Unique in utero identification of fetuses in multi-fetal mouse pregnancies by placental bi-directional arterial spin labeling (BD-ASL) MRI

Abstract

Non-invasive imaging is a critical part of the study of developing embryos/fetuses, particularly in the context of alterations of gene expression in genetically modified animals. However, in litter-bearing animals such as mice, the inability to accurately identify individual embryo/fetus in utero is a major obstacle to longitudinal, non-invasive in vivo studies. Arterial Spin Labeling MRI (ASL-MRI) was adopted here to determine the fetal order along the uterine horns in vivo, based upon the specific pattern of dual arterial blood supply within the mouse uterine horns. Blood enters the mouse uterus cranially through the ovarian artery, and caudally through the uterine artery. Saturation slices were alternately placed on the maternal heart or on the bifurcation point of the common iliac artery, thereby saturating either downward inflow via the ovarian arteries, or upward inflow via the uterine arteries, respectively. Saturation maps provided a unique signature with highly significant correlation between the direction-dependent magnetization transfer and the position of the fetuses/placentas along the uterine horns. The Bi-Directional ASL-MRI (BD-ASL) method reported here opens possibilities to determine and pursue phenotypic alterations in fetuses and placentas in longitudinal studies of transgenic and knockout mice models, and for studying defects in placental vascular architecture.

INTRODUCTION

Over the past few decades, the use of transgenic and knockout mice has extensively increased our understanding of various pathologies. Since a substantial part of the genome participates in embryonic and fetal development, the study of offspring development during pregnancy is an important component of many studies of genetically modified mice (1,2). Gene function is analyzed when it acts alone or in synergy with other genes, and the subsequent phenotype may often be investigated by anatomical, biochemistry, molecular, and physiological studies (3). In addition, due to the homology to the human placenta, mice are commonly used for studying mechanisms underlying pregnancy complications and placental development and function. In particular, both species have a hemochorial placenta (4-6). Moreover, mice offer an important model for the development of novel imaging tools that may help advance clinical imaging of pregnancy.

Most developmental studies of genetically modified mice focused on the fetus itself and ignored the complex interaction between the mother and the fetus. In mice, as in other litter-bearing species, the position of the fetus along the (crowded) uterine horn might add an additional layer of complexity, due to possible effects resulting from the different environments to which the offspring are exposed throughout pregnancy. An additional complication arises from the anatomy of mouse pregnancy (Fig 1a). The uterus of the mouse is duplex, composed of two independent uterine horns that converge at the cervix (7). The origin of the blood vessels feeding the cranial portion of each horn is the uterine branch of the ovarian artery, whereas blood enters at the caudal end from the distal part of the uterine artery (8). The fact that there are two paths for arterial blood arising from the ovarian and internal-iliac arteries, suggests that each fetus might be nourished from one direction alone or from two directions, depending on its location along the uterine horn. Indeed, previous studies claimed that arterial blood flow to the placenta of rodents is ‘bi-directional’, suggesting possible arterial shunts for the two blood supplies (7-9). Although mice provide powerful models for studying the etiology of pregnancy disorders and embryonic development, our understanding is limited by the ability to examine these processes in longitudinal studies in vivo. One of the major obstacles for in vivo imaging of mouse pregnancy is the difficulty of accurate identification of specific embryos/fetuses in utero in longitudinal studies (10).

Figure 1.

Anatomy of mouse pregnancy: (a) Schematic diagram of mouse uterine horns and their blood vessels in the gestation period after placentas have formed (E10.5 to term). Cranial blood supply to the left uterine horn originates from a branch of the renal artery, whereas blood supply to the right uterine horn originates directly from the descending aorta. The caudal uterine artery branches from the common iliac artery on both the right and left sides. F Fetus, K Kidney, Ov Ovary, P Placenta; (b) In vivo coronal and (c) axial T2-weighted MRI images of mouse pregnancy on E17.5. Note fetuses (F) and their placentas (white arrows); (d) An enlarged T2-weighted MRI image of a fetus and its placenta; note the fetal anatomical structures: umbilical cord (black arrow heads), spinal cord (white arrow heads), brain (B), and liver (L).

The aim of this study was to introduce a new approach that would enable non-invasive determination of the fetal position along the uterine horn. To achieve this aim, we made use of bi-directional arterial spin labeling (BD-ASL) MRI for determination of the distance and the contribution of blood flow from the uterine versus the ovarian arteries for each placenta. ASL (11,12) either by inversion or by saturation of water in the arterial blood, has been applied previously for the study of perfusion in a number of organs, particularly in the brain (13-15) and the kidney (16-18). ASL based on steady-state, slice-selective pulsed saturation similar to that applied here, was reported previously by us, for measurement of perfusion in the rat ovary (19,20).

MATERIALS AND METHODS

Animal Models

All experiments were approved by the Institutional Animal Care and Use Committee of the Weizmann Institute of Science. An overview of the study design is illustrated in Fig 2.

Figure 2.

Experimental scheme of ASL MRI in pregnant mice. Pregnant mice at late gestation (E17.5) are first anesthetized using isoflurane, then positioned in a supine position in the MRI. First, fast-spin-echo scans are performed for anatomical identification of all fetuses and placentas (5-10 min), followed by an angiography sequence to locate the heart and the common iliac artery (5 min). After positioning the saturation slabs, the ASL set is repeated, as necessary, to scan all fetuses (1-1.5 hours). At the end of the MRI session, some of the animals are euthanized, and their uterine horns are exposed for counting and numbering of fetuses. Others are taken for subsequent analysis in the stero-microscope, using i.v. injection of FITC-Dextran, and only then euthanized.

Pregnant female ICR mice were analyzed on embryonic day (E) 13.5 (n=4 mothers, 53 placentas/fetuses), or E17.5 (n=10 mothers, 86 placentas/fetuses), or on both E13.5 and E17.5 (n=4 mothers, 37 placentas/fetuses). Timed matings were carried out by pairing strain-matched males and females for one night, with the following morning designated as E0.5 if a vaginal plug was detected as indication that mating had occurred. To further examine the applicability of our ASL methodology, we also employed two well-characterized models for intrauterine fetus crowding: superovulation (21,22) and hemi-overiactomy (23,24).

Superovulation model

Pre-pubertal 24-day-old female ICR mice were injected subcutaneously with 5IU of pregnant mare’s serum gonadotropin (PMSG, National Hormone & Peptide Program, Harbor-UCLA Medical Center, Torrance, CA, USA), resulting in stimulation of multiple follicles to develop to the antral stage. Forty-eight hours later, 5IU of human chorionic gonadotropin (hCG, Sigma-Aldrich, Rehovot, Israel) was injected intraperitonealy to induce ovulation, and males were introduced for overnight mating. The next morning, females were examined to ensure the presence of a vaginal plug (E0.5). Pregnancy was evaluated on E17.5 (n= 5 mothers, 55 placentas/fetuses).

Hemiovariectomy model

Pregnancy following hemiovariectomy was described previously as a technique to create crowded uterine horn pregnancy. This model is effective, as it results in higher number of ovulations in the remaining ovary, while the space in the relevant uterine horn available for maintained embryos/fetuses is unchanged (21). For this model, two-month-old female ICR mice were hemiovariectomized surgically as described elsewhere (23-25). Following at least 10 days recovery, males were introduced for overnight mating. The next morning, females were examined to ensure the presence of a vaginal plug. Pregnancy was evaluated on E17.5 (n= 5 mothers, 51 placentas/fetuses).

In vivo MRI Studies

MRI experiments were performed at 9.4T on a BioSpec 94/20 USR spectrometer (Bruker; Karlsruhe, Germany) equipped with a gradient-coil system capable of producing pulsed gradients of up to 40 gauss/cm in each of three orthogonal directions. A quadrature volume coil, with 72 mm inner diameter and a homogeneous RF field of 100 mm along the axis of the magnetic field, was used for both rf transmit and receive. During the MRI scanning, mice were anesthetized with isoflurane (3% for induction, 1–2% for maintenance; (Abbott Labratories Ltd, England) mixed with oxygen (1 liter/min), delivered through a nasal mask. Once anesthetized, the animals were placed in a supine position in a head holder to assure reproducible positioning inside the magnet. To separate the caudal fetuses from the bifurcation point of the common iliac artery, these fetuses were gently pushed cranially, and the mother’s position was fixed using tape. Respiration rate was monitored using an SA Instruments Model 1025 monitoring and gating system (Stony Brook, NY, USA) and maintained throughout the experimental period at 50–70 breaths per min by adjusting isoflurane levels. Body temperature was maintained at approximately 37°C using a circulating water system.

MRI Data Acquisition

Fast spin-echo MRI was used to allow visualization of all fetoplacental units. Coronal T₂-weighted fast spin echo images were acquired using the following parameters: TR = 3000 ms; effective TE = 85.5 ms; slice thickness 1.0 mm; intersection gap 0.2 mm; FOV = 7 × 5 cm²; matrix 256 × 128, zero filled to 256 × 256 (26); RARE factor = 8. Axial T₂-weighted fast spin echo images were acquired using the following parameters: TR = 4000 ms; effective TE = 85.5 ms; slice thickness 1.0 mm; intersection gap 0.2 mm; FOV = 5 × 5 cm²; matrix 256 × 128, zero filled to 256 × 256 (26); RARE factor = 8. Fifteen to forty slices were acquired to cover all the fetoplacental units.

The longitudinal relaxation rate (R₁ = 1/T₁) of arterial blood was measured using a series of variable-flip-angle 3D coronal gradient-echo (3D-GE) images with the following parameters: pulse flip angle = 5°, 15°, 30°, 50°, 70°; TR = 10 ms; TE = 3.1 ms; two averages; FOV = 6 × 5 × 4 cm³; matrix = 256 × 128 × 64, zero filled to 256 × 256 × 64 (26); 2 averages.

R₁ maps were derived from the variable flip-angle data by a nonlinear fit to the following equation:

$$I\left(\alpha \right)=\frac{M_0\sin \alpha (1-e^{-\mathit{TR}-R_{1\mathit{pre}}})}{1-\cos \alpha \cdot e^{-\mathit{TR}-R_{1\mathit{pre}}}}$$ (1)

where I is the signal intensity as a function of the pulse flip angle α, and the pre-exponent term, M₀, includes contributions from both the spin density and the T₂ relaxation.

Angiography

The location of the major blood vessels was determined by coronal angiography using the following parameters: TR = 12.5 ms; TE = 3.9 ms; slice thickness 0.3 mm; intersection gap 0.1 mm; 60 slices; FOV = 7 × 7 cm²; flip angle = 60 degrees; matrix = 512 × 256, zero filled to 512 × 512 (26); 2 dummy scans; 2 averages and a scan time = 5 min 39 sec. To verify that the experiments are conducted within the steady state saturation approximation the time in the descending aorta was measured using an additional angiography sequence with a presaturation pulse on the heart using the same parameters as for the ASL saturation scheme (vide infra).

The decay of the saturation tag along the arteries as a function of the distance from the heart was measured for different regions of interest along the aorta. The transit time, τ, (time it takes to reach each point from the heart) was assigned to each ROI along the descending aorta according to the following equation:

$$\frac{{\mathit{SI}}_{\text{control}}-{\mathit{SI}}_{\text{sat}}}{\max \left({\mathit{SI}}_{\text{control}}\right)}=\frac{\Delta \mathit{SI}}{\mathit{SI}}=e^{-\tau ∕T_1}$$ (2)

Here, SI_control and SI_sat are the magnetization in the perfused organ with and without saturation pulse, respectively.

Arterial Spin Labeling (ASL)

ASL data were collected using a multi-slice axial 2D-GE sequence with the following parameters: TR = 53 msec; TE = 3.2 msec; slice thickness = 1.0 mm; intersection gap 0.2 mm; 4 slices per scan; FOV = 5 × 5 cm²; matrix = 256 × 128, zero filled to 256 × 256 (26); 10 dummy scans; 32 averages and a scan time with respiratory gating of approximately 6 min (the original scan time without gating was 3 min 39 sec). Two sequential images were collected for each set of fetuses/placentas located at the same axial plane: one with cardiac tagging, leading to saturation of water in the uterine branch of the ovarian artery, and a second with tagging of arterial water in the uterine artery (Fig 3e). The saturation pulses were located according to the 3D maximum intensity projection (MIP) of the angiography sequence (Fig. 3a). This image pair was repeated (usually 6-8 per mouse) as necessary in order to scan all fetuses and their placentas. Presaturation of the MRI signal of blood flowing into the uterine arteries was performed using a steady-state saturation scheme, as described by Tempel, et al (20). A thick axial slice (8 mm), covering either the entire heart (denoted as upper) or the bifurcation point of the common iliac artery (denoted as lower), was saturated by a 90° hermit pulse (2 msec) followed by a crusher gradient. Rapid repetition of this saturation for each phase-encoding step in a four-slice imaging protocol resulted in effective steady-state saturation of the inflow, while a low flip angle excitation pulse (15 degrees) ensured minimal saturation of the observed axial slices. This experimental scheme resulted in steady-state saturation of the arterial blood in which the magnetization did not recover between phase-encoding steps.

Figure 3.

In vivo MRI measurements of arterial transit time in the descending aorta of a pregnant ICR mouse (E17.5). (a) 3D Maximal Intensity Projection (MIP) angiography of all major blood vessels; (b) 3D MIP with an 8 mm width presaturation slab positioned on the heart; (c) Subtraction of a-b reveals the arterial blood vessels. Regions of interest (ROI) along the carotid arteries are marked by white arrows; (d) Plot of the distance from the heart as a function of calculated transit time, derived using the T₁ relaxation time of blood; (e) MIP angiography (red) overlaid on a coronal anatomical image. The white lines represent the location of the 8 mm saturation slabs and the orange lines define the imaging plane, which was moved systematically across the animal to cover all fetuses/placentas. Two arterial presaturation images were obtained in separate scans: one by placing the saturation slab on the heart (top white lines) and the other by placing the saturation slab on the bifurcation point of the common iliac artery (bottom white lines).

ASL data was analyzed using Matlab (Mathworks; Natick, MA, USA). Bi-Directional Arterial Saturation Labeling values (BD-ASL) were determined using the normalized percentile difference between the two acquired images, which was calculated according to the following equation:

$$\mathit{BD}-\mathit{ASL}=(\frac{{\mathit{SI}}_{\text{lower}}}{\max \left({\mathit{SI}}_{\text{lower}}\right)}-\frac{{\mathit{SI}}_{\text{upper}}}{\max \left({\mathit{SI}}_{\text{upper}}\right)})\ast 100$$ (3)

where SI_upper and SI_lower are the magnetization in the perfused organ with upper and lower steady-state saturation pulses, respectively. This difference produces saturation transfer maps, which were color coded to generate maps of flowing blood that depend on directionality with a threshold value of ±1: red represents pixels showing positive change (BD-ASL > 1) and blue represents negative pixels (BD-ASL < −1). In order to correlate the fetus/placenta position along the uterine horn and the BD-ASL signal, regions of interest (ROI) were drawn around the entire placenta, and histogram plots of the saturation transfer maps were then generated for each ROI. The peak of the histogram was plotted for each placenta as a function of its position along the uterine horn, numbered starting from 1 onwards, where 1 is the placenta closest to the cervix.

Intravital Microscopy of the Uterine Blood Supply

The uterine blood supply was imaged using a Zoom Stereo Microscope SZX-RFL-2 (Olympus, Tokyo, Japan) equipped with a fluorescence illuminator and a Pixelfly QE charge-coupled device (CCD) camera (PCO, Kelheim Germany). The excitation and emission for the FITC filter set were 460 - 490 nm (excitation) and 510 - 550 nm (emission). Images were acquired using the Camware camera-controlling software (PCO). Image analysis was performed using ImageJ (http://rsbweb.nih.gov/ij/). Mice were anesthetized with intraperitoneal injection of Ketamin (Ketaset®, Fort Dodge Laboratories, Fort Dodge, IA, USA) and Xylazine (XYL-M, VMD, Arendonk, Belgium). The uterus was surgically exposed through a ventral midline celiotomy and the two uterine horns were spread out so their intact blood vessels (uterine branch of the ovarian artery and the uterine artery) could be visualized. For visualization of the blood supply pattern, high–molecular weight dextran–fluorescein isothiocyanate (6 mg diluted in 300μl PBS per animal; FITC, 500 kDa; Sigma Aldrich) was injected via a tail-vein catheter. During the injection, a movie was recorded (1200 frames/min for 0.5 min with fluorescence exposure time of 50 ms).

Validation of Fetal Location

At the end of the scanning series, the animals were euthanized by anesthesia overdose, and a midline ventral laparotomy was performed to expose the abdominal cavity. The positions of the uterus and all fetuses and their placentas were recorded by digital camera. This information was used as validation for the number and location of fetuses/placentas along the uterus horn, as derived from the MRI data.

Statistical Analysis

Statistical analyses were performed with analytic computerized software (STATISTIX 8 Student Edition, Analytical Software, Tallahassee, FL, USA). Pearson’s Correlation test was used to determine the correlation between ASL signal and the relative location of the placentas along each uterine horn determined after the animal was euthanized. Results are presented as mean ± SEM; or, in some figures, Box-and-whisker plots are presented with the 25% and 75% percentile ranges (box depth) and the maximum and minimum (T-bars). P-values smaller than 0.05 were considered significant.

RESULTS

Fetoplacental MRI

Figure 1a represents a schematic diagram depicting the mouse uterine horns and their blood vessels in the gestation period following placental formation (E10.5 to term). The cranial blood supply of the left uterine horn originates from a branch of the renal artery, whereas the blood supply to the right uterine horn originates directly from the descending aorta. The caudal uterine artery branches from the common iliac artery on both the right and left sides. Using a fast spin-echo MRI sequence, we were able to acquire high resolution MR images of mouse fetuses and placentas on E17.5. Serial coronal (Fig. 1b) and axial (Fig. 1c) T₂-weighted images enable identification of all placentas and fetuses, as well as detection of various fetal anatomical structures including the umbilical cord, spinal cord, brain, and liver (Fig. 1d).

In vivo MRI Measurements of Arterial Transit Time

Measurement of perfusion by arterial spin labeling is limited by the time window defined by the T₁ relaxation of blood. To verify that the magnetization tag of blood exiting the heart is not lost before it reaches the placenta, we measured the transit time in the descending aorta using saturation recovery based 3D angiography MRI (Fig. 3). The transit times were derived from the difference between the magnetization of blood with and without a saturation slab located on the heart (Fig. 3a and 3b), taking into account the T₁ relaxation of arterial blood, which was measured independently to be 2.1 ± 0.2 sec (mean ± SEM, n = 4; in good agreement with previous measurements of T₁ in the mouse arterial blood (27)). Different points along the descending aorta and their derived transit times are depicted in Fig. 3c and 3d, respectively. All calculated transit times were small relative to T₁ of blood, thus verifying that the blood is at least partially saturated under the steady-state approximation in all placentas.

Non-Invasive Identification of Fetal Order in Pregnant Mice by Bi-Directional Arterial Supply (BDASL) ASL MRI of the Placentas

ASL methodology was employed here to explore the effect of fetal position along the uterine horn in pregnant mice at late gestation on the pattern of transfer of saturated blood. Two sequential 2D GE images were collected for each set of fetuses located at the same axial plane, with the pre-saturation slab alternating between the maternal heart and the maternal common iliac artery (Fig. 3e). The resulting images were used to produce directional Bi-Directional ASL (BD-ASL) contrast maps, in which a change in signal intensities revealed the directional contribution of water exchange between the maternal blood and the fetus/placenta (Fig 4).

Figure 4.

Placental BD-ASL MRI of ICR pregnant mice (E17.5). (a-c) Saturation transfer maps of a pregnant mouse at E17.5. BD-ASL values were calculated to produce color-coded saturation transfer maps. Red represents pixels showing positive change in the saturation transfer maps (BD-ASL>1), and blue represents negative change (BD-ASL<−1). Note the fetal positions marked on the maps. Placentas near the cervix (a) had negative BD-ASL values, whereas those near the ovary placentas had positive values (c). Placentas located in the middle of the left uterine horn, namely L3 and L4, (out of 8 fetuses in the left uterine horn) had a dispersive pattern of BD-ASL values with both negative and positive values. By contrast, the placentas in the upper most part of the uterine horn, namely R4 and R5, displays mostly positive values. Note the high positive BD-ASL values in the cortex of the kidneys (K) in panel c; (d) Box plots of placental BD-ASL as a function of placental position along the left and right uterine horns; (e) Peak placental BD-ASL as a function of placenta position along the left and right uterine horns acquired from a live pregnant mouse and post euthanasia (with isoflurane overdosing). Note the low BD-ASL values for all placentas without any position dependency in the dead mouse. (f-g) BD-ASL saturation transfer maps for six different fetuses, collected alive (f) and then after euthanasia (g).

Beyond the placentas, significant saturation transfer was also observed in the fetuses, consistent with efficient exchange of water between the maternal and fetal circulation. For placentas positioned closer to the cervix, a negative BD-ASL contrast was observed (Fig. 4a), consistent with the predominant contribution of maternal blood flow through the uterine artery. In placentas closer to the ovary, the BD-ASL contrast was mainly positive (Fig. 4c), implying that in the cranial portion of the uterine horn the major blood supply is through the uterine branch of the ovarian artery. Interestingly, for some of the placentas located in the central region of particularly crowded horns, the BD-ASL contrast inside the placenta included both negative and positive voxels (Fig. 4b), consistent with a dual supply from both the uterine artery and the uterine branch of the ovarian artery. High positive BD-ASL signals were recorded in the cortex of the maternal kidneys (Fig. 4c), consistent with previous BD-ASL experiments measuring perfusion in kidneys.

To correlate fetal position along the uterine horn and BD-ASL signal, we generated histogram plots of the saturation transfer maps for each placenta (Fig. 4 and 5). Regions of interest (ROI) were selected as the entire placenta, and the peak value of the BD-ASL histogram was plotted for each placenta as a function of its position along the uterine horn. In all cases, placentas near the ovary had positive BD-ASL values and placentas near the cervix had negative values, while placentas located in the middle of the uterine horn had a more disperse pattern of BD-ASL values (Fig. 4b). Control images acquired from a dead pregnant mouse (anesthesia overdose; Fig. 4e) showed very low background BD-ASL values (−0.92 ± 0.38, n = 8 placentas) for all placentas without any position dependence. Figure 4 shows BD-ASL parametric maps derived from six live fetuses (Fig. 4f) and the same fetuses after euthanasia of the dam by anesthesia overdose (Fig. 4g). The BD-ASL parametric revealed the significant attenuation of maternal perfusion of the placentas upon maternal death, which is clearly detectable for all fetuses, including the central ones in which the average BD-ASL is close to zero.

Figure 5.

Placental ASL MRI data from pregnant mice (E17.5) in: (a) ICR control mouse; (b) ICR mouse after hemiovariectomy; and (c) ICR mouse following superovulation treatment. Box plots of placental BD-ASL values as a function of placental position along the uterine horns (left panels), and the corresponding histogram plots of the placental saturation transfer maps (right panels) are shown. Overall, BD-ASL values were negative for placentas near the cervix and increased gradually, with the placentas closest to the ovary having the largest positive values.

To further examine the applicability of our BD-ASL methodology, we also employed two well characterized models for intrauterine fetal crowding; hemiovariectomy and superovulation. Figure 5 shows representative examples of BD-ASL values as a function of fetus/placenta location along the uterine horn. In all models, the peak BD-ASL values showed significant positive correlation (Fig. 6) to the placental position along the uterine horn (WT ICR model: r=0.86, p = 0.0001; Hemiovariectomy model: r= 0.85, p = 0.0001; Superovulation model: r = 0.85, p = 0.0001). Overall, peak BD-ASL values were negative for placentas near the cervix and increased gradually, with the placentas closest to the ovary having the largest positive values.

Figure 6.

Correlation between peak BD-ASL values and the relative location of fetuses along the uterine horn in (a) ICR control mice (r = 0.86, p = 0.0001); and (b) ICR mice after hemiovariectomy (r= 0.85, p = 0.0001). Peak BD-ASL values showed significant positive correlation to the position of placentas along the uterine horn. (c) Examples of placental BD-ASL values (color coded) along the uterine horns (relative location; 0= closest to the cervix; 1= closest to the ovary; black = no data) of ICR pregnant mice (E17.5) carrying different number of fetuses per horn. Note that BD-ASL values may range between −10 to +10. There was no association between the range of these values and the number of fetuses per horn (p = 0.5790).

Dual Arterial Supply to the Placentas Located near the middle of the Uterine Horn

In all models, BD-ASL values were between −10 and 10. There was no correlation between the range of these values and the number of fetuses per horn (p = 0.5790; for specific examples see Fig. 6c). Also, there were no differences in BD-ASL values between the left and the right uterine horn (p = 0.9566). The spatial distribution of arterially labeled water is presented in histograms of regions of interest corresponding to the entire placenta. Unlike most placentas located near the cervix or near the ovary, whose distributions of BD-ASL values are exclusively negative and positive, respectively, most of the central placentas exhibited a broad and bimodal distribution with both positive and negative voxels, consistent with dual perfusion. To verify that all placentas are indeed well perfused and that blood arrival time to the placenta was short relative to the T₁ of blood, we employed intravital fluorescence imaging, as described below.

Longitudinal Studies of Mouse Pregnancy

In order to determine the usefulness of this BD-ASL method in longitudinal studies of mouse pregnancy, we first confirmed that the method is valid in earler days of gestation. Pregnant female ICR mice were scanned on E13.5 (n=4 mothers, 53 placentas/fetuses), and the location of their placentas/fetuses was confirmed following euthanasia, as described. BD-ASL method on E13.5 showed similar relationships between BD-ASL value and the location of the placenta/fetuses, as described for E17.5 (Fig. 7a).

Figure 7.

Longitudinal studies of mouse pregnancy. (a) Box plots of placental BD-ASL as a function of placental position along the left and right uterine horns in an ICR pregnant mouse on E13.5. The location of each fetus/placenta was determined by laparotomy following euthanasia. (b) Box plots of placental BD-ASL as a function of placental position along the left and right uterine horns in an ICR pregnant mouse on E13.5 and again on E17.5. Note that in this followup experiment the position of each fetus/placenta on E13.5 was deduced from the BD-ASL values on E13.5, while it was independently determined by laparotomy on E17.5. Note the larger spread of BD-ASL values on E17.5 as compared with E13.5.

Subsequently, pregnant female ICR mice (n=4 mothers, 37 placentas/fetuses) were scanned twice, at two different gestation days, E13.5 and E17.5. Figure 7b shows an example of such a longitudinal study. In these experiments, while the positional order of the fetuses at termination of the experiment on E17.5 could be validated independently, the positional order at E13.5 was derived from BD-ASL. Interestingly, the spread of BD-ASL values at the later gestation day (E17.5) was larger, consistent with increased perfusion values.

Intravital Fluorescence Microscopy of the Arterial Blood Supply to Placentas Along the Uterine Horn

To measure the transit time of arterial blood to the different placentas along the uterine horns in pregnant mice (E17.5), a fluorescent dye (FITC-dextran) was injected intravenously into the tail vein while its path along the uterine blood vessels was recorded using a fluorescent stereomicroscope (Figure 8). The entry of FITC-dextran to the two arteries of the uterine horn occurred at approximately the same time (~500 msec post injection). No significant differences were detected in the FITC-dextran flow-pattern between the left and the right uterine horns and no fluorescent material was found to penetrate into fetuses. FITC-dextran passed simultaneously into the uterine artery and the uterine branch of the ovarian artery, supplying placentas according to their order from the cervix and the ovary, respectively. First, placentas closest to the ovary and the cervix were supplied, and only then placentas of the fetuses located in the middle were supplied. Arterial arrival times were determined from the normalized signal enhancement in each placenta as a function of time (Fig. 8h, 8i). Experimentally measured arrival times varied between 0.5 to 0.7 sec; longer arrival times were recorded to placentas in the middle of the uterine horn (compared to placentas closer to the cervix and ovary), but in all cases the arrival time was short relative to the T₁ of blood.

Figure 8.

Intravital Fluorescence Microscopy of the arterial blood supply to the uterine horns after intravenous administration of FITC-dextran in a pregnant ICR mouse (E17.5). (a) Light image of an exposed uterine horn before FITC-dextran injection. (b-f) Fluorescent images of the same uterine horn in subsequent time points after FITC-dextran administration. The entry of FITC-dextran to the uterine artery and the uterine branch of the ovarian artery occurred simultaneously (white arrows in panel b), and supply to the placentas occurred according to their order from the cervix and the ovary, respectively. Placentas of fetuses located in the middle were supplied last; (g) Enlarged fluorescent image of the red box in panel f shows a placenta and a fetus within the uterine horn; (h) Enhancement in fluorescent signal of individual placentas plotted as a function of time post FITC-dextran administration; (i) Arterial arrival times of FITC-dextran to individual placentas along the uterine horn as measured from panel h data (white arrow). Experimentally measured arrival times varied between 0.5 to 0.7 sec and longer arrival times were recorded to placentas in the middle of the uterine horn compared to placentas closer to the cervix or ovary.

DISCUSSION

Non-invasive, in utero imaging of developing fetuses and placentas adds important dynamic and physiological information that complements postpartum or ex vivo analysis of mammalian development. In litter-bearing animals such as the mouse, the ability to non-invasively determine fetal location along the uterine horn and therefore to accurately identify individual fetuses throughout the course of pregnancy, as reported here, opens the possibility of monitoring the development of specific fetuses in longitudinal studies. The work reported here demonstrates the feasibility of utilizing ASL MRI as a tool for individualized determination of the fetal order along the uterus horn in vivo, via the specific pattern of blood flow within the mouse uterine horns.

Arterial Spin Labeling MRI, using different labeling approaches, has been applied previously for the study of perfusion in a number of organs (11-20). However, this is the first application for monitoring the directional dependent contribution of the dual arterial supply of maternal blood to the perfusion of each individual placenta. Classical ASL maps are derived from two images: a labeled image (flow sensitive) and a control image, in which the saturation is positioned symmetrically on the other side of the imaging plane, so that the static components remain identical, but magnetization of the inflowing blood differs. The relative signal difference (ΔS/S) reflects local tissue perfusion (11,12). In contrast, in the study reported here, BD-ASL MRI of the mouse placenta was designed to place the saturation pulses on either the ovarian or the uterine arteries, thus allowing us to probe the complex, dual arterial blood supply (uterine artery and the uterine branch of the ovarian artery) to each of the uterine horns. For that purpose, in the current study the saturation of the magnetization of water in the arteries in the upper image was achieved by placing the saturation slab on the heart, perpendicular to the flow in the aorta; this resulted in the tagging of all arterial blood input to the body. The saturation pulse in the lower image was positioned on the bifurcation point of the common iliac artery, such that the caudal entry of the uterine artery was saturated. Approximate steady-state saturation was obtained by slice-selective pulses and gradient dephasing with appropriate setting of the saturation parameters for each phase encoding and slice-select pulse. The method relies on rapid transfer of the magnetization label between the artery and the perfused placenta. In addition, the clearance rate of water from the placenta is assumed to be slow relative to the rate of pulse labeling. This new design allows us to distinguish between fetuses located closer to the ovary and those located farther (caudally) along the uterine horn. The normalized difference between the two images (lower minus upper) yields directionally-dependent BD-ASL maps: negative BD-ASL pixels represent regions in the placentas that are perfused more dominantly by the uterine artery and positive pixels regions perfused more dominantly by the uterine branch of the ovarian artery. Interestingly, for some of the placentas located in the central region of particularly crowded horns, the BD-ASL maps showed bimodal distribution with both negative and positive voxels.

Assessing the anatomical positional order of the fetuses from their locations in 3D images is prone to error, particularly for the middle fetuses, as the fetuses are not positioned according to their order along the uterine horn. The relative location of each embryo/fetus depends on the available free space within the maternal abdomen and changes in longitudinal studies with the relative movement of the embryos/fetuses.

Using the current ASL protocol, careful interpretation of the results is necessary. Quantitative analysis of perfusion or derivation of kinetic parameters of water exchange in the uteroplacental-fetal unit will require independent determination of the degree of saturation and the recovery of magnetization during the transit time. Thus, for all placentas, BD-ASL values are related to perfusion; however, perfusion could be underestimated due to incomplete saturation and saturation-recovery effects. Notably, the placental BD-ASL maps are not only perfusion-weighted, but also significantly position-weighted.

The correlation between the saturation-transfer maps and the fetus position along the uterine horn could be affected by a number of confounding factors: 1) The dominant effect, as shown here, is the position-dependent supply of blood to the placenta, via the ovarian and the uterine arteries. 2) T₁ relaxation effect, which is related to the transit time along the descending aorta and the relevant arteries, will lead to increased magnetization transfer to fetuses at the upper and lower ends of the uterine horn relative to the central fetuses; and 3) Magnetization transfer (MT) between water and large macromolecules, which could lead to reduction in signal intensity in the image, regardless of perfusion, but still be dependent on the distance between the saturation and the imaging plane (28). The current in vivo fluorescent microscopy analyses of the arterial blood flow validate that the longest transit times for water were for placentas located in the middle of the uterine horn. These placentas displayed smaller BD-ASL values, independent of perfusion rate. However, even for these placentas, the transit times were much shorter than T₁. In our experimental setting, in each set of ASL pulses the offset frequencies are different, i.e., δ_{heart–imaging} ≠ δ_{bifurcarion–imaging}, with offset frequency values of ±4kHz≤δ≤±16kHz for distances in the range of 1 to 5 cm. MT, which is effective over this range of frequencies, might reduce the signal intensity of the images. Nevertheless, control images acquired from a dead mouse showed a mean BD-ASL value of −0.9, attributed mainly to MT effects, confirming that the contribution of MT is small relative to that of perfusion.

In summary, we report here a novel approach for noninvasive in utero identification of fetuses utilizing directional-dependent coding of placental perfusion by BD-ASL MRI. The distinctive signature of each fetus is due, predominantly, to the relative contribution of each of the two arteries, ovarian and uterine, to the placental perfusion. Such a unique placental perfusion pattern could assist in distinguishing and following each fetus in utero. The ability to non-invasively determine fetus location along the uterine horn opens possibilities for determining and pursuing phenotypic alterations in genetic, as well as developmental, longitudinal studies. Furthermore, this new methodology can be applied to animal models having targeted defects in placental function, as it might provide insight into serious human prenatal conditions such as fetal growth restriction, preeclampsia, and fetal death in utero, conditions characterized by abnormal placental function.