Twenty-seven patients with GDM and 27 normal pregnant women were recruited from Chungnam National University Hospital. All patients with GDM were diagnosed according to the American Diabetes Association guidelines.23 Additionally, we used the following enrolment criteria: singleton, age < 40, and not complicated by other diseases. Pregnant women with body mass index (BMI > 30 kg/m²) or those undergoing insulin treatment were excluded. There was no participant with gestational hypertension, preeclampsia, proteinuria, or history of polycystic ovary syndrome. Nine cases (six GDM and three controls) were used for RNA-seq, and other cases were used for validation experiments for FACS analysis of placental mononuclear cells (MNCs), immunohistochemical analysis, real-time polymerase chain reaction (PCR), and western blot analysis. Placenta tissue was collected from the subjects during cesarean delivery. Each of the four placenta quadrants was sampled approximately 1.5 cm away from the umbilical cord attachment from the fetal side of the placenta. Fetal membranes and visible large vessels were removed, and phosphate-buffered saline (PBS) was used to wash placenta samples before separating into maternal- and fetal-side samples.

First, the maternal placentas were collected en bloc as 1 × 1 cm pieces for transcriptomic analysis. Then, the tissue samples were homogenized using a mortar and pestle, and total RNA was extracted using the easy-spin Total RNA Extraction Kit (iNtRON, Seongnam, Korea) following the manufacturer’s protocol. All experiments were then conducted under clean conditions, and the equipment was pre-autoclaved. The quality of the extracted RNA was evaluated using an Agilent 2100 Bioanalyzer RNA Nano Chip (Agilent Technologies Inc., Santa Clara, CA, USA). The extracted RNA was used to construct RNA libraries using the TruSeq Stranded mRNA Sample Prep Kit v2 (Illumina, San Diego, CA, USA), according to the manufacturer’s protocol. The quality was analyzed using an Agilent 2100 Bioanalyzer and an Agilent DNA 1000 kit. Finally, the samples were sequenced using an Illumina HiSeq2500, which yielded an average of 38 million paired-end 100 nucleotide reads.

The reads were aligned to the UCSC Homo sapiens reference genome (GRCh37/hg19) using TopHat2 v2.1.5. The default TopHat parameter options were used. To analyze the differentially expressed gene (DEG) profiles between the compared groups (Control vs. GDM), the Tuxedo protocol was used.24 The aligned reads were processed using Cufflinks v2.2.1, which is based on the fragments per kilobase of exon model per million reads mapped (FPKM), and unbiased, normalized RNA-seq fragment counts were used to analyze the relative transcript levels. Gene transfer format (GTF) files were generated to quantitatively compare the transcript levels in each sample to those in a reference GTF file. Next, we used Cuffdiff to calculate the differences in the FPKMs between each group. Heat maps were generated using PermutMatrix Version 1.9.3 (LIRMM, Montpellier, France).

The GSE154377 dataset was analyzed using GEO RNA-seq Experiments Interactive Navigator (GREIN, http://www.ilincs.org/apps/grein/). The counts table was normalized to the gene level and downloaded to draw Supplementary Fig. 1. Differential expression analysis was performed using GREIN to compare gene expressions of GDM vs. control.

First, the placentas were collected at the time of delivery and immediately washed with isotonic sodium chloride solution. The placental villi were scraped from the chorionic membrane using a scalpel. Single-cell suspensions were obtained by homogenizing placental samples in Dulbecco’s phosphate-buffered saline (DPBS) (Welgene, Daegu, Korea), and digested with 5 mL of RPMI digestion solution in a gentleMACS dissociator (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s protocol, and then filtered using a 70 μm nylon mesh (BD Falcon, Frankin Lakes, NJ, USA). Placental MNCs were isolated by centrifugation on a Ficoll-Paque density gradient (GE Healthcare Life Science, Buckinghamshire, UK) at room temperature. After centrifugation, the MNC layer was collected and washed with DPBS (Welgene). The isolated MNCs were resuspended in DPBS (Welgene) containing 0.5% BSA and 0.05% sodium azide. Trypan blue dye exclusion testing was used to determine the number of viable cells in the suspension, which was used for flow cytometry analysis.

As described previously,25 placental MNCs were incubated with fluorochrome-conjugated monoclonal antibodies for 40 minutes at 4°C. MNCs were pre-incubated with an anti-human BD Fc blocker (BD Pharmingen, San Diego, CA, USA), followed by staining with the live/dead marker anti-FVD-APC-Cy7 (eBioscience, San Diego, CA, USA). The antibodies used in this study were anti-CD3-PerCP-Cy5.5, anti-CD3-PE-Cy7, anti-CD4-AF700, anti-CD8-PE, anti-CD8-APC, anti-CD28-APC, anti-CD45RA-FITC, anti-CD45RO-PE-Cy7, anti-CD57-FITC, and fixable viability dye-APC-Cy7 (all supplied by eBioscience). MNCs were stimulated with phorbol-myristate acetate/ionomycin/brefeldin A/monensin for 5 hours. Cells were fixed and permeabilized using a Fixation/Permeabilization Buffer kit (eBioscience). The permeabilized cells were washed and resuspended in 1% formaldehyde and further stained for intracellular cytokines with anti-interferon gamma (IFN-γ)-PE-Cy7 and anti-IL-17A-APC. Multicolor flow cytometry was performed using a BD LSRFortessa flow cytometer (BD Biosciences, San Jose, CA, USA), and data were analyzed using FlowJo software (FlowJo, LLC, Ashland, OR, USA).

PD-1 and PD-L1 immunohistochemical staining was performed on the placental tissues of 68 patients, including control and patients with GDM. Whole tissue sections were cut from representative placental paraffin blocks using a microtome and mounted onto coated slides, which were then transferred to Ventana (Tucson, AZ, USA) and Dako (Glostrup, Denmark) automated immunostainers. Staining was performed according to the manufacturer’s protocol using anti-PD-L1 (Ready-to-Use, clone SP263; Ventana) and anti-PD-1 (NAT105, 1:100; Cell Marque, Rocklin, CA, USA) antibodies. Tonsil tissues were used as the controls. The immunostaining pattern of PD-L1 in the placenta was diffusely expressed on the outer surface of syncytiotrophoblasts of the chorionic villi. The cytotrophoblastic cells and villous stroma were negative, and the intermediate trophoblastic cells were very weak. The pattern of PD-L1 expression was membranous, without cytoplasmic or nuclear expression. PD-L1 was expressed diffusely on the syncytiotrophoblasts of the placenta, and only the intensity of the immunostaining was scored and categorized as weak, moderate, or marked. PD-1 immunohistochemistry (IHC) in the same placentas was evaluated and expressed on immune cells. PD-1 positive immune cells were scattered at the implantation sites of the placenta. The number of positive cells was counted per 50 high-power fields. PD-1 scores were categorized as “Negative (positive cells/50 HPF < 1), Weak (positive cells low (1 ≤ positive cells/50 HPF < 5) and Moderate (5 ≤ positive cells/50 HPF).”

Real time PCR was performed as described previously.26

The primer sequences were: PDCD1-F: 5′-CGTGGCCTATCCACTCCTCA-3′, PDCD1-R: 5′-ATCCCTTGTCCCAGCCACTC-3′; PD-L1-F: 5′-AAATGGAACCTGGCGAAAGC-3′, PD-L1-R: 5′-GATGAGCCCCTCAGGCATTT-3′; PD-L2-F: 5′-GGACCCACAGAATTCCTTCA-3′, PD-L2-R: 5′-GGCAACCCATCTGTTTCTGT-3′; CTLA-4-F: 5′-CAGGGAAGTTTTGTGGAGGA-3′, CTLA-4-R: 5′-CCAAAGCACATGTCAACACC-3′; IFN-γ-F: 5′-AGCTCTGCATCGTTTTGGGT-3′, IFN-γ-R: 5′-CGCTACATCTGAATGACCTGC-3′.

Cells were lysed in buffer containing 150 mM NaCl, 1.0% Nonidet-P 40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate, 50 mM Tris, pH 8.0, and a protease inhibitor cocktail (Roche Applied Science, Vienna, Austria). Electrophoresis was performed as described previously.27 The following primary antibodies were used for western blot analysis: anti-phospho-STAT1 (S727) (1:1,000; Cell Signaling Technology, Danvers, MA, USA), anti-total STAT1 (1:1,000; Cell Signaling Technology), anti-phospho-STAT3 (S727) (1:1,000; Cell Signaling Technology), anti-phospho-STAT3 (T705) (1:1,000; Cell Signaling Technology), anti-total STAT3 (1:1,000; Cell Signaling Technology), anti-phospho-PI3K-p85 (1:1,000; Cell Signaling Technology), anti-PI3K-p85 (1:1,000; Cell Signaling Technology), and anti-β-actin (1:1,000; Cell Signaling Technology). Following incubation with the corresponding horseradish peroxidase-conjugated secondary antibodies (1:1,000; Santa Cruz Biotechnology, Dallas, TX, USA), immunoreactive bands were visualized using enhanced chemiluminescence detection. software (Bio-Rad, Hercules, CA, USA).

The results were expressed as mean ± standard deviation (SD). Chi-square and paired t-tests were used to compare clinicopathological data. P value (< 0.05) were used to select genes differentially expressed between the two groups. All in vitro experiments were repeated three times, and statistical significance was analyzed using the two-tailed Student’s t-test or one-way analysis of variance followed by Tukey’s post hoc test. Data are expressed as mean ± SD, and a value of P < 0.05 was deemed statistically significant (P < 0.05). SPSS statistical software for Windows, version 20 (SPSS, Chicago, IL, USA) was used for all statistical analyses.

The study protocol was reviewed and approved by the Institutional Review Board of Chungnam National University Hospital (approval No. 2020-08-025). Written informed consent was obtained from all participants when they were enrolled.

Transcriptomics analysis were performed in 6 GDM patients and 3 control participants. The clinical information of the participants enrolled in the study is described in Fig. 1A. Compared to the control group, we identified 758 significantly differentially expressed genes (DEGs) (P value < 0.05) between GDM placentas and controls (Supplementary Table 1). Volcano plot analysis revealed the significant DEGs with fold change ≥ 1.5 and P value < 0.05, such as SCN3A, ACTC1, ASBS2, APOB, and CXCL9 (Fig. 1B). The heatmap analysis revealed the distinct gene expression patterns between two groups (Fig. 1C). Gene ontology analysis based on DEGs revealed the upregulation of diverse mitochondria related pathways, including oxidative phosphorylation, ATP synthesis, and oxygen transport (Fig. 1D). Interestingly, the placenta of GDM exhibited down regulation of various immune related signaling pathways, including regulation of lymphocyte activation, immune effector process, T cell differentiation, and immune response signaling pathway. Collectively, these data suggest that altered T cell immunity related with GDM in pregnant women.

To identify the immunophenotype of T cells in the placenta of women with GDM, we first investigated the frequency and activation status of T cells from the pregnant women’s placenta with or without GDM. Six patients with GDM and eight controls were included in the FACS analysis. Various clinical parameters of participants are described in Table 1. The GDM group demonstrated higher pre-pregnancy BMI and BMI at delivery, compared to controls; however, there was no significant difference between two groups. Four patients among the GDM group (66.7%) revealed the post-partum complications, whereas, no such complication were encountered in the control group (Table 1). Of the four patients who experienced complications, one patient experienced both maternal and fetal complications, including preterm labor and premature fetus; two patients experienced maternal complications; and one patient experienced a fetal complication of congenital genitourinary tract anomaly (Table 1).

Placental MNCs from normal controls and patients with GDM were first gated single cells and lymphocytes, and the lymphocyte population was then further analyzed for uptake of a fixable viability dye. To investigate the phenotypic characterization of CD4⁺ T cells, we evaluated the frequency of CD45RA⁺CD45RO⁻ and CD45RA⁻CD45RO⁺ among CD4⁺ T cells in the placental MNCs from normal controls and patients with GDM. The naïve CD4⁺ T cell population, CD45RA⁺CD45RO⁻, tended to decrease, along with an increased population of memory CD4⁺ T cells and CD45RA^-CD45RO⁺ CD4⁺ T cells in patients with GDM (Fig. 2A). The population of IFN-γ⁺CD4⁺ T cells was also significantly higher in patients with GDM (Fig. 2B). We also analyzed the population of Th17 T cells using IL-17A⁺CD4⁺ T cells between the GDM and control groups. There was a significant increase in the population of IL-17A⁺CD4⁺ T cells in the placental MNCs of patients with GDM (Fig. 2C). These data suggest that patients with GDM have a larger population of memory CD4⁺ T cells and proinflammatory cytokine-producing CD4⁺ T cells in the maternal placenta.

Next, we analyzed the phenotypic characteristics of CD8⁺ T cells. The frequency of CD45RA⁺CD45RO⁻ and CD45RA⁻CD45RO⁺ among CD8⁺ T cells and the IFN-γ⁺ expression in CD8⁺ T cells in the placental MNCs from normal controls and patients with GDM were investigated. Comparison of the phenotype CD4⁺ T cells in the placenta between GDM and controls showed that the naïve CD8 T cell population, CD45RA⁺CD45RO⁻, tended to be decreased, along with the increased population of memory CD8⁺ T and CD45RA⁻CD45RO⁺ CD8⁺ T cells in patients with GDM (Fig. 3A). The expression of IFN-γ in CD8⁺ T cells was significantly higher in pregnant women with GDM than in controls. These findings indicate a larger quantity of memory CD8⁺ T cells and proinflammatory cytokine-producing CD8⁺ T cells in the placenta (Fig. 3B).

A previous study reported that senescent T cells have the ability to secrete abundant proinflammatory cytokines in diverse metabolic diseases.28 Moreover, our transcriptomic analysis revealed upregulation of T cell exhaustion signaling in GDM. For these reasons, we investigated the functional phenotypic changes in senescent CD4⁺ and CD8⁺ T cells from the placenta of patients with GDM and controls. Among CD4⁺ T cells, the CD28⁻CD57⁺ senescent population tended to be larger in patients with GDM, but this difference was not statistically significant (Fig. 4A). The population of CD28⁻CD57⁺CD8⁺ T cells and senescent CD8⁺ T cells were significantly increased in the placenta of the GDM group (Fig. 4B). We also analyzed the population of CD279 (PD-1)-positive senescent T cells, since PD-1 is precise molecules related immune-escape signaling. Although the CD279⁺CD28^-CD57⁺ population tended to be larger in the GDM group without statistical significance (Fig. 4C), CD279 (PD-1) was predominantly expressed in the placenta of patients with GDM in senescent CD8⁺ T cells (Fig. 4D), indicating that there are more immunosenescent CD8⁺ T cells, which predominantly express CD279 (PD-1) in the GDM placenta group than in the control group. To validate these results in RNAseq, we analyzed the expression of PD-1, PD-L1, and PD-L2 in patients with GDM and controls; however, there was no expression value for PD-1 and no significant difference between PD-L1 and PD-L2 (Supplementary Fig. 1A). Notably, in GSE154377, a study focusing on the role of the placenta in GDM, PD-L1 expression decreased significantly in the first trimester of pregnancy and tended to increase in the third trimester (Supplementary Fig. 1B). These results suggest that immune-escape signaling has different effects on the placenta in early and late pregnancy and also provides a potential explanation for the increase in PD-1 in senescent T cells in this study.

Accumulating evidence suggests the significance of immunological tolerance in the placenta in GDM pathophysiology29; however, the precise role of PD-1/PD-L1 signaling has not been clearly demonstrated. Since we first identified the upregulated PD-1 expression in senescent T cells in the maternal placenta of patients with GDM, we validated the expression of PD-1/PD-L1 in maternal placentas using immunohistochemical analysis. A total of 61 participants were enrolled in the IHC analysis (Table 2). The GDM group demonstrated significantly more increased gestational weight gain and decreased gestational age, compared to the control group (Table 2). As results of the FACS analysis, postpartum complications were significantly increased in patients with GDM, compared to the control group. PD-L1 immunoreactivity was significantly detected in the syncytiotrophoblasts (Fig. 5A). PD-L1 immunoreactivity was significantly more intense in the syncytiotrophoblasts of chorionic villi in the placenta of patients with GDM (Fig. 5A, Supplementary Fig. 2A). PD-1 expression was detected in some decidua lymphocytes (Fig. 5B, Supplementary Fig. 2B). We defined the absence of PD-1 positive cells as negative, the presence of 1 to 4 cells as weak, and the presence of 5 or more cells as moderate in the 50 high-power field. PD-1 expression showed a tendency to increase in GDM group compared to in the control group (Fig. 5B). These findings indicate that the higher intensity of PD-L1 expression is consistent with increased PD-1⁺ senescent CD8⁺ T cells in the maternal placenta of patients with GDM. Moreover, the mRNA expression of PD-1, PD-L1, and CTLA-4 was significantly increased in placentas of patients with GDM (Fig. 5C). Since we hypothesized that the PD-1/PD-L1 signaling in the placenta is dysregulated in GDM based on our transcriptomics and FACS data, we investigated STAT signaling, a known regulator of PD-L1 expression in other tissues. There were more patients with increased phosphorylation of STAT1, phosphorylation of STAT3, and phosphorylation of PI3K in placentas from patients with GDM, even though the expression patterns of PD-1-PD-L1-STAT-PI3K were not consistent and there were differences between individuals (Fig. 5D, Supplementary Fig. 2C). These data suggest that immuno-escape signaling is dysregulated in the placenta of patients with GDM, although the interrelationship and mechanism could not be elucidated.