**3. Discussion**

Although the etiology of preeclampsia is still elusive, recent evidence has demonstrated that MSCs of different origins, when directly infused in animal models, are able to ameliorate PE-like symptoms, thus suggesting their potential role as therapeutic agents [20,21,24]. Nevertheless, systemic administration of living cells implies significant biosafety and ethical issues that must be considered when designing therapies for such a sensitive environment as pregnancy. Intravenously injected GFP-labeled umbilical cord MSCs were detected in the renal parenchyma and placenta of PE pregnant rats and in fetal kidneys, liver, lungs, and heart [21]. Moreover, locally administered cells often die before they significantly contribute to the healing response due to poor diffusion of nutrients and oxygen [29]. Alternative approaches are therefore mandatory.

The conditioned media obtained from MSCs consist of biologically active molecules whose function is to simultaneously modulate key biological mechanisms such as inflammation and immune response, angiogenesis, cell proliferation, apoptosis, and senescence [19,30–33]. Therefore, MSCs-derived CM can be investigated as an alternative approach to cell therapy.

MSCs have been successfully isolated from a variety of tissues, including adipose tissue, pancreas, and umbilical cord blood [34–36]. In the present study, we used physio-

logical MSCs derived from placental chorionic villi because of their immunosuppressive, pro-angiogenic, anti-inflammatory, and cytoprotective activities potentially useful against the aberrant placentation typical of PE [19,31,35,37–39]. In our previous works we found that PDMSCs-CM is able to modulate in vitro the expression of inflammatory cytokines, angiogenic factors, senescence markers, and cell cycle modulators in the placental villi [19,31].

Herein, we performed intravenous PDMSCs-CM administration to investigate the paracrine effects of placental mesenchymal cells in an LPS-induced mouse model of PE. LPS was chosen to mimic PE symptoms since it induces generalized endothelial dysfunction via the activation of inflammatory pathways [40]. In accordance with previous reports, in our model, LPS exposure led to gestational hypertension and proteinuria [24,41,42], as observed in PE pregnancies. Importantly, we demonstrated that systemic administration of PDMSC-CM, and not of living cells, significantly reduced LPS-induced gestational hypertension and proteinuria.

In line with our data, other groups reported the efficacy of MSCs in reducing hypertension in vivo [22,24,43]. It was suggested that the mechanism by which MSCs may ameliorate hypertension is through the modulation of endothelium-derived factors that control vasodilatation, vasoconstriction, and microvascular density increase [44]. In twokidney, one-clip rats, MSCs minimized hypertension at least in part by interfering with sympathetic nerve activity leading to reduction of sympathetic activity in the cardiovascular system [45].

As mentioned above, our results were obtained by injecting PDMSCs conditioned media and not living cells, thus opening the door to less invasive, more ethical, and safe therapeutic approaches for preeclampsia. PDMSCs-CM composition is complex and it includes free proteins, small molecules, and extracellular vesicles which can be further divided into apoptotic bodies, microparticles, and exosomes [46,47]. We previously published PDMSCs-CM partial characterization performed by cytokine array technology [19], demonstrating the presence of cytokines and chemokines in the conditioned media. In particular, we described Interleukin 8 (IL-8), Osteopontin, Tissue Inhibitor of Metalloproteinases-2 (TIMP-2), Neutrophil Activating Peptide 2 (NAP-2), Monocyte Chemotactic Protein-1 (MCP-1), Osteoprotegerin, Transforming Growth Factor-b2 (TGF-b2), Interferon-inducible protein-10 (IP-10), GRO, Vascular Endothelial Growth Factor (VEGF), Placental Growth Factor (PlGF) and Interleukin 10 (IL-10) as components of physiological PDMSCs-CM [19]. Despite that some of these molecules were classified as proinflammatory and Th1 mediators increased in PDMSCs and maternal blood from PE patients, they are also pivotal for physiological embryo implantation, placentation, and maternal–placental vascular remodeling [48–50]. For example, IL-8 is able to specifically counteract inflammation at the endothelial level [51], while TIMP-2 and IL-10 are powerful anti-inflammatory molecules, suggesting a multifactorial action directed against the exacerbated inflammation typical of PE. Moreover, since VEGF/sFlt-1 unbalance is widely accepted as the main trigger for the endothelial dysfunction, leading to hypertension in preeclampsia [52–54], we hypothesize that PDMSCs-CM counteracted endothelial dysfunction in our LPS-induced PE model via pro-angiogenic VEGF modulation and anti-angiogenic sFlt-1 inhibition. This hypothesis is in line with our previous findings showing that PDMSCs-CM promoted placental VEGF accumulation and sFlt-1 downregulation in physiological human villous explants [19].

Next, we described that PDMSCs-CM administration induced a significant reduction in LPS-induced urinary protein excretion. Similar effects were observed when human placental expanded (PLX-PAD) mesenchymal-like cells were injected in two models of innate immunity-induced PE, thus confirming PDMSCs potential to reduce PE-associated proteinuria [22]. Chatterjee et al. suggested that the release of paracrine factors led to the observed decrease in oxidative stress, angiogenesis, inflammation, and endothelial dysfunction, without the need of cell–cell contact. Since endothelial damage underlies many PE manifestations, including proteinuria [55], our data are in line with the above mentioned results, confirming the ability of human PDMSCs to counteract vascular dysfunction throughout the release of trophic mediators, as VEGF that it is able to stimulate endothelial repair and reduce circulating sFlt-1 levels. In our model, LPS-induced proteinuria slowly decreased with advancing gestation also in animals infused with plain media, even if to a significantly lesser extent relative to CM animals. This effect is likely due to spontaneous LPS clearance and consequent inflammation remission.

Moreover, we reported that LPS followed by plain media infusion in pregnant mice induced reduction in fetuses' number, fetal absorption, and decrease in fetal and placental weight, thus mimicking placenta development anomalies typical of PE. Similar results were reported by Rivera et al., showing that a 100 µg/kg day LPS for 7 days in pregnant rats significantly reduced fetal size and increased fetal demise [56]. In stark contrast, we demonstrated that PDMSCs-CM treatment resulted in a higher number of fetuses, increased fetal–placental weight, and no fetal absorption. These outcomes are likely due to decreased endothelial damage and placental inflammation as demonstrated by placental sFlt-1, TNF-a, and IL-6 downregulation, reduced maternal blood pressure and proteinuria derived from improved placental functionality, and fetal nutritional status as previously suggested [43,57]. Importantly, we specifically investigated TNF-α, Il-6, and sFlt-1 because they are key players in PE pathogenesis [58] and severe endothelial dysfunction [54,59].

To investigate the impact of PDMSCs-CM on liver and renal functions, serum AST-ALT and creatinine–urea levels were monitored. It was previously described that after MSCs injection (e.g., bone marrow MSC, hematopoietic stem cells, umbilical cords MSC, amniotic fluid MSCs), AST, ALT, creatinine, and urea were normalized to physiological levels restoring compromised liver and kidneys activities [60–62]. In our model, AST, ALT, creatinine, and urea serum levels were not modified by LPS injection nor by PDMSCs-CM/plain media infusion, in line with data reported by Oludare and colleagues that described no significant differences in liver enzymes' levels in LPS mice compared to controls [63]. Importantly, our results demonstrated that PDMSCs-CM did not affect liver and kidney functionalities, thus excluding adverse effects.

In conclusion, our findings demonstrated that human PDMSCs-CM administration significantly ameliorated PE-like symptoms and improved fetal–placental outcomes in pregnant LPS mice. Our data strongly suggested that PDMSCs-CM acted through the restoration of endothelial function and the suppression of the proinflammatory cascade, thus contrasting systemic and placental injury. A limitation of our study was that we have not yet completed PDMSCs conditioned media characterization; therefore, we could not indicate its exact mechanism of action. Due to CM complexity, different pathways were most likely promoted and/or suppressed by PDMSCs trophic mediators, thus explaining the multitarget therapeutic activity demonstrated by our results. Indeed, PDMSCs-CMbased therapy could be considered a promising tool due to its ability to simultaneously target different drivers of the preeclamptic syndrome. This approach could minimize the biological variability, biosafety, and ethical issues of cell-based therapies, thus leading to the development of a safe and ethical cell-free strategy against preeclampsia. Further investigations are required.
