Human pregnancy, critical for our species survival, is inefficient and prone to complications such as infertility, spontaneous miscarriages and preeclampsia (PE). Immunological factors may be important as the embryo is 50% paternal and foreign to the mother. Mouse pregnancy models, and in particular the murine CBA/J x DBA/2 mating combination, has been widely used to investigate mechanisms causing and preventing partner-specific recurrent miscarriages (RM) and PE. Occult losses can represent T cell-mediated rejection, and antigen-specific regulatory T cells (Tregs) with classical αβ T cell receptors (TcR) activated by semen antigens at the time of mating are protective. If there is no occult loss, an inadequate Treg response can also predispose to RM. In RM, proinflammatory cytokines from natural killer (NK)-type cells and macrophages of the innate immune system are responsible and cells with γδ TcR protect via release of TGF-β-type molecules. Immunization of abortion-prone female CBA/J mice or administration of cell-associated or soluble CD200, an immune check point inhibitor, can prevent abortions by augmenting uterine decidual suppressor cell activity. Human studies suggest that is also true in couples with RM. Environmental activators of the innate immune system, such as bacterial LPS and stress, can cause abortions as well as occult losses. The endogenous level of Tregs and activation of Tregs specific for the male H-Y antigen may determine success rates and alter the male:female birth ratio. Intralipid alters LPS clearance, prevents abortions in the CBAxDBA/2 model, and is effective in increasing live birth rates in couples undergoing IVF treatment.

Introduction

Animal models are commonly used when human physiological and pathological processes are studied in a non-human species. The assumption is that animal model accurately recapitulates human physiology and pathophysiology. Much current scientific activity involves collecting observational and experimental data in support of the prevailing model (i.e. theory), and granting agencies find it ‘safe' to support such work. A large body of literature may accumulate that supports the popular model, in part because findings that do not fit are hard to publish. But as Karl Popper has rightly instructed, models and theories are actually hypotheses that must be validated by scientific testing, and a hypothesis is only scientific if one can do an experiment where the result could invalidate and replace it with something new [1]. Whenever one uses an animal analog model similar to a human disorder, there is a cost. For example, an inbred laboratory mouse or rat is unlikely to perform identically to all members of an outbred human population. To determine if information and insights obtained from studying the model, one must test whether the findings that provide insights into pathogenesis are actually replicated in the human disorder. In this process one first looks for similarities and differences pertinent to diagnosis and prognosis. Second, one conducts a therapeutic clinical trial based on the outcome of administering a particular treatment to the animal analog. This process has been extensively applied in the search for effective remedies for human malignancies, and is illustrated in Figure 1.

Operational paradigm for efficient use of an animal model to understand human disease.

Figure 1.
Operational paradigm for efficient use of an animal model to understand human disease.

Correlational data is available from studies in either animals or human, but experimental manipulations that can be done in animals can elucidate causation and the pathophyiological pathways that result in what is similar to human disease. In vivo data may lead to in vitro investigation which must then be replicated in vivo. But then one must test the understanding by investigating humans to see if similar events occur and to show that treatments that worked in the animal model produce the same outcome in humans. The inbred laboratory mouse is the preferred in vivo model for investigating immunological factors in specific diseases. There are significant differences between mice and humans, and clinical trials predicted by mouse studies to work in humans do not always work [2].

Figure 1.
Operational paradigm for efficient use of an animal model to understand human disease.

Correlational data is available from studies in either animals or human, but experimental manipulations that can be done in animals can elucidate causation and the pathophyiological pathways that result in what is similar to human disease. In vivo data may lead to in vitro investigation which must then be replicated in vivo. But then one must test the understanding by investigating humans to see if similar events occur and to show that treatments that worked in the animal model produce the same outcome in humans. The inbred laboratory mouse is the preferred in vivo model for investigating immunological factors in specific diseases. There are significant differences between mice and humans, and clinical trials predicted by mouse studies to work in humans do not always work [2].

The remainder of this review will show the successful and continuous application of this strategy to understand human pregnancy failure using mouse models. A more extensive commentary on models of other human reproductive problems may be found elsewhere [3].

Human pregnancy failure

Determinants of species survival and self non-self discrimination

Reproductive success is crucial for propagation and survival of our species, but offspring must survive to sexual maturity and be able to take care of their offspring until they are independent. Much depends on social and environmental factors [5]. Survival also requires an adequate immune defense against infectious agents and parasites. In the adaptive immune system, which generates effector thymus-derived lymphocytes (T cells) and antibody-forming B cells (which usually require T cell help), macrophage phagocytosis and digestion of the invader's tissues creates small peptide antigens that are presented to the antigen-specific receptor on T cells by being bound in a grooves of a major histocompatibility antigens (MHC, H-2 in the mouse, HLA in humans). HLA-A, -B, -C and -D antigens are polymorphic and consequently, bind a spectrum of different 8–12 amino acid peptides. Self peptides bound in the groove of one's own MHC must be displaced by foreign peptides in order to be engage the T cell receptor (TcR). CD4+ T cells recognize peptides bound to Class 2 MHC (HLA-D) which is expressed on the surface of antigen-presenting lymphomyeloid cells, whereas CD8+ T cells recognize peptides bound to Class 1 MHC (HLA-A, -B, -C) which is expressed on the surface of all nucleated somatic cells. CD4+8+ T cells can do both. For more detailed information, the reader should consult some recent reviews [3,4].

In an outbred population, each individual (unless they are identical twins) possesses a different MHC-peptide phenotype, and that results in transplanted tissue from another being rejected, absent treatments to suppress the recipient's immune response. When inbred mammalian individuals with different sets of MHC antigens (e.g. HLA-A1,A2 and HLA-A3,A4) and associated self peptides mate, the progeny will have a polymorphic MHC that is different from each parent (e.g. HLA-A1,A3, HLA-A2,A3, HLA-A1,A4, HLA-A2,A4). If the parents are homozygous but have different MHC antigens (e.g. HLA-A1,A1 and HLA-A3,A3), the progeny (e.g. HLA-A1,A3) will mount an immune response to a smaller number of antigenic peptides of an invading organism such as a virus. That decreases the chance the progeny will survive if the virus is lethal to one of the parental MHC phenotypes that fails to stimulate a robust T cell response. The same consideration applies to loci for HLA-B, HLA-C, and HLA-D alleles. Humans do not select their mate on the basis of HLA typing to ensure children will have disparate HLA phenotypes. Disease pressure in certain geographic regions has led to development of different frequencies of particular advantageous HLA alleles and molecules on host immune cells that recognize MHC determinants, but that population may have a larger number of individuals with increased susceptibility to fatal outcomes should an entirely new infectious agent arrive (e.g. resistance to coronaviri) [6,7].

There is a non-antigen specific set of lymphomyeloid cells that includes macrophages and natural killer (NK) cells that act immediately mediating innate immunity before the adaptive peptide-MHC antigen-specific T cell immune response can generate effector cells. This innate immune system is older from the evolutionary standpoint and primitive self non-self discrimination based on surface recognition molecules such as in the colonial ascidian, Botryllis, which sexually reproduces, began eons before development of an antigen-specific adaptive immune system [8]. NK cells express receptors for Class 1 MHC antigens that can activate or deactivate function [7].

During evolution, some species reproduced by laying eggs, and others reproduced by viviparity,- a process evident as early as development of sharks [8]. The problem with viviparity is the embryo inherits 50% of its DNA coding for antigens of the father, and must survive within the mother whose immune system is quite capable of rejecting cells expressing the father's MHC-peptide antigens. Curiously, optimal reproduction occurs when the embryo does express paternal alloantigens foreign to the mother, which suggests the embryo may elicit a pregnancy-protective immune response in the mother [3,7]. Female mice have olfactory mechanisms and preferentially mate with the male with the more foreign MHC (H-2) but evidence for a human equivalent is scant so societal and legal prohibitions discourage obvious matings of closely related individuals. Reproductive behavior and success if undertaken drops when foreignness is too great, as with a xenogeneic male, or too similar, as with incestuous inbreeding. Nevertheless, there are inbred human populations, such as Parsi and Hutterites, and of course, inbred strains of mice, but long term survival of inbred mouse strains were the exception, and maintaining a small degree of heterozygosity was and is necessary.

Human matings are not always successful. There is infertility and failure of recognizable pregnancies. About 15% of pregnancies miscarry, 5% with two losses and 1–2% of couples with three losses have recurrent miscarriages (RM) which importantly was partner-specific [3,9]. In addition to emotional suffering, there are health care expenses, and the latter can be augmented by desperate couples seeking expensive treatments such as in vitro fertilization (IVF). Recurrent embryo loss also occurs in domestic animals, and that reduces the food supply. Therefore, it is important to understand the causes of pregnancy loss.

Human reproduction is inefficient

Roberts and Lowe [10] were among the first to report that a human conception did not always result in a live birth. In fact, they estimated only ¼ conceptions succeeded, and this was not due to elective terminations. In 1975, Boué et al. [11] reported 61.5% of 1498 first trimester abortuses in women age 20–40 had an abnormal chromosome complement (karyotype). If the abortus tissue had a normal karyotype, the probability of miscarriage in the next pregnancy increased by a factor of 1.35–1.56, depending on maternal age [11]. Subsequently, with advent of IVF, 50% oocytes and 10% of spermatozoa were found to be abnormal, and the fate of 1000 matings could be calculated as shown in Figure 2.

Timing of human reproductive failure.

Figure 2.
Timing of human reproductive failure.

A 22.5% live birth rate per cycle is consistent with Roberts and Lowe, and Evers [10,12]. *Calculations are explained in Clark [11]. The fetal trophoblast produces β-hCG which is detectable in the circulation on the day the blastocyst implants. **In normal fertile women, occult loss of normal karyotype embryos could be as low as 0. In infertile women, it could be as high as 159. In recurrently aborting women, if 50% of abortuses are karyotype normal, it could be as high as 148. Measured occult loss rates in non-RM women range from 17–45% depending on stringency of criteria used to diagnose implantation in women attempting conception [3,13]. Drawings depicting the peri-implantation embryo and subsequent feto-placental stage have been reproduced with permission from the publisher from Clark DA. The importance of being a regulatory T cell in pregnancy. J Reprod Immunol 2016;116:60–69 [25]. In peri-implantation drawing, ICM, inner cell mass, (1) denotes site of fetal antigen-mother interaction. In post-implantation drawing, EVT, extravillous trophoblast, ML, maternal lymphoid cells, FL, fetal lymphoid cells, CT chorionic cytotrophoblasts. Site (2) shows site where HLA-G+ HLA-C+ extravillous fetal trophoblast cells invade decidua and the walls of maternal spiral arteries where trophoblast forms endovascular plugs and can contact lymphoid cells in maternal blood. At site (3), maternal blood cells may contact antigen-negative fetal syncytiotrophoblast cells, but more importantly, maternal cells may cross into fetal tissue and traffic to fetal liver and other organ sites. Fetal lymphoid cells also cross into maternal blood and gain access to various maternal tissues.

Figure 2.
Timing of human reproductive failure.

A 22.5% live birth rate per cycle is consistent with Roberts and Lowe, and Evers [10,12]. *Calculations are explained in Clark [11]. The fetal trophoblast produces β-hCG which is detectable in the circulation on the day the blastocyst implants. **In normal fertile women, occult loss of normal karyotype embryos could be as low as 0. In infertile women, it could be as high as 159. In recurrently aborting women, if 50% of abortuses are karyotype normal, it could be as high as 148. Measured occult loss rates in non-RM women range from 17–45% depending on stringency of criteria used to diagnose implantation in women attempting conception [3,13]. Drawings depicting the peri-implantation embryo and subsequent feto-placental stage have been reproduced with permission from the publisher from Clark DA. The importance of being a regulatory T cell in pregnancy. J Reprod Immunol 2016;116:60–69 [25]. In peri-implantation drawing, ICM, inner cell mass, (1) denotes site of fetal antigen-mother interaction. In post-implantation drawing, EVT, extravillous trophoblast, ML, maternal lymphoid cells, FL, fetal lymphoid cells, CT chorionic cytotrophoblasts. Site (2) shows site where HLA-G+ HLA-C+ extravillous fetal trophoblast cells invade decidua and the walls of maternal spiral arteries where trophoblast forms endovascular plugs and can contact lymphoid cells in maternal blood. At site (3), maternal blood cells may contact antigen-negative fetal syncytiotrophoblast cells, but more importantly, maternal cells may cross into fetal tissue and traffic to fetal liver and other organ sites. Fetal lymphoid cells also cross into maternal blood and gain access to various maternal tissues.

Paternal MHC Class 1 antigens and minor antigens such as H-Y (male specific) could be detected as early as the 2 cell stage of oocyte division, and continued to be expressed on the blastocyst which was protected by a cocoon of zona pellucida [3]. When blastocysts hatch in preparation for implantation, MHC Class 1 antigen expression disappears. In the mouse, mRNA for class 1 MHC reappears 3 days after implantation, on day 7.5 of gestation, when ectoplacental cone trophoblast begins the placentation process, and this timing is comparable to a human embryo at 5 weeks gestation (12–13 days post implantation) [3]. MHC antigens are not expressed on fetal trophoblast cells which directly contact maternal blood in the placenta, but are expressed on trophoblast cells in contact with maternal decidual cells. The fetus expresses both Class 1 and Class 2 MHC, with Class 2 being particularly immunogenic when expressing antigenic peptides foreign to the host, but fortunately, the fetus, and its circulation, are separated from maternal blood by a barrier of fetal trophoblast cells, so cellular traffic into and out of the fetus is limited [3].

The transplantation immunology paradigm

In 1953, Medawar, a transplantation immunologist known for his contribution to the study of immunological tolerance, had discussed successful human pregnancy as an example of paradoxical survival of an allograft for 9 months within the body of an immunocompetent mother [3]. When it became known that karyotype normal embryos were being lost, interest of immunologists in mechanisms determining survival or failure of early pregnancy was piqued. Furthermore, rejection of organ transplants occurred in spite of MHC (HLA antigen) matching of donors and recipients who were not identical twins, probably due to differences in donor and recipient non-MHC peptides that bound to the MHC antigen groove. Transfusing the donor's blood cells into the recipient before transplantation reduced the rejection rate. Adventurous clinicians decided to test for a similar effect in women with RM where a loss could not be associated with a maternal or paternal abnormality, and initial success in pilot studies lead to the first prospective controlled trial by Professor James Mowbray at St. Mary's Hospital in London [3,14]. An Armitage design paired paternal mononuclear leukocyte-injected (LIT) women with autologous leukocyte-injected controls when pregnancy occurred. In couples who conceived quickly, a significant benefit of paternal cells was reached by the 22nd pair, with a significant effect size (69% success with paternal cells, 41% success with autologous cells, mean effect size and SD = 28 ± 11.3%).

Subsequently, two randomized controlled trials (RCTs) done to ‘reproduce the Mowbray result' failed to achieve statistical significance [14]. To compound the controversy, women in the Mowbray trial likely knew which treatment they were getting, and the Stray Pedersens reported an observational study suggesting that psychotherapy might be effective in preventing RM [14,15]. The Ethics Committee of the American Society for Reproductive Immunology then conducted a rigorous meta-analysis of primary patient data from controlled trials of allogeneic leukocyte immunotherapy (LIT) done in multiple centers with two independent statistical analyses to minimize bias. Some statisticians were horrified that the results of the two analyses would differ, but they did not, and why would one trust a result that was not independently reproducible? The effect size, while statistically significant, was only an 8–10% increase in the live birth rate, but was 16% in primary RM women without anti-cardiolipin (ACL) or anti-nuclear (ANA) or anti-paternal HLA antibodies [14]. Interestingly, a larger effect size occurred in double blinded randomized controlled trials (DB-RCTs) compared with unblinded RCTs, and cohort controlled studies showed the least benefit contrary to the preconceived notion that cohort studies must be biased in favor of a positive result [15]. In the RMITG study, which analyzed live birth rate by intention-to treat (whether or not the woman became pregnant), the original Mowbray result lost statistical significance [14]. In part, this was due to inclusion of women who took longer to achieve pregnancy (24 of the 69 entered). Mowbray had reported that immunized women who had not developed anti-paternal HLA antibodies and took >80 days to become pregnant aborted and required a booster injection within 10–14 days of their missed period [3]. A primary immunization within 10–14 days of a pregnancy (missed menses and positive β-hCG) was effective and has recently been confirmed [3,16]. Anti-paternal HLA antibodies, implicated in blocking rejection some models of transplantation tolerance, were unnecessary for a live birth, but were important for prolonged protection >3 months [3]. In the RMITG study, women who already had anti-paternal HLA antibodies (i.e. had had an immune response) manifest a high success rate without immunization which was not increased by LIT [3,14]. The RMITG study concluded the treatment was either partially effective in all RM patients or it was highly effective in a subset. In support of the latter idea, women with antinuclear and anti-cardiolipin auto-antibodies had a worse outcome with immunization, and women with primary RM and no antibodies had a better outcome [14,15]. When the outcome in this group was reevaluated correcting for the potential impact of losses caused by embryo karyotype abnormalities, Figure 3A emerged [17].

Panel A shows the probability of a live birth of offering allogeneic blood mononuclear cell immunotherapy (LIT) to women with primary RM and no serum antinuclear or anti cardiolipin or paternal HLA antibodies.

Figure 3.
Panel A shows the probability of a live birth of offering allogeneic blood mononuclear cell immunotherapy (LIT) to women with primary RM and no serum antinuclear or anti cardiolipin or paternal HLA antibodies.

Shaded area represents estimated loss of karyotype abnormal embryos based on 55% rate from Coulam et al. for primary RM. The figure is reproduced with permission from the publisher from Clark DA, Daya S, Coulam CB, Gunby J, and the Recurrent Miscarriage Immunotherapy Trialists Group. Implications of abnormal human trophoblast karyotype for the evidence-based approach to the understanding, investigation, and treatment of recurrent spontaneous abortion. Am J Reprod Immunol 1996;35:495–498 [17]. Panel B compares karyotype data of Coulam et al. with that of Ogasawara et al. [9,18]. The confidence limits of % abnormal karyotypes for 3–6 prior losses described by Coulam et al. includes the mean value for the same population described by Ogasawara et al. and vice versa, so any difference in the mean % is not statistically significant.

Figure 3.
Panel A shows the probability of a live birth of offering allogeneic blood mononuclear cell immunotherapy (LIT) to women with primary RM and no serum antinuclear or anti cardiolipin or paternal HLA antibodies.

Shaded area represents estimated loss of karyotype abnormal embryos based on 55% rate from Coulam et al. for primary RM. The figure is reproduced with permission from the publisher from Clark DA, Daya S, Coulam CB, Gunby J, and the Recurrent Miscarriage Immunotherapy Trialists Group. Implications of abnormal human trophoblast karyotype for the evidence-based approach to the understanding, investigation, and treatment of recurrent spontaneous abortion. Am J Reprod Immunol 1996;35:495–498 [17]. Panel B compares karyotype data of Coulam et al. with that of Ogasawara et al. [9,18]. The confidence limits of % abnormal karyotypes for 3–6 prior losses described by Coulam et al. includes the mean value for the same population described by Ogasawara et al. and vice versa, so any difference in the mean % is not statistically significant.

The risk of a subsequent loss in the next pregnancy increased with the number of previous losses, and obviously with maternal age as those with many losses were older. Since few of the RM women in the RMITG study had had fetal tissue from abortuses karyotyped, the expected % of losses with an abnormal karyotype was computed and is shown as a grey area based on Coulam et al. who had a ≥94% success rate in karyotyping failing pregnancies [9]. The grey shading that tracks the success rate closely that suggests immunization was highly effective. Subsequently, Ogasawara et al. [18] claimed the likelihood of loss of a normal karyotype pregnancy increased with the number of prior miscarriages. However, Ogasawara et al. only had a 51.1% success rate in karyotyping electively terminated miscarriages, included secondary RM (which from Coulam et al. had only a 35% abnormal karyotype incidence compared with 55% in primary RM), and Ogasawara's data for those with 3–6 consecutive abortions had an abnormal karyotype rate that was not statistically different for the pooled primary and secondary RM data of Coulam et al. as shown in Figure 3B. Conclusions about the increasing likelihood of a normal karyotype with increasing numbers of prior abortions by Ogasawara et al. seems to reflect inclusion of 31 young women with 7–20 losses who were younger (≤34 years of age) compared with patients of Coulam et al. and their mean % abnormal karyotype rate was only 10/31 (32.3 ± 8.4%, upper 95% confidence limit 48.8%). Details of each case were not provided by Ogasawara et al. so one does not know if the RM in every case was partner-specific or not. Lack of karyotype data on each abortus in sequential miscarriages also limits one's ability to draw firm conclusions.

The rigor of the conclusions reached in the RMITG study did not persuade skeptics who believed one had to understand the mechanism by which otherwise normal embryos were ‘saved’. Since improving treatments and selecting patients most likely to benefit from an immunomodulatory intervention was and remains an important goal for physicians, the need for insights from an animal analog model was evident.

The first mouse model of spontaneous abortion

In 1980, Clark et al. [19] reported that CBA/J (H-2k) females mated to DBA/2 (H-2d) males had an unexpectedly high rate of resorption (a form of miscarriage) associated with absence of suppressor cells in the lymph nodes draining their uterus. These suppressor cells inhibited generation of anti-H-2d cytotoxic T cells thought at that time to cause allograft rejection [19]. In 1983, Chaouat et al. reported that preimmunization of the CBA/J females (H-2k) before mating with cells bearing paternal H-2d on a BALB/c background could reduce the abortion rate from 23% (in CBA/J pre-immunized female) to 5.2% (95% confidence interval 0.5–10.9%), P < 0.001 [20]. The spleen cells of naïve CBA/J females 4 days after BALB/c cell foot-pad immunization dramatically suppressed an in vitro mixed lymphocyte reaction of CBA/J stimulated by splenocytes expressing H-2kd. The nature of the putative spleen suppressor cells induced by BALB/c, but not DBA/2 cells, was not further elucidated. Subsequently it was noted that a CBA/JxBALB/c pregnancy prevented abortions when that female was subsequently mated to DBA/2, so prevention of abortion was not an artifact arising from the immunization procedure. The rationale for use mouse models for human pregnancy loss is set out in Table 1.

Table 1
Why choose an inbred CBAxDBA/2 mouse animal anlog model
ReasonsProblems
1. Inbred strains advantage: (a) defined genetics
(b) treated and controls same
(c) recombinant inbred allows study of different male and female MHC antigen phenotypes, (c) knock-outs and knock-ins
(d) whilst CBAxDBA/2 now dominates RM research, other models exist, such as B10xB10.A where B10.A cells immunize [3]
(e) resorb but can abort. 
1. (a) Humans are not inbred so model = a small subset.
(b) Even in an assumed population of ‘identical' individuals, the immune response is variable.
(c) In DBA/2 mated CBA/J, DBA/2 splenocytes do not immunize against abortion: need MHC-identical BALB/c cells or a CBAxBALB/c pregnancy.
(d) Low abortion rate in B10xB10.A where paternal cell immuization prevents abortion is too low. 
2. Immunology similar & have tools and techniques to elucidate mechanisms of allograft rejection and mechanisms causing ‘tolerance'. 2. (a) Mouse immune system is not identical with humans [2]
(b) Abortion in mouse is not = allograft rejection [3]. Pregnancy success not = systemic tolerance [7
3. (a) Mouse has a haemo-chorial labyrinthine placenta.
(b) Two waves of H-2K+ trophoblast invasion of decidua [21] and endovascular plugs
(c) Two types of decidual NK cells are present [22]
(d) Vγ1δ6.3T cells make a novel TGFβ2-related molecule. B cells IL-10.
(e) Spiral arteries modifed by trophoblast invasion + uNK IFN-γ remodeling. 
3. (a) Human placenta haemo-chorial with villous syncytio-trophoblast
(b) Two waves of extravillus HLA-G+ HLA-C+ trophoblast invasion of decidua and endovascular plugs (b) Novel ‘true' uterine NK cells (TuNKs) dominate, unlike in the mouse [3,22]
(c) Vγ9δ2T cells make IL10 and TGFβ1[23]
(d) Arteries modified by trophoblast invasion 
4. Invasive experiments and animal sacrifice ethical  
ReasonsProblems
1. Inbred strains advantage: (a) defined genetics
(b) treated and controls same
(c) recombinant inbred allows study of different male and female MHC antigen phenotypes, (c) knock-outs and knock-ins
(d) whilst CBAxDBA/2 now dominates RM research, other models exist, such as B10xB10.A where B10.A cells immunize [3]
(e) resorb but can abort. 
1. (a) Humans are not inbred so model = a small subset.
(b) Even in an assumed population of ‘identical' individuals, the immune response is variable.
(c) In DBA/2 mated CBA/J, DBA/2 splenocytes do not immunize against abortion: need MHC-identical BALB/c cells or a CBAxBALB/c pregnancy.
(d) Low abortion rate in B10xB10.A where paternal cell immuization prevents abortion is too low. 
2. Immunology similar & have tools and techniques to elucidate mechanisms of allograft rejection and mechanisms causing ‘tolerance'. 2. (a) Mouse immune system is not identical with humans [2]
(b) Abortion in mouse is not = allograft rejection [3]. Pregnancy success not = systemic tolerance [7
3. (a) Mouse has a haemo-chorial labyrinthine placenta.
(b) Two waves of H-2K+ trophoblast invasion of decidua [21] and endovascular plugs
(c) Two types of decidual NK cells are present [22]
(d) Vγ1δ6.3T cells make a novel TGFβ2-related molecule. B cells IL-10.
(e) Spiral arteries modifed by trophoblast invasion + uNK IFN-γ remodeling. 
3. (a) Human placenta haemo-chorial with villous syncytio-trophoblast
(b) Two waves of extravillus HLA-G+ HLA-C+ trophoblast invasion of decidua and endovascular plugs (b) Novel ‘true' uterine NK cells (TuNKs) dominate, unlike in the mouse [3,22]
(c) Vγ9δ2T cells make IL10 and TGFβ1[23]
(d) Arteries modified by trophoblast invasion 
4. Invasive experiments and animal sacrifice ethical  

Studies in mice continued, and the so called ‘rejection' in the CBAxDBA/2 model was reported in 1998 by Clark et al. to be mediated by the proinflammatory cytokines TNF-α + interferon-γ produced by cells bearing an NK cell surface marker interacting with macrophages rather than by T cells that caused allograft rejection [24]. Simmons and Russell had reported that preimplantation embryos placed under the kidney capsule of recipients presensitized against paternal antigens showed rejection of fetal tissue that was minor histocompatibility antigen specific with no rejection of fetal trophoblast cells there was abundant evidence that induction of transplantation immunity against paternal cells failed to cause resorptions, and in fact, the fetoplacental units underwent growth stimulation [3]. Ferry et al. [3] also reported that freshly isolated fetal trophoblast cells were impervious to lysis in vitro by highly activated cytotoxic effector cells. The demise of embryos in the CBAxDBA/2 model was readily explained by an inflammatory process that terminated the developing blood supply to the placenta just as maternal blood circulation was beginning [22,24]. Subsequent research implicated a variety of other molecules involved in activation of the coagulation and complement system [3,25].

With respect to prevention of abortions, several populations of regulatory T cells (Treg cells with αβ T cell receptor for antigen, and suppressor T cells with γδ receptors) were also identified [3,23,25]. In the CBAxDBA/2 model, in vivo depletion of T cells on using anti-CD8 increased abortion rates but only if administered on gestation day 6.5 and to a lesser extent 7.5, corresponding to the time on occult losses in mice (and humans) [3]. The effect of anti-CD8 implied a suppressor cell able to recognize foreign paternal peptide bound to a Class 1 MHC molecule during the peri-implantation period which set the stage for onset of resorptions or abortions in the post-implantation period. As trophoblast did not express Class 1 at this time point in pregnancy, the paternal peptide had to be released in soluble form and taken up by maternal antigen-presenting cells (APC) in implantation site decidua and combined with maternal MHC antigens and exported to the cell surface where it could be recognized by maternal T cells.

T cell-mediated peri-implantation embryo death is possible

Although embryo death in the CBAxDBA/2 model was not T cell mediated, several mouse models demonstrated that antigenic embryos could be destroyed between day 6.5 and 8.5 of gestation in utero provided protection by Treg cells was abrogated [3]. Minor paternal antigens were involved in activating maternal T cells to produce Th1 cytokines causing a local inflammatory process wherein innate immune system cells could be activated [3]. Based on mouse model data, occult human pregnancy loss illustrated in Figure 2 could represent a form of T cell-mediated immune rejection. Validation using human implantation site tissue was not possible.

Non-specific activators of the innate immune system are key mediators of pregnancy failure

Ligands to TLR3, TLR4 and TLR5 and TLR9, such as polyI:C an analog of viral nucleic acids, bacterial lipopolysaccharides (LPS), or CPG directly activate the innate immune system cells, and pregnant mice were particularly sensitive to LPS up until day 7.5 after which a higher dose was required [3]. There were also strain differences determining which types of LPS were abortogentic. For example pregnant C57Bl/6 mice injected with E. coli O55:B155 LPS had only occult losses and required Salmonella enteritidis LPS, a TLR5 agonist, to manifest resorptions, or, E. coli O26:B6 LPS [3]. Increasing doses of E. coli O55:B155 LPS given on day 7.5 of gestation in the CBAxDBA/2 model initially boosted abortion rates, and then caused complete embryo elimination so that no implantations were evident at day 13.5 [3]. All TLR agonists required the presence of MD-1, but that did not explain the mechanism underlying the apparent strain specific effects of different types of bacterial LPS [3]. The relevance of LPS to abortions was provided by dramatic reduction in CBAxDBA/2 abortion rates by antibiotic pretreatment of the CBA/J females [3]. Variation in abortion rates among different animal colonies using the CBAxDBA/2 model was explainable in part by environmental changes in bacterial flora occurring in clean cage and SPF housing conditions [3].

Mechanisms preventing pregnancy failure

How does allogeneic mononuclear leukocyte immunization work?

How did LIT given to women with unexplained RM prevent miscarriage? The 1983 observation that abortions in the CBAxDBA/2 model could be prevented by immunization with BALB/c cells provided an opportunity to dissect the mechanism of protection by alloimmunization [20]. Immunization could be done as late as day 7.5 of pregnancy, but most of the data was obtained from CBA/J pre-immunized before mating [3]. BALB/c pre-immunized CBA/J mice pregnant by DBA/2 showed an increase in small sized γδT and γδNKT lymphoid cells in decidua after on day 8.5 and persisting at non-resorbed implantation sites on days 12.5–14.5 of pregnancy [3]. These CD4CD8 cells produced immunosuppressive TGF-β2-related molecules [3]. In vivo, monoclonal anti-TGF-β2 antibody boosted resorption rates [3]. To determine why immunization of CBA/J with BALB/c but not DBA/2 prevented abortions, nine (DBA/2xBALB/c) x BALB/c recombinant inbred lines were developed all of which were MHC-H2d. Three lines prevented abortion, 4 had no effect, and 2 increased abortion rates. In pre-immunized mice, these effects correlated with increases or decreased in decidual suppressor cell activity [3]. A mathematical model employing 2 genetic loci, P, for protection against abortions and S for stimulating abortions fit the data and suggested two minor paternal histocompatibility antigens were important, but γδ TcR+ T cells should not react with peptides in the groove of MHC antigens.

As previously mentioned, the spleens of BALB/c immunized CBA/J mice contained cells able to suppress a CBA/J anti H-2 reaction in vitro, but spleen cells from DBA/2 immunized mice were minimally active [20]. Twenty-two years later, Zenclussen et al. showed CD25+ Treg cells in the day spleens of day 14.5 CBA/J females pregnant by BALB/c could prevent abortions in naïve CBA/J females mated to DBA/2 if adoptively transferred within 2.5 days of mating, and these cells were CD4+ Foxp3+ αβTcR+ Treg cells These Tregs had to be given by day 2.5 of gestation, but continued to act until to day 7.5 of pregnancy [3]. On day 8.5, when sensitivity to LPS drops, these Tregs were replaced by CD48 γδT suppressor cells which produced a TGF-β2-related immunosuppressive factor [3]. Low levels of suppression on day 8.5 were associated with increased rates of resorption beginning on day 9.5 of pregnancy [3]. Female CBA/J with high or low rates of resorption in a first pregnancy as determined from litter size manifest a similar high or low rates of loss in a second mating, so the abortions were recurrent and correlated with small lymphoid suppressor cell activity in decidua on days 8.5–10.5 of gestation [26]. At days 12.5–14.5 of pregnancy, these cells at non-resorbed implantation sites were shown to be 89% NK γδT and 11 γδT cells %, with few if any NK-only cells [3]. In 2010, Shima et al. used a different allogeneic mouse mating combination to show Treg cell depletion on day 2.5 of pregnancy caused failure in the peri-implantation period, whereas depletion on day 4.5 and 7.5 of gestation increased the post-implantation abortion rate [3]. Depletion of Tregs on 10.5 and 13.5 days of gestation had no effect [3]. Therefore, different regulatory cells, αβ Tregs and γδ T and NKT function at different time periods during pregnancy. Treg cells once activated by specific antigen can suppress without antigen specificity as does TGF-β [3].

In mice, the critical events leading to resorption first appear on day 9.5 of gestation when the spiral arterioles that will deliver maternal blood to the placenta are plugged by intravascular trophoblasts [3]. In resorption, either those arterioles underwent an inflammation-induced thrombosis, or the plugs were destroyed so that the placenta was flooded with highly oxygenated blood causing free radical damage and destruction of the trophoblasts interposed between fetal capillaries and maternal blood [3]. The latter mechanism has been suggested by Burton's group to occur in human pregnancies that go on to abort, but has not been investigated in the mouse model [3].

Seminal plasma TGF-β plus peptides are both important for activating pregnancy protective Treg cells in the CBAxDBA/2 model

In 2011, using a non-CBA strain of mice native to Australia, Guerin et al. reported mating with an allogeneic male activated Foxp3+ Tregs in the lymph nodes draining the uterus and these cells homed to the decidua 1 day prior to implantation [3]. Differentiation and maintenance of Foxp3+ Treg cells is known to be facilitated by TGF-βs, which has a precursor in seminal plasma, and intravaginal administration of recombinant TGF-β3 on day 0.5 of pregnancy prevented abortions in the CBAxDBA/2 model [3]. Indeed, intravaginal TGF-β3 administered to unmated mice at the time of estrus (the time when they usually mate) promoted local development of Foxp3+Tregs [3]. These Tregs were predominantly CD8+ and CD4+8+ Treg cells which should see peptide antigens associated with Class 1 MHC [3]. What the TcR of those Treg cells can recognize remains unknown, but antigen-specific Treg generation in the uterine draining nodes occurs when paternal antigen(s) should be present together with TGF-β. The mathematical model defining S and P peptides explaining the phenotypes of the backcross recombinant (BALB/cxDBA/2)xBALB/c inbred lines described above suggested one should look in seminal plasma. Preimmunization in the CBAxDBA/2 model using DBA/2 splenocytes did not prevent abortions, but replacing DBA/2 paternal peptides bound in the groove their Class 1 MHC with BALB/c paternal peptides present in BALB/c seminal plasma rendered them protective [27]. Conversely, when BALB/c splenocytes that elicit a protective response had their MHC-bound peptides replaced by the peptides in DBA/2 seminal plasma, the BALB/c splenocytes lost the ability to immunize against abortion [27].

The importance of an Treg response to paternal peptides in human pregnancy is suggested by epidemiological evidence that RM, and subsequent development of growth restricted embryos of surviving embryos with maternal preeclampsia, may be prevented by oral/upper GI tract mucosal immunization with seminal plasma [28]. However, evidence that oral tolerance results and is mediated by Treg cells has not yet been shown, and the tolerance effects of oral antigen administration could be due to Th3 type T cells (which make TGF-β, as distinct from Th1 T cells which make proinflammatory cytokines and/or can by cytolytic and Th2 cells which make IL-10 and cytokines favoring antibody production and Th17 cells (which make proinflammatory IL-17).

There was more to an anti-abortive immune response than administering enough of the right peptide antigens

Transplantation immunologists had discovered that portal vein administration of allogeneic cells was very effective in enabling mice to tolerate renal allografts, and this was mediated by γδT suppressor cells and TGF-β [3,29]. The similarity to successful pregnancy in mice was astonishing. The key molecule required in the liver for induction of γδT suppressor cells was CD200, an immune check-point inhibitor [3,29]. CD200 was also expressed in decidua and placenta in successful CBAxDBA/2F1 embryos that did not abort, and administration of a soluble CD200Fc was able to reduce the abortion rate to 2.1% even though the CBA/J female had been exposed to putatively inadequate DBA/2 seminal plasma peptides at the time of mating [29]. In fact, administration of soluble CD200Fc, where Fc made the CD200 molecule soluble, could stop abortions even if given in a single dose on day 9.5 of pregnancy. The effect was probably direct since there was insufficient time for an immune response to occur. Anti-asialoGM1 antibody that ablates blood-type NK cells also arrested the abortion process promptly, even if given as late as day 10.5 of pregnancy in the CBAxDBA/2 model [3]. These were important discoveries, because efficacy of prevention of abortion in RM women given Mowbray-type LIT depended on the dose of CD200+ cells, not just cells with paternal antigens, and a repeat abortion could be prevented if LIT was given after pregnancy had occurred but before it had reached 6 weeks gestation [3]. Importantly, a large expensive DB-RCT that had shown no effect of paternal leukocyte immunotherapy on the live birth rate, and the paternal blood mononuclear cells had been stored at 4°C overnight, a condition which results in loss of surface CD200 [3,30]. The stored cells could still induce anti-paternal antibodies, but anti-paternal HLA antibodies per se have no effect on the underlying pathogenesis of pregnancy loss as already mentioned. Another LIT trial using cells stored at 37°C, a condition which does not lead to loss of cell-surface CD200, reported a high pregnancy success rate [3,30]. These same results were replicated in the CBAxDBA/2 mouse model by storing BALB/c splenocytes at 4°C overnight before immunization [3]. Treatment using stored refrigerated cells failed, and the cells lost CD200 into the supernatant (which interestingly had a small anti-abortive effect) [3,30]. Other ‘problems’ with the large expensive human DB-RCT of refrigerated cells had been identified [30]. A subsequent meta-analysis of RCTs of LIT claiming to disprove the efficacy of LIT combined abortion rates in patient who became pregnant with intention-to-treat live birth rates which diluted positive data [3,30]. As well, these authors ignored significant heterogeneity contraindicating performance of a meta-analysis and other flaws in the study [3,30,31]. Numerous other studies done since the RMITG publication confirm efficacy of LIT and have also support Mowbray's conclusion about how best to treat and whom to treat [16,30–34].

Additional mechanistic insights into anti-abortion mechanisms discovered using mice

One of the more important contributions of the CBAxDBA/2 model was addressing the suggestion that LIT was merely a placebo that reduced stress as a basis for preventing RM [14,15]. CBA/JxDBA/2 matings were done and the female was subjected to restraint stress. Restraint stress increased the resorption rate, and preimmunization with BALB/c cells abrogated the stress-induced loss [35]. Stress increases intestinal permeability allowing endotoxin absorption, and endotoxin, acting on TLR4 triggers production of abortogenic cytokines [3]. The optimal day to stress a mated CBA/J female was day 4.5 or 5.5, the days when embryos implant and begin nidation on the antimesometrial side of the uterus, 2 days before ectoplacental cone trophoblast begins to implant on the mesometrial side of the uterus where the placenta will be formed [3,22,35]. Preimmunization of A/J female before mating and stress did not lead to resorptions but rather to occult losses which were particularly striking with DBA/2 used for preimmunization then mating with DBA/2 [35]. These data were compatible with stress-reduction in CD8+ Treg cell protection [36].

As already mentioned, antibiotic treatment of CBA/J females before mating reduced the endogenous resorption rate and stress can increase LPS absorption from the intestinal tract [3]. Indeed, differences in abortion rates in CBAxDBA/2 matings where fecal LPS levels are the similar may be explained by differences in stress levels [3]. Fortunately, in the study of the effect of alloimmunization on stress triggered abortion rates, C3H/HeJ females, which lack TLR4 and are endotoxin resistant, were also mated to DBA/2 males. Stress increased their resorption rate, and preimmunization with DBA/2 cells abrogated the increase in resorptions [35]. Subsequently Arck et al. showed that stress triggered intrauterine release of substance P which the boosted TNF-α levels, depleting CD8+ T cells in vivo augmented TNF-α-dependent loss, and Markert et al. showed stress caused activation and degranulation of mast cells in the endometrium at the time of implantation [3,36,37]. Therefore there could be an LPS-independent pathway able to explain stress-induced abortions in LPS-resistant mice. In human RM patients, endometrial mast cells also show evidence of activation at the time in the secretory phase of their menstrual cycle that the embryo will implant (comparable to day 4.5 in the mouse) [38]. Therefore, humans with RM show abnormalities similar to those implicated in the mouse model of stress-induced pregnancy loss.

Some strains of mice did not have abortions after exposure to stress [35]. A/J females mated to DBA/2 or C3H/HeJ males and stressed manifest only occult losses, and these were made worse by alloimmunization! [35]. In contrast, C56Bl/6xDBA/2 mated mice proved resistant to stress [35] A recent systematic review and meta-analysis confirms certain types of stress may be associated with RM [39]. However, it is not yet known if this cohort is more likely to benefit from administration of allogeneic paternal cells, or which couples correspond to the different strain-specific responses observed in mice.

A further advance arising from the CBAxDBA/2 model concerns the intravenous nutritional supplement Intralipid. An unsuccessful trial of human syncytiotrophoblast membrane vesicles employed Intralipid as a placebo, and those RM women receiving Intralipid appeared more successful that those receiving trophoblast membrane vesicles. Intralipid was then demonstrated in the CBAxDBA/2 model [40]. Subsequently, it was shown that Intralipid infusion could suppress human peripheral blood NK cells. Two recent DB-RCTs have shown that Intralipid can improve the live birth rate in RM couples with elevated blood NK cells and in primary infertility patients with a previous implantation failure [41,42].

The importance of environment as a determinant of the outcome of mating

Figure 4 depicts the original restricted within-the-box focus of immunologists in understanding materno-fetal and feto-maternal immune interactions [3]. There is more than one materno-fetal interface. During the peri-implantation phase, paternal minor histocompatibility antigen peptides can enter the primary decidua from the implanted blastocyst (shown in Figure 2 as ①) and when taken up by antigen-presenting cells, activate an embryo-destructive T cell-dependent inflammatory process [3,25]. In the post-implantation phase, extravillous trophoblast (EVT) and chorionic trophoblast is in direct contact with decidua and a potential confrontation with maternal immune system cells (Figure 2, location ②). EVT also invade and replace endothelium of maternal spiral arteries that provide blood to the placenta and form endovascular plugs which prevent too much blood flow that could expose the syncytiotrophoblast to oxidative damage (Figure 2, location ③). Additionally, transplacental traffic of fetal lymphoid cells capable to reacting to and harming maternal tissue and maternal lymphoid cells enter the embryo, are present in liver and bone marrow, and could potentially harm the developing fetus [3]. The fetus is exposed to non-inherited maternal MHC antigens (NIMA), and these appear to create some degree of systemic tolerance which persists after birth and into adulthood [43]. Fetal cells entering the pregnant mother have been implicated in remission of autoimmune diseases such as rheumatoid arthritis, may persist for many years, and in certain types of women may lead to scleroderma which has been likened to a graft-versus-host condition [44]. Obviously both the mother and fetus must have mechanisms to mitigate harm. There are, however, external factors which impact on the immunology of the materno-fetal relationship [3,4,39].

Traditional conceptualization of materno-fetal immune interactions during pregnancy.

Figure 4.
Traditional conceptualization of materno-fetal immune interactions during pregnancy.

The traditional paradigm holds that outcome of a non-syngeneic pregnancy is determined by maternal innate and adaptive immune systems. Mother recognizes antigens on her semiallogeneic intrauterine conceptus, both MHC and non-MHC minor transplantation antigens from the fetus or its trophoblast. She may then react. As LIT with allogeneic paternal cells produces larger babies, both in mice and humans, mother's immune response may be beneficial [3]. Seminal plasma antigen accompanying the gametes (a spermatozoa, which also expresses antigens is shown) may trigger pre-implantation generation of Treg cells which prevent harm to the implanted blastocyst up until the time a distinct placenta and fetus have formed and onset of placental perfusion by maternal blood (day 9.5 in the mouse, 6 weeks gestation in humans which is 2 weeks after the missed menstrual period). Events prior to day 9.5 set the stage for subsequent spontaneous abortion, and where the fetoplacental unit survives, impairment of a plentiful flow of maternal blood predisposes to development of pre-eclampsia and/or fetal growth restriction. Note occurrence of bidirectional cellular traffic between fetus and mother. Syngeneic matings, such as repeated in brother-sister matings in mice after successive generations leads to inbreeding depression which is attributed to deleterious recessive genes.

Figure 4.
Traditional conceptualization of materno-fetal immune interactions during pregnancy.

The traditional paradigm holds that outcome of a non-syngeneic pregnancy is determined by maternal innate and adaptive immune systems. Mother recognizes antigens on her semiallogeneic intrauterine conceptus, both MHC and non-MHC minor transplantation antigens from the fetus or its trophoblast. She may then react. As LIT with allogeneic paternal cells produces larger babies, both in mice and humans, mother's immune response may be beneficial [3]. Seminal plasma antigen accompanying the gametes (a spermatozoa, which also expresses antigens is shown) may trigger pre-implantation generation of Treg cells which prevent harm to the implanted blastocyst up until the time a distinct placenta and fetus have formed and onset of placental perfusion by maternal blood (day 9.5 in the mouse, 6 weeks gestation in humans which is 2 weeks after the missed menstrual period). Events prior to day 9.5 set the stage for subsequent spontaneous abortion, and where the fetoplacental unit survives, impairment of a plentiful flow of maternal blood predisposes to development of pre-eclampsia and/or fetal growth restriction. Note occurrence of bidirectional cellular traffic between fetus and mother. Syngeneic matings, such as repeated in brother-sister matings in mice after successive generations leads to inbreeding depression which is attributed to deleterious recessive genes.

Figure 5 looks ‘outside the box' to the role of the environment [3]. Treg cells are implicated in rendering the antigenic feto-placental implant shown in Figure 4 ‘not dangerous' [25,45,46]. Studies of mice with CNS1 gene knock-out, which promotes Foxp3 Treg cells, show that Tregs also play a an important role in dampening effect of infectious not-self danger signals from the normal flora in the environment [25]. That is possible because Treg cells activated by a specific antigen can suppress cells reacting to unrelated antigens or even non-specific danger signals acting via TLRs [3]. Up-regulation of CD200 expression on embryonic trophoblast, which in the mouse appears to become increasingly relevant on day 7.5, 8.5 and 9.5 in mouse pregnancy, also prevents LPS-driven rejection, so trophoblast expression of CD200 can serve as a mechanism signaling which embryos ‘deserve’ to be saved from termination [47]. The more antigenic embryos generate a better Treg response and are the last to be annihilated if the mother has a life threatening infection. CD200 on fetal trophoblast can enhance the protective Treg response, but a sufficiently severe TLR stimulus in early pregnancy will cause all intrauterine embryos to resorb or abort; in some strains of mice, the process in more efficient and occult loss occurs.

The new paradigm of feto-maternal interactions based on danger.

Figure 5.
The new paradigm of feto-maternal interactions based on danger.

The new paradigm is based on assertion attributed to Poly Matzinger that the semiallogeneic fetoplacental unit is not rejected because it is not dangerous [45,46]. However, circumstances can make it so. Within the box, danger results if the antigenic difference between the male and female is too great, the equivalent of a xenogeneic mating. In ungulates and in nature, interspecies matings normally defined by a reproductive barrier, can succeed, but success rate is low [3,45]. Thinking outside the box, pregnancy is a physiological burden on the mother and a societal burden impacting care for her children and male mate. A mild systemic impairment in immunity increases her susceptibility to infectious agents, especially viral pneumonia and mycobacteria such as tuberculosis. Her ability to engage in fight or flight when predators appear is compromised. Thus, a mechanism is needed to terminate reproductive events. Additionally, mother's resources should not be squandered on producing babies that are unlikely to thrive and mature to reproduce and care for their offspring. Potts has shown stress favors birth of H-2 heterozygous embryos, and stress and TLR-mediated danger signaling (e.g. bacterial LPS, viral nucleic acids) trigger mechanisms that eliminate the less heterozygous embryos in utero [3]. This prenatal natural selection, or intrauterine Darwinism, is consistent with abortion and occult loss being natural physiological events. Maternal stress during pregnancy can make offspring more alert to danger, and may program the fetus for post-natal development of asthma. Mother can also be subject to danger by fetal cells persisting within her body and leading to a low grade graft-versus-host reaction manifest as scleroderma. If she is excessively immunosuppressed by conceptus antigen-activated suppressor cells, sudden withdrawal at parturition can trigger post-partum autoimmune disorders such as thyroiditis or rheumatoid arthritis, and fetal cells persisting in mother may eventually lead to scleroderma. (Adapted from Clark, D.A., Chaouat, G., Gorczynski, R.M. (2002) Thinking outside the box: mechanisms of environmental selective pressure on the outcome of the materno-fetal relationship. Am. J. Reprod. Immunol.47, 275–282, with permission of the publisher.)

Figure 5.
The new paradigm of feto-maternal interactions based on danger.

The new paradigm is based on assertion attributed to Poly Matzinger that the semiallogeneic fetoplacental unit is not rejected because it is not dangerous [45,46]. However, circumstances can make it so. Within the box, danger results if the antigenic difference between the male and female is too great, the equivalent of a xenogeneic mating. In ungulates and in nature, interspecies matings normally defined by a reproductive barrier, can succeed, but success rate is low [3,45]. Thinking outside the box, pregnancy is a physiological burden on the mother and a societal burden impacting care for her children and male mate. A mild systemic impairment in immunity increases her susceptibility to infectious agents, especially viral pneumonia and mycobacteria such as tuberculosis. Her ability to engage in fight or flight when predators appear is compromised. Thus, a mechanism is needed to terminate reproductive events. Additionally, mother's resources should not be squandered on producing babies that are unlikely to thrive and mature to reproduce and care for their offspring. Potts has shown stress favors birth of H-2 heterozygous embryos, and stress and TLR-mediated danger signaling (e.g. bacterial LPS, viral nucleic acids) trigger mechanisms that eliminate the less heterozygous embryos in utero [3]. This prenatal natural selection, or intrauterine Darwinism, is consistent with abortion and occult loss being natural physiological events. Maternal stress during pregnancy can make offspring more alert to danger, and may program the fetus for post-natal development of asthma. Mother can also be subject to danger by fetal cells persisting within her body and leading to a low grade graft-versus-host reaction manifest as scleroderma. If she is excessively immunosuppressed by conceptus antigen-activated suppressor cells, sudden withdrawal at parturition can trigger post-partum autoimmune disorders such as thyroiditis or rheumatoid arthritis, and fetal cells persisting in mother may eventually lead to scleroderma. (Adapted from Clark, D.A., Chaouat, G., Gorczynski, R.M. (2002) Thinking outside the box: mechanisms of environmental selective pressure on the outcome of the materno-fetal relationship. Am. J. Reprod. Immunol.47, 275–282, with permission of the publisher.)

TLR3 binds ligands such as poly I:C, thought to mimic viral nucleic acids, and causes pregnancy failure in all mouse strains including C3H/HeJ. Poly I:C given on day 8.5 of pregnancy to BALB/cxBALB/c (H2dxH2d) or BALB/cxC57Bl/6 (H2dxH2b) matings caused abortions. Here, poly I:C directly bound to proliferating cytokeratin 7+ trophoblast cells in vitro and reduced the percentage that were CD200+. In vivo, poly I:C treatment given on day 8.5 of pregnancy dramatically reduced their number of CK7+ CD200+ trophoblasts invading the decidua of day 13.5 implantation sites [48]. In poly I:C-induced fetal loss studied in female C57Bl/10 IL-10−/− at the time a distinct fetus and placenta had formed, decidua-invading trophoblast cells were undergoing apoptosis, and that process was dependent on activated NK cells [49]. Anti-IL-10 increases the resorption rate in CBAxDBA/2 matings which have an inadequate level of Tregs, but not in CBAxBALB/c matings or in allogeneically mated C57Bl/6 females, so suppression of activation of the NK-macrophage inflammatory process by IL-10 only emerges as important when the level of danger is augmented, or Treg activity is dimished.

IL-10 is an anti-inflammatory T cell cytokine, distinct from the Th1 cytokines TNF-α and interferon-γ, and all these cytokines may also be produced by non-T cells. Subpopulations of Treg cells, γδ T cells, or uNK (with receptors for Class I MHC) make IL-10 develop in decidua by gestation day 7.5 when paternal MHC is first made by the mouse embryo and persist beyond day 12.5–14.5 when bidirectional fetal-maternal cellular exchanges begins, remain until term [3,22,44]. Similar types of regulatory cells may be found in human decidua [3,22,50]. Indeed, CD56bright uterine NK cells isolated from first trimester human termination decidua make IL-10 and a TGF-β2-type immunosuppressive factor [3,50]. IL-10-producing cells have been implicated in prevention of LPS and CPG-induced abortions as well as LPS and CPG-induced preterm birth [49–53]. IL-10−/− mice are 10-fold more susceptible to LPS-induced preterm birth that intact mice [52]. Th1 cytokine-driven abortions resulting from CPG engaging TLR9 may also involve inactivation of Treg cells and that would make an IL-10 immunoregulatory backup by Th2 and uNK cells quite important [53]. Alloimmunization in the CBAxDBA/2 model, which generates anti-abortion Treg cells, abrogates LPS-induced abortions mediated by Th1-cytokine producing NK cells but not preterm delivery [52,54]. NK cells have also been implicated in causing preterm birth, but prevention of preterm birth requires a different mechanism from that provided by alloimmunization-induced Tregs,- specifically maternal lymphoid cells that produce IL-10 [50,52].

There are other environmental threats to successful pregnancy. Mother must survive and thrive in order to care for other offspring and her male mate, so if her life or health is threatened pregnancy must be avoided or terminated. Stressors include, but are not limited to war, climate, food shortage, socioeconomic deprivation, predatory animals, and pandemics. These can occur prior to conception or after a pregnancy has begun, and if early on, can affect which embryos survive to term [55]. Potts noted a deficiency of H-2 homozygous offspring born by females who migrated from their usual territory into a less familiar and potentially more dangerous foreign one, and according to Calhoun, male and female mice must emigrate to escape a fecundity-suppressing high density social environment [3,5]. A heterozygous MHC may provide a survival advantage when there are new infectious agents about, and in outbred strains of mice, ‘hybrid vigor' is associated with larger litter sizes than occur when inbred stains of mice gestate semi-allogeneic embryos.

Another manifestation of an intrauterine selective process is change in the male:female sex ratio at birth in a population,- apparently an object of interest since 1662 when an excess of male births was noted, and that also occurs amongst inbred mice [55–61]. Human males are considered ‘more fragile’ than females as reflected in post-natal survival and lifespan, so in times of prolonged war, or in stressful occupations, or where the woman is experiencing more stress as reflected in salivary cortisol, more females are born. Population growth is optimal when the number of males and females are equal, and too may or too few males can be problematic. Could immunological mechanisms determine sex ratio at birth?

In mice and in humans, the male H-Y antigen is expressed by spermatozoa and embryos including trophoblasts. Kahn and Baltimore have shown that H-Y-specific Treg cells are generated and ablation of those Tregs in vivo results in fewer male pups being born [3]. In recurrent miscarriage women, there are fewer Treg cells specific for sperm antigens [62]. Primary RM couples have no live births, but couples can have a live birth followed by a series of spontaneous abortions, and this is called secondary recurrent miscarriage [63]. Women with serum anti-HLA antibodies had an increased miscarriage risk, Serum anti-H-Y antibodies were slightly more common if the first live birth had been male, and the presence of anti-H-Y antibodies predicted a reduced chance of a live birth with a male:female sex ratio of 0.14 suggesting maternal rejection of H-Y+ embryos [63]. Curiously, serum anti-HLA antibodies were more likely and those women also had an increased risk of miscarriage [63]. IVIg treatment appeared to increase the live birth rate, but since 1/3 of the immunosuppressive activity of IVIg is due to bound CD200 IVIg, one doesn't know if the benefit results from interference with anti-H-Y antibodies or suppression of anti-H-Y T cell activity [3,64]. (Unfortunately, rodents such as mice have multiple intrauterine embryos and to study the effect of fetal sex on the maternal immune system would require all of the embryos be either male or female. That is not technically possible at present.) Human IVF data indicates the male:female ratio at conception is 1 : 1 and more preimplantation male embryos have karyotype abnormalities [56]. From Figure 2, many of these will fail in the peri-implantation period. However, more spontaneous miscarriages are female. Miscarriages are prevented by Tregs acting from shortly after conception until the peri-implantation phase is completed, although Tregs may still be found in decidua of first trimester human miscarriages [65]. Female embryos could be aborted because they are not so well protected. However, stress can inhibit peri-implantation CD8+ Tregs (in the mouse), so H-Y-specific Tregs should be ablated thereby enabling a maternal anti-H-Y T cell response to become ‘hostile’. The predicted result would be occult loss of male embryos followed by further loss by miscarriages [36]. There are minor antigenic paternal peptide in addition to H-Y of which mother is aware, and she is also aware of MHC antigen differences that together enable preferential intra-uterine survival [3]. If a female is going to invest in producing offspring, those offspring must be able to survive to adulthood and reproduce.

Objections to the use of mouse data for diagnosis and treatment of pregnancy failure

Table 2 summarizes relevant similarities of the mouse to humans and some important differences. A competing model recently proposed to explain human RM is disrupted endometrial selection [71]. Brosens’ group suggested human decidual stromal embryo can detect and eliminate poor quality embryos, such as those with karyotype abnormalities, at the time of implantation by reducing production of pro-implantation cytokines and widening the implantation window [71–74]. It was suggested the decidua of RM women had a sensing defect allowing poor quality normal karyotype embryos to implant.

Table 2
Relevant homology issues of CBAxDBA/2 model for human RM
Observation in CBAxDBA/2 modelObservation in human pregnancy success & failure
1. Pathogenesis of failure is inflammation at feto-maternal interface:
(a) Activated coagulation-FGL2 prothrombinase
(b) activated complement
(c) Driven by TNF-α + INF-γ
(d) Necrosis due to Th17 recruited neutrophils
(e) Insufficient control by IL-10 and regulatory T cells (both αβ and γδ Tregs + B cells) 
1. (a) Increased FGL2 in karyotype normal abortus tissue.
(b) Complement activation by MBL-A & procoagulants [25]
(c) PBL T cells in RM over-produce TNF-α and under-produce IL10 [3,66]
(d) Human missed abortions (MA) lack Th17 cells that recruit neutrophils [67]
(e) Low blood αβT regulatory cells in RM [65
2. Allogeneic immunization with lymphoid cells can prevent resorptions (abortion): antigen-specific & CD200 related mechanisms. 2. (a) Positive data in several meta-analases. LIT increases live birth rate [14,30,32–34]
(b) CD200 cell dose correlates with livebirth in primary RM cases without autoimmunity [3]. 
3. Intralipid is effective and suppresses innate immune system (both NK cells & macrophages are altered) [403. Intralipid improves livebirth rate in IVF of primary infertility & RM with elevated blood NK cells [41,42
4. Seminal plasma antigens & TGF-β may activate protective Treg cells [3,27(a) Seminal plasma improves IVF success [68]
(b) Seminal plasma in upper GI tract may prevent RM & PE [26
5. Stress and other danger signals activate abortogenic mechanisms to protect mother. 5. Stress associated with risk of RM [39
6. CD200 tolerance-signaling molecule mRNA is expressed in ectoplacental trophoblast and decidua at day 8.5 and prevents abortion process on day 9.5 [29] transgenic ↑ CD200 in trophoblast can prevent LPS-induced abortions in B6 mice [476. Mouse day 8.5 ≍ human 5.4 week gestation. Villus trophoblast CD200 (and in decidua) linked to missed abortion (MA) via prevention Th17 cell infiltrate, CD200 normal in trisomy 18 MA [69]. Villus CD200 reduced in non-trisomic MA as early as 5 weeks hestation [70
Observation in CBAxDBA/2 modelObservation in human pregnancy success & failure
1. Pathogenesis of failure is inflammation at feto-maternal interface:
(a) Activated coagulation-FGL2 prothrombinase
(b) activated complement
(c) Driven by TNF-α + INF-γ
(d) Necrosis due to Th17 recruited neutrophils
(e) Insufficient control by IL-10 and regulatory T cells (both αβ and γδ Tregs + B cells) 
1. (a) Increased FGL2 in karyotype normal abortus tissue.
(b) Complement activation by MBL-A & procoagulants [25]
(c) PBL T cells in RM over-produce TNF-α and under-produce IL10 [3,66]
(d) Human missed abortions (MA) lack Th17 cells that recruit neutrophils [67]
(e) Low blood αβT regulatory cells in RM [65
2. Allogeneic immunization with lymphoid cells can prevent resorptions (abortion): antigen-specific & CD200 related mechanisms. 2. (a) Positive data in several meta-analases. LIT increases live birth rate [14,30,32–34]
(b) CD200 cell dose correlates with livebirth in primary RM cases without autoimmunity [3]. 
3. Intralipid is effective and suppresses innate immune system (both NK cells & macrophages are altered) [403. Intralipid improves livebirth rate in IVF of primary infertility & RM with elevated blood NK cells [41,42
4. Seminal plasma antigens & TGF-β may activate protective Treg cells [3,27(a) Seminal plasma improves IVF success [68]
(b) Seminal plasma in upper GI tract may prevent RM & PE [26
5. Stress and other danger signals activate abortogenic mechanisms to protect mother. 5. Stress associated with risk of RM [39
6. CD200 tolerance-signaling molecule mRNA is expressed in ectoplacental trophoblast and decidua at day 8.5 and prevents abortion process on day 9.5 [29] transgenic ↑ CD200 in trophoblast can prevent LPS-induced abortions in B6 mice [476. Mouse day 8.5 ≍ human 5.4 week gestation. Villus trophoblast CD200 (and in decidua) linked to missed abortion (MA) via prevention Th17 cell infiltrate, CD200 normal in trisomy 18 MA [69]. Villus CD200 reduced in non-trisomic MA as early as 5 weeks hestation [70

As summarized in Table 3, there are many problems with this hypothesis. There is extensive data associating reduced levels of decidual and peripheral blood Treg levels in human RM, and also in mouse models, and the Brosens model does not account for this [65]. A shorter time-to-pregnancy interval (TTP) in RM cases was used to support the argument that more poor quality embryos were surviving elimination to become clinical pregnancies that aborted [74]. That could have been verified by a simple established methods to detect implanted embryos before the diagnosis of clinical pregnancy [75]. A mathematical model was used to determine expected times for achieving three consecutive pregnancies in the general population and by 3 months, 41% of RM cases achieved a pregnancy compared with an expected value of 8%. The authors admit the majority of RM couples did not report short TTP and many factors impact chance of a pregnancy. The abortion rate in those RM cases with a short TTP was not compared with the rate in those taking longer to become pregnant (and the % normal karyotype losses as a function of TTP). In the Mowbray trial of paternal leukocyte immunotherapy, 23 immunized RM and 23 control RM cases achieved pregnancy (67% of 69 in the study) and as already mentioned, the immunized RM patients had a significantly increased live birth rate which would not have been predicted if those embryos were euploid but ‘poor quality' [3,14]. RM couples with a short TTP might be a subpopulation who are more likely to have more normal karyotype blastocysts which would survive in a supportive uterine environment (as has been demonstrated in mice) [80]. Finally, the chance of a sperm-oocyte interaction might be increased in RM couples due to a lower frequency of oral sex, as described by Meuleman et al., where the probability sperm-oocyte interactions would not be reduced [28].

Table 3
Defective decidual stromal cell sensing of ‘poor quality' embryos as a basis for recurrent miscarraiges
Observations and hypothesis (HΦ)Problems with the hypothesis
1. In normal fertile women, secretory phase decidualizing endometrial stromal cells (DESC) migrate towards healthy embryos and support their growth, but move away from ‘poor quality' embryos and shut down. RM patient cells fail to recognize and retreat from poor quality embryos, and thus, favor their successful implantation and subsequent miscarriage between 6–12 weeks gestation [71–741. (a) ‘Poor quality' (PQ) embryos were either blastocysts with 3 pronuclei or arrested at the morula stage. No euploid PQs [73]
(b) HΦ predicts decreased frequency of aneuploid occult losses and shortening time to pregnancy (TTP). Key data is missing [72–75]
(c) HΦ predicts ↑ ratio of abnormal/euploid abortuses. Figure 2 shows if no livebirths, abortion rate will be 7%, and occult euploid embryo loss rate must increase! but how?
(d) HΦ predicts RMs not a partner-specific problem. Not so! but could it be the man? Who knows?
(e) Reduced RM must be due to T cell ‘rejection' of PQ embryos in occult loss time window (ok for item (c)), but lit (and intralipid) increase number of ‘normal' live births [14,15,41,42]. 
2. Defective DESC development due to senescence during implantation window in RM leads to fewer uNK cells [742. (a) In RM, eGL (uNK) are proposed to be effector cells, activated NK cells can cause trophoblast apoptosis and impair the spiral arterial supply to placenta. but suppressing elevated uNK cells in RM can lead to a livebirth [76,77]
(b) Blood NK and T cell changes that relate to RM unexplained. [78,79]
(c) How can senescent DESCs migrate to and support PQ embryo survival? [74
Observations and hypothesis (HΦ)Problems with the hypothesis
1. In normal fertile women, secretory phase decidualizing endometrial stromal cells (DESC) migrate towards healthy embryos and support their growth, but move away from ‘poor quality' embryos and shut down. RM patient cells fail to recognize and retreat from poor quality embryos, and thus, favor their successful implantation and subsequent miscarriage between 6–12 weeks gestation [71–741. (a) ‘Poor quality' (PQ) embryos were either blastocysts with 3 pronuclei or arrested at the morula stage. No euploid PQs [73]
(b) HΦ predicts decreased frequency of aneuploid occult losses and shortening time to pregnancy (TTP). Key data is missing [72–75]
(c) HΦ predicts ↑ ratio of abnormal/euploid abortuses. Figure 2 shows if no livebirths, abortion rate will be 7%, and occult euploid embryo loss rate must increase! but how?
(d) HΦ predicts RMs not a partner-specific problem. Not so! but could it be the man? Who knows?
(e) Reduced RM must be due to T cell ‘rejection' of PQ embryos in occult loss time window (ok for item (c)), but lit (and intralipid) increase number of ‘normal' live births [14,15,41,42]. 
2. Defective DESC development due to senescence during implantation window in RM leads to fewer uNK cells [742. (a) In RM, eGL (uNK) are proposed to be effector cells, activated NK cells can cause trophoblast apoptosis and impair the spiral arterial supply to placenta. but suppressing elevated uNK cells in RM can lead to a livebirth [76,77]
(b) Blood NK and T cell changes that relate to RM unexplained. [78,79]
(c) How can senescent DESCs migrate to and support PQ embryo survival? [74

Non human primates have problems with karyotype-abnormal embryos [81]. However, there artificial methods where chromosomally abnormal mouse embryos can be created and where decidual stromal cell elimination of abnormal karyotype embryos might occur [82]. Using such models one may be able to determine if a decidual cell surveillance mechanism exists and if so, is immunologically modifiable. The Brosens’ model provides no means to identify poor quality normal karyotype embryos which are proposed to lead to pregnancies that will abort. However, 1/3 of CBAxDBA/2 mouse decidua on day 8.5 or pregnancy lacked mRNA for CD200 (OX-2), and possibly these decidual cells are akin to those the Brosens’ model is describing? [29].

Ethical and technical issues

The use of animals in research is currently a matter of significant scrutiny, particularly among animal rights advocates. In the U.K., the National Centre for the Replacement Refinement and Reduction in Animals in Research (NCRRRAR) states opinion polls of public support for animal research depends upon putting into practice the 3 R's, (1) Replacement, (2) Reduction, and (3) Refinement [83].

Replacement

It is well known that in vitro phenomena may prove to not occur in vivo. Neither the phenomenon of cancer nor the phenomenon of pregnancy can be currently reproduced in a culture dish. Furthermore, many types of investigations cannot be done in humans as explained earlier in this article. As illustrated in Figure 1, interventional studies in animals provide the preclinical data required for approval of a human study. Single cell immunological technology has been recently applied to obtain important observational data on human pregnancy without the use of animals. This is satisfying to those who believe human pregnancy is unique (as is man) and animal analog models are not relevant [22]. A US-based 2017 study of 28 women undergoing normal pregnancy obtained peripheral blood samples in each trimester and a post-partum control [84]. A 41 one parameter mass cytometry whole blood assay of endogenous activity of white cell subsets and response to LPS or interferon-α2 or a mixture of IL-2 + IL6 showed a common and consistent gestational age-specific pattern enabling prediction of gestational age from the data. Over the course of normal pregnancy, neutrophils expanded and sensitized to multiple stimuli consistent with increased STAT1-mediated activation of innate immune system cells. CD8+ T cells were also activated but in early pregnancy, suppression of the dendritic (antigen-presenting) cell response to activation by endotoxin (TLR4) was dampened. A second 2018 UK-based study used single cell droplet RNA sequencing (scRNAseq) of cells isolated from 11 first trimester elective termination patients’ villous placenta and adjacent decidua [85]. An encyclopedia of 12 immune and 12 non-immune populations were identified. Functional interactions were inferred from signal-ligand pairs. The major focus were 3 populations decidual NK cells (dNK) none of which were making angiogenic factors. dNK1 were more granulated, interacted with extravillous cytotrophoblast (EVT) HLA-G, expressed CD39 which, with CD73, could degrade extracellular ATP. dNK1 were less frequent in a first pregnancy. dNK2 interacted with dNK3 cells which were high levels of anti-inflammatory protein mRNA, and appeared to recruit type 1 dendritic cells (DC1) and control invasiveness of EVT. DC1 appeared to recruit CD8+ T cells but kept them quiescent via T cell PD1. Curiously, no data on CD200 and its receptors was obtained notwithstanding its physiological importance in preventing pregnancy loss in mice and in humans, but possibly due to the difficulty of detecting isoforms using scRNAseq [29,45,67,69,70,86]. Both studies were labor intensive (requiring 26 and 24 co-authors), were essentially descriptive anatomy, excluded patients with pregnancy complications, and failed to obtain data on the first 6 weeks of gestation which in the time period when pregnancy problems such as miscarriage and preeclampsia have their genesis. Both studies were enthusiastic about the physiological importance of their data. Inferences of pathophysiological function based on form is a hypothesis, and definitive proof of mechanistic importance will require data on abnormal pregnancies and in vivo manipulations. Lucas et al. carried out a similar droplet scRNAseq study comparing 90 normal controls to 89 recurrent first trimester miscarriage patients [74]. In the pregnancy loss group, the median uNK frequency per 100 stromal cells measured by immunohistochemical staining was lower [74]. Unfortunately, the authors failed to provide details about the patients and strategy for obtaining decidua unaffected by the miscarriage process, and did not provide segregated data for each of the 3 uNK subsets. In vivo manipulations to dissect the functional significance of NK cell subsets will likely require use of an animal model (Figure 1) such as the pregnant mouse which has at least 2 distinct uNK subsets [22].

Reduction

This item concerns the number of animals used, experimental designs which result in robust and reproducible studies, and results that contribute to knowledge. That requires a sufficient number of animals, such as mice, per group to achieve statistical significance. The number of animals per study group is usually based on an effect size achieving a Pα ≤ 0.05, and a 1 tail test (that is justified if the direction of the difference is known in advance). The statistical power (Pβ) of a 1 tail test is only 0.5, so 50% of the time, an attempt to reproduce the result in an experiment of the same size will not achieve significance even if the original result was true. Therefore a larger sample size or meta-analysis of several experiments will be necessary to validate or invalidate the original conclusion. In mouse pregnancy studies, pooling the data from >1 experiment is commonly used to obtain achieve robustness. One aspect of robustness, which seems to be ignored, is acknowledging and explaining an opposite outcome with an experimental intervention. For example, NK cells were proposed to prevent abortions by regulating IL-17+ pro-abortive T cells in humans and in mice, and in the CBAxDBA/2 model, tail vein injections of anti-asialoGM1 antibody on days E0.5, E4.5, and E8.5 decreased the number of live fetuses per uterus! [87]. But in the CBAxDBA/2 model, abortions are triggered by NK cells and anti-asialoGM1 antibody treatment reproducibly decreases abortion rates in the widely used CBAxDBA/2 model, albeit not to the lowest possible rate [3,25,29,88]. The anti-abortive effect of anti-asialoGM1 antibody could, however, have been overcome if for some reason endotoxin had been present in the diluted antibody preparation. Study of the role of neutrophils in abortion employed a specially prepared lot of low endotoxin anti-GR1 (RB6-8C5) antibody dramatically reduced abortion rates in the CBAxDBA/2 model, although again, not to the lowest possible rate [24]. Nevertheless, Ren et al. reported RB6-8C5 treatment of CBAxBALB/c and BALBcxB6 pregnancies dramatically increased abortion rates, and slightly increased the abortion rate in CBAxDBA/2 matings! [89]. Anti-GR1 also depletes endotoxin-induced myeloid-derived suppressor cells (CD11b+ GR1+) and therefore the more neutrophil-specific anti-Ly-6G was tested [89–91]. In the CBAxDBA/2 model, anti-Ly-6G also increased the abortion rate, but Gelber et al. reported both anti-GR1 and anti-Ly-6G decreased the abortion rate (albeit using a syngeneic pregnancy, BPH5xBPH5, which as B6xBPH/2 subline) [90]. Pan et al. also reported increased abortion rates with anti-Ly-6G in allogeneic B6xBALB/c pregnancies 5% → 45%) and in syngeneic B6xB6 pregnancies (5% → 33%) [92]. Activated neutrophils are associated with pregnancy loss in humans [3]. A final example is an increased abortion rate in allogeneic but not syngeneic pregnancies in B6 mice injected with anti-PDL-1 antibody, but not anti-PDL-2 antibody [93]. Here PDL-1 binds to PD-1 and suppresses innate and T cell-mediated immunity [94]. (CD200, PDL-1 and galectin 1 must all be present to prevent abortions [3,95]). However, PD-1−/− and PDL-1−/− (also known as B7-H1−/−) B6 females mated to allogeneic BALB/c males failed to abort! [96]. In all cases, the unexpected result in the original study or studies can be explained by presence of endotoxin, which is abortogenic and which must be present in adequate quantity and quality in gut flora for abortions to occur in the CBAxDBA/2 model [3]. This is also true for allogeneically-mated B6 female mice [97]. Failure of investigators to acknowledge their anomalous results and explain them will inevitably lead to the need for more studies and increased consumption of animals. In studies of the role of CD200 in the CBAxDBA/2 model, the increase in the abortion rate after injecting antibody to CD200 was greater at a time in pregnancy when endotoxin sensitivity had dropped, and administration of a soluble CD200 achieved a maximal reduction in abortion rates thereby confirming a role for CD200 in the CBAxDBA/2 model [3,29]. Use of CD200 transgenic mice further confirmed the importance of CD200 in endotoxin-treated B6 mice [47].

Refinement

Methods that minimize distress in animals are quite important to prevent the confounding effect of stressors that can boost abortion rates. Abortion rates in the CBAxDBA/2 model in clean-age housing facilities can be surprisingly low, as gut flora are altered, and low abortion rates increase the need for larger numbers of animals in an experiment to achieve statistical significance when effect sizes are small [3]. Some important phenomena such as altered behavior in offspring depend on gut flora [98]. While healthy mice are more likely to give more reliable results, too much hygiene may abrogate the problems one wants to study, and this also applies to spontaneous abortions in CB17 SCID mice [3]. The current epidemic of human allergic disorders such as asthma have been attributed to too much hygiene. Use of the latest and best in vivo technologies is a sine qua non for publication and winning grant support. Animal Care Committees and Animal Research Ethics Committees have a duty to help investigators achieve the 3R's and still perform successful experiments.

In the U.K., the Animals (Scientific Procedures) Act 1986 protects mouse embryos from studies in which they may have lasting harm [83]. That would appear to prohibit studying effects of maternal stress or inflammation on premature parturition or fetal growth restriction (induced by stress and endotoxin), and post-natal problems such as obesity, asthma, or autism spectrum behavior problems, and other health consequences of low birth weight [98–101]. Better a child should suffer than a mouse embryo? To quote Mr. Bumble in Charles Dicken's Oliver Twist Chapter 51 (as applied by Justice Lord Denning), ‘If….the law is an ass.’ As animal experiments on materno-fetal programming are allowed in other jurisdictions given the need to understand, diagnose, prevent and treat serious human disorders, it is clear there are differences of opinion. This author takes no position on the matter, but does recommend public education and debate. Clinical Practice Guidelines are regularly reviewed and updated (i.e. every 5 years). Science and medicine have advanced in the 34 years since 1986.

Conclusions

Data generated using mouse models is likely to continue to be very important for understanding, investigating, and treating human disorders such as pregnancy loss of otherwise normal embryos. The CBAxDBA/2 model has proven useful for 40 years for solving the jigsaw puzzle of immunological mechanisms in spontaneous abortion, and recently has also provided a model for investigating causes and treatment of pre-eclampsia, another serious human pregnancy disorder [3,88] Similar to human couples with RM (Figure 3A), risk of a repeat abortion in the CBAxDBA/2 model in not 100%, and sufficient exposure to paternal antigenic tissues in utero during a first pregnancy may activate anti-abortive protective mechanisms [3,84]. What are those mechanisms? A few of many additional unanswered but important questions are listed in Table 4.

Table 4
Some important unanswered questions
1. What happens to endovascular trophoblasts and blood flow to the placenta between gestation day 8.5 and 9.5 in the CBAxDBA/2 model? 
2. Why are mice on the C57BL background stress-resistant? 
3. How can E. coli LPS O55:B155 and O26:B6 produce different outcomes in C57BL mice? Is it related to effects on galectin 1 triggering of Treg cell generation? [3
4. What do decidual lymphoid cells with γδ TcR recognize? 
5. What triggers C56bright decidual NK cells to make IL-10 and TGF-β2-like molecules? 
6. Is partner specificity in human RM due to a defectiive male who will have miscarriages with another female partner? 
7. What is the seminal plasma peptide that induces an anti-abortion phenotype in CBAxDBA/2 matings and what is the peptide that induces an increased abortion rate in the CBAxDBA/2 model? Are these peptides present in semen of males in an recurrent miscarriage (RM) relationship? 
8. If microRNAs (miRNAs) from a first trimester pregnant woman's blood mononuclear luekocytes correlates with subsequent pregnancy comlications including miscarriage, what do particular miRNAs do? Could exploiting that therapeutically be expedited by studies in mouse models? [102
9. If immunization against abortion in the CBAxDBA/2 model cannot be achieved using paternal DBA/2 lymphoid cells but can be achieved in the B10xB10A model, are there separate groups of human RM couples homologous to each model? Is the pathophysiology of abortion different in C57BL females? 
1. What happens to endovascular trophoblasts and blood flow to the placenta between gestation day 8.5 and 9.5 in the CBAxDBA/2 model? 
2. Why are mice on the C57BL background stress-resistant? 
3. How can E. coli LPS O55:B155 and O26:B6 produce different outcomes in C57BL mice? Is it related to effects on galectin 1 triggering of Treg cell generation? [3
4. What do decidual lymphoid cells with γδ TcR recognize? 
5. What triggers C56bright decidual NK cells to make IL-10 and TGF-β2-like molecules? 
6. Is partner specificity in human RM due to a defectiive male who will have miscarriages with another female partner? 
7. What is the seminal plasma peptide that induces an anti-abortion phenotype in CBAxDBA/2 matings and what is the peptide that induces an increased abortion rate in the CBAxDBA/2 model? Are these peptides present in semen of males in an recurrent miscarriage (RM) relationship? 
8. If microRNAs (miRNAs) from a first trimester pregnant woman's blood mononuclear luekocytes correlates with subsequent pregnancy comlications including miscarriage, what do particular miRNAs do? Could exploiting that therapeutically be expedited by studies in mouse models? [102
9. If immunization against abortion in the CBAxDBA/2 model cannot be achieved using paternal DBA/2 lymphoid cells but can be achieved in the B10xB10A model, are there separate groups of human RM couples homologous to each model? Is the pathophysiology of abortion different in C57BL females? 

Summary

  • Animal model analogs of human diseases have advantages and disadvantages, but discoveries obtained in animals must always be validated by human studies.

  • Inbred strains of laboratory mice are commonly used to investigate immunologic factors that impact reproductive success and failure,

  • The CBAxDBA/2 mating model of recurrent miscarriage is widely used to study mechanisms of recurrent miscarriage, pre-eclampsia, fetal growth restriction, and pre-term birth. Mouse pregnancy has significant homology with humans but differs in details.

  • Findings obtained using mouse pregnancy models have significantly advanced understating of pathophysiological mechanisms of pregnancy success and failure as validated in humans investigation and clinical trials.

Author Contribution

D.A.C. wrote the manuscript.

Competing Interests

The author declares that there are no competing interests associated with this manuscript.

Abbreviations

     
  • DB-RCTs

    double blinded randomized controlled trials

  •  
  • DC

    dendritic cells

  •  
  • EVT

    extravillous trophoblast

  •  
  • IVF

    in vitro fertilization

  •  
  • LIT

    leukocyte-injected

  •  
  • LPS

    lipopolysaccharides

  •  
  • NK

    natural killer

  •  
  • PE

    preeclampsia

  •  
  • RCTs

    randomized controlled trials

  •  
  • RM

    recurrent miscarriages

  •  
  • TTP

    time-to-pregnancy interval

References

References
1
Popper
,
K.
(
2002
)
The Logic of Scientific Discovery
,
Abingdon-on-Thames
,
Routledge, UK
2
Mestas
,
J.
and
Hughes
,
C.C.W.
(
2004
)
Of mice and not men: differences between mouse and human immunology
.
J. Immunol.
172
,
2731
2738
3
Clark
,
D.A.
(
2016
)
Mouse is the new woman? Translational research in reproductive immunology
.
Semin. Immunopathol.
38
,
651
668
4
Efremova
,
M.
,
Vento-Tormo
,
R.
,
Park
,
J.-E.
,
Teichman
,
S.A.
and
James
,
K.R.
(
2020
)
Immunology in the era of single-cell technologies
.
Annu. Rev. Immunol.
38
,
727
757
5
Calhoun
,
J.B.
(
1973
)
Death squared: the exponential growth and demise of a mouse population
.
Proc. R. Soc. Med.
66
,
80
88
PMID:
[PubMed]
6
Li
,
X.
,
Geng
,
M.
,
Peng
,
Y.
,
Meng
,
L.
and
Lu
,
S.
(
2020
)
Molecular immune pathogenesis and diagnosis of COVID-19
.
J. Pharm. Anal.
10
,
102
108
7
Clark
,
D.A.
(
2014
)
Popular myths in reproductive immunology
.
J. Reprod. Immunol.
104–105
,
54
62
8
Chaouat
,
G.
(
2013
)
Inflammation, NK cells and implantation: friend and foe (the good, the bad, and the ugly?): replacing viviparity in an evolutionary perspective
.
J. Reprod. Immunol.
97
,
2
13
9
Coulam
,
C.B.
,
Stephenson
,
M.
,
Stern
,
J.J.
and
Clark
,
D.A.
(
1996
)
Immunotherapy for recurrent pregnancy loss: analysis of results from clinical trials
.
Am. J. Reprod. Immunol.
35
,
352
359
10
Roberts
,
C.J.
and
Lowe
,
C.R.
(
1975
)
Where have all the conceptions gone?
Lancet
305
,
498
499
11
Boué
,
J.
,
Boué
,
A.
and
Lazar
,
P.
(
1975
)
Retrospective and prospective epidemiological studies of 1500 karyotypes spontaneous abortions
.
Teratology
12
,
11
26
12
Evers
,
J.L.H.
(
2002
)
Female subfertility
.
Lancet
360
,
151
159
13
Clark
,
D.A.
(
2003
)
Is there any evidence for immunologically-mediated or immunologically-modifiable early pregnancy failure?
J. Assist. Reprod. Genet.
20
,
63
72
14
RMITG
. (
1994
)
Worldwide collaborative observational study and meta-analysis on allogeneic leukocyte immunotherapy for recurrent spontaneous abortion
.
Am. J. Reprod. Immunol.
32
,
55
72
15
Clark
,
D.A.
,
Gunby
,
J.
and
Daya
,
S.
(
1997
)
The use of allogeneic leukocytes or IV IgG for the treatment of patients with recurrent spontaneous abortions
.
Transfusion Med. Rev.
11
,
85
94
16
Hou
,
Y.
,
Li
,
J.
,
Liu
,
Q.
,
Zhang
,
L.
,
Chen
,
B.
,
Yuanyuan
,
L.
et al (
2020
)
The optimal timing of immunotherapy may improve outcome in women with unexplained recurrent pregnancy loss: a perspective follow-op study in northeastern China
.
Am. J. Reprod. Immunol.
83
,
e13225
17
Clark
,
D.A.
,
Daya
,
S.
,
Coulam
,
C.B.
Gunby
J
, and
the Recurrent Miscarriage Immunotherapy Trialists Group
.
(
1996
)
Implications of abnormal human trophoblast karyotype for the evidence-based approach to the understanding, investigation, and treatment of recurrent spontaneous abortion
.
Am. J. Reprod. Immunol.
35
,
495
498
18
Ogasawara
,
M.
,
Aoki
,
K.
,
Okada
,
S.
and
Suzumori
,
K.
(
2000
)
Embryonic karyotype of abortuses in relation to the number of previous miscarriages
.
Fetil. Steril.
73
,
300
304
19
Clark
,
D.A.
,
McDermott
,
M.R.
and
Szewczuk
,
M.
(
1980
)
Impairment of host versus graft reaction in pregnant mice. II. selective suppression of cytotoxic T cell generation correlates with a soluble suppressor activity and with successful pregnancy
.
Cell. Immunol.
52
,
106
118
20
Chaouat
,
G.
,
Kiger
,
N.
and
Wegmann
,
T.
(
1983
)
Vaccination against spontaneous abortion in mice
.
J. Reprod. Immunol.
5
,
389
392
21
Adamson
,
S.L.
,
Lu
,
Y.
,
Whiteley
,
K.J.
,
Holmyard
,
D.
,
Hemberger
,
M.
,
Pfarrer
,
C.
et al (
2002
)
Interactions between trophoblast cells and the maternal fetal circulation in the mouse placenta
.
Dev. Biol.
250
,
358
373
22
Clark
,
D.A.
(
2014
)
The use and misuse of animal analog models of human pregnancy disorders
.
J. Reprod. Immunol.
103
,
1
8
23
Exley
,
M.A.
and
Boyson
,
J.E.
(
2011
)
Protective role of regulatory decidual γδ T cells in pregnancy
.
Clin. Immunol.
141
,
236
239
24
Clark
,
D.A.
,
Chaouat
,
G.
,
Arck
,
P.C.
,
Mittruecker
,
H.W.
and
Levy
,
G.A.
(
1998
)
Cytokine-dependent abortion in CBA X DBA/2 mice is mediated by the procoagulant fgl2 prothrombinase
.
J. Immunol.
160
,
545
549
PMID:
[PubMed]
25
Clark
,
D.A.
(
2016
)
The importance of being a regulatory T cell in pregnancy
.
J. Reprod. Immunol.
116
,
60
69
26
Clark
,
D.A.
,
Chaput
,
A.
and
Tutton
,
D.
(
1986
)
Active suppression of host-vs-graft reaction in pregnant mice. VII. Spontaneous abortion of allogeneic CBA x DBA/2 fetuses in the uterus of CBA/J mice correlates with deficient non-T suppressor cell activity
.
J. Immunol.
136
,
1668
1675
PMID:
[PubMed]
27
Clark
,
D.A.
,
Rahmati
,
M.
,
Bensussan
,
A.
,
Markert
,
U.R.
and
Chaouat
,
G.
(
2013
)
Seminal plasma peptides may determine maternal immune response that alters success or failure of pregnancy in the abortion-prone CBAxDBA/2 model
.
J. Reprod. Immunol.
99
,
46
53
28
Clark
,
D.A.
(
2019
)
Oral tolerance and the materno-fetal relationship
.
J. Reprod. Immunol.
134–135
,
36
39
29
Clark
,
D.A.
,
Ding
,
J.W.
,
Yu
,
G.
,
Levy
,
G.A.
and
Gorczynski
,
R.M.
(
2001
)
Fgl2 prothrombinase expression in trophoblasts and decidua triggers abortion but may be countered by OX-2
.
Mol. Hum. Reprod.
7
,
185
194
30
Clark
,
D.A.
(
2012
)
The end of evidence-based medicine
.
Inflammopharmacology
20
,
187
193
31
Clark
,
D.A.
,
Coulam
,
C.B.
,
Daya
,
S.
and
Chaouat
,
G.
(
2001
)
Unexplained sporadic and recurrent miscarriage in the new millennium: a critical analysis of immune mechanisms and treatments
.
Hum. Reprod. Update
7
,
501
511
32
Sarno
,
M.
,
Cavalcante
,
M.B.
,
Niag
,
M.
,
Pimentel
,
K.
,
Luz
,
I.
,
Figueiredo
,
B.
et al (
2019
)
Gestational and perinatal outcomes in recurrent miscarriage couples treated with lymphocyte immunotherapy
.
Eur. J. Obstet. Gyecol. Reprod. Biol.
3
,
100036
33
Liu
,
Z.
,
Xu
,
H.
,
Kang
,
X.
,
Wang
,
T.
,
He
,
L.
and
Zhao
,
A.
(
2016
)
Allogeneic lymphocyte immunotherapy for unexplained recurrent spontaneous abortion: a meta-analysis
.
Am. J. Reprod. Immunol.
76
,
443
458
34
Cavalcante
,
M.B.
,
Sarno
,
M.
,
Niag
,
M.
,
Pimentel
,
K.
,
Luz
,
I.
,
Figueiredo
,
B.
et al (
2018
)
Lymphocyte immunotherapy for recurrent miscarriages: predictors of therapeutic success
.
Am. J. Reprod. Immunol.
79
,
e17833
35
Clark
,
D.A.
,
Banwatt
,
D.
and
Chaouat
,
G.
(
1993
)
Stress-triggered abortion in mice is prevented by alloimmunization
.
Am. J. Reprod. Immunol.
29
,
141
147
36
Arck
,
P.C.
,
Merali
,
F.
,
Chaouat
,
G.
and
Clark
,
D.A.
(
1996
)
Inhibition of immunoprotective CD8+ T cells as a basis for stress-triggered substance P-mediated abortion in mice
.
Cell. Immunol.
171
,
226
230
37
Markert
,
U.
,
Arck
,
P.C.
,
McBey
,
B.A.
,
Manuel
,
J.
,
Croy
,
B.A.
,
Marshall
,
J.S.
et al (
1997
)
Stress-triggered abortions are associated with alterations of granulated cells in the decidual
.
Am. J. Reprod. Immunol.
37
,
94
100
38
Derbala
,
Y.
,
Elazzamy
,
H.
,
Bilal
,
M.
,
Reed
,
R.
,
Garcia
,
M.D.S.
,
Skariah
,
A.
et al (
2019
)
Mast cell-induced immunopathology in recurrent pregnancy losses
.
Am. J. Reprod. Immunol.
82
,
e13128
39
Qu
,
F.
,
Wu
,
Y.
,
Zhu
,
Y.H.
,
Barry
,
J.
,
Ding
,
T.
,
Baip
,
T.
et al (
2017
)
The association between psychological stress and miscarriage: a systematic review and meta-analysis
.
Sci. Rep.
7
,
1731
40
Clark
,
D.A.
(
1994
)
Intralipid as treatment for recurrent unexplained abortion?
Am. J. Reprod. Immunol.
32
,
290
293
41
Singh
,
N.
,
Davis
,
A.A.
,
Kumar
,
S.
and
Kriplani
,
A.
(
2019
)
The effect of administration of intravenous intralipid on pregnancy outcomes in women with implantation failure after IVF/ICSI with non-donor oocytes: a randomized controlled trial
.
Eur. J. Obstet. Gynecol. Reprod. Biol.
240
,
45
51
42
Dakhly
,
D.M.R.
,
Bayoumi
,
Y.A.
,
Sharkawy
,
M.
,
Allah
,
S.H.G.
,
Hassan
,
M.A.
,
Gouda
,
H.M.
et al (
2016
)
Intralipid supplementation in women with recurrent spontaneous abortion and elevated levels of natural killer cells
.
Int. J. Gynecol. Obstet.
135
,
324
327
43
Hirayama
,
M.
,
Azuma
,
E.
and
Komada
,
Y.
(
2012
)
Tolerogenic effect of non-inherited maternal antigens in hematopoietic stem cell transplantation
.
Front. Immunol.
3
,
135
44
Hahn
,
S.
,
Hasler
,
P.
,
Vokalova
,
L.
,
van Breda
,
S.V.
,
Than
,
N.G.
,
Hoesli
,
M.
et al (
2019
)
Feto-maternal microchimerism: the pre-eclampsia conundrum
.
Front. Immunol.
10
,
659
45
Chaouat
,
G.
and
Clark
,
D.A.
(
2015
)
Are animal models useful or confusing in understanding the human feto-maternal relationships? A debate
.
J. Reprod. Immunol.
108
,
56
64
46
Clark
,
D.A.
,
Arck
,
P.C.
and
Chaouat
,
G.
(
1999
)
Why did your mother reject you? Immunogenetic determinants of the response to environmental selective pressure expressed at the uterine level
.
Am. J. Reprod. Immunol.
41
,
5
22
47
Yu
,
G.
,
Sun
,
Y.
,
Foerster
,
K.
,
Manuel
,
J.
,
Molina
,
H.
,
Levy
,
G.A.
et al (
2008
)
LPS-induced murine abortions require C5 but not C3, and are prevented by upregulating expression of the CD200 tolerance signaling molecule
.
Am. J. Reprod. Immunol.
60
,
135
140
48
Lin
,
Y.
,
Zeng
,
Y.
,
Di
,
J.
and
Zeng
,
S.
(
2005
)
Murine CD200+ CK7+ trophoblasts in a poly (I:C)-induced embryo resorption model
.
Reproduction
130
,
529
537
49
Thaxton
,
J.E.
,
Nevers
,
T.
,
Lippe
,
E.O.
,
Blois
,
S.M.
,
Saito
,
S.
and
Sharma
,
S.
(
2013
)
NKG2D blockade inhibits poly(I:C)-triggered fetal loss in wild type but not in IL10−/− mice
.
J. Immunol.
190
,
3639
3647
50
Saito
,
S.
,
Akitoshi
,
N.
,
Subaru
,
M.H.
and
Shiozaki
,
A.
(
2008
)
The balance between cytotoxic NK cells and regulatory NK cells in human pregnancy
.
J. Reprod. Immunol.
77
,
14
22
51
Lin
,
Y.
,
Liu
,
X.
,
Shan
,
B.S.
,
Wu
,
J.
,
Sharma
,
S.
and
Sun
,
Y.
(
2014
)
Prevention of CpG-induced pregnancy disruption by adoptive transfer of in vitro-induced regulatory T cells
.
PLoS ONE
9
,
e94702
52
Robertson
,
S.A.
,
Skinner
,
R.J.
and
Care
,
A.S.
(
2006
)
Essential role for IL-10 in resistance to lipopolysaccharide-induced preterm labor in mice
.
J. Immunol.
177
,
4888
4896
53
Kang
,
X.
,
Zhang
,
X.
,
Liu
,
Z.
,
Xu
,
H.
,
Wang
,
T.
and
Zhao
,
A.
(
2015
)
Excessive TLR9 signaling contributes to the pathogenesis of spontaneous abortion through impairment of treg cell survival by activation of caspase 8/3
.
Int. Immunopharmacol.
29
,
285
292
54
Chaouat
,
G.
(
1994
)
Synergy of lipopolysaccharide and inflammatory cytokines in murine pregnancy: alloimmunization prevents abortion but does not affect the induction of preterm birth
.
Cell. Immunol.
157
,
328
340
55
Weir
,
J.A.
(
1960
)
A sex ratio factor in the house mouse that is transmitted by the male
.
Genetics
45
,
1539
1552
PMID:
[PubMed]
56
Orzak
,
S.H.
,
Stubblefield
,
J.W.
,
Akmaev
,
V.R.
,
Colls
,
P.
,
Munne
,
S.
,
Scholl
,
T.
et al (
2015
)
The human sex ratio from conception to birth
.
Proc. Natl Acad. Sci. U.S.A.
112
,
E2102
E2111
57
Anonymous
. (
2020
)
Human sex ratio. Wikipedia
https//en.wikipedia.org/wiki/Human_sex_ratio
58
Chason
,
R.J.
,
McLain
,
A.C.
,
Sundaram
,
R.
,
Chen
,
Z.
,
Segars
,
J.H.
,
Pyper
,
C.
et al (
2012
)
Preconception stress and the secondary sex ratio: a prospective cohort study
.
Fertil. Steril.
98
,
937
941
59
Ruckstuhl
,
K.E.
,
Colijn
,
G.P.
,
Amiot
,
V.
and
Vinish
,
E.
(
2010
)
Mother's occupation and sex ratio at birth
.
BMC Public Health
10
,
269
60
Ryan
,
C.P.
,
Anderson
,
W.G.
,
Gardiner
,
L.E.
and
Hare
,
J.F.
(
2012
)
Stress-induced sex ratios in ground squirrels: support for a mechanistic hypothesis
.
Behav. Ecol.
23
,
160
167
61
Schacht
,
R.
,
Tharp
,
D.
and
Smith
,
K.R.
(
2019
)
Sex ratios at birth vary with environmental harshness but not maternal condition
.
Sci. Rep.
9
,
9066
62
Liu
,
C.
,
Wang
,
X.Z.
and
Sun
,
X.B.
(
2013
)
Assessment of sperm antigen specific T regulatory cells in women with recurrent miscarriage
.
Early Hum. Dev.
89
,
95
100
63
Christiansen
,
O.B.
,
Sterrensen
,
R.
and
Nielsen
,
H.S.
(
2011
)
Anti-HY responses in pregnancy disorders
.
Am. J. Reprod. Immunol.
66
,
93
100
64
Clark
,
D.A.
(
2011
)
Intravenous immunoglobulin and idiopathic secondary recurrent miscarriage: methodologic problems
.
Hum. Reprod.
26
,
2586
2587
65
Robertson
,
S.A.
,
Care
,
A.S.
and
Moldenhauer
,
L.M.
(
2018
)
Regulatory T cells in embryo implantation and the immune response to pregnancy
.
J. Clin. Invest.
10
,
4224
4235
66
Clark
,
D.A.
(
2009
)
Should anti-TNF-α therapy be offered to patients with infertility and recurrent spontaneous abortion?
Am. J. Reprod. Immunol.
61
,
107
112
67
Clark
,
D.A.
,
Dmetrichuk
,
J.M.
,
McCready
,
E.
,
Dhesy-Thind
,
S.
and
Arredondo
,
J.L.
(
2017
)
Changes in expression of the CD200 tolerance-signaling molecule and its receptor (CD200R) by villus trophoblasts during first trimester missed abortion and in chronic histocytic intervillositis
.
Am. J. Reprod. Immunol.
78
,
e12665
68
Saccone
,
G.
,
Di Spiezio Sardo
,
A.
,
Ciardilli
,
A.
,
Spinelli
,
M.
,
Surbek
,
D.
and
von Wolff
,
M.
(
2019
)
Effectiveness of seminal plasma in in vitro fertilization treatment: a systematic review and meta-analysis
.
BJOG
126
,
220
225
69
Clark
,
D.A.
,
McCready
,
M.E.
,
Harris
,
K.
,
Malloy
,
L.
and
Arredondo
,
J.L.
(
2018
)
Trophoblast CD200 expression in successful human pregnancies and missed abortions
.
J. Reprod. Immunol.
127
,
55
57
70
Clark
,
D.A.
,
Arredondo
,
J.L.
and
Dhesy-Thind
,
S.
(
2016
)
Corrigendum to ‘The CD200 tolerance-signaling molecule and its receptor, CD200R1, are expressed in human placental villus trophoblast and in peri-implant decidua by 5 weeks gestation’ [J Reprod Immunol 2015;112:20-23]
.
J. Reprod. Immunol.
117
,
89
90
71
Larsen
,
E.C.
,
Christiansen
,
O.B.
,
Kolte
,
A.M.
and
Macklon
,
N.
(
2013
)
New insights into mechanisms behind miscarriage
.
BMC Med.
11
,
154
72
Salker
,
M.
,
Teklenburg
,
G.
,
Molokhia
,
M.
,
Lavery
,
S.
,
Trew
,
G.
,
Aojanepong
,
T.
et al (
2010
)
Natural selection of human embryos: impaired decidualization of endometrium disables embryo-maternal interactions and causes recurrent pregnancy loss
.
PLoS ONE
5
,
e10287
73
Weimar
,
C.H.E.
,
Kavelaars
,
A.
,
Brosens
,
J.J.
,
Gelersen
,
B.
,
de Vreeden-Elbertse
,
J.M.T.
,
Heijnen
,
C.J.
et al (
2012
)
Endometrial stromal cells of women with recurrent miscarriage faile to discriminate between high- and low-quality human embryos
.
PLoS ONE
7
,
e41424
74
Lucas
,
E.S.
,
Vrlijcak
,
P.
,
Muter
,
J.
,
Diniz-da-Costa
,
M.M.
,
Brighton
,
P.J.
,
Kong
,
C.S.
et al (
2020
)
Recurrent pregnancy loss is associated with a pro-senescent decidual response during the peri-implantation window
.
Nat. Commun. Biol.
3
,
37
75
Wilcox
,
A.J.
,
Baird
,
D.D.
and
Weinberg
,
C.R.
(
1999
)
Time of implantation of the conceptus and loss of pregnancy
.
N. Engl. J. Med.
340
,
1796
1799
76
Quenby
,
S.
,
Nik
,
H.
,
Innes
,
B.
,
Lash
,
G.
,
Turner
,
M.
,
Drury
,
J.
et al (
2009
)
Uterine natural killer cells and angiogenesis in recurrent reproductive failure
.
Hum. Reprod.
24
,
45
54
77
Quenby
,
S.
,
Farquharson
,
R.
,
Young
,
M.
and
Vince
,
G.
(
2003
)
Successful pregnancy outcome following 19 consecutive miscarriages: case report
.
Hum. Reprod.
18
,
2562
2564
78
King
,
K.
,
Smith
,
S.
,
Chapman
,
M.
and
Sacks
,
G.
(
2010
)
Detailed analysis of peripheral blood natural killer (NK) cells in women with recurrent miscarriage
.
Hum. Reprod.
25
,
52
58
79
Wang
,
W.J.
,
Garcia
,
M.D.S.
,
Deutsch
,
G.
,
Sung
,
N.
,
Yang
,
X.
,
He
,
Q.
et al (
2020
)
PD-1 and PD-L1 expression on T-cell subsets in women with unexplained recurrent pregnancy loss
.
Am. J. Reprod. Immunol.
83
,
e13230
80
Muzikova
,
E.
and
Clark
,
D.A.
(
1995
)
Is spontaneous resorption in the DBA/2-mated CBA/J mouse due to a defect in ‘seed’ or in ‘soil"?
Am. J. Reprod. Immunol.
33
,
81
85
81
Dupont
,
C.
,
Segars
,
J.
,
DeCherney
,
A.
,
Bavister
,
B.D.
,
Armant
,
D.R.
and
Brenner
,
C.A.
(
2010
)
Incidence of chromosomal mosaicism in morphologically normal non-human primate preimplantation embryos
.
Fertil. Steril.
93
,
2545
2550
82
Balbach
,
S.T.
,
Jauch
,
A.
,
Bohm-Steuer
,
B.
,
Cavaleri
,
F.M.
,
Han
,
Y.M.
and
Boiani
,
M.
(
2007
)
Chromosome stability differs in clones mouse embryos and derivative ES cells
.
Dev. Biol.
308
,
309
321
83
National Centre for the Replacement Refinement and Reduction of Animals in Research. NC3Rs
. (
2020
) www.understandinganimalsinresearch.org.uk
84
Aghaeepour
,
N.
,
Ganio
,
E.A.
,
Mcilwain
,
D.
,
Tsai
,
A.S.
,
Tingle
,
M.
,
van Gassen
,
S.
et al (
2017
)
An immune clock of human pregnancy
.
Sci. Immunol
2
,
eaan2946
85
Vento-Tormo
,
R.
,
Efremova
,
M.
,
Botting
,
R.A.
,
Turco
,
M.Y.
,
Vento-Tormo
,
M.
,
Meyer
,
K.B.
et al (
2018
)
Single-cell reconstitution of the early maternal-fetal interface in humans
.
Nature
563
,
347
353
86
Westoby
,
J.
,
Artemov
,
P.
,
Hemberg
,
M.
and
Ferguson-Smith
,
A.
(
2020
)
Obstacles to detecting isoforms using full-length scRNA-seq data
.
Genome Biol.
21
,
74
87
Fu
,
B.
,
Li
,
X.
,
Sun
,
R.
,
Tong
,
X.
,
Ling
,
B.
,
Tian
,
Z.
et al (
2012
)
Natural killer cells promote immune tolerance by regulating inflammatory TH17 cells at the human maternal-fetal interface
.
Proc. Natl Acad. Sci. U.S.A.
110
,
E231
E240
88
Bonney
,
E.A.
and
Brown
,
S.A.
(
2014
)
To drive or be driven: the path of a mouse model of recurrent pregnancy loss
.
Reproduction
147
,
R153
R167
89
Ren
,
J.
,
Zeng
,
W.
,
Tian
,
F.
,
Zhang
,
S.
,
Wu
,
F.M.
,
Qin
,
X.
, et al (
2019
)
Myeloid-derived suppressor cells depletion may cause pregnancy loss via upregulating the cytotoxicity of decidual natural killer cells
.
Am. J. Reprod. Immunol.
81
,
e13099
90
Gelber
,
S.E.
,
Brent
,
E.
,
Redecha
,
P.
,
Perino
,
G.
,
Tomlinson
,
S.
,
Davissor
,
R.L.
et al (
2015
)
Prevention of defective placentation and pregnancy loss by blocking innate immune pathways in a syngeneic model of placental insufficiency
.
J. Immunol.
195
,
1129
1138
91
de Wilde
,
V.
,
van Rompaey
,
N.
,
Hill
,
M.
,
Lebrun
,
J.F.
,
Lemaitre
,
P.
,
Lhomme
,
F.
et al (
2009
)
Endotoxin-induced myeloid-derived suppressor cells inhibit alloimmune responses via heme oxygenase-1
.
Am. J. Transplant.
9
,
2034
2047
92
Pan
,
T.
,
Liu
,
Y.
,
Zhong
,
L.M.
,
Shi
,
M.H.
,
Duan
,
X.B.
,
Wu
,
K.
et al (
2016
)
Myeloid-derived suppressor cells are essential form maintaining feto-maternal tolerance via STAT3 signaling in mice
.
J. Leuk. Biol.
100
,
499
511
93
Guleria
,
I.
,
Khosroshahi
,
A.
,
Ansari
,
M.J.
,
Habicht
,
A.
,
Azuma
,
M.
,
Yagita
,
H.
et al (
2005
)
A critical role for the programmed death ligand 1 in fetomaternal tolerance
.
J. Exp. Med.
202
,
231
237
94
Rui
,
Y.
,
Honjo
,
T.
and
Chikuma
,
S.
(
2013
)
Programmed cell death 1 inhibits inflammatory helper T-cell development through controlling the innate immune response
.
Proc. Natl Acad. Sci. U.S.A.
110
,
16073
16078
95
Blois
,
S.M.
,
Ilarregui
,
J.M.
,
Tometten
,
M.
,
Garcia
,
M.
,
Orsal
,
A.S.
,
Cordo-Russo
,
R.
et al (
2007
)
A pivotal role for galectin-1 in fetomaternal tolerance
.
Nat. Med.
13
,
1450
1457
96
Taglauer
,
E.S.
,
Yankee
,
T.M.
and
Petroff
,
M.G.
(
2009
)
Maternal PD-1 regulates accumulation of fetal antigen-specific CD8+ T cells in pregnancy
.
J. Reprod. Immunol.
80
,
12
21
97
Lee
,
A.L.
,
Kandia
,
N.
,
Karimi
,
K.
,
Clark
,
D.A.
and
Ashkar
,
A.A.
(
2013
)
Interleukin-15 is required for lipopolysaccharide-induced abortion
.
J. Leuk. Biol.
93
,
905
912
98
Kim
,
S.
,
Kim
,
H.
,
Yim
,
Y.S.
,
Ha
,
S.
,
Atarashi
,
K.
,
Tan
,
T.G.
et al (
2017
)
Maternal gut bacteria promote neurodevelopmental abnormalities in mouse offspring
.
Nature
549
,
528
532
99
Reed
,
M.D.
,
Yim
,
Y.S.
,
Wimmer
,
R.D.
,
Kim
,
H.
,
Ryu
,
C.
,
Welch
,
G.M.
et al (
2020
)
IL-17a promotes sociability in mouse models of developmental disorders
.
Nature
577
,
249
252
100
Choi
,
G.B.
,
Yim
,
Y.S.
,
Wong
,
H.
,
Kim
,
S.
,
Kim
,
H.
,
Kim
,
S.V.
et al (
2016
)
The maternal interleukin-17a pathway in mice promotes autism-like phenotypes in offspring
.
Science
351
,
933
939
101
Duros
,
K.
,
Moustaki
,
M.
,
Tsabouri
,
S.
,
Papadopoulou
,
A.
,
Papadopoulos
,
M.
and
Priftis
,
K.N.
(
2017
)
Prenatal maternal stress and the risk of asthma in children
.
Front. Pediatr.
5
,
202
102
Winger
,
E.E.
,
Reed
,
J.L.
and
Ji
,
X.
(
2015
)
First-trimester maternal cell microRNAs is a superior pregnancy marker to immunological testing for predicting adverse pregnancy outcome
.
J. Reprod. Immunol.
110
,
22
25