Clinical characteristics of the participants are presented in Table 1. The distribution of age and BMI were similar across EOPE, LOPE and healthy pregnancy groups, whereas GA at delivery was significantly lower in EOPE (31.3 ± 2.5 weeks) and LOPE (37.03 ± 3.4 weeks), compared to healthy pregnancy groups (39.4 ± 5.5 weeks; p˂0.05), in line with the frequent earlier delivery in EOPE compared to LOPE or healthy controls18. Maternal blood pressure was higher in EOPE (156.4 ± 26.5) and LOPE (144.4 ± 16.8) compared to healthy pregnancies (113.9 ± 8.5), p˂0.05. Heart rate was also increased in women with EOPE (94.6 ± 29.8) compared to healthy controls (74.4 ± 6.3, p˂0.05).
Plasma proteins differentially abundant between different phenotypes of preeclampsia, and compared to healthy controls
Quantitative proteomic analysis of plasma samples was conducted by measuring tryptic peptides using data-dependent acquisition (DDA) mass spectrometry. The generated outputs contained 370 proteins detected among all 31 samples with >85% data completion (Supplementary data 1).
Initially, the clustering of groups was assessed through observing the principal component analysis (PCA) plot (Fig. 2a), which is consistent with the unsupervised hierarchical clustering of heatmap (Fig. 2b). Both PCA and heatmap revealed that the proteomes are heterogeneous between EOPE, LOPE and healthy pregnancy groups. Following PCA, the differential expression analysis was performed by three separate comparisons, including i) EOPE vs healthy pregnancy, ii) LOPE vs healthy pregnancy and iii) EOPE vs LOPE, while adjusting for GA at delivery. In total, there were 26 and 20 differentially abundant proteins in EOPE vs healthy pregnancy and LOPE vs healthy pregnancy, respectively, and one protein was differentially expressed between EOPE and LOPE (Fig. 2c-e; Fig. 3a; Supplementary data 2).
Overall, the differences in proteome profiles between EOPE or LOPE and healthy pregnancy were largely associated with impaired angiogenesis, and included perturbance in the abundance of inter-alpha-trypsin inhibitor heavy chain 3 (ITIH3)19, insulin-like growth factor-binding protein 4 (IGFBP4)20–23 and histidine-rich glycoprotein (HRG)24–26 (Fig. 3a), proteins. Immunoglobulin heavy variable 1/OR15-1 (IGHV1OR15-1), the only protein differentially abundant between two subtypes of preeclampsia, was dramatically overexpressed in EOPE compared to LOPE (FC =14.29, FDR =4.86 x 10-3) (Fig. 3b).
Furthermore, there were 11 proteins commonly shared between EOPE and LOPE. Immunoglobulin lambda variable 3-21 (IGLV3-21) was the top differentially abundant protein shared between EOPE (FDR =2.28 x 10-3) and LOPE (FDR =3.96 x 10-3) with approximately 50% decrease in its abundance compared to healthy pregnancy. Well-characterised preeclampsia biomarkers including fibronectin 1 (FN1)27,28 and complement factor D (CFD)29,30, were ~2-fold increased and among the most significant abundant proteins in both EOPE and LOPE. HRG is another well-studied biomarker of preeclampsia31–33 that was increased ~2-fold in both phenotypes of preeclampsia (EOPE: FDR=7.8 x 10-3; LOPE: FDR =4.1 x 10-3). Interestingly, there were two different ITIH proteins, differentially abundant in each preeclampsia subtype, where ITIH3 was increased in EOPE (FC =1.60, FDR =1.18 x 10-2), and ITIH2 was increased in LOPE (FC =1.29, FDR =3.30 x 10-2).
In line with the Venn diagram, the proteins with outstanding perturbations were highlighted in a three-dimensional plot (Fig. 3b; Supplementary data 3). Among the proteins with significant fold changes, a series of proteins were previously reported as biomarkers of preeclampsia, including serpin family A member 5 (SERPINA5)34, pappalysin 2 (PAPPA2)35,36, hepatocyte growth factor activator (HGFAC)37,38 and thymosin beta-4 (TMSB4X)39. IGFBP4, a protein significantly overexpressed in EOPE (log2FC=4.25, FDR =0.30 x 10-3) and LOPE (log2FC=4.91, FDR =3.96 x 10-3), with anti-angiogenic properties20–23, confirming the central role of impaired angiogenesis in preeclampsia.
Pathogenic pathway associated with different phenotypes of preeclampsia
Following identification of differentially abundant biomarkers of different phenotypes of preeclampsia, pathway enrichment was performed in EOPE or LOPE group, compared to healthy pregnancy with annotation by Reactome database14. Pathways altered in EOPE and LOPE (FDR <0.05) were presented using a triple Venn diagram (Fig. 4a; Supplementary data 4). The pathway Venn diagram revealed a range of pathways associated with altered haemostasis and immune system.
A total of 13 pathways were significantly different, with 6 pathways shared between EOPE and LOPE groups (Fig. 4a). We demonstrated the perturbations of differentially abundant proteins in the biological pathway regulating insulin-like growth factor (IGF) transport and uptake through IGF binding protein (IGFBP; Fig. 4b), which plays a significant role in both EOPE (FDR =9.90 x 10-5) and LOPE (FDR =1.69 x 10-3). Given the pro-angiogenic function of IGF signalling pathway40,41, our findings emphasised the importance of impaired angiogenesis in the pathophysiology of preeclampsia. In addition, we also observed a series of perturbed haemostatic pathways, particularly platelet degranulation in response to the elevated intra-platelet Ca2+ being more pronounced in EOPE (FDR =1.42 x 10-5; Supplementary data 4). EOPE was more closely associated with inflammatory profile than LOPE, due to the presence of a number of unique pathways of acute inflammation.
Signalling networks in EOPE and LOPE
Pairwise correlation network analysis was next performed to investigate protein-protein interactions (PPIs) in EOPE and LOPE. Networks were highlighted with the most correlated nodes (Pearson correlation r >0.7 or <-0.7), where the colour and the length of the edge are proportional to the Pearson correlation coefficient r (Fig 4a&b). To compare the PPIs in our data with broad reference evidence, the edge width was coded proportionally to the PPI-confidence scores derived from the STRING database15. Therefore, if consistent with the references, a pair of nodes are presented with a thick and opaque edge in between, or vice versa. Overall, our findings in terms of PPIs were consistent with those reported by broad evidence (Fig. 5a and 5b; Supplementary data 5).
To aid the readability of networks, the size of each node was programmed proportional to the corresponding eigencentrality (the influence in the entire network). Additionally, we also annotated the functional classification of each protein using DAVID16,17, as indicated by the node colours.
As showed in the networks, all pairwise correlations were positive. The PPI network in EOPE is simple, suggesting a more direct pathogenesis in EOPE than LOPE (Fig. 5a and 5b). Proteins associated with lipid metabolism and extracellular matrix (ECM) were the key proteins perturbed in EOPE, and ECM proteins appear to play a more important role in LOPE. ITIH3 and APOC4-APOC2 were identified as having a substantial influence within the PPI network in EOPE. LOPE was more closely associated with a range of ECM proteins including von Willebrand factor (vWF), IGFBP4 and ITIH2.