INTEGRINS MEDIATE PLACENTAL EXTRACELLULAR VESICLE TRAFFICKING TO LUNG AND LIVER IN VIVO

Membrane-bound extracellular vesicles (EVs) mediate intercellular communication in all organisms, and those produced by placental mammals have become increasingly recognized as significant mediators of fetal-maternal communication. Here, we aimed to identify maternal cells targeted by placental EVs and elucidate the mechanisms by which they traffic to these cells. Exogenously administered pregnancy-associated EVs traffic specifically to the lung; further, placental EVs associate with lung interstitial macrophages and liver Kupffer cells in an integrin-dependent manner. Localization of EV to maternal lungs was confirmed in unmanipulated pregnancy using a transgenic reporter mouse model, which also provided in situ and in vitro evidence that fetally-derived EVs, rarely, may cause genetic alteration of maternal cells. These results provide for the first time direct in vivo evidence for targeting of placental EVs to maternal immune cells, and further, evidence that EVs can alter cellular phenotype.


INTRODUCTION
in regulating immune and cardiovascular responses to pregnancy and parturition. Further, 23 interest in placental EVs as biomarkers for these and other complications has burgeoned. 10 24 In vitro studies have suggested that placental exosomes can be internalized by and 25 affect activity of multiple immune cell types. Placental exosomes can induce migration and 26 cytokine secretion by macrophages, 11 suppress NK cells and T cells, [12][13][14] protect cells against 27 viral infection, 15 and may promote spiral artery remodeling. 16 However, only a few studies have 28 queried their trafficking patterns and functional effects in vivo. Tong et al. found localization of 29 human placental exosomes in the lungs, kidney, and liver of mice 24 hours after intravenous 30 injection. 17 On the other hand, intraperitoneally injected exosomes purified from plasma of 31 pregnant mice traffic to the uterus and cervix, and further, can cross into the fetus. 18 In this 32

A.
Plasma EVs (2.5x10 10 ) isolated from the plasma of nonpregnant (NP) or GD14.5 pregnant mice were labeled with PKH26 and administered i.v. into NP mice. Mice were sacrificed after 30 minutes, and lung and liver were analyzed by epifluorescence microscopy. B.
Representative fluorescence microscopy of lung from a mouse treated with plasma EVs from nonpregnant (NP; upper left) or GD14.5 pregnant (upper right) mice (Arrowheads, EVs). C.
Quantification of plasma EVs as evidenced by presence of red fluorescence in recipient lung. Dots represent biological replicates, Wilcoxon test. D.
Representative fluorescence microscopy of liver from a mouse treated with plasma EVs from nonpregnant or GD14.5 pregnant mice EVs (Arrowheads, plasma EVs). E.
Quantification of plasma EVs as evidenced by presence of red fluorescence in recipient liver. Dots represent biological replicates, Wilcoxon test.
Placental EVs traffic to lung interstitial macrophages in vivo 1 We next sought to identify the type(s) of cells in the lung with which placental EVs associate. To this end, we developed an explant system in which placental EVs are purified following 18 hours of culture at ambient oxygen concentration (Suppl. Fig. 2). EVs from the supernatant of GD14.5 placental explants were isolated, labeled with PKH26, and administered intravenously into nonpregnant female mice ( Fig. 2A).

Figure 2. Placental EV trafficking in vivo.
A. EVs were purified from the medium of GD14.5 placental explants; a representative TEM is shown. After labeling with PKH26, 2.5x10 10 EVs were administered i.v. into into nonpregnant females. After 30 minutes mice were sacrificed, and lung and liver harvested and processed for downstream analysis. B.
Representative t-SNE analysis of dispersed and murine lung cells after treatment of mice with GD14.5 placental EVs. Cell suspensions were stained with the fluorophoreconjugated antibodies indicated above each panel and analyzed by flow cytometry. C.
Immunofluorescence microscopy of lungs from mice treated with GD14.5 placental EVs. Arrows, pEV foci (red), one of which in this image colocalizes with LYVE1 (green) and CD68 (yellow). Representative images of random fields chosen from three independent experiments.

Figure 3. Flow cytometric analysis of pEV trafficking in murine lung.
A. Representative flow cytometry plots identifying macrophages (M0), alveolar macrophages (AM), and interstitial macrophages (iM) from murine lung after treatment of mice with GD14.5 placental EVs. B.
Autofluorescence signature of macrophage, alveolar macrophage, and interstitial macrophage populations.
Outer membrane proteins including integrins influence trafficking 1 of placental EVs to the lung and liver 2 We next asked whether outer membrane components, specifically integrins, influence 3 trafficking of placental EVs, as these proteins can mediate the trafficking of tumor-derived EVs 4 to specific organs. 21 A survey of multiple integrins revealed that placental EVs express ITG 3, 5 V, 5, 1, and 3, whereas ITG 6 was not detected in EVs despite its expression in the 6 placenta ( Fig. 4A). As expected, proteinase K treatment of EVs abolished immunoreactivity for 7 all membrane-associated integrins examined, although neither expression of the exosome 8 marker CD9 nor exosome morphology were affected (Fig 4A, B). 9 Western Blot analysis of integrin expression in murine placenta, placental EVs (pEV), and placental EVs treated with proteinase K (pEV + proK). B.
Representative transmission electron microscopy of untreated EVs (left) or EVs treated with proteinase K (right). C.
Confocal microscopy of murine lung after treatment of mice with GD14.5 placental EVs (upper left), proteinase K-treated EVs (upper right), or EVs pre-incubated with HYD-1 (lower left) or RGD (lower right) peptides. D.
Quantification of EV positive cells in lungs treated with placental EVs. E.
Confocal microscopy images of the liver after treatment of mice with placental EVs (upper left), proteinase K-treated EVs (upper right), or EVs pre-incubated with HYD-1 (lower left) or RGD (lower right) peptides. F.
Quantification of EV-positive cells in liver of mice treated with placental EVs. G.
CUBIC-cleared lung from mouse treated with placental EVs. White arrowheads represent placental EV foci. H.
To determine whether pEV localization to lung and liver was mediated by outer 1 membrane proteins, EVs were digested with proteinase K, labeled, and administered into 2 nonpregnant mice. Proteinase K abolished localization of EVs to the lung and liver (Fig. 4C-F) 3 of EV-treated mice. To account for the possibility that proteinase K caused degradation of EVs, 4 we also administered 10-fold more EVs, and still observed vastly reduced numbers of EVs in 5 liver and lung (Fig. 4D, F). Consistent with this observation, treatment with proteinase K to 6 remove outer membrane proteins reduced the localization of EVs to lung interstitial 7 macrophages (Suppl. Fig. 4). 8 To test the hypothesis that integrins mediate placental EV targeting to the lung, we pre-9 incubated EVs with RGD and HYD-1 peptides, which block binding of integrins 5 1/ V 3 and 10 3 1 respectively, to their substrates 22,23 prior to intravenous administration into nonpregnant 11 recipients. Mice were sacrificed 30 minutes after treatment, and lung and liver were analyzed 12 by fluorescence microscopy (Fig. 4 C, E). Although proteinase K ablated localization of 13 placental EVs to both lung ( placental EVs remained detectable in both tissues after pretreatment with integrin-blocking 15 peptides (Fig. 4C, E, lower left panels). In the lung, neither peptide significantly reduced the 16 total numbers of placental EVs (Fig. 4D). In the liver, however, RGD pretreatment resulted in a 17 significant reduction of placental EV localization (Fig. 4F). 18 Although the total numbers of placental EVs that localized to the lung were not reduced 19 by HYD-1, we noticed that the tissue distribution of the vesicles was altered, with EVs appearing 20 to remain within large vessels (Fig. 4C, lower left panel). To better characterize the effect of 21 HYD-1 treatment on placental EV localization to the lung, 200-micron lung sections were 22 cleared by CUBIC and imaged by 3D confocal microscopy. 24 Strikingly, we saw localization of 23 placental EVs exclusively within the vasculature, suggesting that they were blocked from 24 entering the interstitial tissue (Fig. 4G, H). 25 Since HYD-1 pretreatment inhibited rapid migration of placental EVs to lung interstitium, 26 we asked whether this inhibition is sustained. Using Li-Cor whole organ imaging as a high 27 throughput screening method, we labeled EVs with near-infrared (NiR) dye, administered them 28 i.v. into nonpregnant recipients, and quantified vesicle localization (Fig. 5A). Untreated EVs 29 remained detectable in the lung after 24 hours, but pretreatment with HYD-1 significantly 30 decreased their presence (Fig. 5B). RGD pretreatment tended to reduce localization to the lung 31 at 24 hours, but the result was not statistically significant. 32 Near-infrared imaging of whole lung after in vivo treatment with labeled placental EVs after 24 hours. Image is representative of three independent experiments. B.
Quantification of NIR placental EV positive area from whole lung treated with placental EVs. a.u., arbitrary units; each point represents one biological replicate. C.
Representative whole organ NIR imaging of mice treated with control medium (VEH) or placental EVs (pEV) after 24 hours. Representative of three independent experiments. 1 lymph node, spleen, small intestine, uterus, and ovaries, again reasoning that EVs in the 2 maternal vasculature can potentially access any maternal organ. Twenty-four hours after 3 administration of NIR-labeled placental EVs, we did not observe any significant changes in NIR 4 signal relative to the vehicle control treatments (Fig. 5C). This result suggests that placental 5 EVs are not detectable in these tissues using this model. 6 Placental EVs localize to maternal lung in unmanipulated 7 pregnancy and may genetically modify maternal lung cells 8 The above studies support the notion that placental EVs traffic to the maternal lung. 9 However, a single bolus of purified EVs does not recapitulate the continual release of EVs from 10 the placenta nor their sustained presence in maternal circulation. To address this limitation, we 11 generated a model in which fetal EVs could be detected within the context of normal pregnancy. Using this model, we asked whether fetal/placental mGFP+ EVs are carried to the lung. 1 Using confocal microscopy, we readily detected mGFP fluorescence in maternal lung from 2 GD14.5 CMV-Cre-mated mTmG females (Fig. 6D). Most of the signal we observed was 3 punctate and associated with maternal mTomato-expressing cells, similar to what was observed 4 when labeled placental EVs were administered exogenously. Interestingly, we also observed 5 mT/mG females mated to a CMV-Cre males give rise to female pups ubiquitously expressing mGFP and male pups expressing mTomato. B.
GD14.5 utero-placental interface of a female fetus from a mT/mG x CMV-Cre mating. Fetal mGFP-expressing placental cells (Sp; spongiotrophoblast) are readily distinguishable from maternal mTomato-expressing decidual cells (Dec); invading glycogen trophoblast (GTC) can also be seen. The dashed line demarcates the fetomaternal interface. Representative image of three independent experiments. C.
Confocal image of punctate mGFP-positive foci localization in maternal mT/mG GD14.5 lung after mating to a CMV-Cre male. Representative of n=5. E. E, F. Confocal localization of mG+ recombined cells in maternal mT/mG GD14.5 lung after mating to a CMV-Cre male. Images are representative of five biological replicates. White arrows; punctate mGFP-positive foci. Yellow arrows; mGFP-positive, recombined cells. mGFP expression that was clearly membrane-associated, surrounding distinct DAPI-stained 1 nuclei (Fig. 6E, F). These cells did not express mTomato, indicating that they had undergone 2 Cre-mediated recombination. We observed mGFP-associated foci and cells only in mT/mG 3 dams mated to CMV-Cre males, and not in those mated to WT males, therefore ruling out 4 random recombination of the mT/mG locus in our experimental animals. 5 Placental EVs can induce Cre-mediated recombination in vitro.

6
Recombined mGFP-positive cells observed in the lungs could represent fetal cells that 7 trafficked from the fetus/placenta to the maternal lungs (fetal microchimerism). 26-28 8 Alternatively, they could be maternal cells that underwent Cre-mediated recombination as a 9 result of EVs carrying Cre mRNA or protein to recipient cells. 29,30 To test the latter possibility, 10 we cocultured bone marrow-derived dendritic cells (BMDC) from mTmG mice with GD14.5 11 placental explants from CMV-Cre (colorless) or WT mice (Fig. 7A). 12 Explants were separated from the dendritic cells by a 70 µm insert such that EVs released into 1 the medium could access the underlying BMDC. After 18 hours, placentas were removed, and 2 BMDC were cultured for an additional five days to allow sufficient time for Cre recombination to 3 occur. 25,30 The BMDC were screened for recombined mGFP positive cells by flow cytometry. 4 While no changes in mTomato expression occurred in BMDC cultured with WT placentas (  mT/mG BMDCs were cultured with (colorless) GD14.5 CMV-Cre placentas for 24 hrs and cultured for an additional five days. B.
Representative flow cytometry plots of mT/mG BMDCs cocultured with WT (left) or CMV-Cre (right) placentas. C.
Schematic diagram of forward (mT-F/mG-F) and reverse (mT-R/mG-R) primers for identification of unrecombined mTomato and recombined mGFP. PCR amplification of genomic DNA with these primers results in PCR products of 376 bp for nonrecombined cells and 200 bp for recombined cells. D.
Genotyping electrophoresis gel of mGFP (top) and mTomato (bottom) loci from genomic DNA. WT: tail DNA from a wild type mouse; mT: tail DNA from an mTmG mouse in the absence of Cre; mT+mG: tail DNA from a fully mTmG mouse; WT plac+mT/mG: DNA from dendritic cells cocultured with WT placenta; CMV-Cre plac + mT/mG BMDC: DNA from dendritic cells cocultured with CMV-Cre placenta; NTC, no template control. Representative gel image of five independent experiments.
7B, left), culture with CMV-Cre placentas resulted in a downward shift of events into a new 1 mTomato-negative/mGFP negative population and an increase in the proportion of mTomato-2 negative/mGFP-positive population (Fig. 7B, right). 3 To confirm these results, we isolated the DNA from the cocultured BMDC and performed 4 PCR using primers designed to differentially amplify the non-recombined and recombined 5 mTmG locus (Fig. 7C). 31 We observed the mGFP amplicons in DNA isolated from BMDC 6 cocultured with CMV-Cre, but not WT placentas (Fig. 7D). Finally, confocal microscopy 7 conducted in parallel revealed the presence of cell-associated and non-cell-associated mGFP in 8 BMDC co-cultured with CMV-Cre placentas, but not WT placentas (Suppl. Fig. 5), which we 9 confirmed using immunofluorescence with an anti-GFP antibody (Suppl. Fig. 5B, D). 10 Interestingly, the morphology of dendritic cells co-cultured with CMV-Cre placentas revealed 11 large clusters of cells that were not observed when the cells were co-cultured withWT placentas 12 (Suppl. Fig. 5D). Our results demonstrate a proof of concept model for identifying recipient cells 13 that internalize placenta EVs and provides an additional mechanism for studying how placental 14 EVs induce biological effects on recipient cells. 15

16
Since the discovery of EVs and their effects on distant cells, research interest in the role 17 of placental EVs during pregnancy has greatly increased. The total quantity and concentration 18 of EVs in maternal plasma rise across gestation, with the placenta contributing significantly to 19 this increase. 5,6 While a number of effects of placental EVs have been suggested, few studies 20 have attempted to quantify their biodistribution in vivo. In this study, we show that placental EVs 21 traffic to the lung and the liver. Further, that trafficking to the lung appears to be specific to 22 placental EVs and pregnancy, as only plasma EVs isolated from pregnant dams, but not those 23 from non-pregnant mice, localized to the lung. 24 Our results align with those of Tong et al., who found that human placental EVs 25 administered into mice also localize to the lung and liver. 17 In our study, we add to these 26 findings by using homologous adoptive transfer of murine EVs to identify the cellular targets of 27 placental EVs in these tissues. Using multiparameter flow cytometry together with both targeted 28 and untargeted analyses, as well as immunofluorescence microscopy, we found that murine 29 placental EVs target lung interstitial macrophages and liver Kupffer cells. While this result is not 30 especially surprising, it reveals for the first time the bona fide in vivo targets of EVs and aligns 31 with earlier data showing that human trophoblast EVs are internalized by macrophages in 1 vitro. 11 2 EVs derived from Swan-71 cells, an extravillous trophoblast cell line, were previously 3 shown to induce migration and proinflammatory cytokine production by cultured macrophages. 11 4 Similarly, Southcombe et al. showed that syncytiotrophoblast microparticles, possibly including 5 exosomes, induce pro-inflammatory cytokine release from monocytes. 32 While the precise role 6 of this induction of cytokines by monocytes is uncertain, studies have supported the notion that 7 pregnancy is associated with a shift in inflammatory environment in general, with a bias towards 8 a proinflammatory milieu in early and late gestation, and anti-inflammatory milieu during mid-9 gestation. 33 10 Whether these observations hold true for the effects of EVs on interstitial macrophages 11 of the lung, which arise from monocytes and play a role in lung homeostasis, 19,34 is as of yet not 12 known, and the impact of placental EVs on the pulmonary physiology and pathology of the lung 13 during pregnancy warrants investigation. Pregnancy alters maternal pulmonary function 14 dramatically, with up to 20% increase in maternal oxygen consumption by term, 35 and 15 differential susceptibility to respiratory pathology caused by influenza virus; 36,37 varicella 16 virus; 38,39 asthma; 40 and cigarette smoking. 41 The notion that placenta-and pregnancy-17 associated EVs mediate these physiological and pathological adaptations to pregnancy via 18 effects on pulmonary immune cells including macrophages is untested but intriguing. 19 Another important function of macrophages is antigen presentation. Maternal T cells are 20 made aware of fetal antigen through indirect antigen presentation by maternal antigen 21 presenting cells, 42 which may include macrophages. Maternal T cells that recognize fetal 22 antigen do not mount an adverse immune response to fetal antigen, even when artificially 23 stimulated with high concentrations of adjuvant, 42 suggesting that antigen presenting cells in 24 pregnancy convey powerful tolerogenic signals. We and others have postulated that EVs are 25 the source of these tolerogenic signals, possibly through cargo that include potent suppressors 26 such as PDL1 and FASL, as well as fetal antigen itself. 13,14,43 Surprisingly, our current results 27 show minimal or no localization of placental EVs to the maternal spleen and lymph nodes, 28 where fetal antigen recognition by maternal T cells occurs, 42,44 but rather to lung interstitial 29 macrophages and Kupffer cells of the liver. Lung interstitial macrophages together with resident 30 dendritic cells are potent antigen presenting cells. 19 Future work can examine the role of 31 placental EVs in antigen presentation by these macrophages as a possible mechanism for 32 maternal immune system exposure and tolerance to fetal-placental antigens. 33 Our results highlight possible roles for integrins on placental EVs. The integrin profile in 1 placental EVs did not fully recapitulate integrin expression in the placenta: integrins 3, V, 5, 2 1, and 3 were expressed in both placenta and EVs, while 6 was found only in the placenta. 3 This suggests that proteins are selectively loaded into placental EVs during their biogenesis, a 4 notion supported by numerous other studies. 45 Additionally, removal of surface proteins, 5 including integrins, using proteinase K pre-treatment disrupted the trafficking pattern of placental 6 EVs. Further, pre-incubation of EVs with RGD peptide, which blocks ITG 5 1 and V 3 7 binding to its receptor, fibronectin, abrogated appearance of placental EVs in the liver, which is 8 rich in fibronectin. 46 Similarly, HYD-1, which blocks ITG 3 1 binding to the pulmonary 9 basement membrane glycoprotein laminin (LN)-5, 23,47 prevented entry of placental EVs into the 10 lung. This observation was supported by 3-dimensional imaging, which highlighted association 11 of EVs with lung endothelial cells, and confirmed that EVs remain restricted to large vessels in 12 the lung when pre-incubated with the HYD-1 peptide. 13 Placental EV adhesion and extracellular matrix proteins may also play an important role 14 in metastasis of choriocarcinoma. Trophoblast cells -the source of EVs in our studies -share 15 many properties with cancer cells, including epithelial-to-mesenchymal transition and invasion 16 into adjacent tissues during the physiological process of embryo implantation. 48 When 17 trophoblast cells become malignant, the lung is a primary site for metastasis. In a murine 18 xenograft model of cancer, EVs derived from metastatic breast cancer tumors selectively 19 trafficked to the lung and liver, which served to establish a niche for future metastasis to these 20 tissues. 21 Moreover, integrins played a major role in the selective targeting of EVs to the lung. 21 21 Thus, placental EV trafficking to the lung could play an unfortunate role in establishing a niche, 22 parallel to that found for other metastatic tumors. 23 A limitation of our studies is that bolus injection of purified EVs does not recapitulate the 24 physiological process of continuous EV release during pregnancy. We sought to address this 25 caveat in multiple ways. First, we used a quantity of EV that mimicked quantities found during 26 pregnancy. 6 Second, we administered EVs into the tail vein, reasoning that this route most 27 closely mimics hematogenous release of placental EV in situ, as both this and the uterine vein 28 ultimately drain into the inferior vena cava. Thus, although we were surprised that our results 29 did not show trafficking of placental EVs to the uterus, we believe that intravenous 30 administration simulates the hematogenous route that EVs travel during pregnancy. 31 A third way we addressed the limitations of bolus EV administration was to develop an in 32 vivo model without the need to isolate, label and administer them. In this system, fetal tissues 33 and EVs derived thereof express a mGFP reporter that enables their in vivo tissue and cellular 34 targets. EVs could be detected as punctate foci within the maternal lung pregnancy, mimicking 1 the pattern observed after intravenous injection. We propose that this model is most 2 representative date of continuous placental EV release and trafficking, and that it will allow 3 further studies on the effects of EVs on maternal physiology in vivo. 4 We also identified whole cells in the lung that had recombined to express mGFP but not 5 mTomato. Recombined mGFP-expressing cells in the maternal lung may arise from either or 6 both of two mechanisms: genetic recombination of maternal lung cells by placental EVs, or fetal 7 microchimerism. In our model, Cre recombinase is inherited paternally by the fetus, causing 8 recombination of the mTmG locus that switches cells from red to green fluorescence. Using 9 similar models, others have shown that Cre mRNA is carried by exosomes secreted by cancer 10 cells, and that these exosomes could thus induce recombination of neighboring or distant 11 cells. 30 To test whether this is possible in principle, we used a model system in which red 12 fluorescent mTmG dendritic cells were co-cultured with CMV-Cre placentas. Data generated by 13 immunofluorescence microscopy, flow cytometry, and genetic analysis support the notion that 14 shed vesicles from the explants induced recombination in the dendritic cells. seek to identify Cre recombinase protein or mRNA in EVs and will use in vivo approaches to 25 analyze directly whether placental EVs can induce genetic changes in maternal cells. 26 Collectively, we have established that placental EVs preferentially traffic to maternal 27 pulmonary interstitial macrophages and Kupffer cells in vivo. We demonstrate that integrin 3 1 28 is necessary for localization to the lung interstitial tissue, and that integrins 5 1 and V 5 are 29 necessary for localization to the liver. Future work will seek to identify how placental EVs 30 influence maternal interstitial macrophages and Kupffer cell function, as well as the physiology 31 of the lung and liver, during pregnancy. Additionally, we demonstrate a new model for further 32 expanding the field of understanding placental EV interactions in vivo and provide a framework 33 for visualizing maternal-fetal interactions without the use of exogenous purification and labeling, 34 providing a strong advantage to traditional methods of tracking placental EV kinetics in vivo in a 1 manner consistent with their physiological release across pregnancy. Penicillin/Streptomycin, 1mM Sodium pyruvate, 10% fetal bovine serum) was ultracentrifuged at 25 100,000 x g for 18 hrs at 4°C and sterilized through a 0.22 m filter. Placental explant culture 26 was performed as previously described; 53 briefly, individual GD14.5-16.5 placentas were cut into 1 four pieces and placed in 74 m mesh 15 mm net well inserts in a 12-well plate filled with 3 ml 2 of EV-depleted media for 18 hours at 37°C, 5% CO2. To determine optimal culture conditions, 3 viability of tissue explants was assessed by a board-certified veterinary pathologist (DA) by 4 examination of hematoxylin and eosin stained sections using a double blinded scoring system 5 for placental tissue necrosis, using a scale of 0 (no necrosis) to 5 (complete necrosis). 6 Following culture, the culture supernatant was centrifuged twice at 500 xg and 2000 xg for 15 7 minutes to pellet cells and cellular debris, respectively. Supernatant was filtered through a 8 0.22 m PVDF membrane filter and concentrated to a volume of 500 l at 3320 xg using 100 9 kDa molecular weight cut off Vivaspin centrifugal ultrafiltration columns (Sartorius, Stonehouse, 10 UK). The resulting filtrate was stored at -80°C until processed for EV isolation. 11 To ensure EVs were isolated from high quality tissue, we first tested the viability of the 12 cultured explants under various conditions, as well as the quality and quantity of EVs (Suppl. 13 Fig. 2A). We saw no significant differences in necrosis or EVs between freshly isolated and 14 cultured placentas, although the latter exhibited more variation. We observed the highest yield 15 of EVs from placentas cultured for 24-hour at ambient oxygen, and therefore used these 16 conditions for further studies. For low oxygen culture, tissues were placed in a sealed gas 17 chamber and purged of ambient air and charged with a gas mixture containing 8% oxygen, 5% 18 carbon dioxide before being sealed and placed in an incubator . 19 Extracellular vesicle isolation and labeling 20 Extracellular vesicles were isolated from plasma and concentrated supernatant of 21 explant cultures using qEVsingle or qEVoriginal size exclusion chromatography columns (Izon 22 Sciences, Medford, MA) following the manufacturer's instructions. Briefly, plasma or 23 supernatant were loaded onto columns equilibrated with PBS and fractions 6-8 or 7-9 (200 l for 24 qEVsingle, 500 l for qEVoriginal columns, respectively) were collected and concentrated to 50 25 l using Amicon Ultra 4 10kDa centrifugal filters for further analysis. 26 Purified EVs were labeled with PKH26 lipophilic red fluorescent dye (Sigma-Aldrich, St. 27 Louis, MO) using the manufacturer's instructions. Briefly, purified EVs or PBS (negative control 28 for injection) were diluted in Diluent C and pipetted into ultracentrifuge tubes containing PKH26 29 dye and Diluent C. The labeling reaction was stopped after 5 minutes with the addition of 10% 30 BSA and EV-depleted culture medium. To eliminate excess unbound dye, 0.971 M sucrose 31 was pipetted beneath the labeled EVs and centrifuged for 1.5 h at 150,000 xg at 4°C as 32 described. 54 The labeled EVs were washed and concentrated with Amicon Ultra 4 10kDa 1 centrifugal filters as described above. Prior to administration, we separated any remaining free 2 dye from labeled EV by centrifugation over a sucrose gradient; as a negative control, PKH26 3 dye alone was processed and administered in parallel. 4 For experiments in which whole organ imaging was performed, purified EVs were 5 labeled with 0.246mM (DiIC18 (7) (1,1'-Dioctadecyl-3,3,3',3'-Tetramethylindotricarbocyanine 6 Iodide)) DiR for 3 minutes and ultracentrifuged on a sucrose cushion as described above. After 7 centrifugation, the labeled vesicles were loaded on IZON single columns and fractions 6-8 were 8 collected and concentrated with 10kDa concentrators as described above and used for 9 downstream analysis. For proteinase K treatment, placental EVs were treated with 10 g/ml proteinase K in 23 PBS; for controls, proteinase K was added to PBS alone. Samples and controls were incubated 24 for 10 minutes at 37°C before heat inactivation at 65°C for 10 minutes followed by addition of 25 100 M phenylmethylsulfonyl fluoride (PMSF) protease inhibitor. 26 For inhibitory peptide experiments, RGD peptide was purchased from Sigma-Aldrich and 27 HYD-1 (KIKMVISWKG) was synthesized (Genscript, Piscataway, NJ). All reagents were 95% 28 or greater purity and resuspended in deionized water, aliquoted, and frozen. Labeled pEVs 29 were incubated with 0.6 M of inhibitory peptide and incubated at 37°C for 30 minutes before 1 intravenous administration into recipient mice. 2 mT/mG BMDC explant co-culture 3 Bone marrow derived dendritic cells (BMDC) were prepared from mT/mG mice as 4 described with modifications. 55 Ten million bone marrow cells/well were plated overnight in 5 medium containing 20ng/ml murine GM-CSF(BioLegend, San Diego, CA), replenishing the 6 medium daily for three days, then culturing for an additional three days before downstream 7 analysis. 8 For co-culture with placental explants, 100,000 BMDC were seeded in 12-well plates, 9 and CMV-Cre or WT placentas were added on Netwell inserts as described above. Following 10 overnight culture, the placentas were removed, and the BMDC were cultured for an additional 11 five days to allow for sufficient time for recombination, 25 replenishing the media every other day. 12 For microscopy, BMDC were seeded onto coverslips before placental co-culture, then fixed with 13 4% PFA and stained with 1:5000 rabbit anti-EGFP (Origene) and anti-rabbit AF647 ( Torrance, CA). For fluorescence microscopy, 5 m tissue cryosections were mounted with 10 DAPI nuclear stain and imaged on a Nikon Eclipse Ti epifluorescent microscope fitted with a 11 20x NA 0.5 plan fluor objective using Nikon NIS-Elements AR 4.40 software. To quantify EV in 12 tissues, five random fields were imaged, and individual color channels processed using ImageJ. 13 PKH26 and nuclei counts were computationally quantified using a custom pipeline developed in 14 CellProfiler 3.0 software. For immunofluorescence microscopy, slides were stained with 15 BioLegend antibodies and mounted with mounting medium with DAPI. 16 Tissue clearing and immunofluorescence analysis 17 For clearing tissues prior to 3D imaging, paraformaldehyde-fixed organs were embedded 18 in 2% agarose. Sections (200 m) were obtained using a Leica VT1200 S vibratome (Leica,19 Buffalo Grove, IL), placed in 12-well plates, and permeabilized in 0.05% tween 20. Sections 20 were blocked in 10% SuperBlock solution for 4-12 hrs, and stained with antibodies at 37°C in an 21 orbital shaker for 24 hours. DAPI (10ug/ml) was added to the solution and sections were 22 incubated for an additional 24 hours. After a 12-hour wash, secondary antibody was added 23 overnight at 37°C, and samples were again washed for 12 hrs. Samples were then immersed in 24 2-3ml of CUBIC1 solution and incubated overnight at 37° before being washed briefly with PBS 25 before immersed with 2-3ml of CUBIC2 solution for 12hrs. 24 26 For traditional immunofluorescence, paraformaldehyde-fixed tissues were placed in 30% 27 (W/V) sucrose solution in PBS overnight before being embedded in optimal cutting tissue 28 (O.C.T.) compound (Sakura Finetek, Torrance, CA) and freezing. Cryosections (7 m) were 29 placed onto charged glass slides, blocked with 10% SuperBlock and 3% goat serum, and 30 and sections were mounted in DAPI-containing medium (Vector, Burlingame, CA). 3 All tissues and cells were imaged on a Nikon A-1 confocal microscope fitted with a 20X 4 and 40X oil objectives lenses and images were processed using FIJI image analysis software. 5 Images of the 200 m vibratome sections were also processed using the Imaris v9.2 software 6 (Bitplane) for 3D reconstruction of vasculature and EV localization. 3D mesh of CD31 positive 7 vasculature was created using the Surface plugin and background subtraction was applied. 8 Placental EVs were detected using the Spots plugin based on PKH26 labeling. The diameter of 9 EV was estimated to be 0.75um in the XY slice for creating spots. Images were taken using the 10 Snapshots function and 3D Video was generated using the animation function of the software. Whole organ imaging 8 To quantify EV localization in whole organs, 2.5 x 10 10 GD14.5 DiR labeled pEVs were 9 injected intravenously into NP female mice. After 24 hours, mice were euthanized and the lung, 10 liver, spleen, thymus, brain, uterus, spleen, lymph node, kidney, and heart were harvested and 11 fixed in 4% PFA in PBS for four hours before being stored in PBS. Whole organs were imaged 12 on a LiCor odyssey infrared imager on manual scan mode (21 microns) under automatic 13 acquisition settings (LiCor BioSciences, Lincoln, NE). Raw tifs were analyzed using Fiji and 14 converted to 8 bit before mean intensity was measured in each tissue . 15 Recombination genotyping 16 To detect genomic recombination in co-cultured mT/mG dendritic cells, DNA was 17 isolated with a Quick DNA/RNA miniprep kit (D7005, Zymo, Irvine, CA) following the 18 manufacturer's instructions. DNA was amplified using a modified two-step polymerase chain 19 reaction protocol targeting the unrecombined mTomato locus or recombined mGFP locus in 20 mT/mG reporter ( Figure 7C) mice. 31 The mTomato locus was amplified with mT-F 5'-21 GCAACGTGCTGGTTATTGTG-3' and mT-R 5'-TGATGACCTCCTCTCCCTTG-3' primers, 22 yielding a 200bp amplicon. The mGFP locus was amplified with mG-F 5'-23 GTTCGGCTTCTGGCGTGT-3' and mG-R5'-TGCTCACGGATCCTACCTTC-3' primers, yielding 24 a 376 amplicon. Genomic DNAfrom dendritic cells was amplified for 35 cycles, and 3 l of the 25 PCR product was used as the template for a second round of PCR. Genomic tail DNA from 26 WT, mTomato, and mGFP mice were used as positive and negative controls for both PCR 27 reactions. PCR products were run on 1.5% agarose gel with 100bp ladder (New England 28 Biolabs, Cambridge, MA) and visualized on an iBright digital gel imager (Thermo Fisher,29 Waltham, MA). 30 1 For boxplots, the middle line represents median value, upper and lower box regions 2 correspond to third and first quartiles (75th and 25th percentiles) and whiskers represent 1.5 3 times the interquartile range. All plots and analyses were generated in R v4.0. All raw data, 4 analysis, and scripts for generation of figures are available on GitHub. Fiji/imageJ macros and 5 CellProfiler pipelines are available on GitHub. Data were subjected to a Shapiro normality test 6 before selecting the appropriate parametric or non-parametric statistical test.

A.
Representative concentration and size distribution analysis of plasma EVs from non-pregnant (NP) and gestational day (GD) 14.5 mice obtained by ultracentrifugation and analyzed by NanoSight. The thick middle line in each distribution curve represents mean concentration, and the shaded area bound by narrower lines represents SEM. Data represent five technical replicates of three different animals in each group. B.
Western blot analysis of plasma EVs from a GD14.5 mouse. Lane numbers represent size exclusion chromatography fractions; red box indicates fractions used for in vivo experiments. WP, whole plasma. Representative data from three independent experiments. C.

Supplemental Figure 4. Proteinase K inhibits pEV localization to lung interstitial macrophages.
Flow cytometric quantification of placental EV localization to interstitial macrophages in lung. Points represent individual biological replicates; data were analyzed by Welch's T-test.