Synthesis of unlabeled OLA and FITC-labeled OLA

Pure OLA, with a molecular weight ~1,000 Da, was provided by Shanghai Sur-Release Biotech (Shanghai, China). FITC-conjugated OLA were produced by reacting FITC to the primary amine-terminated PLA oligomers in anhydrous dimethyl sulfoxide based on our previous research [13]. The detailed procedure is provided in the S1 Text.

In vivo challenge with PLA oligomers

Adult (8-week) CD-1 mice were purchased from Beijing Vital River Laboratory Animal Technology Co. (Beijing, China), and fed in a specific-pathogen-free (SPF) grade animal facility with ad libitum food and water. To examine the effect of environmentally relevant doses of unlabeled PLA oligomers, pregnant mice were divided into 3 dose groups randomly (N = 12 dams/group): control [corn oil (Aladdin, China, C116025)], and low and high-OLA exposure group (0.01 mg/kg/day and 0.1 mg/kg/day OLA in corn oil, respectively). OLA was administered by oral gavage at 4 μL/g body weight in the morning. The experimental dosage was computed in accordance with the actual weekly consumption of PLA at 0.1–5.0 g by humans [14]. In this study, FITC-labeled OLA was used exclusively to trace biodistribution following gestational exposure, whereas liquid chromatography-tandem mass spectrometry (LC-MS/MS) quantification and developmental toxicity were assessed in dams administered unlabeled OLA. Experimental animals were handled with care and body weight and food/water consumption were measured daily. All animal experiments were conducted in accordance with the Guidelines for the Ethical Review of Laboratory Animal Welfare, People's Republic of China National Standard GB/T 35892‐2018. The study protocol was approved by the Anhui Medical University Animal Care and Use Committee (Approval No. LLSC20221266, Hefei, China). All efforts were made to minimize animal suffering and reduce the number of animals used.

Pregnant dams were sacrificed at different stages of pregnancy, fetal weight and body length, placental weight and diameter were measured by the same technician to minimize variation. Maternal blood, placenta, and fetal serum were collected for experiments. Fig 1A was the overall study design.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Gestational exposure to environmentally relevant dose of unlabeled oligomeric lactic acid (OLA) passes through placenta and accumulates in maternal and fetal organs.

(A) Overall study design, pregnant dam of the control group were administered with corn oil, the exposure groups received daily doses of 0.01 and 0.1 mg/kg/day of OLA administered in corn oil (N = 12 dams per group). (B) Profiling the distribution of OLA following gestational exposure to FITC-labeled OLA using the IVIS Spectrum in utero, including signals from the uterus with fetuses, the placenta, and the fetuses. Dams of the control and OLA exposure group (0.1 mg/kg/day) were sacrificed on gestational day (GD) 14, 16, and 18, where the blue and orange dashed square framed the maternal and fetal side of placenta, respectively (N = 3 dams per group). (C) In vivo OLA-FITC fluorescence density distribution of fetal brain, liver, and kidney at GD18 (N = 3 dams per group). (D and E) Relative fetal OLA abundance measurement by mass spectrometry (MS) at GD18 (N = 5 dams per group). Data are expressed as mean ± SEM. Data for this figure is presented in S1 Data folder. Schematic (A) was created in Figdraw.

https://doi.org/10.1371/journal.pbio.3003676.g001

In vivo OLA tracing

FITC-labeled OLA biofluorescence imaging: between gestational day (GD) 0–17, pregnant dams were administered with FITC-OLA dissolved in corn oil by oral gavage at 4 μL/g body weight. At GD 14, 16, and 18, dams were first anesthetized with 2.5% Isofluran (Forene, Abvie, Ludwigshafen, Germany) for in vivo FITC-labeled OLA measurement by IVIS Spectrum (PerkinElmer, USA), and then sacrificed by cervical dislocation for quantitative analysis of OLA in major maternal and fetal organs, including uterine with fetuses, placentas, fetuses, fetal brain, liver, spleen, kidneys, small intestine, and colon.

Determination of OLA concentration by LC-MS/MS: To verify the fluorescence signal of OLA, pregnant dams were also administered with unlabeled OLA during gestation, the concentration of which was measured in fetal liver by liquid chromatography mass spectrometry as described below. Fetal liver (50 mg) was homogenized in saline, and 200 µL of the homogenate was used for extraction. Following addition of 400 µL of acetonitrile (CAS: 75-05-8, LC-MS grade, Merck, USA), samples were centrifuged at 15,000 × g for 30 min at 4 °C and then placed at −20 °C freezer for 30 min to overnight to facilitate protein precipitation. After centrifugation at the same conditions described above, the supernatant was collected and 200 µL of ethyl acetate containing 0.1% formic acid was added, then the upper organic layer was separated and evaporated to dryness under a gentle stream of nitrogen in a water bath at 30 °C. We successfully synthesized OLA standards (n = 5, 6, and 7) with 95% purity, following our previous study [25], purified oligomers are with well-defined chain lengths. Samples were reconstituted in 50 µL of acetonitrile and analyzed on 1,290 Infinity II LC system coupled with 6,495 triple quadrupole mass spectrometer (Agilent, Santa Clara, CA, USA) and C18 column (50 mm × 2.1 mm, 1.8 µm, Agilent, CA, USA). A typical gradient using water containing 0.1% formic acid (A) and acetonitrile (B) was as follows (regards to B): 0–2.0 min, maintain 5% B; 2.0–12.0 min, 5%−100% B; 12.0–16.0 min, maintain 100% B; 16.0–18.0 min, 100%−5% B; 18.0–21.0 min, maintain 5% B.

RNA sequencing of mouse placentas

In each of the control and OLA exposure groups, RNA-seq was initially performed on eight placentas per group, balanced for fetal sex. One control sample failed library quality control, resulting in a final dataset of n = 7 controls and n = 8 OLA-H placentas. To ensure equal group sizes and balanced sex representation, six placentas (three male and three female) were ultimately selected from each group for RNA sequencing. Placental tissue (50 mg) was homogenized in Invitrogen TRIzol Reagent (15596026, Thermo Fisher Scientific, USA) and total RNA was extracted according to the manufacturer's instructions. Reverse transcription of RNA and library construction were performed using Truseq RNA sample prep Kit (FC-122–1001, Illumina, USA) according to the manufacturer's instructions. The DESeq2 v1.40.2 software was used to conduct differentially expressed gene (DEGs) analysis [26] with a standard of |Fold change| ≥ 1.2 and false discovery rates (FDR) < 0.05. Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) and Reactome analysis were analyzed by Goatools v1.3.1 [27], KOBAS v3.0.0 [28], and ReactomePA v1.44.0 software [29] with FDR value < 0.05 significantly enriched. Gene set enrichment analysis (GSEA) was performed by GSEA v4.1.0 [30,31]. The pathways with |normalized enrichment score (NES)| > 1 and FDR < 0.25 in GSEA were significantly enriched.

Single-cell RNA sequencing of mouse placentas

Mouse placentas were perfused under anesthesia to remove red blood cells and immediately placed in tissue preservation buffer. Single-cell suspensions were prepared, and only samples with cell viability ≥80% were used. Endothelial cells were enriched via Fluorescence-Activated Cell Sorting (FACS) using CD31⁺/CD45⁻ markers (eBioscience 30086, FITC Anti-Mouse CD31 Antibody, Invitrogen, USA) and subsequently mixed with the original cell suspension in 1:1 ratio. The final cell concentration was adjusted to 700–1,200 cells/μL for single-cell capture using the 10x Genomics Chromium platform. Individual cells were encapsulated with barcoded Gel Beads in Emulsion (GEMs), followed by cell lysis, mRNA release, reverse transcription, cDNA amplification, and library construction. Libraries were sequenced on an Illumina platform (FC-122-1001, Illumina, USA). Raw sequencing data were processed with Cell Ranger (v9.0.1) for alignment and gene quantification. Downstream analyses were performed using Seurat (v4.4.0) with quality control filters (≥500 genes per cell, mitochondrial gene content <15%) and DoubletFinder to remove doublets. Highly variable genes were identified for principal components, and the first 25 components were used for graph-based clustering. Clusters were visualized using uniform manifold approximation and projection (UMAP) and t-distributed stochastic neighbor embedding (tSNE). Marker genes for each cluster were identified using the Wilcoxon rank-sum test (Log₂^FC > 0.26, expression >10%, FDR < 0.01) and subjected to GO, KEGG, and Reactome enrichment analyses. Cell types were annotated with SingleR (v1.10.0) and manually curated based on canonical marker genes. Differential expression analysis between OLA-treated and control groups was performed on endothelial cell subpopulations using thresholds of |Log₂^FC| > 0.26, expression >10% and FDR < 0.05, followed by pathway enrichment to identify key OLA-responsive biological processes.

Simulation of protein 3D structure

Since the complete 3D structure of GATA2 is unavailable, we employed AlphaFold2 (ver. 2.3.1) and ColabFold (ver. 1.5.2 compatible with AlphaFold 2.3.1) for structural prediction [32,33]. The amino acid sequence of GATA2 (UniProt P23769) was utilized to predict its 3D structure. Multiple sequence alignment (MSA) was generated using JackHMMer and HHBlits through AlphaFold2 and provided to ColabFold to generate 100 structural models (20 random seeds, with each seed generating five models) [34,35]. Templates were not used in the structural predictions by ColabFold. For MSA subsampling, we set the number of sequence clusters (max_msa_clusters) to 512 and the number of extra sequences (max_extra_msa) to 1,024. The number of prediction recycles was set to three, and structural relaxation after prediction was not performed. Additionally, we utilized ColabFold to generate MSA and predict the structure of the transporter OxlT, where MSA generation was executed using MMseqs2 instead of JackHMMer and HHBlits, to investigate the effect of different MSA methods on structural prediction. Inter-residue coevolution coupling within the MSA was analyzed using pseudo-likelihood maximization direct coupling analysis (plmDCA) with the software pydca [36–38]. To calculate the coevolution score, we employed the Frobenius norm of the inter-residue couplings and applied average product correction.

Molecular docking

The predicted structures of GATA2 obtained from AlphaFold2 were utilized for molecular docking studies. The initial protein structures were prepared using AutoDockTools version 1.5.6, ensuring the maintenance of the original protein charges, and pdbqt files were generated for docking purposes [39]. Small molecules comprising 2–9 OLA units were individually optimized for their molecular structures using the MOPAC program, and AM1 atomic charges were calculated to facilitate subsequent molecular docking. Following optimization, the ligand structures were processed with AutoDockTools version 1.5.6 to generate the corresponding pdbqt files required for docking simulations.

Molecular dynamics simulation

The protein-ligand complex conformations obtained from molecular docking were utilized to construct the dynamic system. Molecular dynamics simulations were performed using GROMACS version 5.1.4 with the AMBER15 force field [40,41]. Each complex was placed in a standard cubic box and neutralized with counterions to mimic physiological conditions. Initially, the protein was restrained to prevent large conformational changes during solvent energy optimization and to eliminate steric clashes. Subsequently, positional restraints were applied to the protein backbone to stabilize the structure during equilibration. Finally, an unrestrained simulation of 10 picoseconds (ps) was conducted to observe the dynamic behavior of the protein-small molecule complex. Long-range van der Waals interactions were calculated using a cutoff distance of 0.8 nm with the cutoff method for classical interactions.

Dual-luciferase reporter assay

To assess whether GATA2 regulates VEGFR2 transcription, the human VEGFR2 promoter was cloned into the pGL3-basic vector (E1751, Promega, USA), and full-length GATA2 cDNA was inserted into the pcDNA3.1 expression vector (V79020, Invitrogen, USA). All constructs were sequence-verified. HEK293T cells were cultured in DMEM with 10% FBS (CGM101.05, Cellmax, USA) and co-transfected with VEGFR2-WT, GATA2-pcDNA3.1, and the Renilla control plasmid pRL-TK (E2241, Promega, USA) using Lipofectamine 2000 (11668019, Invitrogen, USA). After 48 h, Firefly and Renilla signals were quantified using a Dual-Luciferase Reporter Assay Kit (JKR23008, Genekai, China). Experiments were independently repeated five times, and data are presented as mean ± SD.

Surface plasmon resonance (SPR)

SPR was used to quantify the binding interaction between GATA2 and OLA. Experiments were performed on a CM5 sensor chip (Cytiva, BR100530, USA) using standard amine-coupling chemistry with EDC/NHS (Cytiva, BR100050) and ethanolamine (Cytiva, BR100086). Recombinant human GATA2 protein (20 μg/mL in 10 mM sodium acetate buffer, pH 4.5, CSB-EP340572HU, CUSABIO) was immobilized at 10 μL/min for 420 s, yielding ~12,600 response units. A reference flow cell without immobilized protein was included for background subtraction. OLA was dissolved in DMSO and diluted in running buffer (PBS supplemented with 1% DMSO and 0.005% Tween-20) to 0–500 μM. Analytes were injected at 20 μL/min with 180 s association and 120 s dissociation phases. Surfaces were regenerated using 10 mM glycine-HCl (pH 2.0; BR100354, Cytiva, USA) at 200 μL/min for 30 s. All experiments were performed at 22 °C. Kinetic parameters (kₐ, kd, and KD) were derived using multi-cycle kinetic fitting after reference subtraction.

Cellular thermal shift assay (CETSA)

CETSA was conducted to verify OLA-induced thermal stabilization of endogenous and overexpressed GATA2 in endothelial cells. Human umbilical vein endothelial cell (HUVEC) and primary human placental vascular endothelial cells (HPVEC) were treated with OLA (500 μM) or vehicle control for 2 h. Cells were harvested and aliquoted into PCR tubes, heated at 10 temperature points (37−65 °C) for 3 min using a thermocycler, and cooled to room temperature for 3 min. Samples were lysed by freeze-thaw cycles and centrifuged at 20,000 × g for 20 min at 4 °C. Soluble protein fractions were analyzed by SDS-PAGE and immunoblotting using anti-GATA2 antibodies (11103-1-AP, Proteintech, China). Melting curves were generated to evaluate OLA-induced protein stabilization.

Quantitative real-time PCR (qRT-PCR) analysis, western blotting (WB), Hematoxylin-eosin (H&E) staining, and immunofluorescence staining

The detailed procedures are provided in the S1 Text.

Cell culture

HUVEC was purchased from the American Type Culture Collection (PCS-100-010, USA). HUVEC and HPVEC were cultured in ECM medium (1001, ScienCell, USA) and supplemented with 10% fetal bovine serum (0025, ScienCell, USA), 1% endothelial cell growth supplement (1052, ECGS, ScienCell, USA) and 1% penicillin/streptomycin solution (0503, ScienCell, USA). To determine the dose of OLA that impairs HUVEC and HPVEC growth, cells were exposed to OLA at different concentrations for 24 h. Then, the cell viability experiment was detected by the CCK8 assay. To investigate the effects of OLA exposure on GATA2 expression level, HUVEC and HPVEC were exposed to OLA for 0, 3, 6, 12, and 24 h. To investigate the effect of gata2 siR on OLA-downregulated GATA2 expression, cells were exposed to OLA for 24 h after GATA2 siR treatment. To investigate the effect of lentivirus vector HBLV-h-GATA2–3xflag-ZsGreen-PURO on OLA-downregulated GATA2 expression, cells were exposed to OLA for 24 h after lentivirus vector HBLV-h- Gata2–3xflag-ZsGreen-PURO treatment.

Cell viability

CCK8 assay (BS350A, Biosharp, Anhui, China) was used to determine the effect of OLA on cell viability. HUVEC and HPVEC were seeded into 96-well plates with 4 × 10³ cells per well and 6 replicate wells for each group. Cells were treated with 0, 25, 50, 100, 200, and 400 ng/mL OLA containing 0.1% DMSO (CAS No.67-68-5, HY-Y0320 MedChemExpress, USA, ≥ 99.0%). The new medium was replaced after treatment for 24 h, and 10 μL CCK8 solution (C0048S, Beyotime, China) was added to each well according to standard steps. The absorbance of each group was detected at 450 nm in a microplate reader. The cell viability of the treated cells was calculated relative to that of the cells in control group. All values are presented as the Mean ±SEM of at least six samples.

Interventions of gene expression

Gene silencing: RNA interferences were performed following the approach of our previous study [42]. Briefly, GATA2 specific small interference RNA (siR, Hanbio Biotechnology Co., https://www.hanbio.net, Shanghai, China) was mixed with Lipofectamine 3,000 for 15 min in serum-free medium. The mixtures were added to the culture medium for 6 h to transfect HUVEC and HPVEC. After this, the medium was replaced with fresh medium and cells were cultured for 24 h. The sequence of GATA2 siRNA was 5′-CCUGUGGCCUCUACUACAATT-3′ (forward) and 5′-UUGUAGUAGUAGAGGCCACAGGTT -3′ (reverse). To investigate the effect of GATA2 siR on OLA-downregulated GATA2 expression, cells were exposure to OLA for 24 h after GATA2 siR treatment.

Gene overexpression: polymerase reaction was used to amplify GATA2-human gene. Amplified-GATA2 products were inserted into lentivirus vector pHBLV. Then we constructed the lentivirus vector pHBLV-GATA2. Agarose gel electrophoresis and DNA sequencing were used to verify the lentivirus vector pHBLV-GATA2. Lentivirus vector pHBLV-GATA2, helper plasmid psPAX2 vector and pMD2G vector were co-transfected into 293 T cells by Lipofiter transfection reagent (Hanbio Biotechnology Co., Shanghai, China). After 48 h and 72 h of co-transfection, the virus supernatant was obtained at 4 °C, 82,700g for 120 min. It was the Gata2 overexpression lentivirus HBLV-h-GATA2-3xflag-ZsGreen-PURO. HBLV-h-GATA2-3xflag-ZsGreen-PURO transfection was performed following the approach of a previous study [43]. Briefly, HBLV-h-GATA2-3xflag-ZsGreen-PURO (LV93071705, Hanbio Biotechnology Co., Shanghai, China) were mixed with polybrene (20210710, Hanbio Biotechnology Co., Shanghai, China). Then, mixtures were added to the medium to transfect HUVEC (MOI = 3) and HPVEC (MOI = 10). After transfected 4 h, the same volume of fresh medium was added to cells and cultured for 24 h. To investigate the effect of lentivirus vector HBLV-h-GATA2-3xflag-ZsGreen-PURO on OLA-downregulated GATA2 expression, cells were exposed to OLA for 24 h after lentivirus vector HBLV-h- GATA2-3xflag-ZsGreen-PURO treatment.

Statistical analyses

All experimental data were analyzed using SPSS software (version 23.0, USA). Results conforming to a normal distribution were compared using one-way analysis of variance (ANOVA) and are presented as mean ± standard error of the mean (Mean ± SEM). When ANOVA results were significant, post hoc pairwise comparisons were performed using the Tukey's tests. Animal phenotype data were analyzed by repeated measures ANOVA, and the dose-response relationship was assessed using the chi-squared trend test. Linear mixed-effects models (LMMs) were applied to evaluate dose-dependent effects on fetal and placental developmental outcomes, adjusting for litter size and fetal sex. Dam ID was included as a random intercept to account for clustering of fetuses within litters, while dose group, litter size, and fetal sex were included as fixed effects. Models were fitted using restricted maximum likelihood (REML), and degrees of freedom for fixed effects were estimated using the Satterthwaite approximation. Adjusted marginal means (emmeans) were calculated for each dose group, and Tukey's method was applied for post hoc pairwise comparisons. Random effect estimates, including variance and standard deviation for dam ID and residuals, were reported. For gene function and pathway enrichment analyses, P values for GO, KEGG, and Reactome analyses were calculated using hypergeometric tests, P values for GSEA were obtained via permutation tests, and P values for differential expression (DEGs) analyses were calculated using Wald tests. FDR were adjusted using the Benjamini–Hochberg method. Statistical significance was set at P < 0.05 or FDR < 0.05, and results are presented as estimate ± standard error mean (SEM). All analyses were performed in R (version 4.3.1) using the lme4 and lmerTest packages.

OLA distribution across placenta and fetus

Our previous study identified a low-molecular-weight (~720 Da) eight-unit oligomer as the most abundant form following intestinal biodegradation, originating from PLA as excess ester oligomers via depolymerization [13]. Based on this, the ~ 1,000 Da OLA were synthesized. After confirmation of pregnancy, 8-week-old CD-1 mice were exposed to an environmentally relevant dose of OLA (0.01 and 0.1 mg/kg/day) for 18 consecutive days (GD0-GD17, Fig 1A). First, we measured maternal and fetal body distribution of OLA through in vivo fluorescence imaging and revealed that OLA accumulated in utero towards the fetuses following a time-dependent manner through GD14 to GD18, the late pregnancy stage for the dams (Figs 1B and S2). Interestingly, we observed the markedly higher fluorescence density on the maternal side than the fetal side of placenta. Further, in fetal organs, the highest fluorescence signal was observed in fetal brain at GD18, relative to other major organs including liver and kidney (Fig 1C). In addition, direct quantitation of OLA oligomers by LC-MS/MS confirmed substantial increase in concentration of OLA in the fetal livers of the treatment group, with the OLA8 isomer showing the highest abundance (Fig 1D and 1E). These quantitative results confirmed that OLA transferred across the placenta to the fetus, and accumulated in the placenta and fetus, forming a gradient distribution in a time-dependent pattern.

Gestational exposure to OLA impairs fetal growth in utero

Following gestational exposure to environmentally relevant doses of OLA (0.01 and 0.1 mg/kg/day), both fetal and placental sizes were significantly reduced compared to controls (Fig 2A–2E, S2 and S3 Tables). LMMs revealed that dose group was significantly associated with multiple fetal and placental outcomes, after adjusting for litter size and fetal sex and including dam ID as a random intercept. Fetuses in the high-dose group exhibited markedly lower weight (−0.107 ± 0.040 g, P = 0.016) and crown-rump length (−3.159 ± 0.262 mm, P < 0.001) compared with controls, while the mid-dose group showed a nonsignificant trend for reduced weight (−0.056 ± 0.036 g, P = 0.139) and significantly reduced length (−1.534 ± 0.236 mm, P < 0.001). Placental weight was decreased in the high-dose group (−0.016 ± 0.006 g, P = 0.019), and placental diameter was significantly smaller (−0.766 ± 0.196 mm, P = 0.001), with trends for reduction in the mid-dose group. Female fetuses had consistently lower weight, length, and placental weight and diameter compared with males, whereas litter size was not significantly associated with any outcome. Random effects analyses indicated that between-litter variance contributed to all outcomes, although residual variance remained larger than litter-level variance, confirming substantial within-litter variability (fetal weight: variance = 0.00397, SD = 0.063; fetal length: 0.1773, SD = 0.421; placental weight: 8.185 × 10 ⁻ ⁵, SD = 0.00905; placental diameter: 0.10055, SD = 0.3171). These findings support the use of dam ID as a random intercept. Together, these findings confirm that gestational OLA exposure impairs placental development and causes IUGR in mice.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Gestational exposure to environmentally relevant dose of unlabeled oligomeric lactic acid (OLA) induced intrauterine growth restriction in the offspring and reduced placental size.

Pregnant dams of control group were administered with corn oil, the exposure groups were 0.01 and 0.1 mg/kg/day of unlabeled OLA administered in corn oil (N = 12 dams per group). (A) Representative fetuses. (B) Crown-rump length (N = 12 dams per group). (C) Fetal body weight (N = 12 dams per group). (D) Placental diameter (N = 12 dams per group). (E) Placental weight (N = 12 dams per group). Data are expressed as mean ± standard deviation (SD), statistically significant differences were determined using Linear mixed-effects models, **P < 0.01. Data for this figure is presented in S1 Data folder.

https://doi.org/10.1371/journal.pbio.3003676.g002

Placental vascular dysplasia following gestational exposure to OLA

To further investigate the potential target and underlying mechanism through which OLA-induced IUGR, we analyzed placental morphology. In mid-sagittal sections at GD18, H&E staining revealed a dose-dependent reduction in the proportion of the labyrinthine zone (LZ) area in exposed groups (Fig 3A and 3B), reflecting relative area estimates within the analyzed plane rather than volumetric proportions of the whole placenta. Additionally, we observed a smaller blood sinus area and lower protein expression of placental vascular markers CD34 and Laminin in the exposure group (P < 0.05, Fig 3C–3H) using immunofluorescence staining. These results confirmed that gestational OLA exposure disrupted placental vascular structure and volume, which may be the potential target explaining IUGR.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Pathophysiological changes of the placental vascular structure after gestational exposure to environmentally relevant dose of unlabeled oligomeric lactic acid (OLA) at GD18.

(A and B) Hematoxylin-Eosin (H&E) staining of the placenta from dam of control and OLA-exposed groups under with magnification at 400× (N = 4) (A), and differences in the relative area of labyrinth zone (LZ, %) between control and OLA-exposed groups (B). (C and D) H&E staining of the LZ with magnification at 400× (N = 4) (C), differences in the relative area of LZ and blood sinus to the placenta (%) between control and OLA-exposed groups, with black arrows pointing towards blood sinuses (D). (E and F) Immunofluorescence staining of placental Laminin under with magnification at 400× (N = 4) (E) and quantification for Laminin (F). (G and H) Immunofluorescence staining of blood vessel index marker CD34 under with magnification at 400× (N = 4) (G) and quantification for blood vessel marker CD34 per field (H). Statistically significant differences were determined using one-way analysis of variance, *P < 0.05, **P < 0.01, ***P < 0.001. Data for this figure is presented in S1 Data folder.

https://doi.org/10.1371/journal.pbio.3003676.g003

Gene expression of biological processes and key signaling pathways in the placenta vascular development

To investigate the causes of placental vascular dysplasia, we first performed transcriptomic analysis of placental tissues. GO analysis and KEGG pathway analysis revealed significant enrichment in biological processes related to placental vascular development, including placenta development, blood vessel morphogenesis, angiogenesis, and VEGF and TGF-β signaling pathways (Figs 4A and S3). GSEA also revealed negative enrichment in the regulation of vasculature development post-OLA exposure (Fig 4B). Because bulk placental RNA-seq reflects a mixture of cell types that may mask cell-type-specific transcriptional alterations, we complemented this approach with scRNA-seq. The scRNA-seq analysis identified seven major cellular subsets within mouse placentas (Figs 4C and S3) and endothelial cell clusters were clearly delineated in both control and OLA-exposed samples (Fig 4D and 4E). We next focused on the endothelial cell populations. Consistent with the bulk transcriptomic findings, GO enrichment analysis of endothelial cell DEGs showed significant involvement of vascular development pathways (Fig 4F), while Reactome pathway analysis indicated selective enrichment of VEGF-associated signaling (Fig 4G). These data provide cell-type-specific evidence of impaired placental vascular development in response to OLA exposure. Additionally, the variable importance in projection (VIP) score identified Cdh5 and Gata2 as key contributors to vascular development (Fig 4H). Gata2, as a transcription factor, plays a critical role in regulating cell and organ development. Thus, we hypothesized that Gata2, by regulating the VEGF signaling pathway, leads to impaired placental vascular development, thereby inducing IUGR following gestational exposure to OLA. Further qRT-PCR analysis revealed significant decreases in the expression of Gata2 and VEGF signaling-related genes, including Vegf-a, Vegf-d, Vegfr1, and Vegfr2, in the placenta after OLA exposure (Figs 4I and S3), supporting this hypothesis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Transcriptomic evidence of disrupted placental vascular development following gestational exposure to human-relevant doses of unlabeled oligomeric lactic acid (OLA).

(A) Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) revealed significant enrichment in biological processes related to placental development. (B) Gene set enrichment analysis (GSEA) showed negative enrichment of placental vasculature development in OLA-exposed placentas compared with controls. (C) Uniform Manifold Approximation and Projection (UMAP) of placental single-cell RNA-seq data, where each dot represents one cell and colors indicate distinct cellular subsets (N = 3 per group). (D–E) UMAP visualization of single-cell transcriptomes from six mouse placentas. Endothelial cell clusters were identified in the control and OLA-H groups using established endothelial marker genes; an example marker gene Rflnb is shown. (F) GO enrichment analysis of endothelial cell DEGs indicated significant involvement of vascular development-related biological processes. (G) Reactome pathway analysis showed that endothelial cell DEGs were specifically enriched in signaling pathways related to vascular endothelial growth factor (VEGF). (H) Top 10 DEGs associated with placental vasculature development ranked by the variable importance in projection (VIP) score, blue squares represent genes with lower expression levels, while red squares indicate higher expression levels. (I) mRNA expression levels of five VEGF-regulatory genes in placental tissues (N = 8 per group); hprt1 rRNA was used as the internal control. P values in (A), (B), (F), and (G) were determined using the hypergeometric test; P values in (A), (B), (F), and (G) were determined by a permutation test. False discovery rates (FDR) were adjusted using the Benjamini-Hochberg method. Statistical significance in (I) was assessed using one-way analysis of variance (*P < 0.05, **P < 0.01). The "rich factor" represents the ratio of DEGs annotated to a given pathway to the total number of genes annotated to that pathway. Data for this figure is presented in S1 Data folder.

https://doi.org/10.1371/journal.pbio.3003676.g004

OLA prevents Gata2 nuclear translocation and function in vivo and in vitro

Next, the mechanism underlying Gata2-mediated VEGF signaling obstruction was explored. Immunofluorescence co-staining of CD34 and Gata2 showed a dose-dependent decrease in Gata2 expression within CD34-positive endothelial cells following OLA exposure (Fig 5A). While most Gata2 signal originated from trophoblasts, the reduction in endothelial Gata2 indicates a direct effect on this population. Consistently, western blot analysis of cytoplasmic and nuclear fractions revealed that OLA exposure reduced nuclear Gata2 levels while increasing cytoplasmic Gata2 (Fig 5B–5D). The levels of VEGF-A in placental tissues were also decreased compared with controls (Fig 5B and 5F), supporting impaired VEGF signaling. Together, these results suggest that OLA inhibits Gata2 in endothelial cells, thereby disrupting VEGF-mediated signaling. We note that changes in trophoblasts cannot be excluded, but our subsequent mechanistic analyses focused on the endothelial cells, confirming functional impairment in this compartment.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Gestational unlabeled oligomeric lactic acid (OLA) exposure induced abnormal placental vascular development-related to nucleus translocation of transcription factor GATA2.

(A) Immunofluorescence staining for placental Gata2 of dam from control and OLA-exposed groups under with magnification at 600×, with red pointing towards GATA2, green pointing towards CD34, blue pointing towards DAPI (N = 4). (B) Immunoblot analyses of nucleus and cytoplasm level of Gata2 and cytoplasm vascular endothelial growth factor (VEGF)-A level in placenta. (C) Quantification for GATA2 integrated density/μm [2] of GATA2. (D and E) Quantification for nucleus (D) and cytoplasm (E) level of Gata2 in placenta, respectively. (F) Quantification for cytoplasm level of VEGF-A in placenta. Statistically significant differences were determined using one-way analysis of variance, *P < 0.05, **P < 0.01. Data for this figure is presented in S1 Data folder. The original uncropped western blotting images are available in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003676.g005

We then examined how impaired GATA2 nuclear translocation affects the expression of VEGF-A and its receptors in HUVEC and HPVEC following OLA exposure. We silenced Gata2 expression (si-GATA2) and observed a significant reduction in nuclear GATA2 and VEGF-A and VEGFR2 levels. Conversely, overexpression of Gata2 (oe-GATA2) resulted in increased nuclear translocation of GATA2 and higher VEGF-A and VEGFR2 levels (Figs 6A and 6B, S4). We also observed a slight upward shift of the nuclear GATA2 band in HPVECs (Fig 6F), likely reflecting cell type-specific post-translational modifications, such as ubiquitination, SUMOylation, which do not affect the overall conclusion [44–46]. Western blotting analysis confirmed consistent changes in GATA2 and VEGF-A and VEGF-R2 protein levels (Figs 6C–6F and 6H–6I, S4), supporting the findings above. After 24 h exposure to OLA, the si-GATA2 group showed a reduced GATA2 nuclear translocation, while oe-GATA2 significantly increased nuclear GATA2 (Fig 6A–6D). Additionally, VEGF-A and VEGFR2 levels were lower in the si-GATA2 + OLA group and higher in the oe-GATA2 + OLA group compared to their respective controls (S4 Fig). To determine whether GATA2 directly regulates VEGF-A or its receptor VEGFR2, we performed in silico promoter binding site prediction. The predicted GATA2 binding sites within the VEGF-A promoter showed low relative scores (none exceeding the commonly accepted thresholds of relative score > 0.8 or absolute score > 8), suggesting that VEGF-A is unlikely to be a direct transcriptional target of GATA2. In contrast, GATA2 appeared more likely to regulate Vegfr2 transcription. Indeed, subsequent dual-luciferase reporter assays demonstrated that GATA2 directly binds to and activates the Vegfr2 promoter (Fig 6G and S4 Table). Collectively, these results indicate that OLA exposure suppresses VEGF signaling primarily by reducing GATA2 nuclear translocation and thereby diminishing Vegfr2 transcriptional activation.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Unlabeled oligomeric lactic acid (OLA) prevented GATA2 nuclear translocation and transcriptional activity.

Human umbilical vein endothelial cell (HUVEC) and primary human placental vascular endothelial cells (HPVEC) stably were transfected with si-GATA2 or oe-GATA2 in the presence or absence of OLA for 24 h, as indicated. (A) Immunofluorescence staining picture for GATA2 in HUVEC, Magnification at 600× , with red pointing towards GATA2, green pointing towards Phalloidin, blue pointing towards DAPI. (B) Immunofluorescence staining picture for GATA2 in HPVEC, Magnification at 600×, with red pointing towards GATA2, green pointing towards Phalloidin, blue pointing towards DAPI. (C) Immunoblot analyses of nucleus and cytoplasm level of GATA2, cytoplasm level of β-actin and Lamin B1 in HUVEC. (D) Quantification for nucleus level of GATA2 in HUVEC. (E) Quantification for cytoplasm level of GATA2 in HUVEC. (F) Immunoblot analyses of nucleus and cytoplasm level of GATA2, cytoplasm level of β-actin and Lamin B1 in HPVEC. (G) Dual-luciferase reporter assay confirming that GATA2 binds to the VEGFR2 receptor promoter and regulates its transcription. (H) Quantification for nucleus level of GATA2 in HUVEC. (I) Quantification for cytoplasm level of GATA2 in HUVEC. Data for this figure is presented in S1 Data folder. The original uncropped western blottting images are available in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003676.g006

During nuclear translocation, GATA2 undergoes conformational changes by exposing its nuclear localization signal (NLS) region, which is then recognized and bound by nuclear-transport receptor, importin α-6 protein (KPNA6), facilitating nucleocytoplasmic transport through the nuclear pore complexes (NPCs) [47,48]. To identify the potential mechanism of GATA2 nuclear translocation obstruction following OLA exposure, we performed simulated docking of GATA2 with OLA oligomers at 2–9 units, as shown in Fig 7A and 7B. The docking scores for GATA2_OLA8 and GATA2_OLA6 were −7.64 kcal/mol and −6.83 kcal/mol, respectively, indicating stronger binding than the other oligomers examined. Molecular dynamics simulations further revealed that the structures of GATA2_OLA8 and OLA6 remained stable within the 0–100 ns range (S4 Fig). To experimentally validate the in silico predictions, we performed SPR analyses to assess the interaction between purified GATA2 protein and OLA. SPR revealed that GATA2 immobilized on a CM5 chip binds OLA with high affinity, exhibiting a dissociation constant (KD) of ~2.75 nM and a slow dissociation rate, indicative of stable complex formation. These findings provide direct biochemical evidence that GATA2 can specifically interact with OLA, supporting the computational predictions. We next performed CETSA in HUVEC and HPVEC to assess whether OLA binding alters the thermal stability of endogenous GATA2 in a compound-specific manner. Consistent with the SPR findings, CETSA demonstrated a clear stabilization of GATA2 upon OLA treatment, further confirming the interaction and its functional relevance (Fig 7C–7G). Hence, these data substantially strengthen our mechanistic hypothesis by providing convergent computational and biochemical evidence for a direct OLA-GATA2 interaction. We therefore propose that OLA directly binds to GATA2, causing abnormal conformational changes that prevent its NLS from being recognized and bound by the KPNA6, thereby preventing GATA2 nuclear translocation and reducing its activity (Fig 8).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. OLA-GATA2 interaction is specific and unique.

HUVEC and HPVEC were treated with OLA for 24 h, as indicated. Molecular Docking for (A) GATA2 and OLA with different units. (B) Conformation of the complex formed between GATA2 and 6 OLA units or 8 OLA units at 0 and 100 ns. GATA2 is shown as a gray ribbon model; OLA are shown as a spherical model. Carbon, light blue (OLA8) or yellow (OLA6); oxygen, red; hydrogen, white. (C–F) CETSA analyses in HUVEC and HPVEC showing changes in GATA2 thermal stability following OLA treatment for 24 h. (C, E) Representative immunoblots of GATA2 and β-actin. (D, F) Quantification of GATA2 signal intensities. (G) SPR sensorgram demonstrating direct binding between OLA and purified GATA2 protein. Data for this figure is presented in S1 Data folder. The original uncropped western blotting images are available in S1 Raw Images.

https://doi.org/10.1371/journal.pbio.3003676.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Mechanistic schemes of the disruption of the VEGF signaling pathway regulated by GATA2, leading to impaired placental vasculature development induced by OLA released from orally ingested polylactic acid plastics during pregnancy.

Mechanistic schemes created in Figdraw.

https://doi.org/10.1371/journal.pbio.3003676.g008