Ethics

The local ethical committee (N°45 in the French National register) approved the experimentation under N°12/102.

Animal exposure

Twenty-eight pregnant New-Zealand white female rabbits (INRA1077 line, 1-year old) (F0) were exposed by nose-only inhalation in custom made plexiglas tubes to either diluted DE (1 mg/m³) (exposed group) or clean air (control group) for 2 h/day, 5 days/week, from the 3rd to the 27th day post-conception (dpc) (i.e., 20 days altogether over a 31-day gestation) (Fig. 1). DE exposure was performed with the Mobile Ambient Particle Concentrator Exposure Laboratory [18] connected to a 25KVA Loxam engine, with a 500 nm particle filter (Additional file 1: Figure S1).

DE is a complex mixture of hundreds of constituents in either a gas or particle form. Gaseous components of DE include carbon dioxide, oxygen, nitrogen, water vapour, carbon monoxide, nitrogen compounds, sulphur compounds, and numerous low-molecular-weight hydrocarbons (some of them individually known to be toxic, such as aldehydes, benzene, 1,3-butadiene, polycyclic aromatic hydrocarbons (PAHs) and nitro-PAHS). The particles present in DE are known to be composed of center core of elemental carbon with absorbed organic compounds and small amounts of sulphate, nitrate, metals, and other trace elements [19]. The measured components of the exposure mixture in the present experiment are shown in Additional file 2: Table S1.

Intra-uterine growth of the F1 generation with Ultrasound/Doppler monitoring

No sex-specific difference was observed in any of the analyses.

At 7 dpc, no effect of exposure to DE was observed on embryo diameter, perimeter and volume (Additional file 3: Table S2).

At 21 dpc, DE exposure was not associated any more with any variation in abdominal perimeter, femur and head length, biparietal diameter, heart rate, nor Doppler (umbilical resistance index and systolic and diastolic velocities) or placental measurements (Additional file 4: Table S3 and Additional file 5: Table S4).

Contemporary to the groups used for fetoplacental exploration, two groups of exposed (N = 7) and control (N = 9) dams (F0) were allowed to give birth. At birth, litter weight tended to be lower in exposed vs. control dams (−18.3 %, p = 0.065) (Additional file 7: Table S6).

Materno-placental exchanges and placental vascularization

Quantitative Power Doppler analysis of placental vascularization at 28 dpc

Near term, DE exposure induced reductions in VI (−66.1 %, p = 0.017), FI (−9.3 %, p = 0.009) and VFI (−70.1 %, p = 0.015) (Fig. 2b, c, d). Altogether these data show that maternal exposure to diluted filtered DE induced a reduction in placental blood flow near the term.

Placental stereology

The rabbit placenta develops as a labyrinthine and hemodichorial placenta close to that of the human (villous and hemochorial) [16] and is made up of three compartments: the decidua (of maternal origin), and the junctional and labyrinthine areas of fetal origin (Fig. 2a). Materno-fetal exchanges take place in the labyrinthine area, which cellular composition was explored by stereology (Fig. 3a and b). At 28 dpc, the relative volume fractions of trophoblastic cells (−43 %, p = 0.004) and fetal capillaries (−42 %, p = 0.002) were significantly decreased in exposed placentas compared to controls (Fig. 3c and d). The relative surface density of fetal capillaries was significantly reduced in the labyrinthine area (−31.7 %, p = 0.014) from exposed compared to control fetuses (Fig. 3f). In contrast, the relative volume of maternal blood space was significantly larger (+116 %, p = 0.0001) in the exposed group compared to controls (Fig. 3e). The relative volume (+370 %, p = 0.001) and surface of circulating immune cells (+469 %, p = 0.001) in the maternal space were significantly increased in exposed placentas compared to controls (Additional file 8: Table S7).

These measurements confirmed the in vivo observations made with 3D-PDA.

Transplacental transfer of NP

To determine the ability of inhaled diesel NP to reach the fetus, ultrathin sections of maternal lungs and placental labyrinthine area collected at 28 dpc were examined by Transmission Electron Microscopy (TEM). Since the identification of NPs in the placental tissues was only based on TEM observations, the observed black particles are described below as “NP-like”.

NP-like were observed in lungs of exposed does, located specifically in type 1 pneumocytes within both the cytoplasm and the nucleus (Additional file 9: Figure S2). Isolated, non-aggregated NP-like were also observed within blood vessels (in maternal erythrocytes and plasma) (Additional file 10: Figure S3).

In the placenta, diffuse aggregated NP-like were observed along the microvilli of syncytial membranes by TEM analysis in the maternal blood spaces of exposed placentas (Fig. 4b), while no NP-like could be identified in control placentas (Fig. 4a). Free “finger-print” like NP-like were also found in maternal blood spaces (Fig. 4c and d). In the cytoplasm of trophoblastic cells, isolated NP-like were located in endosomes (Fig. 4f), in figures of autophagy (Fig. 4g) and in lysosomes (Fig. 4h). No similar observation was made in control placentas (Fig. 4e). Compact NP-like were observed scarcely scattered in the nucleus of some trophoblastic cells whose cytoplasm was well-preserved (Fig. 4j). In contrast, numerous small NP-like (diameter 40 ± 10 nm, N = 60) were present in the nuclei of other trophoblastic cells (Fig. 4k and l), together with morphological organelle abnormalities indicating cellular death (Fig. 4k and l). This was not observed in control animals (Fig. 4i). Other “finger-print” like particles were also observed in the trophoblast (data not shown). Aggregated NP-like were visualized in the lumen of fetal vessels from exposed animals (Fig. 4n) but not in controls (Fig. 4m). Finally, NP-like were observed in endocytotic vesicles in fetal erythrocytes (Fig. 4o) with some suggestion of on-going formation of “finger-print” like structures (Fig. 4p). The presence of numerous organelles in the fetal erythrocytes from exposed dams suggests a delay in erythrocyte maturation as erythrocytes should contain only hemoglobin at 28 dpc [25].

Fetoplacental biometry and metabolism of second generation (F2)

At adulthood, F1 female rabbits (6.5 months) were mated to generate a second generation. F2 fetoplacental units were collected at 28 dpc.

Exposure to DE

The DE exposure (1 mg/m³, 2 h/day, 5 days/week for 20 days over 31-day gestation, with mean diameter = 69 nm) in the present study equates to a mean exposure of 80 μg/m³ over 24 h, which is much higher than the 25 μg/m³ WHO daily average recommendation for PM_2.5 [26], whereas there are no recommendations for NP. This DE exposure is similar to the daily human exposure in large European areas when people drive or walk on big roads twice a day. It could be possible that a daily 2 h exposure at 1 µ gram/m³ may have a different impact than a same pulmonary dose delivered over a 24 h period, but most studies consider the accumulation effect. Several studies in rats and mice, however, have been published in which similar or even higher doses and concentrations were used for longer exposure durations without observing moderate to severe detrimental effects in the lung [27, 28].

Fetal growth

In this study, DE exposure induced signs of growth retardation at mid-pregnancy in the first generation, with a reduction in head length and umbilical pulse, suggesting fetal distress, which is in agreement with previous observations in humans [5]. In the present rabbit model, although signs of hypotrophy were observed by ultrasound at 14 dpc, no hypotrophy was observed at 21 dpc, suggesting that catch-up growth occurred in the second half of pregnancy. This is possibly due to placental adaptations aimed at overcoming growth retardation as described in the context of maternal over nutrition in mice, where initial IUGR induced by maternal excess nutrition around mid-pregnancy was subsequently compensated for by an increase in the expression of placental transporters [29]. Here, on 28 dpc, i.e., 3 days before term, however, the decreased placental blood flow due to reduced placental vascularization was associated with a limited hypotrophy, i.e., a reduction in head length with a tendency for a reduced abdominal perimeter, which indicates that these adaptations were not sufficient to sustain fetal growth up to term. Overall, these results are in agreement with those of Slama et al. [5], in humans, showing decreased biparietal diameter in relation to maternal individual exposure to benzene (a marker of traffic-related pollutants in non-smoking women) in the second and third trimester of pregnancy, measured by ultrasound. Effects on head circumference at birth have also been observed for PM_2.5 in other human studies [4].

The reduced fetal head length associated to a weak decrease of the abdominal perimeter at 28 dpc may, as well as the reduced placental efficiency, suggest that placental compensation mechanisms that developed at mid-gestation became insufficient to maintain optimal fetal development near term, as observed in protein-restricted pregnant mice [30].

Fetal insulin is known to act as a growth factor [31]. Hypoinsulinemia correlated with a decrease in IGF-1 concentration, as observed here in F1 fetuses, has been associated with growth retardation in rat fetuses [31]. These observations are also supported by another study in rabbits showing a positive correlation between plasma insulin concentrations and fetal bodyweight, reflecting the involvement of insulin in the regulation of fetal growth [32]. In IUGR humans, hypoinsulinemia was only observed at adulthood [33].

Placental structure and vascularization

Fetal growth is controlled by placental function, and particularly dependent on placental perfusion. In this study, placental vascularization was decreased with a reduction in relative volume and surface of fetal vessels. This reduction in placental vascularization and perfusion as well as the presence of NP-like in all placental compartments, and in particular on the microvillous membrane, could contribute to the reduced nutrient exchanges between the maternal and the fetal blood. Moreover, using TEM, we observed trophoblastic cells with nuclei filled with NP-like together with altered organelle structure (data not shown), suggesting cell degeneration. Although they could not be quantified (in terms of number of cells affected), placental function could possibly be affected. In mice, in utero exposure to 300 μg/m³ (6 h/day, 5 days/week) of inhaled DE did not affect fetal weight but decreased placental weight [38]. The authors mentioned an increase in stromal density (not quantified) which could have induced a reduction in the labyrinthine vascular space, in agreement with our observations [38]. In the present study, placental perfusion, as measured by 3D-PDA, was affected by DE maternal exposure and associated with the decreased vascular density observed through the stereological approach [20]. IUGR and decreased placental vascularization were also observed in a mice study with maternal exposure to PAH before conception [39]. In terms of effects of DE on placental function, the placental immune function has been explored [40] but so far, in vivo or histological exploration of placental vascularization as performed in the current study have, to our knowledge, rarely been performed. Moreover, most studies used rodent models, for which the placenta is not as close to that of humans compared to rabbits [16, 17, 41, 42].

Transplacental transfer of NP

The transplacental transfer of NP has been demonstrated using animal models mostly after intravenous or intraperitoneal NP injections [11, 13, 43]. Blum et al. [15], however, showed that inhaled cadmium oxide NP could reach the placenta but not the fetus. In the present study using inhaled NP-rich DE, NP-like were observed in lungs of exposed does in type 1 pneumocytes and within blood vessels (in maternal erythrocytes and plasma), which we assume to be of DE origin. This would demonstrate that NP are able to cross the lung barrier and reach maternal organs, including the placenta. Indeed, in the cytoplasm of placental trophoblastic cells, similar NP-like were observed as isolated structures in endosomes, in figures of autophagy and in lysosomes, indicating that endocytosis may be one way for the transplacental transfer of NP. Moreover, we demonstrated that NP-like are able to reach the fetal erythrocytes, as also shown by Soler et al. [44]. Nevertheless, the mechanism through which NP-like cross the placenta remains unknown and may depend on their chemical composition [43]. We hypothesize that NP-like could cross placenta by endocytosis, but could also be conveyed through by simple diffusion or facilitated transport, also depending on their size. This deserves to be further investigated. Moreover, the different forms for isolated, aggregated NP-like or “finger-print” like particles observed in the placenta could reflect an ongoing transformation process of the inhaled NP, in relation to the repeated exposure throughout gestation. Small NP-like could be the result of recent DE exposure whereas the “finger-print” like particles could result from the degradation of NP inhaled at the beginning of the experiment. These “finger-print” like particles have been previously observed in mouse spleen and liver after injection of magnetic iron NP and have been suggested to result from a lysosomal degradation process [45]. Further analyses will be performed to determine the chemical composition of these NP in order to confirm that they originate from inhaled diesel exhaust and study if iron metabolism is involved in this degradation.

Intergenerational effects

At birth, exposed F1 litter weight tended to be lower compared to controls. They had caught up at adulthood (data not shown). Whether intergenerational effects could be conveyed through contamination of the maternal milk remains to be determined. Nevertheless, so far, preliminary studies on the maternal mammary gland do not provide clear evidence of the presence of NP in mammary tissue (data not shown).

In this study, intergenerational effects were observed on the second generation. The plasma concentrations of triglycerides were higher and that of cholesterol were lower in “exposed” F2 fetuses compared to controls, indicating metabolic dysfunction. This metabolic impairment during in utero life could predispose offspring to the onset of metabolic syndrome at adulthood, as shown by Picone et al. [46] and Tarrade et al. [47] in a context of maternal high fat diet in rabbits. Moreover, in a study in mice, cumulative effects were observed after continuous exposure of males and females to Sao Paulo air pollution for 3 generations: in the third generation, IUGR and a reduction in volume, caliber and surface of maternal placental space together with an increase in placental fetal vessel surface were observed [48, 49]. These experiments, however, did not discriminate between pre-conceptional, gestational and postnatal effects, nor between paternal and maternal effects. In the present study, the intergenerational effects, as observed in F2 fetuses, could be due to metabolic modifications during fetal and post-natal growth of the F1 animals and thus could possibly not be directly caused by NP but result from indirect effects.

Relative role of NP and other DE components

Altogether, the effects observed here could be due to either NP or Polycyclic Aromatic Hydrocarbons (PAH) among other volatile components and most probably to the combined effects of several components of the mixture composing DE. Indeed, one study in humans compared the vascular effects of the inhalation of total DE, filtered DE or pure carbon nanoparticulate in healthy adult patients [50]. Only total DE significantly affected plasma concentrations of vasodilators, whereas filtered DE and pure carbon nanoparticulate did not have a significant effect [50]. The effect of the combined elements could be due to the adsorption of the volatile compounds on the corona of the NP.

Strengths and limitations

Rabbits were used because of their hemodichorial placenta closer to humans than those of rodents, with a body size enabling ultrasound monitoring during gestation like in humans. Moreover, genetic diversity is maintained in rabbits as opposed to the inbred genetic lines available in rodents. This experiment focused on feto-placental development of the first and the second generations after exposure of F0 females but did not aim at studying intergenerational effects (involving the F3). The technical option of nose-only exposure was selected in order to avoid oral absorption of NP through self-grooming. The use of filters blocking particles with a diameter larger than 500 nm mimicked the diesel particulate filters installed in current diesel vehicles.

Nevertheless, the results obtained here should be confirmed in other studies in humans and/or animals. Rabbits, like rodents, are polycotous animals, in contrast to the unique or twin pregnancy in humans. Their gestation length is short (31 days), thus reducing the likelihood to observe chronic adaptive mechanisms as could be developed during a 9 month long human pregnancy. Furthermore, dams were exposed twice a day to a peak concentration rather than being exposed to lower concentrations throughout the day. Our experimental model is also not totally applicable to the human situation because exposure occurred only during gestation with no preconceptional or postnatal exposure. The sample size was also obviously limited by technical constraints. Our choice was not to formally rely on significance testing, but to look for coherent patterns in associations with DE exhaust. Finally, we have not so far characterized the chemical composition of NP, even though a carbon core is most probable according to the literature. We are aware that, given the novelty of the field, our study represents a first step to highlight potential risk factors for birth outcomes, and should be seen as hypothesis-generating. These hypothesis need to be confirmed by future studies.