Establishment of a pregnant HIS mouse model. We first examined whether Thy/HSC HIS mice generated using the conventional protocol can be used for the human materno-fetal immunity study. Six- to 8-week-old female NSG mice were preconditioned with sublethal TBI, followed by transplantation of human fetal thymic tissue (under the renal capsule) and intravenous injection of human CD34+ fetal liver cells (24, 32, 33). High levels of multilineage human lymphohematopoietic cells were detected in peripheral blood mononuclear cells (PBMCs) of the mice from week 6 after humanization (Figure 1A); this was consistent with our previous reports (32). However, these HIS mice failed to get pregnant after mating with NSG males for 2 months. This was presumably due to the TBI treatment, which causes severe damage to mouse reproductive capabilities (34, 35). Therefore, we conditioned the mice by limb local irradiation to construct the HIS mouse model to avoid irradiating the reproductive organs. All 5 HIS mice got pregnant and delivered healthy pups after mating with NSG males; however, all mice had very low levels (<1.5%) of human lymphohematopoietic cells (Figure 1B).

Construction of a pregnant HIS mouse model with high levels of human lymphohematopoietic cell reconstitution. (A–C) Ratios (mean ± SEM) of multilineage human lymphohematopoietic cell reconstitution within human CD45+ plus mouse CD45+ cell population of HIS mice made by intravenous injection of human CD34+ fetal liver cells and transplantation of human fetal thymic tissue under the renal capsule after preconditioning by TBI (A; n = 5), limb local irradiation (B; n = 8), or busulfan treatment (C; n = 19). Percentages of human CD45+ leukocytes, CD3+ T cells, CD19+/CD20+ B cells, NK cells, and CD33+ myeloid cells in PBMCs at indicated weeks were shown. An unpaired t test was used to analyze the differences in the ratios of human CD45+ cells, T cells, and B cells in the PBMCs at week 9 between the TBI and busulfan groups: CD45+ cells: P < 0.01; T cells: P < 0.001; B cells: P < 0.0001. (D) The average weight of neonatal mice born to NSG or HIS mice that were mated with BALB/c male mice (busulfan treatment, n = 5). Box plots show the interquartile range, median (line), and minimum and maximum (whiskers). (E) The number of neonatal mice born by NSG or humanized mice that were mated with BALB/c male mice (busulfan treatment, n = 5). (F) The average fetus number (mean ± SEM) of pregnant HIS mice made by mating with NSG males (n = 12) or BALB/c males (n = 13) summarized (busulfan treatment) at E14.5. (G) Ratio of placental junctional to labyrinth zone (JZ/L ratio) in humanized mice (hu-mice) (n = 10) and NSG mice (n = 6) on E14.5. (H) Pictures of placentas and fetuses of a representative HIS mouse euthanized at E14.5 were shown. Statistical analyses were performed using unpaired t test (D–G).

After several tests, we found that HIS mice generated using the busulfan-based preconditioning protocol, a bifunctional alkylating agent that kills proliferating hematopoietic stem/progenitor cells in recipients by crosslinking guanine bases in DNA double-helix strands when used as a myeloablative regimen before hematopoietic stem cell transplantation (36), with less toxicity against reproductive organs (37), have high levels of human lymphohematopoietic cells, composed of reconstituted human CD3+ T cells, CD20+ B cells, CD56+ NK cells, and CD33+ myeloid cells (Figure 1C). The HIS female mice efficiently got pregnant and delivered healthy pups. A comparable number and weight of well-developed neonatal mice were observed between the HIS mice and NSG mice after mating with BALB/c males (Figure 1, D and E). In addition, the fetus number of HIS female mice after mating with NSG or BALB/c males was comparable at E14.5 (Figure 1, F and H). There was no significant difference in the ratio of placental junctional to labyrinth zone between HIS mice and NSG mice (Figure 1G). No fetal demise or resorption events were observed in busulfan-treated pregnant mice. These data demonstrate that pregnant HIS mice can be constructed by busulfan preconditioning.

Composition of multilineage human immune subsets at mouse materno-fetal interface. Pregnant HIS mice were euthanized at E14.5 after mating with syngeneic NSG or allogeneic BALB/c males to examine the human immune cell composition and phenotype at decidua. Multilineage human immune cells were detected in the mouse decidua, PBMCs, and spleen under either mating condition (Figure 2, A and B, and Supplemental Figure 1, A and B; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.176527DS1). The vast majority were T and B cells, and only small populations were macrophages and especially NK cells. While human CD14+HLA-DR+ macrophages and CD56+ NK cells were markedly enriched in the decidua compared with that in the PBMCs and spleen (Figure 2, D and E, and Supplemental Figure 1, D and E), similar to what is observed in humans. Human maternal immune cells at the decidua are naturally exposed to semiallogeneic paternally derived antigens (38). Therefore, our study focused on pregnant HIS mice made by mating with BALB/c males in the following experiments.

Human immune subset composition in pregnant HIS mice at E14.5. HIS mice (n = 10) made by the busulfan protocol were mated with BALB/c males and euthanized at E14.5 for analysis. Summarized data about the chimerism (mean ± SEM) of human lymphohematopoietic cells (A) (%huCD45+/[%hCD45+ + %mCD45+]) and ratios (B) of CD3+ T cells, CD20+ B cells, CD14+ macrophages, and CD56+ NK cells within human CD45+ cells in PBMCs, spleen, and decidua were shown. (C) Summarized results (mean ± SEM; n = 7) of the percentages of CD49a+ and CD49a+CD103+ tissue-resident NK cells in human CD56+CD16– NK cells. (D) Summarized results (mean ± SEM; n = 7) of the percentages of CD56+CD16– NK cells and representative flow cytometric profiles were shown. (E–G) Summarized results (mean ± SEM; n = 7) of the percentages of CD14+HLA-DR+ human macrophages (E), CD206+ M2 macrophages (F), and CD80+ M1 macrophages (G) were shown. (H) Summarized results (mean ± SEM; left; n = 5) and representative flow cytometry (FCM) profiles (right) of the ratios of Treg cells. (I) Immunohistochemical (IHC) examination of human CD45+ lymphohematopoietic cells (left) and CD4+ T cells (right) in decidua. D, decidua. Statistical differences were determined with 1-way ANOVA for multiple-variable comparisons (A and C–H).

Immune subsets at the materno-fetal interface have specialized phenotypes and functions (39). We found an approximately 7-fold higher number of CD56+CD16– NK cells (a specific NK subset that secretes cytokines) in the decidua compared with that in the PBMCs and spleen (Figure 2D). Moreover, markedly higher levels of CD49a and CD103 (2 molecules that play crucial roles in retaining tissue characteristics and can be used to identify human dNK cells) were detected in human NK cells in the decidua than in the PBMCs and spleen (Figure 2C). Additionally, human immune subsets with potent immune-modulatory function were significantly enriched at the materno-fetal interface. Higher ratios of immunosuppressive M2 macrophages (CD206+), but not inflammatory M1 macrophages (CD80+), were observed in human CD14+HLA-DR+ macrophages in the decidua than those in the PBMCs or spleen (Figure 2, F and G, and Supplemental Figure 1, E–G). Furthermore, significantly higher ratios of CD25+Foxp3+CD4+ Treg cells (the immune subset playing key roles in immune tolerance induction and maintenance) were found in the decidua than in the PBMCs and the spleen (Figure 2H). Marked reduction of CD45RA+CCR7+ naive T cell and elevation of CD45RA–CCR7+ effector memory T cell (CD4+ and CD8+ T cells) counts were observed in the decidua (Supplemental Figure 2, B and C, and Supplemental Figure 3, A and B), which appeared to acquire an experienced and differentiated phenotype. Interestingly, the proportions of decidua human CD49a+ dNK, CD49a+CD103+ dNK, and CD25+Foxp3+CD4+ Treg cells were markedly higher under allogeneic mating conditions than under syngeneic mating conditions (Figure 2, C and H; Supplemental Figure 1C; and Supplemental Figure 2A). IHC examination of HIS mouse decidua samples further substantiated the distribution of human CD45+ lymphohematopoietic cells and human CD4+ T cells at the decidua (Figure 2I). These data demonstrate that pregnant HIS mice host a specialized HIS at the materno-fetal interface.

Tolerogenic to inflammatory conversion of the materno-fetal immune profile across the mid to late gestation stages. Traditionally, materno-fetal immunity was considered under static immunosuppressive status to inhibit maternal versus semiallogeneic fetal immune responses. However, recent studies demonstrated dynamic alternations in the HIS at the materno-fetal interface across the period of pregnancy (3). Human CD45+ lymphohematopoietic cells were purified from mouse decidua and spleen at the mid (E14.5) and late (E19) gestation stages and characterized using single-cell RNA-Seq (scRNA-Seq) to examine whether a pregnant HIS mouse model can simulate this process (Figure 3A). A total of 8,217, 9,932, 4,258, and 4,304 qualified single-cell transcriptomes were acquired from E14.5 decidua, E14.5 spleen, E19 decidua, and E19 spleen samples, respectively. We identified 13 cell clusters, including B cells, T cells, basophils, plasmablasts, dendritic cells (DCs), progenitor cells, CD16+ NK cells, CD56+ NK cells, NKp cells (proliferative NK cells), monocytes, macrophages, M1 macrophages, and M2 macrophages, in E14.5 decidua based on Seurat and annotated with SingleR. Interestingly, identification of the lineage-negative human lymphohematopoietic progenitor subset (cluster 6) at the decidua supports the hypothesis that partial human immune subsets at the decidua might locally differentiate (40). The E19 decidua contained 9 human immune cell clusters, including T cells, B cells, NKp cells, CD56+ NK cells, CD16+ NK cells, monocytes, M1 macrophages, M0 macrophages, and DCs (Figure 3, B and C, and Supplemental Figure 4). Integration analysis revealed that the composition of human immune cells in the E14.5 decidua from HIS mice was similar to that of human samples collected at similar gestation stage (41), though there were fewer human macrophages, monocytes, and NK cells but more human T and B cells in HIS mice than humans (Supplemental Figure 5). The disappearance of M2 macrophages at the E19 decidua implies a shift in the immune profile from tolerogenic to inflammatory at the materno-fetal interface in this period, as indicated in a previous study (42).

Human immune profile alternation in the decidua and spleen of pregnant HIS mice at E14.5 and E19. Human CD45+ cells were purified from the spleen and decidua of pregnant HIS mice made by mating with BALB/c males at mid and late gestation and examined by scRNA-Seq (mother N = 1 per time point; E14.5, n = 6 decidua, pooled; E19, n = 5 decidua, pooled). Each sample represents a single pregnant mouse. Decidua from the same mouse were pooled together. (A and B) Workflow diagram (A) and the t-distributed stochastic neighbor embedding (t-SNE) plots of main cell types (B) were shown. GEMs, gel beads in emulsion; DC, dendritic cells; M0, M0 macrophages; M1, M1 macrophages; M2, M2 macrophages; NKp, NK proliferative cells. (C) Expression of representative markers for different cell clusters are plotted onto the t-SNE map. Color key from gray to purple indicates relative expression levels from low to high. (i) CD79A (B cells), (ii) CD3D (T cells), (iii) HDC (basophils), (iv) FKBP11 (plasmablasts), (v) LILRA4 (DCs), (vi) SPARC (progenitor cells), (vii) GZMM (CD16+ NK), (viii) XCL2 (CD56+ NK), (ix) TYMS (NKp), (x) S100A8 (monocytes), (xi) C15orf48 (macrophages), (xii) LYZ (M1), and (xiii) FCER1G (M2). (D) The cell-cell interaction profiles at E14.5 decidua and E19 decidua made by CellPhoneDB were shown. (E and F) Heatmaps show average count of the genes annotated as cytokines (E) and chemokine receptors (F, upper panel) and chemokine ligands (F, lower panel) in different cell types.

The cell–cell interaction pattern was analyzed using CellPhoneDB (20) to understand the regulation of the human immune profile. Distinct human immune cell–cell interaction patterns (e.g., close crosstalk between macrophages and T cells) were seen between the decidua (Figure 3D) and spleen, which is similar to what was observed in human sample analysis (43) (Supplemental Figure 6). The human M0 and M2 subsets in the E14.5 decidua offered intensive immune-modulating signals toward other lineages of human immune subsets, whereas M1 macrophages were the key node within the cell–cell interaction network at E19 decidua (Figure 3D). The detailed roles of different immune subsets were determined by examining their cytokine and chemokine production pattern. Human decidua M0 macrophages produced a much broader range of cytokines and chemokines compared with other lineages of human immune subsets (Figure 3E). This was consistent with CellPhoneDB data. Interestingly, human decidua macrophages at E14.5 secreted a series of functional immunosuppressive cytokines (such as IL-4, IL-10, IL-13, and TGFB1) and angiogenic mediators (such as IL-8) that contribute to immune tolerance induction and decidua development. However, they changed to produce a set of inflammatory cytokines (such as IL-6, TNF, IL-1A, and IL-1B) (42, 44, 45) at E19, which are crucial for pregnancy termination and fetal delivery. In addition, human decidua macrophages showed potent capabilities for chemokine secretion, wherein the counts of macrophages known for promoting endocytic activity/cell growth and tissue-repairing functions (characterized by mediators such as CCL18, CXCL16, and CCL2) were increased at E14.5, whereas the counts of M0 macrophages known for inflammation induction (characterized by mediators such as CXCL8 and CXCL9) (46) were increased at E19 (Figure 3F). Upregulation of CXCR4 expression on human E14.5 decidua macrophages suggests that they may be recruited by mouse trophoblast cells that are capable of producing CXCL12 (a CXCR4 ligand) from the periphery (47) (Figure 3F). Decidual Treg (dTreg) cells exhibit different cytokine production patterns between E14.5 and E19; there is a reduction in the expression of TGFB1 (an immune-inhibitory mediator; ref. 48) and elevation in the expression of IL21 (a cytokine that negatively regulates Treg cell stability/function; ref. 49). Because TGF-β1 protein undergoes posttranslational modifications to become an active form (50) and no reports showed that Treg cells can produce IL-21 protein (51), we measured TGF-β1 and IL-21 at the protein level by FCM. We found that both IL-21 and TGF-β1 proteins were significantly upregulated in E19 dTreg cells compared with E14.5 dTreg cells (Supplemental Figure 7). These data imply a dynamic transition of the human immune profile from tolerogenic status to inflammatory status at the materno-fetal interface of HIS mice during pregnancy, while the alternation of dTreg cell function in this period requires further examination.

Decidual Treg cells switch from immune-inhibitory status to activated status from mid to late gestation. Treg cells are an immune subset playing key roles in materno-fetal immune response regulation. Abnormal number or functionality of human Treg cells is clinically associated with several reproductive complications (16, 52). Thus, we focused on characterizing human dTreg cell composition, differentiation, activation, and function alteration in pregnant HIS mice. Initial analysis of the TCR repertoire of Treg cells revealed that around 15% of splenic Treg (sTreg) cells were monoclonal, while approximately 50% of dTreg cells were monoclonal, including approximately 10% of dTreg cells detected with over 10 identical clones. The proportions of monoclonal splenic conventional CD4+ T (sTconv) cells and decidua conventional CD4+ T (dTconv) cells were much lower (Figure 4A). This finding demonstrated the existence of robust clonal expansion in dTreg cells. TCR overlaps between dTreg and dTconv, sTreg and sTconv, and dTreg and sTreg cells were analyzed. Only 1 out of 107 sTreg cell shared the same TCR with sTconv cells, whereas 16 out of 372 dTreg cells possessed the same TCR as dTconv cells, indicating that increasing numbers of human induced Treg (iTreg) cells were developed at the materno-fetal interface. However, 20 out of 372 dTreg cells shared the same TCR with sTreg cells, suggesting human thymic derived Treg (tTreg) cells that migrated from lymphoid organs also existed at the decidua (Figure 4B). Consistently, IKZF2 (a classical gene expressed in tTreg cells) was found selectively highly expressed in decidual tTreg cells but not iTreg cells (Figure 4E). Additionally, FCM examination further verified that there were significantly fewer Helios-positive (the protein coded by IKZF2) cells within dTreg cells compared with sTreg cells (Figure 4, C and D).

TCR repertoire and immune profile transition of human dTreg cells at E14.5 and E19. (A) Quantification of percentage of T cells per clone size. (B) Venn diagrams showing the number of TCR clonotypes in dTreg, sTreg, dTconv, and sTconv cells. (C) Ratio of Helios+ cells in CD4+CD25+Foxp3+ Treg cells in decidua and spleen at E19 (n = 4). Significance was determined using paired t test. (D) Representative flow cytometric profiles of Helios expression in decidual and splenic Treg cells. (E) Violin plots showing the expression of IKZF2 in tTreg and iTreg cells. (F) Volcano plot of differential gene expression between dTreg cells at E19 and E14.5. (G) Heatmaps show genes related to Treg activation/differentiation (left panel), tissue resident (middle panel), and transcription factors (right panel). (H) Pseudotime trajectories of Treg, EM CD4+ T, and naive CD4+ T cells in decidua and spleen were analyzed using Monocle 2.

The top 10 genes enriched in dTconv cells, dTreg cells, sTconv cells, and sTreg cells at E14.5 (Supplemental Figure 8A) and E19 (Supplemental Figure 8B) were then clustered. A series of genes that play crucial roles in maintaining Treg cell immune-regulatory function and stability (including FOXP3, TIGIT, IKZF2, RTKN2, ISG20, PERP, SAT1, CXCR4, DUSP4, and RP11-1399P15.1) were markedly upregulated in E14.5 dTreg cells compared with that in E14.5 sTreg or E14.5 Tconv cells (Supplemental Figure 8A and Supplemental Figure 9A). Specifically, higher expression of DUSP4 and RGS1 in E14.5 dTreg cells than in E14.5 sTreg cells may enhance their immunosuppressive function and maintain their retention in the decidua (53, 54). The gene enrichment patterns in E19 dTreg cells were vastly different from that of E19 sTreg cells, wherein a number of genes that are relevant in inflammation, activation, and proliferation (such as CD52, LGALS1, S100A4, S100A6, IL-32, and GNLY) showed a prominent upregulation in E19 dTreg cells (Supplemental Figure 8B and Supplemental Figure 9B). Direct comparison of the gene expression profiles of E14.5 dTreg cells and E19 dTreg cells indicated that the genes involving Treg cell activation and instability (including HLA-DRB1, CD52, IL7R, and LGALS1) were upregulated, while the genes critical for Treg cell stability and inhibitory function (including KLF2, TIGIT, DUSP4, and CD27) were downregulated (Figure 4F). This finding suggests that a large immune status transition occurred from mid to late gestation. Gene Ontology analysis of differentially expressed genes for E19 dTreg and E14.5 dTreg subsets also revealed that E19 dTreg cells were activated and proliferating (Supplemental Figure 10).

The differences between E14.5 dTreg cells and E19 dTreg cells were analyzed by assessing the differentially expressed genes that are associated with activation/differentiation, tissue location, and transcription factors. E19 dTreg cells upregulated a set of activation/differentiation-related genes (such as LAG3 and PDCD1) and tissue resident–related genes (such as ITGAE, S1PR1, CX3CR1, CXCL13, and KLF2) while downregulating transcription factor genes known for negatively affecting FOXP3 stability and immune-inhibitory function (such as MAF and NFIL3). This suggested that E19 dTreg cells may be actively confronting the inflammatory microenvironment at the decidua right before labor (Figure 4G). E19 sTreg cells also showed upregulation of a series of genes associated with immune regulation (such as IL2RA, TIGIT, and CTLA4) and transcription factors (such as FOXP3 and IKZF2) that are essential for enhancing Treg cell immune-inhibitory function, which differed from that of E19 dTreg cells. This might suppress potential autoimmune disorders in the whole body driven by the immune reactions during fetal delivery. Temporal dynamic trajectory analysis was performed by Monocle 2 assays (55) to better understand the differentiation status of dTreg cells. Naive T cells were found at the start point of the trajectory and gradually transitioned toward effector memory (EM) T cells, which was consistent with previous reports (56). E14.5 dTreg cells and E14.5 sTreg cells exhibited a transitional distribution in the trajectory, with the differentiation of E14.5 sTreg cells being closer to the start point than that of E14.5 dTreg cells. E19 dTreg cells and E19 sTreg cells were concentrated in different terminally differentiated regions (Figure 4H). These data revealed an immunological feature transition of human dTreg cells in pregnant HIS mice during pregnancy.

Human Treg cells suppress maternal versus fetal T cell reactions and maintain immune tolerance. Last, we investigated whether the human materno-fetal immune system in HIS mice was functional. Human decidua and splenic CD4+CD25+ Treg cells were purified from pregnant HIS mice and incubated with the CFSE-labeled human CD45+CD25– splenic cells (responder cells from pregnant HIS mice) in the presence of 30 Gy irradiated BALB/c mouse splenic cells (stimulator). The proliferation of responder cells was quantified 4 days later using FCM. Proliferation of maternal human splenic CD4+ effector T cells after stimulation with paternally (BALB/c) derived antigen reflects the presence of fetal reactive human maternal T cells (Figure 5A). STreg cells were capable of suppressing Tconv cells at 1:2 ratio (Treg/Tconv) (Supplemental Figure 11). However, under a low Treg/Tconv ratio (1:16) condition, the maternal versus fetal T cell reactions (MvFRs) of maternal T cells were significantly inhibited by dTreg but not by sTreg cells (or were only minimally suppressed; Figure 5A).

Inhibition of human Treg cells causes pregnancy termination and fetal rejection in pregnant HIS mice. Treg cell inhibitory function examination. (A) Summarized data (n = 3; mean ± SEM) of the immune-inhibitory effect of human dTreg or sTreg cells on human sTconv cell proliferation in the presence of 30 Gy–irradiated BALB/c splenic cells. Representative FCM profiles of Tconv cell proliferation. Treg/Tconv = 1:16, BALB/c stimulator cells/Tconv = 1:2. Statistical differences were determined with 1-way ANOVA for multiple-variable comparisons. (B) Schematic outline of the experimental design. Anti-CD25 mAb (basiliximab; n = 6) or equal volume of PBS (n = 5) was i.p. injected into pregnant HIS mice (200 μg/mouse) made by mating with BALB/c males at E2.5 and E5.5, and the mice were euthanized at E14.5. Percentages (mean ± SEM, C) and representative FCM profiles (D) of Treg cells after anti-CD25 mAb injection were shown. Statistical analyses were performed using unpaired t test. (E) The number of implantation sites of pregnant HIS mice treated with PBS (n = 5) or basiliximab (n = 6). Statistical differences were calculated with 2-way ANOVA. (F) Representative photos of the uteri harvested from HIS mice treated with basiliximab (upper panel) or PBS (lower panel).

We then evaluated the roles of human Treg cells in maintaining immune tolerance in vivo during pregnancy. Basiliximab is a monoclonal antibody of the IgG1κ subtype that specifically targets CD25 (the α subunit of the IL-2 receptor) and can inhibit the interaction of IL-2 with Treg cells, causing IL-2 starving (57). When administered to humans, basiliximab causes a significant reduction in the number of circulating Treg cells (58, 59). Using a conventional pregnant mouse model, it has been shown that Treg cells are necessary for maintaining materno-fetal immune tolerance during early stages, but not late stages, of pregnancy (60, 61). Pregnant HIS mice were treated at E2.5 and E5.5 with pretitrated basiliximab (Supplemental Figure 12) or with the same volume of PBS as controls and euthanized at E14.5 for examination (Figure 5B). The number of human CD25+Foxp3+CD4+ Treg cells was markedly reduced in the PBMCs, spleen, and decidua (Figure 5, C and D). Severe embryo resorption was observed in basiliximab-treated mice but not in control mice (Figure 5, E and F). Moreover, markedly elevated levels of inflammatory mediators (such as TNF-α, IFN-γ, IL-17A, granzyme B, perforin, and IL-4), but not antiinflammatory molecules (such as IL-10), were detected in the sera of basiliximab-treated mice, but not control mice (Figure 6A). The majority of these inflammatory mediators (except IFN-γ) did not increase in nonpregnant HIS mice after basiliximab treatment (Figure 6A). This finding indicated that elevation of these inflammatory cytokines is attributed to materno-fetal immunological rejection. Moreover, histological examination of the embryonic resorption site demonstrated severe tissue damage and human T cell infiltration in the embryo of basiliximab-treated mice but not in control mice (Figure 6, B and C). These results verified that the human materno-fetal interface immunity in pregnant HIS mice was functional and capable of developing MvFRs and rejecting fetuses after inhibition of Treg cell activity.

Inhibition of human Treg cells causes an inflammatory response in pregnant HIS mice. (A) Concentration (mean ± SEM) of human TNF- α, IFN-γ, IL-17A, granzyme B, perforin, IL-10, IL-4, granzyme A, granulysin, IL-6, and IL-2 in mouse serum. Statistical differences were determined with 1-way ANOVA for multiple-variable comparisons. NP, nonpregnant HIS mice. (B) Representative H&E images of the decidua in PBS group (left) and the resorbed embryos collected from the basiliximab treatment group (right). D, decidua; Jz, junctional zone; L, labyrinth zone. (C) IHC examination of the placenta (PBS) or uterus (basiliximab treatment) section stained with anti-human CD45 and CD4 antibodies. Scale bars for each picture are shown at lower right corner.