Noninvasive Dual-Modality Photoacoustic-Ultrasonic Imaging to Detect Mammalian Embryo Abnormalities after Prenatal Exposure to Methylmercury Chloride (MMC): A Mouse Study

Abstract

Background:

Severe environmental pollution and contaminants left in the environment due to the abuse of chemicals, such as methylmercury, are associated with an increasing number of embryonic disorders. Ultrasound imaging has been widely used to investigate embryonic development malformation and dysorganoplasia in both research and clinics. However, this technique is limited by its low contrast and lacking functional parameters such as the ability to measure blood oxygen saturation (${\text{SaO}}_2$) and hemoglobin content (HbT) in tissues, measures that could be early vital indicators for embryonic development abnormality. Herein, we proposed combining two highly complementary techniques into a photoacoustic-ultrasound (PA-US) dual-modality imaging approach to noninvasively detect early mouse embryo abnormalities caused by methylmercury chloride (MMC) in real time.

Objectives:

This study aimed to assess the use of PA-US dual-modality imaging for noninvasive detection of embryonic toxicity at different stages of growth following prenatal MMC exposure. Additionally, we compared the PA-US imagining results to traditional histological methods to determine whether this noninvasive method could detect early developmental defects in utero.

Methods:

Different dosages of MMC were administrated to pregnant mice by gavage to establish models of different levels of embryonic malformation. Ultrasound, photoacoustic signal intensity (PSI), blood oxygen saturation (${\text{SaO}}_2$), and hemoglobin content (HbT) were quantified in all experimental groups. Furthermore, the embryos were sectioned and examined for pathological changes.

Results:

Using PA-US imaging, we detected differences in PSI, ${\text{SaO}}_2$, HbT, and heart volume at embryonic day (E)14.5 and E11.5 for low and high dosages of MMC, respectively. More important, our results showed that differences between control and treated embryos identified by in utero PA-US imaging were consistent with those identified in ex vivo embryos using histological methods.

Conclusion:

Our results suggest that noninvasive dual-modality PA-US is a promising strategy for detecting developmental toxicology in the uterus. Overall, this study presents a new approach for detecting embryonic toxicities, which could be crucial in clinics when diagnosing aberrant embryonic development. https://doi.org/10.1289/EHP8907

Introduction

Ultrasound imaging has become ubiquitous in obstetrics departments to monitor the development and well-being of the fetus. Although ultrasound waves can penetrate several centimeters of tissue, there is a lack of strong image contrast and an inability to measure certain functional parameters, making it one of the most challenging imaging modalities to interpret. Moreover, it has been reported that more than 50% of congenital heart abnormalities were not found in routine fetal ultrasound in the second trimester (Tegnander and Eik-Nes 2006). In past decades, many different types of transabdominal ultrasound systems became available in the clinic, including three-dimensional (3D) ultrasound, Doppler ultrasound, and fetal echocardiography. However, they also have some limitations. For example, 3D ultrasound relies on dedicated 3D transducers and tremendous computational power (Gonçalves 2016); Doppler ultrasound analysis uses stored offline video recording review (Dhillon et al. 2020), which differs from clinical practice based on instantaneous visualization; fetal echocardiography technology gives little information on cardiac structures (DeVore et al. 2017). Therefore, it is urgent to develop a sensitive, noninvasive, and quantitative imaging method with high penetration depth to monitor the developmental processes in the uterus.

Photoacoustic imaging, as a newly developed ultrasound application, combines the high contrast of pure optics imaging modalities with the high ratio of imaging depth to spatial resolution of the ultrasound (Wang et al. 2016). Photoacoustic imaging has broad application prospects in diseases such as cancer monitoring (Heijblom et al. 2016), imaging-guided surgery (Levi et al. 2014), and functional and metabolic brain imaging (Laufer et al. 2009). The noncontact characteristics of optical imaging technology and the wavelength selectivity of exploring specific targets in the tissue are the advantages of optical imaging over acoustic imaging technology. Still, the lower penetration depth of optical imaging technology relative to acoustic imaging technology is its disadvantage (Manwar et al. 2020). Dual-modality photoacoustic-ultrasound (PA-US) imaging is a real-time noninvasive imaging method, which combines laser pulse tissue excitation and ultrasonic detection of the tissue response (Needles et al. 2013). A PA-US dual-modality imaging system uses a hybrid detector. Photoacoustic imaging is generated by irradiating the sample with a short-pulsed laser followed by detecting photoacoustic signals induced by converting absorbed laser energy into pressure waves due to the thermoelastic expansion. Ultrasound imaging is produced using the same detector to detect ultrasound waves reflected from tissues (Lee et al. 2020). Thus, dual-modality PA-US imaging technology not only offers the advantages of strong image contrast and high tissue penetration depth up to a few centimeters, but it can also complement those with both structural and functional information of organs (Huang et al. 2020; Li et al. 2018). Sato et al. applied PA-US imaging to visualize the vasculature of living chicken embryos (Sato et al. 2013). Arthuis et al. also used PA-US dual-modality imaging to evaluate placental oxygenation in pregnant mice under hypoxic conditions (Arthuis et al. 2017). Although it is a promising biomedical imaging technique, it has not yet been used to assess developmental abnormalities in the uterus.

Many factors can cause developmental malformation, such as radiation (Amer et al. 2013), genetics (Zhang et al. 2009), and heavy metals (Chen et al. 2015). With the increasingly severe environmental pollution in our lives, the chances of heavy metals inducing malformations in embryo development continue to increase. Mercury is an accumulative heavy metal pollutant and ranked as the third most toxic chemical substance on the planet by the U.S. Government Agency for Toxic Substances and Disease Registry (ATSDR 2019). Methylmercury (MeHg) was the primary source of organic mercury in ecosystems, where mercury-containing compounds can be converted to methylmercury by microorganisms through a variety of pathways, including atmospheric and water cycles (Clarkson and Magos 2006). The risk of MeHg to humanity was demonstrated by the Minamata tragedy in Japan (Murakami 1970) and the epidemic of MeHg poisoning in Iraq (Bakir et al. 1973). MeHg was considered liposoluble and often presented in aquatic ecosystems as methylmercury chloride (MMC) (Mahaffey 1999). MMC can be easily absorbed by lower organisms, gradually passes up the food chain, and eventually showed bioaccumulation in humans (Hall et al. 1997; Clayden et al. 2013). There is evidence that MMC may affect neural stem cells (Tamm et al. 2006) and may be associated with nervous system outcomes based on in vivo (Onishchenko et al. 2007; Wu et al. 1985) and epidemiological studies (Geier and Geier 2003); in addition, there is evidence for an association between MMC and cardiovascular outcomes based on in vitro (Naganuma et al. 1980), in vivo (Duan et al. 2016; Moreira et al. 2012), and epidemiological (Drescher et al. 2014; Frustaci et al. 1999; Roman et al. 2011) studies. In recent years, the concern regarding people consuming contaminated fish has been increasing (Sunderland et al. 2018; Lepak et al. 2019) . How to balance the risks and benefits of fish intake during pregnancy is still debated (Stratakis et al. 2020; Starling et al. 2015). Despite the increasing concern about mercuric ramifications, studies for prenatal exposure to low dosages of MMC are still lacking.

The study of study heavy metals poisoning with regard to pregnancy and newborn health is of considerable significance. In this study, we focused on exploring the potential of dual-modality PA-US to detect embryotoxicity resulting from prenatal exposure to different dosages of MMC. The light absorption characteristics at the animal tissue level reflected by photoacoustic signal intensity (PSI) were closely related to the physiological characteristics of the mice (Rowland et al. 2012), metabolic status of the mice (Yao et al. 2011), disease characteristics of rats and mice (Zhang et al. 2006), and even neural activity in monkey brain tissue (Yang and Wang 2008). Detecting blood oxygen saturation (${\text{SaO}}_2$) and hemoglobin content (HbT) levels of the embryo in rats was helpful for clinical evaluation of the embryonic state (Lawrence et al. 2019). Therefore, we used dual-modality PA-US imaging to measure the organ volumes, PSI, ${\text{SaO}}_2$, and HbT to assess the development of Institute of Cancer Research (ICR) mouse embryos prenatally exposed to MMC and compare to normal embryonic development in control-exposed mice. To further confirm the accuracy of this noninvasive diagnosis, the results were compared with the traditional pathological analysis. This study could reveal whether PA-US imaging system is able to diagnose chemical-induced ICR mouse embryo malformations precisely and could be a new noninvasive strategy to assess the process of embryo development.

Materials and Methods

Ethics Statement

This study was conducted following the rules established by the Center for Molecular Imaging and Translational Medicine at Xiamen University. The protocol was approved by the Ethics Committee for Animal Experimentation of Xiamen University.

Animal Preparation

The ICR outcrossing mice were purchased from the Experimental Animal Center of Xiamen University (Huang et al. 2019; Yasuda et al. 1985). ICR mice ($\sim 30–35\mathrm{g}$) were housed in a temperature-controlled room (23°C) with a 12:12-h light:dark cycle and freely available food and water. Timed mating was set up overnight with 1:1 male mice to female mice ratio. Noon on the day of finding a vaginal plug was designated as embryonic day 0.5 (E0.5) (Huang et al. 2019). The pregnant ICR mice were weighed daily from E0.5 until the end of imaging. A total of 198 female mice were used in this study. A total of 27 mice (three experimental groups and nine mice analyzed for each group) were used for assessing daily weight changes from E0.5 to E15.5. A total of 108 mice (three experimental groups, six time points from E10.5 to E15.5 and 6 mice analyzed each time) were used for PA-US imaging and then sacrificed by cervical dislocation to harvest the embryos for bright view measurements and histopathological analysis. A total of 54 mice (three experimental groups and three mice analyzed for each group from E10.5 to E15.5) were used for flame-heated furnace atomic absorption spectrometry (FHF-AAS) analysis. A total of nine mice (three experimental groups and three mice analyzed for each group at E15.5) were sacrificed by cervical dislocation, and their hearts, livers, and kidneys were harvested and sectioned for continued hematoxylin and eosin (H&E) staining to assess the maternal toxicity of MMC.

Administration of MMC

MMC with a purity of greater than 96% (CAS #115-09-3; Aladdin Chemical Company) was dissolved in saline (CAS #7,647-14-5, Sigma). The MMC of the high-dose group was configured to be $1.05\mathrm{mg}/\mathrm{mL}$, and the low-dose group was configured to be $0.525\mathrm{mg}/\mathrm{mL}$. To investigate the prenatal effects of MMC during postimplantation, MMC was administrated by gavage once a day from gastrulation to the early organogenesis stages (E5.5–E10.5), which was considered to be the most sensitive stages for birth defects in mice (Savolainen et al. 2009). To assess the effects of different dosages, $7.5(\mathrm{mg}/\mathrm{kg}/\mathrm{d})$ (low dosage, LD) and $15(\mathrm{mg}/\mathrm{kg}/\mathrm{d})$ (high dosage, HD) MMC was administrated by gavage, respectively. The control group was given saline. Therefore, there were three experimental groups in total.

In Vivo PA-US Dual-Modality Imaging

The schematic of this study is shown in Figure 1. The laser wavelength of the PA-US imaging system (Vevo LAZR-X; Fujifilm VisualSonics Inc.) had a tunable range from 680 to $970\text{\hspace{0.17em}nm}$. PA-US signal collection was achieved through the hybrid PA-US transducer (Model: LS-550, central frequency: 40 MHz) simultaneously. This transducer consisted of two beams of fiber bundles with each piece inserted into either side of the transducer for photoacoustic signal excitation (Figure 1C). The incident laser energy density at the pregnant mouse abdomen surface was about $36.284{\text{\hspace{0.17em}mJ}/\mathrm{cm}}^2$. All the experiments used the following parameters: imaging depth: $15\mathrm{mm}$; width: $20\mathrm{mm}$; wavelength: $808\text{\hspace{0.17em}nm}$ for photoacoustic imaging, two wavelengths $750/850\text{\hspace{0.17em}nm}$ for ${\text{SaO}}_2$ and HbT imaging; gain value: 29 dB for ultrasound, 40 dB for photoacoustic imaging. During the imaging process, the ultrasound transducer was placed on the abdomen of the pregnant mouse, and the embryonic heart (beating area) was used as the region of interest (ROI) for scanning. After the imaging, the pregnant mouse was sacrificed by cervical dislocation, and three mouse embryos were selected randomly for fixation and testing. Visualsonics® workstation suite (version 3.1.0; VevoLab) was used for postprocessing of all imaging data. To calculate the concentration of ${\text{SaO}}_2$ and HbT based on PSI, the previously reported two-wavelength calculation method was used. In brief, for biotissue, the absorbance at the wavelengths used in this study is almost exclusively due to blood. Therefore, we can consider tissue as a very sparse absorber to generate allowedly approximate estimates of oxygen saturation. For this preliminary evaluation, based on these assumptions, a minimum of two wavelengths were collected to estimate relative total hemoglobin and oxygen saturation:

$$\mathrm{S}a{O}_2=\frac{\left({\text{HbO}}_2\right)}{\left({\text{HbO}}_2\right)+\left(\text{Hb}\right)}=\frac{A_{{\mathrm{\lambda }}_2}{\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_1}-A_{{\mathrm{\lambda }}_1}{\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_2}}{A_{{\mathrm{\lambda }}_1}{\mathrm{\Delta }\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_2}-A_{{\mathrm{\lambda }}_2}{\mathrm{\Delta }\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_1}},$$ (1)

$$\text{HbT}=\left({\text{HbO}}_2\right)+\left(\text{Hb}\right)=\frac{A_{{\mathrm{\lambda }}_1}{\mathrm{\Delta }\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_2}-A_{{\mathrm{\lambda }}_2}{\mathrm{\Delta }\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_1}}{{\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_1}{\mathrm{\Delta }\mathrm{\epsilon }}_{{\text{HbO}}_2}^{{\mathrm{\lambda }}_2}-{\mathrm{\epsilon }}_{\text{Hb}}^{{\mathrm{\lambda }}_2}{\mathrm{\Delta }\mathrm{\epsilon }}_{{\text{HbO}}_2}^{{\mathrm{\lambda }}_1}},$$ (2)

where $\mathrm{\Delta }\mathrm{\epsilon }\text{Hb}={\mathrm{\epsilon }\text{HbO}}_2-\mathrm{\epsilon }\text{Hb}$. Brief, by imaging at least two wavelengths, ${\mathrm{\lambda }}_1$ and ${\mathrm{\lambda }}_2$, it was possible to estimate the optical energy deposition at each location in space, and (${\mathrm{A}}_{\mathrm{\lambda }}$) was photoacoustic signal (Needles et al. 2013; Xia et al. 2013).

Figure 1.

Experimental design. (A) Mice were given MMC by gavage from embryonic day (E5.5 to E10.5). (B) Experiments were divided into control, LD, and HD groups. Mice in the control group were given saline by gavage, those in the LD group were given a dose of $7.5\mathrm{mg}/\mathrm{kg}/\mathrm{d}$, and those in the HD group were given a dose of $15\mathrm{mg}/\mathrm{kg}/\mathrm{d}$. (C) A mouse was placed on the heating pad, ready for imaging. (D) Through use of the dual modality PA-US imaging system, ultrasound images, PSI images, ${\text{SaO}}_2$, and HbT signals near the mouse embryonic heart, functional and structural information of the mouse embryonic heart can be obtained. The embryonic heart areas are outlined with white circles, based on heartbeat position. Note: HbT, hemoglobin content; HD, high dosages; LD, low dosages; MMC, methylmercury chloride; PA-US, photoacoustic-ultrasound; PSI, photoacoustic signal intensity; ${\text{SaO}}_2$, blood oxygen saturation.

Imaging Protocol

Before imaging, the pregnant mouse was anesthetized with 2.5% isoflurane (RWD Life Science) with $2\mathrm{L}/\text{min}$ flow rate and then chemically depilated with a soft depilatory cream (WT200; Solarbio). To avoid the effect of the fluctuation caused by mouse, we custom-built a facial oxygen, anesthesia mask, and mouse body fixed device (Figures S1A, S1B, and S1C). Moreover, for all of the in vivo experiments, pregnant mice were immobilized by taping their limbs. In addition, a physiological monitoring system (THM-150 Advanced Physiological Monitoring Unit) was used to monitor physiological status of the mouse (Figure S1D). During PA-US image acquisition, the mouse was scaffolded on a multifunctional imaging platform with a 37°C thermostatic pad under it and supplied with 1.5% isoflurane and $3\mathrm{L}/\text{min}$ air. The rectal probe (after lubrication) was gently inserted to monitor body temperature because the constant body temperature was essential for the stability of maternal and infant hemodynamic. The ultrasonic coupling gel (CAS #JY0001; Shanghai Xinyu Biological Technology Co. Ltd.) was smeared on the surface of the pregnant mouse abdomen to promote the transmission of the ultrasound signal. For each developmental time point, six mice were used in each MMC dosage group for imaging, and three embryos were carefully imaged and analyzed for each mouse. The mice were sacrificed by cervical dislocation right after imaging, and three embryos from each mouse were collected randomly and fixed for further histopathological analysis. The mice were not kept alive for continuously imaging, considering that the frequent inhalation of isoflurane could have effects on the embryos (Chetkowski and Nass 1988; Warren et al. 1992). To test the consistency of PSI, we compared the in utero imaging results to ex vivo imaging. Immediately after the dams were sacrificed, embryos were delivered by cesarian section and quickly imaged using PSI. Specifically, after a mouse embryo was removed from the womb quickly, saline was added to the mouse embryo to keep it alive. The mouse embryos were placed in a custom-made plastic dish to prevent them from moving randomly during the imaging process. The mouse embryos were then coated with an ultrasonic coupling agent, and the PA/US probe was gently placed on the embryos by a professional operator. The position of the mouse embryonic heart was identified by the beating heart and the PSI signal was detected in the embryo using an $808\text{\hspace{0.17em}nm}$ laser (Figure S2).

For ultrasound imaging, the maximum cross-section setting was used to obtain information about the number and location of the embryos. The number of the embryos was counted using ultrasound imaging according to the heartbeat. The number of embryos was recorded more clearly by splitting the screen and comparing the images to determine whether they had been repeated. The length of the embryos was measured by the Visualsonics® workstation suite. Crown–rump length (CRL) was used to determine the length of the embryos, which was one of the critical factors to assess the general growth and health of the embryos (Weiss 1973). In addition, the PA-US modality imaging was also used to measure the heart rate of the embryo. Due to the small size of mouse embryos and their different positions, a nontraditional ultrasound imaging plane was used to determine the body axis of embryos (Kim 2013). Besides, the PSI and ${\text{SaO}}_2$ were also investigated at embryonic day from E10.5 to E12.5 after administration of different doses of MMC (Figures S3–S5). Moreover, we also checked the maternal skin tissue ${\text{SaO}}_2$ levels on the abdomen from E10.5 to E15.5 among the control group, LD group, and HD group after administration of different doses of MMC (Figure S6).

Furthermore, all the 3D embryo images were obtained by setting the xyz-axis position of the hybrid PA-US transducer. After the midpoint of scanning, the transducer was selected, moved to both sides, and then returned to the scanning origin (step size: $0.5\mathrm{mm}$; time elapsed: 30 s). When the embryo slice just appeared on the screen as the starting point of the scan, it could be determined that the entire embryo was within the scan range, and the scan ended after the embryo completely disappeared. The obtained images were loaded into the Visualsonics® workstation suite for 3D reconstruction. Manual definition of embryonic heart was required for volumetric quantification. The maximum cross-section of the beating heart was selected as the ROI. Then, the volume function of the software was used to determine the 3D structure of the heart through the beating heart frame by frame. Finally, the software automatically calculated the volume of the mouse embryonic heart. It was not necessary to define the ROI object in every frame, but it was defined every 5–10 frames (Freeling and Rezvani 2016).

Embryo Dissection

After PA-US imaging, the pregnant ICR mice were sacrificed by cervical dislocation. More details of the embryo dissection procedure were described according to the previous study (Copp and Cockroft 1999). Briefly, the skin was cut with ophthalmological scissors to expose to the intact abdominal cavity of the pregnant mouse, the cervix was cut off, and embryos were placed in a petri dish (705001; Nest) containing DMEM medium (11965092; ThermoFisher). The myometrium was carefully torn apart using two pairs of fine forceps (AGT5034; Dumont). Subsequently, the deciduae were peeled away carefully to prevent the amniotic cavities from being punctuated. Last, Reichert’s membrane was removed by pinching the membrane with forceps and then slowly separated from the embryo. To further investigate the morphological changes of embryos after treatment with MMC, we obtained white light photos of embryos from E10.5 to E12.5 (Figure S7) and E13.5 to E15.5 (Figure S8) in the control group, the LD group, and HD group with optical microscope (ISH500; Nikon), respectively. Furthermore, severe malformations such as delayed brain development and dysplasia of limbs were also investigated in the HD group (Figure S9).

Embryo Section and Staining

The embryos were fixed with 4% paraformaldehyde (CAS #158127; Sigma) for 48 h at 4°C and then washed with phosphate-buffered saline (PBS; CAS #10010023; ThermoFisher), followed by dehydration stepwise with 30:70, 50:50, 70:30, 80:20, 90:10, and 100:0 mixtures of ethanol (CAS #E7023; Sigma) and xylene (CAS #534056; Sigma) for 40 min per step. The embryos were then embedded in a paraffin block (CAS #107150; Sigma) and cut into $4\text{-}\mathrm{\mu }\mathrm{m}\text{-thick}$ slices by a paraffin microtome (Leica-RM2245). Before staining, the paraffin sections were dewaxed and rehydrated stepwise in xylene, 100% alcohol, and 95% alcohol for 30 s per step. The sections were rinsed in water and then stained with conventional H&E (Li et al. 2011). Briefly, the slices were stained by hematoxylin solution (CAS #H9627, ThermoFisher) for 5 min and differentiated by hydrochloric acid alcohol. Next, the slices were stained with eosin solution (CAS #E4009; Sigma) for 8 min. The maximum cross-section of the heart of embryos was selected to quantitatively analyze the change of heart wall in different study groups. To reduce the impact on individual differences, at least three embryos were chosen in each pregnant mouse. Three fields of view on each slice were chosen using a microscope (IX51; Olympus) for pathological observation. In addition, the average thickness of the heart wall was used to assess the normality of the heart development with H&E images (Figure S10). This quantitative method was briefly described with the representative sample (Figure S10A). According to the previous studies, at least five different positions of the heart wall were chosen and labeled using the yellow lines as shown in Figure S10A and then quantified by ImageJ software. (version 2.1.4.7; U.S. National Institutes of Health) (Chen et al. 2020; He et al. 2020).

Flame-Heated Furnace Atomic Absorption Spectrometry (FHF-AAS) Analysis for Measuring the Mercury Concentration in the Embryos

More details regarding the measuring the mercury concentration in the embryos were described in our previous study (Huang et al. 2019). Briefly, first, the embryos were placed in 70% nitric acid (CAS #7697-37-2, Sigma) for nitrification overnight and then digested via a microwave digester (CEM MARS6). The program settings were as follows: ramp for 10 min; hold for 15 min; temperature 190°C; pressure 800 psi (${\mathrm{l}\text{\hspace{0.17em}bf}/\text{in}}^2$); power 1,600 W. Then the obtained digestion solution was heated and evaporated to dryness, and 5% diluted nitric acid was used to dissolve the solid residue to form the sample solution. The standard curve was made by diluting $1\mathrm{mg}/\mathrm{mL}$ standard MMC solutions with 3% nitric acid to the following concentrations: 0, 1, 5, 10, 25, 50, 100, and $200\text{\hspace{0.17em}ppm}$. The absorbance (Abs) value of sample solutions was measured using flame-heated furnace atomic absorption spectrometry (FHF-AAS, Shimadzu AA-6300C; Shimadzu). The mercury concentration was calculated according to the standard curve.

Statistical Test and Analysis

Statistical analysis was performed using SPSS software (version 3.1.1; IBM). Quantitative variables were presented as $\text{means}\pm \text{standard}$ deviations (SD). The analysis of variance (ANOVA) was performed on the weight of pregnant mice, embryo body length, PSI, ${\text{SaO}}_2$ and HbT value, and the heart volume of the embryo. In addition, the independent sample $t$-test was used to measure the heart wall thickness of mouse embryos, the PSI of mouse embryonic heart in and out of the uterus, the PSI of mouse embryonic heart and amniotic fluid in utero, and the heart rates of mouse embryos; $p<0.05$ was defined to have a significant difference.

Results

The Effects of MMC on the Pregnant Mice and the Number of Their Embryos

The effects of MMC on the body weight (BW) of pregnant mice were first investigated. As shown in Figure 2A; Table 1, the BW of the control mice gradually increased in the first 8 d of the pregnancy, followed by much faster growth from E8.5 to E15.5 (Figure 2A; Table 1). In the HD group, the weight of the pregnant mice increased in a similar pattern to the control group during the first 8 d (Figure 2A; Table 1). However, the pregnant mice showed a significant reduction in their BW, starting from E10.5 (Figure 2A; Table 1). The average BW in the HD group was only 0.716-fold and 0.660-fold of the control group at E14.5 and E15.5, respectively (Figure 2A; Table 1). In the LD group, the weight of pregnant mice seemed to increase more slowly than that of the control group after E11.5, but there was no significant difference between these two groups from E0.5 to E15.5 (Figure 2A; Table 1). Then pregnant mice were sacrificed by cervical dislocation at E15.5. The heart, liver, and kidney were harvested and sectioned for H&E staining to assess the maternal toxicity of MMC (Figure 3). As shown in the Figure 3D, the liver tissues in the LD group did not show clear pathological differences from control group. Meanwhile, in the HD group, the cardiomyocytes showed increasing eosinophilia, concentrated nuclei, and clear apoptotic regions (Figure 3H). In addition, in the HD group, the nuclei of the liver cells were condensed, and the cell bodies were smaller. The cytoplasm was denser in comparison with that of the control group, suggesting that some liver cells underwent apoptosis (Figure 3G).There were no apparent abnormalities in the kidney in either the LD or the HD groups (Figure 3F and Figure 3I).

Figure 2.

Effects of MMC on pregnant mice and their embryos. (A) Body weight of dams under three treatments ($n=9$ mice per treatment group) from gestational day 0 through gestational day 16 (as indicated on the x-axis). MMC was administered by gavage from embryonic day E5.5 to E10.5. (B) The number of embryos per pregnant mouse from E10.5 to E15.5 ($n=6$ mice per treatment group). (C) A representative ultrasound image showing an E15.5 mouse embryo in the control group. (D) Changes in embryonic CRL among three treatments from E10.5 to E15.5 ($n=6$ embryos per treatment group). Error bars indicated $\text{mean}\pm \text{SD}$. The individual data in (A) are shown in Table 1. The individual data in (B) and (D) are shown in Table 3. Note: ANOVA, analysis of variance; CRL, crown–rump length; HD, high dosages; LD, low dosages; LSD, least significant difference; MMC, methylmercury chloride; SD, standard variation. *$p<0.05$; **$p<0.01$; ***$p<0.001$; all * relative to control [analysis of variance (ANOVA) with one-way ANOVA using LSD test].

Table 1.

Body weight of pregnant mice.

	Control [$\text{mean}\pm \text{SD}$ ($n$)]	LD [$\text{mean}\pm \text{SD}$ ($n$)]	HD [$\text{mean}\pm \text{SD}$ ($n$)]
Day	Body weight (g)	Body weight (g)	Body weight (g)
E0.5	$33.44\pm 0.829$ (9)	$31.52\pm 1.020$ (9)	$31.55\pm 0.970$ (9)
E1.5	$33.64\pm 0.588$ (9)	$32.315\pm 1.182$ (9)	$33.04\pm 1.013$ (9)
E2.5	$33.46\pm 0.828$ (9)	$32.91\pm 0.968$ (9)	$34.49\pm 1.319$ (9)
E3.5	$34.34\pm 0.705$ (9)	$33.47\pm 1.016$ (9)	$35.16\pm 1.437$ (9)
E4.5	$34.61\pm 0.624$ (9)	$34.06\pm 1.592$ (9)	$35.87\pm 1.530$ (9)
E5.5	$35.17\pm 0.511$ (9)	$34.49\pm 1.449$ (9)	$37.44\pm 1.028$ (9)
E6.5	$35.74\pm 0.653$ (9)	$35.18\pm 1.561$ (9)	$38.07\pm 0.998$ (9)
E7.5	$36.25\pm 0.781$ (9)	$36.47\pm 1.379$ (9)	$38.78\pm 1.161$ (9)
E8.5	$37.49\pm 0.983$ (9)	$37.11\pm 1.415$ (9)	$39.32\pm 1.252$ (9)
E9.5	$38.41\pm 1.080$ (9)	$37.77\pm 1.551$ (9)	$39.82\pm 1.386$ (9)
E10.5	$39.21\pm 1.604$ (9)	$38.52\pm 1.128$ (9)	$40.186\pm 0.659$ (9)
E11.5	$41.03\pm 1.071$ (9)	$40.15\pm 1.474$ (9)	$39.22\pm 0.709$ (9)**
E12.5	$43.68\pm 0.913$ (9)	$41.40\pm 1.270$ (9)	$36.06\pm 1.472$ (9)***
E13.5	$45.17\pm 0.685$ (9)	$42.66\pm 1.326$ (9)	$34.93\pm 1.37$ (9)***
E14.5	$46.94\pm 1.567$ (9)	$43.80\pm 1.256$ (9)	$33.62\pm 1.470$ (9)***
E15.5	$49.94\pm 1.408$ (9)	$45.38\pm 1.387$ (9)	$32.96\pm 1.996$ (9)***

Note: ANOVA, analysis of variance; E, embryonic day; HD, high dosage; LD, low dosage; LSD, least significant difference; SD, standard deviation. $n=9$ mice per treatment group. **$p<0.01$; ***$p<0.001$; all * relative to control (ANOVA with one-way ANOVA using LSD test).

Figure 3.

Representative H&E staining images showing maternal liver, heart, and kidney tissues at E15.5 in (A–C) the control group; (D–F) LD group; and (G–I) HD group. The black arrows showed the sites of apoptosis. Note: E, embryonic day; HD, high dosages; H&E, hematoxylin and eosin; LD, low dosages.

We then investigated whether MMC could penetrate the placenta barriers and affect the embryos. The mercury concentration in the embryos was determined by FHF-AAS. The average mercury concentration in the embryos of the LD group increased from $6.33\pm 2.36\text{\hspace{0.17em}ppm}$ at E10.5 to $44.67\pm 4.32\text{\hspace{0.17em}ppm}$ at E15.5. In addition, the average mercury concentration in the embryos of the HD group increased from $13.00\pm 4.08\text{\hspace{0.17em}ppm}$ at E10.5 to $129.67\pm 14.33\text{\hspace{0.17em}ppm}$ at E15.5 (Table 2). The results showed that in both the HD and LD groups, the mercury concentration was significantly higher than those in the control group (Table 2), which meant that MMC could enter the embryo through the placental barrier and present even after the withdrawal of MMC from the pregnant mice. Then, we further looked at whether these levels of MMC would affect the number of embryos. The number of embryos in each pregnant mouse was counted in ultrasound mode using the PA-US imaging system (Figure 2B; Table 3). In the control group, the average numbers of embryos from E13.5 to E15.5 and newborn mice were 10.6, 9.6, 9.6, and 8.6, respectively (Figure 2B; Table 3). In comparison with the control, there was no significant difference in the number of live embryos from E13.5 to E14.5 in the LD group (Figure 2B; Table 3). But the average number of embryos on E15.5 was 6.6, and the number of newborn mice was only 4.3, which was significantly lower than the control group (Figure 2B; Table 3). In the HD group, the average number of embryos was only 6.6 at E10.5 (Figure 2B; Table 3). Moreover, this number kept decreasing; at E15.5 there was less than an average of one embryo in the HD group (Figure 2B; Table 3).

Table 2.

Mercury concentration in the embryos from E10.5 to E15.5 among the control group, LD group, and HD group.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)] (ppm)	LD [$\text{mean}\pm \text{SD}$ ($n$)] (ppm)	HD [$\text{mean}\pm \text{SD}$ ($n$)] (ppm)
E10.5	$0.051\pm 0.07$ (3)	$6.33\pm 2.36$ (3)	$13.00\pm 4.08$ (3)**
E11.5	$0.075\pm 0.01$ (3)	$12.00\pm 1.47$ (3)**	$20.00\pm 1.76$ (3)****
E12.5	$0.054\pm 0.02$ (3)	$13.33\pm 3.68$ (3)*	$26.33\pm 8.49$ (3)**
E13.5	$0.012\pm 0.01$ (3)	$21.33\pm 2.36$ (3)**	$29.67\pm 6.24$ (3)****
E14.5	$0.032\pm 0.01$ (3)	$39.67\pm 6.23$ (3)****	$73.00\pm 4.08$ (3)****
E15.5	$0.042\pm 0.01$ (3)	$44.67\pm 4.32$ (3)****	$129.67\pm 14.33$ (3)****

Note: ANOVA, analysis of variance; E, embryonic day; HD, high dosage; LD, low dosage; LSD, least significant difference; SD, standard deviation. $n=3$ mice per treatment group. *$p<0.05$; **$p<0.01$; ****$p<0.0001$; all * relative to control [analysis of variance (ANOVA) with one-way ANOVA using LSD test].

Table 3.

Effects of MMC on the embryos.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]		LD [$\text{mean}\pm \text{SD}$ ($n$)]		HD [$\text{mean}\pm \text{SD}$ ($n$)]
	($n$)	CRL (mm)	($n$)	CRL (mm)	($n$)	CRL (mm)
E10.5	$10.0\pm 1.4$ (6)	$3.43\pm 0.21$ (6)	$9.6\pm 1.0$ (6)	$3.38\pm 0.20$ (6)	$6.6\pm 2.1$ (6)*	$3.06\pm 0.47$ (6)
E11.5	$9.0\pm 0.7$ (6)	$5.52\pm 0.09$ (6)	$8.0\pm 0.8$ (6)	$5.33\pm 0.09$ (6)	$7.0\pm 1.4$ (6)*	$3.65\pm 0.12$ (6)***
E12.5	$10.0\pm 1.4$ (6)	$6.24\pm 0.25$ (6)	$9.3\pm 2.1$ (6)	$6.14\pm 0.16$ (6)	$4.0\pm 0.7$ (6)*	$3.87\pm 0.03$ (6)***
E13.5	$10.6\pm 1.0$ (6)	$8.91\pm 0.53$ (6)	$9.6\pm 1.0$ (6)	$8.60\pm 0.07$ (6)	$1.3\pm 0.4$ (6)***	$7.133\pm 0.07$ (6)**
E14.5	$9.6\pm 1.0$ (6)	$9.38\pm 0.12$ (6)	$8.6\pm 0.4$ (6)	$9.13\pm 0.09$ (6)	$0.6\pm 0.4$ (6)***	$6.42\pm 0.17$ (6)***
E15.5	$9.6\pm 1.2$ (6)	$12.04\pm 0.10$ (6)	$6.6\pm 0.4$ (6)**	$11.55\pm 0.27$ (6)	$0.3\pm 0.4$ (6)***	$6.00\pm 0.08$ (6)***
Born	$8.6\pm 0.9$ (6)	—	$4.33\pm 0.4$ (6)**	—	$0.0\pm 0.0$ (6)	—

Note: —, no data; ANOVA, analysis of variance; CRL, crown–rump length; E, embryonic day; HD, high dosage; LD, low dosage; LSD, least significant difference; MMC, methylmercury chloride; ($n$), the number of embryos per pregnant mouse; SD, standard deviation. $n=6$ mice per treatment group. *$p<0.05$; **$p<0.01$; ***$p<0.001$; all * relative to control (ANOVA with one-way ANOVA using LSD test).

The Effects of MMC on Embryonic CRL and Fetal Heart Rate

Although the LD of MMC did not seem to influence the number of the embryos before E15.5, it was still necessary to determine whether it affected the morphology of the embryos in the early stages. CRL of the embryos was measured by ultrasound mode in the PA-US imaging system as described in the “Materials and Methods” section (Figure 2C) (Weiss 1973). In the control group, CRL gradually increased from E10.5 to E15.5 (Figure 2D; Table 3). In the LD group, the gross morphology of the embryos showed no difference in comparison with the control group in both in utero ultrasound images (Figure 2D; Table 3) and bright light images (Figure S7–S8) after dissection. In the HD group, the embryos were severely dysplastic, and the size was even reduced in the later stages (Figure 2D; Figures S7G–S7I and Figures S8G–S8I; Table 3). Therefore, the embryo length was significantly shorter than that in the control group from E11.5 to E15.5 (Figure 2D; Table 3). According to the ultrasound scan, the LD of MMC did not seem to cause significant morphological changes in the embryos in comparison with the control group. However, the HD group induced severe embryonic malformation.

Because the fetal heart rate was considered an essential indicator for the health of the embryos, we then recorded it using the PA-US imaging system in ultrasound mode. In the LD group, the heart rate was not significantly different from that of the control group at E13.5 and E14.5 [Figure 4A; Table 4; Supplementary Videos 4 and 5 (LD) compared to Supplementary Videos 1 and 2 (control), respectively], but much lower at E15.5 [Figure 4A; Table 4; Supplementary Video 6 (LD) compared to Supplementary Video 3 (control)]. The results showed that the hearts of the embryos in the HD group stopped beating during E13.5 to E15.5 (Supplementary Videos 7–9).

Figure 4.

(A) The heart rates in the control and LD groups from E13.5 to E15.5 ($n=6$ embryos per treatment group). * relative to control (Independent sample $t$ test was used for analyzing the heart rate). The individual data in (A) are shown in Table 4. (B, C, and D). PSI, ${\text{SaO}}_2$, and HbT changes detected by PA-US imaging ($n=6$ embryos per treatment group). * relative to control (ANOVA with One-way analysis of variance using LSD test). The individual data in (B, C, and D) are shown in Table 5. Note: ANOVA, analysis of variance; E, embryonic day; LD, low dosages; HD, high dosages; LSD, least significant difference; PA-US, photoacoustic-ultrasound; PSI, photoacoustic signal intensity; ${\text{SaO}}_2$, blood oxygen saturation; HbT, hemoglobin content; SD, standard deviation. *$p<0.05$; **$p<0.01$; ***$p<0.001$; ****$p<0.0001$; Error bars indicated mean ± s.d.

Table 4.

The heart rates among the control group and the LD group.

Day	Control [$\text{mean}\pm \text{SD}$ (n)]	LD [$\text{mean}\pm \text{SD}$ (n)]
	Heart rates (beats/min)	Heart rates (beats/min)
E13.5	$102\pm 6.48$ (6)	$80\pm 5.66$ (6)
E14.5	$104\pm 6.2$ (6)	$84\pm 16.97$ (6)
E15.5	$127.33\pm 5.05$ (6)	$16\pm 23.10$ (6)**

Note: E, embryonic day; HD, high dosage; LD, low dosage; SD, standard deviation. $n=6$ mice per treatment group. **$p<0.01$; all * relative to control (independent sample $t$-test was used for analyzing heart rates).

PSI, ${\text{SaO}}_2$, HbT Measures Using Dual-Modality PA-US Imaging in Mice Exposed to MMC

Because heart development is an essential healthy diagnostic indicator for early embryo development, we quantified PSI, ${\text{SaO}}_2$, and HbT signals in the heart region as described in the “Materials and Methods” section to monitor the effects of MMC on the embryos in utero. The reliable PSI, ${\text{SaO}}_2$, and HbT signals detection began at E10.5, E11.5 and E13.5, respectively (Figure 4B–4D). In the control group, all three signals increased with the continuous development of embryos (Figure 4B–4D).

In the LD group, all three signals decreased after E13.5 (Figure 4B–4D; Table 5) which were not statistically significantly different from the control group at day E13.5 (Figure 5P–5R; Table 5). but were significantly lower at E14.5 (Figure 6P–6R; Table 5). The values of PSI, ${\text{SaO}}_2$, and HbT at E15.5 were 0.415-, 0.684-, and 0.595-fold of the control group, respectively. These results suggest that the functional abnormality of the embryos in the LD group started at E14.5, which could cause some level of embryonic defects in the later stages. In the HD group, PSI was no different than the control group at E10.5 (Figure 4A; Figure S3M; Table 5). Although PSI continued to increase from E11.5 to E12.5, they were significantly lower than the control group (Figure S4M and Figure S5M; Table 5). After E12.5, instead of increasing as the in control group, PSI in the HD group decreased (Figure 4B; Table 5). The ${\text{SaO}}_2$ and HbT signals became detectable at E11.5 and E13.5, respectively, the same as in the control group (Figure 4C and 4D; Table 5). However, the signals were significantly lower than those in the control group at this time (Figure 5R; Table 5). The ${\text{SaO}}_2$ increased from E11.5 to E13.5 but decreased afterward (Figure 4C; Table 5), whereas the HbT decreased gradually from E13.5 (Figure 4D; Table 5). Also, we checked the ${\text{SaO}}_2$ levels in maternal abdominal skin from E10.5 to E15.5. Our result showed no significant difference in the ${\text{SaO}}_2$ levels in maternal abdominal skin between the LD and control groups (Figure S6B; Table 6). However, in the HD group, the ${\text{SaO}}_2$ levels in maternal abdominal skin were significantly lower than those in the control group from E14.5 (Figure S6B; Table 6). At E15.5, PSI, ${\text{SaO}}_2$, and HbT in the HD group were 0.244-, 0.230-, and 0.198-fold of those in the control group, respectively (Figure 7P–7R; Table 5). These results suggested that the functional abnormality of the embryos in the HD group started at E11.5 and may have led to more severe defects at later stages with MCC enrichment. Although these results are subject to the performance of detector, it is worth noting that the ${\text{SaO}}_2$ is gradually decreased over time, during which MCC concentrations increased in the embryos. In the HD group, the embryos showed a difference in ${\text{SaO}}_2$ at E11.5, but the maternal difference in ${\text{SaO}}_2$ occurred at E14.5, so it is possible that embryonic toxicity precedes maternal toxicity and perhaps is an early indication of MCC toxicity in embryo.

Table 5.

Changes of PSI, ${\text{SaO}}_2$, and HbT signals in the embryonic heart from E10.5 to E15.5 among the control group, LD group, and HD group.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]			LD [$\text{mean}\pm \text{SD}$ ($n$)]			HD [$\text{mean}\pm \text{SD}$ ($n$)]
	PSI	${\text{SaO}}_2$ (%)	HbT	PSI	${\text{SaO}}_2$ (%)	HbT	PSI	${\text{SaO}}_2$ (%)	HbT
E10.5	$0.13\pm 0.02$ (6)	0	—	$0.12\pm 0.01$ (6)	0	—	$0.11\pm 0.01$ (6)	0	—
E11.5	$0.21\pm 0.01$ (6)	$23.9\pm 1.00$ (6)	—	$0.18\pm 0.03$ (6)	$21.1\pm 0.90$ (6)	—	$0.14\pm 0.02$ (6)*	$4.33\pm 0.86$ (6)***	—
E12.5	$0.24\pm 0.02$ (6)	$26.1\pm 2.10$ (6)	—	$0.23\pm 0.01$ (6)	$25.7\pm 1.7$ (6)	—	$0.18\pm 0.02$ (6)**	$11.4\pm 0.03$ (6)***	—
E13.5	$0.29\pm 0.04$ (6)	$28.1\pm 1.32$ (6)	$12.6\times {10}^3\pm 1.44\times {10}^3$ (6)	$0.23\pm 0.03$ (6)	$24.8\pm 2.27$ (6)	$11.5\times {10}^3\pm 1.23\times {10}^3$ (6)	$0.13\pm 0.04$ (6)****	$13.4\pm 1.33$ (6)***	$4.11\times {10}^3\pm 0.33\times {10}^3$ (6)*
E14.5	$0.37\pm 0.01$ (6)	$30.0\pm 3.12$ (6)	$14.4\times {10}^3\pm 0.84\times {10}^3$ (6)	$0.18\pm 0.03$ (6)****	$23.1\pm 3.15$ (6)***	$9.62\times {10}^3\pm 0.76\times {10}^3$ (6)*	$0.09\pm 0.03$ (6)****	$8.20\pm 2.00$ (6)****	$3.52\times {10}^3\pm 0.53\times {10}^3$ (6)**
E15.5	$0.41\pm 0.03$ (6)	$31.0\pm 2.01$ (6)	$14.8\times {10}^3\pm 0.56\times {10}^3$ (6)	$0.17\pm 0.04$ (6)***	$21.2\pm 4.18$ (6)**	$8.80\times {10}^3\pm 0.53\times {10}^3$ (6)*	$0.1\pm 0.01$ (6)****	$7.12\pm 2.00$ (6)***	$2.93\times {10}^3\pm 0.46\times {10}^3$ (6)****

Note: —, no data; ANOVA, analysis of variance; E, embryonic day; HbT, hemoglobin content; HD, high dosage; LD, low dosage; LSD, least significant difference; PSI, photoacoustic signal intensity; ${\text{SaO}}_2$, blood oxygen saturation; SD, standard deviation. $n=6$ mice per treatment group. *$p<0.05$; **$p<0.01$; ***$p<0.001$; ****$p<0.0001$; all * relative to control (ANOVA with one-way ANOVA using LSD test).

Figure 5.

PSI, ${\text{SaO}}_2$, and HbT signals at E13.5 after administration of different doses of MMC. (A–E) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the control group. (F–J) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the LD group. The heart areas were outlined with a white circle based on heartbeat position. (K–O) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the HD group. (P–R) Comparison of PSI, ${\text{SaO}}_2$, and HbT signals in the heart areas in different groups at E13.5, respectively ($n=6$ embryos per treatment group). *$p<0.05$; ***$p<0.001$; ****$p<0.0001$; all * relative to control (ANOVA with one-way ANOVA using LSD test). Error bars indicated $\text{mean}\pm \text{SD}$. The individual data in (P–R) are shown in Table 5. Note: ANOVA, analysis of variance; E, embryonic day; HbT, hemoglobin content; HD, high dosages; LD, low dosages; LSD, least significant difference; MMC, methylmercury chloride; PSI, photoacoustic signal intensity; ${\text{SaO}}_2$, blood oxygen saturation; SD, standard deviation; US, ultrasound.

Figure 6.

PSI, ${\text{SaO}}_2$, and HbT signals at E14.5 after administration of different doses of MMC. (A–E) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the control group. (F–J) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the LD group. The heart areas were outlined with a white circle based on heartbeat position. (K–O) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the HD group. (P–R) Comparison of PSI, ${\text{SaO}}_2$, and HbT signals in the heart areas in different groups at E14.5, respectively ($n=6$ embryos per treatment group). *$p<0.05$; **$p<0.01$; ***$p<0.001$; ****$p<0.0001$; all * relative to control (ANOVA with one-way ANOVA using LSD test). Error bars indicated $\text{mean}\pm \text{SD}$. The individual data in (P–R) are shown in Table 5. Note: ANOVA, analysis of variance; E, embryonic day; HbT, hemoglobin content; HD, high dosages; LD, low dosages; LSD, least significant difference; MMC, methylmercury chloride; PSI, photoacoustic signal intensity; ${\text{SaO}}_2$, blood oxygen saturation; SD, standard deviation; US, ultrasound.

Table 6.

The ${\text{SaO}}_2$ signals in maternal abdominal skin among the control, LD, and HD groups from E10.5 to E15.5.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]	LD [$\text{mean}\pm \text{SD}$ ($n$)]	HD [$\text{mean}\pm \text{SD}$ ($n$)]
	${\text{SaO}}_2$ (%)	${\text{SaO}}_2$ (%)	${\text{SaO}}_2$ (%)
E10.5	$30.29\pm 1.10$ (6)	$32.14\pm 0.80$ (6)	$32.77\pm 0.35$ (6)
E11.5	$30.56\pm 1.16$ (6)	$31.73\pm 1.03$ (6)	$31.26\pm 0.79$ (6)
E12.5	$31.43\pm 0.99$ (6)	$32.57\pm 0.62$ (6)	$31.25\pm 0.94$ (6)
E13.5	$31.26\pm 1.71$ (6)	$29.16\pm 1.25$ (6)	$30.34\pm 1.86$ (6)
E14.5	$30.43\pm 1.22$ (6)	$30.75\pm 1.74$ (6)	$18.93\pm 1.67$ (6)****
E15.5	$30.64\pm 1.80$ (6)	$29.76\pm 1.29$ (6)	$9.20\pm 0.50$ (6)****

Note: ANOVA, analysis of variance; E, embryonic day; HD, high dosage; LD, low dosage; LSD, least significant difference; ${\text{SaO}}_2$, blood oxygen saturation; SD, standard deviation. $n=6$ embryos per treatment group. ****$p<0.0001$; all * relative to control (ANOVA with one-way ANOVA using least significant difference LSD test).

Figure 7.

PSI, ${\text{SaO}}_2$, and HbT signals at E15.5 after administration of different doses of MMC. (A–E) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the control group. (F–J) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the LD group. The heart areas were outlined with a white circle based on heartbeat position. (K–O) Representative image of PSI, ${\text{SaO}}_2$, and HbT in the HD group. (P–R) Comparison of PSI, ${\text{SaO}}_2$, and HbT signals in the heart areas in different groups at E15.5, respectively ($n=6$ embryos per treatment group). *$p<0.05$; **$p<0.01$; ***$p<0.001$; ****$p<0.0001$; all * relative to control (ANOVA with one-way ANOVA using LSD test). Error bars indicated $\text{mean}\pm \text{SD}$. The individual data in (P–R) are shown in Table 5. Note: ANOVA, analysis of variance; E, embryonic day; HbT, hemoglobin content; HD, high dosages; LD, low dosages; LSD, least significant difference; MMC, methylmercury chloride; PSI, photoacoustic signal intensity; ${\text{SaO}}_2$, blood oxygen saturation; SD, standard deviation; US, ultrasound.

In addition, we studied the ratio of PSI signals between mouse embryonic heart and amniotic fluid; the signal strength of PSI in the heart region was found to be greater than 2-fold that of the amniotic fluid at the same depth in the control and LD groups. However, in the HD group, the ratio of PSI in the heart area to that in the amniotic fluid at the same depth was less than 2-fold, which may be due to the fact that in the HD group, most of embryos had died and were absorbed by the mother, so the PSI near the heart region was the same as that in the surrounding amniotic fluid (Table 7). A 3D reconstruction based on the constitutive sections from ultrasound images was performed to carefully examine the anatomical structure of the embryos (Figure S11). The volume of the embryonic heart was calculated as described in the “Materials and Methods” section. In the control group, the volume of the embryonic heart gradually increased from $0.261{\mathrm{mm}}^3$ to $1.577{\mathrm{mm}}^3$ during E10.5–E15.5 (Figure S12A–S12F; Table 8). The heart volume did not show any significant difference in the LD group in comparison with the control group at E13.5 (Figure S12S; Table 8). However, at E14.5, the heart volume did not increase significantly in the LD group (Figure S12S; Table 8). Moreover, the heart volume in the LD group was 0.811-fold and 0.623-fold lower in comparison with the control group at E14.5 and E15.5, respectively (Figure S12S; Table 8). For the HD group, the size of the heart was significantly smaller than that in the control group at E11.5 (Figure S12S; Table 8). Because the embryonic heartbeat stopped during E13.5–E15.5, there was no 3D reconstruction performed at these stages.

Table 7.

PSI signal ratio between mouse embryonic heart and amniotic fluid.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]			LD [$\text{mean}\pm \text{SD}$ ($n$)]			HD [$\text{mean}\pm \text{SD}$ ($n$)]
	Heart	Amniotic fluid	Ratio	Heart	Amniotic fluid	Ratio	Heart	Amniotic fluid	Ratio
E13.5	$0.34\pm 0.03$ (6)	$0.15\pm 0.03$ (6)	2.26***	$0.29\pm 0.07$ (6)	$0.14\pm 0.03$ (6)	2.07*	$0.16\pm 0.05$ (6)	$0.14\pm 0.02$ (6)	1.17
E14.5	$0.41\pm 0.02$ (6)	$0.16\pm 0.01$ (6)	2.48***	$0.25\pm 0.02$ (6)	$0.12\pm 0.01$ (6)	2.10***	$0.09\pm 0.01$ (6)	$0.09\pm 0.02$ (6)	1.01
E15.5	$0.42\pm 0.01$ (6)	$0.19\pm 0.04$ (6)	2.21**	$0.23\pm 0.01$ (6)	$0.11\pm 0.02$ (6)	2.03***	$0.15\pm 0.02$ (6)	$0.16\pm 0.03$ (6)	0.96

Note: E, embryonic day; HD, high dosage; LD, low dosage; PSI, photoacoustic signal intensity; SD, standard deviation. $n=6$ mice per treatment group. *$p<0.05$; **$p<0.01$; ***$p<0.001$; all * relative to PSI of embryonic heart. (Independent sample $t$-test was used for analyzing the ratio of mouse embryonic heart to amniotic fluid).

Table 8.

The volume of the embryonic heart from E10.5 to E15.5 in different MMC treatments.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]	LD [$\text{mean}\pm \text{SD}$ ($n$)]	HD [$\text{mean}\pm \text{SD}$ ($n$)]
	Heart volume (${\mathrm{mm}}^3$)	Heart volume (${\mathrm{mm}}^3$)	Heart volume (${\mathrm{mm}}^3$)
E10.5	$0.261\pm 0.089$ (6)	$0.248\pm 0.055$ (6)	$0.223\pm 0.014$ (6)
E11.5	$0.424\pm 0.062$ (6)	$0.402\pm 0.092$ (6)	$0.285\pm 0.055$ (6)***
E12.5	$0.620\pm 0.054$ (6)	$0.596\pm 0.091$ (6)	$0.379\pm 0.016$ (6)***
E13.5	$0.783\pm 0.046$ (6)	$0.713\pm 0.066$ (6)	—
E14.5	$1.003\pm 0.054$ (6)	$0.813\pm 0.045$ (6)*	—
E15.5	$1.577\pm 0.045$ (6)	$0.983\pm 0.060$ (6)***	—

Note: —, no data; ANOVA, analysis of variance; E, embryonic day; HD, high dosage; LD, low dosage; LSD, least significant difference; MMC, methylmercury chloride; ${\text{SaO}}_2$, blood oxygen saturation; SD, standard deviation. $n=6$ embryos per treatment group. *$p<0.05$; ***$p<0.001$; all * relative to control (ANOVA with one-way ANOVA using LSD test).

Evaluation of Heart Histology in Mice Exposed to MMC

To validate the PA-US detection results, the embryos were then examined using the traditional histopathological method (Figure 8). H&E staining showed normal histology of the embryos in the control group from E13.5 to E15.5 (Figure 8A–8C). Various tissues and organs were significantly deformed in the HD group during these stages (Figure 8G–8I). Because heart development is one of the essential healthy diagnostic indicators for early embryo development, we zoomed in and carefully examined this area. In the HD group, the hearts were severely underdeveloped, and their size was significantly smaller than those of the control and the LD groups at E13.5–E15.5 (Figure 8P–8R). In the control group, the thickness of the ventricular wall of mouse embryos was $40.57\pm 1.99\mathrm{\mu }\mathrm{m}$, $47.67\pm 3.10\mathrm{\mu }\mathrm{m}$, and $54.18\pm 7.24\mathrm{\mu }\mathrm{m}$ from E13.5 to E15.5, respectively. However, in the LD group, the thickness of the ventricular wall of mouse embryos was $38.47\pm 2.79$, $34.68\pm 2.06$, and $21.91\pm 2.15\mathrm{\mu }\mathrm{m}$ from E13.5 to E15.5, respectively (Figure S10J; Table 9). The histopathological results also showed that in the LD group, the morphology of the heart was not significantly different from that of the control group at E13.5. However, the heart wall tissue was thinner than that of the control group at E14.5 (Figure S10J; Table 9). Moreover, myocardial interstitial congestion was observed in the LD group (Figure S10F′) compared to the control group (Figure S10C′) at E15.5. In addition, severe malformations, such as delayed brain development and dysplasia of limbs, were observed in the HD group (Figure S9; Table 10). These results suggested that the pathological changes of the heart started at E14.5 in the LD group.

Figure 8.

H&E staining of embryos after prenatal exposure to different dosages of MMC. Representative H&E staining images showing embryos from E13.5 to E15.5 in (A–C) the control group; (D–F) the LD group; and (G–I) the HD group. (B′ and E′) the same image as (B) and (E) with anatomic labels. Representative H&E staining images showing embryonic heart from E13.5 to E15.5 in (J–L) the control group; (M–O) the LD group; and (P–R) the HD group. The black line in (J–O) showing the outline of the heart. Note: E, embryonic day; H&E, hematoxylin and eosin; HD, high dosages; LD, low dosages; MMC, methylmercury chloride.

Table 9.

The thickness of embryonic heart wall of ventricles from E13.5 to E15.5 among the control group, LD group, and HD group.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]	LD [$\text{mean}\pm \text{SD}$ ($n$)]	HD [$\text{mean}\pm \text{SD}$ ($n$)]
	Heart wall ($\mathrm{\mu }\mathrm{m}$)	Heart wall ($\mathrm{\mu }\mathrm{m}$)	Heart wall ($\mathrm{\mu }\mathrm{m}$)
E13.5	$40.57\pm 1.99$ (3)	$38.47\pm 2.79$ (3)	—
E14.5	$47.67\pm 3.10$ (3)	$34.68\pm 2.06$ (3)*	—
E15.5	$54.18\pm 7.24$ (3)	$21.91\pm 2.15$ (3)**	—

Note: —, no data; E, embryonic day; HD, high dosage; LD, low dosage; SD, standard deviation. $n=3$ mice per treatment group. *$p<0.05$; **$p<0.01$; ***$p<0.001$; all * relative to control (independent sample $t$-test was used for analyzing heart wall).

Table 10.

The length of embryonic brain and the length of embryonic limbs among the control group, LD group, and HD group.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]		LD [$\text{mean}\pm \text{SD}$ ($n$)]		HD [$\text{mean}\pm \text{SD}$ ($n$)]
	brain	limb	brain	limb	brain	limb
E13.5	$2.37\pm 0.11$ (3)	$1.57\pm 0.40$ (3)	$2.33\pm 0.42$ (3)	$1.51\pm 0.18$ (3)	$1.82\pm 0.23$ (3)***	$0.88\pm 0.38$ (3)***
E14.5	$2.47\pm 0.58$ (3)	$1.70\pm 0.11$ (3)	$2.39\pm 0.58$ (3)	$1.51\pm 0.27$ (3)	$1.86\pm 0.44$ (3)***	$0.72\pm 0.24$ (3)***
E15.5	$3.39\pm 0.12$ (3)	$2.46\pm 0.10$ (3)	$3.23\pm 0.16$ (3)	$2.29\pm 0.36$ (3)	$1.76\pm 0.58$ (3)***	$0.44\pm 0.45$ (3)***

Note: ANOVA, analysis of variance; E, embryonic day; HD, high dosage; LD, low dosage; LSD, Least significant difference; SD, standard deviation. $n=3$ embryos per treatment group. ***$p<0.001$; all * relative to control (ANOVA with one-way ANOVA using LSD test). Error bars indicated $\text{mean}\pm \text{SD}$.

To investigate the effect of laser fluctuations due to animal positioning and depth, the embryonic heart were examined ex vivo (out of uterus) and in vivo (in uterus) scanning (Figure S2l Table 11). The experimental results showed that the ratio of PSI intensity out of uterus to that of in uterus was less than 2. Therefore, we considered the laser to have minor fluctuations and little influence on skin tissue on the imaging process.

Table 11.

PSI in the embryonic heart areas between the inside and outside of the uterus in the control group at E14.5.

Day	Control [$\text{mean}\pm \text{SD}$ ($n$)]	LD [$\text{mean}\pm \text{SD}$ ($n$)]
	PSI (a.u.) in utero	PSI (a.u.) ex utero
E14.5	$0.36\pm 0.31$ (6)	$0.42\pm 0.05$ (6)**

Note: E, embryonic day; LD, low dosage; PSI, photoacoustic signal intensity; SD, standard deviation. $n=6$ embryos per treatment group. **$p<0.01$ relative to in vivo group. (Independent sample $t$-test was used for analyzing the PSI in the embryonic heart areas between the inside and outside of the uterus).

Discussion

In this study, we attempted to explore the application of PA-US in the early prenatal examination. We first established a mouse model of MMC poisoning during early organogenesis and observed that exposure to the HD of MMC resulted in a lower number of embryos per dam in comparison with control animals, beginning at E10.5, whereas in the LD group the number of embryos were similar to control until E15.5, when these animals had significantly fewer embryos. Subsequently, we applied the dual-modality PA-US imaging technique to determine whether it could further detect the developmental abnormalities induced by the LD of MMC. As is well known, the photoacoustic signal of hemoglobin in blood is mainly produced by deoxyhemoglobin and oxygenated hemoglobin (Laufer et al. 2005; Zhang et al. 2007). The ratio of deoxyhemoglobin and oxygenated hemoglobin can be calculated by analyzing the photoacoustic signal. Kiguna et al. used photoacoustic technology to accurately measure the blood oxygen saturation of the rabbit hypoxemia model (Kiguna et al. 2019). Furthermore, we monitored the ultrasound signals, PSI, ${\text{SaO}}_2$, and HbT from E10.5 to E15.5 and found that in the LD group, PSI, ${\text{SaO}}_2$, and HbT signals near the embryonic heart were significantly lower than those in the control group at E14.5. Furthermore, we used H&E staining to demonstrate that the heart wall tissue was thinner than that in the control group at E14.5. Myocardial interstitial congestion was observed in the LD group; this phenomenon was possibly caused by cardiac function abnormity. This finding further suggests that early cardiac diastolic and systolic dysfunction may occur in the embryonic heart at day E15.5 in the LD group. The timing of the differences between the MMC- and control-exposed embryos as determined by the quantitative analysis of PA-US images matched well with the timing of the observed histological results.

We chose to give the MMC from E5.5 because this is an implantation time point, which is equivalent to the second week of human embryo development (Norwitz et al. 2001; Plaks et al. 2006). E10.5 is the time point for the closure of posterior neuropore in mouse embryos, equivalent to the fourth week of human embryo development (Borchin and Barberi 2015; Lay et al. 1999). E5.5–E10.5 is a period from postimplantation to the early organogenesis stage in embryo development (Minkin 2009). In humans, this period is the riskiest period for miscarriage and developmental abnormality. For most women, the end of the month after conception was 6–7 wk since the last period. This means they had missed their period for 2–3 wk. This time frame was the time most people found out about their pregnancy and started to pay attention to their daily lives. Thus this is the reason we chose to give MMC to the mice from E5.5 to E10.5 and started to observe the outcomes afterward. Su and Okita had also tried to induce mouse embryo malformation in the postimplantation stage (Su and Okita 1976). They found that when mice were given a dosage of $8\mathrm{mg}/\mathrm{kg}$ or higher concentration of MMC in the 7 to 12 d of gestation, the BW of the pregnant mice was significantly reduced, and the abortion rate of the embryos was significantly increased. Curle et al. also reported a significant increase in the incidence of embryo abnormalities in pregnant mice after treatment with 10, 15, or $20\mathrm{mg}/\mathrm{kg}$ MMC at E9.5 (Curle et al. 1987). In addition, Fuyuta et al. reported that giving the mice $7.5\mathrm{mg}/\mathrm{kg}$ MMC by gavage from E6 to E13 resulted in a significant decrease in the weight of pregnant mice and a significant increase in abortion rate at the end of pregnancy (Fuyuta et al. 1978). These results were consistent with our results of the HD and LD groups.

The studies in both primates (Mottet et al. 1985; National Research Council Committee on the Toxicological Effects of Methylmercury 2000) and rodents (Cagiano 1990) had shown that MMC could quickly cross the placental barrier and produce significant embryonic toxicity. Our results also demonstrated that MMC could enter the embryo through the placental barrier and resulted in embryonic abnormalities even after the withdrawal of MMC from the pregnant mice. Two possible explanations for MMC-induced embryonic malformations have been put forth. The first explanation was that MMC had a significant effect on the synthesis of DNA and RNA (Persson et al. 1983), indicating that MMC had a high interference with cell proliferation and could cause defects in embryo development. The second was that MMC might lead to abnormalities in the extra-embryonic vasculature, affecting the transport of substances between the maternal and embryo tissues (Gheorghescu and Thompson 2016).

This study showed that using ultrasound alone was not enough to detect apparent morphological abnormalities in the LD group other than the decrease in the heart rate and the number of embryos at E15.5. However, with the PA-US system, we showed that the PSI, ${\text{SaO}}_2$, and HbT signals in the mouse embryonic heart were lower than those of the control group at E14.5. Yamaleyeva et al. demonstrated that early embryonic dysplasia and tissue hypoxia could be detected by monitoring placental ${\text{SaO}}_2$ and HbT signals in mice (Yamaleyeva et al. 2017). Van den Berg et al. also showed that human embryos with abnormal heart rates had reduced ${\text{SaO}}_2$ and significant acidosis compared to the healthy embryos (Van den Berg et al. 1994). Moreover, in rodents, ${\text{SaO}}_2$ and HbT signals were closely related to the occurrence of many diseases, such as tumors (Chen et al. 2017), cardiovascular and cerebrovascular diseases (Sun et al. 2015), and reproductive and developmental diseases (Arthuis et al. 2017). Our preliminary results suggested that the photoacoustic signal intensity has certain differences in the control, LD, and HD groups at same development stage. This difference could be due to the heart abnormality caused by a decline of hemoglobin after LD and HD treatment, but this possibility would require further investigation. The PSI will be dependent on the positioning of the fetus/animal. Although we took care to be consistent with the positioning, it still can and may contribute to variability in signaling. A ratiometric measurement method (like ${\text{SaO}}_2$) would provide a more precisely visualize and quantify the relevance between these single-wavelength photoacoustic signal intensity results (Horvath et al. 2008; Zhang et al. 2020, 2017). Nonetheless, PSI as well as ${\text{SaO}}_2$ need to be analyzed more carefully to be used as fetal toxicity biomarkers. This analysis should be performed in further work.

A significant association has been found between heart disease and MeHg exposure in adults (Inoue et al. 2012; Mozaffarian et al. 2011; Salonen et al. 1995; Yoshizawa et al. 2002). Our histopathological examination also showed that even the LD of MMC could induce the thinning of the ventricular wall during development, which leads us to speculate that it is likely that MMC causes impaired cardiac function in the embryos. In addition, some studies also reported that administration by gavage with MMC to pregnant mice led to a cleft palate, edema, and brain malformations in the newborn mice (Fuyuta et al. 1978). We also observed severe malformations such as delayed brain development and dysplasia of limbs in the results of the HD group.

In past decades, some other imaging techniques such as magnetic resonance imaging (MRI) (Egloff and Bulas 2015), optical coherence tomography (OCT; Huang et al. 2019), computed tomography (CT; Wong et al. 2012), and multiphoton microscopy (MPM; Supatto et al. 2011) have been used to visualize mouse embryo development. However, all these methods have their own disadvantages when applied to clinics. For example, MRI took minutes to hours to collect full 3D volumes of whole embryos (Popescu et al. 2011); due to the limitation in its penetration, OCT was not suitable for human in utero imaging (Huang et al. 2019); micro-CT imaging required a complex microfocus adjustment device and a sophisticated image reconstruction algorithm (Boppart et al. 1996); MPM could not be used to image living mouse embryos (Lucitti et al. 2007). The popularity of photoacoustic imaging is increasing. Moreover, it has even been used to diagnose human diseases (Lin et al. 2018). Such work shows the detailed angiographic structure of the human breast using photoacoustic computed tomography (PACT) scanning technology, and this imaging modality could also show tumors in the breast. Furthermore, PACT modality could not only be applied to diagnose human soft tissue diseases but also to detect the functional activity in the human brain (Na et al. 2021). In this study, we showed that dual-modality PA-US imaging modality was a sensitive, noninvasive and high-penetration quantitative imaging method, which enabled it to accurately detect the abnormal development of mouse embryos in utero. The imaging depth in our PA-US system was not enough to apply to the human body (Hamelmann et al. 2020). However, the imaging depth of a PA-US system could be increased with the application of a photoacoustic contrast agent (Li and Chen 2015) and the optimization of detector bandwidth (Allen and Beard 2012) in the future. This technique will allow the application of PA-US imaging in deep human pathological tissue. Overall, our work brings a new approach to studying developmental toxicity effects and will promote the pursuit of studies where 3D photoacoustic functional imaging with noninvasive measurements of the optical absorption in tissue is desirable. For biomedical applications, the high acoustic resolution of ultrasound imaging can provide fine morphological details. Photoacoustic imaging has better contrast and can acquire functional information about hemoglobin and oxygen saturation. Indeed, the optical absorption of the ${\text{SaO}}_2$ and HbT changes with embryo development, which could be an early indicator for developmental abnormality assessment. For combined application, an ultrasound transducer was used to collect the generated wideband acoustic signals. In addition, photoacoustic imaging used the detected acoustic signals to reconstruct the optical absorption distribution. Because ultrasound and photoacoustic modalities shared the same detector, they were paired with each other. It became a potent dual-modality approach that provided complementary information for developmental toxicity application.

Certainly, the preliminary study has a few limitations. First, our dual-modality PA-US imaging system does not have a higher resolution. In addition, due to the embryo development is dynamic growth process, a regular frequency detector may result in impacts on the photoacoustic signal measure, especially on the HD group. The measurement accuracy of the PSI and ${\text{SaO}}_2$ in the HD group should be investigated in a subsequent study based on larger sample size, more frequency of testing, and several MMC derivatives toxicity models. Therefore, further studies on the embryo matrix toxicity of MMC, the number of apoptoses of embryos in an HD group with flow cytometry, and frequency adjustable dual-modality detectors are underway to further elucidate and optimize this new approach. Overall, our study explored a novel method for investigating the mammalian embryo abnormalities after prenatal exposure to MMC. This dual-modality imaging technique is extremely important to evaluate the risk of environmental pollution on prenatal development in vivo, especially on exposure to air pollution during pregnancy. With further development of imaging techniques, this method also could be used for the assessing the environmental pollutant influences on prenatal nervous system development.

Conclusion

Our results showed that the dual-modality PA-US imaging system could detect early developmental anomalies accurately as the histopathological analysis in a mouse model of MMC toxicity. Therefore, we believed that dual-modality PA-US would have a great potential to be applied in developmental biology research as well as clinical embryology in the future.