Probing morphological, genetic and metabolomic changes of in vitro embryo development in a microfluidic device

Assisted reproduction technologies for clinical and research purposes rely on a brief in vitro embryo culture which, despite decades of progress, remain suboptimal in comparison to the physiological environment. One promising tool to improve this technique is the development of bespoke microfluidic chambers. Here we present and validate a new microfluidic device in polydimethylsiloxane (PDMS) for the culture of early mouse embryos. Device material and design resulted embryo compatible and elicit minimal stress. Blastocyst formation, hatching, attachment and outgrowth formation on fibronectin‐coated devices were similar to traditional microdrop methods. Total blastocyst cell number and allocation to the trophectoderm and inner cell mass lineages were unaffected. The devices were designed for culture of 10–12 embryos. Development rates, mitochondrial polarization and metabolic turnover of key energy substrates glucose, pyruvate and lactate were consistent with groups of 10 embryos in microdrop controls. Increasing group size to 40 embryos per device was associated with increased variation in development rates and altered metabolism. Device culture did not perturb blastocyst gene expression but did elicit changes in embryo metabolome, which can be ascribed to substrate leaching from PDMS and warrant further investigation.


| INTRODUCTION
This work has impact on microfluidics, embryology and metabolomics techniques. With the final goal to develop an alternative, simplified embryo culture system to improve the efficiency of in vitro fertilization (IVF) procedures for breeding genetically altered mice, we completed a thorough study of preimplantation embryo development applying an exhaustive set of analytic techniques. We observed subtle, different developmental characteristics of embryos cultured in drops under oil or in a microfluidic, closed conduit. These data are essential for planning carefully the embryo transfer trial, in compliance with 3Rs principle, to confirm the impact of the microfluidic system on the embryo quality and to extend its application to other research/ domestic species and humans. We linked morphokinetics, genetic and metabolomic profiles with chemical and physical characteristics of the manufacturing plastics. These data are informative for the research community in microfluidics and embryology, providing additional understanding on the safety of manufacturing plastics in IVF.
Genetically altered (GA) mice are widely used as models for developing clinical practice and furthering our understanding of human and animal reproduction and diseases. Almost 50% of the total amount of animals used for scientific research are genetically modified animals, the majority of which are mice. 1 In recent years a resurgence in the use of GA mouse models in medical research and pharmaceutical industry has occurred thanks to the introduction of sequencing methods, and our ability to combine genetic engineering technologies 2 such as CRISPR/Cas9, with methods used for microinjection, nuclear transfer as well as IVF, embryo and stem cell culture and derivation. The animal facilities for breeding GA mice are increasingly reliant on assisted reproduction technologies (ART) to generate embryos in vivo and in vitro for manipulation and cryopreservation to build research banks of GA embryos or to distribute GA embryos or animals to the research community. 3,4 High standards of animal welfare are required to support the processes involved in the generation of GA animals for research in order to generate, robust, high-quality data. 5 Many mouse GA facilities therefore need to implement significant changes in their breeding protocols and production methods in order to maximize efficiency, reduce animal suffering, improve throughput and increase pregnancy and birth rates of genetically modified animals. [6][7][8] The in vitro production of mouse embryos typically involves the culture of multiple embryos in 5-100 μl microdrops of specialized embryo culture medium 9 in Petri dishes covered with a layer of mineral oil to prevent media evaporation and associated changes in pH and osmolarity (Figure 1a). Single cell, fertilized zygotes or early cleavage staged preimplantation embryos derived in vivo or in vitro are transferred to sterile microdrops of pre-equilibrated, defined culture media and allowed to develop undisturbed in a humidified and gassed incubator environment for 4-5 days until they reach the blastocyst stage of development. 10 When ready for transfer, mouse blastocysts are aspirated and injected in the oviduct or uterus by traditional surgical embryo transfer (SET) techniques or by nonsurgical embryo transfer (NSET). 7 Embryo handling for both SET and NSET transfers involves manual pipetting, which is labor intensive, time consuming, and complicated by the presence of the mineral oil over layer in the culture dish. First implemented in 2009, NSET techniques are now adopted globally and are of similar or improved efficiency to SET. 6,7,11 Considerable research effort in a range of species has confirmed that embryo development and implantation efficiency are highly variable and are critically dependent on a range of factors that include the embryo development stage at transfer 12 and the creation and use of species specific, optimal culture environments. 13 Higher implantation and pregnancy rates post transfer have been reported when culturing embryos to the morula or blastocyst stage in vitro prior to transfer, a procedure requiring up to 4 days of culture in the mouse and 5-8 days in other mammalian species. [14][15][16] Since the 1990s, microfluidics has been proposed as a new approach to optimize the specialized in vitro requirements for successful ART. In 2013, Swain summarized the advantages and limitations of the microfluidic approach for clinical procedures in ART (these include oocyte maturation, manipulation, embryo culture, cryopreservation and noninvasive quality assessment). 17 LeGac also reported an overview of existing microfluidic platforms specifically used for embryo culture and characterization, discussing both intrinsic benefits and factors which currently limit the adoption of those innovative techniques in IVF clinical laboratories. 18 Moreover, Esteves et al. assessed the beneficial effect of volume reduction on single and group embryo culture using a microfluidic chamber that supported improved blastocyst development without altering birth rate. 19 Later in 2017, Ferraz et al. attempted to create an oviduct-on-a-chip platform using 3D printing technology. 20 Although this system was successful for sperm penetration of bovine oocytes and reduction of F I G U R E 1 (a) Traditional microdrop culture in a 60 mm dish. (b) Image of the fabricated PDMS microfluidic device sitting within a standard 60 mm culture dish. Red dye indicates inlet and outlet ports and microfluidic channels. Image shown next to a 1 GB pound coin for scale. (c) Schematic of in vitro murine embryo development and noninvasive analytical methods used to monitor and stage the embryos during culture and to evaluate single embryo quality abnormal fertilization compared to standard IVF systems, the material used for 3D printing resulted toxic to fertilized oocytes. 21 The same group recently developed a polydimethylsiloxane (PDMS) oviduct-on-achip device for the culture of oviductal epithelial cells and production of bovine zygotes. 22 However, none of these systems has been translated in commercially available devices. While the interest in the use of these microfluidic systems is still high, extensive scientific analyses to assess safety, consistency and accuracy remain and the long-term impact of culturing embryos in microfluidic devices is not completely understood.
In recent years, the variable developmental competence (quality) of embryos produced using ARTs has led to the development of a range of invasive and noninvasive clinical and research methods that can be used to directly quantify and hence predict embryo implantation and pregnancy potential. 23 These embryo quality assays can now be used to generate an in-depth understanding of the impact of microfluidic culture and the reduced culture volumes on the health and developmental potential of in vitro-derived mouse embryos. Morphokinetics (embryo morphology, cleavage rate, timing, cell number and fragmentation), 23 41 can be used to provide valuable insights into the potential effects of the microfluidic environment on key markers of embryo development and health. 42,43 In the current study we developed and fabricated a novel, oil-free disposable microfluidic device in PDMS, in which fertilized 1 cell/two pronuclei stage murine zygotes can be grown to the expanded blastocyst stage in vitro and retrieved for subsequent embryo transfer ( Figure 1b). The requirement for manual pipetting of embryos is reduced compared to transfer between microdrops under oil, due to the relative ease of loading of embryos in a simple step, as a group, or individually, into the device chamber by flow of media from the inlet to outlet. Further washes of embryos or media changes, if required, can be performed by replacing the media and without direct movement of the embryos. In order to minimize any fluid dynamic shear stress during embryo handling and their injection into the device, the microfluidic design was optimized using finite element modeling; the model also verified the efficiency of the embryo loading and nutrient diffusion in the system. The potential toxicity of PDMS was assessed by performing the mouse embryo assay, in which embryo cleavage, blastocyst development, hatching and outgrowth rate were used as predictive indexes of embryo health and implantation potential. The embryo culture conditions were optimized by studying the effect of group embryo culture on blastocyst development rate, performing metabolic profiling and determining mitochondrial polarization ratios. Furthermore, the gene expression profiles of blastocysts developed inside the system were compared to those cultured in traditional microdrops using real-time PCR analysis ( Figure 1c). These data were used to exclude potential genetic alterations induced by the different environment and culture methods. Finally, global untargeted metabolomics was used to identify PDMS-released compounds from culture media extracted from the microfluidic device at different time points (24 h and 5 days).

| Microfluidic device flow and shear stress analysis
All chemicals were purchased from Sigma Aldrich (St Louis, MO, USA) unless specified otherwise. COMSOL Multiphysics 5.2a was used to evaluate and compare flow rate, velocity field and predict shear stress as function of microfluidic device geometry. To generate a 3D model, the microfluidic design was first created by using computer-aided design software (Autodesk AutoCAD 2017). The design geometry was then imported into a COMSOL library. The fluid inside the device was simulated as an incompressible, homogeneous, Newtonian fluid with density (ρ = 1000 kg m À3 ) and viscosity (μ = 1 Â 10 À3 Pa s). 44     Dishes were pre-equilibrated at 37 C under 5% CO 2 , 5% O 2 in humidified nitrogen for 2 h before use. Embryos were cultured in these conditions throughout early cleavage and blastocyst development. Embryos were checked daily for developmental progression on a Nikon inverted microscope. Blastocyst development was recorded on day 5, while culture was extended by a further 96 h for attachment and outgrowth formation. 46 Outgrowth was imaged and diameters measured and recorded at 200Â power using RI viewer software. At the end of outgrowth culture, all embryos were tested for attachment by gentle pipetting using a 170 μm embryo pipette. Embryos which were not displaced by the flow of medium were considered to be attached.

| Experiment 2: Blastocyst rates, energy substrate consumption and mitochondrial polarization
To define the loading capacity of device culture, groups of 10, 20, 30 and 40 2-cell mouse embryos were cultured to the blastocyst stage in microfluidic devices in parallel to controls (10 embryos in one 10 μl culture microdrop) before metabolic profiling. A total of 220 blastocysts were analyzed in each treatment group from five replicate cultures. Therefore, the embryo: medium ratio varied from the widely used 1:1 μl in microdrop controls to 1:0.01 in the chamber of the 40 group.
Glucose and pyruvate consumption were measured using spent media from devices or microdrops by the method of Guerif et al. 28 and expressed as mean pmol/embryo/h ± SEM. Briefly, 10 μl of reaction mixture was added to the base of a black 384 well microplate. All media from each device was collected in a 500 μl microtube and measured with a pipette to identify and account for any change in total volume. One microliter of spent media from each device or microdrop was added to each reaction mixture well and the difference in NAD, NADH or NADP signal, measured at 340/460 nm, was calculated.
Concentrations were calculated using a 6-point standard curve after correction for the final volume retrieved from each device. Consumption data were expressed in terms of pmol/embryo/h using the recorded culture time.
Day 5 blastocysts were labeled and imaged within devices using the ratiometric mitochondrial dye JC-10 (Molecular Probes). Briefly, a 1 mg/ml JC-10 stock solution was prepared in KSOM culture media. JC-10 stock was diluted to 10 μg/ml in pre-equilibrated M2 media (Millipore). Following removal of spent media for energy substrate assays, media was replaced with this JC-10 solution and incubated at 37 C for 30 min. Control embryos must be moved to glass slides for imaging, while device embryos were imaged within devices. Imaging was performed on a LSM 780 confocal microscope, while image analysis was carried out in ImageJ. JC-10 accumulates in the matrix of polarized mitochondria, forming J-aggregates with punctate red fluorescent signal. In areas of reduced mitochondrial polarization, the dye tends to remain in monomeric form with diffuse, green signal. A ratio of red+green/total signal intensity, thus indicates overall mitochondrial polarization in the images sample and is termed the mitochondrial polarization ratio. 47 Data were represented as red fluorescence/total red + green fluorescence to account for any nonspecific fluorescence across both channels. Higher values indicate a higher level of mitochondrial polarization throughout the measured region of interest.

| Experiment 3: Real-time PCR (qPCR) of single blastocysts
Groups of 10, 1-cell murine embryos were cultured in KSOM medium in 10 μl microdrops (control) and inside the devices. Ninety blastocysts were used in each treatment group from three replicate cultures. From these cultures, 10 stage matched, expanded blastocysts were selected and analyzed for each experimental group. Individual blastocysts were recovered from devices or control microdrops, and immediately transferred into 2 μl RNAGEM lysis buffer (RNAGEM Tissue Plus ® , MicroGem International PLC, Southampton, UK) and frozen at À80 C. For the construction of cDNA libraries of individual blastocysts, we modified an existing protocol. 48 In summary, total RNA from single blastocysts was isolated using an RNAGEMextraction reagent mastermix. The total RNA was reverse-transcribed to cDNA using a first strand cDNA synthesis kit (Thermo Fisher Scientific Inc., UK). Quantitative PCR (qPCR) was performed on 6-fold diluted sample cDNA to analyze expression of 53 selected genes associated with blastocyst development and cell differentiation.
Accession number, primer sequence and product length of target genes are presented in Table S1. Ten individual blastocysts were analyzed in each experimental group. mRNA expression was examined using SYBR Green Master PCR mastermix (Thermo Fisher Scientific Inc., UK) with an ABI 7500 RT-PCR System (Applied Biosystems) over 40 cycles and using the house keeping genes listed in Table 1. Data were analyzed with 7500 Software using relative quantification analysis.

| Experiment 4: Global untargeted metabolomics
To investigate impact of microfluidic culture on the embryo secretome and to establish if the PDMS matrix released low molecular weight species or sequestered hydrophobic biomolecules from culture T A B L E 1 Summary of gene symbol, accession number, product length and primer sequences of the housekeeping genes As a control, spent media from devices without embryos was compared to media collected from microdrops without embryos using the same incubation time performed for embryo culture. These experiments were performed to investigate PDMS leaching and/or molecule absorption and adsorption. Spent culture media was frozen and stored at À80 C prior to sample preparation for analysis. Culture media samples (100 μl) were thawed on ice and prepared using previously described methods. Briefly, 300 μl of dry ice cooled methanol was added to individual culture medium samples and incubated overnight at À80 C. Individual samples were spun down to remove proteins and the subsequent supernatant was used for analyses. Samples were separated and analyzed using reverse-phase liquid chromatography connected to a Thermo Scientific Q Exactive HF (LC-Hybrid Quadrupole-Orbitrap MS/MS) instrument using positive ion mode MS. 50 55 Metlin, 56 MassBank, 57 and the National Institute of Standards and Technology database. 58 Increased confidence in the annotation of many features was achieved by manually assessing spectral match and RT consistencies between experimental data and chemical standards within a curated in-house library.  as the culture chamber diameter, was adopted (i) to optimize the flow profile within the culture chamber ( Figure S2) and (ii) to observe fabrication requirements preventing the PDMS structure from collapse.

| STATISTICAL ANALYSIS
An additional design criterion was used to favor the spreading out of the embryos across the whole culture chamber and to ensure a homogeneous perfusion of medium in the chamber. As shown in Figure 2c and in Figure S2 Figure 5f). This suggests that substrate competition outweighs increased paracrine effects in larger group sizes.
Development rates are a widely used key marker of ongoing developmental competence. However, to examine embryo competence in richer detail, metabolic parameters were investigated. Blastocysts were labeled and imaged directly within the devices using the ratiometric mitochondrial dye JC-10 ( Figure 5a-d,g) to evaluate Pyruvate is the preferred energy substrate during early cleavage, while glucose consumption is low during early cleavage but tends to increase greatly with increased ATP generation through oxidative phosphorylation at the blastocyst stage. 29 The present data suggest increased competition for these substrates within the more concentrated population of 40Â embryos in comparison to groups of 10Â in devices or control drops. Device embryos were seemingly more quiescent overall, with reduced variation between culture groups. The quiet hypothesis of Henry Leese suggested that embryos with moderate, or 'quiet', overall metabolism may be most viable. [64][65][66] In contrast, embryos with insufficiently low metabolic rates and those with high metabolic rates may be stressed and have restricted development. 28 (Table S1) Similarly, among the 48 species that were significantly decreased in spent media from the device (Figure 8b), numerous biological compounds were identified that were sequestered by PDMS from the culture medium, these include: amino acids and dipeptides (e.g., isoleucyl-isoleucine, isoleucyl-leucine, isoleucyl-phenylalanine, n-acetyl-L-methionine, and valyl-leucine).
Among the species that were significantly decreased (57 compounds) in spent media from microdrop culture, peptides and amino acids, such as L-tyrosine, aspartylphenylalanine and phenylacetylglycine were identified. These data suggest that these molecules were absorbed or transformed by the plastic or mineral oil. The remaining detected compounds represent breakdown products of culture media that degraded over time.
Other organic species appeared down-regulated in PS-media and  Tables S2 and S3.
Notably, among the 44 common metabolites present in both embryo culture PDMS-media and embryo culture PS-media, the relative abundance of 42 of these compounds were statistically significantly increased (p < 0.05 and fold change ratios >2) in both media when compared to control, these identified compounds include metabolic markers of preimplantation embryo development, such as pyroglutamic acid, 5 0 -methylthioadenosine, 75,71 hypoxanthine, 76,77 cytosine, n-acetyl-L-methionine, and phenylacetylglycine. Preimplantation embryo development compounds represent metabolites produced by embryos in both culture conditions. Similarly, two compounds significantly decreased in spent embryo culture media compared to control (i.e., muramic acid and meticillin) might be consumed by embryos from the culture medium.
In summary, these results allow us to conlcude that metabolomic changes could be detected by mass spectrometric analysis in samples of media collected during embryo culture, and these were correlated with organic and inorganic compounds available to the embryos during their development in vitro.

| DISCUSSION
Based on this extensive evaluation, we demonstrated the ability of Second-order meta-analysis allowed for biologically important metabolite changes to be observed in the culture medium throughout embryo development. Pathways overrepresentation analysis showed that microfluidic culture had a significant impact on tryptophan metabolism pathway, which could explain the resulting activity of protein synthesis mechanisms fundamental during embryo development.
Importantly, MS data did not reveal alteration of metabolites involved in metabolic pathways of glucose, pyruvate and lactate. This suggests that the microfluidic environment does not alter energy substrate metabolism, as also shown by our metabolic profile data (Figure 5h-j).
Similarly, no significant changes in abundance of metabolites involved in oxidative stress processes were detected from the analyzed datasets, which demonstrates that the plastic does not impact the release of reactive oxygen species into the culture media. Oxygen availability is also a key requirement for embryo development. 80 In this work, we assessed the effect of the confinement of the embryos in a close compartment, completely surrounded by PDMS, which is permeable to gas. 81,82 Oxygen tension of 5% for embryo culture has been widely adopted in both animal and clinical human IVF laboratories and considered a more physiological concentration that can boosts blastocyst development with no detectable adverse effects. 83,84 While it is true that the diffusion of oxygen through the PDMS could be modeled and compared to that through mineral oil and media, with optimal embryo loading we did not observe detrimental effects on the embryo development due to a different oxygen dynamic.

CONFLICT OF INTEREST
The authors do not have any conflict of interest with the described research results.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request. The full collection of raw data on the MASS SPEC analysis has been published on Metabolomics Workbench at https://doi.org/10.21228/M8W99G.