Translocation of (ultra)fine particles and nanoparticles across the placenta; a systematic review on the evidence of in vitro, ex vivo, and in vivo studies

Fetal development is a crucial window of susceptibility in which exposure may lead to detrimental health outcomes at birth and later in life. The placenta serves as a gatekeeper between mother and fetus. Knowledge regarding the barrier capacity of the placenta for nanoparticles is limited, mostly due to technical obstacles and ethical issues. We systematically summarize and discuss the current evidence and define knowledge gaps concerning the maternal-fetal transport and fetoplacental accumulation of (ultra)fine particles and nanoparticles. We included 73 studies on placental translocation of particles, of which 21 in vitro/ex vivo studies, 50 animal studies, and 2 human studies on transplacental particle transfer. This systematic review shows that (i) (ultra)fine particles and engineered nanoparticles can bypass the placenta and reach fetal units as observed for all the applied models irrespective of the species origin (i.e., rodent, rabbit, or human) or the complexity (i.e., in vitro, ex vivo, or in vivo), (ii) particle size, particle material, dose, particle dissolution, gestational stage of the model, and surface composition influence maternal-fetal translocation, and (iii) no simple, standardized method for nanoparticle detection and/or quantification in biological matrices is available to date. Existing evidence, research gaps, and perspectives of maternal-fetal particle transfer are highlighted.

engineering processes [14]. Concerning the latter, Manangama et al. showed a significant association between maternal occupational NP exposure and small for gestational age (birth weight < 10th percentile for gestational age) [15]. Appropriately, the question arises if, during pregnancy, particles can translocate from the mother towards the developing fetus. To our knowledge, this is the first systematic review synthesizing all literature regarding the maternal-fetal transfer of (ultra)fine particles and NPs in in vitro, ex vivo, and in vivo settings. The systematic review aims to (i) evaluate the translocation of (ultra)fine particles and NPs towards and across the placenta in in vitro cellular barriers, ex vivo placental perfusion models, in vivo animal, and in vivo human studies, (ii) summarize the exploited analytical techniques to determine maternal-fetal NP translocation, and (iii) identify gaps and further research needs.

Methods
The systematic review was processed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) statement [16]. In accordance with the guidelines, our systematic review protocol was published by the International Prospective Register of Systematic Reviews (PROSPERO) on April 28th, 2020 (CRD42020167478, Additional file 1).

Search strategy
The search strategy used to identify relevant studies on the maternal-fetal transfer of (ultra)fine particles and engineered NPs was made up of four stages, as depicted in Fig. 1. In the first stage, articles were identified through a comprehensive literature search using two electronic bibliographic databases: MEDLINE (PubMed interface (www.pubmed.ncbi.nlm.nih.gov)) and Science Citation Index Expanded (Web of Science interface (www.webofknowledge.com/WOS)). The full search strategy was based on the search components "placenta", "particles", and "translocation". Literature search strategies based on Boolean operators were developed using related MeSH terms and text words (Additional file 2). To ensure literature saturation, reference lists in key Fig. 1 Flowchart, following the PRISMA statement guidelines, of the search strategy used to identify studies examining the maternal-fetal transfer of (ultra)fine particles and NPs. NPs: nanoparticles review papers and included studies were screened to find additional eligible publications that were not retrieved from our initial database searches. The literature search covered articles published in English between January 1st, 1940 and August 11th, 2020.

Selection criteria
We included human studies and animal studies relevant to human health that addressed the translocation of (ultra)fine particles or NPs across the placenta in an in vitro, ex vivo, and in vivo context. To maintain the focus on air pollutants and engineered nanomaterials in particulate form, we excluded articles examining exposure to tobacco smoke, secondhand smoke, volatile organic components (e.g., benzene, styrene, or xylene) or other volatile substances (e.g., carbon monoxide (CO), ozone (O 3 ), nitrogen dioxide (NO 2 ), or sulfur dioxide (SO 2 )), and nanomaterials characterized by a high aspect ratio (e.g., nanotubes, nanosheets, or nanowires). Additionally, therapeutic NPs were excluded since these systems are specifically fabricated to (i) achieve targeted and increased placental uptake to treat placental complications (e.g., placenta previa-accreta) [17], (ii) limit transplacental transfer to protect the developing fetus while treating the pregnant mother [18,19], or (iii) allow transplacental transfer to enable prenatal treatment of congenital diseases (e.g., congenital adrenal hyperplasia or fetal cardiac arrhythmia) while avoiding severe maternal side effects [20,21]; whereas we want to maintain the focus on unintentional/environmental exposures.

Selection of studies
Two reviewers (EB and TVP) independently screened the titles and abstracts of all identified papers to exclude studies that did not fulfill one or more of the a priori set inclusion criteria. Any disagreement was resolved through discussion. If no consensus was reached, a third reviewer (HB) was consulted. In the third stage, the full text of selected papers was retrieved and underwent a second screening to see which articles were eligible for inclusion.

Data extraction process
In the fourth stage, selected studies were grouped according to model and characterized as in vitro, ex vivo, or in vivo (i.e., human or animal) study. The following data were extracted and registered in a predesigned data extraction form: authors, model characteristics, experimental information (e.g., nature of particle exposure, particle size, particle material, surface modification, exposure route, exposure period, and dose), and main findings (e.g., the observation of translocation, and analytical methods used to visualize or quantify particle transfer).

Synthesis of results
The diversity in, among others, species origin (i.e., rodent, rabbit, or human) or complexity (i.e., in vitro, ex vivo, or in vivo) of the applied model, administration route and particle dose, exposure assessment, and analytical detection method, did not allow to carry out a comparative quantitative analysis. Instead, we provided a qualitative overview of the results on the maternal-fetal transfer of (ultra)fine particles and NPs. Narrative result synthesis was achieved via three different steps: (i) summarizing information on the characteristics of included studies in tables per study model, (ii) identifying the maternal-fetal transfer of a given particle, and (iii) grouping those confirmed translocation situations as quantitative or qualitative.

Study selection
The initial literature search was completed in January 2020 and re-run in August 2020. A total of 647 articles were identified using PubMed (N = 296) and Web of Science (N = 351). Additionally, 29 articles were retrieved from the reference list of included reviews (N = 18) and other studies (N = 11). In total, 105 duplicates were removed using the EndNote software. Titles and abstracts of 571 papers were screened, and 108 full-text papers were assessed for eligibility. Thirty-five articles were excluded because they did not fulfill the following predetermined inclusion criteria; 2 had no full text available, 9 did not measure particle translocation across the placenta, 22 focused on diagnostic/therapeutic NPs, and 2 showed data already published in another included study. The final selection of 73 articles studying transplacental particle transfer included: (i) 21 studies using the ex vivo placental perfusion model and/or in vitro cell line, (ii) 50 animal model studies, and (iii) 2 studies in a human population (Fig. 1).

Study characteristics
We have summarized the existing evidence on the maternal-fetal transfer of (ultra)fine particles and NPs in in vitro placental barriers (Table 1), ex vivo placental perfusion models (Table 2), in vivo animal models (Table 3), and humans ( Table 4). The characteristics of the included studies are summarized in Fig. 2.
In vitro maternal-fetal particle transfer As summarized in Table 1, a total of 15 in vitro cell line studies investigated the maternal-fetal transfer of NPs and (ultra)fine particles. Particle uptake was studied using trophoblast cells grown to confluence on the bottom of a well or Petri dish [29,30]. Moreover, to study transplacental transfer, cells were cultured on Transwell inserts, which consist of a permeable  Data are shown as mean ± standard deviation, a primary particle size stated by the manufacturer, b

ICP-MS
/ Significantly higher levels of Zn in liver and kidneys, but not in blood and brain of rat offspring exposed to ZnO NPs before, during, and after gestation.
[85] CD-1 mice Zn detected in the placentae of mothers exposed to ZnO NPs during early gestation in contrast to mothers exposed to bulk ZnO. Only the smallest ZnO NPs crossed the placenta to reach the fetus.
[  NP-like structures observed in placenta, maternal blood space, trophoblasts and fetal blood of exposed rabbits.
Data are shown as mean ± standard deviation, a primary particle size stated by the manufacturer, b  Abbreviations -BC black carbon, Cu copper, Fe iron, fs femtosecond, ICP-OES inductively coupled plasma-optical emission spectrometry, m3 cubic meter, μm micrometer, n.d. not defined, NPs nanoparticles, SEM/EDX scanning electron microscopy/energy-dispersive X-ray spectroscopy, TEM transmission electron microscopy a predominantly inhaled Fig. 2 Pie charts describing the characteristics of the included studies. Different models (a) were used to assess if (ultra)fine particles and NPs (b) can bypass the placenta (c) using a variety of detection methods (d). NPs: nanoparticles translocation success of 0.2 and 1.3% across a BeWo b30 monolayer, 0.1 and 3.6% across an HPEC-A2 monolayer, and 0.1 and 0.6% across the detailed co-culture, respectively for 3 nm poly(ethylene glycol) (PEG)-coated and 4 nm carboxylated Au NPs [31]. Additionally, the same research group reported a negligible transplacental transfer of carboxylated and aminylated TiO 2 NPs across similar cellular barriers [34]. Multiple studies focused on the effect of particle size and concluded a size-dependent particle transport as larger particles were primarily internalized by the cells with limited transfer to the basolateral compartment [26,27,32]. Moreover, surface modification was found to steer particle transport and cell-particle interactions [22,24,[31][32][33]. Sodium oleate-coating increased both transcellular transport and cellular uptake, while PEGylation or poly(ethyleneimine) (PEI)-coating resulted in enhanced placental cell association and thus less transport. On the other hand, carboxylation, starch-, or dextran-coating were shown to enhance the transport across the cell barrier while decreasing cell-particle interactions and thus internalization [22,24,[31][32][33]. Only 2 studies examined the transfer of ambient air pollution particles across the in vitro placental barrier, more specifically across placental first-trimester trophoblast cells [29,30]. Both wood smoke particles [30] and PM 2.5 [29] accumulated in the exposed placental cells as observed by transmission electron microscopy (TEM).
Ex vivo maternal-fetal particle transfer in placental perfusion models Table 2 gives an overview of the 11 included studies on maternal-fetal particle transfer that employed an ex vivo placental perfusion model to mimic the maternal and fetal blood circulation in the placenta. All placental surrogate models used human term placentae, except for one study that used a rat placenta to investigate transplacental particle transfer [38]. Almost all included studies adopted a 6 h placental perfusion duration (N = 10) with the most frequently used NPs being PS (N = 4) [40][41][42][43] and Au (N = 3) [23,31,38]. A total of 4 studies combined results from in vitro cell barriers and ex vivo placental perfusion studies to gain additional insights on the kinetics of particles at the placenta [23,25,31,34]. Overall, the ex vivo placental transfer studies reported a substantial decrease in particle concentration in the maternal perfusate over time, without a corresponding increase in the fetal circulation, indicating NP accumulation in the placental tissue. Accordingly, accumulation of Au NPs [23,31], SiO 2 NPs [25], and TiO 2 NPs [34] was observed in the human placenta, both in vitro and ex vivo. Au NPs [31] and SiO 2 NPs [25] were mainly found in the outer surface of the chorionic villi (i.e., syncytiotrophoblasts), which is in agreement with the observed apical accumulation of the NPs in a BeWo cell monolayer. After 6 h of perfusion, very limited transfer of small (3-4 nm) PEGylated Au NPs [31], and SiO 2 NPs (25 and 50 nm) [25] was observed in the fetal circulation. For example, after 6 h of perfusion, a placental accumulation of 4-7 μg/g and 2-14 μg/g was observed for 3 nm PEGylated and 4 nm carboxylated Au NPs, respectively, while the respective particle concentration in the fetal perfusate was only 0.0031 μg/mL and 0 μg/mL [31]. As observed for in vitro studies, an association between placental transfer and particle size is present; an increased placental transfer was observed for perfusions with smaller particle sizes [37,[40][41][42][43]. For instance, Wick et al. found that PS NPs up to 240 nm were able to cross the placenta and reach the fetal circulation already after a few minutes of perfusion, while 500 nm PS NPs were mainly retained in the placental tissue and maternal circuit [43]. 4 included ex vivo perfusion studies emphasized the effect of surface modification (e.g., PEGylation and carboxylation) on the transplacental passage of engineered NPs [31,37,41,42]. The overall consensus was that PEGylation increases the transplacental particle transport, while carboxylation (e.g., COONaand COOH-modification) resulted in enhanced placental accumulation. One study also reported increased placental accumulation of aminylated compared to carboxylated TiO 2 NPs [34].
In vivo maternal-fetal particle transfer in animal models In total, 49 animal studies on maternal-fetal particle transfer were included. Forty-three studies showed quantitatively and/or qualitatively that particles reached the placenta after gestational exposure, among which 16 studies observed bioaccumulation of nanosized particles in tissues collected during the embryonic (i.e., organogenesis) and/or fetal period ( Table 3). As depicted in Fig. 3a, rodents were the primary animal model used in the included studies (either mouse (N = 29) or rat (N = 18)), whereas only 2 studies used rabbits as a model for gestational particle translocation [91,92]. Moreover, the latter are the only studies examining transplacental translocation of airborne (ultra)fine particles, whereas the remainder of in vivo animal studies focused on engineered NPs.  Table 3). Fetal/ embryonic NP accumulation was observed in 32 of the included animal studies, more specifically NPs were detected in the fetal brain [27,44,48,52,60,73,78,80,81,91], fetal liver [27,48,50,52,60,69,73,83,84,86], fetal lung [27,48,52,75], fetal kidney [48,52,84], fetal gastrointestinal tract (GIT) [88], and fetal blood [92]. The preferred method for NP administration was intravenous (i.v.) injection (N = 27) or gavage (N = 14) (Fig.  3b). The i.v. route is preferred to examine tissue distribution and elimination of NPs while avoiding absorption by the GIT and first-pass elimination in the liver; on the other hand, the oral route is used to assess the ingestion of NPs. Fennell et al. used both i.v. injection and oral administration to expose pregnant rats to 20 nm and 110 nm Ag NPs, and showed increased placental Ag NP levels compared to fetal levels regardless of the administration route [51]. Despite inhalation exposure being the primary route in environmental and occupational settings, only 6 animal studies [47,66,68,77,91,92] examined particle translocation following this exposure route. While the studies are limited, the conclusion is unambiguous. All studies except two [68,77] showed the translocation of Ag NPs [47], CdO NPs [66], and DEPs [91,92] to the placenta and/or fetus following maternal inhalation exposure. In total, 7 included in vivo animal studies revealed a size-or dose-dependent distribution of NPs from mother to fetus, meaning that smaller particles administered in higher doses tend to accumulate more in placental and embryonic/fetal tissue [27,46,55,57,62,73,85]. For instance, measured cadmium (Cd) levels in mice pups were positively associated with the injected CdTe/CdS QD dose their mothers received. Moreover, a smaller QD size was found to be associated with increased accumulation of Cd in the pups [62]. Also, 1.4, 18, and 80 nm negatively charged Au NPs were found in placentae and amniotic fluid samples of i.v. injected pregnant rats. Only fractions of the initial dose of 1.4 and 18 nm Au NPs reached the fetal tissue, suggesting a size-dependent maternal-fetal translocation of Au NPs [57]. The effect of surface charge on transplacental transfer of engineered NPs was discussed in cited studies [69,74,89]. Polymeric NPs coated with PEI (i.e., cationic polymer) preferentially accumulate in the rat placenta over negatively charged polymeric NPs [89] as confirmed by Di Bona et al. who showed an increased placental crossing and consequent elevated fetal levels of PEI-coated Fe 2 O 3 NPs in i.v. injected pregnant mice as opposed to negatively charged poly(acrylic acid) (PAA)-coated Fe 2 O 3 NPs [69]. Additional studies on surface composition were included [49,61,90]; one showed a reduction in maternal-fetal transfer after chitosancoating of Ag NPs [49]. Other studies reported a timedependent accumulation of Ag NPs, CeO 2 NPs, and fullerenes as they found that the fetal NP concentration increased over time, reached a peak, and then declined [54,67,87]. Moreover, critical exposure windows during gestation were evaluated by defining changes in particle transfer after varying the day of NP exposure [60,61,70,71,87]. Pregnant rats were divided into five groups and differently exposed with radioactively-labeled fullerenes ( 14 C(C60)) via single i.v. injection. The experimental groups covered different stages of pregnancy and varied in exposure duration and period until examination. The percentage of radioactivity recovered in rat placenta was higher at later stages of pregnancy (2% recovery at gestational day (GD) 18, compared to 0.05% at GD11). On the other hand, radioactivity detected in fetuses was lower at later stages compared with earlier stages of gestation (0.2% at GD11 and 0.04% at GD18), which can be explained by the lack of a developed placenta during early gestation [87]. Seven animal studies did not find evidence of maternal-fetal transfer of engineered NPs [56,63,68,72,77,79,82].

Airborne (ultra)fine particles
Only 2 animal studies examined the maternal-fetal translocation of ambient air pollution particles [91,92]. Black particles were observed by TEM in the maternal lungs and blood vessels of pregnant rabbits after exposure for 2 h/day, 5 days/week to 1 mg/m 3 of 69 nm diesel Fig. 3 Pie charts describing the characteristics of the animal studies included in the review. In vivo maternal-fetal particle transfer was assessed in animal models (a) exposed to (ultra)fine particles and NPs via different methods of administration (b). NPs: nanoparticles exhaust particles (DEP) from GD3 to 27. Nonaggregated and NP-like particles were detected in the placenta, maternal blood space, trophoblastic cells, and in fetal blood [92] as well as in the fetal olfactory tissue [91]. However, the identification of these particles in the tissues was solely based on TEM observations of black particles without confirming their origin from the diesel aerosols [91,92].
In vivo maternal-fetal particle transfer in humans Only 2 studies examined the maternal-fetal transfer of (ultra)fine particles and NPs under real-life exposure conditions [1,93] (Table 4). Bové et al. performed a study on a subset of term placentae from 20 healthy, non-smoking mother-newborn pairs enrolled within the Belgian ENVIRONAGE (ENVIRonmental influence ON early AGEing) birth cohort [94]. Ambient black carbon (BC) particles were found in all screened placentae, and the placental BC load was positively associated with the mothers' residential BC exposure during pregnancy. The average (SD) placental BC load was 0.95 × 10 4 (0.66 × 10 4 ) and 2.09 × 10 4 (0.90 × 10 4 ) particles per mm 3 for low-and high-exposed mothers, respectively [1]. Raia-Barjat et al. assessed NP crossing over the human placenta by investigating the NP loading in amniotic fluids collected from 100 pregnant women. A high number of pregnant women with a substantial concentration of the essential trace elements iron (Fe), copper (Cu), and zinc (Zn) in the nano/ion fraction was observed. In contrast, the prevalence of women with a substantial concentration of aluminum (Al), silver (Ag), beryllium (Be), cobalt (Co), chromium (Cr), nickel (Ni), silicon (Si), titanium (Ti), and tungsten (W), was relatively low (i.e., under 20%). Nonetheless, the authors acknowledged that this does not necessarily indicate the presence of NPs since the used technique, ICP-OES, is not able to discriminate NPs from ions [93].
Analytical methods to visualize and quantify maternalfetal particle transfer A variety of analytical methods were used to quantify (i.e., (semi-)quantitative techniques) and/or visualize (i.e., qualitative techniques) the gestational transfer of particles by their identification in relevant tissues (e.g., placental or fetal tissue). Included studies based their results on (semi-)quantitative detection techniques (N = 33), qualitative methods (N = 11), or a combination of both (N = 29). In total, 26 different techniques were used in the studies discussed in this review, as summarized in Table 5.

Qualitative techniques
Qualitative microscopy-based methods employed in the discussed studies included bright-field light, fluorescence, and confocal microscopy as well as transmission and scanning electron microscopy (TEM and SEM), and X-ray microscopy. The most frequently used qualitative technique was TEM (N = 22). Analytical tools, such as energy-dispersive X-ray spectroscopy (EDX) or Raman spectroscopy, can be coupled to electron microscopes for additional elemental composition analysis [45,47,75,80,93]. For example, Takeda et al. used field emission-type SEM (FE-SEM)/EDX to show the transfer of nanosized TiO 2 from pregnant mice into the brain and testis of their offspring [80]. Autometallography, a histochemical technique based on silver enhancement, is also used to allow light and electron microscopic visualization of metallic NPs in biological tissues and cells (N = 4). Using TEM and silver enhancement, 10 and 15 nm Au PEGylated NPs could be visualized in the trophoblastic cell layer of the placenta after 6 h of re-circulating perfusion, yet no particles were quantified in the fetal circulation [23].

Quantitative techniques
One of the included human studies on maternal-fetal particle transfer employed a novel detection technique based on the white-light generation by carbonaceous particles under near-infrared femtosecond pulsed illumination [1,105]. This label-free confocal microscopic technique showed the presence of BC particles originating from ambient exposure at the fetal side of the human placenta [1]. Label-free detection of NPs in cells and tissues is not straightforward and, therefore, NPs are often labeled using fluorophores, radiolabels, or contrast agents. For quantitative visualization of non-fluorescent NPs in biological systems, fluorescent labeling (e.g., using fluorescein isothiocyanate [71] or rhodamine [24,25]) can be employed followed by fluorescence detection methods. Melnik     (i.e., single particle, sector field, and laser ablation ICP-MS), was the most frequently used quantitative technique (N = 29) among the included studies. In addition, 6 of the included studies used atomic absorption spectrometry (AAS) as elemental analysis tool for detection of metallic NPs in maternal-fetal tissue [33,44,48,49,59,66].

Combination of qualitative and quantitative techniques
In total, 29 included studies combined quantitative and qualitative methods to gain complementary insights on the maternal-fetal transfer of NPs. For example, Ho et al. combined findings from quantitative (i.e., magnetic resonance imaging and fluorometry) and qualitative assessments (i.e., fluorescence and confocal imaging) to determine the biodistribution of cationic and anionic multimodal (i.e., fluorescent and paramagnetic) polymeric PGMA NPs in a pregnant rat model at different stages of gestation. While the quantitative methods were unable to unambiguously determine tissue uptake, confocal microscopy confirmed a differential charge-based accumulation of the NPs in the rat placenta [89]. The time-dependent transfer of SPIONs with a differential surface charge through a BeWo/pericyte co-culture was measured by magnetic particle spectroscopy and confirmed by AAS. As previously observed, neutral and negatively charged iron oxide NPs were able to pass the cell layer, whereas positively charged NPs primarily interact with the BeWo cells [33].

Discussion
Placenta models used to study maternal-fetal NP translocation As visualized in Fig. 2a, only 2 included studies examined the transplacental passage of (ultra)fine particles or NPs in a human population under real-life exposure conditions [1,93]. The majority of the cited studies investigated maternal-fetal particle transfer and distribution in animal models, hence, in biological systems under controlled conditions. Mainly rodents (i.e., mice and rats) were used as animal model (Fig. 3a). Both rodents and rabbits are easy to breed and handle and their small size facilitates large scale/high throughput studies, making them cost-efficient models [119,120]. However, placentation differs between species [121]. To compare data on transplacental NP transfer, it is necessary to understand both the physiological and anatomical differences between humans and the employed animal models [122]. Humans, rodents, and lagomorphs (rabbits) all share a discoid, hemochorial placenta, denoting that maternal blood comes in direct contact with fetal trophoblastic tissue. More specifically, hemomonochorial (human at term), hemodichorial (rabbit and human in their first trimester), and hemotrichorial (rat and mouse) placentae, with one, two, and three trophoblastic epithelial layers separating maternal and fetal blood, respectively [121]. Rodents are the most widely used animal model in developmental toxicology, yet there are structural differences between rodent and human placentae. Rodents (i) reach their definitive placental structure in a later stage, (ii) have less invasive trophoblast cells, and (iii) have a labyrinthine as opposed to a villous organization in human placentae [123,124]. Rabbit placentae also have a labyrinthine structure, yet they are hemochorial with two trophoblast layers, a syncytium and a cytotrophoblast layer, which more closely resembles the human placenta. Moreover, critical exposure windows can be defined more precisely in rabbits as ovulation is induced by mating, resulting in the exact timing of fertilization and pregnancy stages. Rabbit placentae appear to resemble more the human placentae than that of rodents. Therefore, rabbits are considered the preferable animal model to study gestational particle exposure [120,125]. Keeping in mind that the placenta is the most species-specific mammalian organ, various human model systems have been developed to mimic the in vivo situation in pregnant women as closely as possible.
Ex vivo perfusion of human term placentae is used as a surrogate to study the transplacental transport of NPs. It is considered as the "gold standard" among currently available translocation models as it preserves the structural complexity of a full-term placenta and resembles its dynamic environment, enabling to study ex vivo transplacental NP passage without harming the mother and/or fetus [126]. Nonetheless, perfusion studies (i) predominantly use full-term placentae and, hence, do not allow to estimate NP transfer during earlier and more vulnerable stages of pregnancy, (ii) are limited to a few hours (4-8 h) of perfusion due to tissue degradation, which is insufficient to observe chronic effects, (iii) have a low success rate of perfusion (i.e., 30%), and (iv) are time-consuming [126,127].
To overcome the aforementioned limitations, in vitro models using human cell cultures (primary cytotrophoblasts or choriocarcinoma cell lines) are attractive as they allow high-throughput testing of transplacental transfer of various NPs. However, models like the commonly employed BeWo b30 Transwell model also have their limitations [128]. In vitro placental transfer models do not fully resemble the physiological structure of the in vivo placenta as they lack anatomical integrity and blood flow. Moreover, monolayers form a simplified placental barrier. Co-cultures, on the other hand, attempt to more closely mimic the complex in vivo placenta that constitutes multiple cell layers (e.g., trophoblast and endothelial cell layer) across which particles have to be transported and where cell-cell interactions among various cell types can happen [129,130]. In this regard, Muoth et al. used a 3D co-culture microtissue model to more closely resemble the human placental structure, and to study Au NP uptake and penetration in an organotypic environment. Higher uptake and deeper penetration were observed for the smaller carboxylated Au NPs in comparison to the larger PEGylated Au NPs [32]. Interestingly, a lower degree of NP transport was observed across co-culture placental models compared to monolayer cultures. For instance, Cartwright et al. exposed a BeWo b30 monolayer to PS NPs with sizes up to 100 nm and was able to detect those particles in the basal compartment after 24 h of exposure [26]. Correspondingly, Aengenheister et al. exposed a BeWo b30/ HPEC-A2 co-culture to PS NPs with particle sizes up to 70 nm under similar conditions and found only low amounts of 49 nm PS NPs and no 70 nm PS NPs in the basal compartment after 24 h of exposure [35]. Additionally, Au NPs (±14 nm) were able to transport across BeWo b30 and HVMF monolayers, while they barely passed the co-culture of both cell types [32]. Nonetheless, most in vitro models lack a physiological microenvironment as they are exposed to NPs under static conditions.
Although a variety of placental models are currently available, the preconditions of each model should be taken into consideration depending on the research objective. High-throughput in vitro transfer models are useful to pre-screen a variety of NPs and to provide mechanistic insights, yet, further improvements (e.g., microfluidic approaches to develop a dynamic model) are needed to enhance their predictive value. Ex vivo placental perfusion studies provide transfer data with high in vivo relevance, at least for term pregnancy. On the other hand, exposure of pregnant animal models can provide important insights on the biodistribution of NPs in a living organism, including potential translocation to the fetus. Nevertheless, data extrapolation from animals to humans remains challenging as the placenta is the most species-specific mammalian organ. Despite technical challenges and ethical constraints, term human placentae form the ideal model to study transplacental NP transfer.

Factors that influence maternal-fetal translocation
Various studies showed the influence of different factors on maternal-fetal particle translocation, including particle size, particle material, dose, particle dissolution, and surface composition, as well as NP administration route and the gestational stage of the study model.
First, the effect of particle size on transplacental transfer was addressed by 6 included studies [27,55,57,62,73,85]. For instance, Yamashita et al. i.v. administered 70, 300, or 1000 nm SiO 2 particles to pregnant mice and concluded that only the 70 nm particles could reach the placentae, fetal liver, and fetal brain tissue [73]. Similar size-dependent transplacental transfer was observed for PS NPs in the ex vivo placental perfusion [40,42,43] and in vitro [26,27,35] studies. Different sizedependent mechanisms of placental NP exchange have been proposed, including passive and facilitated diffusion via transtrophoblastic channels (i.e., canaliculi) and active transport (e.g., receptor-mediated endocytosis), yet mechanistic insights on transplacental NP transfer remain scarce [39]. Despite the use of various placental transporter inhibitors, no major influence on the translocation of 50 nm PS NPs in the in vitro BeWo transfer model was observed, which indicates preferential NP translocation by passive diffusion [28]. In contrast, bidirectional ex vivo placental perfusion studies showed an increased fetal to maternal transfer of 50 nm PS NPs suggesting an active, energy-dependent transplacental transport mechanism for PS NPs [41]. Additionally, upregulation of clathrin-and caveolin-mediated endocytosis was observed following i.v. administration of 20 and 50 nm Au NPs [55] and oral administration of 16 nm ZrO 2 [86] in pregnant mice. Noteworthy, the particle cut-off size, above which no transplacental particle translocation is observed, appears to be influenced by the particle material. Aengenheister et al. showed the in vitro transplacental transfer of 50 nm PS [35] but not 4-8 nm TiO 2 NPs [34] across a BeWo/HPEC co-culture. The study also showed that the 4-8 nm TiO 2 NPs were not able to cross the placenta in an ex vivo perfusion model [34], whereas PS NPs up to 240 nm have been shown to reach the fetal circuit [43] in a similar perfusion study. In agreement, in pregnant mice, 5 nm TiO 2 did not cross the placenta [79], while PS NPs up to 500 nm could be observed in various organs of fetuses [27]. Moreover, Kloet et al. observed a difference in translocation behavior of PS NPs with a similar size and surface charge but acquired from different manufactures [28]. This reflects the difficulty to compare results between, and even within, studies due to variations in NP characteristics on top of differences in the employed detection technique.
Second, numerous studies reported that the surface composition of the particles has a tremendous influence on their in vivo translocation [22, 24, 31-34, 37, 41, 42, 61, 62, 69, 74, 89, 95, 96]. Accordingly, Yang et al. exploited three types of surface modifications to assess the effect of surface functionality on maternal-placentalfetal biodistribution of 13 nm Au NPs in fetuses of mice; coating with (i) endogenous proteins (i.e., ferritin) for optimal biocompatibility, (ii) stealth groups (i.e., PEG polymer chains) to increase circulation time by avoiding recognition and phagocytosis by the mononuclear phagocytic system, and by reducing NP-cell and NPprotein interactions, and (iii) stabilizing anionic material (i.e., citrate) to explore the effects of negative charge on placental and fetal distribution. Substantially less uptake of 13 nm citrate-capped Au NPs in fetal tissues has been found compared to ferritin-modified or PEGylated Au NPs with an identical size [61]. A similar effect of surface charge on the transplacental crossing of metal oxide NPs was demonstrated by coating 14 nm Fe 2 O 3 NPs with either negatively charged PAA-groups or positively charged PEI-groups. Pregnant mice intraperitoneally exposed to 10 mg/kg Fe 2 O 3 -PEI for eight consecutive days (i.e., GD9 to 16) had significantly increased iron levels in their placentae and the livers of their offspring compared to mice exposed to Fe 2 O 3 -PAA [69]. In general, an increased transplacental transfer for PEGylated NPs was observed among the included studies, while carboxylated NPs were mainly retained inside placental tissue [31,37,41,42]. In accordance, Aengenheister et al. showed that both carboxylated and PEGylated Au NPs crossed the placental co-culture in low amounts. Despite the higher cellular uptake of carboxylated Au NPs, increased translocation was observed for PEGylated Au NPs. In contrast, only PEGylated particles reached the fetal circulation in the dynamic ex vivo placental perfusion model [31]. Possible explanations given by the authors for the absence of carboxylated particles in the fetal circulation were: (i) the agglomeration behavior of carboxylated Au NPs, and (ii) the non-specific adherence of carboxylated Au NPs to the perfusion system, both markedly reducing the cellular available dose. In all cases, the carboxylated particles were larger than the PEGylated NPs, which can partially explain the reduced transfer.
Third, another factor with considerable influence on maternal-fetal NP transfer is particle dose as discussed in 9 included studies [24,45,48,62,66,76,79,81,86]. By increasing the dosage of i.v. injected 3mercaptopropionic acid (MPA)-coated QDs in pregnant mice, Chu et al. observed a corresponding increase in Cd concentration in the pups [62]. Similar results were observed for TiO 2 NPs [76,81] and ZrO 2 NPs [86] in a pregnant mouse model. However, a lack of doseresponse was observed for CdO [66] NPs and Ag NPs [45,48], possibly because the tissue saturation limit was not reached by the administered NPs, as suggested by Austin et al. [45,48]. In general, NP uptake and translocation do not only depend on their physicochemical properties, surface modification, and particle concentration but also on particle dissolution. The latter is an important property to consider because it alters the particle presence. Accordingly, translocation may be observed which cannot only be attributed to intact NPs (e.g., Ag NPs, CdO NPs, and ZnO NPs), but also released ions, precipitates, or a combination. NP dissolution has been reported by 7 [22,37,45,47,48,66,85] of the included studies yet only 2 distinguished between the translocation and uptake of actual particles or dissolved Ag in an in vitro placental barrier model [22] and ex vivo placental perfusion model [37]. Both studies showed a favorable transplacental transport of ionic over particulate Ag and highlighted the need to consider the uptake of Ag ions and/or dissolution of Ag NPs in the cellular barrier or tissue followed by re-precipitation to Ag NPs in the basolateral compartment or fetal circulation, respectively. On the other hand, Wang et al. showed that the integrity of ZrO 2 NPs was not altered upon encountering biological barriers in a pregnant mouse model as the ionic Zr content was fewer than 1% after incubation of ZrO 2 NPs with water or artificial gastric fluid for 5 h [86].
Fourth, the administration route in in vivo animal studies is an important factor influencing particle translocation. As depicted in Fig. 3b, inhalation exposure (N = 6) remains fairly understudied in gestational translocation studies despite being the main entry route in environmental and occupational settings. Among the included studies, i.v. administration (N = 27) is mainly used as route of exposure. i.v. NP injection allows to control the systemically available dose as NPs do not have to cross primary biological barriers (e.g., lung or intestinal epithelium), which would result in limited transfer across the placenta and thus unfeasible detection limits that restrain the visualization and quantification of NPs in embryonic/fetal tissue [132]. However, the data from i.v. route studies cannot directly be extrapolated to the real-life scenario of inhalation exposure due to differences in the systemic distribution pattern [133]. For example, the differences in the overall distribution patterns might be determined by the different protein coronae that arise when particles come into contact with bronchoalveolar fluid compared to plasma [134][135][136].
Fifth, the previously highlighted interspecies differences are further complicated by dissimilarities in placental anatomy and physiology between the various stages of pregnancy, making it necessary to focus on a critical exposure period in animals that reflects an equivalent stage in human development [137]. Alignment of rodent and lagomorph reproductive timelines with that of humans is based on Theiler (mouse) [138], Witschi (rat) [139], Edwards (rabbit) [140], and Carnegie (human) [141] stages of development. The included studies employed various gestational windows of NP exposure ranging from exposure during early or late gestation [60,67,70,71,74,85,89] to continuous exposure during the whole pregnancy [1,52,68,72,76,81,92,93] to assess the time-dependence of transplacental particle transfer. In this regard, polymeric NPs accumulated in different compartments of the rat conceptus during early (GD10), but not late (GD20) gestation, consistent with the lack of a developed placenta during early gestation [89]. Similarly, Wang et al. reported a higher translocation of ZrO 2 NPs to the mouse placenta, fetal brain, and fetal liver following oral exposure during early stages of pregnancy (i.e., GD9-11 and GD13-15) compared to NP administration later in gestation (i.e., GD16-18) [86]. On the other hand, Pietroiusti et al. detected 25 and 60 nm SiO 2 NPs in the mouse conceptus after i.v. administration at GD5.5 and GD12.5. At later stages (GD16.5), when the placenta acquired higher permeability, larger 115 nm SiO 2 NPs were able to reach the mouse conceptus [74].

Methods used to assess maternal-fetal NP translocation
Studies on gestational NP biodistribution demand highly sensitive visualization and quantification methods because the accurate detection of the fetoplacental accumulation of nanomaterials at very low levels is required.
The uptake of particulates into the systemic circulation is limited and often amounts less than a few percent of the total administered dose [142]. For example, Hesler et al. showed a lack of transport of 50 nm carboxylated PS NPs across a placental co-culture using an asymmetrical flow field-flow fractionation (AF4) system. Still, they recognize that it cannot be excluded that a small number of particles, below the detection limit of the AF4 system, could have been translocated [36]. This detection limit is possibly confirmed by other studies that were able to show a size-dependent translocation of similar PS NPs in mice [27] as well as in an ex vivo placental perfusion model [41,42].
As summarized in Table 5, a multitude of imaging approaches was used to visualize NPs in an in vitro, ex vivo, and in vivo context. Some included studies investigated maternal-fetal NP distribution at the wholebody level by using magnetic resonance imaging [71,89], in vivo/ex vivo fluorescence imaging [61,62,73,89], or radiolabeling techniques [53,57,70,87]; whereas other reports focused on the subcellular localization of NPs by exploiting transmission [23, 24, 27, 29-32, 41, 43, 45, 47, 49, 52, 54, 57, 58, 60, 61, 73, 75, 91, 92] or scanning electron microscopy [75,80,93]. Fluorescence microscopy is generally the method-of-choice to monitor the cellular and tissue-level distribution of fluorescent NPs [143]. However, visualization of fluorescent NPs, even with high-resolution confocal microscopy, is limited because of their sizes (typically being between 1 and 100 nm), which is below the Abbe's diffraction limit of 250 nm. An additional consideration when using fluorescent NPs is dye leakage [144]. Stable incorporation of dyes within the NPs is a prerequisite as the fluorescence from leaked dyes may cause wrong interpretation of NP biodistribution [145]. Accordingly, the in vitro stability of the fluorescent dye in different solutions, including perfusion medium [42,43], and phosphate buffered saline [73], was measured in some of the included studies to exclude false-positive results. Moreover, the detection of fluorescent NPs may be hindered by the autofluorescence of the biological samples, which hampers the visualization of small aggregates [143]. Kenesei et al. employed spectral imaging fluorescence microscopy to distinguish fluorescence, originating from the i.v. injected carboxylated or PEGylated PS NPs, from tissue autofluorescence in mice. The study revealed no placental penetration but retention of PS NPs by the reticuloendothelial system regardless of surface functionalization [90]. In general, fluorescence microscopic techniques are hampered by the risk of sample photobleaching combined with limited particle size detection and the necessity to fluorescently couple the NP under study. The latter is facilitated in the case of (i) QDs [62][63][64][65], which possess inherent fluorescence, or (ii) carbonaceous particles, which generate white-light under femtosecond pulsed laser illumination [1,105]. Such label-free detection methods allow the direct analysis of maternal-fetal NP uptake. Chu et al. used fluorescence imaging to investigate whether intact QDs can transfer across the rodent placenta to reach the fetus. Penetration of the placenta by intact QDs was hard to demonstrate using in vivo fluorescence imaging, possibly due to a low fetal concentration combined with a reduction in fluorescence intensity due to background tissue-endogenous fluorescence (i.e., autofluorescence). Nonetheless, increased Cd levels, originating from QDs, were measured in mouse fetuses and pups using ICP-OES [62]. Another study indicated that the Cd ions originating from CdO NPs, instead of the particles themselves, were able to translocate across the placenta of a pregnant mice model [66]. In agreement, favorable transport of ionic silver over pristine Ag NPs was observed across the in vitro placental cell layer and in the ex vivo placental perfusion model [22,37]. High concentrations of dissolved ions may overestimate the concentration of materials present in nanoparticle form when using chemical analysisbased methods (e.g., ICP-OES and ICP-MS) that measure the total elemental concentration in a sample [146]. In contrast, single-particle ICP-MS (spICP-MS) allows determination of the elemental composition of single particles and provides information on the NP size distribution. Furthermore, spICP-MS can distinguish between dissolved and nanoparticulate forms of a certain material [147]. In this regard, Abdelkhaliq et al. showed the transport of ionic silver and Ag NPs across an in vitro BeWo b30 model using ICP-MS and spICP-MS, respectively [22]. In addition, separation techniques are often used to remove target particles from interfering matrix components to meet the quantification limits of the detection methods. Accordingly, chromatography-based techniques such as field-flow fractionation and highperformance liquid chromatography (HPLC) are used for high-resolution sizing and separation of a wide range of diverse particles [148,149]. Coupling of the chromatographic instruments with other spectroscopic techniques, such as light scattering or light absorption methods, can be used to enhance the characterization power as it allows the physicochemical characterization and elemental analysis of the size-separated particles [150]. In this regard, Hesler et al. used an AF4-UV method for size analysis and quantification of carboxylated PS NPs across an in vitro placental co-culture [36].
To summarize, quantification and reliable detection of NPs in maternal-fetal tissue remain one of the main challenges in transplacental transfer studies, especially under realistic exposure conditions where transfer rates are expected to be lower. Quantification of maternalfetal NP uptake is often achieved at the cost of spatial resolution, whereas intracellular NP localization is often qualitatively defined. Hence, techniques are best combined to gain complementary and profound insights in transplacental particle transfer. Moreover, progress should be made towards standardization and validation of methods used to detect (ultra)fine particles and NPs with high specificity to ensure reliable detection of the particles of interest. Such implementation of standardized protocols would facilitate the comparison between different studies on particle exposure and the assessment of associated health and safety risks.

Conclusion and future directions
Exposure to (ultra)fine particles and NPs in daily life is unavoidable, and this is no different for pregnant women. This systematic review summarizes the evidence that particles can bypass the placenta as observed in (i) in vitro monolayers and co-cultures of different placental cells (e.g., trophoblasts, fibroblastic cells, and endothelial cells), (ii) ex vivo placental perfusion models, and (iii) in vivo rodent and rabbit models and humans. Almost all types of particles tested, ambient or engineered, seemed to be able to reach and/or cross the maternalfetal barrier, even if merely in trace amounts. Transplacental particle transport is affected by the particle size, particle material, dose, particle dissolution, and surface modification, as well as the NP administration route and gestational stage of the employed model. Results on placental NP transfer must be interpreted with care seeing differences in the species origin (i.e., rodent, rabbit, or human) and complexity (i.e., in vitro, ex vivo, or in vivo) of the applied model as well as in particle properties and routes of NP exposure/administration. To obtain substantial and complementary results on developmental toxicity following prenatal exposure, it will be essential to test realistic doses (extrapolated from populationbased studies) and exposure routes. The number of studies on transplacental NP transfer remains limited. More studies on the topic are paramount and in particular on inhalation exposure since this is the primary route of environmental and occupational exposure and largely understudied.
Moreover, to date, little is known about the kinetics and bioavailability of NPs as no simple nor standardized method for NP detection and/or quantification in biological settings is available. This urges the need to come up with standardized protocols and to develop state-ofthe-art methods to accurately detect and quantify the fetoplacental accumulation of low particle levels in tissues, such as in fetal samples.
To conclude, further research on particle uptake, accumulation, and translocation at the placenta is indispensable to predict potential fetal exposure and adverse health effects during fetal development and later in life.