Distinct gene expression patterns for CD14++ and CD16++ monocytes in preeclampsia

Preeclampsia (PE) is a serious gestational complication affecting the life of a mother and child. The immunophenotype and gene expression profile of isolated blood monocyte subpopulations of pregnant women with PE have not been studied before. In this work, we assessed changes in CD14++ and CD16++ monocyte subpopulations in PE and physiological pregnancy (n = 33). Immunophenotyping, immunomagnetic sorting of monocytes and analysis of the transcriptional profile of their genes were carried out. The percentage of classical monocytes was significantly lower, while the intermediate fraction of monocytes was significantly higher in late-onset PE compared to control. Transcriptome analysis of late-onset PE classical CD14++ monocytes revealed significant activation of inflammation mediated by chemokine and cytokine signalling pathways; apoptosis; regulation of transcription from RNA polymerase II promoter in response to stress and others. The most suppressed signalling pathways were associated with T cell activation and selection. In CD16++ monocytes of late-onset PE cases, positive regulation of cell–cell adhesion, integrin signalling pathway, blood coagulation cascade were the most activated ones. The inflammation mediated by chemokine and cytokine signalling pathway and p53 pathway were the most down-regulated in CD16++ monocytes. The obtained results indicate profound changes occurring to two most polar monocyte subpopulations in PE and their different roles in the pathogenesis of this disease.

Preeclampsia (PE) is a serious obstetric pathology remaining the leading cause of maternal and fetal morbidity and mortality 1 . PE complicates on average 8% of pregnancies and is characterized by hypertension in combination with proteinuria/fetoplacental or maternal organ dysfunction manifesting after 20 weeks of gestation. Monocyte-macrophage lineage plays an important role in maintaining physiological pregnancy since macrophages, along with NK cells, are a major cell population in the decidua basalis, the maternal part of the placenta 2 . Monocytes originate in the bone marrow and give rise to tissue macrophages. In humans, they are divided into two main subpopulations: classical (CD14++) and nonclassical (CD16++) monocytes, depending on the expression of the mentioned above markers on their surface 3 . Researchers also distinguish an intermediate subpopulation that carries both markers (CD14++CD16+). CD14 is a membrane protein component of the lipopolysaccharide receptor complex. CD16 is the FcγRIII receptor responsible for antibody-dependent cellular cytotoxicity 4,5 . The functional differences between these two subpopulations are now being actively studied. Classical monocytes are the main sources of the macrophage pool in tissues. It is believed that non-classical monocytes are responsible for vascular patrolling by scanning the surface of the endothelium in search of pro-inflammatory signaling molecules 6 . Functional differences in monocytes are emphasized by their different responses to pro-inflammatory stimuli 7,8 . Plasma levels of pro-inflammatory cytokines are increased in several pathologies, in particular in PE: elevated concentrations of different cytokines represent a laboratory feature accompanying PE that has also been demonstrated in our previous study and other works [9][10][11] . These changes in the plasma cytokine profile certainly could affect monocytes functions.
It can be assumed that a change in the profile of monocytes, their phenotype, and the genes they express could affect those organs where monocytes come from the bloodstream, in particular the placenta. The aim of Monocyte profiles. The first step of our study was to compare the absolute number of monocytes according to clinical diagnostic reports. There were no significant differences in monocyte count between the groups (Fig. 1a). Then we performed general monocyte phenotyping before their isolation by magnetic sorting. Monocytes were gated both in dot-plot of forward and side scatter (FSC-SSC, R1 in Fig. 1b) and SSC-CD45 dot-plot (R2 in Fig. 1b) with subsequent gating on CD14-HLA-DR dot-plot (R3 in Fig. 1b). After that, we determined the percentage of cells stained positively for CD14 and CD16 markers. Significant differences were demonstrated for classical and intermediate monocyte fractions (Fig. 1c, left and middle plots in Fig. 1d:), with the percent of CD14++CD16− monocytes decreased (p = 0.003) and the percent of CD14++CD16+ monocytes increased (p = 0.03) in loPE group compared to control. We found no differences in non-classical (CD14+CD16++) monocytes level between the studied groups. The content of monocytes stained positively for CD80, CD40, CD86, CD163, CX3CR1, and HLA-DR markers was similar between the groups and close to 100% (Fig. 2). The content of CD206+ monocytes in control group was lower compared to PE groups, albeit non-significantly. As significant differences in monocyte composition were demonstrated only in loPE group, we used monocytes sorted by CD14 and CD16 markers only from loPE and control patients for subsequent microarray analysis. Monocytes were sorted using anti-CD14 and anti-CD16 magnetic beads; each fraction was stained with anti-CD14 and anti-CD16 antibodies, respectively, to determine the purity of subpopulations, which reached ≥ 90% (Fig. S1 in Supplementary Material).
The down-regulated genes included the oxidized low-density lipoprotein receptor 1 (OLR1), linc-BOLA2, and small nucleolar RNA C/D box 56B (SNORD56B). The list of ten most up-and down-regulated transcripts is given in Table S3 ( (Table 3). Enrichment analysis for down-regulated genes was successfully performed for molecular pathways only (Fig. 4c). And again, cholecystokinin receptor (CCKR) signaling map www.nature.com/scientificreports/

Effects of blood plasma (control vs loPE) on CD14++ monocytes isolated from matched healthy donors.
We further set to figure out whether the observed changes in monocyte expression profiles in loPE patients were induced by an exogenous source (e.g. by the plasma cytokines) or occurred autonomously through realization of intrinsic program. We exposed the sorted CD14++ monocytes from gestational agematched pregnant donors to cell culture media supplemented with blood plasma obtained from loPE vs control groups (Fig. 5a). The cultures were harvested for RT-PCR analysis of genes found to be the most up-(CXCR1, JUN, CXCL8) or down-(ADGRE4P, IL17R, ARL17A) regulated according to microarray experiments (Table S1, Supplementary Material). We observed a significant increase in CXCR1 expression by CD14++ monocytes in response to incubation with loPE plasma samples (Fig. 5b), whereas JUN expression levels were similar between the groups. The loPE plasma exposure promoted a decrease in CXCL8 expression by CD14++ monocytes, contrary to expectations based on microarray data. Significant decrease for ADGRE4P, was consistent with CD14++ monocyte profiles for loPE patients (Table S1, Supplementary Material). For IL17R and ARL17A, expression levels were similar between the groups.

Discussion
Recent decades were marked by intensive and comprehensive research on PE and its immunological aspects, but no decisive contributions to the field of therapy and prediction of this severe pregnancy complication. PE continues to be the leading cause of maternal and perinatal mortality and morbidity worldwide. The involvement of the monocyte-macrophage system in PE was confirmed by many experimental studies demonstrating an increase in pro-inflammatory and a decrease in anti-inflammatory macrophage numbers in the placenta, both fetal and maternal parts of the organ 10,12-16 . A possible role of blood monocytes as macrophage precursors in PE remains an open issue. A shifted balance of maternal blood monocyte subpopulations in PE, towards the predominance of non-classical or intermediate subpopulations, has been demonstrated. Our results are consistent with these findings. However, the immunophenotype is only one of the facets of monocytes. In this work, we for the first time report PE-associated changes of subset-specific gene expression signatures in CD14++ and CD16++ monocytes. The imbalanced ratio of classical, intermediate, and non-classical macrophage subpopulations is a hallmark of PE. Such imbalances are known to be more indicative of particular physiological or pathological processes than the total monocyte population dynamics 17,18 , which was also confirmed for PE. In this study, we observe a  24 . The non-classical CD14+CD16++ monocytes function as patrol cells that regulate endothelial repair 25 . Considering the association of PE with endothelial dysfunction, the increase in CD16++ monocyte numbers can be viewed as reflecting the process of endothelial repair stimulation under conditions of endothelial damage in PE. Immunophenotyping by flow cytometry revealed no PE-related differences in CD80, CD40, CD86, CD163, CD206, CX3CR1 and HLA-DR expression by monocytes. However, Nunes et al. describe down-regulation of scavenger receptor CD163 on monocytes in patients with PE. This inconsistency can be explained by the use of different assay: the authors analyzed not the percentage of CD163+ monocytes but the CD163's mean of fluorescence intensity of CD14-positive cells after short-time cell adhesion to culture plastic 26 .
For subsequent transcription profiling, two gestational age-matched groups were selected-loPE and control. We performed immunomagnetic sorting of PBMC by CD14 and CD16 markers, followed by gene expression microarray hybridization and enrichment analysis. The routine procedure of CD14++ monocyte isolation yields the classical subset constituting about 80-90% of total blood monocytes 7,27 . Classical monocytes are the main source of macrophages to which they give rise after migrating into the tissue for bridging innate and adaptive immune responses. Functionalities of classical monocytes include phagocytosis, antibody-dependent cell-mediated cytotoxicity, participation in coagulation, and cytokine secretion [28][29][30] . Our data indicate that, in loPE, CD14++ monocytes are subject to stress and catabolic restructuring, given that the up-regulated signaling pathways were mostly associated with chemokine-and cytokine-mediated inflammation, stress response, macromolecular changes in cell catabolism, ubiquitinylation/de-ubiquitinylation, apoptosis, and intracellular proteolysis. These inferences are consistent with other studies. Matias et al. demonstrate higher endogenous activation of NLRP1/NLRP3 inflammasomes responsible for caspase activation and TLR4/NF-κB pathway upregulation in PE, as well as higher gene/protein expression for IL-1β, IL-18, and TNF-α 31,32 . In a recent study Xu et al. show that CD14+ monocytes from patients with PE produce higher levels of TNF-α, IL-1β, and IL-6 compared to normotensive pregnant women and associate these observations with down-regulation of α7 nicotinic acetylcholine receptor 33 . Interestingly, according to our results, most of the significantly suppressed signaling pathways were associated with T lymphocyte activation or differentiation. It is difficult to explain the reason for such changes in the transcriptome of loPE CD14++ monocytes. The decrease in the activity of signaling pathways associated with T lymphocytes probably occurs as a compensatory response to avoid their hyperactivation in PE. Still, there is no data unequivocally confirming this hypothesis. At early stages of lung cancer, as well as in B cell non-Hodgkin lymphoma, CD14+ monocytes exert immunosuppressive properties, which allow them to suppress the cytotoxic activity of T-lymphocytes 34,35 . A similar effect has been observed in Kawasaki disease characterized by pronounced immune mobilization, activation of monocytes and neutrophils, and elevated synthesis of IL-1, IL-6, and TNFα 36 . The elevated plasma levels of pro-inflammatory cytokines in Kawasaki disease www.nature.com/scientificreports/ are accompanied by elevated levels of anti-inflammatory cytokine IL35, which suppresses the ability of CD14+ monocytes to activate naive CD4+ T cells and has a protective effect 36 . Consistently with our previous experience, CD16++ monocyte isolation procedure yields a mixture of the intermediate and non-classical monocyte subsets 7 , both of them involved in antigen processing and promotion of neutrophil adhesion to endothelium through TNFα production, and also specifically associated with wound healing processes 37,38 . In loPE, CD14++ and CD16++ monocytes differ by gene expression profiles, having only few signaling pathways in common. Signaling pathways up-regulated in CD16++ monocytes are associated with intercellular adhesion, cell-to-matrix adhesion, leukocyte activation, blood coagulation, apoptosis, and ubiquitinylation/de-ubiquitinylation. Several pathways are found in both the 'activated' and 'suppressed' enrichment lists, notably cholecystokinin receptor (CCKR) signaling map and gonadotropin-releasing hormone receptor pathway, which may indicate their participation in a complex bilateral crosstalk, as both are involved in cell survival, angiogenesis, and invasion 39 . Interestingly, the inflammatory cascade mediated by chemokine and cytokine signaling pathway has been identified as one of the most inhibited. These results suggest that, in loPE, the activated CD16++ monocytes follow their native program of vascular patrolling and innate immunity activation, which is consistent with their increased maternal blood counts 21,22,40 . The results indicate pronounced differences in functional activity and molecular signatures of CD14++ and CD16++ monocytes, particularly in loPE. Interestingly, four lincRNAs (linc-FCGR1B-7, linc-RAB23-2, linc-PARD6G-2 and linc-POTED-9) are found among the most up-regulated transcripts in both CD14++ and CD16++ monocytes of patients with loPE as compared to control group. The involvement of lincRNAs in pathogenic processes has not been studied in detail. Linc-FCGR1BP affiliates with a Fc Receptor FCGR1BP pseudogene apparently encoding a non-functional protein non-detectable at the cell surface and binding the ligand with low affinity 41 . Linc-PARD6G-2 is downregulated in multiple myeloma resistant to proteasome inhibitors 42 . Specific increase in lncRNA expression in PE, observed in several studies, places lncRNAs as prospective markers of PE. However, the exact list of lncRNAs up-and down-regulated in PE is controversial 43,44 , probably due to heterogeneity of PE and other factors that require further investigation. The nature of these lincRNA activation in monocytes in loPE remains obscure.
Our results raise the question of whether such molecular signature of CD14++ and CD16++ monocytes persists upon entering peripheral tissues or whether the cells undergo apoptosis. Monocytes are known to lose a number of markers (e.g. Ly6C in mice) upon infiltration and differentiation into macrophages 45 , but whether specific activation of pro-inflammatory and ubiquitinylation-related pathways persists after monocyte migration is unclear 17 . Several studies based on animal models suggest that monocytes extravasating constitutively into lymph nodes and tissues retain their transcriptional profiles and may even remain monocytes 23,46 . In addition, the activated monocytes may provide a source of massive cytokine secretion and thus modulate inflammation in the body 30,47,48 . The subset-specific gene expression signatures of monocytes in loPE may reflect the exposure to signaling agents (cytokines, chemokines, free nucleic acids, placental exosomes) circulating in maternal blood at elevated levels confirmed by multiple studies [49][50][51][52][53] . This hypothesis is partially supported by our data: expression of several genes up-and down-regulated in monocytes was differentially modified by exposure to blood plasma of patients with PE vs control group.
Overall, our study strengthens the concept of maternal immunological stress as a major contribution to PE, tracked by altered immunophenotypes and transcriptomic signatures of blood monocytes. Isolation of CD14++ and CD16++ monocytes was performed using magnetic sorting with beads to CD14 and CD16 receptors as was previously described 7 . Briefly, blood was mixed in a 1:1 ratio with autoMACS ® Rinsing Solution (Miltenyi Biotec, Germany), and subjected to gradient centrifugation on Lympholyte-H Cell Separation Media (Cedarline, Canada) at 800g at room temperature for 20 min. The mononuclear cell fraction was collected and washed twice with Rinsing Solution (300g at 11 °C for 20 min). At this stage, an aliquot (100,000 cells) of PBMC was collected for further flow cytometry analysis. The rest of the cells were subjected to immunomagnetic sorting on a manual MidiMACS™ Separator by using LS, LD columns and magnetic anti-CD14, anti-CD16 beads (Miltenyi Biotec, Germany) in accordance to the manufacturer's protocols. To obtain CD14++ monocytes, www.nature.com/scientificreports/ positive selection using CD14 magnetic beads was performed. CD16++ monocytes were obtained using negative selection with FC-blocking reagent and depletion cocktail (the stage of depletion of granulocytes and NK cells), and further positive selection of pre-enriched target cell fraction using CD16 magnetic beads. The purity of each CD14++ and CD16++ monocyte sample was at least 90% (Figs. S1, S2). Cell count and viability was assessed on a TC20 analyzer (Bio-Rad, USA). For the experiment with the patient's plasma CD14++ monocytes were placed in 6-well cell-culture plates (Costar, USA), at a density of 0.5 × 10 6 cells per well and incubated for 24 h in cell medium (RPMI supplemented with 2 mM l-glutamine, 25 U/mL penicillin, 25 μg/mL streptomycin and 10% patient's plasma instead of FBS). RNA extraction and microarray assay. Peripheral venous blood samples were collected prior to the caesarian section in EDTA-treated tubes. CD14++ and CD16++ monocytes were isolated using magnetic sorting as described above and immediately lysed in 1 mL QIAZOL reagent for the following RNA extraction according to the manufacturer's instructions (Qiagen). Biotinylated cRNA was prepared according to the standard Affymetrix protocol. Following fragmentation, 10 μg of cRNA were hybridized for 16 h at 45 °C on GeneChip™ Human Gene 2.0 ST Array. GeneChips were washed and stained in the Affymetrix Fluidics Station 450. GeneChips were scanned using the Affymetrix GeneChip Scanner 3000 7g. The data were analyzed with Transcriptome Analysis Console (TAC) Software version 4.0.2 using Affymetrix default analysis settings and RMA algorithm.

Enrichment analysis. Panther (Protein ANalysis THrough Evolutionary Relationships) gene classification
analysis was used for relating gene sequence to specific molecular functions, biological processes, and pathways 39 .
Fisher's exact test with FDR correction was performed for estimation of statistical over-representation.

Real-time polymerase chain reaction (RT-PCR) and gene expression assay.
Total RNA was isolated from CD14++ cells using QIAZOL (Qiagen); the corresponding cDNA was synthesized by using the MMLV-RT kit (Evrogen, Russia). The polymerase chain reactions were set in triplicates with qPCRmix-HS SYBR master mixes containing Sybr Green I dye (Evrogen, Russia). The target-specific primers (Table S4 in Supplementary Materials) were designed by using the Primer-BLAST tool (NCBI, USA) in accordance with the general rules and custom ordered from Evrogen. Expression was quantified by the threshold cycle (Ct) approach; the relative expression levels were evaluated via an algorithm proposed by M. Pfaffl with B2M as a housekeeping reference target 54 .
Statistical analysis. The data is presented as mean ± standard error mean (SEM) and mean ± standard deviation (SD). The Shapiro-Wilk test was used to estimate the normality of distributions. One-way analysis of variance (ANOVA) followed by Tukey's post-hoc test was used to identify differences among multiple groups with normal distribution, while the Kruskal-Wallis non-parametric ANOVA followed by the post-hoc Dunn test was used to calculate statistical differences for non-normal distributions. Data from two groups were compared by Mann-Whitney test or t test depending on the distribution. All calculations were performed by Prism 7.0 software (GraphPad, USA). p-value < 0.05 was considered significant i.e. indicative for differences in comparison to the control.

Data availability
This study did not generate new unique reagents. The original contributions presented in the study are included in the article/supplementary files; further inquiries can be directed to the corresponding author. The data reported in this paper have been deposited into the Gene Expression Omnibus (GEO) database, https:// www. ncbi. nlm. nih. gov/ geo (GEO accession: GSE186819). www.nature.com/scientificreports/