Recent advances in sex-specific reproductive microphysiological systems for the systematic evaluation of assisted reproductive technology.

Conventional sperm selection methods such as swim-up and density gradient centrifugation effectively isolate highly motile sperm but often induce mechanical stress and oxidative damage due to centrifugal forces. These effects increase DNA fragmentation, which can subsequently lead to reduced fertilization rates and impaired embryo development. Due to these limitations, microfluidic-based sperm concentration and selection technologies have recently gained attention in both animal reproduction research and clinical ART procedures. Funahashi and Nagai introduced a “climbing-over-a-wall” sperm selection and IVF system using porcine oocytes. This design required sperm to migrate over a physical barrier, effectively selecting for those with superior motility. As a result, the incidence of polyspermy was significantly reduced, and fertilization efficiency improved, as only the most motile sperm were able to reach the oocyte [ 177 ]. Following this work, Clark et al. developed a biomimetic microfluidic chip that optimized sperm concentration and motility by controlling sperm kinematics and interactions within a flowing medium ( Fig. 4 A) [ 178 ]. The chip incorporated microstructures mimicking the oviductal isthmus, which guided rheotactic sperm movement while filtering out immotile sperm. This system maintained a high rate of monospermic fertilization even at low sperm concentrations, providing a more stable and physiologically relevant fertilization environment compared with conventional high sperm count approaches. Fig. 4 Examples of MPS for IVF stages. (A) A microfluidic fertilization system that guides rheotactic sperm through microstructures to enrich motile sperm and maintain high monospermic fertilization even at low sperm concentrations. Reproduced with permission from Ref. [ 178 ], © 2005 Royal Society of Chemistry. (B) A closed-loop automated microfluidic platform for oocyte recovery from follicular fluid. Reproduced with permission from Ref. [ 164 ], © 2026 Springer Nature. (C) A microfluidic fertilization platform with transparent ITO electrodes that traps oocytes and concentrates motile sperm around them under an AC electric field. Reproduced with permission from Ref. [ 179 ], © 2018 Wiley. Fig. 4 Examples of MPS for IVF stages. (A) A microfluidic fertilization system that guides rheotactic sperm through microstructures to enrich motile sperm and maintain high monospermic fertilization even at low sperm concentrations. Reproduced with permission from Ref. [ 178 ], © 2005 Royal Society of Chemistry. (B) A closed-loop automated microfluidic platform for oocyte recovery from follicular fluid. Reproduced with permission from Ref. [ 164 ], © 2026 Springer Nature. (C) A microfluidic fertilization platform with transparent ITO electrodes that traps oocytes and concentrates motile sperm around them under an AC electric field. Reproduced with permission from Ref. [ 179 ], © 2018 Wiley. Suh et al. further demonstrated that applying microfluidic channels to human IVF significantly reduced both the total sperm usage and concentration while maintaining high fertilization efficiency. In this system, laminar flow allowed controlled and limited contact between oocytes and sperm, thus preventing polyspermy while ensuring effective fertilization [ 180 ]. Collectively, these studies demonstrate that microfluidic-based systems offer a technological alternative capable of optimizing sperm concentration under physiological conditions without imposing excessive physical stress. Such systems improve both fertilization outcomes and embryo quality while reducing polyspermy, a frequent issue in animal IVF. Moreover, the ability to achieve high-efficiency fertilization with minimal sperm concentration represents a significant advancement in MPS and reproductive bioengineering. Recent technological advances have targeted automated microfluidic strategies to enhance oocyte isolation from FF. In a recent multicenter clinical study involving 582 patients across four IVF centers, a fully integrated microfluidic platform was applied to discarded FF samples following routine manual screening ( Fig. 4 B) [ 164 ]. The device, which incorporates sequential filtration, denudation, concentration, and capture modules within a closed automated workflow, recovered additional oocytes in more than half of the processed samples, indicating that conventional manual screening may underestimate the recoverable oocyte pool. These systems have been reported to recover additional oocytes that may be overlooked during conventional manual screening, potentially offering particular benefit to patients with low oocyte yield. Notably, the additional oocytes retrieved from discarded FF exhibited comparable maturation status and developmental competence to manually collected oocytes, including similar fertilization rates and blastocyst formation efficiency. In at least one reported case, an embryo derived from an oocyte recovered using a microfluidic device resulted in a successful live birth, underscoring the potential clinical relevance of automated retrieval. Collectively, these findings suggest that microfluidic automation may extend beyond improving retrieval efficiency and provide a technological foundation capable of influencing downstream IVM and IVF outcomes. Furthermore, automation of the oocyte retrieval process may reduce operator dependency and inter-laboratory variability, thereby contributing to improved standardization and broader accessibility of ART in the long-term. While automated recovery maximizes the available oocyte pool, the focus of microfluidic integration extends to the subsequent stage of precise downstream manipulation. To address the structural fragility and large scale of oocytes, recent MPS technologies have introduced specialized features for accurate positioning, trapping, and orientation control, thereby enhancing the efficiency of fertilization and micromanipulation. Passive positioning methods typically rely on geometric design or fluid dynamic characteristics [ 181 ]. For example, microarrays of wells or columns with diameters similar to those of oocytes can be fabricated, allowing the oocytes to naturally settle into designated positions [ 182 ]. Another passive approach involves density-based sedimentation, in which sucrose concentration gradients are utilized to separate healthy oocytes according to their sedimentation rates. In a bovine oocyte chip, this method enabled automatic selection of viable oocytes based on density-dependent settling velocity [ 183 ]. Active control methods allow more precise manipulation. A common technique involves micro-suction, which immobilizes oocytes at specific positions to maintain consistent polar body orientation, thereby increasing the reproducibility of ICSI [ 184 ]. Dielectrophoretic control can also be used to trap or transport oocytes without physical contact by applying electric fields. More recently, magnetic microrobot-based oocyte manipulation has been introduced, enabling the rotation and translation of oocytes within a chip through externally applied magnetic fields [ 185 , 186 ]. Acoustic microfluidic systems employing vibration-induced microflows have also been developed to achieve 3D rotation of oocytes [ 185 ]. Collectively, these technologies enhance spatial precision during procedures such as ICSI, assisted hatching, and biopsy, while also contributing to procedural consistency and reduced operator-dependent variability. Dielectrophoresis (DEP) is a technique that manipulates cells within non-uniform electric fields according to their intrinsic electrical properties such as membrane charge and permittivity, enabling precise and contact-free control of cell movement. In IVF systems, DEP has been applied to both sperm and oocytes for operations such as sperm concentration, viable sperm slection, and spatial cell arrangement. Alternating current electric potentials within the range of several hundred kHz to a few MHz, with amplitudes between 5 and 20 V p-p , are typically utilized. The direction of the DEP response, either positive or negative, depends on the electrode geometry and the conductivity of the medium [ 187 , 188 ]. Huang et al. developed a dielectrophoretic chip with transparent indium tin oxide electrodes that immobilized oocytes and concentrated motile sperm around them under an electric potential of 10 V p-p at 1 MHz ( Fig. 4 C) [ 179 ]. This configuration created an on-chip fertilization environment that achieved approximately a 5% higher fertilization rates compared with conventional IVF, and about 20% of the embryos developed to the blastocyst stage even when the sperm to oocyte ratio was very low. Zhang et al. demonstrated that DEP can distinguish live sperm from damaged ones through differences in their electrical properties [ 189 ]. Viable and nonviable sperm exhibit distinct dielectrophoretic behaviors at specific frequencies due to variations in membrane charge distribution and capacitance. By tuning the electrical conditions, intact sperm can be selectively captured on electrodes. Building on this concept, Kim et al. reported that subtle electrical differences between sperm carrying X chromosomes and those carrying Y chromosomes could enable sex-specific sperm separation [ 190 ]. Liao et al. further advanced electrode design from planar interdigitated configurations to 3D nanoelectrode arrays [ 191 ]. They fabricated titanium nitride nanoelectrode arrays using a complementary metal oxide semiconductor compatible process and successfully captured approximately 60% of motile sperm at 3 MHz and 20 V p-p . The system achieved a processing throughput of several milliliters per hour, demonstrating the practical potential of DEP-based sperm separation in ART. The integration of DEP with microfluidic control is expected to lead to fully automated IVF chips capable of performing continuous processes such as debris removal, sperm trapping, oocyte positioning, and controlled fertilization within a single platform.

MPSs associated with ART are predominantly integrated with microfluidic technologies to recapitulate complex in vivo environments. MPSs using microfluidics facilitate continuous nutrient supply and metabolic waste removal while providing precisely controlled shear stress as a mechanical stimulus [ 70 ]. Microfabrication technology enables the construction of microfluidic networks on a physiological scale. It facilitates the guided three-dimensional (3D) organization of cells when coupled with hydrogel matrices [ 71 ]. In this chapter, we provide detailed strategies for microfluidic MPS design, covering four key aspects: material selection, compartmentalization, fluid flow control, and long-term culture ( Fig. 2 ). We also introduce sex-specific design considerations for reproductive MPS. Fig. 2 MPS design strategies for different ART processes. Design strategies to be considered for modeling each ART stage are presented using a color code. (A) Four key MPS strategies are shown: materials selection (blue), compartmentalization (red), fluid flow control (green), and long-term culture (yellow). (B) Stages of the ART process include in vitro sperm production, sperm selection, in vitro oocyte production, in vitro maturation, oocyte selection, in vitro fertilization, embryo implantation, embryo development, reproductive toxicity, and body-on-a-chip incorporating reproductive system. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Fig. 2 MPS design strategies for different ART processes. Design strategies to be considered for modeling each ART stage are presented using a color code. (A) Four key MPS strategies are shown: materials selection (blue), compartmentalization (red), fluid flow control (green), and long-term culture (yellow). (B) Stages of the ART process include in vitro sperm production, sperm selection, in vitro oocyte production, in vitro maturation, oocyte selection, in vitro fertilization, embryo implantation, embryo development, reproductive toxicity, and body-on-a-chip incorporating reproductive system. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Materials used in reproductive MPS can be broadly divided into device substrates and hydrogel constructs. Device substrates determine structural stability and fluid control, while hydrogels recreate the biochemical and mechanical microenvironment that regulates cellular behavior. Therefore, material selection should be strategically aligned with the intended objectives and applications of the platform. A comparative summary of representative core device substrates and hydrogel constructs used in reproductive MPS is provided in Table 1 , Table 2 , respectively. Table 1 Comparison of core device substrates in reproductive MPSs. Table 1 Category Material Advantages Limitations Modified Performance Improvements Applications Reference Elastomer PDMS ∙ Optical transparency ∙ Gas permeability ∙ Biocompatibility ∙ Easy microfabrication ∙ High-resolution microstructure replication ∙ Absorption of hydrophobic small molecules ∙ Hydrophobic surface ∙ Limited scalability ∙ Oligomer leaching affect sensitive cells ∙ Parylene coating ∙ PEG/polydopamine grafting ∙ Surface oxidation stabilization ∙ Short-term culture ∙ Flow-controlled screening ∙ Imaging-based assays [ 72 , 73 ] Thermoplastic PMMA ∙ Mechanical robustness ∙ Chemical stability ∙ Compatible with injection molding and hot embossing ∙ Low gas permeability ∙ Hydrophobic surface ∙ Complex bonding ∙ Autofluorescence in imaging applications ∙ Plasma treatment ∙ ECM coating ∙ Thermal/solvent bonding optimization ∙ Scalable, structurally stable chips ∙ Industrial manufacturing platforms [ 74 ] Thermoplastic PC ∙ Chemical resistance ∙ High mechanical strength ∙ Dimensional stability under flow ∙ Compatible with industrial-scale injection molding ∙ Low gas permeability ∙ Hydrophobic surface ∙ Lower fabrication flexibility ∙ Bonding difficulty ∙ Plasma treatment ∙ ECM coating ∙ High-pressure perfusion systems [ 75 , 76 ] Table 2 Comparison of core hydrogels in reproductive MPSs. Table 2 Category Material Advantages Limitations Modified Performance Improvements Applications Reference Natural hydrogel Collagen ∙ Biologically relevant fibrillar architecture ∙ Supports 3D cell structure ∙ Tunable mechanical stiffness ∙ Limited mechanical stability under long-term culture ∙ Batch variability ∙ Depending on source ∙ Cell-mediated matrix contraction ∙ GMP-grade collagen ∙ Recombinant human collagen ∙ Stromal remodeling ∙ Trophoblast invasion modeling [ 77 ] Natural hydrogel Fibrin ∙ Pro-angiogenic properties ∙ Supports endothelial network formation ∙ Rapid degradation ∙ Limited long-term structural stability ∙ Tunable stiffness via fibrinogen and thrombin concentration ∙ Implantation-like environments ∙ Placental interface model ∙ Vascularized endometrium model [ 78 , 79 ] Natural hydrogel Matrigel ∙ Supports epithelial polarity ∙ Organoid formation ∙ Batch-to-batch variability ∙ Limited tunability of stiffness ∙ Tumor-derived and xenogenic origin ∙ Growth factor–reduced (GFR) Matrigel ∙ Hormone-responsive epithelial models ∙ Organoid chip [ 80 , 81 ] Natural hydrogel dECM ∙ Preserves native ECM composition ∙ Organ-specific biochemical cues ∙ Natural cell–matrix interactions ∙ Donor variability ∙ Mechanical tuning challenges ∙ Variable growth factor content ∙ Standardized processing ∙ Long-term endocrine models ∙ Ovarian ∙ Endometrial Decidualization Model [ 82 , 83 ] Synthetic hydrogel GelMA ∙ Intrinsic bioactivity ∙ Photocrosslinkable ∙ Microfabrication compatibility ∙ Restricted growth factor sequestration ∙ Photoinitiator-related cytotoxicity ∙ Limited tissue specificity ∙ Reduced biological fidelity ∙ Growth factor immobilization ∙ Follicle encapsulation model ∙ Stromal cell mechanobiology [ 84 , 85 ] Synthetic hydrogel PEG-based ∙ High reproducibility ∙ High structural stability ∙ Lack of intrinsic bioactivity ∙ Limited cell infiltration ∙ Heparin conjugation ∙ Photopatterning ∙ peptide grafting ∙ Hormone diffusion kinetics study ∙ Long-term endocrine cycling platform [ 86 , 87 ] Comparison of core device substrates in reproductive MPSs. Optical transparency Gas permeability Biocompatibility Easy microfabrication High-resolution microstructure replication Absorption of hydrophobic small molecules Hydrophobic surface Limited scalability Oligomer leaching affect sensitive cells Parylene coating PEG/polydopamine grafting Surface oxidation stabilization Short-term culture Flow-controlled screening Imaging-based assays Mechanical robustness Chemical stability Compatible with injection molding and hot embossing Low gas permeability Hydrophobic surface Complex bonding Autofluorescence in imaging applications Plasma treatment ECM coating Thermal/solvent bonding optimization Scalable, structurally stable chips Industrial manufacturing platforms Chemical resistance High mechanical strength Dimensional stability under flow Compatible with industrial-scale injection molding Low gas permeability Hydrophobic surface Lower fabrication flexibility Bonding difficulty Plasma treatment ECM coating High-pressure perfusion systems Comparison of core hydrogels in reproductive MPSs. Biologically relevant fibrillar architecture Supports 3D cell structure Tunable mechanical stiffness Limited mechanical stability under long-term culture Batch variability Depending on source Cell-mediated matrix contraction GMP-grade collagen Recombinant human collagen Stromal remodeling Trophoblast invasion modeling Pro-angiogenic properties Supports endothelial network formation Rapid degradation Limited long-term structural stability Tunable stiffness via fibrinogen and thrombin concentration Implantation-like environments Placental interface model Vascularized endometrium model Supports epithelial polarity Organoid formation Batch-to-batch variability Limited tunability of stiffness Tumor-derived and xenogenic origin Growth factor–reduced (GFR) Matrigel Hormone-responsive epithelial models Organoid chip Preserves native ECM composition Organ-specific biochemical cues Natural cell–matrix interactions Donor variability Mechanical tuning challenges Variable growth factor content Standardized processing Long-term endocrine models Ovarian Endometrial Decidualization Model Intrinsic bioactivity Photocrosslinkable Microfabrication compatibility Restricted growth factor sequestration Photoinitiator-related cytotoxicity Limited tissue specificity Reduced biological fidelity Growth factor immobilization Follicle encapsulation model Stromal cell mechanobiology High reproducibility High structural stability Lack of intrinsic bioactivity Limited cell infiltration Heparin conjugation Photopatterning peptide grafting Hormone diffusion kinetics study Long-term endocrine cycling platform Polydimethylsiloxane (PDMS) is a commonly used material in microfluidic MPS designs. As a silicon rubber, PDMS offers high transparency, gas permeability, biocompatibility, and low cost [ 88 ]. Although it enables the fabrication of high complex structures through soft lithography, large-scale production remains challenging [ 72 , 75 ]. Due to its inherent hydrophobic surface, surface treatment is necessary to improve hydrophilicity. Small molecules such as steroid hormones tend to be absorbed into PDMS, which can alter compound concentrations and result in inaccurate measurements [ 73 , 89 ]. This is a significant limitation for reproductive models that require quantitative analysis of endocrine responses. Furthermore, PDMS is characterized by hydrophobic recovery, meaning that the surface can return to a hydrophobic state even after hydrophilic treatments such as plasma treatment [ 90 ]. Such constraints can be partially addressed by surface coatings with polyethylene glycol (PEG), polydopamine (PDA), or extracellular matrix (ECM) proteins [ 91 , 92 ]. Consequently, PDMS is better suited for short-term flow control MPS applications like sperm selection or IVF. These applications prioritize gas exchange and precise microchannel architecture, rather than long-term modeling of hormonal cycles. PDMS can be substituted with thermoplastics such as polymethyl methacrylate (PMMA) and polycarbonate (PC). PMMA is highly optically transparent and has a low degree of autofluorescence. Its superior mechanical strength and chemical inertness allow precision fabrication techniques such as computer numerical control (CNC) milling [ 74 ]. Similarly, PC is a thermoplastic polymer characterized by high mechanical strength, excellent chemical stability, and optical clarity. Its high thermal stability makes it suitable for large-scale manufacturing processes, including hot embossing and injection molding [ 76 ]. However, the gas permeability of these thermoplastics is lower than that of PDMS, which may limit oxygen and carbon dioxide exchanges [ 93 ]. Since PMMA and PC also possess hydrophobic surfaces, surface treatments are required for cell adhesion [ 94 ]. This necessitates more complex integration procedures such as thermal or solvent bonding [ 95 ]. Thermoplastic materials generally exhibit lower sorption of hydrophobic small molecules than PDMS [ 96 ], which can be advantageous for reproductive MPS applications regulated by hormones. Hydrogels provide the biochemical and mechanical microenvironment that directly regulates cellular behavior within MPSs [ 97 ]. In reproductive models, ECM composition influences hormone responsiveness, stromal differentiation, trophoblast invasion, follicular maturation, and immune cell recruitment. Therefore, the selection of hydrogels is not merely a technical consideration but a central determinant of physiological relevance. Hydrogels can be classified into natural and synthetic types according to the origin and chemical characteristics of their constituent polymers. Natural polymer-based hydrogels are used in biomedical applications due to their intrinsic biocompatibility, biodegradability, and low toxicity, which facilitate cell-matrix interactions and tissue integration [ 98 ]. In contrast, synthetic polymer-based hydrogels are typically more hydrophobic and form stable covalently crosslinked networks, providing enhanced mechanical strength, structural stability, and a longer functional lifespan [ 99 ]. Natural hydrogels such as collagen type I, fibrin, and Matrigel have been used due to their inherent bioactivity and cell-adhesive motifs [ 100 ]. Collagen-based matrices provide structural support and enable the modeling of endometrial remodeling and trophoblast invasion [ 77 ]. Type I collagen self-assembles into fibrillar networks under physiological temperature and pH, forming physically entangled structures stabilized primarily by non-covalent interactions [ 101 ]. Matrix stiffness is primarily regulated by collagen concentration and fibril density and it can be modified via enzymatic crosslinking to increase fiber connectivity and bulk elasticity [ 102 ]. However, collagen matrices display source-dependent batch variability and polymerization sensitivity, leading to inconsistent mechanical properties [ 103 ]. Nevertheless, collagen scaffolds remain mechanobiologically responsive substrates capable of undergoing dynamic remodeling in response to cellular activity. For example, in an implantation model, extravillous trophoblast (EVT) invasion increased the Young's modulus of the collagen ECM from 390.24 ± 146.25 Pa to 826.26 ± 164.84 Pa after 6 days of culture, remaining within the physiological range [ 77 ]. Fibrin-based hydrogels are frequently employed in vascularized reproductive models owing to their pro-angiogenic properties and compatibility with endothelial network formation [ 78 , 104 , 105 ]. Fibrin gels are generated through thrombin-induced polymerization of fibrinogen, forming a fibrous network whose stiffness can be modulated by fibrinogen and thrombin concentrations [ 106 ]. Despite this modular control, fibrin matrices are susceptible to rapid proteolytic degradation and cell-mediated remodeling when stromal, mesenchymal, or trophoblast cells are embedded, which may compromise structural stability during long-term culture [ 107 , 108 ]. Matrigel, rich in laminin and growth factors, supports epithelial organoid formation and hormone-responsive differentiation [ 109 ]. Despite its widespread use, Matrigel presents significant limitations including undefined composition, high batch variability, a tumor-derived origin, and limited tunability of mechanical properties, which restrict translational reproducibility and regulatory acceptance [ 110 ]. Recently, decellularized extracellular matrix (dECM) has emerged as a tissue-specific alternative to conventional hydrogel systems, retaining native biochemical cues and organ-derived structural proteins [ 111 ]. dECM derived from reproductive tissues has been reported to enhance lineage-appropriate differentiation, sustain long-term hormonal responsiveness, and preserve physiologically relevant phenotypes in vitro [ 112 ]. In hormone-regulated reproductive systems, dECM-based matrices are particularly advantageous for modeling cyclic endocrine dynamics and tissue remodeling during extended culture periods. Nevertheless, dECM matrices inherently exhibit batch-to-batch variability and offer limited control over mechanical properties, complicating the precise regulation of matrix stiffness [ 113 ]. Moreover, cell-driven contractile forces and proteolytic activity can compromise structural stability during prolonged culture [ 114 ]. To overcome these limitations of natural hydrogels, synthetic or chemically modified hydrogels, including PEG-based systems and gelatin methacryloyl (GelMA), have been developed [ 115 , 116 ]. PEG provides a bioinert and highly tunable backbone that enables precise control over mechanical properties, degradation kinetics, and biochemical functionalization. For instance, PEG hydrogels conjugated with collagen-mimetic peptides (CMP) have been used to reconstruct the niche for spermatogonial stem cells [ 117 ]. In addition, PEG-based hydrogels can be engineered to incorporate bioactive adhesive ligands that support cell attachment and tissue-specific functionality. These matrices allow precise control over stiffness, degradability, and ligand density through controlled chemical functionalization [ 86 ]. Such tunability enables systematic investigation of mechanobiological parameters, including stiffness-dependent decidualization, follicular growth, or blood-testis barrier formation [ 118 ]. Nevertheless, some chemical polymerization approaches may leave residual unreacted crosslinking agents, raising potential cytotoxicity concerns that can limit broader clinical translation [ 119 ]. MPS architectures featuring compartmentalization enable the subdivision of culture environments for gametes and embryos. This provides optimized conditions for specific cell types and developmental stages. Such designs are particularly advantageous for modeling multilayered organ structures and investigating interactions between distinct cell populations. In these systems, cellular layers are separated through porous membranes or hydrogel-based micropatterning. For membrane-based compartmentalization, a porous membrane is typically positioned horizontally within the microfluidic device, creating independent microchannels above and below the interface. Porous membranes utilized primarily for cell attachment feature nanoscale pores with diameters of approximately 0.4 μm, which restrict cell migration between channels to maintain compartmentalization [ 120 ]. Thermoplastic membranes composed of PC or polyethylene terephthalate (PET) produced via track etching allow for uniform pore distributions and are characterized by mechanical rigidity [ 121 ]. In the porous membrane-based compartmentalization method, cellular separation occurs vertically, which can limit high resolution imaging due to optical scattering [ 120 ]. Alternatively, PDMS is frequently utilized to ensure optical transparency or to culture tissues requiring cyclic mechanical stimulation, such as the lung or intestine [ 122 ]. Sufficient cell adhesion can be secured through the surface treatment previously discussed in chapter 3.1.1 . Despite this physical separation, secreted factors including proteins, exosomes, and microvesicles can diffuse through the pores, thereby facilitating biochemical communication and paracrine interactions between cell populations in adjacent microchannels [ 123 ]. For instance, Blundell et al. modeled the placental barrier by adhering trophoblasts and endothelial cells to opposite sides of PC membranes with pores 1 μm coated with fibronectin [ 124 ]. By providing independent flow environments in a countercurrent direction between the two channels, they effectively recapitulated the physiological interface between the maternal and fetal circulations. In contrast, compartmentalization based on hydrogel micropatterning is achieved using microfluidic chips that incorporate micropillar arrays or phase guides capable of patterning sol-state hydrogels. These architectures allow precise gel-based patterning of natural ECM hydrogels along with embedded cells, enabling accurate spatial positioning of cells to form multilayered tissues that closely resemble in vivo structures. Furthermore, chemical gradients can be generated within the hydrogel to regulate cell migration and differentiation [ 125 , 126 ]. Unlike the vertical stacking inherent in porous membrane-based methods, hydrogel micropatterning allows for the lateral arrangement of cells on the same focal plane, which facilitates unobstructed microscopic observation and real-time monitoring [ 127 ]. However, this approach presents several technical challenges. Maintaining the structural integrity of the hydrogel during long-term culture remains difficult due to cell-induced gel contraction, which can lead to detachment from the microchannel walls [ 128 ]. Additionally, compared to standardized membranes, lateral configurations complicate trans-epithelial electrical resistance (TEER) measurements because of electrode placement and non-uniform current distribution [ 129 ]. Both porous membrane-based and hydrogel micropatterning-based compartmentalization offer the advantage of not only separating cell populations but also independently supplying culture media optimized for each compartment. This enables the creation of cell-specific microenvironments and overcomes one of the major challenges of conventional 2D co-culture systems, which is the difficulty of maintaining appropriate media conditions for multiple cell types simultaneously. Fluid flow is determined by various fluid parameters, including flow patterns, media recirculation, channel geometry such as width and height, and fluid viscosity. The combination of MPS architecture and fluid parameters directly influences shear stress, which is the mechanical stimulus experienced by cells. Through mechanotransduction, this stimulus affects cell morphology, adhesion, and gene expression [ 130 ]. Furthermore, fluid flow alleviates diffusion limits to ensure adequate nutrient supply. It also facilitates the continuous removal of metabolic waste generated by cellular activity. Fluid flow is commonly implemented by integrating external pumps, such as syringe pumps or peristaltic pumps. Generally, syringe pumps provide continuous flow, whereas peristaltic pumps generate pulsatile flow conditions. However, pump-based systems often suffer from poor throughput and complicated in fluid handling due to the extensive use of tubing [ 131 ]. Consequently, passive flow methods driven by gravity, osmolarity, or surface tension are increasingly used [ 132 ]. While pumpless methods have relative limitations in enabling media recirculation or achieving precise flow control, the specific driving mechanism and flow conditions should be selected based on the intended application. Additionally, pneumatic valve systems can be integrated to achieve automated flow control and complex media delivery. This is advantageous for precisely programming physiological and hormonal cycles within microchannels [ 133 ]. In this regard, microfluidic-based MPS and bioprinting differ significantly in their core priorities. While bioprinting excels at replicating biomimetic tissue architecture through precise spatial control of cell distribution and matrix composition, it faces technical challenges in implementing controlled perfusion within macroscale hydrogel constructs [ 134 ]. As a result, flow distribution and local shear stress remain heterogeneous and difficult to quantify, while transport relying predominantly on diffusion often leads to poorly regulated oxygen and nutrient gradients [ 135 ]. In contrast, microfluidic MPSs are inherently engineered for the quantitative regulation of fluid dynamics, offering predictable laminar flow, microscale control, and reproducible shear stress calculations, alongside the precise generation of spatiotemporal molecular gradients. Accordingly, microfluidics-assisted bioprinting has emerged to address these perfusion control limitations by integrating microfluidic components that enable more reproducible flow delivery and support downstream culture and monitoring [ 136 ]. MPS fluid parameters should strictly be guided by physiological thresholds [ 137 ]. For example, Xie et al. observed that exposing mouse blastocysts to a maximum shear stress of 1.2 dyn/cm 2 resulted in embryo death. This value is ten times higher than the estimated in vivo shear stress of less than 0.12 dyn/cm 2 in the mouse oviduct. This suggests that excessive physical stimuli outside the physiological range can damage embryonic development [ 138 ]. Ferraz et al. designed an oviduct-on-a-chip with a flow rate of 5 μl/h. Mouse embryos experienced an average shear stress of 0.70 ± 0.46 dyn/cm 2 [ 139 ]. Embryos trapped between micropillars due to perfusion were exposed to a maximum shear stress of 2.06 dyn/cm 2 . This led to an observed developmental arrest. It is therefore necessary for most reproductive MPSs to be validated not only for average shear stress, but also for maximum shear stress, duration of exposure, and shear hotspots caused by internal structural elements. Reproductive processes are characterized by time-dependent features, making long-term culture essential for MPS modeling. In the human male reproductive system, spermatogenesis requires approximately 64 days and necessitates the stable maintenance of complex 3D architectures and vascularized stroma niches within the seminiferous tubules [ 140 ]. Similarly, the human female reproductive system undergoes follicle maturation and endometrial remodeling governed by 28-days of the hormonal cycle. Furthermore, the transition from fertilization to implantation takes about one week. These processes involve phased development and continuous microenvironmental changes over an extended period. Thus, it is necessary to design appropriate MPSs capable of recapitulating stable niches for long-term culture and providing the required temporal stimulation. The feasibility of long-term culture is influenced by MPS material properties and the interactions between cells. For instance, steroid hormones, which are essential for reproductive system modeling, frequently suffer from non-specific adsorption to PDMS or the tubing used for media circulation. In addition, the degradation of ECM scaffolds and variations in viscoelasticity over time are primary factors that undermine the structural stability of engineered tissues [ 141 ]. Endometrial stiffness changes from approximately 1.97 to 3.34 kPa during the menstrual cycle, which regulates endometrial epithelial cell phenotypes to ensure appropriate scaffold selection [ 137 ]. It is also important to consider MPS materials in long-term embryonic development studies in order to ensure adequate oxygen transport. Aguilera-Castrejon et al. demonstrated this by adjusting oxygen levels during the ex vivo culture of mouse embryos, maintaining 5% O2 during early gastrulation from embryonic day (E) 5.5 to E7.5, 13% O2 during organogenesis (E7.5 to E9.5), and 21% O2 during advanced organogenesis (E9.5 to E11) [ 142 ]. This indicates that as embryos grow, oxygen demand increases. This necessitates the design of oxygen supply and gas exchange conditions tailored to developmental stages to minimize hypoxia. Beyond these technical limitations, ethical constraints and economic costs determine the practical scope of long-term culture research. In particular, the 14-day rule applied to human embryo research serves as an ethical constraint for investigating post-implantation development in vitro [ 143 ]. Moreover, the continuous perfusion of high-cost growth factors and culture media for in vitro germ cell production imposes a substantial financial burden. As a result, successful reproductive MPS modeling requires a comprehensive approach that ensures chemical inertness, maintains physical properties over time, and adheres to regulatory guidelines. The design of a sex-specific reproductive MPS necessitates the selection of appropriate cell sources and materials. Although the sex of a lineage may be specified, cell lines are susceptible to losing their sex chromosome (XX or XY) configurations or undergoing genetic modifications during long-term culture, which complicates the reliable maintenance of sex identity in the model [ 144 ]. Specifically, female derived cells may exhibit higher expression levels of immune regulatory genes, such as Toll-like receptor 7 (TLR7) and CD40 ligand (CD40L), due to X chromosome inactivation [ 145 , 146 ]. The use of primary cells or patient-derived iPSCs with clearly defined sex and hormonal status is recommended for reliable investigations into sex differences [ 147 ]. Stiffness control is also essential for physiological environment simulation. For instance, 3D testicular organoids cultured in a mixed hydrogel of Matrigel and collagen I demonstrated increased ATP synthesis and testosterone secretion in a higher concentration (1.25 mg/mL) collagen matrix compared to a lower concentration (0.25 mg/mL) environment [ 148 ]. Within the female reproductive system, stiffness plays a pivotal role in follicle maturation and hormone synthesis. Studies using alginate models have shown that matrices with low stiffness promote estrogen production and follicle growth, while matrices with higher stiffness increase the production of progesterone (P 4 ) and androgen [ 149 ]. Mechanical stimuli within reproductive MPS can be engineered to align with the biological sex and specific objectives of each ART stage. In male models, MPS platforms featuring flow countercurrent direction to sperm movement or tortuous channel geometries resembling the oviduct have been reported to actively select high quality sperm by inducing rheotaxis and enhancing motility [ 150 , 151 ]. Conversely, female oocytes and embryos exhibit high sensitivity to shear stress, necessitating protection through precise flow control. Furthermore, reproductive MPS can recapitulate organ-specific movements by implementing quantitative mechanical inputs. A representative example is myometrial peristaltic waves, which increase during the ovulatory phase to facilitate sperm movement [ 152 ]. Elad et al. developed a platform that mimics in vivo peristaltic wall shear stress by integrating a peristaltic pump into a microfluidic chip comprising layered architectures of endometrial epithelial cells and myometrial smooth muscle cells [ 153 ]. Exposure to shear stress at levels lower than those in medium sized blood vessels (<0.05 Pa) for only 60 to 120 min induced significant structural changes, such as increased F-actin expression and enhanced barrier function, not only in the epithelial layer but also in the muscle layer that was not directly exposed to flow. MPS platforms provide an alternative framework to overcome ethical and scientific limitations associated with animal testing and human embryo research. Due to the 14-day rule, human embryo research is legally and strictly restricted with respect to implantation and post-implantation development [ 143 ]. Hence, there is a growing need to utilize MPS to recapitulate early pregnancy processes using stem cell-derived embryo models, such as blastoids, as alternatives to human embryos [ 154 ]. Ethical constraints are not confined to the embryonic stage but extend across the entire developmental process of sex-specific germ cells. For instance, in vitro spermatogenesis or folliculogenesis modeling possesses the potential to generate functional germ cells from donor-derived tissues. This requires rigorous ethical oversight regarding the handling and directed differentiation of male and female germ cells [ 155 ].

Minkyeong Jeong: Data curation, Formal analysis, Writing – original draft. Kiyeon Park: Conceptualization, Writing – original draft. Dawit Jung: Data curation. Sieun Shin: Formal analysis. Jungseub Lee: Formal analysis. Jihoon Ko: Writing – review & editing. Seokyoung Bang: Writing – review & editing. Jungho Ahn: Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.

In vitro production, selection, and maturation of gametes represent core areas of ART, offering new solutions for infertility and subfertility. In these fields, MPSs enable the precise replication of the physiological microenvironments essential for gametogenesis, allowing detailed investigation of gamete formation and functional assessment [ 156 ]. This chapter introduces microengineering-based platforms for sperm and follicle formation, high-quality gamete selection, and gamete maturation. A summary of representative microphysiological models of gametogenesis and gamete selection in the pre-IVF stage is provided in Table 3 . Table 3 Microphysiological models of gametogenesis and gamete selection in Pre-IVF stage. Table 3 Chip Name Biological Stage Key Contribution Technical Challenges Potential Application Scenarios References Testis-on-a-chip Spermatogenesis Long-term ex vivo germ cell differentiation ∙ Maintenance of blood–testis barrier architecture ∙ Hormonal microenvironment control ∙ Long-term viability ∙ Male infertility modeling ∙ Spermatogenic toxicity screening ∙ Endocrine disruption studies [ 157 ] Pumpless spermatogenesis chip Spermatogenesis Enhanced maintenance of the spermatogonial population ∙ Hydrostatic flow control without external pumps ∙ Device fabrication precision ∙ Ex vivo long-term spermatogenesis [ 158 ] Testis-on-a-chip Spermatogenesis Successfully induced in vitro spermatogenesis using mesenchymal stem cells ∙ Unverified fertilization competence ∙ Uncertain long-term culture stability ∙ Pediatric cancer survivor fertility preservation ∙ Therapeutic development [ 159 ] Cervix-inspired tortuous channel chip Selection Improved selection of high-motility, High-integrity sperm ∙ High channel fabrication ∙ Precision requirements ∙ Clinical small-sample sperm selection [ 160 ] Spiral channel-based sperm sorting chip Selection Motility-independent sperm separation ∙ Flow-rate optimization ∙ Stable size-based focusing ∙ TESE/mTESE sperm retrieval ∙ Blood-contaminated sample processing [ 161 ] Ovary-on-a-chip Folliculogenesis Demonstrates that mechanical heterogeneity governs follicle development ∙ Reproducible control of droplet size and spatial organization ∙ Drug screening for ovarian function and toxicity [ 162 ] Follicle-on-a-chip Folliculogenesis Long-term hormone-secreting 3D follicle model ∙ Core–shell microfluidic droplet fabrication control ∙ Ovarian reserve assessment ∙ Follicular toxicity testing [ 163 ] Isolation and denudation chip IVM Enhances oocyte recovery from follicular fluid ∙ Fixed size-cutoff architecture ∙ No real-time oocyte detection ∙ Low-yield IVF optimization ∙ Embryo pool expansion ∙ Cross-clinic workflow standardization [ 164 ] Dynamic IVM chip IVM Low-shear dynamic microfluidic IVM enhancing oocyte development ∙ Precise ultra-low shear control with stable oocyte trapping ∙ Shear-modulated small-volume IVM [ 165 ] Field-deployable microfluidic IVM chip IVM Automated microwell-based COC trapping ∙ Maintaining stable temperature ∙ Efficient COC trapping without laboratory-grade equipment ∙ On-site oocyte collection ∙ Maturation in portable ART workflows [ 133 ] Microphysiological models of gametogenesis and gamete selection in Pre-IVF stage. Maintenance of blood–testis barrier architecture Hormonal microenvironment control Long-term viability Male infertility modeling Spermatogenic toxicity screening Endocrine disruption studies Hydrostatic flow control without external pumps Device fabrication precision Ex vivo long-term spermatogenesis Unverified fertilization competence Uncertain long-term culture stability Pediatric cancer survivor fertility preservation Therapeutic development High channel fabrication Precision requirements Clinical small-sample sperm selection Flow-rate optimization Stable size-based focusing TESE/mTESE sperm retrieval Blood-contaminated sample processing Reproducible control of droplet size and spatial organization Drug screening for ovarian function and toxicity Core–shell microfluidic droplet fabrication control Ovarian reserve assessment Follicular toxicity testing Fixed size-cutoff architecture No real-time oocyte detection Low-yield IVF optimization Embryo pool expansion Cross-clinic workflow standardization Precise ultra-low shear control with stable oocyte trapping Shear-modulated small-volume IVM Maintaining stable temperature Efficient COC trapping without laboratory-grade equipment On-site oocyte collection Maturation in portable ART workflows Spermatogenesis is a complex process involving multiple stages of cell proliferation, differentiation, and maturation. Microfluidic systems allow precise control of fluid flow velocity, facilitating nutrient delivery and waste removal within the culture environment [ 166 ]. By incorporating microfluidic platforms, several studies have sought to improve the differentiation efficiency of germ cells into mature sperm during in vitro spermatogenesis. A representative example is the work by Sharma et al., who applied continuous perfusion using a syringe pump to maintain flow within seminiferous tubule tissue cultured in a testicular MPS [ 157 ]. Continuous perfusion provided gonadotropin stimulation to the testicular tissue, resulting in sustained tissue morphology and the secretion of testosterone and estradiol (E 2 ) necessary for spermatogenic induction. Compared with static cultures, external syringe pump-based perfusion systems demonstrated advantages in maintaining tissue structure, enhancing viability, and promoting hormone responsiveness. However, the use of external syringe pumps in microfluidic applications can limit experimental scalability due to the restricted number of samples that can be processed simultaneously. To overcome this limitation, Komeya et al. developed a spermatogenesis platform that controlled medium flow using hydrostatic pressure and resistance-based microfluidic circuits without reliance on external power or pumping systems ( Fig. 3 A) [ 158 ]. The culture medium flowed under gravity-induced hydrostatic pressure, and the flow rate was adjusted by modifying the length and diameter of the resistance channels. The testicular tissue undergoing spermatogenesis was vertically separated from the microfluidic channel by a porous membrane, allowing oxygen and nutrients to diffuse while facilitating waste removal. This system effectively supported mouse testicular tissue culture and maintained spermatogenesis for over three months, comparable to conventional pump-driven systems. Similarly, Önen et al. utilized hydrostatic pressure to drive a pumpless medium flow rate of 0.5-1.0 μL/min in a microfluidic spermatogenesis chip [ 159 ]. In this platform, the mouse testicular tissue was laterally compartmentalized from the perfusion compartments by arrays of rectangular micropillars, enabling material exchange through the gaps between the pillars. Fig. 3 Examples of MPS for pre-IVF stages. (A) A hydrostatic pressure-driven resistance-based microfluidic circuit system for in vitro sperm production. Reproduced with permission from Ref. [ 158 ], © 2017 Springer Nature. (B) A microencapsulated follicle culture system for in vitro oocyte production. Reproduced with permission from Ref. [ 162 ], © 2014 Elsevier. (C) A 3D printed multi-chamber microfluidic system mimicking the female reproductive tract for sperm selection. Reproduced with permission from Ref. [ 167 ], © 2026 Springer Nature. (D) A microwell microfluidic perfusion system for in vitro maturation and oocyte selection by a 3D printed syringe pump, Reproduced with permission from Ref. [ 168 ], © 2025 Royal Society of Chemistry. Fig. 3 Examples of MPS for pre-IVF stages. (A) A hydrostatic pressure-driven resistance-based microfluidic circuit system for in vitro sperm production. Reproduced with permission from Ref. [ 158 ], © 2017 Springer Nature. (B) A microencapsulated follicle culture system for in vitro oocyte production. Reproduced with permission from Ref. [ 162 ], © 2014 Elsevier. (C) A 3D printed multi-chamber microfluidic system mimicking the female reproductive tract for sperm selection. Reproduced with permission from Ref. [ 167 ], © 2026 Springer Nature. (D) A microwell microfluidic perfusion system for in vitro maturation and oocyte selection by a 3D printed syringe pump, Reproduced with permission from Ref. [ 168 ], © 2025 Royal Society of Chemistry. To isolate non-motile sperm from samples containing impurities such as blood, Son et al. developed a spiral-channel-based sperm sorting system using inertial microfluidic technology [ 169 ]. Unlike conventional motility-based methods, this approach can retrieve non-motile sperm from testicular sperm extraction (TESE) or microdissection testicular sperm extraction (μTESE) samples from patients with azoospermia. The system exploits Dean drag and inertial lift forces generated within the spiral channel to achieve size-based cell separation. Smaller sperm migrate toward the outer wall, while larger red blood cells are directed toward the inner wall, allowing sperm isolation from samples obtained from patients with severely impaired fertility. There are several barriers to clinical translation. The platform consists of two syringe pumps and six syringes that must be operated to balance injection and withdrawal flows, increasing complexity. Moreover, it exhibits a dead volume of ∼0.2 mL per 1 mL injection, which can cause a substantial loss of the already scarce sperm present in (μ)TESE samples. In clinical scenarios with rare sperm, retention within the device and tubing can undermine the clinical utility of the entire sorting process. While the platform quantified enrichment performance at each outlet, it did not provide clinically relevant evaluations of sperm quality such as the DNA fragmentation index (DFI) and reactive oxygen species (ROS) levels. Alternatively, geometry-driven designs have been explored to enrich progressively motile sperm by mimicking key transport features of the female reproductive tract. Dadkhah et al. developed a curved microfluidic chip designed to mimic the anatomical structure of the cervix [ 161 ]. The chip design was inspired by the natural migration of sperm through the cervix and oviduct, where only highly motile sperm can reach the oocyte after navigating tortuous paths. Within this device, the flow is directed opposite to the forward movement of the sperm, allowing separation based on progressive motility. Sperm with low motility are pushed backward or stagnate under the counterflow, whereas highly motile sperm track along the curved channel walls and reach the outlet efficiently. This platform allows the direct use of raw semen samples without preprocessing, and its parallel configuration of 85 microchannels enables the high-throughput processing of 0.5 mL of semen within 20 min. Despite these advancements, the overall workflow requires 65 min of preparatory steps. This includes 20 min of vacuum degassing to prevent bubble formation and 45 min of pre-incubation for thermal and pH equilibration. These steps increase the total turnaround time and may reduce practical efficiency in time-sensitive clinical settings. Also, quantitative comparisons of the device's flow rate and shear stress with physiological values in the human oviduct remain limited. Building on this concept, multistage platforms combine several tract like barriers within one integrated system to further enrich sperm with higher genetic integrity. Dai et al. developed a 3D printed sorting system that sequentially recapitulates anatomical barriers of the female reproductive tract with multiple chambers ( Fig. 3 C) [ 167 ]. In this platform, sperm first undergo swim-up, then pass through a porous membrane mimicking cervical penetration, migrate along curved microstructures, and are collected only after traversing a mucus layer that models cumulus penetration. The DFI was assessed using the sperm chromatin structure assay (SCSA). In a patient cohort with a high mean DFI of 41.93%, the system yielded a DFI of 0.71%. For the same samples, conventional swim-up and a commercial membrane-based selection device yielded mean DFI values of 5.39% and 5.25%, respectively, indicating improved genetic integrity. However, the stringent multistep selection results in lower sperm recovery than conventional methods. In vitro follicle culture is a valuable reproductive technology for cancer patients at risk of losing future fertility due to ovarian damage caused by chemotherapy or radiotherapy. This technique induces the maturation of immature follicles in vitro to produce fertilizable oocytes [ 170 ]. Successful follicle culture requires replication of the 3D ECM environment found in vivo , along with efficient hormonal and nutrient supply essential for follicular development. However, conventional 2D culture systems fail to recapitulate the native biomechanical and structural properties of ovarian follicles. For example, in 2D follicle cultures using ECM protein-coated membranes, theca and granulosa cells spread laterally rather than forming spherical 3D aggregates, limiting the establishment of physiologically relevant architecture [ 171 ]. Moreover, folliculogenesis depends heavily on hormones and growth factors. However, essential paracrine and autocrine factors secreted by theca and granulosa cells are often diluted in large culture volumes, disrupting the biochemical balance required for proper oocyte maturation [ 172 ]. While such approaches emphasize the regulation of mechanical stiffness, equal attention needs to be directed toward reconstructing physiologically relevant cellular organization. To address these challenges, Choi et al. developed a microfluidic device capable of encapsulating follicles within a two-layered hydrogel structure consisting of a 2% alginate shell and a 0.5% collagen core to mimic the stiffness gradient of ovarian tissue ( Fig. 3 B) [ 162 ]. This microcapsule system effectively promoted follicular development to the antral stage and induced cumulus-oocyte complex (COC) release even in the absence of luteinizing hormone (LH) and EGF, demonstrating that mechanical stiffness differences within the ovarian stroma play a crucial role in follicle growth and ovulation. Healy et al. further designed a 3D co-culture system that replicated the cellular organization of follicles by embedding murine theca cells in a 2% alginate shell and murine granulosa cells in a 0.8 mg/mL collagen core containing 0.05% alginate core [ 163 ]. Using a microfluidic encapsulation process, uniform localization of oocytes at the center of the two-layer capsules was achieved. During 27 days of culture, stable secretion of E 2 , P 4 , and androstenedione was observed, indicating active endocrine function of the theca and granulosa cells. These findings confirm that 3D capsule systems provide structural support and promote hormone production during follicular development. IVM is the process of inducing immature germinal vesicle (GV) stage oocytes to reach the metaphase II (MII) stage. It can be performed without ovarian stimulation and is therefore beneficial for patients at risk of OHSS or those with PCOS [ 173 ]. Conventional IVM involves culturing COCs in petri dishes. Under these conditions, the cumulus cell layer does not fully expand and exhibits a morphology that differs from that of in vivo matured oocytes. [ 174 ]. The 3D organization of the COCs and enhanced nuclear maturation, mitochondrial function, and the expression of maternal effect genes. In vivo , ovulated oocytes and early embryos are constantly exposed to gentle fluid movement and mechanical stimuli within the oviduct. However, such dynamic conditions are not reproduced in conventional static culture systems. Moreover, ovulated oocytes and preimplantation embryos exist in a suspended state without direct vascular supply and are subjected to mechanical stimuli from ciliary motion and muscular contractions, conditions that are not replicated in static culture systems. To overcome these limitations, microfluidic devices have been designed with a partially constricted channel that mimics peristaltic muscle contraction, resulting in significantly higher 8-cell development rates compared to straight-channel culture, which suggests that dynamic physical stimulation is vital for embryo development beyond static conditions [ 175 , 176 ]. To better mimic the dynamic in vivo environment, microfluidic-based IVM systems have been proposed. Oskouei et al. developed a microfluidic device that allowed dynamic culture of immature oocytes and demonstrated that this approach protected oocytes from oxidative damage caused by ROS during maturation [ 165 ]. Compared with static culture in Petri dishes, oocytes matured under dynamic microfluidic conditions showed higher MII maturation rates, fertilization rates, blastocyst formation rates, as well as increased glutathione content. These results are based on in vitro measures such as maturation, fertilization rate, blastocyst development, and glutathione (GSH) levels. However, establishing translational relevance will require evaluation of pregnancy outcomes. Meanwhile, Franko and Ferraz designed a microfluidic chip featuring 135 microwells to capture bovine oocytes directly from follicular fluid (FF) and support their maturation under continuous perfusion at a rate of 20 μL/h ( Fig. 3 D) [ 168 ]. Perfusion was driven by a custom 3D printed syringe pump. The device demonstrated a high COC capture efficiency of 87.7 ± 4.4% and includes an inlet serving as both an oocyte-loading site and a medium reservoir, with an outlet connected to a syringe pump that allows for automated collection and a post-maturation recovery rate of 91.7 ± 11.5%. Under perfused conditions, oocytes exhibited improved chromosomal alignment and reduced spindle abnormalities compared with those matured under static conditions, highlighting the role of dynamic culture in enhancing cytogenetic stability and oocyte quality. Although the platform's efficacy has been demonstrated in bovine models, human clinical validation is yet to be established. Also, pump precision can vary between units, necessitating individual flow-rate calibration for each unit to ensure consistency.

Although implantation and early placentation take place in vivo after embryo transfer, they ultimately determine whether assisted reproductive technologies lead to a sustained pregnancy. The post-IVF stage involves a tightly coordinated sequence of events, including embryo-endometrium interaction, decidual transformation, vascular remodeling, and maternal immune adaptation. Together, these processes establish the conditions required for successful implantation and early placental development. Even when morphologically high-quality embryos are transferred, implantation failure remains common, with per-cycle success rates typically reported at 30–40% [ 192 ]. This limited efficiency reflects more than embryo competence alone. Endometrial receptivity, developmental timing, and synchronization between the embryo and uterine environment all contribute to the final outcome. For this reason, the post-IVF period cannot be viewed simply as the continuation of embryo culture. It represents a biologically decisive phase that warrants dedicated mechanistic investigation. Recent MPSs have begun to address this challenge by reconstructing key elements of the implantation microenvironment and incorporating endocrine, vascular, and immune components within controlled platforms. The following chapters outline advances in implantation, placenta, and immune-integrated maternal-fetal interface models, with an emphasis on how these systems are reshaping the study of post-IVF biology in reproductive medicine. We summarize the sequential cellular and microphysiological events spanning the IVF stage through the post-IVF stage in Table 4 . Table 4 Sequential cellular and microphysiological events from IVF stage to Post-IVF stage. Table 4 Chip Name Biological Stage Key Contribution Technical Challenges Potential Application Scenarios References Artificial uterus on a microfluidic chip Implantation Development of a perfusion-based endometrium–embryo co-culture platform ∙ Multilayer integration and stable bonding ∙ Shear stress control during perfusion ∙ Oocyte capture microstructure design ∙ ART embryo quality improvement platform ∙ Implantation mechanism research tool [ 193 ] Human implantation-on-a-chip Implantation Reconstruction of human EVT–maternal endothelial interactions ∙ Diffusion-mediated signaling control ∙ Precise implementation of 3D compartmentalization ∙ Optimization and batch consistency of ECM stiffness ∙ Studying human implantation mechanisms ∙ Personalized implantation modeling [ 77 ] Endometrium-on-a-chip Implantation Patient-specific platform to predict endometrial receptivity ∙ Incomplete immune microenvironment ∙ Patient-specific assessment of endometrial receptivity ∙ Preclinical drug screening for the restoration of impaired receptivity ∙ Personalized therapeutic stratification in infertility patients [ 79 ] Endometrium-on-a-chip Early endometrial changes Identifies endothelial shear stress–induced prostaglandin signaling ∙ Lack of embryo-derived signals ∙ Limited multicellular complexity ∙ Static hormonal modeling without full menstrual cycle dynamics ∙ Patient-derived testing platform for personalized assessment ∙ Screening tool for drugs modulating prostaglandin signaling [ 194 ] Human implantation-on-a-chip Immune adaption Demonstrates that type I IFN signaling disrupts EVT differentiation trajectory ∙ Lack of immune and stromal components ∙ Limited dose-response and temporal resolution of IFN exposure ∙ Absence of physiological vascular flow and shear stress. ∙ Studying immune-driven implantation dysfunction ∙ Enabling patient-specific implantation risk assessment [ 164 ] Microfluidic EVT chemotaxis platform Immune adaption Quantitative 3D microfluidic for immune-regulated primary EVT chemotaxis ∙ Matrigel-based ECM limitations ∙ Stable microfluidic gradient generation ∙ Quantitative 3D migration analysis with limited primary EVT availability ∙ Screening of cytokine-mediated placentation disorders ∙ Mechanistic investigation of maternal–fetal immune crosstalk [ 193 ] Choriodecidual interface-on-a-chip Immune adaption Modeling infection-driven maternal–fetal immune crosstalk ∙ Lacking full decidual immune heterogeneity ∙ Absence of physiological flow or oxygen gradient control ∙ Pregnancy-related infection and inflammation modeling ∙ Preclinical screening platform for anti-inflammatory therapeutics [ 195 ] Sequential cellular and microphysiological events from IVF stage to Post-IVF stage. Multilayer integration and stable bonding Shear stress control during perfusion Oocyte capture microstructure design ART embryo quality improvement platform Implantation mechanism research tool Diffusion-mediated signaling control Precise implementation of 3D compartmentalization Optimization and batch consistency of ECM stiffness Studying human implantation mechanisms Personalized implantation modeling Incomplete immune microenvironment Patient-specific assessment of endometrial receptivity Preclinical drug screening for the restoration of impaired receptivity Personalized therapeutic stratification in infertility patients Lack of embryo-derived signals Limited multicellular complexity Static hormonal modeling without full menstrual cycle dynamics Patient-derived testing platform for personalized assessment Screening tool for drugs modulating prostaglandin signaling Lack of immune and stromal components Limited dose-response and temporal resolution of IFN exposure Absence of physiological vascular flow and shear stress. Studying immune-driven implantation dysfunction Enabling patient-specific implantation risk assessment Matrigel-based ECM limitations Stable microfluidic gradient generation Quantitative 3D migration analysis with limited primary EVT availability Screening of cytokine-mediated placentation disorders Mechanistic investigation of maternal–fetal immune crosstalk Lacking full decidual immune heterogeneity Absence of physiological flow or oxygen gradient control Pregnancy-related infection and inflammation modeling Preclinical screening platform for anti-inflammatory therapeutics Implantation represents one of the most biologically complex and clinically decisive stages in assisted reproduction. Following embryo transfer, successful implantation requires coordinated trophoblast adhesion, lineage commitment, and controlled invasion into the decidua. The process is coupled with ECM remodeling, angiogenesis and spiral artery modification, and tightly regulated spatiotemporal immune signaling at the maternal-fetal interface [ 196 , 197 ]. Early microfluidic embryo culture systems aimed to recreate uterine microenvironments by evolving from epithelial-stromal co-culture models to dynamic multi-compartment platforms. 3D ECM-based platforms have enabled quantitative analysis of trophoblast invasion. A dextran methacrylate (DexMA) and PEG hydrogel chip quantified trophoblast invasion and profiled vascular associated gene programs. In this system, embryos were positioned within 100 to 150 μm of endothelial cells to visualize trophoblast endothelium interactions in real-time [ 198 ]. However, these platforms were developed using murine models, and species-specific differences in placentation and trophoblast invasion constrain their direct translational relevance to human implantation. To address this translational gap, a subsequent human derived platform was engineered to investigate the influence of pre-implantation maternal immune cells on the invasive behavior of human EVTs. The device recapitulated the 3D architecture of the maternal-fetal interface, with a central channel filled with a collagen-Matrigel composite hydrogel to reconstruct the decidual ECM of the maternal endometrium. Although much of the current understanding of maternal cellular function during early placentation has been derived from studies using first-trimester tissues, this system was designed based on the premise that the earliest stages of EVT invasion are initiated in response to a pre-established uterine microenvironment prior to embryo arrival. Thus, the platform prioritized reconstruction of the native maternal niche to enable mechanistic interrogation of immune-trophoblast crosstalk. Beyond invasion-focused studies, MPSs have been used to model the implantation process. Blastoid-based implantation assays quantify embryo attachment efficiency under shear flow and spatial constraints, enabling controlled assessment of early implantation dynamics [ 199 ]. A 3D endometrial model was constructed using a collagen and fibronectin hydrogel scaffold in a transwell insert. This approach reconstructed luminal, glandular, and stromal compartments, establishing an in vitro model of the receptive endometrium. Human blastoids derived from ESCs were subsequently transferred onto the engineered endometrial scaffold to evaluate implantation potential. The system reproduced early implantation and initial placental development, accompanied by stage-specific secretion of human placental lactogen (hPL), pregnancy-associated plasma protein A (PAPP-A), and placental growth factor (PlGF). Despite these advances, the models lacked endometrial vasculature. This limitation restricts their ability to recapitulate vascular functions that regulate implantation, including angiogenic remodeling and endothelial stromal signaling that influences endometrial receptivity and trophoblast invasion. A patient-derived vascularized human endometrium MPS has been reported to reconstruct the spatial organization and temporal dynamics of native endometrial tissue, integrating multicellular architecture with functional biomarker readouts under controlled hormonal and mechanical regulation ( Fig. 5 A) [ 79 ]. This system enables implementation of an endometrial receptivity scoring system (ERS 2 ), offering a potential tool for personalized assessment of implantation competence and therapeutic monitoring in infertility care. To better recapitulate pre-implantation states, hESC-derived embryoid bodies were integrated into the platform. The embryoid bodies exhibited reproducible invasion-like behavior within the engineered endometrial microenvironment, supporting their utility as stem cell-based surrogates for early implantation studies. Fig. 5 Examples of MPS for post-IVF stages. (A) A patient-derived vascularized human endometrium MPS for modeling endometrial receptivity and implantation like invasion. Reproduced with permission from Ref. [ 79 ], © 2025 Springer Nature. (B) A perfused endometrium-on-a-chip that models vascular shear-driven prostaglandin signaling and embryo-derived factor-mediated modulation of endometrial remodeling and decidualization. Reproduced with permission from Ref. [ 200 ], © 2019 Oxford University Press. (C) A human implantation-on-a-chip model for interrogating type I IFN–mediated regulation of EVT invasion across an ECM barrier toward maternal endothelial cells. Reproduced with permission from Ref. [ 201 ], © 2025 Elsevier. Fig. 5 Examples of MPS for post-IVF stages. (A) A patient-derived vascularized human endometrium MPS for modeling endometrial receptivity and implantation like invasion. Reproduced with permission from Ref. [ 79 ], © 2025 Springer Nature. (B) A perfused endometrium-on-a-chip that models vascular shear-driven prostaglandin signaling and embryo-derived factor-mediated modulation of endometrial remodeling and decidualization. Reproduced with permission from Ref. [ 200 ], © 2019 Oxford University Press. (C) A human implantation-on-a-chip model for interrogating type I IFN–mediated regulation of EVT invasion across an ECM barrier toward maternal endothelial cells. Reproduced with permission from Ref. [ 201 ], © 2025 Elsevier. Complementary approaches combine human blastoids with endometrial organoids to model attachment and invasion. In a recent 3D implantation model, a matrix metalloproteinases (MMP)-degradable hydrogel matrix combined with Matrigel supported invasion while mimicking ECM remodeling during implantation [ 202 ]. Using organoids derived from patients with recurrent implantation failure (RIF), the study reported significantly reduced blastoid attachment and developmental progression compared with controls. This platform captured sequential stages of implantation, including initial contact, adhesion, invasion, and early post-attachment development. However, the structural complexity and individualized organoid preparation limit scalability and present challenges for high-throughput screening. Although scalability remains limited, the platform enables patient-specific functional analysis from a single endometrial biopsy. Given the heterogeneous and individualized nature of RIF, such personalized platforms highlight the necessity of tailored therapeutic strategies and underscore the translational potential of endometrium-integrated implantation models. In parallel, recent comprehensive analyses of female reproductive MPSs emphasize the integration of vascular networks, immune components, and multi-organ connectivity. Collectuvely, these advances represent a progression from isolated mechanistic modeling toward translationally oriented systems that bridge implantation biology with clinical application. Immediately after embryo implantation, the maternal endometrium undergoes rapid structural and functional transformations through the process of decidualization, creating an environment suitable for the maintenance of pregnancy [ 194 ]. During this period, the endometrium is precisely regulated by hormonal cues, hemodynamic stimuli, and signaling molecules secreted by the embryo. To reproduce these complex processes in vitro , MPSs have been increasingly utilized. Microfluidic endometrium-on-a-chip systems have provided important mechanistic insights into maternal regulation during early implantation ( Fig. 5 B) [ 200 ]. In this microfluidic model, continuous perfusion was applied to mimic uterine blood flow–associated shear stress, revealing that endothelial cells respond by increasing secretion of prostaglandin E 2 (PGE 2 ) and prostacyclin (PGI 2 ). These vascular mediators subsequently promoted stromal decidualization, as evidenced by elevated expression of prolactin (PRL) and insulin-like growth factor binding protein-1 (IGFBP-1). This study highlighted the contribution of endogenous mechanobiological cues and vascular secretory activity to endometrial remodeling in early pregnancy. Complementary findings from a 3D bovine endometrium-on-a-chip model demonstrated that exposure to embryo-conditioned medium or purified embryo-derived proteins, including macrophage capping protein (CAPG) and protein disulfide isomerase (PDI), altered the transcriptomic and secretory responses of endometrial cells [ 203 ]. Activation of immune modulation pathways, interferon-stimulated gene expression, and prostaglandin synthesis underscored the critical role of embryo-derived signaling in implantation stability. Collectively, these complementary models of maternal hemodynamic regulation and embryo-derived signaling establish a more comprehensive representation of early endometrial adaptation. By reconstructing immune-vascular crosstalk, endocrine responsiveness, and embryo–maternal communication within a controlled microenvironment, contemporary endometrium-on-a-chip platforms provide powerful experimental tools for investigating implantation failure, early pregnancy loss, and mechanisms underlying ART outcomes. During early pregnancy, the maternal immune system undergoes coordinated adaptation that permits tolerance toward the semi-allogeneic fetus while preserving host defense [ 195 ]. At the maternal-fetal interface, reciprocal interactions among decidual stromal cells, EVTs, and specialized immune populations, particularly decidual natural killer (dNK) cells and macrophages, regulate placentation. Rather than functioning as cytotoxic effectors, dNK cells secrete cytokines and growth factors that modulate trophoblast invasion and spiral artery remodeling. Recent MPSs have enabled dynamic reconstruction of these immune–trophoblast interactions. A 3D microfluidic invasion platform quantified primary human EVT migration under defined chemokine gradients [ 193 ]. EVTs embedded in a hydrogel matrix responded to cytokines produced by activated dNK cells, with granulocyte-macrophage colony-stimulating factor (GM-CSF) significantly enhancing trophoblast directionality and motility. Activation of dNK cells increased GM-CSF secretion and trophoblast invasion, while GM-CSF neutralization attenuated this migratory response, demonstrating functional immune regulation of trophoblast behavior. More complex immune-integrated placental interface chips replicate the layered choriodecidual structure [ 204 ]. These platforms separate maternal and fetal compartments while permitting paracrine signaling and immune cell trafficking, enabling investigation of donor-specific and allogeneic immune interactions. Controlled levels of interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α) together with regulatory cytokines such as interleukin-10 (IL-10) sustain physiological migration, whereas excessive inflammatory signaling reduces invasion and is associated with pregnancy complications including preeclampsia and preterm birth. Aberrant elevation of type I interferon (IFN) signaling at the time of embryo transfer has been associated with pathological placentation and adverse pregnancy outcomes. To functionally dissect this relationship, a human implantation-on-a-chip was employed in which EVTs were engineered to migrate across an ECM barrier toward maternal endothelial cells ( Fig. 5 C) [ 201 ]. Amplification of IFN signaling within the device markedly restricted EVT invasion and suppressed the emergence of invasive EVT subpopulations. Molecular analyses further demonstrated that sustained IFN stimulation disrupted the EMT-associated transcriptional program in EVTs and broadly downregulated stromal and mesenchymal gene expression signatures. Although IFN signaling was experimentally modulated in a reductionist manner, the decidual immune microenvironment in vivo is shaped by complex interactions among diverse immune cell populations. Accordingly, immune cell sourcing remains a key translational challenge. Primary decidual immune cells provide physiological relevance but are limited by restricted availability, donor heterogeneity, gestational variability, and potential activation drift during ex vivo culture. iPSC-derived macrophages offer scalability and standardization. However, their ability to reproduce tissue-imprinted transcriptional and functional states of decidual macrophages remains under investigation [ 205 , 206 ]. The decidual immune niche extends beyond the conventional MI and MII framework, raising concerns regarding functional fidelity in simplified in vitro differentiation systems. Rigorous benchmarking, including single-cell transcriptomic comparison, spatial immune profiling, and quantitative cytokine analysis, will be required to validate immune-integrated placental MPS platforms and ensure reliable modeling of post-IVF immune adaptation.

ART has become an integral part of modern reproductive medicine by addressing diverse infertility etiologies, improving pregnancy outcomes, and enabling fertility preservation. To provide a biological and clinical context for the reproductive MPS models discussed in this chapter, we outline the ART workflow in three sequential stages: pre-IVF, IVF, and post-IVF ( Fig. 1 ). Fig. 1 Schematic illustration of different ART steps and toxicity assessment models along the human reproductive development process. The figure depicts the overall human reproductive development process from gametes to the zygote and embryo stages. ART is categorized into three stages along the biological and clinical timeline: pre-IVF, IVF, post-IVF. Toxicity assessment is included separately to highlight MPS-based evaluation of reproductive toxicity across ART-relevant contexts. (A) Male pre-IVF models cover in vitro sperm production and sperm selection. (B) Female pre-IVF models include in vitro oocyte production, IVM, and oocyte selection. (C) IVF models correspond to the fertilization step. (D) Post-IVF models extend to embryo implantation and subsequent embryonic development. (E) Toxicity assessment models encompass reproductive and developmental toxicity studies, including body-on-a-chip incorporating the reproductive system. Fig. 1 Schematic illustration of different ART steps and toxicity assessment models along the human reproductive development process. The figure depicts the overall human reproductive development process from gametes to the zygote and embryo stages. ART is categorized into three stages along the biological and clinical timeline: pre-IVF, IVF, post-IVF. Toxicity assessment is included separately to highlight MPS-based evaluation of reproductive toxicity across ART-relevant contexts. (A) Male pre-IVF models cover in vitro sperm production and sperm selection. (B) Female pre-IVF models include in vitro oocyte production, IVM, and oocyte selection. (C) IVF models correspond to the fertilization step. (D) Post-IVF models extend to embryo implantation and subsequent embryonic development. (E) Toxicity assessment models encompass reproductive and developmental toxicity studies, including body-on-a-chip incorporating the reproductive system. The causes of infertility are diverse, including ovulatory dysfunction, fallopian tube obstruction, uterine abnormalities, sperm defects, and idiopathic factors [ [24] , [25] , [26] ]. To effectively address infertility, pharmacological therapy, surgical intervention, and ART should be appropriately integrated. Pharmacological treatments have limitations due to potential alterations in the menstrual cycle and endocrine system, while surgical approaches, such as the removal of endometriosis or uterine fibroids, do not always lead to an immediate increase in conception rates and may be associated with recurrence or persistent infertility [ [27] , [28] , [29] , [30] , [31] , [32] ]. When ART is incorporated into infertility treatment, such as intracytoplasmic sperm injection (ICSI), the chances of conceiving through IVF are significantly enhanced [ 33 , 34 ]. ART refers to a group of medical interventions developed to overcome infertility and improve reproductive success. Representative techniques include IVF, ICSI, cryopreservation of gametes and embryos, and assisted hatching [ 35 ]. IVF involves the fertilization of oocytes with sperm outside the human body, followed by the transfer of the resulting embryos into the uterus. The embryos obtained during this process can be cryopreserved and reused if pregnancy does not occur after the initial transfer or when additional pregnancies are desired in the future. Moreover, preimplantation genetic diagnosis can be performed prior to embryo transfer to select embryos without genetic abnormalities, thereby improving implantation success and supporting pregnancy safety [ 36 ]. ICSI differs from IVF in that a single sperm is directly injected into the cytoplasm of an oocyte to induce fertilization. Because this method requires only a minimal number of sperm, it is particularly effective for cases of male infertility characterized by low sperm count or poor motility [ 37 , 38 ]. The cryopreservation of gametes and embryos also serves as an essential approach for fertility preservation. This technique offers an opportunity to preserve reproductive potential for patients at risk of reproductive damage from treatments such as chemotherapy, as well as individuals who wish to postpone childbirth for social or personal reasons. [ 39 , 40 ]. Successful implantation requires the embryo to escape from the surrounding zona pellucida. However, in older women or patients who have experienced repeated implantation failure, the zona pellucida may become thickened, impeding this process. To address this issue, assisted hatching techniques using laser, microneedle, or chemical methods are applied to thin or partially remove the zona pellucida, thereby enhancing implantation rates [ 41 , 42 ]. Overall, ART not only plays a crucial role in overcoming various causes of infertility but also contributes to fertility preservation and improved pregnancy outcomes. With ongoing technological innovation and clinical refinement, ART continues to advance as a cornerstone of modern reproductive medicine. Before describing the established models of reproductive system MPS, the ART process that reproductive MPSs are designed to recapitulate is outlined in three stages: pre-IVF, IVF, and post-IVF. In vitro gametogenesis (IVG) is a technology that enables the generation of gametes, including sperm and oocytes, outside the human body by culturing cells derived primarily from induced pluripotent stem cells (iPSCs) or embryonic stem cells (ESCs) [ 43 ]. In mouse studies, functional oocytes produced through IVG have been successfully fertilized with sperm to generate viable offspring, but research on its application to humans remains in an early stage [ 44 ]. Gamete selection refers to the process of identifying and isolating high quality gametes that meet specific criteria for generating healthy embryos in ART. The quality of sperm and oocytes critically influences successful fertilization and subsequent embryonic development [ 44 ]. Therefore, selecting viable and functionally competent gametes prior to IVF is essential. During gamete selection, sperm are evaluated for motility, morphology, and DNA integrity to identify the healthiest candidates [ 45 , 46 ]. Mature oocytes are chosen by assessing parameters such as size, morphology, and cytoplasmic characteristics [ 47 ]. Through this selection process, embryos with the highest fertilization potential can be generated. In vitro maturation (IVM) is a process that enables immature oocytes to mature outside the body, making them suitable for fertilization. In conventional IVF, hormonal stimulation using gonadotropins is required to induce ovulation, whereas IVM allows oocyte maturation without such stimulation. In this approach, immature oocytes are retrieved during the early follicular phase and subsequently cultured under controlled conditions to induce maturation [ 48 ]. During oocyte maturation, cumulus cells are maintained and the oocytes are cultured in media supplemented with maturation-promoting growth factors such as epidermal growth factor (EGF), human chorionic gonadotropin (hCG), and follicle stimulating hormone (FSH), thereby promoting both nuclear and cytoplasmic maturation and improving fertilization competence [ 49 ]. For patients with polycystic ovary syndrome (PCOS), who are at high risk of ovarian hyperstimulation syndrome (OHSS) during conventional ovulation induction, IVM offers a safer alternative by minimizing hormonal exposure [ 50 ]. Jurema et al. reported that IVM can improve safety and cost-effectiveness while maintaining near-physiological sex hormone levels in women at risk of hyperstimulation, including those with PCOS [ 51 ]. Early-stage rescue IVM initially produced oocytes of suboptimal quality, but subsequent approaches that cultured oocytes collected from small antral follicles measuring two to 5 mm together with their companion cumulus cells improved the synchrony between nuclear and cytoplasmic maturation, leading to higher implantation rates. For sperm activation, sperm with reduced motility or abnormal morphology often fail to penetrate the oocyte under natural conditions, necessitating capacitation to enhance their fertilization ability [ 52 ]. When sperm count is extremely low or motility is severely compromised, spontaneous penetration of the oocyte may not occur. In such cases, ICSI is employed to directly introduce a single sperm into the cytoplasm of an oocyte, thereby enabling fertilization [ 53 ]. This technique is widely used for the treatment of male-factor infertility. To achieve fertilization in vitro , particularly in cases of male infertility, ICSI is commonly employed. This technique involves the direct injection of a single, preselected sperm into the cytoplasm of a mature oocyte [ 54 , 55 ]. The success of ICSI largely depends on selecting sperm of high quality from the semen sample. In current clinical practice, semen samples are processed through centrifugation in specialized media to separate motile sperm from debris and nonviable cells [ 56 ]. Embryologists then manually select a single sperm using a glass micropipette and inject it directly into the oocyte [ 57 ]. Recent studies have explored sperm selection methods based on microfluidic technologies, which utilize microscale fluid mechanics to isolate sperm with superior motility. In these systems, semen is introduced into a microchannel where physical and chemical cues such as inertia, chemotaxis, thermotaxis, and acoustic wave propagation enable the selective migration of motile sperm [ 58 ]. However, because ICSI requires the creation of a microscopic opening in the oocyte membrane for sperm injection, this procedure may induce oocyte degeneration and potentially compromise the genetic stability of the resulting embryo [ 59 ]. Furthermore, as sperm selection during ICSI is performed manually by technicians based primarily on morphology, studies have reported a higher risk of autism spectrum disorder and developmental delay in offspring conceived through this method compared with natural conception [ 60 ]. Therefore, there is a growing need for ART that enables sperm to autonomously interact and fuse with oocytes without human intervention, a goal that can be achieved through the application of microfluidic technology [ 61 ]. In humans, the embryo develops into a blastocyst within five to six days after fertilization and implants into the endometrium approximately one week later [ 62 , 63 ]. During the first three days following fertilization, the embryo undergoes three cleavage divisions and begins to organize into a compact cell cluster. By the fourth day, compaction occurs and internalized cells emerge, forming the morula stage, which subsequently gives rise to the inner cell mass (ICM). The outer epithelial layer differentiates into the trophectoderm, while the accumulation of fluid leads to the development of the blastocoel, and the ICM becomes localized at the embryonic pole. The formation of the blastocyst is regulated by endometrial differentiation controlled by ovarian steroid hormones, which supports coordinated embryonic maturation [ 64 ]. After implantation, characteristic morphological changes occur sequentially, including apical-basal polarization of the epiblast, lumenogenesis, formation of the pro-amniotic cavity, and trophoblast diversification [ 65 , 66 ]. In ART, embryos fertilized in vitro are typically cultured for three to five days before being transferred into the maternal endometrium [ 67 ]. Due to ethical constraints, research on human embryo culture and development in vitro is limited and rarely performed beyond clinical contexts [ 68 ]. Clinically, improving embryo developmental success, defined as the proportion of fertilized embryos that reach a transferable blastocyst stage, is essential for achieving successful implantation and live birth. This is because blastocyst-stage transfers often result in higher implantation rates than those performed at earlier cleavage stages. Additionally, the larger size of the blastocyst reduces the likelihood of ectopic pregnancy and facilitates clinical decision-making regarding embryo transfer and cryopreservation [ 69 ]. Therefore, although embryo loss within the uterine environment cannot be completely excluded, improving the quality and developmental potential of embryos generated through ART is directly linked to higher rates of implantation, pregnancy, and live birth.

The rapid evolution of MPS has established a transformative foundation for ART, bridging the gap between traditional in vitro models and the intricate physiological complexity of human reproduction. The future of the field now lies in the seamless convergence of bioengineering, digital sensing, and AI to enhance the personalization, objectivity and predictability of reproductive care. This concluding chapter provides a strategic roadmap for the development of ART-based MPS, categorized into short-term, mid-term, and long-term milestones ( Fig. 7 ). These milestones collectively define a staged progression from technology-driven analysis to translational standardization and ultimately to scalable, functionally mature systems with regenerative potential. The following chapters delineate the scientific, technical, and regulatory requirements necessary at each stage to enable the responsible and clinically meaningful deployment of MPS for ART. Fig. 7 Roadmap for Integrating Reproductive MPS into Future ART Workflows. Reproductive MPS perspectives are organized into three phases. In the short-term goal, platforms prioritize multimodal data acquisition and AI-assisted decision support. In the mid-term goal, standardization and benchmarking enable integration into ART workflows with clinically relevant performance metrics, including gamete yield, implantation, and live birth linked outcomes. In the long-term goal, advances in tissue engineering and manufacturing are envisioned to support therapeutic organ-scale constructs and, ultimately, artificial reproductive organs. Fig. 7 Roadmap for Integrating Reproductive MPS into Future ART Workflows. Reproductive MPS perspectives are organized into three phases. In the short-term goal, platforms prioritize multimodal data acquisition and AI-assisted decision support. In the mid-term goal, standardization and benchmarking enable integration into ART workflows with clinically relevant performance metrics, including gamete yield, implantation, and live birth linked outcomes. In the long-term goal, advances in tissue engineering and manufacturing are envisioned to support therapeutic organ-scale constructs and, ultimately, artificial reproductive organs. The integration of MPS and digital sensors enables real-time biological data collection. This provides a foundation for monitoring and digitizing subtle changes in reproductive organs, germ cells, and embryos, as well as variations in their surrounding environment [ 232 , 233 ]. In particular, biosensors embedded within microfluidic-based MPSs allow analysis using extremely small sample volumes [ 234 ]. Conventional embryo quality assessment and genetic aneuploidy assessments such as embryo biopsy are problematic because they are invasive and can potentially damage the embryo [ 235 , 236 ]. As an alternative, sensor integrated methods have emerged to evaluate embryo quality in a non-invasive manner. Obeidat et al. developed an electrochemical-based multisensor system capable of simultaneously measuring mitochondrial respiration and glycolysis in bovine embryos in real-time [ 237 ]. By integrating dissolved oxygen, glucose, and lactate sensors into a microchamber, they non-invasively captured the metabolic shift toward lactate production as embryonic development progressed. Vargas-Ordaz et al. implemented a light-sheet-on-a-chip using image sensors to obtain 3D images of mouse two-cell embryos [ 238 ]. While maintaining embryo viability, they predicted blastocyst formation by measuring the autofluorescence of nicotinamide adenine dinucleotide [NAD(P)H], an embryonic metabolite. Taken together, these technologies are currently in the proof-of-concept stage. Further research is required to ensure sensor signal stability by addressing signal fluctuations caused by drift and biofouling while minimizing embryo handling. Sensor integration may also be extended upstream to the oocyte retrieval stage. During FF processing, oocytes are particularly susceptible to shear-induced mechanical stress [ 239 ]. Incorporating flow-sensitive sensors within microfluidic handling systems could enable real-time detection of excessive shear forces and trigger automated reduction of flow rates, thereby minimizing mechanically induced oocyte damage [ 240 ]. Similarly, in sperm sorting procedures, integration of ROS sensors may allow continuous monitoring of oxidative stress levels experienced by spermatozoa during microfluidic selection [ 241 ]. Real-time quantification of ROS could provide a functional indicator of biological stress, supporting optimization of sorting parameters to preserve sperm integrity and fertilization potential [ 242 ]. AI is being implemented to standardize morphological evaluations, such as strict morphology, which are often prone to human error [ 243 , 244 ]. Deep learning-based image analysis techniques learn patterns from germ cell and embryo images to minimize inter-observer variability and enhance objectivity, serving as a form of decision support [ 245 , 246 ]. Currently, clinical sperm morphology analysis relies primarily on computer assisted semen analysis (CASA). However, unlike oocyte or embryo assessments, sperm evaluation results can fluctuate significantly depending on sample processing and staining methods [ 247 ]. In this context, research efforts are focused on integrating AI with pre-IVF models to automate the entire process from data acquisition to analysis. Suryawanshi et al. obtained high speed phase contrast images of semen samples from patients with non-obstructive azoospermia while processing them in a microfluidic chip at a flow rate of 400 μl/h, performing real-time analysis using an AI model [ 248 ]. Once sperm were identified through deep learning-based object detection, they were collected via microfluidic gating for use in ICSI. This approach automates ART workflows by implementing sample processing, analysis, and detection within a closed-loop system. Nevertheless, AI integration within the ART context must be approached with careful consideration of ethical factors, such as data privacy, algorithmic bias, and fairness, alongside comprehensive clinical validation frameworks [ 249 ]. Meanwhile, cases of clinical applications involving embryo time-lapse video-based algorithms have been reported following regulatory approval [ 250 , 251 ]. However, current time-lapse systems typically involve placing embryos in microwells and covering them with drops of media overlaid with oil. These oil drops can induce artifacts during imaging due to the meniscus and optical scattering. It has been proposed that the integration of MPS, which provides a transparent and sealed environment that is free of oil, may facilitate the standardization of data acquisition conditions [ 252 ]. While embryologists evaluate embryos from multiple perspectives, most AI models are currently specialized in the analysis of 2D static images. To enhance the reliability of AI, it is essential to expand the scope of data to include kinetic information and morphokinetic profiles captured during embryonic development. Such advancements necessitate standardized image acquisition environments to ensure data consistency and accuracy [ 253 ]. AI applications can also be extended to the assessment of endometrial receptivity in patients with RIF. When patient-derived endometrial cells are cultured within an implantation-on-a-chip platform, continuous monitoring of cellular morphology, barrier integrity, secretory activity, and immune-epithelial interactions can generate high-dimensional phenotypic datasets. Integration of these longitudinal features with transcriptomic profiles may enable predictive modeling of individualized receptivity states and support optimization of embryo transfer timing. From physiological reconstruction to clinical validation, the selection of cell sources emerges as a critical determinant of translational reliability. While early reproductive MPS development prioritized structural fidelity and multicellular architecture, mid-term progression necessitates reproducibility and regulatory alignment. In this phase, the biological origin of cells, including ESC-derived lineages, iPSC-derived populations, and primary human tissues, directly influences phenotypic stability and the clinical interpretability of ART-relevant readouts. ESC-derived cells offer relatively consistent differentiation trajectories for modeling early reproductive processes [ 239 , 240 ]. However, restricted genetic diversity and ethical and regulatory constraints limit their applicability in patient-specific validation frameworks [ 241 ]. iPSC-derived cells enable modeling across diverse genetic backgrounds and infertility-associated phenotypes in a personalized context [ 242 ]. This adaptability renders iPSCs highly relevant for translational ART platforms designed to capture inter-individual variability [ 243 ]. Nonetheless, inter-line variability, residual epigenetic memory, and incomplete maturation remain significant barriers. In the absence of rigorous functional benchmarking, iPSC-based systems may generate biologically coherent yet clinically nonpredictive outputs. Thus, the mid-term objective is functional concordance across cell sources, which will require standardized differentiation protocols and quantitative quality control criteria to ensure reproducibility and regulatory alignment. Technical refinement alone does not ensure clinical translation. Meaningful mid-term progress will require microphysiological systems in ART to be explicitly positioned within defined steps of the clinical workflow, with a clearly articulated intended use. Validation strategies must then be tailored to the specific clinical decision each platform aims to inform, rather than relying solely on technical performance metrics. Ultimately, translational credibility will depend on prospective, cohort-based concordance studies that link predefined chip-derived outputs to established ART endpoints, including oocyte yield, implantation rate, and live birth outcomes [ 254 ]. Beyond clinical validation, regulatory feasibility is similarly shaped by a bounded claim of use. A staged qualification strategy that prioritizes analytical validity, reproducibility, and a clearly defined clinical indication represents the most pragmatic path forward [ 255 ]. Successful integration of MPS into ART requires the concurrent assurance of biological consistency at the cellular level and engineering reproducibility in device design and material performance [ 256 ]. Progress along this pathway will depend on the establishment of performance benchmarks, inter-laboratory quality control standards, and rigorously defined endpoints. Despite rapid technical advances, consistently generalizable clinical benefit has yet to be established, leaving the translational significance uncertain. In microfluidic sperm selection, studies have variably reported improvements in fertilization, clinical pregnancy, and live birth outcomes. However, substantial heterogeneity in study design, patient stratification, and endpoint definitions continues to impede consensus on clinical utility and adoption into standard practice [ 257 ]. Ethical governance introduces additional constraints, particularly for platforms involving human embryos, embryo-like constructs, or implantation-relevant interfaces. In such contexts, outputs should be framed as mechanistic or associative indicators rather than definitive clinical forecasts, while patient-derived models require stringent attention to consent, data stewardship, and responsible communication around clinical decision-making [ 258 ]. Across these considerations, analytical standardization remains central. Harmonized reporting of cell provenance, hormonal exposure conditions, ECM configuration, microfluidic operating parameters, and statistical definitions of success is essential for cross-study comparability. Within this framework, mid-term validation becomes the disciplined task of quantitatively linking engineered readouts to real-world ART outcomes under ethically sound and analytically standardized conditions. While the early objectives of MPS focused primarily on drug screening and toxicity assessment, recent developments have expanded their role into regenerative medicine, aiming to restore or replace reproductive functions in infertile patients. The core of this transition lies in scaling up from microscale cell culture platforms to organ- or tissue-level constructs that replicate physiological size and function. The previously described MPS platforms are primarily constructed with microfluidic channel architectures, which inherently limit the total volume of the cell-laden hydrogel to the microscale. In most microfluidic-based MPS, the cell-laden hydrogel is localized within a closed channel, rendering the retrieval of cultured tissues from the device difficult. This poses significant limitations for downstream applications that require recovering the cultured tissue for transplantation or further analytical characterization [ 259 ]. Achieving such organ-scale systems requires biomaterials that accurately reflect the mechanical properties of native tissues, the establishment of perfusable vasculature to ensure long-term tissue viability, and the integration of biocompatible components that closely mimic in vivo conditions [ 260 , 261 ]. Moreover, the use of patient-derived stem cells presents significant potential for developing personalized therapeutic strategies tailored to individual infertility cases [ 262 ]. The creation of implantable artificial organs and tissues holds particular promise for addressing infertility caused by premature ovarian failure, PCOS, or the adverse effects of anticancer therapies. This concept can be realized through the fabrication of bioengineered artificial ovaries using advanced bioprinting and ART technologies [ 263 ]. Such implantable artificial ovaries are expected to restore hormonal balance and mitigate reproductive toxicity-associated dysfunction. The development of scaled-up organ and tissue models can also serve non-implantable research purposes by enhancing the efficiency of ART-related investigations. For example, to overcome the limitations of current embryo culture techniques, artificial embryo and artificial uterus systems are being explored to allow stable, in vitro observation of early embryonic development. These systems enable precise evaluation of the effects of drugs and toxic agents on embryos, as well as real-time tracking of developmental dynamics, thereby improving embryo quality assessment methodologies [ 264 ]. Ultimately, the development of artificial uterus platforms capable of replicating the implantation environment and their integration into controllable systems represent a future direction in ex utero and synthetic embryo research within the ART field [ 142 ].

The global prevalence of infertility is estimated to affect approximately 8 to 12% of couples, equivalent to about 48 million couples. This is due to the biological constraints of natural conception, the rising average maternal age, and increased exposure to environmental toxicants [ [1] , [2] , [3] , [4] ]. As a result, clinical demand for assisted reproductive technology (ART) has continued to grow. ART refers to all infertility treatment techniques that manipulate sperm, oocytes, or embryos in vitro to enhance the likelihood of pregnancy when natural conception is difficult [ 5 ]. Because ART procedures include the handling and culture of gametes and embryos, their effectiveness is dependent on how well in vitro conditions mimic the reproductive microenvironment [ 6 ]. However, the current procedures show relatively low pregnancy rates per embryo transfer. They are associated with an increased risk of preterm birth, low birth weight, and fetal developmental abnormalities [ 7 , 8 ]. Therefore, physiologically relevant ART models are required to improve pregnancy rates and enhance the safety of both mothers and infants. Centering on in vitro fertilization (IVF) as the principal procedure, we have categorized the ART process into three temporal stages. The pre-IVF stage includes the in vitro production, selection and maturation of gametes, while the post-IVF stage covers embryo implantation, early embryo development, and maintenance of pregnancy [ 9 ]. Conventional ART studies have primarily relied on animal models and traditional two-dimensional (2D) cell culture systems [ 10 ]. Animal models are advantageous because of their short life cycles and the ability to investigate systemic interactions among multiple organs [ 11 ]. 2D cell culture systems are relatively simple to handle, cost-effective, and easy to analyze [ 12 ]. However, animal models have limited clinical applicability due to fundamental genetic and physiological differences from humans. Meanwhile, 2D culture systems lack the physiological microenvironment required for cell growth and cannot accurately reproduce complex cell-to-cell interactions within or between tissues over extended periods [ [13] , [14] , [15] ]. To overcome the limitations of animal experiments and 2D cell culture, the development of microphysiological systems (MPSs) has become increasingly important [ 16 , 17 ]. MPSs, including bioengineered tissues and organ-on-a-chips, are in vitro models designed to recapitulate the microenvironmental characteristics of organs [ 18 , 19 ]. These systems enable the simulation of pre-IVF, IVF, and post-IVF steps. The research in the field of ART has generally advanced through the integration of microfluidic technology [ 20 ]. Accordingly, we focus on microfluidic-based MPS within ART. Recently, sex-specific medicine has emerged as a paradigm in precision healthcare. It recognizes physiological differences between males and females to provide more accurate diagnostic and therapeutic strategies [ 21 ]. Therefore, it is necessary to design reproductive MPS that account for sex-specific microenvironments. This approach can be extended to toxicity assessments linked to infertility [ 22 , 23 ]. Sex-specific reproductive MPSs will serve as essential tools for improving the personalization of ART and toxicity screening. We focus on sex-specific reproductive MPS models and provide a detailed discussion organized into three temporal stages: pre-IVF, IVF, and post-IVF. We also analyze sex-specific reproductive toxicity assessment platforms incorporating multi-organs axis. Finally, we explore perspectives for enhancing the reproducibility in ART research and improving clinical success rates. Through a comprehensive analysis of MPS models related to both male and female reproductive systems, we aim to advance ART research and accelerate the development of next-generation reproductive platforms.

ART is performed under physicochemical and biochemical conditions that differ substantially from the in vivo reproductive environment. Variations in oxygen concentration, cytokine composition, growth factor levels, and hormonal balance can impose stress on gametes and early embryos, leading to abnormal gene expression and epigenetic reprogramming [ 207 , 208 ]. Placentas formed following ART pregnancies often exhibit aberrant phenotypes compared with those from natural conception, including increased thickness, a higher incidence of hematoma formation, and greater overall weight [ 209 ]. Such placental dysplasia reflects altered placental adaptive responses that affect embryonic microenvironments, placental gene expression, and fetal development [ 210 ]. Moreover, women exposed to environmental toxicants frequently experience subfertility or infertility, and even when pregnancy is achieved through IVF, the likelihood of successful live birth is reduced [ 211 ]. Elevated maternal blood concentrations of heavy metals such as arsenic, lead, and mercury have been correlated with increased rates of early embryonic arrest during IVF cycles [ 212 ], while exposure to air pollutants including NO 2 and O 3 has been linked to decreased birth rates among ART patients [ 213 ]. Similarly, exposure to endocrine-disrupting chemicals (EDCs) disrupts normal ovulatory processes and compromises ovarian function [ 214 ]. Collectively, these findings indicate that toxic environmental exposure not only contributes to infertility but also reduces the resilience of ART-induced pregnancies, resulting in adverse outcomes even after clinical intervention. Reproductive toxicology requires an approach rooted in sex-specific medicine to address biological differences between males and females [ 21 ]. These differences, driven by a complex interplay of genetic, physiological, and environmental factors, influence pharmacokinetics, toxicokinetics, and toxicodynamics [ 215 ]. Distinct patterns in drug metabolism and clearance lead to sex-specific adverse drug reactions (ADRs) and varying levels of reproductive toxicity [ 216 ]. Accordingly, developing physiologically representative in vitro systems to assess reproductive toxicity relevant to ART is essential. This chapter summarizes recent advances in male and female reproductive MPS and body-on-a-chip platforms for studying infertility and ART-associated toxicity. For clarity, we organize reproductive toxicity models along the ART timeline into pre-IVF and post-IVF categories. Representative examples are summarized in Table 5 , Table 6 , respectively. Table 5 Microphysiological and toxicological models for ART in Pre-IVF stage. Table 5 Chip Name Biological Stage Key Contribution Technical Challenges Potential Application Scenarios References Testicular Spermatogenic Epithelium-on-a-Chip Pre-meiotic spermatogenesis Functional SSC niche responsive to oxidative and busulfan toxicity ∙ Limited to pre-meiotic stages ∙ lacking peritubular and immune cells ∙ Male reproductive toxicant screening ∙ SSC niche mechanism studies ∙ Early spermatogenic survival testing [ 217 ] Testis-on-a-Chip Testicular endocrine crosstalk Human sertoli–leydig crosstalk chip with integrated SERPINB2 toxicity reporter ∙ Lacks germ cell incorporation ∙ Limited to static endocrine exchange ∙ Biomarker-dependent toxicity detection ∙ Human-based screening of male reproductive toxicants ∙ Evaluation of early toxic signaling responses [ 218 ] Integrated reproductive tract-on-a-chip Multi-organ reproductive endocrine cycle 28-day functional menstrual cycle recapitulated in a five-organ integrated endocrine microfluidic platform ∙ Cross-species integration Hormone dilution and inter-module consumption effects ∙ Complex electromagnetic actuation system ∙ Female reproductive toxicant screening across hormonal cycle Endocrine-disrupting chemical (EDC) evaluation Cycle-phase-specific drug response assessment [ 133 ] Ovary–Endometrium dual chip Ovarian–endometrial endocrine regulation Recapitulation of physiologic E 2 /P 4 -driven ovarian–endometrial feedback loop ∙ Limited long-term cycle validation ∙ Female reproductive toxicant screening targeting endocrine disruption ∙ Cycle-dependent drug response evaluation [ 219 ] Liver–Testis Multi-Organ-Chip Testicular endocrine Human liver–testis co-culture MPS demonstrating bioactivation-dependent germ cell toxicity ∙ Short-term culture ∙ Simplified organ equivalents ∙ Human-relevant reprotoxicant screening ∙ Prodrug germ cell toxicity assessment [ 220 ] Table 6 Microphysiological and toxicological models for ART in Post-IVF stage. Table 6 Chip Name Biological Stage Chip Name Key Contribution Technical Challenges Potential Application Scenarios References Maternal–placental–embryonic co-culture platform Early pregnancy Maternal–placental–embryonic co-culture platform Placental transport coupled to embryonic response in a microfluidic maternal–fetal axis ∙ Limited gestational duration ∙ Incomplete endocrine integration ∙ Drug safety evaluation during early pregnancy [ 221 ] Placental barrier-on-chip Early pregnancy Placental barrier-on-chip Recapitulates early placental development and function ∙ Limited long-term barrier maintenance assessment ∙ Drug safety assessment during pregnancy ∙ Modeling pregnancy complications [ 222 ] hiPSC-Derived brain organoid-on-a-chip Early fetal neurodevelopment hiPSC-Derived brain organoid-on-a-chip Prenatal nicotine–induced neurodevelopmental defects recapitulated in a perfused brain organoid chip ∙ Organoids represent early developmental stages only ∙ Prenatal neurotoxicity screening [ 223 ] Microphysiological and toxicological models for ART in Pre-IVF stage. Limited to pre-meiotic stages lacking peritubular and immune cells Male reproductive toxicant screening SSC niche mechanism studies Early spermatogenic survival testing Lacks germ cell incorporation Limited to static endocrine exchange Biomarker-dependent toxicity detection Human-based screening of male reproductive toxicants Evaluation of early toxic signaling responses Cross-species integration Hormone dilution and inter-module consumption effects Complex electromagnetic actuation system Female reproductive toxicant screening across hormonal cycle Limited long-term cycle validation Female reproductive toxicant screening targeting endocrine disruption Cycle-dependent drug response evaluation Short-term culture Simplified organ equivalents Human-relevant reprotoxicant screening Prodrug germ cell toxicity assessment Microphysiological and toxicological models for ART in Post-IVF stage. Limited gestational duration Incomplete endocrine integration Drug safety evaluation during early pregnancy Limited long-term barrier maintenance assessment Drug safety assessment during pregnancy Modeling pregnancy complications Organoids represent early developmental stages only Prenatal neurotoxicity screening Male infertility is often characterized by impaired spermatogenesis resulting from endocrine dysfunction within the testis. This failure primarily stems from a functional imbalance between the somatic cells and germ cells that constitute the testicular microenvironment. Sertoli cells, the essential somatic component of the testis, form the blood-testis barrier within the seminiferous tubules. They provide the physical support and nutrients necessary for germ cell development. Leydig cells in the testicular interstitium secrete testosterone to regulate spermatogenesis through endocrine signaling. Consequently, assessing male reproductive toxicity necessitates platforms that recapitulate the interactions between germ cells, Sertoli cells, and Leydig cells. Li et al. developed a three-layer microfluidic-based MPS for spermatogenesis using mouse-derived spermatogonial stem cells (SSCs), Sertoli and Leydig cells to model the spermatogenic epithelium [ 217 ]. The device architecture consisted of a top spermatogenesis layer containing SSCs and Sertoli cells and a bottom nutrition layer containing Leydig cells, with the two compartments separated by a porous membrane. To mimic the physiological environment where flow within the seminiferous tubules is nearly static, a slow flow rate was maintained in the upper layer. A relatively rapid flow rate was applied to the bottom nutrition layer to simulate vascularity. The system was maintained for 28 days using syringe pumps. On day 7 of culture, SSCs began to proliferate and form colonies. By day 14, differentiation into the spermatogonia stage was confirmed through increased CD117 (c-Kit) expression. Following functional validation, the platform was utilized for reproductive toxicity screening by treating the spermatogenesis layer with cytotoxic agents. Exposure to hydrogen peroxide (H 2 O 2 , 0.1 μmol/mL) to simulate oxidative stress and treatment with the pesticide busulfan (0.1 μmol/mL) both suppressed SSC proliferation, supporting the utility of the system as a toxicity screening tool. Notably, while the model successfully recapitulated SSC proliferation and early differentiation stages, complete spermatogenesis was not achieved. This was indicated by the lack of significant expression of the meiotic marker synaptonemal complex protein 3 (SYCP3). Additionally, the toxicity assessments were primarily limited to observing reduced SSC proliferation without providing a detailed discussion of the specific toxicological mechanisms. Park et al. designed a PDMS-based testicular MPS featuring separate chambers for Sertoli and Leydig cells obtained from normal human testicular tissue, which were connected via channels coated with vascular endothelial cells to facilitate the exchange of growth factors and steroid hormones ( Fig. 6 A) [ 218 ]. To recapitulate the complex microenvironment of the seminiferous tubules, the external spaces of both chambers were loaded with vascular endothelial cells and macrophages. All cell types were embedded in natural polymers, specifically hyaluronic acid and collagen, prior to application. As a result of this co-culture, the system demonstrated the secretion of androgen binding protein (ABP) in the Sertoli chamber and testosterone in the Leydig chamber, effectively mimicking the endocrine functions of the physiological testis. For the quantitative assessment of reproductive toxicity, the researchers identified SERPINB2 as a biomarker induced in both Sertoli and Leydig cells through RNA sequencing following dioxin exposure. Subsequently, a reporter system was established by coupling fluorescent proteins, such as mCherry and GFP, to the SERPINB2 promoter. By integrating this reporter system into the chip, the platform was engineered to convert increases in SERPINB2 activity into detectable fluorescent signals. This was done upon exposure to various toxic substances. This approach demonstrated the feasibility of an intuitive and quantitative evaluation of male reproductive toxicity. However, since the endocrine functions and toxicity assessments in this study were conducted after only seven days of culture, further validation through long-term exposure is necessary to enable comprehensive chronic toxicity evaluations. Fig. 6 Examples of MPS for reproductive toxicity assessment. (A) A PDMS-based human testicular MPS with coupled Sertoli and Leydig chambers connected by an endothelial channel to support hormone exchange, integrating a SERPINB2 fluorescent reporter. Reproduced with permission from Ref. [ 218 ], © 2024 Springer Nature. (B) A pneumatic-actuated multi-organ MPS that models 28-day menstrual cycle hormonal dynamics for systemic toxicity assessment. Reproduced with permission from Ref. [ 133 ], © 2017 Springer Nature. (C) A placental barrier-embryoid body co-culture chip for embryo-maternal interface toxicity assessment. Reproduced with permission from Ref. [ 221 ], © 2021 Wiley. (D) A male reproductive MPS that co-cultures human liver spheroids and testicular organoids under shared perfusion to assess metabolism-mediated male reproductive toxicity. Reproduced with permission from Ref. [ 220 ], © 2020 Oxford University Press. (E) A pneumatically actuated multi-organ MPS integrating an endometrium module with up to ten organ chips to support long-term culture and quantitative drug assessment. Reproduced with permission from Ref. [ 224 ], © 2018 Springer Nature. Fig. 6 Examples of MPS for reproductive toxicity assessment. (A) A PDMS-based human testicular MPS with coupled Sertoli and Leydig chambers connected by an endothelial channel to support hormone exchange, integrating a SERPINB2 fluorescent reporter. Reproduced with permission from Ref. [ 218 ], © 2024 Springer Nature. (B) A pneumatic-actuated multi-organ MPS that models 28-day menstrual cycle hormonal dynamics for systemic toxicity assessment. Reproduced with permission from Ref. [ 133 ], © 2017 Springer Nature. (C) A placental barrier-embryoid body co-culture chip for embryo-maternal interface toxicity assessment. Reproduced with permission from Ref. [ 221 ], © 2021 Wiley. (D) A male reproductive MPS that co-cultures human liver spheroids and testicular organoids under shared perfusion to assess metabolism-mediated male reproductive toxicity. Reproduced with permission from Ref. [ 220 ], © 2020 Oxford University Press. (E) A pneumatically actuated multi-organ MPS integrating an endometrium module with up to ten organ chips to support long-term culture and quantitative drug assessment. Reproduced with permission from Ref. [ 224 ], © 2018 Springer Nature. Successful implantation and pregnancy require precise bidirectional endocrine crosstalk between the ovary and the endometrium [ 197 ]. To reproduce this reciprocal interaction in vitro , Park et al. developed a dual reproductive MPS that integrates the cellular components of the endometrium and ovary [ 219 ]. In this dual-compartment platform, the ovarian and endometrial tissues were functionally coupled through a shared microfluidic circulation, enabling reciprocal endocrine signaling rather than isolated tissue responses. Hormone-producing granulosa and theca cells in the ovarian chamber synthesized E 2 and P 4 , which were transported through the interconnected medium to the endometrial chamber. Upon exposure to these ovarian-derived steroids, endometrial stem cells, fibroblasts, and endothelial cells exhibited receptor-mediated responses, including activation of E 2 and P 4 signaling pathways and induction of PGE 2 synthesis. Conversely, endometrial-derived soluble factors, including PGE 2 , diffused back into the ovarian compartment, where corresponding receptors were robustly expressed, thereby modulating ovarian cellular metabolism and steroidogenic activity. To enable quantitative assessment of reproductive toxicity, the researchers incorporated a SERPINB2 luciferase reporter system. In the ovary-uterus axis, dioxin introduced into the chip medium is distributed between the ovarian and endometrial chambers through the shared media channel, thereby exposing both tissue modules to the toxicant and inducing SERPINB2 reporter activation in each compartment. Consistent with this, dioxin exposure increased SERPINB2 expression together with apoptotic and growth-inhibitory responses, supporting the utility of SERPINB2-linked signaling as a readout of toxicant-mediated injury. Additional chemical stressors likewise increased SERPINB2 luciferase activity, supporting the platform as a quantitative biosensor for female reproductive toxicity screening. Xiao et al. established a multi-organ microfluidic platform that reconstituted the hormonal dynamics of a 28-day menstrual cycle across the female reproductive tract ( Fig. 6 B) [ 133 ]. The model contained separate chambers for the ovary, fallopian tube, uterus, and cervix, which were connected by microfluidic flow controlled with a pneumatic actuator. Hepatic microtissues were added to assess systemic toxicity, forming a five-organ configuration. To reproduce the pituitary hormone cycle, FSH was supplied at 10 mIU/mL for 14 days, followed by a strong hCG pulse to simulate the LH surge. When the concentration of hCG was later reduced, the system entered a phase similar to the luteal phase. Pituitary cues (FSH and hCG) delivered into the common recirculating medium were distributed across the interconnected modules. Ovarian-secreted hormones (E 2 , P 4 , and peptide hormones) are likewise carried through the same circulation to other tissues, while intra-module and whole-system recirculation promote mixing and relatively uniform exposure across modules. Under these conditions, isolated follicles produced increasing levels of E 2 during the follicular phase, which peaked at ovulation. P 4 levels reached their maximum two days after hCG exposure, resembling the luteal phase. When ovarian explants were cultured instead of isolated follicles, the pattern of hormone secretion was similar. Finally, murine ovary and human fallopian tube, endometrium, ectocervix, and liver tissues were cultured together for 28 days, allowing communication between species. Compared with single tissue culture, the concentrations of steroid and peptide hormones decreased over time. This suggested that inter-organ interactions, such as changes in upstream ovarian hormone expression or downstream hormone consumption by other tissues, affected overall endocrine regulation. The platform can culture up to ten tissues simultaneously, allowing the analysis of systemic toxicity and physiological responses in a single interconnected reproductive model. To evaluate reproductive toxicity under pregnancy-related conditions, Boos et al. developed a systemic toxicity test chip that mimics the embryo-maternal interface by co-culturing a placental barrier with mouse ESCs ( Fig. 6 B) [ 221 ]. In this platform, trophoblast cells were cultured on the upper surface of a porous membrane to form the placental barrier representing the maternal side, while mESCs were cultured on the opposite surface using a hanging drop method to generate embryoid bodies. Carboxyl-modified polystyrene microparticles (PS MPs) or their metabolites were introduced to the maternal side to examine whether these materials could pass through the trophoblast layer and affect early embryonic development on the embryonic side. The results showed that the microparticles accumulated within trophoblast cells, and that as a secondary response from the placental layer, ATP levels in the embryoid bodies decreased, indicating impaired early embryonic metabolism. Beyond mouse-based models, several studies have used hiPSCs to investigate reproductive toxicity. The use of hiPSCs has clear advantages over human ESCs because they can be derived from individual patients and avoid ethical concerns associated with embryonic sources [ 225 ]. Wang et al. cultured hiPSC-derived embryoid bodies in a PDMS-based microfluidic chip to form brain organoids, which were used to analyze the effects of nicotine exposure on early fetal brain development [ 223 ]. Nicotine was supplied through a perfusion channel connected to a syringe pump along with nutrient medium. Organoids exposed to high concentrations of nicotine exhibited abnormal expression patterns of forebrain, hindbrain, and cortical layer markers. These abnormalities indicated disrupted brain regionalization and impaired cortical development. In addition, irregular neurite outgrowth was observed, confirming that nicotine exposure interferes with early neurodevelopment during gestation. Another promising approach involves hiPSC-derived trophoblast models, which can differentiate into primitive syncytia and are well suited for personalized placental modeling [ 226 ]. Lermant et al. developed a hiPSC-derived placental barrier-on-a-chip model that effectively reproduced structural and functional features of the human placenta [ 222 ]. During culture within the microfluidic platform, hiPSCs were treated with bone morphogenetic protein 4 (BMP4) in combination with inhibition of Activin/Nodal signaling and fibroblast growth factor, a protocol known as the BAP treatment. This treatment induced differentiation toward the primitive syncytium. The differentiated cells invaded the collagen I gel and expressed lineage-specific markers of cytotrophoblasts, syncytiotrophoblasts, and EVTs, confirming progressive structural maturation. After BAP treatment, increased permeability to TRITC-dextran and upregulation of glucose transporter 1 (GLUT1) and placental amino acid transporters were observed, demonstrating the establishment of a functional placental barrier. These findings suggest that the hiPSC-based placental barrier-on-a-chip can serve as a valuable platform for drug toxicity testing and maternal-fetal interface research. Reproductive toxicity is not confined to the reproductive organs themselves but is also closely linked to systemic organs responsible for metabolism and homeostasis, including the liver, kidneys, and endocrine glands. Multi-organ MPS, often referred to as body-on-a-chip, have been developed to reproduce inter-organ communication and simulate systemic pathophysiological processes [ 227 ]. In such systems, individual single-organ modules are interconnected through shared microfluidic circuits that permit the exchange of metabolites, soluble ligands such as growth factors, cytokines, and hormones, as well as extracellular vesicles and exosomes [ 228 ]. These circulating factors mediate crosstalk between tissue modules, thereby enabling dynamic studies of inter-organ physiology. The principal applications of body-on-a-chip include chemical toxicity assessment, drug efficacy testing, and modeling of multi-organ disease processes. Integration of a hepatic module, for example, allows simultaneous evaluation of drug metabolism and bioactive metabolite toxicity under physiologically relevant conditions [ 229 ]. Thus, body-on-a-chip can simulate the dynamic in vivo ADME (Absorption, Distribution, Metabolism, and Excretion) processes, offering a predictive tool for compound toxicity and pharmacological responses [ 230 ]. In this context, sex-specific body-on-a-chip design can further improve translational relevance by capturing sex-dependent metabolic and hormonal differences in systemic toxicological responses. In male reproductive toxicology, Baert et al. developed a liver-testis axis-based model by co-culturing human liver spheroids and human testicular organoids to study systemic reproductive toxicity ( Fig. 6 C) [ 220 ]. The device consisted of a PDMS-based microfluidic channel. Above this channel, an adapter plate was mounted to accommodate transwell type cell culture compartments that also functioned as reservoirs and connectors for the perfusion pump. The circulation of culture medium through three interconnected compartments established physiological communication between the organ modules. To identify the optimal co-culture condition, they compared different media compositions for maintaining both hepatic and testicular functions. They found that the liver spheroid-specific medium best supported endocrine activity in Sertoli and Leydig cells, as indicated by elevated inhibin B production. The platform was then evaluated for its toxicological application using cyclophosphamide (CPA), a well-known testicular toxicant. CPA is metabolized in the liver by cytochrome P450 enzymes (CYPs) into its active metabolite phosphoramide mustard, which binds to DNA, disrupts replication, and induces apoptosis in rapidly dividing germ cells, leading to oligozoospermia and azoospermia [ 231 ]. The resulting active metabolites are transported through the shared microfluidic circulation to the testicular organoid module, where they induce germ cell toxicity. Following CPA exposure, both single-cultured and co-cultured testicular organoids were analyzed. Although the metabolite itself was not directly detected, the co-cultured group exhibited a greater loss of germ cells compared with the single organoid culture. The liver spheroids showed no evident cytotoxicity, but a significant increase in the CYP3A4 and CYP2B6 mRNA expression confirmed active hepatic metabolism of CPA. However, a limitation remains that the study could not clearly establish a dose-response relationship between metabolite exposure and toxic phenotypes in testicular tissue, as the CPA metabolites were not directly quantified. In addition, testicular organoids were limited in recapitulating the seminiferous tubule spatial structure. An example of a multi-organ platform incorporating the female reproductive system was demonstrated by Edington et al., who developed a quantitative drug assessment platform integrating multiple distinct MPS modules ( Fig. 6 D) [ 224 ]. The researchers designed configurations ranging from a 4-way system consisting of liver, intestine, lung, and endometrium, to a 7-way system that included brain, heart, and pancreas, and finally to a 10-way system that further incorporated kidney, skin, and skeletal muscle. These platforms were used to analyze drug responses involving the female reproductive system. Each organ module was implemented either as a flow-through unit or as a transwell insert, both connected downstream to a central mixing chamber. Mixing and distribution of culture media between MPS modules were precisely controlled using a pneumatically-actuated microfluidic pumping system with a high degree of freedom. In terms of functionality, the four-way platform maintained stable performance for two weeks, the 7-way for three weeks, and the 10-way for four weeks, enabling long-term culture across all connected tissues. The platform was further validated using quantitative systems pharmacology (QSP) modeling to predict theoretical drug distribution across the MPS network. The model accounted for drug dosage, sampling intervals, and media flow rates both between and within individual modules. Using a seven-MPS configuration, they simulated the physiologically based pharmacokinetic (PBPK) distribution of diclofenac (DCF). When DCF was introduced into the apical gut MPS at a dose equivalent to oral administration, the predicted concentration detected in the central mixing chamber matched clinically observed levels. After 48 h, the experimentally measured DCF concentrations closely aligned with PBPK model predictions, confirming the system's capacity to reproduce pharmacokinetic dynamics. Nevertheless, the endometrium MPS in this system served primarily as a constituent component to demonstrate the architectural complexity and long-term culture capabilities of the platform. Functional assessments remained largely limited to insulin-like growth factor binding protein 1 (IGFBP-1) secretion. Consequently, sophisticated endocrine dynamics were not fully investigated. Future research should aim to recapitulate hormone feedback loops, exemplified by the hypothalamic pituitary ovarian axis, at a systemic level. Furthermore, integrated studies are required to quantitatively elucidate the long-term impacts of multi-organ metabolic interactions on reproductive function and fertility. This approach will be backbone for developing high fidelity models that accurately predict systemic responses to pharmacological agents and environmental stressors.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.