) X, MT14390 let-418(n3536) V.

NER: RB1801 csb-1(ok2335) X, FX03886 xpc-1(tm3886) IV, RB864 xpa-1(ok698) I, FX04539 csa-1(tm4539) II, BJS21 xpc-1(tm3886) IV; csb-1(ok2335) X, BJS631 lin-52(n771) III; csb-1 (ok2335) X, BJS629 lin-52(n771) III; xpc-1 (tm3886) IV, BJS630 lin-52(n771) III; xpc-1 (tm3886) IV; csb-1 (ok2335) X, BJS772 xpa-1(ok698) I; lin-52(n771) III, BJS825 csa-1(tm4539) II; lin-52(n771) III.

HRR: DW102 brc-1 (tm1145); brd-1(dw1) III, BJS890 brc-1 (tm1145) III; brd-1(dw1); lin-52(n771) III, BJS868 lin-52(n771) III; brc-1(tm1145) III; brd-1(dw1) III; cku-70(tm1524) III.

NHEJ: FX1524 cku-70(tm1524) III, BJS887 lin-52(n771) III; cku-70(tm1524) III.

MMS: XF132 polh-1(lf31) III, BJS722 polh-1 (lf31) III; lin-52 (n771) III.

UV-irradiation assay during somatic development

The effects of UV-B in worm development were analyzed as previously described79 after bleach synchronization79. UV-B irradiation was performed with a 310-nm PL-L 36W/UV-B UV6 bulb (Waldmann, 451436623-00005077). OP50 Escherichia coli was added to the plates, and worms were incubated for 48 h. Larval stages were determined using a dissecting microscope. For the strain MT15107 lin-53(n3368) I/hT2 [bli-4(e937) let-?(q782) qIs48] (I;III), lin-53(n3368) homozygotes were distinguished using a fluorescence microscope (Leica M165 FC) and assessing the larval stage of worms that did not express green fluorescent protein.

UV-irradiation assay for germline development

Synchronized late-L4 worms were irradiated with different doses of UV-B and allowed to recover for 24 h. Irradiated and mock-irradiated worms were transferred to a fresh seeded NGM plate to lay eggs for 4 h (5 worms per plate). Upon removal of the adults, the plates were incubated for 24 h, after which the number of eggs laid and percentage of eggs that survived and hatched were evaluated.

NHEJ-dependent IR-sensitivity assay

As previously described13,14, L1 worms repair DSBs mainly through NHEJ repair. To analyze worm sensitivity to IR in a NHEJ-dependent way, synchronized L1 worms were irradiated with different doses of IR using an IR-inducing cesium-137 source, and were left for 48 h to allow development. The different larval stages were determined using a dissecting microscope.

HRR-dependent IR-sensitivity assay

Early embryos highly rely on HRR to repair DSBs13,14. To study the capacity of the different strains to tolerate IR-induced DSBs during embryogenesis, day-1 adults were left to lay eggs on seeded NGM plates for no longer than 1.5 h. Upon removal of the adult worms, early eggs were irradiated using an IR-inducing cesium-137 source. After 24 h, the percentage of surviving embryos that hatched was evaluated using a dissecting microscope.

Alkylation-damage induction by using MMS

Synchronized L1 worms were incubated with different concentrations of MMS (Sigma, 129925) diluted in M9 buffer for 1 h. Worms were washed three times with M9 buffer, and plated in seeded NGM plates. After incubation for 48 h, worm development was evaluated with a dissecting microscope.

ICL induction by using cisplatin

Synchronized L1 worms were exposed to different concentrations of cisplatin in dimethylformamide (DMF) diluted in M9 buffer or were mock-treated with DMF (Sigma, 227056)-diluted M9 buffer for 2 h. Worms were washed three times with M9 and incubated for 48 h in NGM plates, and the larval stages were quantified using a dissecting microscope.

Lifespan assay

Synchronized day-1 adult worms were irradiated or mock-irradiated with UV-B light using a 310-nm PL-L 36W/UV-B UV6 bulb (Waldmann, 451436623-00005077) or with IR using an IR-inducing cesium-137 source, and then were placed on fresh OP50-seeded NGM plates. At the beginning of the experiment, the worms were transferred to new plates every other day to avoid progeny overgrowth. Worms presenting internal hatching or protruding or ruptured vulvas were censored and removed from the experiment, and worms were scored as dead when no movement or pumping was observed even upon physical stimulus. Lifespan curves were analyzed with Graphpad Prism 7.03 log-rank test.

DNA-repair capacity assay in L1 worms

The quantification of DNA repair via immunostaining of CPDs and 6-4PPs of DNA samples in a slot blot was performed as has been described, with slight changes79. Bleach-synchronized L1 worms (at least 30,000 per plate) were irradiated with UV-B light and split in two groups, one to be immediately quick-frozen in liquid nitrogen, to serve as controls with unrepaired damage, and the other one was left in seeded plates for 24 h to allow for DNA repair to occur. After this, worms were washed 5 times, incubated for 2 h to permit the removal of intestinal bacteria, washed another 5 times and quick-frozen.

DNA extraction was performed using the Gentra Puregene Tissue Kit (Qiagen, 158667) and the protocol for DNA purification from tissue. The protocol was adapted to increase the volumes, but we still used a 1.5-ml Eppendorf tube to aid the supernatant extraction and pellet formation. That is, instead of the specified amounts, we used 500 µl cell lysis solution, 2.5 µl Puregene Proteinase K, 2.5 µl RNase A solution, 170 µl protein precipitation solution, 500 µl isopropanol and 500 µl 70% ethanol at the respective steps in the protocol. Cell lysis solution was directly added to the thawed sample, and an additional step with Proteinase K was performed. The DNA concentration was measured using the Qubit dsDNA HS Assay Kit (Invitrogen, Q32851). Serial dilutions of the DNA were denatured at 95 °C for 5 minutes (min) and transferred onto a Hybond nylon membrane (Amersham, RPN119B) using a Convertible Filtration Manifold System (Life Technologies, 11055). DNA crosslinking to the membrane was achieved by incubating the membrane at 80 °C for 2 h. The membrane was blocked for 30 min in 3% milk/phosphate-buffered saline (PBS)-T (0.1%) at room temperature. The membrane was incubated overnight at 4 °C with anti-CPD antibodies (Clone TDM-2, 1:10,000, Cosmo Bio, CAC-NM-DND-001) or anti-6-4 PP antibodies (Clone 64M-2, 1:3,000, Cosmo Bio, CAC-NM-DND-002), then washed three times with PBS-T (5 min at room temperature), and blocked for 30 min with 3% milk/PBS-T. The secondary antibody was a goat anti-mouse AffiniPure peroxidase-conjugated secondary antibody (1:10,000, Jackson Immuno Research, 115-035-174). Addition of the secondary antibody was followed by three washes in PBS-T and incubation with ECL Prime (Amersham, RPN2232). The DNA lesions were visualized by using a Hyperfilm ECL (Amersham, 28906836).

In order to quantify the total amount of DNA per sample, the membrane was incubated overnight at 4 °C in PBS with 1:10,000 SYBR Gold Nucleic Acid Stain (Invitrogen, S11494), then washed in PBS at room temperature and imaged using a BIO-RAD Gel Dox XR + Gel Documentation System (BIO-RAD, 1708195).

Adult somatic and germline DNA-repair assay

Synchronized day-1 adult worms were irradiated or mock-irradiated with a 310-nm UV-B light Philips UV6 bulb in a Waldmann UV236B irradiation device. Half of the worms were left in seeded NGM plates for 60 h to allow DNA repair to occur, whereas the others were collected directly after the irradiation.

After irradiation or incubation, worms were picked and placed in a drop of M9 buffer on top of a HistoBond+ Adhesion Microscope Slide (Marienfeld, 0810461). Using a hypodermic needle, we cut the worms close to the head, which also releases one of the germline arms, and then placed a coverslip over the slide and kept it at −80 °C for at least 30 min. After this, the coverslip was removed quickly to perform freeze-cracking80. Worms were fixated in liquid methanol at −20 °C for 10 min, then washed for 5 min in PBS. Seventy microliters of 2 M HCl were added on top of the worms for 30 min at room temperature to denature the DNA. Slides were washed three times with PBS, and blocked with 70 µl of 20% fetal bovine serum (FBS) in PBS for 30 min at 37 °C. The slides were incubated with 70 µl of 1:10,000 anti-CPD antibodies (Clone TDM-2, 1:3,000, Cosmo Bio, CAC-NM-DND-001) in PBS containing 5% FBS, at 4 °C overnight in a humid chamber. After subjecting the slides to three PBS washes, 5 min each, 70 µl of secondary anti-mouse Alexa Fluor 488-conjugated antibody (1:300, Invitrogen, A21202) in 5% FBS PBS was added for 30 min at 37 °C. The slides were washed three times, 5 min each, and mounted using 5 µl of Fluoromount-G with DAPI (Invitrogen, 00495952). Images were obtained using a SP8 Confocal Microscope by Leica using LAS X 3.5.7 software.

Image quantification of CPDs

Image stacks of the heads and germlines of adult worms were analyzed using the analysis software Imaris 9.9 (Oxford Instruments). Nuclei in the area anterior to the pharyngeal-intestinal valve and germlines were determined by using DAPI staining and setting a threshold of size and intensity. False-positive nuclei (due to bacteria in the pharynx) were manually discarded. The CPD signal was quantified using the maximum spherical volume fitting inside each of the nuclei. Owing to variable background signal in the germlines, background intensity was subtracted in these samples.

Cell culture and treatments

U2OS (ATCC, HTB-96) were cultured in DMEM, high-glucose GlutaMAX supplement, pyruvate (Thermo Fisher Scientific, 31966047) with 10% fetal bovine serum (FBS; Biochrom, S0615) and 1% penicillin–streptomycin (Thermo Fisher Scientific, 15140112). Cells were kept at 37 °C in a 5% CO2 incubator (Binder). Cell dissociation was performed with Accutase (Sigma, A6964). To promote quiescence, cells were cultivated in FBS-free medium for 48 h before genotoxic treatment. After 24 h of culture in FBS-free medium, cells were mock treated or received harmine hydrochloride (diluted in water) or INDY (diluted in DMSO) (Sigma, SMB00461 and SML1011) at 10 or 25 µM, respectively. Before the genotoxic treatment, cells were washed with FBS-free medium. For the UV treatment, medium was removed from the plates and cells were irradiated using 254-nm UV-C light Philips UV6 bulbs with 2 mJ/cm2. The MMS treatment was performed by adding MMS at 2 mM for 2 h, followed by 3 washes with FBS-free medium. Then, FBS-free medium was added. Quantification via flow cytometry of cell death and apoptosis was performed 24 h after genotoxic treatment.

Flow cytometry analysis

Collected cells were incubated in annexin V binding buffer (BioLegend, 422201) with Pacific Blue annexin V (BioLegend, 640917) and 7-AAD (Thermo Fisher Scientific, 00699350) at 4 °C for 15 min. Cells were measured using a MACSQuant VYB (Miltenyi Biotec) using MACSQuantify software 2.13.0 and analyzed using FlowJo v10.7.1 (BD). The gating strategy can be found in Supplementary Figure 1.

Animal handling

All animals were maintained in their breeding cages on a 12-h light/dark cycle. Mice were kept on a regular diet and had access to water ad libitum. Body weight was measured weekly. Animals were housed in a temperature- (18–23 °C) and humidity-controlled (40–60%), pathogen-free animal facility at the Institute of Molecular Biology and Biotechnology (IMBB), which operates in compliance with the ‘Animal Welfare Act’ of the Greek government, using the ‘Guide for the Care and Use of Laboratory Animals’ as its standard. All experiments were performed under the Animal license 6ΛΤΑ7ΛΚ-ΚΚΘ, issued by the Veterinary Medicine Directorate of Greek Republic.

Mice experiments

Male and female FVB/nj:C57BL/6j Ercc1–/– and their respective control WT mice81, on the third day after birth (postnatal day P3), were injected intraperitoneally 3 times per week with 10 mg/kg body weight of harmine hydrochloride (SMB00461, Sigma) diluted in 0.9% sodium chloride. Mice were euthanized at postnatal day P15 for retina tissue isolation. Tissues were embedded in optimal cutting temperature (OCT) compound, cryosectioned and stained using the in situ cell death detection kit (TUNEL staining) (11684817910, Roche), according to the manufacturer’s instructions.

For the immunostaining experiments against γH2AX (Millipore, 05–636), retina slices were fixed in 4% formaldehyde in 1× PBS for 10 min at room temperature, permeabilized with 0.5% Triton X-100 in 1× PBS for 10 min, on ice, and blocked with 1% BSA in 1× PBS for 1.5 h at room temperature. After overnight incubation with the primary antibody (1:12,000, in 1% BSA/1× PBS, 4 oC), a secondary fluorescent antibody was added (goat anti-mouse IgG-Alexa Fluor 555, 1:2,000, Invitrogen, A-21422) and DAPI (1:20,000, Thermo Fisher Scientific, 62247) was used for nuclear counterstaining.

Samples were visualized with an SP8 TCS laser scanning confocal microscope (Leica). The detection of nuclei and signal intensity from retinas was performed utilizing Imaris 9.9 (Oxford Instruments).

RNA extraction for RNA-seq and qPCR experiments

For the qPCR and RNA-seq of L1 worms, around 10,000 (qPCR) or 40,000 (RNA-seq) bleach-synchronized L1 worms in triplicates (qPCR) or quadruplicates (RNA-seq) per strain and condition were placed in seeded NGM plates for 3 h. They were mock-treated or UV-B irradiated, and left for 6 h to allow the DNA-damage-related transcriptional changes to take place. Worms were collected and washed three times with M9 buffer, and the pellet was placed in a tube containing 1 ml TRIzol (Invitrogen, 15596018) and 1 mm zirconia/silica beads (Biospec Products, 11079110z).

To extract the RNA, worms were disrupted with a Precellys24 (Bertin Instruments, P000669-PR240-A), and the RNA isolation was performed by using the RNeasy Mini Kit (QIAGEN, 74106) following the manufacturer’s specifications, except we used 1-bromo-3-chloropropane (Sigma, B9673) instead of chloroform. The RNA was quantified using NanoDrop 8000 (Thermo Fisher Scientific, ND-8000-GL).

RNA extraction from U2OS cells was performed after 24 h of harmine or INDY treatment of cells that had been starved for a total of 48 h by using the RNeasy Mini Kit (QIAGEN, 74106), following the manufacturer’s specifications. Cells were disrupted with RLT buffer and homogenized with QIAshredder spin columns (QIAGEN, 79656).


Reverse transcription to form complementary DNA (cDNA) was performed using Superscript III (Invitrogen, 18080044). The obtained cDNA was used to perform qPCR by using SYBR Green I (Sigma, S9460) and Platinum Taq polymerase (Invitrogen, 10966034) in a BIO-RAD CFX96 real-time PCR machine (BIO-RAD, 1855196). The analysis of the results was performed by using the comparative CT method82.

All C. elegans qPCR experiments were done in biological triplicates, and the data were normalized to three housekeeping genes. qPCR CT values were obtained using Bio-Rad CFX Manager 3.0.

C. elegans qPCR primers

Primers used for PCR were as follows:

Housekeeping genes:

Forward primer

Reverse primer










Genes of interest:
























A triplicate of RNA samples from lin-52 mutant and WT L1 worms were rRNA-depleted using Ribo-Zero Plus rRNA Depletion Kit (Illumina, 20037135) and sequenced using a Hiseq4000 (Illumina) with PE75 read length. For RNA quality control, the RNA integrity number was ≥9.4 for all samples. RNA-seq data were processed through the QuickNGS pipeline83, Ensembl version 85. Reads were mapped to the C. elegans genome using Tophat84 (version 2.0.10) and abundance estimation was done using with Cufflinks85 (Version 2.1.1). DESeq2 (ref. 76) was used for differential gene expression analysis.

The human RNA-seq data were processed with Salmon-1.1 (ref. 86) against a decoy-aware transcriptome (gencode.v37 transcripts and the GRCh38.primary_assembly genome) with the following parameters: –validateMappings –gcBias –seqBias. The output was imported and summarized to the gene-level with tximport (1.14.2)87, and differential gene analysis was done with edgeR (3.28.1)88.


WT and lin-52(n771) L1 worms were plated in OP50-containing NGM plates and left to feed for 9 h. Worms were collected and washed 5 times with M9 buffer to remove the OP50, and 8 M urea buffer mixed in 50 mM TEAB with 1× Protease Inhibitor cocktail (Roche) was added to the sample before quick freezing.

Chromatin was degraded using a Bioruptor (Diagenode) for 10 min with cycles of 30/30 seconds. Upon centrifugation, the concentration of protein in the supernatant was calculated using Qubit Protein Assay Kit (Thermo Fisher Scientific). Twenty-five micrograms of protein per sample were transferred to a new tube, dithiothreitol was added to a concentration of 5 mM followed by vortexing and incubation at 25 °C for 1 h. Chloroacetamide was added to a final concentration of 40 mM, and the samples were incubated at room temperature for 30 min. Protein digestion with lysyl endopeptidase was done at an enzyme:substrate ratio of 1:75 and incubated at 25 °C for 4 h. Samples were diluted with 50 mM TEAB to reach a urea concentration of 2 M, and trypsin protein digestion was performed by adding trypsin at an enzyme:substrate ratio of 1:75; the samples were kept at 25 °C overnight.

After protein digestion, SDB RP StageTip purification was performed89. The protein samples were then analyzed utilizing liquid chromatography–mass spectrometry by the CECAD Proteomics Facility on a Q Exactive mass spectrometer that was coupled to an EASY nLC 1000 (Thermo Fisher Scientific). The differential protein levels were obtained by CECAD’s Proteomics Facility. Briefly, a predicted spectrum library was generated using the Prosit webserver90, and data were processed using DIA-NN 1.7.16 (ref. 91) and imported into Perseus (ref. 91) for analysis.

Four replicates per strain and condition were used, each containing around 20,000 worms.


The list of 211 genes belonging to the GO term ‘Cellular response to DNA damage stimulus’ was obtained by using data from the GO Consortium92,93 (database released on 8 October 2019) (Supplementary Table 1). GO analysis was performed using the PANTHER 15.0 over-representation test and Venn Diagrams were created using Venn Diagram Plotter 1.5 and GIMP 2.10.12.

Gene IDs from previously published datasets were updated to current databases. Duplicated or dead IDs were eliminated accordingly. Overlap analysis were done by using Fisher’s exact test in R v3.6.3 (ref. 94). Gene set enrichment analysis (GSEA) was done in R v3.6.3 (ref. 94) with the GSEA function of clusterProfiler v3.14.3 (ref. 95) and the parameter settings minGSSize = 3, maxGSSize = 5000, and nPerm = 20000. To calculate the adjusted P values for the GSEA results, statsmodels96 v0.11.1 multipletests methods with the parameter method = ’fdr_bh’ or method = ’bonferroni’ in Python 3.6 (ref. 97) was used.

Promoter analysis: C. elegans

The set of 211 DDR genes was used as input for the findMotifs function of HOMER-4.11-2 (ref. 98) with the parameters -len 8,10 -start −1000 -end 0. Wormbase IDs were converted to the sequence name with WormBase’s SimpleMine99. These identifiers were searched in the ‘worm.description’ file of HOMER to gain the corresponding RefSeq IDs. The P values were calculated with the hypergeometric tests function in scipy(1.5.1). HOMER’s seq2profile function98 was used to convert the previously reported CDE + CHR DREAM complex motif35 with one mismatch and three random base pairs in between to a motif file usable by HOMER with the following parameter: BSSSSSNNNTTYRAA 1 (ref. 35). The constructed motif was searched with the -find function for the 211 DDR genes with the parameters -start −1000 -end 0. The background enrichment of the motif was calculated for all 20,174 protein-coding genes with a RefSeq ID included in the worm.description file of HOMER. The P values were calculated with the hypergeometric tests function in scipy (1.5.1)100.

Promoter analysis:human

Homer’s seq2profile function was used to convert previously reported DREAM complex motifs56, with no allowed mismatch, with the following parameters:




  • CWCGYG 0

The motifs were searched with the -find function with the parameters -start −1000 -end 0 for the up and downregulated genes, after harmine and INDY treatments, respectively, with an FDR cutoff of 0.01.

The background enrichment of the motif was calculated for all protein-coding genes included in the homer.description file of HOMER.

The P values were calculated with the hypergeometric test function in scipy(1.5.1)100 and Python’s statsmodels (0.11.1)96 was used to calculate the Benjamini–Hochberg FDR.

Statistics and reproducibility

In C. elegans, development growth assays and egg-laying assays were performed a minimum of 3 times, each of which included 3 biological replicates per condition with an average of around 40–50 individuals per sample. Slot blots were performed at least three times. Flow cytometry assays were done at least three times, each having three biological replicates. For experiments from which data were obtained from single individuals, such as mice experiments, worm imaging studies, and lifespan assays of worms, the sample sizes are indicated. All attempts at replication were successful.

The sample sizes have been well established in similar experiments in other scientific publications (refs. 9,13,39,79, among others). The statistical analysis performed in each experiment can be found in the figure legend. Two-tailed t-tests were done in Microsoft Excel 2019 and GraphPad Prism 7.03. log-rank, Mann–Whitney and two-way ANOVA tests for the germline image quantification tests and unpaired t-tests with Welch’s correction were done in GraphPad Prism 7.03. Two-way ANOVA for the quantification of worm heads was done with Python’s pingouin v0.3.6. Two-tailed Fisher’s exact tests were done in RStudio 1.2.5019. C. elegans experiments were not randomized was not applied because the group allocation was guided on the basis of the genotype of the respective mutant worms. Worms of a given genotype were nevertheless randomly selected from large strain populations for each experiment without any preconditioning. In mice experiments, allocation was random.

Blinding was generally not applied, as the experiments were carried out under highly standardized and predefined conditions to avoid investigator-induced bias. Developmental assays upon DNA damage with small observed effects were performed blinded to exclude any bias. This affects the developmental growth upon IR, MMS and cisplatin treatment.

Starvation assay

A pool of synchronized starving lin-52 mutant and WT L1 worms was maintained in M9 buffer rolling at 20 °C. From this pool, around 30 worms were transferred to seeded NGM plates over consecutive days, and the number of L1 worms per plate was counted. After 48–72 h in the seeded plates, the number of worms that recovered and survived the ongoing starvation was evaluated. A biological triplicate for each strain was used for each timepoint. The experiment was discontinued after 14 days because all worms from days 13 and 14 had died. Each plate per condition, replicate and day had an average of 30 worms.

Total-egg-hatching assay

Single synchronized day 1 adult worms were transferred to seeded NGM plates and left to lay eggs for 24 h. Each day, the worms were transferred to a new seeded NGM plate, until egg laying stopped for all individuals. The total number of eggs laid and hatching/surviving eggs were evaluated 24 h after the removal of the adult from the plate. Twelve adult WT and lin-52 mutant worms were used. Internal hatching or exploding worms were excluded from the day the event occurred onwards.

Motility assay

Synchronized day 1 adult lin-52 mutant or WT worms were UV-irradiated or mock-treated and incubated at 20 °C for 72 h. Next, 30 worms were transferred to unseeded small NGM plates and left for 30 min to avoid worms accumulating in areas with food, promote movement and facilitate image analysis. To obtain video footage, the plates containing the worms were left under the microscope light without the lid for 30 seconds to allow the worms to get used to the conditions. Thirty-second videos were taken by using a Zeiss Axio Zoom V.16 and Zeiss ZEN 2.3 pro software. Worm footage was analyzed using the plugin wrMTrck in ImageJ 1.53q.

EdU-incorporation assay in L1 and adult worms

Thymidine-deficient E. coli (strain MG1693) were grown in M9 containing 1 % glucose, 1 mM MgSO4, 1.25 µg/ml vitamin B1, 0.5 µM thymidine and 20 µM 5-ethynyl-2′-deoxyuridine (EdU) at 37 °C overnight in darkness. These bacteria were used to seed M9-agar plates (M9 with 1.2% agar and 0.6% agarose) and were left to incubate overnight at room temperature.

Synchronized L1 or adult lin-52 mutant and WT worms were UV-irradiated or mock-treated and transferred to the plates containing the EdU-labeled MG1693 E. coli. Worms were collected after 6 h, 12 h and 24 h for L1 worms, and after 24 h for adults, washed three times in M9 buffer and transferred to fixing buffer (1× egg buffer with 0.1% Tween and 3% PFA). Fixed worms were placed on top of a HistoBond+ Adhesion Microscope Slides (Marienfeld, 0810461), adult worms were cut open, and a coverslip was placed above the worms and slight pressure was applied. Next, the slides were placed on dry ice to allow freeze-cracking80. We used the Click-iT EdU imaging kit (Invitrogen, C10337), following the manufacturer’s instructions, for the preparation of the Click-iT reaction cocktail. Upon washing the worms 3 times for 5 min in PBS, 50 µl of reaction cocktail was added to the slides, followed by incubation for 30 min at room temperature in darkness. Slides were washed once in3 % BSA in PBS and mounted using 5 µl of Fluoromount-G with DAPI (Invitrogen, 00495952). Images were obtained using a SP8 Confocal Microscope by Leica using LAS X 3.5.7 software. Positive EdU nuclei were counted manually from the obtained image stacks.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.