Integrative analyses of maternal plasma cell-free DNA nucleosome footprint differences reveal chromosomal aneuploidy fetuses gene expression profile

The study was performed using low-throughput sequencing data of 140 healthy pregnant women and the cfDNA library of 101 pregnant women with aneuploid fetuses with trisomy 21, 18, and 13 by Guangzhou Darui Biotechnology Company. The selection of cfDNA sequencing libraries was based on the gestational age at plasma collection and follow-up results for pregnant women with trisomic pregnancy. All selected samples were collected at 12–28 weeks of gestation and the selected participants had singleton pregnancies. The definitive diagnosis of each fetus was confirmed by chromosome testing of the fetus in utero via the collection of amniotic fluid cells, chorionic tissue biopsy, or umbilical cord blood cell culture.

A total of 6 mL peripheral blood was collected in EDTA tubes from each patient and immediately centrifuged at 1600 g for 10 min at 4 °C. The plasma was then transferred to a low-adsorption centrifuge tube (Eppendorf, EP tubes), and re-centrifuged at 16,000 g for 10 min at 4 °C, and the supernatants were collected into new EP tubes and stored at − 20 °C before use. At least 10 ng cfDNA was extracted from plasma for sequencing using the QIAamp DNA Blood Mini Kit (Qiagen). A starting amount of approximately1–5 ng DNA was prepared for library construction. The libraries for sequencing were prepared using the Ion Xpress^™ Plus Fragment Library Kit and the following steps: end repair, adapter ligation, and amplification with 12 amplification cycles PCR. Libraries were analyzed using a 2100 Bioanalyzer (Agilent Technologies, Singapore) to observe DNA size distribution. Sequencing was performed using the Ion PI^™ Hi-Q^™ OT2 200 Kit and Ion PI^™ Hi-Q^™ Sequencing 200 Kit.

All samples were sequenced using the Ion Proton platform (Thermo Fisher Scientific, USA). For trisomic pregnancy samples, each cfDNA sample had approximately 20 M sequencing raw data, the sequenced fragment length is 150–160 bp, and the sequencing depth is about 1X. For healthy pregnancy samples, the sequencing raw data of each sample is 5 M, the length of the sequencing fragment is 150–160 bp, and the sequencing depth is about 0.25X. To serve as the control, every 4 low-coverage NIPT data of 140 healthy pregnant women were pooled randomly using “merge” function in SAMtools [26] as the control sample data according to gestational age and a total of 35 bam files were obtained that simulate the high-depth data of healthy pregnant women. Base calling and alignment were performed using Torrent Suite (Version 5.4.0). All reads were mapped to the human reference genome hg19. All results of cfDNA testing and clinical data were transferred to an independent data-coordinating center for consolidation.

Raw bam files processed GC correction using deeptools [27] (Version 3.5.1) to compute GCBias and correct GC bias. The read counts of regions and single base depths ranging from − 1000 bp to + 1000 bp around TSSs were calculated using bedtools coverage [28] (Version 2.17.0) by switching the program argument to “-d”. All gene information and TSS positions were obtained from RefSeq [29]. The read counts were normalized using the reads per kilobase per million mapped reads (RPKM) method.

The fold change of each TSS was calculated between groups (N vs. T13, N vs. T18, N vs. T21), and only the TSS with fold change  > 1.5 was used in the Wilcoxon rank sum test (two-sided). The false discovery rate (FDR) method of Benjamini and Hochberg was used for multiple testing corrections. Finally, the TSSs with p-value  < 0.05, |log₂(FoldChange) |≥ log₂1.5, and FDR  < 0.2 were determined as differential TSSs. Volcano plots were generated using the ggplot2 [30] package, and heatmaps were plotted with pheatmap. All data preparation and screening processes were based on R 4.1.1.

To explore the function of the corresponding genes of differential TSSs, GO and KEGG analyses were performed using the R package, clusterProfiler [31] (Version 4.2.0). The GO terms and KEGG pathways were obtained from the QuickGo [32] and KEGG [33] websites. Immune-related genes were determined using ImmPort database [34]. All gene aliases and ID in the different databases were downloaded from GeneCards [35].

To determine the cfDNA footprint profiles that reflect intracellular nucleosome positioning and predicted gene expression, we selected 101 samples (42 trisomy 21, 28 trisomy 18, and 31 trisomy 13; Additional file 1: Table S1) of the next-generation sequencing library from the Darui Bio-Clinical Examination Center and low coverage sequencing data of cfDNA derived from 140 healthy controls according to the gestational age of plasma collection and their follow-up results (Fig. 1). The gestational age of the healthy controls was matched to the gestational age of each trisomic pregnancy. The study flowchart of the retrospective analysis of whole-genome resequencing of all collected libraries for trisomic pregnancies is shown in Fig. 1. By comparing cfDNA coverage at the primary TSS (pTSS) for the 500 highest and 500 lowest expressed genes in the placenta and maternal blood (Fig. 2A, C), the sequence coverage depth around the TSSs was found to be significantly lower for highly expressed genes than lowly expressed genes in the placenta (Fig. 2B). Similar patterns were evident for genes with high or low expression levels in the maternal blood (Fig. 2D), aligning with the conclusions of some existing studies on the use of nucleosome footprints to identify differential expression genes [22‐24]. As the development of trisomy pregnancy is related to the function of the placenta and maternal immune system, these results suggest that the coverage of cfDNA in gene TSS can be used as a biomarker to explore changes in gene expression during trisomy pregnancy. PCA was performed on GC-adjusted TSS profiles, and samples were divided into four groups according to pregnancy status to reflect the differences in TSS profiles between healthy pregnant women and women with trisomy pregnancies (trisomy 21, trisomy 18, and trisomy 13) (Fig. 2E). The resulting score plot showed that trisomy and healthy pregnancies occupied distinct locations in the PCA plot. However, each trisomy was cross-reactive with all other trisomic pregnancies.

We compared the sequence coverage depth around the cfDNA TSSs between pregnant women with trisomy and healthy pregnant women. Among the 50,800 tested TSSs, 158 were significantly different (p < 0.05, |log₂(foldchange)|≥ log₂1.5 and FDR < 0.2), including 101 TSSs with relatively high coverage and 57 TSSs with relatively low coverage in pregnant women with trisomy (n = 101) compared with healthy pregnant women (n = 38) (Fig. 2F, Additional file 1: Table S1). Hierarchical clustering analyses revealed an obvious separation between pregnant women with trisomy and healthy controls (Fig. 2G). Simultaneously, we compared TSS profiles between pregnant women with different trisomy pregnancies (T21, T18, and T13) and healthy controls. The TSS regions of 242 genes (p < 0.05, |log₂(FoldChange)|≥ log₂1.5, and FDR < 0.2) were significantly differentially covered in trisomy 21 pregnancies, 73 of which were upregulated and 169 were downregulated (Additional file 6: Fig. S1A; Additional file 1: Table S1). The TSS regions of 517 genes (p < 0.05, |log₂(FoldChange)|≥ log₂1.5, and FDR < 0.2) were significantly differentially covered in trisomy 18 pregnancies, 157 of which were upregulated and 360 were downregulated (Additional file 6: Fig. S1B, Additional file 1: Table S1). The TSS regions of 293 genes (p < 0.05, |log₂(FoldChange)|≥ log₂1.5 and FDR < 0.2) were significantly differentially covered in trisomy 13 pregnancies, 105 of which were upregulated and 188 were downregulated (Additional file 6: Fig. S1C; Additional file 1, Table S1). Conversely, the depth of promoter coverage of these genes were not altered in healthy pregnant women. We proceeded to perform unsupervised hierarchical clustering analysis on the coverages for these trisomy syndromes. Distinctive coverage patterns were found for T21 (Additional File 6, Fig. S1D), T18 (Additional file 6: Fig. S1E), and T13 (Additional file 6: Fig. S1F), revealing distinct coverage patterns for T21, T18, and T13. The heatmaps showed distinct patterns of promoter coverage between healthy pregnancies and pregnancies with trisomy (Additional file 6: Fig. S1D-F).

To study the main functions of the differential genes between pregnant women with trisomy and healthy pregnant women, we performed gene function annotation analysis of the differential genes. Functional annotation analyses revealed that these genes were related to the top 15 terms (Fig. 3A). Trisomy 21, 18, and 13 fetuses commonly have severe mental retardation, prenatal multiple organ malformations, reproductive organ dysplasia, and growth retardation [36‐38]. These fetuses result from chromosomal aberrations, the non-disjunction of homologous chromosomes or chromatids that occurs during oocyte meiosis process. Our method enriched many signaling pathways involved in the formation and maintenance of chromatin structures. For example, genes related to chromatin remodeling and nucleosome aggregation (ARID1A, SATB1, SETD5, EHMT1, TSPY8, and TSPY10) were concentrated in the signaling pathways of chromatin formation and structural changes (GO:0,006,333, GO:0,006,334, GO:0,034,728, GO:0,006,338, GO:0,006,325); the related genes are listed in Supplementary Material (Additional file 2: Table S2). Some differential genes were also found to be enriched in pathways related to specific reproductive system development. For example, genes related to gonadal mesoderm development (TSPY8, TSPY4, TSPY10, TSPY3, RBMY1B, SF1) were found to be concentrated in the signaling pathways of gonadal development, reproductive system development, and sex differentiation (GO:0,008,406, GO:0,048,608, GO:0,045,137, GO:0,061,458, GO:0,007,548, GO:0,007,498); the related genes are listed in Additional file 2: Table S2. Diseases that have been reported to be associated with these genes include global stunting with or without mental retardation; autosomal dominant non-syndromic mental retardation; and heart, eye, and genital syndromes [39‐41].

We searched the literature to elucidate the functional relevance of the top enriched gene functions and pathways to the diseases. According to the results, each set of enriched pathways was associated with the development of the corresponding disease (Fig. 3B). For example, the Hippo signaling pathway (hsa04390) and TGF-β signaling pathway (hsa04350) play important roles in the regulation of cell growth, differentiation, and development in the human body, which involve different key physiological processes [42, 43]. Based on recent research [44], the Hippo pathway and the TGF-β signaling pathway have a new role in coordinating developmental signals. Hippo signaling plays a key role in early embryonic development, as low Hippo activity is required for trophoblast differentiation whereas high Hippo activity allows inner cell mass formation. Hippo signaling also plays an important role in the embryonic developmental stages of multiple organs, such as the heart, craniofacial development, lungs, eyes, brain, and kidneys [45, 46]. The TGF-β signaling pathway inherently contributes to a wide range of early developmental events, including embryonic patterning, cell fate determination, and dynamic motility [47]. The Ras signaling pathway (hsa04014) controls many cellular responses, such as proliferation, survival, and differentiation. Developmental disorders associated with mutations in the Ras pathway share common phenotypic features, including facial abnormalities, cardiac defects, and impaired growth and development [48]. Ubiquitin-mediated proteolysis pathway (hsa04120) regulate nearly every aspect of cellular events in eukaryotes. Dysregulated ubiquitination can lead to serious consequences and human diseases, including cancer, growth defects, and immune disorders [49, 50]. In addition to these signaling pathways, which play important roles in the overall regulation of embryonic growth and development, we found that these major functional enrichments are concentrated in signaling pathways that regulate embryonic neurodevelopment. The MAPK signaling pathway (hsa04010) is critical for normal development of the nervous system from neural progenitor cells derived from the embryonic mesoderm. Studies have shown that MAPK signaling is important for oligodendrocyte differentiation. Children with reduced levels of MAPK signaling exhibit microcephaly, cognitive impairment, and developmental delay [51, 52]. The axon guidance pathway (hsa04360) is critical for the formation of neuronal networks [53]. The thyroid hormone signaling pathway (hsa04919) affects brain development by regulating gene expression, and is important for neuronal development, and fetal size and survival. Maternal hypothyroidism or deficiency can have severe effects on fetal brain and heart development [54, 55]. Individual genes with significantly different coverages from these pathways are listed in the supplementary material (Additional file 3: Table S3). These signaling pathways were found to display significant differential enrichment in trisomy pregnancies.

In addition to the common differential analysis of these chromosomal aneuploid pregnancies, we also performed functional enrichment analysis of differential genes between cfDNA nucleosome footprints of healthy pregnant women and pregnant women carrying a fetus with trisomies 21, 18, and 13, respectively. Although these differentially gene-enriched pathways were not significantly different among the three trisomic pregnancies, the number of differentially enriched genes in the pathways was found to differ in different trisomic pregnancies (Fig. 3C–H). In trisomy 18 pregnancies, the differential genes were mainly enriched in reproductive system and mesoderm development (GO:0,008,406, GO:0,048,608, GO:0,061,458, GO:0,007,498, GO:0,045,137, GO:0,060,485) (Fig. 3E). Interestingly, the number of enriched differential genes in trisomy 18 pregnancies was significantly higher than the number of differential genes in the same signaling pathway in the other two aneuploid pregnancies; the related genes are listed in Supplementary Material (Additional file 2: Tables S2, Additional file 3: Tables S3). In trisomy 13 pregnancies, differential genes were enriched for multiple biological processes, from embryo implantation to embryonic organ development, during embryonic development in utero. These pathways, which play important roles in the regulation of embryonic development (GO:0,007,566, GO:0,016,331, GO:0,048,568, GO:0,001,838, GO:0,048,568, GO:0,001,701) (Fig. 3G), were not significantly enriched in the other two aneuploid pregnancies. Compared to other children with chromosomal aneuploidy, children with trisomy 18 have obvious genitourinary abnormalities and abnormal developmental delays. Fetuses with trisomy 13 have fewer stable embryos. The up- and down-regulation functions and pathways of differential gene enrichment in the three trisomic pregnancies are presented in the Supplementary Material (Additional file 6: Figs. S2, S3), and the related genes are listed in the Supplementary Material (Additional file 6: 3–5).

According to the literatures, survivors of common trisomy aneuploidies have obvious growth and developmental disorders and mental retardation. Our differential gene enrichment results precisely focused on the pathways of cell growth and development and neurodevelopmental regulation, which correspond well with the shared genetic phenotypes of fetuses with trisomy. In addition, our differential genes in different chromosomal aneuploid pregnancies were focused on their respective and most characteristic genetic phenotypes, indicating that our method may reveal good evidence related to fetal gene expression profiles in trisomic pregnancy.

To determine whether these differential genes were associated with some immune responses in trisomic fetuses during pregnancy, all genes with significantly differential TSS obtained from trisomic pregnancy samples that were screened out were entered into the immune gene database, ImmPort database [34], for comparison and search for the immune genes among these significantly different genes by generating intersections. Fourteen distinct immune genes were identified, and the boxplot shown the relative depth of the region around the TSS of some immune-related genes in the cfDNA of pregnancies in different groups (Fig. 4A–F). Among them, trisomy 13 pregnancy had six distinct genes, trisomy 18 pregnancy had seven distinct genes, and trisomy 21 pregnancy had three distinct genes. We searched the literature for correlations between different immune genes and fetal and maternal immune responses. According to our findings, these significantly different immune genes are related to cell differentiation, growth arrest, and apoptosis, including embryogenesis and tissue homeostasis. The expression of these immune genes, especially the differential genes that play important roles in embryogenesis, differentiation, and development, was significantly different in the promoter profiling of different trisomic pregnancies and healthy pregnant women (Fig. 4A–F).

For example, Sp1 (Sp1 Transcription Factor) has an important role in the development of blood cells. Loss of Sp1 results in general dysregulation of the timing and coordination of differentiation trajectories during hematopoietic specification, such as uncoordinated and unstable embryonic development [56]. FGF18 (Fibroblast Growth Factor 18) is expressed in embryo and is required for osteogenesis and chondrogenesis. Studies have shown that deletion or mutation of FGF18 results in reduced proliferation of calvarial osteogenic mesenchymal cells during embryonic development, and terminal differentiation to calvarial osteoblasts is specifically delayed [57]. CYLD (Deubiquitinating Enzyme) helps regulate cell survival, proliferation, and differentiation through its effect on NF-κB activation. CYLD is important for the TNF signaling pathway, and a lack of the CYLD gene leads to early embryonic lethality [58]. HLA-C (Human Leucocyte Antigen-C) is the only classical MHC molecule expressed by extravillous trophoblasts (EVTs). During pregnancy, HLA-C mediates the interaction of EVT with maternal KIR and regulates trophoblast invasion necessary for placentation and overall fetal growth, which in turn affects placental formation and pregnancy outcomes [59, 60]. The Notch signaling pathway is essential for proper embryonic development, and mutations in its components disrupt embryonic development and cause inherited diseases in humans. The Jagged2 (JAG2)-mediated Notch signaling is essential for proper craniofacial morphogenesis and proper thymic and limb development in embryos [61]. SBDS (SBDS ribosomal maturation factor) is an essential gene for early mammalian development, and its expression pattern is consistent with its critical role in cell proliferation. Deletion of SBDS results in early embryonic lethality and ectoderm development failure [62]. All related genes are listed in the Supplementary Material (Additional file 1: Table S1).

Interestingly, the expression of these immune genes, which play important roles in regulating cell growth and development, embryonic differentiation, and maturation in our analysis, is consistent with those reported in previous studies. In the early and second trimesters of pregnancy, the fetal DNA population only accounts for approximately 10% to 20% of the maternal plasma free DNA, called the fetal part [19, 63]; the remaining population is mainly maternal DNA. Since the cfDNA of mother was mainly derived from immune cells, these results suggested that these immune genes may be associated with the mother’s immune response to the fetus.