Hypoxia-induced small extracellular vesicle proteins regulate proinflammatory 2 cytokines and systemic blood pressure in pregnant rats

Small extracellular vesicles (sEVs) released from the extravillous trophoblast (EVT) are known to regulate uterine spiral artery remodeling during early pregnancy. The bioactivity and release of these sEVs differ under differing oxygen tensions and in aberrant pregnancy conditions. Whether the placental cell-derived sEVs released from the hypoxic placenta contribute to the pathophysiology of preeclampsia is not known. We hypothesize that, in response to low oxygen tension, the EVT packages a specific set of proteins in sEVs and that these released sEVs interact with endothelial cells to induce inflammation and increase maternal systemic blood pressure. Using a quantitative MS/MS approach, we identified 507 differentially abundant proteins within sEVs isolated from HTR-8/SVneo cells (a commonly used EVT model) cultured at 1% (hypoxia) compared with 8% (normoxia) oxygen. Among these differentially abundant proteins, 206 were upregulated and 301 were downregulated (p < 0.05), and they were mainly implicated in inflammation-related pathways. In vitroincubation of hypoxic sEVs with endothelial cells, significantly increased (p<0.05) the release of GM-CSF, IL-6, IL-8, and VEGF, when compared to control (i.e., cells without sEVs) and normoxic sEVs. In vivo injection of hypoxic sEVs into pregnant rats significantly increased (p < 0.05) mean arterial pressure with increases in systolic and diastolic blood pressures. We propose that oxygen tension regulates the release and bioactivity of sEVs from EVT and that these sEVs regulate inflammation and maternal systemic blood pressure. This novel oxygen-responsive, sEVs signaling pathway, therefore, may contribute to the physiopathology of preeclampsia.

Small extracellular vesicles (sEVs) released from the extravillous trophoblast (EVT) 22 are known to regulate uterine spiral artery remodeling during early pregnancy. The 23 bioactivity and release of these sEVs differ under differing oxygen tensions and in 24 aberrant pregnancy conditions. Whether the placental cell-derived sEVs released 25 from the hypoxic placenta contribute to the pathophysiology of preeclampsia is not 26 known. We hypothesize that, in response to low oxygen tension, the EVT packages 27 a specific set of proteins in sEVs and that these released sEVs interact with 28 endothelial cells to induce inflammation and increase maternal systemic blood 29 pressure. Using a quantitative MS/MS approach, we identified 507 differentially 30 abundant proteins within sEVs isolated from HTR-8/SVneo cells (a commonly used 31 EVT model) cultured at 1% (hypoxia) compared with 8% (normoxia) oxygen. Among 32 these differentially abundant proteins, 206 were upregulated and 301 were 33 downregulated (p < 0.05), and they were mainly implicated in inflammation-related 34 pathways. In vitro incubation of hypoxic sEVs with endothelial cells, significantly 35 increased (p<0.05) the release of GM-CSF, IL-6, IL-8, and VEGF, when compared to 36 control (i.e., cells without sEVs) and normoxic sEVs. In vivo injection of hypoxic sEVs 37 into pregnant rats significantly increased (p < 0.05) mean arterial pressure with 38 increases in systolic and diastolic blood pressures. We propose that oxygen tension 39 regulates the release and bioactivity of sEVs from EVT and that these sEVs regulate 40 inflammation and maternal systemic blood pressure. This novel oxygen-responsive, 41 sEVs signaling pathway, therefore, may contribute to the physiopathology of 42 preeclampsia. Optimal pregnancy outcome is dependent upon successful fertilization, endometrial 82 implantation, and placentation to support blastocyst development (1). Extravillous 83 trophoblast plays a significant role in establishing feto-maternal circulation via 84 remodeling of the uterine spiral arteries and placentation (2). During early 85 pregnancy (< 10-12 weeks), endovascular extravillous trophoblasts occlude uterine 86 spiral arteries to maintain a low oxygen environment (~2-3% O 2 ), which is essential 87 for normal embryogenesis and organogenesis (2). Subsequently, extravillous 88 trophoblast replaces the vascular endothelial and smooth muscle cells to remodel 89 the uterine spiral arteries with the formation of high capacitance and low resistance 90 vessels, enabling adequate placental perfusion (3). In addition, extravillous 91 trophoblast invades the uterine glands and veins and connect all these luminal 92 structures to form the inter-villous space (4). When extravillous trophoblast invasion 93 fails to occur or is dysfunctional, uterine spiral arterial remodeling is inadequate, and 94 placental function is suboptimal, resulting in placental hypoxia and the development 95 of pregnancy pathologies such as preeclampsia (5). 96

97
Preeclampsia affects approximately 8% of pregnancies worldwide and is recognized 98 to cause 60,000 maternal deaths and 500,000 neonatal deaths from preterm delivery 99 each year (6). This condition is characterized as early-onset (that develops before 34 100 weeks of gestation), or late-onset (develops at or after 34 weeks of gestation). The capacity of exosomes to induce changes in the target cells is mediated by the 137 specific delivery of bioactive molecules, such as proteins and miRNAs (23, 24). 138 Recently, using a longitudinal study design, we reported that the miRNA content 139 within exosomes changes in preeclamptic compared to normotensive pregnancies 140 (14). In addition, oxygen tension regulates the miRNA profile of EVT-derived 141 exosomes (21). The biological effects of oxygen tension on the protein profile of 142 EVT-derived exosomes, however, have yet to be described. Sammar et al. 143 investigated the level of expression of placental protein 13 in syncytiotrophoblast- Poor placentation associated with a failed invasion of the EVT is a feature of 155 preeclampsia and is associated with hypoxia and oxidative stress. We hypothesize 156 that, in response to low oxygen tension, the EVT packages a specific set of proteins 157 in sEVs and that these released sEVs interact with endothelial cells to induce 158 inflammation and increase maternal systemic blood pressure. To test this 159 hypothesis, small EVs were isolated from a transformed extravillous trophoblast cell PBS. Five-hundred-microliter of 12 fractions were collected, and particle 213 concentration determined using nanoparticle tracking analysis (NAT, NanoSight). 214 High particle fractions were pooled and stored at -80°C until sEVs analysis. sEVs 215 were characterized by size distribution, the abundance of proteins associated with 216 sEVs (i.e., CD63, sc15363 [1:1000] and TSG101, EPR7130 [1:1000]) and 217 morphology using Nanoparticle Tracking Analysis (NTA), Western blot analysis and 218 electron microscopy, respectively as previously described (16). sEVs were quantified 219 using an electrochemical exosome detection method, as we previously described 220 (28). Samples were suspended in PBS and divided into several aliquots after the 221 isolation and stored immediately at -80°C. To thaw the sEVs, samples were taken 222 out from -80°C and maintained at 4°C in ice until completion of the thawing process. 223 The protein concentration and the number of vesicles were quantified immediately 224 after the isolation and also after thawing at 4°C to evaluate the stability and yield of 225 the vesicles under the storage conditions. No differences were observed in the 226 protein concentration and yield (i.e., vesicles/protein) after the thawing process. This 227 data is consistent with our previously published studies in which no significant 228 differences were observed using fresh or frozen plasma in exosome quantification, 229 exosomal marker expression, microRNA expression and protein content(13). All 230 samples were stored and thawing with the same procedure, discarding that the 231 differences observed at the endpoint experiments are due differences to stored and 232 thawing protocols. 233 Quantitative Mass spectrometry analysis of exosomes 234 In-gel Digestion. A local ion library was generated to use in the Sequential Window 235 Acquisition of All Theoretical mass spectra (SWATH) mass spectra analysis using 236 an in-gel digestion method. Briefly, two protein pools were prepared from exosomes 237 obtained from 8% and 1% oxygen. The samples were mixed with Bolt™ LDS 238 sample buffer (ThermoFisher), sonicated for 5 min and heated at 95°C for 5 min. Microapp: 2 peptides per protein, 3 transitions per peptide, peptide confidence 299 threshold corresponding to 1% global FDR and FDR threshold of 1% was used. 300 The retention time was then manually realigned with a minimum of 5 peptides with 301 high signal intensities and distributed along the time axis. The resulting peak area 302 for each protein after SWATH processing was exported to MarkerView 1.3.1 303 (SCIEX) for statistical analysis. The resulting data were normalized using the Total 304 Area Sums (TAS) approach. The coefficient of variation in the abundance of 305 peptides across the samples was established by comparing SWATH peptide ion 306 against the IDA library. For independent samples, t-tests were used to compare 307 protein expression between exosomes from cells cultured to 8% and 1% oxygen. 308 The proteins with p<0.05 were considered as statistically significant.    (Table S1) and analyzed using IDA and SWATH. To evaluate whether 382 hypoxia changes the protein profile within sEVs from HTR-8/SVneo cells, we 383 analyzed data using an unsupervised principal component analysis (PCA) with 384 Qlucore Omics Explorer. With the first three PCA components explaining >90% of 385 the total variance, the generated PCA plot revealed that the sEVs from hypoxic and 386 normoxic groups had distinct protein contents ( Figure 2B). The variation in the 387 relative abundance of exosomal proteins between sEVs from hypoxic and normoxic 388  Figure 3A; with the most 402 significant difference in the sEVs protein profiles between these groups were 403 associated with Eukaryotic Initiation Factor 2 (EIF2; a signaling pathway that 404 activates vascular endothelial growth factor, VEGF signaling, and with glucocorticoid 405 receptor signaling pathway). Interestingly, the majority of the pathways were 406 associated with inflammation, and the top 25 canonical pathways with the common 407 genes (network/overlap) are presented in Figure 3B. Many of the differentially 408 expressed genes are present in multiple pathways related to inflammation. Finally, 409 GSEA of the total protein profile revealed several gene sets that were significantly 410 enriched in sEVs derived from hypoxic compared with normoxic cells. This is 411 illustrated by the normalized enrichment score. There was an enrichment of proteins 412 involved in MYC targets, hypoxia, and epithelial to mesenchymal transition 413 suggesting that these biological processes might be regulated by the hypoxic sEVs 414 ( Figure 3C). 415 416

Effect of HTR-8/SVneo cells-derived sEVs on cytokines releases from 417 endothelial cells 418
The effect of hypoxic and normoxic sEVs on the release of IL-6, IL-8, VEGF, and 419 Figure 4. sEVs derived from hypoxic 420 EVT dose-dependently increased (p <0.05) the release of all cytokines from 421 endothelial cells when compared to controls (without sEVs) or sEVs from cells 422 cultured at 8% oxygen (normoxic control). 423 424

Effect of EVT-derived exosome in systemic blood pressure in pregnant rats. 425
The mean litter size and maternal weights were similar between hypoxic (1% 426 oxygen) and control (8% oxygen) groups. Fetal weights (8% O 2 : 2.59 ± 0.06 g; 1% 427 O 2 : 2.47 ± 0.05 g), placental weights on GD 21 (1%: 0.48 ± 0.09 g; 8%: 0.50 ± 0.05 g) 428 were comparable between the two groups. Rats are nocturnal animals, and 429 continuous monitoring of blood pressure by telemetry revealed a characteristic 430 circadian pattern with higher arterial pressure and heart rate values during the dark 431 cycle (active phase) compared to the light cycle. In animals injected with normoxic 432 sEVs, MAP progressively decreased from GD 16 and reached a nadir on GD21, 433 which was comparable to the MAP in the saline-injected group. Pregnant rats 434 injected with hypoxic sEVs had significantly higher MAP starting from GD18 to GD21 435 compared to the respective time point in the control group ( Figure 5A; n = 6 rats in 436 each group; P < 0.05). The changes in MAP correlated with a significant increase in 437 systolic blood pressures in the hypoxic compared to the control group ( Figure 5B; n= 438 6 rats in each group). The diastolic blood pressure increased only in the later part of 439 gestation (i.e., GD 20-21) in the hypoxic compared to the control group ( Figure 5C; 440 n= 6 rats in each group). No differences in heart rate were observed between the 441 hypoxic and control groups ( Figure 6; n = 6 rats in each group). 442

443
The data obtained in this study are consistent with the hypothesis that the protein 444 content of EVT sEVs is programmed by low oxygen tension to be pro-inflammatory 445 (i.e., increasing the release of the IL-6, IL-8, VEGF and CS-GMS from target cells) 446 and to promote hypertension. 447 The bioinformatic analysis revealed that in normoxic (8% oxygen) conditions, the 448 proteins in EVT sEVs are associated with EIF2 signaling that activates the VEGF 449 signaling pathway. VEGF is a protein mediator that is synthesized and secreted by 450 placental macrophages. VEGF binds as a ligand with the soluble fms-like tyrosine 451 kinase (sFLT-1) receptor (also described as VEGF receptor 1) expressed on the 452 surface of vascular endothelial and smooth muscle cells (33). It also binds with the 453 kinase insert domain (KDR) receptor (also described as VEGF receptor 2), which is 454 which are known primary EVT (epithelial) markers (57-59). Furthermore, genome-529 wide gene expression profiles showed that the molecular signature of HTR8/SVneo 530 cells was vastly different from that of primary EVTs (60). Therefore, results obtained 531 from HTR-8/SVneo cells must be further verified using the appropriate primary EVT 532

cells. 533
Based on the data obtained, we suggest that hypoxia alters the content of EVT sEVs 535 and that these changes contribute to the physiopathology of preeclampsia. The 536 changes in protein and miRNA expression promote an inflammatory environment 537 within uterine spiral arteries and cause hypertension during pregnancy. These 538 changes are likely to occur in pregnancies characterized by compromised placental 539 perfusion and ischemia (expressed as preeclampsia and intrauterine growth 540 restriction) as an adaptive response aiming to improve placentation.