A hypomorphic mutation in Pold1 disrupts the coordination of embryo size expansion and morphogenesis during gastrulation

Gastrulation is a critical developmental process required for germ layer formation and the establishment of the body plan (Arnold and Robertson, 2009). Gastrulation initiates with the emergence of the primitive streak in the proximal posterior epiblast. As the streak extends to the distal tip of the embryo, epiblast cells undergo an epithelial-mesenchymal transition (EMT) to form the mesoderm layer between the epiblast and the visceral endoderm (VE) (Kinder et al., 1999; Lawson, 1999; Lawson et al., 1991). Epiblast cells ingressing through the anterior region of the elongating primitive streak intercalate into the VE to form the definitive endoderm layer, which will give rise to the gut tube, and subsequently, the epithelium of endodermal organs, such as the pancreas and intestine (Kwon et al., 2008; Lawson et al., 1986; Lawson and Pedersen, 1987). Gastrulation requires tight spatiotemporal coordination of cell number expansion, cell migration, and cell fate determination. Despite extensive research on these topics, it has been challenging to untangle the complex interplay among these three key components. Previous studies used embryological methods in pre-implantation embryos to investigate size regulation during the pre- and early-gastrulation stages. These experiments found that double-sized embryos, formed by aggregating two eight-cell-stage morula, underwent size regulation before gastrulation. The double-sized embryos showed an increase in cell-cycle length compared to controls; in addition, they lacked the proliferative burst that normally occurs before gastrulation. These two modes of regulating cell proliferation allowed the aggregated embryos to reach a normal size and cell number before E7.0 and then to gastrulate (Buehr and McLaren, 1974; Lewis and Rossant, 1982). Conversely, undersized mouse embryos generated by removing one or two blastomeres from the four-cell-stage preimplantation embryo, sustained a prolonged proliferative burst, leading to an increase in cell number before the initiation of gastrulation (Power and Tam, 1993). Another study on size regulation in the mammalian embryo examined the response to reduced cell numbers in the early post-implantation embryo, an experimentally more refractory stage. Following treatment with mitomycin to inhibit cell proliferation, E7.0 embryos, with ∼80% of their cells eliminated, could still recover and complete gastrulation (Snow and Tam, 1979). These elegant studies suggest that intrinsic mechanisms operate within the pre- and early-post-implantation embryo to monitor and control cell numbers both before and at the onset of gastrulation, supporting regulative development as an important feature of early mammalian embryogenesis. However, it is technically challenging to apply standard embryological and pharmacological methods to investigate how altered cell number in the gastrulating embryo impacts tissue patterning and morphogenesis. Therefore, to ask whether mechanisms of size regulation continue to act during gastrulation, it is beneficial to explore alternative approaches to perturbing cell numbers, such as genetic manipulation.

Although many genes encoding cell-cycle-related proteins have been genetically inactivated to explore the effects of cell proliferation on embryo size and morphogenesis, the resulting phenotypes are generally not suitable for studies of gastrulation. For example, Cyclin A2 (Ccna2) homozygous null mutants can be recovered only up to E5.5 (Murphy et al., 1997), whereas embryos lacking all D-type cyclins survive past gastrulation, with no overt phenotypes (Kozar et al., 2004). Other targets for genetic perturbation of cell proliferation are the polymerases that replicate DNA. DNA Polymerase Delta (Pol δ), the subject of this report, plays multiple critical roles in DNA replication, with functions in DNA synthesis and repair (Jain et al., 2018). In mammalian cells, Pol δ contains four subunits: the catalytic subunit, p125 (Pold1), and three regulatory subunits, p50 (Pold2), p66 (Pold3) and p12 (Pold4). Pold1 consists of two functional domains: an N-terminal 3′-5′ exonuclease with DNA proofreading activity and a C-terminal DNA polymerase that catalyses DNA synthesis. Several mutant alleles have been generated for Pold1 but, like the targeted cell-cycle-related genes mentioned above, the homozygous mutants either fail to survive beyond the onset of gastrulation or show no phenotypic defects during gastrulation. Null mutations in Pold1 cause peri-implantation lethality (Uchimura et al., 2009). Two missense mutations have been reported for Pold1: a D400A substitution in the exonuclease domain and an L604K substitution in the DNA polymerase active site (Fig. S1A) (Uchimura et al., 2009; Venkatesan et al., 2007). While developmentally normal, Pold1^D400A/D400A mice frequently died with swollen thymuses 3 months after birth (Uchimura et al., 2009). Pold1^+/L604K heterozygous mice underwent normal development but had a reduced lifespan and developed multiple tumour types, including lymphoma, adenoma, and carcinomas of the liver and lung (Venkatesan et al., 2007). Although no specific phenotypes were reported for Pold1^L604K/L604K embryos, notably they died around E8.5, suggesting that missense mutations in the polymerase domain might perturb cell proliferation at a level compatible with the investigation of gastrulation phenotypes.

In this study, we report a Pold1 hypomorphic mutation identified in a phenotype-based genetic screen for recessive mutations causing gastrulation defects in mouse embryos. This mutation altered a conserved residue (D939Y) in the Pold1 DNA polymerase domain, caused reduced Pold1 protein expression, and resulted in compromised DNA synthesis. Mutant embryos could be retrieved up to E8.5; at this stage they were small with a siren-like morphology; hence we named the mutant tiny siren (tyrn). We investigated embryo growth and cell lineage differentiation in tyrn mutants at developmental stages between E6.5 and E8.5. The tyrn mutation impaired cell proliferation without affecting anterior–posterior (A–P) patterning, but severely disrupted tissue morphogenesis during gastrulation. Our findings suggest normal cell proliferation is essential for mesoderm lineage allocation and is required to coordinate embryo size with cell movement for proper morphogenesis.

To study the genetic regulation of gastrulation, we performed mouse ENU mutagenesis screens to uncover genetic disruptions in embryos with abnormal morphology at E8.5 (García-García et al., 2005; Hernández-Martínez et al., 2019; Huangfu et al., 2003; Migeotte et al., 2011). We isolated a mutant (later named tyrn) that exhibited not only a smaller overall size, but also a striking body shape and orientation (Fig. 1A) that was distinct from other mutant phenotypes we had observed at this stage (Bazzi et al., 2017; García-García et al., 2005; Hernández-Martínez et al., 2019; Huangfu et al., 2003; Migeotte et al., 2011; Zhou and Anderson, 2010). Instead of forming a U-shaped embryo with a well-extended A–P axis, the tyrn mutant embryos had a short A–P axis and lay relatively flat along the posterior side of the yolk sac, with the head misoriented towards the distal tip (Fig. 1A). To determine if the irregularly shaped mutants had established the A–P axis, we performed whole-mount in situ hybridisations (WISH) at E8.5 to a diagnostic set of markers. We found that although tyrn mutant embryos did not form a well-structured head, they did express Foxg1 and Otx2, markers of forebrain and forebrain/midbrain, respectively, in discrete overlapping regions (Fig. 1B,C). In addition, WISH detected relatively normal expression of T (Brachyury), which marks the posterior tail bud and notochord (Fig. 1D), and of Foxa2, which labels the floor plate of the neural tube, posterior notochord, notochordal plate and gut endoderm (Fig. 1E). Therefore, despite causing a highly unusual body shape, the tyrn mutation does not affect A–P patterning.

To identify the causative mutation for the tyrn phenotype, we collected both wildtype and tyrn mutant embryos at E8.5 and performed whole-exome sequencing (Jain et al., 2017). We found a G to T transversion at nucleotide position 2815 of the Pold1 open reading frame that generated an aspartate to tyrosine substitution (D939Y) within the DNA polymerase domain (Fig. 2A). The mutated aspartate residue is highly conserved across eukaryotic organisms from budding yeast to humans (Fig. 2B). Western-blot analysis showed a reduced level of Pold1 in tyrn mutant embryos (Fig. 2C). To confirm that Pold1 is the causative gene underlying the tyrn mutant phenotype, we performed a complementation test using the Pold1^tm1b null allele, derived from embryonic stem cells carrying a ‘knockout-first’ tm1a allele (Skarnes et al., 2011) (Fig. 2D, Fig. S1B). No Pold1^tm1b/tm1b embryos were recovered at post-implantation stages, consistent with the phenotype previously reported for Pold1-null mutants (Uchimura et al., 2009). The Pold1^tyrn/tm1b embryos produced from a Pold1^tm1b/+ and tyrn/+ cross failed to survive past ∼E7.5 (Fig. 2D), demonstrating that the tm1b and tyrn mutations failed to complement. Moreover, the phenotype displayed by Pold1^tyrn/tm1b embryos was more severe than that of the Pold1^tyrn/tyrn embryos (Fig. 2E), but milder than that of Pold1^tm1b/tm1b-null mutants. These findings indicate that perturbation of Pold1 function is responsible for the phenotypes observed in tyrn mutants and that Pold1^tyrn is a hypomorphic allele.

Based on the published cryo-electron microscopy structure of the human Pol δ (Lancey et al., 2020), the aspartate residue mutated in tyrn embryos resides in the thumb domain of the C-terminal DNA polymerase domain of POLD1 (Fig. 2A, Fig. S2A-C). Because of the important role of the thumb domain in stabilizing Pol δ at the primer–template junction during DNA synthesis (Jain et al., 2018), we hypothesised that the tyrn mutation may impair and reduce the DNA synthesis activity of Pol δ. We used a primer extension assay and mouse EdU labelling to test the Pol δ polymerase activity in vitro and in vivo, respectively. For the primer extension assay, we modelled the mouse Pold1 D939Y missense mutation in budding yeast Pol δ by introducing the equivalent mutation, D941Y, in the yeast catalytic subunit, Pol3 (Devbhandari and Remus, 2020) (Fig. 3A). We purified wildtype Pol δ or Pol δ^D941Y (with Pol3^D941Y) after overexpression in budding yeast (Fig. 3B). Analysis of the DNA synthesis products by denaturing gel analysis reveals that both overall DNA synthesis and the level of full-length DNA products were significantly reduced in the presence of Pol3^D941Y when compared to the wildtype Pol δ (Fig. 3C-E). These data demonstrate that the D941Y substitution in Pol3, and by extension the corresponding D939Y substitution in Pold1, impairs the DNA polymerase activity of Pol δ, possibly by decreasing its processivity.

To test if DNA synthesis is affected in tyrn embryos, we performed in vivo EdU labelling at E6.0, E6.5, and E7.0, the window of time during which the size difference between wildtype and tyrn embryos emerges. We observed a significant reduction in EdU incorporation in tyrn mutants compared to wildtype embryos at all stages (Fig. 4A-B). Both wildtype and tyrn embryos showed a substantial increase in cell number from E6.0 to E7.0, but tyrn embryos exhibited a slower growth rate starting from E6.5; by E7.0 tyrn embryos had significantly lower numbers of cells compared to wildtype embryos (Fig. 4C). The reduction in cell number is not due to increased cell death; based on cleaved Caspase-3 staining, levels of cell apoptosis were similar between wildtype and tyrn embryos during this period (Fig. 4D). However, tyrn embryos showed increased cell apoptosis at around E7.5, with apoptotic cells concentrated at the distal tip, the prospective location of the abnormal small head at E8.5 (Fig. S3). Taken together, we conclude that the D939Y missense mutation impairs Pold1 polymerase activity, which, together with the reduced Pold1 protein expression in tyrn embryos, impedes cell proliferation and leads to a reduction in embryo size during gastrulation.

The mispositioned head, and the abnormal morphology of E8.5 tyrn mutants, indicated that the A–P axis, although established, was shifted in orientation along the proximal–distal axis. Proper formation of the A–P axis depends on the anterior migration of a morphologically distinct population of VE cells from their position at the distal tip of the E5.5 embryo (Rivera-Pérez et al., 2003); by E5.75-E6.0 this cell population will reach the anterior-epiblast–extraembryonic-ectoderm boundary and form the anterior visceral endoderm (AVE) (Srinivas et al., 2004). Multiple studies have shown that defects in AVE migration can lead to abnormal positioning of the A–P axis and mislocalised head phenotypes (Acampora et al., 2009; Clements et al., 2011; Kojima et al., 2014; Lu et al., 2001; Martinez-Barbera et al., 2000; Rossant and Tam, 2009). Therefore, we asked whether the AVE resided in its normal anterior position in tyrn mutants at E6.5. At this stage, the tyrn mutants were morphologically indistinguishable from wildtype littermates. WISH showed that tyrn mutants expressed two archetypal AVE markers, Dkk1 (a Wnt antagonist) and Cerl (a Nodal antagonist), in a pattern comparable to that of wildtype embryos (Fig. 5A,B). We validated these findings by crossing in the Hhex-GFP transgene reporter to fluorescently label AVE cells in tyrn mutants (Rodriguez et al., 2001). Consistent with the Dkk1 and Cer1 in situ hybridisation results, we detected the Hhex-GFP expressing AVE cells at the anterior boundary between the embryonic and extraembryonic regions in tyrn embryos at E6.5 (Fig. 5C). These results indicate that AVE migration and A–P axis establishment proceed normally in pre-gastrulation stage tyrn mutants.

At E6.5, the AVE resided at the expected location in tyrn mutants; yet the position of the tyrn embryo's anterior region was shifted toward the distal tip at E7.5. We examined the expression of multiple anterior-specific markers to visualise the organisation of the anterior region in both wildtype and tyrn mutant embryos at E7.5. As depicted in Fig. 6A-B, wildtype embryos expressed Sox2 uniformly throughout the anterior epiblast, whereas the tyrn mutants expressed Sox2 in a discontinuous pattern, with more intense staining in the distal epiblast than in the proximal/anterior region. In the wildtype E7.5 embryo in the left panels of Fig. 6C,D, the Hhex-GFP transgene labelled anterior definitive endoderm (ADE) as well as remaining AVE cells. In contrast, Hhex-GFP expression was evident predominantly in the distal region of the E7.5 tyrn mutant, with no obvious A–P asymmetry (Fig. 6C,D, right panels). Similarly, cells expressing Otx2, a head organiser marker, lay distally in tyrn embryos compared to their anterior position in wildtype embryos (Fig. 6E).

The AVE serves as a transient anterior signalling centre at ∼E6.5. As gastrulation progresses, AVE cells disperse into the extraembryonic–embryonic boundary and the anterior portion of the extraembryonic yolk sac (Lawson and Pedersen, 1987; Rivera-Pérez et al., 2003; Shimono and Behringer, 2003; Tam and Loebel, 2007). Meanwhile, the ADE and axial mesoderm (AME), both of which strongly express Nodal and Wnt antagonists (Arnold and Robertson, 2009), gradually migrate anteriorly and become new sources of anterior signalling (Arkell and Tam, 2012). We hypothesised that the abnormal orientation of the A–P axis might reflect the distal positioning of AME and ADE. This hypothesis is consistent with the aforementioned aberrant expression pattern of Hhex-GFP, which labels both ADE and AVE cells. Furthermore, Fig. 6A,D and F examined the expression of Foxa2, a marker of AME and ADE, in wildtype and tyrn embryos at E7.5. Emerging Foxa2-expressing ADE and AME cells resided in the distal anterior region in wildtype embryos, but these cells were observed more posteriorly in tyrn mutants.

The aberrant posterior positioning of ADE and AME, both derivatives of the anterior primitive streak (APS) (Arnold and Robertson, 2009; Lawson, 1999; Lawson et al., 1991), suggested that primitive streak extension was defective in tyrn mutants. To compare the organisation of the primitive streak between wildtype and tyrn embryos, we performed section immunofluorescence for T (Brachyury), a marker of nascent mesoderm (including AME) emanating from the primitive streak (Fig. 6B,C). In E7.5 wildtype embryos, T staining showed that the primitive streak had extended to the distal tip and generated AME derivatives of the anterior primitive streak. In contrast, T staining of E7.5 tyrn embryos indicated that the primitive streak had extended only to the midpoint of the posterior side (Fig. 6B,C). Also pointing to impaired primitive streak elongation in tyrn embryos, WISH detected irregularities in the production and positioning of paraxial and extraembryonic mesoderm. Whereas expression of Tbx6, a marker of paraxial mesoderm, was found along the posterior side of the wildtype embryo, down to the distal tip, it was restricted to the posterior–proximal region in tyrn, consistent with a short primitive streak (Fig. 6G). Twist expression labels two distinct mesodermal populations at E7.5: anterior mesoderm precursors to cranial mesenchyme and extraembryonic mesoderm of the allantois (Bildsoe et al., 2009). Of note, the E7.5 tyrn mutant generated Twist-expressing anterior mesoderm but lacked Twist-expressing cells of the allantois (Fig. 6H). The tyrn embryos also expressed greatly reduced levels of Tbx4, another marker of allantoic mesoderm (Fig. 6I). These findings indicate that tyrn mutants have very few extraembryonic mesoderm cells inside the allantois. Taken together, these results suggest that impaired extension of the primitive streak in tyrn embryos causes defects in multiple streak derivatives, including a severe deficiency of allantoic mesoderm and aberrantly positioned AME and ADE in the distal region of the E7.5 tyrn embryo. The development of the head distally, rather than anteriorly, in tyrn mutants, is potentially a consequence of impaired primitive streak extension as well. In wildtype embryos, the AME of the notochordal plate lies posterior to the prechordal plate and head process, midline mesodermal populations that provide signals governing the growth and patterning of neuroectoderm (Balmer et al., 2016; Costello et al., 2015). If their placement is coordinated with the position of the AME and ADE, then the prechordal plate and head process likely reside in an abnormal distal position, where they direct the formation of the head in tyrn mutants.

The regulation of cell number plays an integral role in mouse embryo size determination during the pre-gastrulation stages. Earlier studies used embryo manipulation or pharmacological methods to alter cell numbers in pre- and early post-implantation mouse embryos. The findings showed, that prior to gastrulation, mouse embryos have the ability to sense and respond to an acute, global alteration in cell number and achieve normal size. Thus, robust size regulatory mechanisms must operate intrinsically in the mouse embryo to ensure proper size before the onset of gastrulation (Lewis and Rossant, 1982; Power and Tam, 1993). These classic studies raise many questions about when, in what cell populations, and how size-regulatory mechanisms function. Such questions are not amenable to approaches involving the global loss or gain of cells, especially for investigations during and after gastrulation. More suitable would be fine-tuned genetic methods that allow manipulation of candidate components of the size regulatory process over a defined time period. Here, we assessed the impact on embryo growth and morphogenesis of tyrn, an ENU-induced hypomorphic mutation that reduced levels of DNA synthesis at gastrulation stages. The tyrn mutation disrupted the polymerase function of Pold1, causing reduced cell proliferation in vivo, starting at ∼E6.0. tyrn embryos did not merely exhibit a wildtype-like morphology with a proportionally reduced size, or a random shape with no underlying logic. Instead, they showed a siren-like morphology, with the head located near the distal tip rather than in the anterior region, as seen in wildtype embryos. Phenotypic analyses performed at stages between E6.5 and E7.5 revealed that defective extension of the primitive streak likely underlies the mispositioned head. The AME and ADE produced by the short primitive streak resided in a distal position, causing the A–P axis to misalign with the proximal–distal axis (Fig. 7). We propose that in the tyrn mutant, decreased rates of DNA synthesis resulted in reduced levels of cell proliferation during gastrulation, which, in turn, disrupted the coordination of embryo growth with lineage specification and tissue morphogenesis. Therefore, the abnormal orientation of the tyrn mutant inside the decidua highlights the pivotal importance of the mechanisms orchestrating embryo growth with cell differentiation and movement during gastrulation.

Our study also suggests that reduced levels of cell proliferation during gastrulation perturb allocation of mesoderm lineages. Primitive streak formation initiates at the posterior–proximal pole where high levels of BMP, FGF, Wnt and Nodal signals are present (Brennan et al., 2001; Ciruna and Rossant, 2001; Conlon et al., 1994; Huelsken et al., 2000; Mishina et al., 1995). As the primitive streak elongates and extends to the distal tip, epiblast cells move into the streak and undergo EMT to form the mesoderm layer. The time and site of ingression through the streak determines the cell fate of mesoderm derivatives (Vincent et al., 2003; Winnier et al., 1995). Epiblast cell proliferation drives streak extension by providing a continuous supply of cells to populate the streak. In addition, normal rates of cell proliferation guarantee coordination of embryo size expansion with the formation of signal gradients; thus, exposing cells along the streak to different signal combinations and enabling them to contribute to different mesoderm subtypes. Since the primitive streak failed to fully extend in tyrn mutants, reduced cell proliferation likely disrupted such coordination. We found that tyrn mutants generated axial and paraxial mesoderm subtypes but contained them at a more posterior position. While the greatly reduced allantoic mesoderm may result from an altered posterior signalling environment, it could also reflect a general developmental delay, as allantoic mesoderm normally arises late during gastrulation, after the streak has reached the distal tip. Another possibility is that the extraembryonic mesoderm of the allantois originates from a highly proliferative cell population that is more sensitive to the inhibition of DNA synthesis. These questions may be addressed in the future through the perturbation of embryonic cell numbers in specific cell populations during gastrulation.

Cell number regulation occurs throughout different stages of embryonic development. The effects of cell number perturbation on tissue and organ size and morphogenesis can vary greatly based on developmental stage and tissue or organ type. The mandibular hypoplasia, deafness, progeroid and lipodystrophy (MDPL) syndrome, a multisystem disorder, has been associated with a heterozygous in-frame deletion of Ser605 that causes loss of POLD1 polymerase activity (Elouej et al., 2017; Fiorillo et al., 2018; Sasaki et al., 2018; Weedon et al., 2013). The reported patients presented growth retardation, sensorineural deafness, loss of subcutaneous adipose tissue and insulin resistance, indicating that decreased POLD1 activity exerts organ-specific effects. Other studies have also identified organ-specific responses to the loss of tissue-specific progenitors. Removing the embryonic limb field in amphibians or chickens does not affect the final limb size since the limb is capable of robust compensatory proliferation (Holder, 1981; Summerbell, 1981), and similar compensatory growth has been observed for the developing liver progenitors (Bort et al., 2006). In contrast, the final pancreas size is determined by the initial pancreatic progenitor pool and there is a lack of significant compensatory growth (Stanger et al., 2007). These studies demonstrate the importance of studying cell number regulation in diverse tissue and organ types at different stages of embryonic development. Our findings underscore the power of unbiased phenotype-based forward genetic screening approaches and highlight the value of using ENU-induced hypomorphic alleles to uncover new biological functions of well-characterised genes such as Pold1. The developmental biology field may benefit from revisiting unsolved questions from the ‘forgotten classics’ on embryo size regulation (https://thenode.biologists.com/forgotten-classics-regulating-size-mouse-embryo/research/) (Buehr and McLaren, 1974; Lewis and Rossant, 1982; Snow and Tam, 1979) by integrating newer methods and tools. For instance, one may use carefully chosen genetic models coupled with modern developmental biology techniques such as live imaging and single-cell genomics to investigate this important topic in diverse developmental contexts beyond the early stages of embryogenesis.

The parental JM8.N4 mouse ES cell strain (strain origin: C57BL/6N, male, black coat, non-agouti, MGI ID: 4431772) (Skarnes et al., 2011) carrying one Pold1^{tm1a(EUCOMM)Wtsi} (Pold1^tm1a) allele was imported from the European Mouse Mutant Cell Repository (EUMMCR). The selected Pold1-G09 ES cell clone passed the karyotype analysis performed by Molecular Cytogenetics Core Facility, Memorial Sloan Kettering Cancer Center (MSKCC). The Pold1-G09 ES cell clone was injected into the female B6(Cg)-Tyr^c-2J/J (albino C57BL/6J, or B6-albino, non-agouti, The Jackson Laboratory) donor mice by MSKCC Mouse Genetics Core Facility. Fourteen out of the 19 pups alive were chimeric. Male chimeras were crossed with albino FVB/NJ (FVB, containing homozygous dominant agouti locus A/A) females to generate heterozygous offspring carrying the Pold1^tm1a allele. Germline transmission was determined by the presence of agouti pups and genotyping for the presence of the LacZ cassette present in the tm1a allele. Mice carrying a Pold1^tm1b-null allele were generated by crossing Pold1^+/tm1a mice with CAG-Cre transgenic mice (The Jackson Laboratory) to remove the critical exons between exon 3 to exon 10. The lacZ cassette remained in the Pold1^tm1b allele and was used for genotyping. The Hhex-GFP strain was a gift from Anna-Katerina Hadjantonakis (Developmental Biology Program, Sloan Kettering Institute) (Rodriguez et al., 2001). Mice that were 8-16 weeks old were used to generate E6.5 to E8.5 embryos. Analysis of the mutant phenotype was performed in the FVB or the mixed FVB-B6 genetic background. Mice were housed and bred under standard conditions in accordance with Institutional Animal Care and Use Committee (IACUC) guidelines. All experimental procedures were approved by the MSKCC IACUC.

The tyrn allele was generated by mouse ENU mutagenesis screens using C57BL/6J males and was identified based on its embryonic phenotype at E8.5, as previously described (García-García et al., 2005). To obtain genomic DNA, E8.5 tyrn mutant embryos were pooled into three sample groups and snap frozen on dry ice. Gentra Puregene kit (QIAGEN) was used for the genomic DNA extraction of tyrn mutant embryos. Whole-exome sequencing was performed at the MSKCC Integrated Genomics Operation. Exome capture was performed using the SureSelectXT kit (Agilent Technologies) and SureSelect Mouse All Exon baits (Agilent Technologies). An average of 100 million 75-bp paired reads were generated. Sequencing data analysis was performed using the methods described in (Jain et al., 2017). We examined homozygous exonic sequence variants shared among all pools of phenotypic embryos, but which were not identified in cohorts of wildtype embryo samples or separate lineages of ENU screen-derived embryos, and which were not present in a publicly available mouse genomic polymorphism database (dbSNP). Eleven potential phenotype-causing lesions were found in eight gene annotations, six nonsynonymous, of which Pold1 was one: Pold1: NM_011131: exon23: c. G2815T: p.D939Y. After examining the known phenotypes of published alleles or functional annotations of these candidate genes, we re-examined non-exonic and heterozygous mutations and eliminated the read depth threshold to explore other possibilities. After going through the whole dataset (developed by Devanshi Jain from the Scott Keeney laboratory, Sloan Kettering Institute) (Jain et al., 2017) containing all our exome sequencing submissions over the years to catalogue universal variants we could eliminate, we confirmed that the sole candidate variant was in Pold1. The tyrn allele has a single G to T transversion in the 2815 nucleotide position of exon 23 in the Pold1 coding sequence (G2815T), causing a nonsynonymous missense mutation in the amino acid position 939 resulting in the substitution of aspartic acid to tyrosine, D939Y. The G2815T mutation created a RsaI restriction site used for genotyping. Embryos displaying the tyrn morphological defect were homozygous for the G2815T mutation.

Pold1^tm1b/+ females were crossed in timed matings with tyrn/+ males to produce embryos. Embryos at E7.5 and E8.5 stages were harvested and grouped into wildtype and mutant cohorts based on phenotype. No tyrn/tm1b mutant embryos were recovered at E8.5 stage. The genotypes of E7.5 embryos corresponded with their morphological phenotypes, verifying that a lesion in Pold1 was the causative mutation of the tyrn phenotype. Embryos with abnormal morphological phenotype genotyped Pold1^tyrn/tm1b, whereas embryos with Pold1^+/+, Pold1^tyrn/+ and Pold1^tm1b/+ genotypes displayed wildtype phenotypes.

Pregnant FVB female mice bearing embryos at E6.5-10.5 stages were euthanised by cervical dislocation. Uteri were harvested based on an IACUC-approved mouse protocol. Embryos were dissected from decidua inside the uterus in cold 0.4% Bovine Serum Albumin (BSA) (Sigma-Aldrich) in phosphate-buffered saline (PBS) using fine forceps and Leica dissection scope. Embryos were fixed in 4% paraformaldehyde (PFA) on ice for at least 4 h and washed with PBS for 5 min three times at room temperature (RT). Fixed embryos were stored in PBS at 4°C for downstream experiments.

Fixed embryos were sequentially dehydrated in 25, 50%,75% methanol-DEPC PBS and 100% methanol and stored at −20°C. Embryos were sequentially rehydrated in 75%, 50%, and 25% methanol-DEPC PBS before experiments. In situ hybridisation was performed following the standard protocol (Eggenschwiler and Anderson, 2000). Briefly, after rehydration, embryos were incubated in 1 μg/ml Proteinase K/PBS solution for 3 to 7 min and hybridised in hybridisation solution with RNA probes at 70°C overnight. A series of washes in 2X SSC and MAB solutions were performed the next day. The embryos were incubated in 1:10000 anti-Digoxigenin antibody (Roche) in blocking buffer [1% blocking reagent (Roche), 10% heat-inactivated goat serum, 0.1% Tween-20 in PBS] overnight and washed extensively in PBS+0.1% Tween-20+0.1% BSA. Embryos were incubated in BM purple solution (Roche) at RT protected from light until the purple colour was developed.

EdU (5-ethynyl-2′-deoxyuridine) powder (Invitrogen) was dissolved in sterile PBS into a working concentration of 2.5 mg/ml. Mice were weighed and injected with EdU solution (25 mg/kg) intraperitoneally. Embryos were harvested 2 h after injection and fixed in 4% PFA overnight. Fixed embryos were washed with PBS+3% BSA twice and then permeabilised in PBS+0.5%Triton X-100 at RT for 20 min. Embryos were washed in PBS+3% BSA after permeabilisation. Embryos were incubated with the reaction cocktail made from Click-iT™ EdU Cell Proliferation Kit for Imaging, Alexa Fluor 637 dye kit (Invitrogen) at RT for 30 min, protected from light. The cocktail was removed after incubation and embryos were washed in PBS+3% BSA twice.

Fixed embryos were stored in PBS at 4°C before use. Embryos for cryosection were dehydrated in 30% Sucrose-PBS at 4°C overnight. Embryos were embedded in Tissue-Tek O.C.T Compound (Sakura Finetek) and frozen on smashed dry ice immediately after embedding. Embedded embryos were stored at −80°C before use. Embryos were sectioned in 10 μm using Leica CM1520 Cryostat. Section slides were stored at −80°C before use. For whole-mount staining embryos, embryos were permeabilised with PBS+0.5% Triton X-100 for 1 h at RT. For section staining, slides were dried for 30 min and incubated in blocking buffer [0.1% TritonX-100, 1% heat-inactivated donkey serum (Gemini Bio) in PBS] for 1 h at RT. Primary antibodies were diluted with the optimised dilution ratio in blocking buffer: Foxa2 (Abcam, catalogue number ab108422, RRID:AB_11157157,1:300), T (Cell Signaling Technology, catalogue number 81694, RRID:AB_2799983,1:400), Cleaved Caspase-3 (Cell Signaling Technology, catalogue number 8202, RRID:AB_1658166,1:300), Sox2 (R and D Systems, catalogue number AF2018, RRID:AB_355110,1:300). Embryos and slides were incubated with primary antibodies at 4°C overnight. Embryos and slides were washed with PBS for 10 min three times the next day. After that, section slides or embryos were incubated in the blocking buffer containing specific secondary antibodies (Invitrogen,1:500) and DAPI (1:1000) for 2 h at RT. For whole-mount staining embryos, embryos were washed in PBS for 5 min three times, incubated in FocusClear (CelExplorer.Co) for 20 min at RT, and protected from light. Embryos were mounted with MountClear (CelExplorer.Co) and stored at 4°C. For cryosection staining, slides were washed in PBS for 5 min three times. Slides were mounted with ProLong Gold Antifade Mountant (ThermoFisher Scientific) and stored at 4°C. Both embryos and sections were imaged using Leica SP8 inverted laser scanning confocal microscope. Confocal images were reconstructed using Fiji (ImageJ) open-source image processing software.

E8.5 wildtype and tyrn embryos were harvested and pooled separately and stored at −80°C before experiments. Tissues were homogenised in cold lysis buffer [0.1% NP-40, 50 mM Tris-HCl (pH 7.2), 250 mM NaCl, 2 mM EDTA, phosphatase inhibitor mixture I and II (Calbiochem) and one tablet of Minicomplete (Roche) per 10 ml] on ice. Lysate was left on the shaker at 4°C for another 30 min. Lysate was centrifuged at maximum speed (12,000 rpm) to collect supernatant. Protein concentration was determined through BSA-based protein assays using Quick Start Bradford 1X Dye Reagent (BIORAD) and adjusted to a final concentration of 2 μg/μl. Samples were mixed 1:1 with 2X SDS loading buffer for denaturing at 95°C for 5 min. Then, samples were loaded in equal amounts onto 8% SDS-PAGE gels and run for 2 h under 150 V in 1X SDS buffer at RT. Proteins were transferred to PVDF membranes under 15 V overnight at 4°C. Membranes were incubated in blocking buffer (TBST+5% BSA) for 1 h at RT and incubated with primary antibodies in blocking buffer at 4°C overnight: POLD1 (Abcam, catalogue number ab168827,1:500); GAPDH (Santa Cruz Biotechnology, catalogue number sc-32233, RRID: AB_627679, 1:1000). Membranes were washed with TBST and incubated with specific secondary antibodies for 1 h at RT. Finally, membranes were washed with TBST and incubated with Pierce ECL Western Blotting Substrate (ThermoFisher Scientific) for 5 min and were film-exposed to show target bands.

Primer extension reactions were performed at 30°C in polymerisation buffer (25 mM Tris-HCl pH 7.5, 8 mM Magnesium Acetate, 5 mM Potassium Glutamate, 5% Glycerol). Purified proteins used in the primer extension assays were purified as previously described (Devbhandari and Remus, 2020). DNA template for the assay was generated by annealing a primer (5′-CCCAGTCACGACGTTGTAAAACG-3′) to M13mp18 single-stranded DNA (New England Biolabs, N4040S). The assay was initiated by incubation of 1 nM of DNA template with 1mM ATP, 1mM DTT, 80 μM dATP, 80 μM dGTP, 80 μM dCTP and 400 nM of RPA for 5 min. PCNA and RFC were then added to 70 nM and 4 nM, respectively, and incubation was continued for 5 min. Then, 33 nM of α−³²P-dATP (3000 Ci per mmol) and 4 nM of either Pol δ^WT or Pol δ^D941Y was added to the reaction, resulting in a primer extension by nine base pairs (due to lack of dTTP). After 5 min, 80 μM dTTP were added to the mix for synchronous primer extension. Equal volume aliquots of this reaction (18 μl) were removed from the master reaction (100 μl) at indicated times and stopped by adding EDTA and SDS to final concentrations of 40 mM and 0.25%, respectively. Products were fractionated on a 0.8% alkaline agarose gel (30 mM NaOH and 2 mM EDTA), dried and imaged using Typhoon FLA 7000. Quantification of the gel images was performed using ImageJ.

E6.0, E6.5, and E7.0 whole-mount EdU-labeled embryos were imaged using Leica SP8 inverted laser scanning confocal microscope and confocal z stacks of embryos were generated. tyrn mutant and wildtype embryos of the same stage were imaged under the same conditions. For each stage, three embryos per genotype were imaged for quantification. Optical sections of each embryo were imported into Imaris (version 9.5, Oxford Instruments), and three-dimensional reconstitution along Z-axis was performed for data analysis. We used the spots function in Imaris to automatically segment all DAPI-positive cells (in both embryonic and extraembryonic tissues) to obtain a total cell count for each embryo. EdU signal intensity of each segmented cell was automatically measured, and the background signal was subtracted. Cells with EdU signal above 10 arbitrary units were considered EdU positive and counted as EdU positive cells by the software. We used Prism 9 (GraphPad) to perform normalisation of EdU signal intensity. At each stage, the EdU signal of each EdU positive cell was normalised to the maximum EdU signal intensity. Data points were presented as box plots to show the distribution of EdU signal intensity among EdU positive cell population. Within the box plot, the median is represented by the horizontal dividing line and the top and bottom of the box represent the 75th and 25th percentiles, with the whiskers indicating the maximum and minimum points. A two-tailed Student's t-test was performed to evaluate the significance of all measurements.

For all embryo imaging experiments, n=3. For primer extension assays, three biological replicates were performed. We used Prism 9 to perform two-tailed Student's t-tests to evaluate the significance of all measurements and generate box plots and non-linear fitted curves. For the structure of human DNA polymerase δ, the original structure was imported and processed in PyMOL (PyMOL Molecular Graphics System, Version 1.2r3pre. Schrödinger, LLC).

The structure of human DNA polymerase δ is from the Protein Data Bank (PDB). PDB DOI: 10.2210/pdb6TNY/pdb. EM Map EMD-10539: EMDB EMDataResource.

The authors thank Dr Devanshi Jain for sharing the methods of data analysis. The authors thank MSKCC Integrated Genome Operation for performing whole-exome sequencing, MSKCC Molecular Cytogenetics Core Facility for karyotype analysis, MSKCC Mouse Genetics Core Facility for ESC clone injection, MSKCC Molecular Cytology Core Facility (Dr Boyko, Dr Tipping, Mr Feng) for assistance with confocal imaging and image processing. The authors thank Dr Anna-Katerina Hadjantonakis for providing Hhex-GFP mouse strain, and Jonathan Pai from Hadjantonakis’ lab for sharing embryo clearing methods. The authors thank Dr Jennifer Zallen and Dr Maria Jasin for helpful suggestions on this project.