Cross-species incompatibility between a DNA satellite and a chromatin protein poisons germline genome integrity

Satellite DNA spans megabases of eukaryotic genome sequence [1]. These vast stretches of tandem DNA repeats undergo high rates of sequence turnover, resulting in radically different satellite DNA landscapes between closely related species [2–4]. Such extreme evolutionary plasticity suggests that satellite DNA accumulates mutations with no functional consequence. Paradoxically, satellite-rich genomic regions support essential, conserved nuclear processes, including chromosome segregation, dosage compensation, and nuclear structure [5–10]. A leading resolution to this paradox is that deleterious alterations to satellite DNA trigger adaptive evolution of chromatin proteins to preserve these essential functions [11]. Here we experimentally test this model of coevolution between chromatin proteins and DNA satellites by conducting an evolution-guided manipulation of both protein and satellite. We focused on an adaptively evolving, ovary-enriched chromatin protein, called Maternal Haploid (MH) from Drosophila. MH co-localizes with an 11 Mb 359-bp satellite array present in Drosophila melanogaster but absent in its sister species, D. simulans [12]. Using CRISPR/Cas9-mediated transgenesis, we swapped the D. simulans version of MH into D. melanogaster. We discovered that D. melanogaster females encoding only the D. simulans mh (“mh[sim]”) do not phenocopy the mh null mutation. Instead, MH[sim] is toxic to D. melanogaster ovaries—we observed elevated ovarian cell death, reduced ovary size, and subfertility in mh[sim] females. Using both cell biological and genetic approaches, we demonstrate that MH[sim] poisons oogenesis through a DNA damage pathway. Remarkably, deleting the D. melanogaster-specific 359 satellite array from mh[sim] females completely restores female germline genome integrity and fertility. This genetic rescue offers experimental evidence that rapid evolution resulted in a cross-species incompatibility between the 359 satellite and MH. These data suggest that coevolution between ostensibly inert repetitive DNA and essential chromatin proteins preserves germline genome integrity.


32
Satellite-enriched genomic regions evolve rapidly and yet support strictly conserved nuclear 33 functions [2][3][4][5][6][7][8][9][10]. A classic resolution to this paradox is that DNA satellite-associated proteins 34 evolve adaptively to mitigate deleterious changes to DNA satellite sequence [11]. Repeated 35 bouts of DNA satellite evolution and host chromatin protein adaptation result in exquisitely 36 coevolved satellites and satellite-associated chromatin proteins. This model of coevolution 37 predicts pervasive incompatibilities between satellite DNA and chromatin proteins from closely 38 related species: adaptively evolving chromatin proteins from one species should fail to package 39 or process DNA satellites from another [11,13,14].

41
Evidence for this coevolution model has emerged from engineering "evolutionary mismatches" 42 between the chromatin protein(s) of one species and the DNA satellite landscape of a close 43 relative. Under one approach, a diverged chromatin protein is introduced into a closely related 44 species, generating an evolutionary mismatch between the manipulated protein and one or 45 more DNA satellites [14][15][16][17]. Consistent with disrupted DNA satellite:chromatin protein 46 coevolution, the naïve chromatin protein typically perturbs a satellite-mediated function, such as 47 chromosome segregation or nuclear organization [14,16,17]. In these cases, however, the 48 incompatible DNA satellites are unknown. A second approach crosses sister species to 49 generate evolutionary mismatches between chromatin proteins and DNA satellites. Consistent 50 with disrupted DNA satellite:chromatin protein coevolution, interspecies hybrid inviability has 51 been linked to DNA satellites [18,19]. In these cases, however, the incompatible chromatin 52 proteins are unknown. To date, there are no cases of experimental identification of both 53 chromatin protein and satellite putatively engaged in coevolution.

55
To experimentally probe both sides of the coevolution model, we searched for a rapidly evolving 56 DNA satellite that colocalizes with an adaptively evolving chromatin protein. In Drosophila 57 melanogaster, the 359-bp satellite occurs in an 11 Mb array at the base of the X chromosome 58 [20,21]. Close relatives of D. melanogaster, including D. simulans and D. erecta, lack this 59 satellite array on the X chromosome [22]. Instead, these close relatives of D. melanogaster 60 have shorter arrays of "359-like" sequence dispersed throughout heterochromatin and 61 euchromatin [3,23,24] Figure 1A). The dynamic evolution of the 359 satellite and 77 adaptive evolution of a 359-associated protein, MH, raise the possibility that mh recurrently 78 evolves to preserve a biological function compromised by 359 satellite proliferation.

80
To test the possibility of MH-359 coevolution, we first conducted an evolution-guided 81 manipulation of mh. We predicted that generating an "evolutionary mismatch" between the 82 D. simulans mh and the D. melanogaster 359bp X-linked array would compromise an essential 83 nuclear function. To perform this swap, we used CRISPR/Cas9 to integrate into the native mh 84 locus either a 3xFLAG-tagged D. melanogaster mh coding sequence (our control fly, "mh[mel]") 85 or a 3xFLAG-tagged D. simulans mh coding sequence (our experimental fly, "mh[sim]", Figure   86 1B). Both the D. melanogaster and the D. simulans coding sequences were codon-optimized for 87 D. melanogaster. We observed equivalent expression of the two transgenes ( Figure S1A).

93
An oogenesis defect was unexpected: previous reports suggest that mh null alone yields no 94 ovary phenotype [12,25]. Consistent with these data, we detected no difference in ovary size of 95 mh null mothers compared to heterozygous controls ( Figure S1B).

103
( Figure S1C). These data, combined with the unexpected oogenesis phenotype, suggest that 104 mh [sim] does not behave as a loss-of-function allele. Instead, MH[sim] might be toxic.

106
To explore the possibility that MH [sim]

138
Applying these phenotypic data to the coevolution model, we hypothesized that the expanded should restore germline genome integrity and fertility of mh [sim] females. To directly test this 142 prediction, we took advantage of a fly strain that lacks the 11 Mb array of X-linked 359 satellite 143 [33]. We recombined this 359-deletion, called Zygotic hybrid rescue, or "Zhr," onto both the

154
The molecular events that trigger 359-dependent DNA damage in mh[sim] ovaries remain to be 155 elucidated. We suspect that MH[sim] interferes with a chromatin-mediated pathway required to

178
The 359-mediated toxicity to oogenesis highlights the catastrophic functional consequences of 179 DNA satellite evolution. Importantly, 359-mediated toxicity is also apparent in D. melanogaster-

222
We crossed the single males, injected as embryos, to an FM7 (X-chromosome balancer) 223 female. We screened F1 females to identify positive transformants using forward primer  replacements. Primers can be found in Table S1.

237
We used the ΦC31 integrase-mediated transgenesis system to introduce into the same landing 238 site mh from D. melanogaster or D. simulans downstream of an "upstream activating sequence" 239 or "UAS" [47]. Using the HDR plasmids as a template (see above), we PCR-amplified the

253
To generate stocks that encode both the X-linked mh-transgene and the X-linked 359 satellite 254 deletion (Zhr 1 , BDSC #25140), we first generated trans-heterozygote females. We crossed 255 these trans-heterozygote females to FM7 males and used PCR to assay individual recombinant

294
We conducted immunofluorescence on ovaries following the protocol described in [49]. We

301
We conducted immunofluorescence on embryos from mh [mel]

331
We thank Isabella Farkas and Courtney Christopher for technical assistance. We also thank the              Figure S1 A B