Female-biased embryonic death from genomic instability-induced inflammation

Genomic instability (GIN) can trigger cellular responses including checkpoint activation, senescence, and inflammation 1,2. Though extensively studied in cell culture and cancer paradigms, little is known about the impact of GIN during embryonic development, a period of rapid cellular proliferation. We report that GIN-causing mutations in the MCM2–7 DNA replicative helicase 3,4 render female mouse embryos to be dramatically more susceptible than males to embryonic lethality. This bias was not attributable to X-inactivation defects, differential replication licensing, or X vs Y chromosome size, but rather “maleness,” since XX embryos could be rescued by transgene-mediated sex reversal or testosterone (T) administration. The ability of exogenous or endogenous T to protect embryos was related to its anti-inflammatory properties 5. The NSAID ibuprofen rescued female embryos containing mutations not only in MCM genes but also Fancm, which like MCM mutants have elevated GIN (micronuclei) from compromised replication fork repair 6. Additionally, deficiency for the anti-inflammatory IL10 receptor was synthetically lethal with the Mcm4Chaos3 helicase mutant. Our experiments indicate that DNA replication-associated DNA damage during development induces inflammation that is preferentially lethal to female embryos, whereas male embryos are protected by high levels of intrinsic T.

DNA replication requires the heterohexameric minichromosome maintenance complex (MCM2-7), constituting the catalytic core of the replicative helicase. Reduction of MCMs causes RS by decreasing dormant ("backup") origins that are important for completing DNA replication when replication forks stall or collapse [11][12][13] . Mice bearing the Chaos3 allele of Mcm4 (abbreviated Mcm4 C3 ) have elevated micronuclei and are highly cancer prone 4 . This allele causes GIN by destabilizing the MCM2-7 helicase and triggering post-transcriptional pan-reduction (~40%) of Mcm2-7 mRNAs and protein 3  MEFs (mouse embryonic fibroblasts) bearing MCM mutations exhibit reduced dormant replication origins 14,15 . To test if dormant origin reduction contributes to the female-biased lethality, Mcm3 heterozygosity was introduced into the semilethal genotypes. Mcm3 heterozygosity ameliorates several deleterious phenotypes of MCM-deficient mutant mice and cells by increasing chromatin-bound MCMs (MCM3 participates in nuclear export of MCMs) 3 . This dramatically rescued viability of Mcm4 C3/Gt and Mcm4 C3/C3 Mcm6 Gt/+ female embryos preferentially, increasing female viability from 0% to 27% in the former, and from 3% to 42% in the latter (Fig. 1a, Tables S1,S2). Mcm3 heterozygosity also increased viability of Mcm4 C3/C3 Mcm2 Gt/+ newborns from 30% to 72%, but both sexes were rescued approximately proportionately ( Fig. 1a Table S6). These results indicate that maleness, and not the presence of two X chromosomes per se, protects embryos from MCM deficiency. These data are consistent with the finding that preferential female embryo death occurs after sex determination (E9.5-12.5).
Since sex reversal rescued XX lethality, we hypothesized that testosterone (T) might be responsible. It is produced at high levels by Leydig cells in embryonic testes from ~E12.5 onward 16 . We injected pregnant females daily with T beginning at E7.5, and found that the viability of XX Mcm4 C3/C3 Mcm2 Gt/+ E19.5 fetuses increased dramatically from 22% to 54% ( Fig. 2a; Table S7). We speculated that T might protect MCM-deficient embryos by increasing replication capacity, given a report that the androgen receptor stimulates proliferation of prostate cancer cells by acting as a replication factor 17,18 . However, we observed no increase of Mcm mRNA or chromatin-bound MCMs in T-treated MEFs (Fig.   2b, c), and no sex-specific differences in MCM2 or MCM4 protein levels in E13.5 fetuses or placentae of various genotypes (Extended Data Fig. 2).
Next, we hypothesized that T was ameliorating certain consequences of GIN in the Mcm mutants. In particular, elevated micronuclei, the signature phenotype of Mcm4 Chaos3 mice 4 , can trigger inflammation via the cGAS-STING pathway 19 . T, a steroid hormone, suppresses the expression of pro-inflammatory cytokines while increasing the anti-inflammatory molecule IL10 5,20,21 . Indeed, T treatment of Mcm4 C3/C3 Mcm2 Gt/+ MEFs caused >2-3 fold decreases in mRNAs for the pro-inflammatory cytokine IL6 and also PTGS2 (COX2), which is central for production of prostaglandins that cause inflammation and pain (Fig. 2b). Strikingly, administration of ibuprofen to pregnant females (from 7. While these data indicate that GIN-driven inflammation underlies preferential female embryonic lethality, we considered the possibility that unrelated alterations in gene expression by ibuprofen and the androgen receptor (which is strongly induced by T; Fig. 2c) 22,23 were responsible. We therefore took the orthogonal approach of increasing inflammation by ablating the receptor (Il10rb) for the anti-inflammatory molecule IL10, hypothesizing that this would exacerbate Mcm4 C3/C3 Mcm2 +/− lethality or sex bias. Remarkably, the genotype of Mcm4 C3/C3 Il10rb −/− caused highly penetrant lethality to embryos of both sexes (Extended Data Table 1c). IL10 mediates a feedback loop under conditions of inflammation to induce degradation of Ptgs2/COX2 transcripts 24 , and also counters the inflammation response triggered by the STING pathway 25 . This synthetic lethality was rescued by treating pregnant dams with NSAID, increasing viability of Mcm4 C3/C3 Il10rb −/− offspring (both sexes) from 8.9% to 94% (Extended Data Table 1c).
Successful pregnancy requires suppression of inflammation at the maternal:fetal interface. Because homozygosity for Chaos3 alone causes a ~20 fold increase in micronucleated erythrocytes without decreasing viability in the C3H background 4 , and IL10 is thought to play a role in suppressing maternal inflammation at the fetal:maternal interface 26  males (all data presented heretofore were from reciprocal crosses). Surprisingly, this cross abolished the sex bias against Mcm4 C3/C3 Mcm2 Gt/+ females ( Fig. 3a; Extended Data Table  1b). We hypothesized that maternal homozygosity for Chaos3 imposes additional stress on the placentae of genetically susceptible female embryos, possibly via DNA damage-induced inflammation. We examined double strand break (DSB) levels (marked by γH2AX) in placentae of E13.5 embryos produced in various control and mutant reciprocal crosses. Regardless of fetal genotype, placentae from embryos within Mcm4 C3/C3 dams had more γH2AX-positive cells than when dams were of any other genotype (Fig. 4). NSAID treatment did not reduce the level of γH2AX staining, consistent with the rescue effect being related to inflammation, not GIN per se (Fig. 4). We therefore hypothesized that GINinduced placental inflammation might underlie the lethality in our mice. Consistent with this, we observed significant reductions in placental, but not embryonic MCM2 and MCM4 (especially MCM4) in Chaos3 mutant genotypes, regardless of maternal genotype or whether the dams were NSAID-treated (Extended Data Fig. 2a-c). Thus, placental cells may be particularly sensitive to DNA replication defects that trigger downregulation of MCM production and consequent increases in GIN and inflammation 27,28 29 ), allograft rejection, and interferon gamma response. All three of these categories contain genes involved in inflammation and the innate immune response (Extended Data Fig. 3). Overall, the results indicate that the combination of maternal and fetal GIN causes lethal levels of inflammation. However, it remains possible that the parental genotype-dependent, sex biased lethality may have an epigenetic component (i.e. imprinting).
While the data presented thus far demonstrate that MCM depletion (e.g. Mcm2 hemizygosity) in conjunction with a destabilized replicative helicase in Chaos3 mice trigger inflammation and embryonic death, it is unclear exactly what defects are primarily responsible, and whether the sex-bias phenomena are entirely unique to these models. We therefore attempted to parse the key proximal defects that trigger the sex bias by exposing WT embryos to either exogenous RS alone or DSBs alone. Pregnant females, treated with hydroxyurea to induce RS, delivered pups without significant sex skewing (M:F 1.08; Table  S8). Chronic exposure to ionizing radiation, which causes DSBs, also failed to produce a sex bias (M:F 1.00; Table S8). We then conjectured that replication-associated DNA damage that causes micronuclei might underlie the inflammation-driven lethality. Mice deficient for FANCM, involved in DNA replication fork repair, display elevated micronuclei 6 and underrepresentation of females 30 . We also observed a bias against  Table 1d).
Our results indicate that DNA damage caused by defective DNA replication and/or replication-associated repair cause a level of inflammation compromising female embryos lacking anti-inflammatory protection of testosterone. We hypothesize that since both genetic models tested have elevated micronuclei, a known trigger of the cGAS-STING cytosolic DNA sensing pathway, that this may precipitate lethal inflammation in a key compartment(s) of the embryo and/or uterine environment. Future experiments exploiting mouse mutants and mosaics will help resolve these questions, and guide studies into whether similar phenomena occur in humans.

Mice.
All breeding and husbandry all crosses were performed in the same animal facility and room at Cornell's Veterinary College (East Campus Research Facility), and under the same environmental conditions and health status. Use of mice was performed in compliance with all relevant ethical regulations, having been conducted under a protocol (0038-2004) approved by Cornell University's Institutional Animal Care and Use Committee (IACUC). Sample sizes for original sex skewing observations, since they were taken from historical colony breeding data, were not planned, and selection of individuals was entirely genotypebased, thus not randomized. Sexing of animals was done before genotyping, thus there was no blinding. Sample sizes with T and ibuprofen were also not pre-determined, as potential effect size was unknown yet proved to be dramatic.

Testosterone Injections and Sex Reversal.
Mcm4 C3/C3 Mcm2 Gt/+ males were mated to Mcm4 C3/C3 females and 100uL of a 3mg/ml solution of testosterone propionate (Sigma) was injected sub-cutaneously into the hind leg of pregnant females daily from E7.5 to E.16.5 (20µg/g/day). This dose has been shown to increase female fetal testosterone by 80% in a rodent model without serious toxicological effect 31 . The testosterone propionate was dissolved in corn oil and filter sterilized prior to injection. MEFs were derived from E13.5 embryos using Mcm4 C3/C3 Mcm2 Gt/+ males mated to Mcm4 C3/C3 females. MEFs were genotyped and treated with Plasmocin (InvivoGen) to prevent mycoplasma. For treatment of MEFs, a 50mM solution of testosterone propionate was prepared in ethanol and cells were treated with 10nM for 1 hour. The media was then removed and the cells collected at indicated timepoints. Sex reversal of XX Mcm4 C3/C3 Mcm2 Gt/+ embryos was carried out using an autosomally-linked Sry transgene (Tg(Sry129)4Ei) 32 .

Genotyping.
Genomic DNA was isolated from animal tissue using the hot-shot lysis procedure 33 . Genotyping PCR was carried out using Taq1 and gene-specific primer pairs (Table S9). For Chaos3 genotyping, the PCR products were digested with MboII to identify mutant alleles as Chaos3 but not wild-type alleles are digestible with this enzyme. For Mcm5, ES cells were verified using primers containing regions outside of the gene trap insertion to verify. To determine the sex of early embryos, primers for Sry (Sex-determining region Y) were used to identify males, females are Sry negative. Genotyping for Mcm2-7 genetraps has been previously described 3 .  (Table S9). Following germline transmission, the mutation was backcrossed into C3H for ≥ 4 generations before crossing to C3H-Mcm4 Chaos3 mice.

Generation of FancM mice.
Fancm em1/Jcs was generated using CRISPR/Cas9-mediated genome editing. In summary, an optimal guide sequence targeting the first exon of Fancm was designed using the mit.crispr.edu website. Oligos to generate the sgRNA DNA template were ordered from Integrated DNA Technologies (IDT) and the sgRNA was in vitro transcribed as described previously 34 (CRISPR-FancF: GAAATTAATACGACTCACTATAGGCCAGCTGGTAGTCGCGCACGGTTTTAGAGCTA GAAATAGC, CRISPR-FancR: CAAAATCTCGATCTTTATCGTTCAATTTTATTCCGATCAGGCAATAGTTGAACTTTT TCACCGTGGCTCAGCCACGAAAA). Embryo microinjection into C57BL/6J zygotes was performed as described previously 35 using 50ng/uL of sgRNA and 50ng/uL of Cas9 mRNA (TriLink Biotechnologies). The resulting 7bp deletion was identified via Sanger sequencing and subsequent genotyping was performed with primers sets specific to the mutant and wildtype alleles. (Table S9).

Flow cytometry to monitor X-inactivation.
A transgenic mouse 36 containing an X-linked EGFP was crossed to Mcm4 Chaos3 mice, and FACS analysis of embryos was carried out as described in that citation. Mcm4 C3/+ Mcm2 Gt/+ males bearing an ubiquitously-expressed X-linked GFP transgene were bred to Mcm4 C3/C3 females. E10.5 female embryos (littermates from 7 different pregnancies), all of which must bear the GFP transgene, were genotyped, dispersed into single cells, and analyzed by flow cytometry to determine the fraction of GFP+ cells. Theoretical maximum of GFP-positive cells in controls is 50%.

Quantitative real-time reverse transcription-PCR (qRT-PCR).
RNA was isolated from cells using a kit per manufacturer's instructions (Zymo or Qiagen RNeasy). 500ng of RNA was reverse transcribed into cDNA using qScript (Quanta) and analyzed on an ABI7300 or a Bio-Rad CFX96 using the following primers and iTaq (Bio-Rad). All reactions were normalized to Gapdh and/or Tbp. Primer sequences are available in Table S10

RNA-seq.
Total RNA was isolated from E13.5 placentas by homogenizing placentas in RNA lysis buffer followed by column purification per manufacturers' instructions (Omega Biotech). RNA sample quality was confirmed by spectrophotometry (Nanodrop) to determine concentration and chemical purity (A260/230 and A260/280 ratios) and with a Fragment Analyzer (Advanced Analytical) to determine RNA integrity. Ribosomal RNA was subtracted by hybridization from total RNA samples using the RiboZero Magnetic Gold H/M/R Kit (Illumina). Following cleanup by precipitation, rRNA-subtracted samples were quantified with a Qubit 2.0 (RNA HS kit; Thermo Fisher). TruSeq-barcoded RNAseq libraries were generated with the NEBNext Ultra II Directional RNA Library Prep Kit (New England Biolabs). Each library was be quantified with a Qubit 2.0 (dsDNA HS kit; Thermo Fisher) and the size distribution was be determined with a Fragment Analyzer (Advanced Analytical) prior to pooling. Libraries will be sequenced on a NextSeq500 instrument (Illumina). At least 20M single-end 75bp reads were generated per library. For analysis, reads were trimmed for low quality and adaptor sequences with cutadapt v1.8 using parameters: -m 50 -q 20 -a AGATCGGAAGAGCACACGTCTGAACTCCAG --match-readwildcards. Reads were mapped to the mouse reference genome/transcriptome using tophat v2.1 with parameters: --library-type=fr-firststrand --no-novel-juncs -G <ref_genes.gtf>. For gene expression analysis: cufflinks v2.2 (cuffnorm/cuffdiff) was used to generate FPKM values and statistical analysis of differential gene expression 37 . For the GSEA analysis, all expressed genes were analyzed using the Hallmarks dataset 38
Crescendo ECL substrate(Millipore) was used and immunoblots digitally scanned using a cDigit scanner. Quantification of immunoblots was performed using ImageStudio software.
Placentae from E13.5 embryos were dissected from individual embryos and washed in PBS. Decidua were separated from placenta and uterine tissue with fine forceps. Genotyping was carried out using a piece of the embryo. Placentae were flash-frozen in OCT and 10μM sections cut on a cryostat and affixed to slides. Sections were fixed for 10 minutes with 4% paraformaldehyde in PBS, and stained with mouse anti-γH2ax-phospho ser41 (Millipore) using a M.O.M kit and Biotin-Streptavidin blocking kit (Vector Labs) according to manufacturer's instructions. Alexa-488 or Alexa 647-streptavidin (Invitrogen) was used to visualize. Slides were scanned using a Scanscope FL with a 20X objective. Images were quantified using Fiji or HALO(Indica Labs) and foci were detected as described 39 with an added size parameter to differentiate between nuclei and cytoplasmic signals 40

Hydroxyurea and IR treatment of embryos.
For irradiation experiments, pregnant females were irradiated with 5 Rads (50mGy), 3 times a week during gestation, beginning at E1.5. For HU experiments, hydroxyurea (Sigma) was dissolved at 10mg/ml in sterile 1X PBS for injection. Pregnant C3H females were subjected to daily i.p. injections of 30-50ug/kg beginning at E3.5. Control females received daily i.p injections of sterile 1X PBS alone. All pregnancies were carried to term and the number and sex of animals determined at birth.

Data Availability.
All data underlying the findings of this study are presented in the paper, except for RNA-seq data, which has been deposited into the GEO database (accession number GSE119710). Note that a source data file is online for the γH2AX and MCM protein quantifications.

Extended Data
Extended Data Fig. 1

. X-inactivation is not perturbed in MCM mutant embryos.
Mouse female embryos bearing one X-linked GFP transgene were dispersed into single cells and examined by flow cytometry for GFP fluorescence. Graphed are percent viability at birth of males and females for each of the indicated genotypes versus C3/C3 littermates. For Fancm, the viability is versus WT littermates. The numbers on or over the bars = # males or females of the indicated genotype, and the N values below equal the total number of newborns with that genotype. Some of the data for all genotypes except that involving M5 were reported in 3 , but broken out by sex here and with added data that are enumerated in Extended Data Table 1 and Tables S1-S5     γH2AX staining in placentae from the indicated maternal genotypes. Each dot represents a single placenta. A minimum of 2 litters was examined per mating, total number of placentae analyzed is indicated. Significance was by unpaired, 2-tail t-tests. Centre value=mean, Error bars = standard deviation. ns = not significant.