Embryonic/fetal mortality and intrauterine growth restriction is not exclusive to the CBA/J sub-strain in the CBA × DBA model

Inbred strains of mice are powerful models for understanding human pregnancy complications. For example, the exclusive mating of CBA/J females to DBA/2J males increases fetal resorption to 20–35% with an associated decline in placentation and maintenance of maternal Th1 immunity. More recently other complications of pregnancy, IUGR and preeclampsia, have been reported in this model. The aim of this study was to qualify whether the CBA/CaH substrain female can substitute for CBA/J to evoke a phenotype of embryonic/fetal mortality and IUGR. (CBA/CaH × DBA/2J) F1 had significantly higher embryonic/fetal mortality mortality (p = 0.0063), smaller fetuses (p < 0.0001), and greater prevalence of IUGR (<10th percentile; 47% vs 10%) than (CBA/CaH × Balb/c) F1. Placentae from IUGR fetuses from all mating groups were significantly smaller (p < 0.0001) with evidence of thrombosis and fibrosis when compared to normal-weight fetuses ( > 10th percentile). In addition, placentae of “normal-weight” (CBA/CaH × DBA/2J) F1 were significantly smaller (p < 0.0006) with a greater proportion of labyrinth (p = 0.0128) and an 11-fold increase in F4/80 + macrophage infiltration (p < 0.0001) when compared to placentae of (CBA/CaH × Balb/c) F1. In conclusion, the embryonic/fetal mortality and IUGR phenotype is not exclusive to CBA/J female mouse, and CBA/CaH females can be substituted to provide a model for the assessment of novel therapeutics.


Results
To establish whether CBA/CaH × DBA/2J pregnancies show the same propensity for embryonic/fetal mortality and IUGR as CBA/J females, 150 CBA/CaH females were placed with DBA/2J, Balb/c or CBA/CaH males overnight. Both mid-gestation (14.5 dpc, n = 20 matings per group) and term (18.5 dpc, n = 30 matings per group) time points were assessed to directly compare the fetal outcomes. Harvesting of gravid uteri, including the cervix, revealed the presence of resorption sites but showed no macroscopic uterine abnormalities in CBA/ CaH × DBA/2J pregnancies at mid-gestation (Fig. 1a) or term (Fig. 1b).
A range of factors have been found to influence embryonic/fetal mortality in mice and specifically the CBA/J × DBA/2J murine model. These include maternal nutrition, primigravidity, immunological environment, maternal age and overcrowding of one uterine horn 19,20 . However, irrespective of time-point we found no differences in maternal weight gain, uteri wet weight, or total number of concepti when compared to CBA/ CaH × Balb/c (Fig. 2). The number of concepti in left vs right uterine horns of CBA/CaH × DBA/2J pregnancies was uneven (≥ 2 concepti difference between left and right horns; Χ 2 = 7.370, p = 0.0251) when compared to CBA/CaH × Balb/c, but the 'overcrowded' horn was not associated with greater mortality when compared to the 'uncrowded' horn (Χ 2 = 2.249, p = 0.3248; and shown Fig. 1).
Proportion of IUGR fetuses is increased in CBA/CaH × DBA/2J pregnancies. Studies using the CBA/J × DBA/2J murine model have reported that surviving fetuses show consistent and significant 'IUGR' 6 , based on a statistical reduction in average fetal weight. To utilise a biologically relevant criterion for IUGR, we quantified the prevalence of IUGR in our murine pregnancies using the human clinical definition of < 10 th percentile for gestational age, with fetuses from allogeneic (CBA/CaH × Balb/c) F1 as the population standard. Fetal wet weight and crown-rump length (CRL) 21 were assessed and demonstrated a strong correlation at both time points (14.5 dpc r = 0.77 to 0.89; p = 0.0406 to p < 0.0001; and 18.5 dpc r = 0.91 to 0.94, p < 0.0001 Fig. 3a). Using our definition, IUGR fetuses were denoted as those with a wet weight < 0.1427g and < 0.4725g and/or CRL < 9.75 mm and < 14.75 mm at 14.5 and 18.5 dpc, respectively. At mid-gestation, 47% of (CBA/ CaH × DBA/2J) F1 fell below the 10 th percentile confirming a substantial IUGR phenotype at this time point (Fig. 3b). From 20 pregnant dams this equated to 46 IUGR fetuses, compared to 31 for (CBA/CaH × CBA/CaH) F1 and 13 for (CBA/CaH × Balb/c) F1. The number of IUGR fetuses dropped markedly at term with only 5-6 IUGR fetuses in all pregnancy groups. In contrast, the number of resorptions at term remained high; 46 in (CBA/ CaH × DBA/2J) F1, 24 in (CBA/CaH × CBA/CaH) F1 and 14 in (CBA/CaH × Balb/c) F1. We propose that IUGR fetuses present at 14.5 dpc become non-viable by term (present as resorptions) resulting in a 'perceived' absence of IUGR at term in the CBA/CaH × DBA/2J pregnancies. There is precedence for this, with pregnant mice treated with 6 mg of the yeast extract Malucidin at 16-17 dpc, showing complete resorption of a fetus after 48 hr 22 . Even in large mammals, such as dogs, complete resorption of a fetus takes only 2-5 days and is preceded by reduced growth and/or development of the embryo and fetus 23 . Placentae from IUGR fetuses show histopathology. Macroscopically, the murine IUGR fetuses were developmentally delayed with smaller, thinner and/or pale placentae in comparison to normal-weight littermates (> 10 th percentile; Fig. 4a). Quantitation of placental characteristics confirmed that IUGR placentae from (CBA/CaH × DBA/2J) F1 weighed on (mean) average 24% less at mid-gestation (p = 0.011) and 47% less at term (p < 0.0001) than their normal-weight littermates (Fig. 4b). Representative histological sections of IUGR placentae are shown in Fig. 4c (c.f. normal-weight in Fig. 4e). Evidence of pathological changes including thrombosis and fibrosis were present in placentae from IUGR fetuses of all groups (Fig. 4d), consistent with placental patho-histological observations of human IUGR placentae 24,25 .
Among the IUGR fetuses of all strains there were no differences in placental dimension at either time point (Table 1). In mice the primary site of feto-maternal exchange is the labyrinth, with the maternal decidua providing anchorage to the uterine wall and the fetal chorionic plate delivering blood to/from the umbilical cord. The proportion of IUGR placentae dedicated to the labyrinth layer was significantly less in (CBA/CaH × DBA/2J) F1 compared to (CBA/CaH × CBA/CaH) F1 and (CBA/CaH × Balb/c) F1 at mid-gestation, but did not reach significance at term (Table 1, and see Fig. 4f for colour rendering of placental layers). The trophospongium and maternal decidual layers were highly variable among the (CBA/CaH × DBA/2J) F1 cohort at 14.5 dpc indicating there may be a range of dysfunction among the placentae in this cohort (i.e. mild, moderate, severe).
One indicator of placental function, or the efficiency of placental nutrient transport, is the ratio of fetal wet weight to placental wet weight in grams 26 . In addition, with the placenta acting as a nutrient sensor and storage site for glucose (as glycogen) during pregnancy, we sought to determine whether there are changes in the proportion of glycogen-containing cells, and by extension the energy repository, available for fetal growth and development of (CBA/CaH × DBA/2J) F1. There was no significant difference in placental efficiency or the proportion of glycogen-containing cells at either time point between the (CBA/CaH × DBA/2J) F1 and (CBA/CaH × Balb/c) F1 IUGR placentae, although trends toward reduced efficiency and glycogen storage were present (Table 1). Meanwhile, the IUGR placentae of (CBA/CaH × CBA/CaH) F1 showed the highest placental efficiency of the three strains ( Table 1).

Comparison of placental morphology in IUGR and normal-weight fetuses. Normal-weight data
is provided in Table 2 and micrographs in Fig. 4e. The p-values for the statistical comparison between placentae of IUGR and normal-weight fetuses of the same strain are provided in Supplementary Table I. Consistent with reduced placental weight, the IUGR placentae of all strains were significantly smaller in dimension (diameter, thickness and/or cross-sectional midline area) than their normal-weight littermates with the differences more pronounced at term than mid-gestation (Fig. 4a,e, and Supplementary Table I). When compared to normal-weight fetuses at mid-gestation, the placentae of IUGR fetuses from (CBA/CaH × DBA/2J) F1, but not the other strains, had smaller labyrinth (39.7 vs 52.4%, p < 0.0001) and larger trophospongium (41.4 vs 24.7%; p < 0.0001) (Supplementary Table I). Combined with the aforementioned analysis of placental structure of At term all strains demonstrated significantly greater placental efficiency for normal-weight fetuses compared to IUGR (all p < 0.0001; Table 1), though only the (CBA/CaH × Balb/c) F1 IUGR placentae showed differences in placental layer composition compared to normal-weight fetuses; less labyrinth (48.6 vs 63.5%; p < 0.0001), and larger trophospongium ( 29.1 vs 19.8%; p < 0.0001), maternal decidua (22.3 vs 16.7%; p = 0.0004) and glycogen storage (14.1 vs 9.5%; p < 0.0001; Table 1).
Normal-weight fetuses of CBA/CaH × DBA/2J pregnancies show placental abnormalities. The increase in embryonic/fetal mortality, IUGR and abnormal placental structure in (CBA/CaH × DBA/2J) F1 IUGR placentae raised the question of whether there are also perturbations of the placenta evident for fetuses that macroscopically appear normal and have normal weight. In short, is there a predisposition in the (CBA/ CaH × DBA/2J) F1 placenta that leads to IUGR and embryonic/fetal mortality? In the normal-weight fetal population, placentae from (CBA/CaH × DBA/2J) F1 were significantly smaller in diameter (p < 0.0001) and midline cross-sectional area (p = 0.0006) at term and significantly higher proportion of labyrinth and lower proportion of trophospongium at mid-gestation and term, when compared to (CBA/CaH × Balb/c) F1 ( Fig. 4e and Table 2). Therein while these fetuses still achieved a normal body weight there are notable differences in the morphology of their placentae which may impact placental function.

Discussion
The data in this study demonstrates that the IUGR and embryonic/fetal mortailty phenotype of the CBA/J × DBA/2J model can also be elicited by using the Commonwealth-derived CBA/CaH substrain. First, prevalence of IUGR and mortality was higher at 14.5 dpc in (CBA/CaH × DBA/2J) F1 than (CBA/CaH × Balb/c) F1. Second, in all strains IUGR fetuses were associated with lighter, smaller placentae with histopathology and reduced materno-fetal exchange as evidenced by smaller labyrinth and lower placental efficiencies when compared to normal-weight fetuses of the same strain. Finally, despite some fetuses obtaining 'normal' body weight, the placentae from these (CBA/CaH × DBA/2J) F1 show altered glycogen storage and considerable T-lymphocyte and macrophage inflammatory infiltration compared to (CBA/CaH × Balb/c) F1. These findings suggest that there is a range of placental pathology in the CBA/CaH × DBA/2J pregnancy which dictates the fetal outcome; severe pathology leads to early resorption, moderate to IUGR at 14.5 dpc with subsequent demise by term, and mild which supports live birth.
Among inbred murine strains there is a spectrum of gestational lengths, fetal and placental weights, maternal weight gain and number of concepti 29 . While gestational length and concepti are largely dependent upon maternal genotype 29 , which was unchanged in our study, fetal and placental weight are dependent on the genetic composition of the embryo. Of our three inbred strains, the syngeneic (CBA/CaH × CBA/CaH) F1 had the lowest fetal and placental weights of our normal-weight cohorts, consistent with the 14-20% lower fetal and placental weight reported in allogeneic vs syngeneic inbred matings [30][31][32] . The acquirement of heterosis in F1 fetuses is shown to precede implantation, with blastocysts of syngeneic pregnancies containing more cells than allogeneic pregnancies at the same gestational age 33 . As placentation continues the mixed genetic background of the fetuses may become more influential with differences in remodelling of the uterine vasculature and maternal immunological response associated with paternal MHC expression 34 . Yet, precisely how the mixed genetic background of (CBA/CaH × DBA/2J) F1 fetuses leads some to achieve normal-weight while others are growth restricted or die is unknown.
Technological advancements and greater appreciation of the placenta as an immunomodulatory organ have provided increasing evidence that dysregulation of placentation, and not a loss of maternal immune tolerance against the semi-allogeneic fetus per se, underlie human and murine (CBA/J × DB/2JA) IUGR and mortality [8][9][10]35 . Following implantation fetal trophoblasts invade the maternal decidua and partner resident natural killer (NK) cells 36 to remodel spiral arteries by inducing apoptosis of endothelial cells and vascular smooth muscle cells by soluble Fas ligand [37][38][39] and tumour necrosis factor-alpha-related apoptosis-inducing ligand. Apoptotic cells are phagocytosed, typically by decidual macrophages, though first trimester trophoblasts are phagocytic 40 and may also play a role in clearing these cells during spiral artery remodelling. Such aberrations in the function of the trophoblasts or NK cells results in a failure to convert the maternal spiral arteries into low resistance, high capacity vessels 39 . Subsequent perfusion of the placenta with maternal blood at high pressure incites trophoblast damage via shear stress and mechanical damage 41 leading to thrombotic and fibrotic events as noted in the IUGR placentae of all murine strains in our study, and infarcts, loss of tertiary villi and increased syncytial knots (associated with syncytiotrophoblast apoptosis) among the placentae of human IUGR newborns 42 . Such injury to the human placental villi or murine labyrinth reduces the surface area of materno-fetal exchange contributing to IUGR.
Delineation of the initial mechanism that renders trophoblasts or decidual NK cells less effective at spiral artery remodelling is a relatively new field of research and is complicated by access to very early human placental tissue and the differences between invasiveness of murine and human trophoblasts. From the CBA/J × DBA/2J model there is evidence of inadequate placentation via poor maternal arterial remodelling at the implantation sites 8 and perturbations of genomic DNA methylation in the maternal decidua 43 . Other studies report inflammatory mechanisms, with complement deposition on trophoblasts with increased neutrophil and monocyte infiltration and TNF-α expression in maternal decidua and reduced trophoblast giant cells in placentae of resorbing fetuses 6 . Both abnormal DNA methylation and complement deposition appear to distinguish normally developing fetuses from those which resorb, even within the same litter 6,43 . Our study compliments this evidence wherein perturbations in placental size and morphology were observed in placentae of IUGR fetuses when compared to normal-weight littermates. In humans, these parameters dictate the number of maternal spiral arteries that can be remodelled and subsequently influences the surface area of nutrient exchange in the placenta 44 affecting fetal growth and development.
It is understandable that the outcome of any fetus is dependent on the severity of impact to the placenta, as the key regulator of both oxygen and nutrients. Mild alterations in placental morphology or inflammation below a threshold do not observably impact on weight or CRL but may evoke epigenetic or anatomical changes that alter ongoing fetal-adult development. To our knowledge the CBA × DBA model has not been examined as far as pup development following birth, largely due to the need to excise uteri and fetuses for assessment of embryonic/ fetal mortality and IUGR. However, as ultrasound biomicroscopy technology becomes more cost-effective, such non-invasive assessments of viable vs non-viable murine fetuses on the basis of heart beat 45 , as well as in situ murine fetal growth measurements 46 will be reported and enable long term outcomes to be assessed in this model. Of particular note would be cognitive function and behavioural impacts as these are commonly associated with children who were IUGR 47 , despite 'catching up' in weight after birth. This will be essential for assessment of novel therapeutics to reduce embryonic/fetal mortality and IUGR using the CBA × DBA/2J model.
In summary, we have provided a comprehensive assessment of fetal outcome and placental dimension and immune infiltrate temporal and spatial anomalies at two gestational time points (14.5 and 18.5 dpc), with both allogeneic and syngeneic control matings, to demonstrate that the CBA/CaH × DBA/J is an equivalent murine model of IUGR and embryonic/fetal mortality to the CBA/J × DBA/2J model for assessment of pregnancy pathology and/or novel therapeutics.

Methods
Animals and Tissue Preparation. All experiments were approved and performed in accordance with the guidelines of the Northern Sydney Local Health District Animal Research Ethics Committee, Australia (ACEC 1311-015A). Twelve-week old virgin female CBA/CaH.Arc and male CBA/CaH.Arc (H2k), Balb/c.Arc (H2d), and DBA/2J.Arc (H2d) mice were purchased from the Animal Resources Centre (Murdoch, Australia). Animals were provided food and water ad libitum and kept under specific pathogen-free conditions with controlled humidity and light conditions. Timed syngeneic CBA/CaH × CBA/CaH, and allogeneic CBA/CaH × Balb/c or CBA/CaH × DBA/2J matings were performed by pairing females with males overnight. Detection of the vaginal plug denoted 0.5 days post coitum (dpc). At 14.5 dpc (mid-gestation; N = 20 dams per strain) and 18.5 dpc (term; N = 30 dams per strain) pregnant mice were sacrificed by cervical dislocation and the gravid uteri, including the cervix, were harvested and weighed. Subsequently, each fetus and placentae were dissected, weighed, the fetal crown-rump length measured, and the litter photographed. The number of concepti and embryonic/fetal mortality (resorption) rate were also recorded. Embryonic/fetal mortality was defined as including 'macerated pale fetuses (embryonic)' , bag of cells, and 'clot' 18 . Pink/perfused fetuses were considered viable. Overall there was ≥ 80% pregnancy success in each group. In the term population, two pregnant dams from each group delivered pre-term at 17.5 dpc and were excluded from subsequent analyses.
Placental Structure Quantification. Quantitation of the placental dimensions (diameter, thickness, and midline cross-sectional area) were analysed using haemotoxylin and eosin stained sections and the placental layers (labyrinth, trophospongium and maternal decidua) using PAS-stained sections. All histological analyses were performed using two parallel sections per IUGR and normal-weight placentae taken at least 100 μ m apart. Low magnification micrographs were taken using a Nikon Eclipse Ci microscope fitted with a DS-Fi1 digital camera and DS-U2 control unit and 2x Plan Achromat objective lenses. From the scale bar placental diameter, thickness, and midline cross-sectional area were measured using ImageJ 1.48v software (http://imagej.nih.gov/ij) and expressed in millimetres. The area attributed to the labyrinth, trophospongium and maternal decidual layers was assessed and expressed as a percentage of the total area of the placenta (excluding fetal circulation structures). The percentage of glycogen-containing trophoblasts in PAS-stained sections was quantified by magenta-coloured pixels in micrographs using ImageJ 1.48v software (http://imagej.nih.gov/ij). A macro to automate the analysis was modified from the freely available ImageJ resource at http://imagej.net/docs/examples/stained-sections/ index.html (NIH) and utilised the green channel and a magenta colour threshold of (0, max/1.4). For each section the data was calculated as the area of magenta coloured pixels/placental midline cross-sectional area*100 and expressed as glycogen-containing cells as a proportion (%) of the placental area.
Statistical Analyses. Normality of the data for each outcome was determined by the D' Agostino & Pearson omnibus K2 normality test (GraphPad Prism; NIH), except the IUGR 18.5 dpc cohorts due to low sample size (normal distribution was assumed using the Central Limit Theorem as the placentae are derived from normally distributed populations; see Fig. 3b). Data are expressed as mean ± standard error of the mean (SEM). Comparison between groups at 14.5 dpc and 18.5 dpc was performed using unpaired two-tailed Student t-test and between groups of a single time point by One-way ANOVA with Tukey's multiple comparison test. Association between crown-rump length and fetal weight was determined by Pearson correlation. The Chi square test was used to compare the number of concepti between left vs right uterine horns, the frequency of perinatal mortality in 'crowded' vs 'uncrowded' horns, incidence of placentally-fused twins, and differences in proportion of litters with resorbing fetuses vs those without. A p-value < 0.05 was considered statistically significant for all analyses.