Microfluidic Biofabrication of a Hydrogel Vessel‐Like Structure for Interrogating Tumor Cell Propagation in a Breast Cancer‐on‐a‐Chip Model

ABSTRACT

Engineering physiologically relevant vascular models remains a challenge in cancer biology and tissue engineering. Here, we present a microfluidic wet‐spinning methodology for the biofabrication of cell‐laden alginate‐gelatin methacryloyl (GelMA) hydrogel‐based vessel‐like structures with controlled geometry and perfusable architecture. Using a sacrificial gelatin core within a core–shell microfluidic device, tubular scaffolds were produced in a single continuous step under mild, cell‐compatible conditions. The alginate‐GelMA shell is fine‐tuned and loaded with human umbilical vein endothelial cells (HUVEC), yielding an endothelialized‐like, perfusable construct at the small‐vein scale (∼0.6–0.7 mm lumen diameter). As a proof‐of‐concept use case, the biofabricated vessel‐like structure was integrated into the breast cancer‐on‐a‐chip (BCoC) platform to showcase its feasibility in a 3D vascularized‐like tumor model for interrogating cancer cell propagation. Breast cancer spheroids were positioned in the surrounding GelMA matrix (perivascular‐like compartment), and THP‐1 monocytes were circulated through the perfusion platform. Cancer cell dissemination, quantified as circulating tumor cells (CTCs), was modulated by the immune microenvironment: interleukin‐4 (IL‐4, anti‐inflammatory‐like) conditions promoted peak CTC release, whereas lipopolysaccharide (LPS, pro‐inflammatory‐like) stimulation suppressed this dissemination. Our model integrates the vascular, tumor, and immune compartments within a single construct, providing a versatile platform for investigations of breast cancer dissemination and tumor‐immune crosstalk.

A microfluidics‐assisted breast cancer‐on‐a‐chip (BCoC) model integrates tumor, vascular, and immune components within a perfusable three‐dimensional (3D) construct. Cell‐laden alginate‐GelMA vessel‐like structures form tubular conduits that enable the analysis of tumor propagation and monocyte interactions. This physiologically relevant platform provides new opportunities to study and target early propagation‐like processes in breast cancer. Created with BioRender.com.

1. Introduction

Engineering physiologically relevant vascular models with controlled geometry, perfusable architecture, and living cellular components remains a central challenge in tissue engineering and biofabrication [1, 2]. Despite significant advances in three‐dimensional (3D) tissue‐engineered constructs, existing approaches rarely combine robust, geometrically controllable vessel‐like structures with stable long‐term perfusion within a modular on‐chip format [3, 4, 5, 6]. As a result, the fabrication of vascularized tissues featuring tubular networks under continuous flow, together with a reproducible interface to the surrounding tissue, remains a major challenge.

Hydrogels have emerged as biomaterials for 3D cell culture and vascular modeling because of their hydrated, extracellular matrix (ECM)‐like architecture, which supports cell encapsulation, morphogenesis, and engineering of perfusable vascular networks [5, 7]. Commonly used hydrogels include alginate, gelatin methacryloyl (GelMA), collagen, and fibrin, each offering distinct advantages and limitations [8, 9]. Alginate provides excellent printability and rapid ionic crosslinking but lacks native cell‐adhesive motifs, limiting endothelial cell attachment and proliferation without chemical modification [10, 11]. GelMA, a photocrosslinkable gelatin derivative, offers tunable mechanical properties and inherent cell‐adhesive RGD sequences that promote endothelial morphogenesis and vascular network formation [8, 9]. Collagen and fibrin are natural ECM proteins that strongly support angiogenic sprouting and capillary formation but exhibit limited mechanical stability for recreating large‐scale perfusable architectures [12, 13]. Other hydrogels, such as Matrigel, serve as a functional benchmark for vasculogenesis assays but suffer from batch‐to‐batch variability and poor mechanical properties [14]. In this study, an Alginate‐GelMA formulation was chosen to synergistically combine the rapid gelation and structural printability of alginate with the cell‐adhesive, enzymatically degradable, and mechanically tunable properties of GelMA, enabling the fabrication of perfusable, endothelialized‐like constructs with both macroscopic stability and microscopic adaptability for vascular tissue engineering [15, 16, 17]. Alternative biofabrication strategies for perfusable vascular constructs include sacrificial templating, microfluidic organ‐on‐chip systems, and volumetric bioprinting, each with distinct trade‐offs. Sacrificial approaches, such as SWIFT (i.e., sacrificial writing into functional tissue) and Pluronic F‐127‐ or polyvinyl alcohol (PVA)‐based writing, can create channels post‐hoc via template removal, which can compromise (sub‐)millimeter geometric fidelity and are generally not designed for direct on‐chip integration [18, 19]. Microfluidic chip platforms offer reproducibility and imaging compatibility, but rely on synthetic channel walls rather than a living hydrogel matrix, limiting endothelial–matrix interactions and 3D perivascular positioning of surrounding tissues [20]. Volumetric bioprinting enables rapid fabrication of geometrically complex, centimeter‐scale constructs [21, 22]; however, to our knowledge, such a biofabrication strategy has not yet been demonstrated for producing perfusable hollow tubular structures suitable for use cases such as on‐chip immune cell circulation.

To address these limitations, we propose a microfluidics‐assisted biofabrication approach to generate cell‐laden alginate‐GelMA hydrogel‐based vessel‐like structures (referred to as tubular scaffolds). Our microfluidic wet‐spinning approach produces cell‐laden core–shell tubular scaffolds with a defined lumen in a single, continuous step. It is specifically designed for immediate on‐chip integration, albeit at the cost of geometric simplicity compared to volumetric methods, a limitation we explicitly acknowledge. Microfluidic wet‐spinning was chosen over standard co‐axial extrusion because of several key advantages that are critical for our application. First, it offers enhanced control over the geometry of biomimetic vessel‐like structures by modulating flow rates and microchannel design. This allows for precise adjustment of the tubular diameter(s) and the creation of complex hollow morphologies, such as perfusable channels, which are challenging to achieve with macroscopic co‐axial extrusion setups [23, 24]. Second, the laminar flow conditions within microfluidic channels provide mild processing environments that can promote high cell viability and uniform cell distribution. The controlled shear forces and rapid, gentle gelation preserve cell functionality and enable consistent cell encapsulation [23]. Moreover, microfluidic wet‐spinning is particularly effective at producing perfusable hollow channels, which are crucial for developing vessel‐like structures capable of sustaining long‐term perfusion and supporting endothelialization [6, 24]. Although standard co‐axial extrusion offers simplicity and can produce mechanically robust core–shell biomimetic vessel‐like structures, it lacks the precise geometric control and seamless integration with perfusion microfluidics required by our platform [25, 26]. The sacrificial gelatin core was designed as a hollow conduit to recapitulate a native blood vessel. The alginate‐GelMA shell formulation was fine‐tuned and loaded with human umbilical vein endothelial cells (HUVEC) to produce and characterize a vessel‐like structure suitable for sustained perfusion. The resulting hydrogel‐based tubular scaffold recapitulated an engineered blood vessel at the millimeter scale and allowed its integration into a novel 3D breast cancer model for the interrogation of tumor cell propagation across a vascular‐like interface. In detail, the engineered vessel‐like construct was designed to approximate a core sub‐millimeter scale, allowing its perfusion once integrated into the on‐chip platform, enabling both shell endothelialization‐like process and inner immune cell circulation. The engineered tubular scaffold exhibited a lumen diameter of ∼0.6–0.7 mm and an overall diameter of ∼1.4–1.5 mm, corresponding to the size range of small veins rather than microvasculature [27]. It should be noted that native vasculature exhibits hierarchical branching and a multilayered wall architecture comprising endothelial, mural, and perivascular compartments. The simplified, single‐layer endothelialized‐like conduit presented here represents a recognized limitation; however, it enables reproducible and well‐controlled fabrication with clearly defined geometric and biological parameters. As a proof‐of‐concept use case, the biofabricated vessel‐like structure was integrated into a breast cancer‐on‐a‐chip (BCoC) platform to illustrate a potential application in interrogating cancer cell propagation across a vascular‐like interface. This use case is motivated by the fact that, despite considerable progress in cancer biology, the mechanisms driving breast cancer cell dissemination and survival at metastatic secondary sites remain insufficiently understood [28, 29, 30, 31].

Although conventional cell culture and animal‐based models have contributed to our understanding of cancer progression [30, 31], the mechanisms driving breast cancer cell dissemination and survival at metastatic secondary sites are unclear. Moreover, neither cell culture systems nor animal models can fully recapitulate the pathology [30]. In addition, the critical steps of cancer invasion and intravasation are insufficiently understood, largely due to the lack of physiologically relevant experimental systems to study these processes. Similarly, the absence of immune system components in most existing metastatic models represents a key drawback [32, 33, 34]. Circulating tumor cells must survive in the bloodstream under constant immune attack [35, 36]; however, the current limited in vitro systems account for this critical aspect [36, 37]. Recent sacrificial‐based cancer models have demonstrated vascularized tumor niches with custom perfusion [38, 39], none combine circulating immune cell perfusion with tumor intravasation‐associated readouts in a single modular device. Taken together, existing 3D tissue‐engineered models rarely integrate the tumor microenvironment (TME) with robust, perfusable, and geometrically controllable vascular structures, limiting their ability to mimic the complexity of metastatic progression [1, 2].

In the BCoC use case presented here, breast cancer cells were introduced as pre‐formed multicellular spheroids positioned in the perivascular tumor compartment surrounding the biomimetic vessel‐like structure, thereby remaining spatially separated from circulating immune cells at the initial stage. THP‐1 monocytes were subsequently introduced into the circulation of the perfusable platform via a reservoir‐connected perfusion system. This design enabled feasibility‐level observation of the interplay between circulating THP‐1 monocytes and disseminating breast cancer cells and the associated formation of circulating tumor cells (CTCs) in the perfusion reservoir. Importantly, this use case application is intended to illustrate the potential of the engineered tubular scaffold as a tool for further exploration of tumor‐immune interactions, rather than to draw definitive mechanistic conclusions. To our knowledge, while individual elements such as hydrogel wet‐spinning, microfluidic perfusion, and breast cancer models have been reported, their rational integration into a single, controllable system has not been demonstrated. By combining advanced hydrogel biofabrication with microfluidic perfusion, our model offers a versatile platform in which the synergistic combination of vascular, tumor, and immune compartments provides a unique opportunity to interrogate the dynamic interplay between cancer invasion, propagation, and survival in circulation. More broadly, this study aims to highlight the biomimetic nature of engineered vessel‐like constructs while recapitulating key events of metastatic‐like progression, specifically tumor invasion through the surrounding tissue and propagation into the vascular lumen, thereby contributing to the future development of effective therapeutic approaches, framed specifically as a proof‐of‐concept demonstration.

2. Results and Discussion

2.1. Microfluidics‐Assisted Wet‐Spinning Biofabrication of Hydrogel‐Based Biomimetic Vessel‐Like Structures

A core–shell design was chosen to biofabricate hollow tubular scaffolds for vascularization of the BCoC model. To this end, a microfluidic co‐axial extrusion nozzle (MiCoEN) was used. To closely mimic the structure of a blood vessel, the core composition was designed as a sacrificial template of gelatin (3% w/v) to enable the formation of a tubular architecture. Similarly, a HUVEC‐laden dual‐phase hydrogel consisting of alginate (4% w/v) and GelMA (5% w/v) was selected for the shell formulation to provide a robust and cell‐friendly environment (Figure 1A). A core–shell configuration of the nozzle was employed to replicate the native architecture of the vessel‐like structures. The MiCoEN employed in this study was characterized by two independent inlets, designated to independently provide the core and shell inks. This configuration is reflected in the MiCoEN outlets and allows compartmentalization of the extruded tubular scaffold (Figure 1B).

FIGURE 1.

Microfluidics‐assisted wet‐spinning biofabrication of biomimetic hydrogel‐based vessel‐like structures. (A) Design of the wet‐spun tubular scaffold forming a core–shell vessel‐like structure. The HUVEC‐laden alginate‐GelMA shell was arranged concentrically around the sacrificial gelatin‐based core to generate a hydrogel conduit that mimicked vascular architecture. (B) Configuration of the microfluidic co‐axial extrusion nozzle (MiCoEN) employed for the biofabrication of the perfusable biomimetic vessel‐like structure shown in (A). (C) Schematic representation of the experimental setup and workflow. Hydrogel precursors are processed via the wet‐spinning technique, followed by immediate crosslinking in an ice‐cold CaCl₂ coagulation bath to stabilize the alginate matrix and physically gel the gelatin phase. Subsequent UV photopolymerization was performed to crosslink the GelMA network. The cell‐laden construct was then incubated at 37°C to support HUVEC maturation and induce thermal removal of the gelatin core, triggering the creation of a hollow structure suitable for perfusion in the on‐chip platform. Created with BioRender.com.

The overall biofabrication platform is based on the wet‐spinning technique and relies on three main components: i) the MiCoEN for the extrusion of a meter‐long hydrogel‐based vessel‐like tubular conduit, ii) a dual digital pump to dispense the hydrogel precursors toward the tip of the MiCoEN, and iii) an icy CaCl₂ coagulation bath (∼4°C) for both the immediate polymerization of the alginate phase, which interacts with the calcium ions of the CaCl₂ buffer, and the gelation of the gelatin core at low temperature (Figure 1C). Once extruded, the fibrous hydrogel scaffold was gently collected from the icy tank and sectioned into standardized 3 cm‐long tubular conduits to obtain experimental specimens, which were subsequently placed under UV light to cross‐link the GelMA phase of the shell. Subsequently, the biofabricated constructs were incubated (5% CO₂, 37°C) for up to 21 days for two main reasons. The first was to fine‐tune and characterize the in vitro 3D tubular vessel‐like model, and the second was to demonstrate its capability to sustain cell culture for a minimum of 14 days, thereby supporting the envisioned integration of a single perfusable and biomimetic vessel‐like structure segment into individual BCoC platforms (i.e., one vessel‐like conduit per platform). This also allowed experiments to be conducted in parallel across multiple chips while maintaining statistical robustness and reproducibility.

The physiological temperature allowed for sustained dissolution of the gelatin core within the first 6 h. This step was crucial for establishing the dynamic microenvironment experienced by breast cancer cells when the biofabricated hollow vascular channel was integrated into the on‐chip platform. This integration enabled the direct interrogation of tumor cell propagation across a vasculature‐like interface, including interactions with circulating THP‐1 monocytes within the proposed 3D model (Sections 2.6 and 2.7).

A significant advantage of the wet‐spinning biofabrication platform is its ability to modulate fiber dimensions through the precise adjustment of core and shell flow rates [26, 40]. This is remarkably important for tissue engineering applications in the (micro)vascularization field [6]. Such controllability allows for the tuning of the overall diameter and shell thickness, which are directly associated with mass transport properties, such as oxygen and nutrient diffusion, and consequently, cell viability within the construct. In addition to this dimensional flexibility, the system offers rapid prototyping capabilities. Specifically, meter‐long, fiber‐based tubular scaffolds can be produced and subsequently sectioned into multiple smaller specimens. This approach not only facilitates the efficient generation of experimental replicates, thereby avoiding batch‐to‐batch variation from a single fabrication run, but also enhances the reproducibility and scalability of the biofabrication process for future applications.

2.2. Fine‐Tuning of Wet‐Spinning Parameters and Vessel‐Like Structure Characterization

The selection of an appropriate hydrogel formulation for the wet‐spinning of tubular scaffolds involves the consideration of various parameters, which are independently analyzed and subsequently integrated. The effect of multiple wet‐spinning parameters was systematically investigated to identify an optimal spinnability window while ensuring reproducible tubular scaffold formation, structural integrity, and compatibility with on‐chip perfusion. In fact, the hydrogel‐based vessel‐like structure should exhibit sufficient stability when placed into the on‐a‐chip platform. Parallelly, needle insertion and flow injection should proceed without compromising the shell structure, thereby ensuring smooth flow throughout the tubular conduit. To this end, two distinct co‐axial nozzles were evaluated: 15G‐19 and 17G‐22G configurations (where xxGG‐yyGG represents the gauge dimensions of the outer (yy) and inner (xx) needle diameters, respectively). The choice of the MiCoEN is pivotal in determining the characteristics of wet‐spun tubular scaffolds. The 15G‐19G system, characterized by a wider lumen for both the core and shell compartments, facilitates the extrusion of fiber‐based tubular scaffolds with larger diameters and thicker shells, which can be advantageous when greater stability or increased cell‐loading capacity is desired. Conversely, the 17G‐22G configuration possesses narrower internal and external diameters, allowing the fabrication of thinner hydrogel conduits with reduced shell thickness, enhancing the diffusion of nutrients and oxygen across the construct, and mimicking microvascular dimensions more accurately. Thus, the selection of the nozzle size directly influences the tubular morphology, structural integrity, and mass transport properties, offering flexibility in tailoring constructs for specific biological applications. Additionally, two different combinations of core–shell flows (Q_s:640/Q_c:320, Q_s:640/Q_c:160 µL/min) and four distinct shell formulations (i.e., alginate 2% and 4%, in combination with GelMA 5% and 7% (w/v); herein named as AxGy, where “x” and “y” are the variable alginate and GelMA concentration (% w/v) used, respectively) formulations) were tested to assess their impact on the properties of the tubular scaffolds, allowing for the mapping of a formulation‐ and biofabrication‐dependent spinnability window guiding the final hydrogel selection (Figure 2A). The rationale behind these variations was to effectively investigate the stability and overall handling of the biomimetic vessel‐like structures (Section 4.1.2) while ensuring a supportive environment for cell culture. Importantly, in all tested conditions, the concentration of GelMA was higher than that of the alginate. This strategy was employed to leverage the rapid crosslinking ability of alginate for tubular scaffold formation while simultaneously prioritizing the bioactivity and cell‐adhesive characteristics of GelMA, thereby promoting cell attachment, proliferation, and a more favorable microenvironment for sustained culture. Notably, the results indicated that wet‐spinning via the 15G‐19G nozzle did not negatively affect the extrusion and spinnability of the vascular channels for both Q_s/Q_c flows (Figure 2B). However, the 640‐160 µL/min condition presented a slight challenge in needle insertion into the tubular structures owing to the reduced core diameter (lower flow rates and lower volume extruded). The smaller size of the 17G‐22G nozzle resulted in a reduced shell thickness, which in turn limited the performance in terms of both needle insertion and tubular channel injection, which are critical for allowing the perfusion of the vascular channel (Figure 2C; Figure S1A,B). While thinner channels may more closely mimic microvascular dimensions, they exhibit limited robustness during post‐fabrication manipulation. In particular, hydrogel scaffolds fabricated with the 17G‐22G nozzle could not reliably withstand manual needle insertion or subsequent tubular channel injection, which are crucial for enabling perfusion within the on‐chip platform. Conversely, scaffolds produced using the 15G‐19G nozzle provided a thicker shell and increased structural stability, allowing reproducible needle insertion and sustained injection without inducing the collapse or rupture of the vessel‐like conduit. These practical limitations, along with the fine‐tuned spinnability analysis integrating material formulation and wet‐spinning parameters such as nozzle gauge and core–shell flow rates (Figure 2A), led to the selection of the 15G‐19G configuration as the optimal compromise between geometric fidelity, mechanical robustness, and perfusion capability for subsequent biological studies on the overall 3D on‐chip platform. For instance, tubular constructs fabricated using 2% (w/v) alginate exhibited weaker stability than those extruded with a 2‐fold higher concentration (4% w/v). This is likely attributable to the strengthening effect of the sodium alginate material [41]. Therefore, the material selection favored the alginate 4%‐GelMA 5% (w/v) formulation (i.e., A4G5) wet‐spun with the MiCoEN 15G‐19G. This configuration provided flow stability and facilitated tubular scaffold perfusion owing to the larger inner diameters of both the shell and core compartments.

FIGURE 2.

Fine‐tuning and characterization of wet‐spun hollow tubular vessel‐like structures. (A) Wet‐spinnability window for the fabrication of vessel‐like scaffolds, showing the combination of different shell‐core flow rates (Q_s‐Q_c (µL/min) – i.e., 640–320, 640–160, 320–320, 320–160) used to produce vascular channels with two microfluidic co‐axial extrusion nozzles (MiCoEN), i.e., 15G‐19G (1519) vs. 17G‐22G (1722) tested for various alginate‐GelMA (AxGy, where “x” and “y” are the variable alginate and GelMA concentrations (% w/v) used, respectively) formulations. The red box highlights the fine‐tuned A4G5‐1519 hydrogel formulation for the wet‐spinning of the proposed vessel‐like structures. (B) Details of (i) the obtained A4G5‐1519 tubular scaffold (water was injected to enhance the visualization of the lumen and shell architecture) and performance testing of (ii) initial injectability and (iii) overall perfusability of the (iv) tubular vessel‐like structure. A red dye water‐based solution was used for the perfusion trials. C) Comparison of scanning electron microscopy (SEM)‐based cross‐sectional images of the fine‐tuned A4G5 (Q_s:640‐Q_c:320 µL/min) tubular conduits wet‐spun with two different MiCoENs. Scale bar: 200 µm. D) Swelling and E) degradation behavior of the proposed vessel‐like conduits A4G5‐1519 640‐320 up to 21 days in both HUVEC medium and HEPES 25 mm (n = 3).

2.3. Swelling and Degradation Dynamics as Indicators of Hydrogel Physical Properties

The swelling capacity of the fine‐tuned A4G5 (Q_s‐Q_c: 640‐320 µL/min) hydrogel tubular scaffolds was evaluated for up to 21 days in HUVEC medium to determine their water absorption potential (Figure 2D). Specimens in 25 mm HEPES buffer were used as controls. After 1 h, both conditions exhibited a rapid increase in water uptake, reaching ∼866.24% in HUVEC medium and ∼313.32% in HEPES buffer. In HUVEC medium, rapid swelling continued until day 1, when the vessel‐like structures reached maximum water uptake (∼993.86%), followed by a plateau‐like phase with minor fluctuations throughout the 21‐day culture period. The pronounced swelling of tubular scaffolds in HUVEC medium can be attributed to the strong osmotic gradient created by its enriched composition, whereas salts, growth factors, and serum proteins likely increase water influx into the hydrophilic hydrogel network [42]. Moreover, the presence of Na⁺ and Mg^{2 +} ions in HUVEC culture medium may have contributed to both the pronounced swelling and the faster degradation of the vascular channels compared to HEPES buffer, which lacks divalent ions. While Ca^{2 +} initially stabilizes alginate through ionic crosslinking, it may potentially undergo competitive exchange with abundant Na⁺ and Mg^{2 +} ions in the cell culture medium, thereby leading to enhanced swelling and accelerated mass loss [42, 43]. Conversely, the absence of such ions in HEPES buffer resulted in limited swelling and slower degradation. The subsequent slight decline likely reflects partial matrix degradation and dynamic ion‐polymer interactions, resulting in fluctuations around the equilibrium swelling state. The swelling of the HEPES‐buffered controls gradually increased over the entire culture period, reaching ∼700% by day 14. After two weeks, the profile entered a plateau‐like phase with minor fluctuations up to day 21, including a slight decrease on day 17 (∼682.03%), followed by a modest rise to ∼739.38% on day 21. The overall lower and more gradual swelling observed for the specimens in HEPES can be explained by its simple buffering composition, in contrast to the enriched components of the HUVEC medium. Consequently, the swelling in HEPES likely reflected the intrinsic hydration capacity of alginate–GelMA, resulting in a steadier and less pronounced profile than that observed in the HUVEC medium. This behavior was further confirmed by a degradation test conducted over 21 days (Figure 2F), which revealed trends consistent with swelling data. In HEPES, the specimens exhibited a slow and steady mass loss, decreasing from ∼92.90% residual mass on day 1 to ∼85.80% on day 21. Conversely, tubular scaffolds incubated in HUVEC medium degraded more rapidly, retaining only ∼77.37% of their mass on day 1 and declining to ∼69.22% by day 21. This higher degradation rate compared to HEPES‐buffered controls can be linked to the cell culture medium‐enriched composition, which likely weakens the alginate crosslinks and accelerates the overall hydrogel matrix breakdown. This trend aligns with the swelling results, as the same components that enhanced water uptake in the HUVEC medium also contributed to faster degradation, whereas the HEPES buffer maintained greater structural stability than the HUVEC medium. A comparison of both the swelling and degradation of the fine‐tuned A4G5 tubular scaffolds wet‐spun at 640–320 µL/min was also assessed against the same vessel‐like structures wet‐spun at 640–160 µL/min (Figure S1C). Interestingly, the optimal conditions revealed a higher swelling rate in HUVEC culture medium. Conversely, the fine‐tuned specimens in HEPES buffer exhibited a lower swelling profile over time, although the difference was not significant.

2.4. Microstructural Evaluation of the Vessel‐Like Tubular Structure

The microstructure of the cell‐laden tubular hydrogel structures was investigated for up to 21 days with minimal specimen preparation after cell fixation (Figure 3). The tubular cavities of the scaffolds were visible and well‐defined throughout the culture period. The cross‐sectional images clearly outline the regular circularity of the outer diameter, dictated by the co‐axial nozzle (Figure 3A.i). This can be attributed to the alginate‐based shell, which undergoes crosslinking almost instantaneously (i.e., at the tip of the MiCoEN) when in contact with the Ca^{2 +} present in the coagulation bath, thereby freezing a smooth outer profile. Conversely, the gelatin core, which is softer than the shell formulation and undergoes markedly different gelation dynamics, gels more slowly within the icy tank and can deform before full thermal polymerization, thus producing an irregular lumen and volumetric deformation, as shown on day 7 (Figure 3A.ii). In addition, the combined effect of a soft sacrificial core and a potential slight tilt between the inner and outer needles could produce an off‐center core. These findings were supported by micro‐computed tomography (µCT) 3D reconstructions, which highlighted a consistent outer hydrogel profile, whereas the inner lumen appeared less regular and occasionally displaced from the central axis. (Figure 3B). These effects highlight the sensitivity of coaxial wet‐spinning to both formulation parameters and process stability. These fluctuations in the biomimetic vessel‐like structure dimensions (i.e., overall diameter, core (conduit) diameter, and shell (wall) thickness) were confirmed by analyzing 2D images (Figure 3C). The average size of each fiber‐like dimension was calculated over the 3‐week culture period (overall diameter: 1470.41 ± 58.12 µm, core diameter: 638.71 ± 97.51 µm, and wall thickness: 415.85 ± 30.51 µm). The overall diameter of the tubular scaffold demonstrated a moderate temporal variation across the time points (day 7: 1415.34 ± 26.81 µm, day 14: 1531.16 ± 11.82 µm, and day 21: 1464 ± 7.80 µm). The initial enlargement observed up to day 14 may be attributed to progressive extracellular matrix deposition or cellular proliferation, contributing to fiber‐like thickening. The subsequent decrease on day 21 could reflect matrix compaction or remodeling processes associated with tissue maturation and structural reorganization. Similarly, the core variability was statistically significant when comparing day 14 and day 21 to day 7 (day 7: 526.14 ± 17.97 µm, day 14: 692.91 ± 10.53 µm, and day 21: 697.07 ± 5.89 µm). Similarly, the same trend was detected for the shell (day 7: 444.59 ± 4.51 µm, day 14: 419.12 ± 8.26 µm, day 21: 383.83 ± 6.41 µm), whose inner interface was also influenced by the soft, thermally polymerized gelatin core. The progressive enlargement of the core and thinning of the shell also suggest dynamic remodeling of the microstructure of the tubular scaffold during culture. Here, HUVEC deposited ECM within the hydrogel, leading to the overall compaction and contraction of the biomimetic vessel‐like structure network. This cell‐mediated contraction could be most pronounced around day 14, resulting in a decrease in the overall diameter [44]. These findings indicate that the above dimensional fluctuations can be attributed not only to the intrinsic softness and partial degradation of the thermally polymerized gelatin‐based core but also to possible cell‐mediated remodeling processes and swelling/dehydration phenomena of the hydrogel matrix. Such dimensional shifts are particularly relevant when designing perfusable conduits, because lumen stability directly influences nutrient transport and long‐term structural fidelity.

FIGURE 3.

Geometry Evaluation of the vessel‐like tubular structure via X‐ray microcomputed tomography (µCT) scan reconstruction. (A) 2D view (XY‐ and XZ‐plane) of both (i) cross‐ and (ii) longitudinal‐section and (B) µCT‐based 3D scan reconstruction of the hydrogel‐based hollow tubular conduits at 7, 14, and 21 days of culture. Scale bar: 400 µm. (C) Analysis of the tubular scaffold dimensions (overall diameter, core diameter, wall (shell) thickness) over 21 days (n = 3). Significant differences: ^* p ≤ 0.05, ^** p ≤ 0.01, ^*** p ≤ 0.001, and ^**** p ≤ 0.0001.

2.5. Morphological Features and Cell Viability Assessment of the Biofabricated Vessel‐Like Structure

The morphological characteristics of the biomimetic cell‐laden hydrogel‐based vessel‐like structures were evaluated after dehydration. Both cross‐sectional and top views were examined using scanning electron microscopy (SEM) for 21 days (Figure 4). In detail, SEM imaging was performed on dehydrated specimens. Thus, the measured wall thickness corresponds to the dry state and cannot be directly extrapolated to physiological (hydrated) conditions. In this frame, SEM was primarily employed to confirm the presence of the hollow architecture and to qualitatively assess shell delamination and time‐dependent microstructural evolution of the alginate‐GelMA vessel‐like structures. The hollow geometry was quantitatively evaluated under wet conditions using µCT analysis (Section 2.4). The dry wall thickness was consistently measured at approximately 20 µm throughout the culture period in cell culture medium, with no significant variation (Figure 4A). This size reduction is mainly attributed to the shrinkage effect induced by the dehydration process [45]. Although alginate lacks cell‐specific binding sites to facilitate cell adhesion, the A4G5 scaffold demonstrated HUVEC attachment to the hydrogel, likely due to the geometrical shape of the scaffold induced by the shell compartment. Other studies have reported favorable cell adhesion on alginate‐based scaffolds when combined with other biomaterials or coatings [46, 47]. By day 7, a few cells were qualitatively observed on the exposed A4G5 surface, which showed a honeycomb‐like pattern at high magnification (i.e., 20 µm), with a visible increase by day 14 (Figure 4B). This process may result from the timely degradation of GelMA over alginate within the shell compartment, leading to the gradual exposure and attachment of encapsulated cells to alginate upon GelMA delamination, as well as from the overall material degradation, as previously discussed [48, 49]. After two weeks, a greater number of cells were observed, material degradation became more pronounced, and the microstructure appeared smoother and less rough. It is possible to speculate that this may reflect the dynamic remodeling processes occurring within the alginate‐GelMA matrix during the culture period, as well as SEM‐related artifacts [50]. Also, both polymeric components are known to undergo gradual physicochemical transformations, as evidenced by the swelling and degradation experiments (Section 2.3). Taken together, these factors can affect the integrity and surface properties of the network. The hydrogel network likely experienced partial relaxation or structural reorganization by day 14, temporarily resulting in a smoother surface appearance. At later stages, increased matrix compaction may have reintroduced surface irregularities, whereas similar time‐dependent morphological changes have been documented in composite hydrogels of GelMA and alginate, where local remodeling, swelling‐deswelling behavior, and degradation kinetics were found to influence the apparent topography and mechanical stability of the construct [51, 52, 53]. By day 21, the shell formulation underwent further degradation, as evidenced by cross‐sectional images displaying an overall thinner tubular diameter, although a fully tubular structure was not retrievable. In addition, the honeycomb‐like pattern observed at day 7 (Figure 4B, 20 µm‐magnification) was no longer clearly visible. The top view did not confirm a substantial cell population, suggesting that shell delamination stabilized and preserved the innermost part of the conduit. Parallelly, the dehydration process may have potentially washed cells out. However, a few niches with encapsulated cells were observed on day 21. A comparison of fiber‐like diameters in both dry (SEM) and wet (µCT) states was conducted to assess the synergistic effects of dehydration and delamination on the A4G5 hydrogel formulation (Figure S1D).

FIGURE 4.

Morphological features of biomimetic vessel‐like structures. SEM‐based images of the HUVEC‐laden hydrogel conduits at 7, 14, and 21 days in both (A) cross‐sectional and (B) longitudinal views. Light blue arrows indicate the region of interest displayed in the top view at 8 µm. Blue arrows indicate both cell and surface patterns, which are subsequently zoomed in on the following micrographs. Scale bars: 300, 30, 20, and 8 µm.

SEM observations were confirmed by optical images, whereas HUVEC did not show characteristic cell proliferation features over the culture time. This is potentially due to the high cell density and limited space provided by the thin‐shell compartment (Figure 5A). In this constrained environment, the cells tended to maintain a rounded morphology rather than displaying a distributed spindle shape. This behavior can be observed in hydrogels with a restricted spreading capacity due to stiff matrices, such as our A4G5 composition [54] (Figure S2A,B), while those in softer matrices are more likely to spread and elongate [55]. For instance, mesenchymal stem cells encapsulated in soft hydrogel matrices can retain spherical shapes, only elongating when the hydrogel permits deformation or degradation [56].

FIGURE 5.

Cell viability of biomimetic vessel‐like structures. (A) Representative optical images of the HUVEC population within the shell compartment of the tubular scaffold. Scale bar: 300 µm. (B) Cell viability of encapsulated HUVEC in terms of (i) qualitative (Day 7) and (ii) quantitative analysis (the dotted red line represents the cut‐off of 75 % cell viability according to the ISO 10993–5 standard). Viable cells were stained in green, and dead cells were depicted in red. Scale bars: (i) 200 µm, (ii) 300 µm.

To demonstrate the feasibility of maintaining endothelial cell viability within the compartmentalized shell of the tubular scaffolds, quantitative cell assessment was performed across the culture period. In this framework, the AO/PI assay on the hollow HUVEC‐laden vessel‐like structures confirmed that the encapsulated cells remained viable (Figure 5B.i). HUVEC maintained high cell viability (> 80%) up to day 21, indicating that the A4G5 hydrogel microenvironment supported cell survival, in agreement with ISO standard 10993–5 for cytotoxicity studies, even in the absence of extensive cell spreading. No statistical difference was observed among all three time points (day 7: 87.62% ± 8.92, day 14: 88.96% ± 6.79, day 21: 90.73% ± 7.68) (Figure 5B.ii). Such behavior is not uncommon in densely packed alginate‐based or alginate‐containing hydrogels, such as the A4G5‐based shell of the proposed biomimetic tubular scaffolds. Here, the bioinert alginate, despite the presence of RGD motifs in GelMA, may locally limit cell‐matrix adhesion, as well as delay proliferation and elongation until sufficient remodeling or matrix degradation potentially occurs [57, 58].

Moreover, the observed round morphology at this stage did not adversely affect the integrity or stability of the engineered constructs. However, the endothelial barrier function was not quantitatively evaluated in the present study; therefore, it may represent a limitation of the current platform. The proposed biomimetic vessel‐like structure serves as a simplified, perfusable, endothelial‐like conduit intended to facilitate integration within an on‐chip system rather than recapitulating a fully mature, multilayered vascular barrier.

Subsequently, each tubular construct was independently injected with HUVECs and integrated into a single, dedicated BCoC platform containing a matrix with Human Dermal Fibroblasts (HDFs) and either Human Metastatic Breast Adenocarcinoma (MDA‐MB‐231) monoculture spheroids or HUVEC co‐cultured spheroids for further evaluation.

2.6. A Use Case for Integration of a Vessel‐Like Structure to A Microphysiological System To Recapitulate Propagation of Breast Cancer Cells

In the context of existing breast cancer models, a wide range of spheroid‐based cultures, static co‐culture systems, microfluidic tumor‐on‐chip platforms, and vascularized tumor‐on‐chip platforms have been reported to study tumor progression, invasion, and metastasis [59]. While these approaches have provided important biological insights, many lack integrated perfusable vasculature, pre‐patterned endothelial channels, and incorporation of immune and stromal cells. Leveraging tissue engineering approaches, the present platform translates a perfusable biomimetic vessel‐like hydrogel construct into a modular on‐chip architecture that enables both controlled flow and dynamic tumor‐vascular‐immune interactions within a single system. Such integration provides a use case by complementing existing vascularized tumor microenvironment platforms and stands out as a reproducible biofabrication workflow while maintaining flexibility for adaptation to different tumor types and experimental designs.

HDF cells, a key component of the TME, were incorporated into a GelMA hydrogel matrix that mimicked the native ECM of the TME (Figure 6). To form the tumor niche, spheroids based on monocultures of MDA‐MB‐231 breast adenocarcinoma cells and co‐cultures with HUVEC cells, respectively, were generated in the rotary cell culture system (RCCS) bioreactor under dynamic conditions for 14 days (Figure S3). On day 7, bright‐field microscopy revealed that the co‐cultured spheroids exhibited a compact and well‐defined morphology (Figure S3A). In contrast, the aggregate structure in monoculture spheroids gradually dispersed toward the periphery, which clearly demonstrated the regulatory and tissue‐organizing effects of endothelial cells in the TME [60]. Live/Dead staining qualitatively indicated a tendency toward greater cell death in monoculture spheroids than in co‐culture spheroids (Figure S3B). This suggests that the tumor structure was better nourished by increased endothelium‐mediated diffusion in the presence of HUVEC, while the typical necrotic zone persisted [61]. Considering the parameters of advancement, efficiency, quality, size, morphology, and time consumption, optimal spheroid formation was achieved on day 7 of RCCS culture with HUVEC co‐culture. Therefore, co‐culture spheroids harvested after 7 days of culture were used for subsequent on‐chip experiments. Quantitative analysis of spheroid dimensions demonstrated that the average diameter was significantly greater in the co‐culture spheroids (492.8 ± 74.6 µm, p < 0.0001) than in monoculture counterparts (309.0 ± 53.6 µm), consistent with observations from bright‐field imaging and Live/Dead staining assays (Figure S3C). HIF‐1α is an indicator of cell responses to hypoxic conditions, where oxygen is insufficient, while Ki67 is a marker expressed in highly proliferating cells [62, 63]. Immunofluorescence analyses further revealed elevated HIF‐1α signaling in monoculture spheroids, whereas co‐culture spheroids exhibited a partial reduction in HIF‐1α expression, indicative of a more balanced oxygen gradient. Despite this decrease, the hallmark hypoxic characteristics were retained (Figure S3D). Additionally, Ki67 expression was predominantly localized to the peripheral regions of the co‐culture spheroids, whereas monoculture spheroids displayed comparatively weaker fluorescence signals. This is consistent with enhanced proliferative cellular activity due to increased access of spheroids to oxygen and nutrients in the co‐culture environment [64, 65].

FIGURE 6.

Schematic representation, experimental setup, and timeline of the breast cancer‐on‐a‐chip platform. (A) Illustration of the stepwise assembly of the microchip platform: (i) HDFs encapsulated in GelMA hydrogel were placed into the chamber, followed by (ii) the integration of a hydrogel‐based vessel‐like structure containing encapsulated HUVECs. (iii) MDA‐MB‐231/HUVEC tumor spheroids that were prematurely matured in the RCCS bioreactor were positioned adjacent to the vessel‐like structure to mimic tumor‐vascular interactions. (iv) The complete microchip was connected to a perfusion system to establish a dynamic microenvironment. Representative display of all experimental groups: (v) control, (vi) THP‐1, (vii) IL4‐induced THP‐1, and (viii) LPS‐induced THP‐1. (B) Design and dimensions of the microchip device, consisting of three layers enclosing a central part (4 × 2 × 0.5 cm) with an integrated lumen‐like channel (3 × 1 × 0.3 cm). (C) Photographic and microscopic images of the experimental components: (i) RCCS bioreactor containing (ii) MDA‐MB231/HUVEC spheroids, (iii) bright‐field microscopic image of the biomimetic hydrogel‐based vessel‐like structure, (iv) complete perfusion setup with tubing and reservoir connections inside the incubator‐assembled perfused microchip platform, (v) top view of the platform, and (vi) bright‐field microscopic image of the vessel‐like structure (white *) with adjacent tumor spheroids (green arrows). Scale bar: 100 µm. Partly created with BioRender.com.

To reconstruct the TME, fibroblasts were embedded within GelMA and positioned as a thin layer on top of the chamber into the platform (Figure 6A.i). Subsequently, HUVEC‐laden vessel‐like structures, which mimic blood vessels through their lumen‐like structure, were integrated into the platform in alignment with the inlet and outlet (Figure 6A.ii). The remaining space was filled with fibroblast‐laden GelMA, and 7‐day‐old co‐cultured breast cancer spheroids harvested from the RCCS bioreactor were positioned in GelMA adjacent to the vessel‐like structure (Figure 6A.iii). This strategic arrangement successfully recapitulates an in vivo‐like microenvironment by mimicking the spatial organization between the tumor stroma and the angiogenic region. For the use case application of this study, a fully perfused system was established by connecting the self‐lumen vessel‐like structure, positioned at the inlet and outlet of the platform, to a dynamic flow circuit via silicone tubing. This configuration enabled mechanotransduction, thereby supporting cell–cell interactions and sustaining metabolic activity within the tumor region (Figure 6.iv). The hydrogel‐based vessel‐like structures, produced in a single fabrication batch, were carefully UV crosslinked, cut into approximately 3 cm long segments, and subsequently integrated into the platforms. This approach enabled the assembly of multiple independent platforms from a single continuously fabricated fiber‐based tubular scaffold. Systematically designed experimental groups were compared with i) control conditions (Figure 6A.v), ii) cultures treated with THP‐1 monocytes alone (Figure 6A.vi), and iii) cultures treated with THP‐1 cells polarized toward either an IL‐4‐induced M2‐like phenotype (Figure 6A.vii) or lipopolysaccharide (LPS) induced M1‐like phenotype (Figure 6A.viii). By using this approach, the formation of CTCs that dissociate from breast cancer spheroids and enter the circulation through a vessel‐like structure was demonstrated. The on‐chip platform further facilitates the observation of the correlation between pro‐inflammatory and anti‐inflammatory conditions and the differences in cancer cell dissemination behavior. Additionally, this system permits a detailed investigation of how immune cells within the TME influence stromal organization and tumor spheroid morphodynamics. The layer‐by‐layer design of the BCoC configuration promoted laminar flow via peristaltic pump‐mediated perfusion and ensured the uniform positioning of cellular components. In addition to its compact and easy‐to‐assemble structure, the platform ensures long‐term sealing using screws and bolts, thereby maintaining cell viability during extended culture periods (up to 14 days). Moreover, its transparent design allowed for real‐time visualization of the TME during experimentation and contributed to enhanced cellular proliferation (Figure 6B). The RCCS bioreactor was used to mature the breast cancer spheroids (Figure 6C.i). This bioreactor allowed the spheroids to undergo less shear stress under microgravity conditions than by conventional methods. Additionally, a balanced distribution of mass transfer across all areas resulted in more uniform spheroids (Figure 6C.ii). We also demonstrated the versatility of the microgravity bioreactor in both high‐throughput organoid and spheroid maturation in our previous studies [66, 67]. Notably, the microgravity environment has been reported to provide significant advantages in the formation of proliferative and hypoxic zones by preserving the characteristic features of tumor spheroids [68, 69].

As described schematically, the vessel‐like structures were examined using a bright‐field inverted microscope before being positioned on the platform, and a homogeneous distribution of HUVEC was observed (Figure 6C.iii). After all the platform elements were successfully assembled, they were moved into an incubator, and the reservoirs provided the cells with a constantly fresh nutrient medium, and mechanical forces were applied via dynamic flow (Figure 6C.iv). Furthermore, a closer look at the platform demonstrated that breast cancer spheroids and vessel‐like structures were successfully positioned within the hydrogel matrix (Figure 6C.v, vi). In this way, propagation‐associated behaviors were successfully addressed in the presence of immune cells.

2.7. Feasibility Assessment of the Breast Cancer‐on‐a‐Chip Platform under Defined Immune Conditions

The next step was to validate the continuously perfused BCoC platform under immune cell co‐culture conditions and demonstrate its feasibility for capturing immune context‐dependent trends in propagation‐associated cell behavior over a 14‐day period. In this context, the quantification of wall shear stress in a vessel‐like structure is critical, as it represents the primary biomechanical cue exerted by fluid flow on endothelial cells and directly regulates endothelial barrier integrity, permeability, junction organization, hemostasis, mechanotransduction pathways, and tumor‐endothelial interactions, thereby controlling vascular function in microphysiological systems [70, 71, 72]. Thus, the theoretical wall shear stress was calculated based on the Hagen‐Poiseuille equation as ∼0.03 dyn/cm² (Section 4.2.4), indicating a low‐shear microenvironment within the vascular‐like structure, which falls within the range commonly reported for tumor‐associated microvasculature and low‐perfusion regions (∼0.01‐1 dyn/cm). Such low shear conditions are known to support endothelial permeability, tumor‐endothelial interactions, and cancer cell dissemination processes, making them suitable for modeling in vitro early metastatic events [70, 73]. For the validation, samples were collected from the reservoirs of four experimental groups: control (without THP‐1), THP‐1, THP‐1+LPS, and THP‐1+IL4 for both CTC and THP‐1 counts on days 7 and 14 (Figure 7). The culture supernatants from the on‐chip reservoirs were viewed under a fluorescence microscope to determine both CTCs and THP‐1 cell concentrations, and the effects of immune cells on cancer propagation‐related features in different phenotypes were qualitatively and quantitatively determined (Figure 7A,B). First, the number of blue fluorescence‐tracked CTCs was evaluated (Figure 7A,B.i). In the control group (i.e., without THP‐1), cells dissociated from breast cancer spheroids entered the perfusion, emulating blood circulation, and successfully formed CTCs by diffusing into the vessel‐like structure. Cell count increased over time from 6.4 × 10⁵ cells/mL on day 7 to 1.1 × 10⁶ cells/mL on day 14. Conversely, in the THP‐1 group, the presence of immune cells was associated with a significant reduction in CTC levels over time (from 4.3 × 10⁵ cells/mL, p < 0.05 on day 7 to 4.9 × 10⁵ cells/mL, p < 0.0001 on day 14 vs control), with an effect similar to that seen in in vivo circulation and remained almost constant over time. Furthermore, in the THP‐1+LPS group, where THP‐1‐derived M1‐like cells were induced with LPS, CTCs were further reduced under pro‐inflammatory‐like conditions, exhibiting a decrease over time from 3.7 × 10⁵ cells/mL on day 7 (p > 0.05 vs. THP‐1) to 2.9 × 10⁵ cells/mL on day 14 (p < 0.05 vs. THP‐1), due to the increase in THP‐1 cell numbers. In the THP‐1+IL4 group, CTCs persisted under anti‐inflammatory‐like conditions and increased significantly (8.9 × 10⁵ cells/mL, p < 0.0001 vs. THP‐1) on day 7 and showed the highest CTC concentration of 1.3 × 10⁶ cells/mL on day 14 (p < 0.0001 vs. THP‐1) due to M2‐like polarization of immune cells via an anti‐inflammatory‐like effect. Accordingly, the density of red fluorescent tracked THP‐1 cells was evaluated (Figure 7A,B.ii). As there were no THP‐1 cells in the reservoir of the control group, no quantification was performed. In contrast to the density of CTCs, the THP‐1 group exhibited a significant increase in immune cell proliferation over time, rising from 8.2 × 10⁵ cells/mL (p < 0.01) on day 7 to 2.1 × 10⁶ cells/mL (p < 0.0001) on day 14 compared to the control. The most pronounced increase was observed in the THP‐1+LPS group, where cell proliferation surged from 1.5 × 10⁶ cells/mL (p < 0.01) on day 7 to 2.9 × 10⁶ cells/mL (p < 0.01) on day 14 compared to THP‐1. Conversely, the presence of IL4 led to immune stability and a reduction in cell density over time, decreasing from 1.4 × 10⁶ cells/mL (p < 0.05) on day 7 to 9.8 × 10⁵ cells/mL (p < 0.0001) on day 14 in the THP‐1+IL4 group compared to THP‐1. Finally, the CTC‐to‐THP‐1 ratio was plotted (Figure 7B.iii), and the highest ratio was observed in the THP‐1+IL4 group on both days 7 and 14, directly reflecting the concurrent increase in CTC numbers (Figure 7B.i) and decrease in THP‐1 cell density observed under anti‐inflammatory‐like conditions (Figure 7B.ii). As a preliminary and correlative observation, this pattern suggests that a reduced immune cell presence under IL‐4 stimulation may be associated with greater cancer cell dissemination within this platform. THP‐1 cells exposed to LPS entered an M1‐like pro‐inflammatory activation state [74], commonly used as a simplified immune model in tumor‐immune interaction studies [75, 76]. As shown by our findings, LPS‐induced THPs exhibited increased proliferation and consequently reduced CTC numbers. In contrast, THP‐1‐derived M2‐like polarization is generated in the presence of IL‐4 or IL‐13 signaling and induces IL‐10 or TGF‐β release. This may increase VEGF release, and the associated anti‐inflammatory‐like effects may promote a propagation‐associated behavior [77]. In turn, this is consistent with the increased CTC concentration in our THP‐1+IL4 group, associated with a decreased THP‐1 cell count.

FIGURE 7.

Characterization of the vascularized‐like breast cancer‐on‐a‐chip platform under defined immune cell co‐culture conditions. (A) Representative time‐lapse (day 7 and day 14) microscopy images of THP‐1 cells and CTCs within the reservoir under control, LPS, and IL4 conditions. CTCs (Hoechst, Blue) and THPs (CellTracker, Red). Scale bar: 200 µm. (B) Quantification of THP‐1 cells and CTC numbers in the reservoir based on image analysis, showing altered cell proliferation and motility profiles in response to LPS (THP‐1‐derived M1‐like cells) or IL4 (THP‐1‐derived M2‐like cells) stimulation. Concentrations (c/mL, where “c” refers to “cells”) of (i) CTC and (ii) THP‐1 cells in reservoirs, and (iii) the ratio of CTCs to THP‐1 cells (n = 3). Significant differences: ^* p ≤ 0.05, ^** p ≤ 0.01, ^*** p ≤ 0.001, and ^**** p ≤ 0.0001. (C) ELISA analysis of cytokines; (i) IL‐6 (pg/mL), (ii) IL‐10 (pg/mL), and (iii) MMP2 (ng/mL) in the reservoir medium, demonstrating pro‐inflammatory (LPS) vs. anti‐inflammatory‐like (IL4) secretome signatures (n = 3). Significant differences: ^* p ≤ 0.05, ^** p ≤ 0.01, ^*** p ≤ 0.001, and ^**** p ≤ 0.0001. (D) Heatmap visualization of Quantitative Real‐Time Reverse Transcription (qRT‐PCR) analysis of gene expression based on fold regulation values from (i) reservoir cells and (ii) cellular chambers (n = 3).

To assess cytokine release, IL‐6, IL‐10, and MMP2, ELISA analyses were performed using samples collected from the reservoirs (Figure 7C). In terms of IL‐6 (Figure 7C.i), low levels were observed in the control group (day 7: 7.06 pg/mL; day 14: 7.10 pg/mL), suggesting that minimal basal inflammation may occur in the absence of immune cells, whereas IL‐6 levels increased significantly in the THP‐1 group (day 7: 24.47 pg/mL; day 14:1 9.95 pg/mL, p < 0.0001 vs. control). Notably, IL‐6 concentration in the THP‐1+LPS group was the highest among all groups on day 7 (39.91 pg/mL, p < 0.0001 vs. THP) and day 14 (20.65 pg/mL, p > 0.05 vs. THP‐1). This may indicate a strong pro‐inflammatory response corresponding to classical THP‐1‐derived M1‐like cells [78]. Conversely, on day 7, IL‐6 levels in the THP‐1+IL4 group were similar to those of the THP‐1 group (25.04 pg/mL vs. 24.47 pg/mL, respectively; p>0.05) but significantly lower on day 14 (11.44 pg/mL vs. 19.95 pg/mL, respectively; p < 0.01); thus, it is possible to speculate that M2‐like polarization conditions (i.e., anti‐inflammatory‐like status) suppressed the IL‐6 expression [79]. Over time, a general decrease in IL‐6 levels was observed in all groups on day 14, suggesting that inflammatory activity had reached equilibrium under long‐term culture conditions. On the other hand, IL‐10 levels reflected the anti‐inflammatory‐like nature of the immune microenvironment (Figure 7C.ii). While the control group had high IL‐10 levels due to the absence of immune cells (day 7: 170.96 pg/mL; day 14: 145.96 pg/mL), these values were slightly decreased in the THP‐1 group (day 7: 107.63 pg/mL, p < 0.001; day 14: 103.46 pg/mL, p < 0.05). This may reflect the natural cytokine secretion capacity of unstimulated THP‐1 cells. In the THP‐1+LPS group, IL‐10 levels were significantly lower than those in the THP‐1 group (day 7: 66.38 pg/mL, p < 0.05; day 14: 51.38 pg/mL, p < 0.01), suggesting the predominant pro‐inflammatory nature of the THP‐1‐derived M1‐like activation state. In the THP‐1+IL4 group, IL‐10 levels were significantly higher than those in all other groups on both days 7 (1217.63 pg/mL, p < 0.0001 vs. THP‐1) and day 14 (963.48 pg/mL, p < 0.0001 vs. THP‐1).

This finding is consistent with previous evidence demonstrating that IL‐4 enhances IL‐10 secretion by promoting the polarization of THP‐1 cells toward a THP‐1‐derived M2‐like phenotype [80, 81]. MMP2 levels showed that ECM remodeling changed depending on the immune conditions (Figure 7C.iii). In the control group, MMP2 levels increased during the culture period (day 7: 2.5 ng/mL; day 14: 3.02 ng/mL). In the THP‐1 group, a significant decrease in matrix destruction in the tumor microenvironment was observed (day 7: 2.13 ng/mL, p < 0.001; day 14: 2.46 ng/mL, p < 0.0001 vs. control group). In the THP‐1+LPS group, MMP2 levels decreased significantly from the early period compared to the THP‐1 group (day 7: 1.25 ng/mL, p < 0.0001), and a similar trend was observed throughout the culture period (day 14: 1.22 ng/mL, p < 0.0001) [82]. Conversely, cytokine levels remained significantly higher in the THP‐1+IL4 group at both time points (day 7: 2.38 ng/mL, p < 0.01; day 14: 2.56 ng/mL, p > 0.05 vs. THP‐1) and were similar to the control group. This suggests that ECM remodeling associated with invasive behavior was enhanced under IL‐4 stimulation [82], and our on‐chip model successfully captured key aspects of this phenomenon.

Furthermore, inflammatory and EMT‐related gene expression was evaluated by quantitative real‐time reverse transcription PCR (qRT‐PCR). Samples were collected from both the cellular chambers and reservoir supernatants of the BCoC platform. (Figure 7D, Table S1).

In the supernatants, several markers associated with endothelial activation, EMT, and matrix remodeling were upregulated, depending on the immune conditions. CD31 expression increased across all groups (THP‐1: 1.02‐fold, THP‐1+LPS: 2.25‐fold, THP‐1+IL4: 8.76‐fold), indicating endothelial involvement under perfusion (Figure 7D.i). EMT‐associated genes showed pronounced upregulation, with Vimentin markedly increased in the THP‐1 (78.14‐fold), THP‐1+LPS (55.95‐fold), and THP‐1+IL4 (128.56‐fold) groups. Similarly, MMP9 expression was significantly elevated in all conditions (THP‐1: 24.5‐fold, THP‐1+LPS: 15.6‐fold, THP‐1+IL4: 49.01‐fold), consistent with enhanced ECM remodeling and propagation‐related activity [83]. E‐cadherin levels also increased in all groups (THP‐1: 10.82‐fold, THP‐1+LPS: 9.94‐fold, THP‐1+IL4: 18.38‐fold), which may reflect dynamic epithelial‐mesenchymal plasticity rather than a unidirectional EMT process. The proliferation marker Ki67 was upregulated in the THP‐1+LPS (2.05‐fold) and THP‐1+IL4 (1.93‐fold) groups, while it remained downregulated in the THP‐1 group (1.03‐fold). In addition, the M2‐like polarization‐associated marker CD206 was strongly upregulated in the THP‐1+IL4 group (9.81‐fold), whereas it was downregulated in the THP‐1 (1.41‐fold) and THP‐1+LPS (3.33‐fold) groups. This may indicate effective immune phenotype modulation [84]. Conversely, certain endothelial junction markers are downregulated under pro‐inflammatory conditions. VE‐cadherin expression was reduced in the THP‐1 (1.60‐fold) and THP‐1+LPS (1.56‐fold) groups, and mildly upregulated in the THP‐1+IL4 group (1.12‐fold). N‐cadherin was upregulated in the THP‐1 (1.04‐fold) and THP‐1+LPS (1.11‐fold) groups, whereas it was downregulated in the THP‐1+IL4 group (3.23‐fold), potentially indicating immune‐dependent modulation of mesenchymal features. Overall, supernatant analyses indicated that IL4 exposure was associated with an invasive and ECM remodeling‐related expression profile characterized by the upregulation of Vimentin and MMP9, which are precursor genes for mesenchymal transition [83]. The pronounced upregulation of CD206 further supported the presence of a THP‐1‐derived M2‐like phenotype [84]. These findings were consistent with ELISA measurements and CTC/THP‐1 quantification results. Conversely, LPS stimulation induced a predominantly pro‐inflammatory gene expression pattern with comparatively limited EMT‐related activity.

Within the cellular chamber (Figure 7D.ii), several invasion‐ and signaling‐related genes were upregulated across all experimental conditions. MMP9 expression was significantly increased in all groups (THP‐1: 8.38‐fold, THP‐1+LPS: 3.06‐fold, THP‐1+IL4: 12.50‐fold), suggesting that active matrix remodeling plays a role at the tumor‐vasculature‐like interface. β‐catenin was also upregulated in all conditions (THP‐1: 6.26‐fold, THP‐1+LPS: 10.51‐fold, THP‐1+IL4: 5.74‐fold); thus, potentially indicating altered cell adhesion dynamics and signaling pathways associated with propagation and vascular interaction. Hypoxia‐related signaling was evident, as HIF1A expression increased in all groups (THP‐1: 2.11‐fold, THP‐1+LPS: 3.74‐fold, THP‐1+IL4: 5.88‐fold), confirming the presence of hypoxic regions in the cellular chamber. Consistent with spheroid‐based tumor characteristics, increased HIF1A expression in the presence of IL4 coincided with elevated MMP9 levels. This correlation may reflect hypoxia‐associated matrix remodeling tendencies rather than a defined mechanistic axis. In contrast, several endothelial and proliferation markers were downregulated in response to immune stimulation. ZO‐1 expression was reduced in the THP‐1 (3.45‐fold) and THP‐1+LPS (1.54‐fold) groups. However, they were upregulated under THP‐1+IL4 conditions (1.82‐fold). Similarly, CD31, which is expressed at endothelial cell‐cell junctions, was downregulated in the THP‐1 (1.96‐fold) and THP‐1+LPS (2.33‐fold) groups, while upregulated in the THP‐1+IL4 group (2.62‐fold), suggesting immune‐dependent regulation of endothelial stability. E‐cadherin expression showed a modest but consistent downregulation in the cellular chamber across all immune conditions (THP‐1: 1.23‐fold, THP‐1+LPS: 1.52‐fold, THP‐1+IL4: 1.47‐fold). While this pattern may be broadly consistent with reduced epithelial adhesion tendencies under perfusion and immune modulation, the relatively small fold changes should be interpreted with caution, and whether this reflects a meaningful shift toward a mesenchymal‐like phenotype remains to be confirmed. VE‐cadherin, which is expressed at the junctions between endothelial cells, was increased in all groups (THP‐1: 1.33‐fold, THP‐1+LPS: 1.52‐fold, and THP‐1+IL4:1.89‐fold). This may indicate the preservation of endothelial junctional components under perfusion. The inflammation‐related gene TNF‐α was downregulated in the THP‐1 (3.70‐fold) and THP‐1+IL4 (2.0‐fold) groups, whereas it was upregulated specifically in the THP‐1+LPS group (1.47‐fold), which may reflect classical M1‐like inflammatory activation. Ki67 expression was downregulated in all cellular chamber conditions (THP‐1: 1.11‐fold, THP‐1+LPS: 1.02‐fold, THP‐1+IL4: 1.18‐fold); thus, indicating that active proliferation within the cellular compartment was limited under these experimental conditions. Taken together, the coordinated regulation of β‐catenin signaling and endothelial markers under continuous perfusion supports the relevance of the vessel‐like structure, although it is a structurally simplified, single‐component endothelialized‐like conduit that does not recapitulate the full multilayered architecture of native vessels, including smooth muscle cells and perivascular supporting elements. This is a recognized scope boundary of the current platform, and more complex mural cell‐inclusive configurations are envisioned for future studies. In the same spirit, the biological experiments presented here are intended as proof‐of‐feasibility demonstrations to validate the ability of the BCoC platform to integrate tumor, vascular, and immune components and capture immune‐associated trends. While THP‐1 cells subjected to LPS or IL‐4 stimulation are widely used to model pro‐inflammatory and anti‐inflammatory‐like immune states, these conditions represent simplified and operationally defined polarization states rather than fully resolved macrophage phenotypes. Consequently, the observed changes in CTC levels, cytokine profiles, and EMT‐associated gene expression should be interpreted as correlative trends that reflect immune context‐dependent modulation within the platform, rather than as definitive mechanistic or pathophysiological conclusions. Future studies incorporating primary immune cells, additional phenotypic validation, and patient‐derived tumor models will further strengthen the translational relevance of this system.

3. Conclusions

In this study, a wet‐spinning biofabrication approach was employed to generate biomimetic, hydrogel‐based vessel‐like structures that support the formation of perfusable architectures within a 3D tumor microenvironment. As a use‐case application, this strategy enabled the development of an innovative breast cancer‐on‐a‐chip platform. By providing a controllable and reproducible microfluidic environment, the proposed platform offers a versatile biofabrication framework that can be leveraged to study propagation‐associated cellular interactions. The overall system enables preliminary observations of tumor‐vascular interactions and the investigation of cancer cell behaviors relevant to circulating tumor cell generation under dynamic perfusion conditions. The proposed on‐chip use‐case application serves as a foundation for future, detailed investigations into breast cancer propagation‐associated features. While recognizing that our platform intentionally employs a simplified conduit rather than a fully hierarchical, multilayered, native‐like vasculature and does not yet incorporate endothelial barrier assessment, such a design choice facilitates experimental controllability and integration within the on‐chip system. In turn, this allows for insightful investigations into propagation‐like processes under physiologically relevant conditions, as a proof‐of‐concept model. Moreover, our findings may serve as a preliminary tool for exploring drug responses and therapeutic testing in complex controlled in vitro settings. In the future, the use of patient‐derived cells could expand their application in personalized medicine, assisting in the selection of tailored therapeutic strategies and fostering the development of more effective and patient‐specific treatments. Additionally, a meaningful direction for future work is the systematic study of the endothelial barrier, particularly in research focused on cancer cell transmigration mechanisms.

4. Materials and Methods

All reagents were purchased from Merck (formerly known as Sigma–Aldrich, Germany) and used without further purification unless otherwise stated. Calcium chloride (CaCl2) was purchased from Eurochem BGD (Poland), and 4‐(2‐hydroxyethyl)‐1‐piperazine ethanesulfonic acid) (HEPES) was purchased from Roth GmbH (Germany). Live/Dead and cell tracker dyes were supplied by Invitrogen (Carlsbad, CA, USA). Primary and secondary antibodies were purchased from Santa Cruz Biotechnology, Affinity Bioscience, and Abcam (Cambridge, MA, USA). qRT‐PCR kits were purchased from Qiagen (USA).

4.1. Biofabrication of Hydrogel‐based Hollow Tubular Vessel‐Like Structures

4.1.1. Microfluidics‐Assisted Wet‐Spinning Setup

4.1.1.1. Shell Compartment

Sodium Alginate (ALG, UP VLVG Pronova, Mw < 75 kDA,) and gelatin methacryloyl (GelMA, high‐grade metacrylation), synthesized as already stated elsewhere [32], were blended and selected in agreement with fine‐tuning experiments (i.e., Alg 4% w/v and GelMA 5% w/v (hereafter referred to as A4G5)). Moreover, a two‐step crosslinking method was required for both ALG and GelMA (ionic and photocrosslinking, respectively). The hydrogel precursor formulation was prepared in HEPES 25 mm buffer and left overnight at 37°C under stirring to dissolve completely. Subsequently, lithium phenyl (2,4,6‐trimethylbenzoyl) phosphinate (LAP) photoinitiator (0.1% w/v) was added to the formulation, and the precursor was stirred further to allow for full dissolution.

4.1.1.2. Core Compartment

3% w/v gelatin (bovine skin, Type B) was chosen and kept as a sacrificial material (dissolution at 37°C) to obtain tubular‐like hollow constructs. Gelatin (3%) was prepared in HEPES (25 mm) and stirred until fully dissolved (50°C) prior to use, thus avoiding any early‐stage polymerization that eventually occurs at room temperature (RT).

4.1.1.3. Microfluidic Parameters

The pH of both the core and shell prepolymer solutions was adjusted to 7.4 to mimic the physiological conditions. Subsequently, the biomaterial inks were sterile‐filtered. The prepolymer solutions were loaded into two different sterile syringes for dual injection (Harvard Apparatus, USA, Dual Digital Pump). The fine‐tuned flow rates were set at 640 µL/min for the shell (Q_s) and 320 µL/min for the sacrificial core (Q_c), with Q_s = 2Q_c.

4.1.1.4. Co‐Axial Wet‐Spinning System and Experimental Setup

Both shell and core hydrogel precursors (Alginate‐GelMA and sacrificial Gelatin, respectively), were sterile‐filtered (0.22 µm syringe filters) prior to wet‐spinning. The complete wet‐spinning procedure, including fiber‐like extrusion, crosslinking, and handling, was conducted within a laminar‐flow biosafety cabinet routinely used for cell culture experiments, ensuring sterility throughout the fabrication process. The selected coaxial nozzle used for the wet‐spinning of the hydrogel‐based tubular conduit was commercially available (15G‐19G type: ID_shell 1.37 mm, ID_core 0.70 mm). The two inlets of the microfluidic co‐axial nozzle were connected using autoclaved polytetrafluoroethylene (PTFE) tubing to two different sterile syringes containing prepolymer solutions (shell and sacrificial core solutions). Subsequently, they were loaded onto a digital pump (Harvard Apparatus, USA). The independent flow rates for the wet‐spinning of the bioinks were set at 640 µL/min and 320 µL/min for the (cell‐laden) shell and cell‐free core, respectively. The prepolymer formulations were purged until they reached the tip of the nozzle. The coaxial nozzle was rapidly immersed in a sterile CaCl₂‐based coagulation bath (0.6 m, pH 7.4) kept at 4°C (liquid tank V = 100 mL) to wet‐spin meter‐long hydrogel‐based vessel‐like structures, allowing for the immediate ionic crosslinking of alginate and parallel physical gelation of gelatin. Herein, a meter‐long tubular scaffold was wet‐spun in the icy bath and subsequently removed from the tank using tweezers to undergo photopolymerization (DYMAX lamp (Torrington, CT, USA), 4 cm, 30 s, 800 mW/cm²) to crosslink the GelMA component. Finally, the hydrogel scaffolds were moved to incubated conditions (37°C, 5% CO₂) for culturing purposes and, in parallel, thermally dissolving the gelatin core (∼6 h). The obtained biomimetic vessels were subsequently cut into standardized 3 cm‐long samples using sterile disposable blades for experimental purposes. Each centimeter‐long segment was later integrated into an individual BCoC platform, thus minimizing batch‐to‐batch variability, as all specimens originated from the same parent construct.

4.1.2. Fine‐Tuning of Vessel‐Like Structures: Assessment of Spinnability, Injection and Perfusability

Two different co‐axial needles (15G19G: ID_shell 1.37 mm, ID_core 0.70 mm, and 17G22G: ID_shell 1.11 mm, ID_core 0.42 mm) and four different flow rates/coupling (640‐320, 640‐160, 320‐320, 320‐160 µL/min (i.e., Q_s‐Q_c shell‐core), respectively) were tested for the acellular wet‐spinning of the following shell hydrogel formulations: A2G5, A2G7, A4G5, and A4G5 (where A stands for Alginate and G for GelMA). Tubular scaffolds were wet‐spun under the above conditions (sacrificial core: 3% w/v gelatin). Subsequently, the vessel‐like conduits were cut (l = 3 cm) and incubated (37°C, 5% CO₂) to allow the gelatin to dissolve and create a hollow conduit. The performance in terms of stability upon manipulation (ST), ease of needle insertion (INS), and effective flow injection (INJ) was qualitatively considered and independently scored on a scale of 0 to 1. Subsequently, these three parameters were weighted according to Equation (1).

$$\mathrm{Score}(\%)=\mathrm{ST}\times (0.4\times \mathrm{INS}+0.6\times \mathrm{INJ})\times 100$$ (1)

These scores were visualized on a heat map to provide a spinnability window for selecting the optimal conditions and formulation (i.e., target material) for the biofabrication of wet‐spun vessel‐like structures.

4.1.3. Morphological Structure of Acellular Tubular Scaffolds via Scanning Electron Microscopy

The tubular scaffolds were washed thrice with deionized water (dH₂O) and dehydrated. To this end, vessel‐like structures were soaked in a series of increasing ethanol (EtOH) solutions (50%, 70%, 90%, and 100%) for 15 min each. The samples were then directly immersed in hexamethyldisilazane (HMDS) for 15 min and kept overnight under laminar flow to dry completely. The samples were mounted on aluminum stubs using double carbon tape and sputter‐coated with a 10 nm gold layer (Leica EM SCD 500, Germany). A 10 kV acceleration voltage and a secondary electron detector were employed. Observations were performed at different magnifications for both the top and cross‐sectional views (SEM, Axia ChemiSEM, Thermo Scientific, USA).

4.1.4. Swelling

The test was performed in HUVEC medium and HEPES buffer as a control. Upon wet‐spinning, the hydrogel‐based structures were cut (l = 3 cm), washed, immersed in dH₂O, and stored at −80°C overnight. Afterward, the constructs were dried by soaking each specimen in a series of concentrated ethanol solutions (50%, 70%, 90%, and 100%, 15 min each) and allowed to dry overnight. The dry mass was weighed (M_o). The samples were then immersed in the selected buffers supplemented with sodium azide (0.05% w/v; POCH S.A., Poland) as a bacteriostatic agent and incubated at 37°C (MaxQ 6000, Thermo Fisher Scientific, United States). The time points were set at 1, 3, 6, 9 h, and 7, 14, and 21 days, respectively. At each time point, the samples (n = 3) all wet samples were gently swabbed, weighed (M_w), and placed back in the buffer.

The punctual water content (W) was calculated using Equation (2).

$$\mathrm{W}(\%)=100·(({\mathrm{M}}_{\mathrm{w}}-{\mathrm{M}}_{\mathrm{d}})/{\mathrm{M}}_{\mathrm{d}})$$ (2)

4.1.5. Degradation

The overall test was performed in HUVEC medium, whereas samples in HEPES 25 mm were used as controls. Upon wetspinning, the tubular scaffolds were cut (l = 3 cm), washed, and immersed in dH₂O for storage at −80°C overnight. The vessel‐like structures were dried by soaking each specimen in a series of concentrated ethanol solutions (50%, 70%, 90%, and 100%, 15 min each) and were allowed to dry overnight. The dry mass was weighed (M_d0) and recorded for each sample. Subsequently, the samples were immersed in HUVEC medium, which was supplemented with 0.05% w/v sodium azide (POCH S.A., Poland) to prevent contamination, and incubated at 37°C (MaxQ 6000, Thermo Fisher Scientific, United States). At each time point (i.e., day 1,3, 7, 14, and 21), pertinent samples (n = 3) were washed thrice in dH₂0 and stored again at −80°C prior to the ethanol drying process (M_dt). The degradation rate (DR) was calculated according to Equation (3).

$$\mathrm{DR}(\%)=100·(({\mathrm{M}}_{\mathrm{d}0}-{\mathrm{M}}_{\mathrm{dt}})/{\mathrm{M}}_{\mathrm{d}0})$$ (3)

The RM was calculated by subtracting the punctual degradation rate from M_d0 (i.e., RM_t = 100 – DR_t).

4.1.6. Microcomputed Tomography of HUVEC‐Laden Tubular Scaffolds

At designated intervals, wet cell‐laden samples were initially fixed with 4% paraformaldehyde for 30 min, followed by three washes in 25 mm HEPES, each lasting 5 min. A disposable syringe equipped with a dedicated single‐use sterile needle was used to gently evacuate any liquid from the inner structure of the tubular scaffold by injecting air into it. The hydrogel conduits were then placed in 0.5 mL microcentrifuge tubes filled with 25 mm HEPES. The specimens were analyzed using a microfocused X‐ray tomographic system (MICRO XCT‐400, Xradia, Zeiss) at 40 kV and 250 µA. For each sample, 1000 projection images were captured with an exposure time of 5 s and a magnification objective of 4X. The volume was reconstructed using the instrument software and subsequently exported to Avizo Fire (FEI Visualization Sciences Group) for further three‐dimensional image analysis. Cross‐sectional and longitudinal images were used to assess the scaffold dimensions, specifically the overall diameter, core diameter, and wall thickness, at various time points (days 7, 14, and 21). The analysis of the microcomputed tomography (µCT)‐based images involved measuring the dimensions in different areas (n = 3) of each tubular scaffold using the ImageJ plugin (NIH, USA).

4.1.7. Scanning Electron Microscopy of HUVEC‐laden Tubular Scaffolds

Wet cell‐laden vessel‐like structures were fixed using paraformaldehyde 4% (30 min) at selected time points (days 7, 14, and 21) and then washed thrice with HEPES 25 mm (5 min/washing). Subsequently, dehydration and gold‐sputtering of the specimens were performed, similar to the acellular scaffolds (Section 4.1.3). The structural morphology of the constructs was investigated from both cross‐sectional and top views (SEM, PhenomProX, Eindhoven, The Netherlands) at various magnifications.

4.1.8. Cell Experiments

Human umbilical vein endothelial cells (HUVEC) were expanded in EGM‐2 cell culture medium supplemented with BulletKit (Lonza, Switzerland) until they reached confluence (P3‐P6). Trypsin‐EDTA 0.25% (Thermo Fisher Scientific, USA) was used to detach the cells. Cells were gently mixed within the A4G5 prepolymer solution prior to wet‐spinning (7.5 × 10⁶ cells/mL). Sterile A4G5 ink was added to the HUVEC pellet to obtain a well‐distributed bioink. After pipetting a few times, the cell‐laden bioink was withdrawn from the respective syringe while avoiding bubble formation. Similarly, sterile gelatin (3% w/v) was loaded into another syringe. Then, the PTFE tubings (ID = 0.8 mm) were connected to 1) the respective syringes (input source) and gently filled with prepolymer solutions, and 2) to the coaxial nozzle (output source: outer needle with Alginate‐GelMA and inner needle with Gelatin). Then, the microfluidic pump was activated for the wet‐spinning process according to the setup established for this research (see Section 4.1.1).

4.1.8.1. Brightfield Images of HUVEC‐Laden Tubular Scaffolds and Cell Viability Assessment

4.1.8.1.1. Cell Distribution

Optical images of cell‐laden tubular scaffolds were taken at three different time points (n = 3) (i.e., day 7, 14, and 21, respectively) in brightfield mode (PrimoVert digital microscope, Zeiss, Germany).

4.1.8.1.2. Cell Viability

At selected time points, 500 µL of a dual‐staining solution containing acridine orange (AO, Merck, USA) and ethidium bromide (EB, Merck, India) (100 µg/mL AO and 100 µg/mL EB) was added to each incubated construct immersed in 500 µL of HUVEC medium. AO stained both live and dead cells green (G), whereas EB stained only dead cells red (R). After staining, the samples were promptly imaged using a fluorescence microscope (Leica TCS SP8; Leica Systems, Switzerland) at the wavelengths specific to the fluorophores of interest. Fluorescence images (n = 3) were analyzed to determine the number of live and dead cells using the ImageJ counting algorithm (NIH, USA) in green and red channels, respectively. Viability was calculated using Equation (4).

$$\mathrm{Viability}(\%)=100\times ((\mathrm{G}-\mathrm{R})/\mathrm{G})$$ (4)

4.2. Fabrication and Assembly of the Breast Cancer‐on‐a‐Chip Model

4.2.1. Fabrication of the Microfluidic Platform

The microfluidic platform designed in our previous study was employed [67]. Briefly, the platform was fabricated using COMSOL Multiphysics software and constructed from laser‐cut polymethyl methacrylate (PMMA) molds (VLS2.30, Universal Laser Systems). Polydimethylsiloxane (PDMS; SYLGARD 184, Germany) was prepared at a 10:1 (v/v) base‐to‐curing agent ratio, degassed under vacuum, and cast into the molds. After curing at 50°C for 6 h, the PDMS components were demolded, and inlet/outlet silicone tubing (Longer Biosicon, 05.50.766) was integrated. The final platform, featuring a 10 mm × 30 mm × 3 mm (W×L×H) cellular chamber in the PDMS layer with total dimensions of 20 mm × 40 mm× 5 mm (W×L×H), was sterilized using ethylene oxide gas (16 h) and UV prior to use.

4.2.2. Formation of Breast Cancer Spheroids

4.2.2.1. Cell Culture

MDA‐MB231, HUVEC, HDF, and human monocytic cells (THP‐1) were obtained from the Biomimetic Microsystems Laboratory Biobank of the Department of Bioengineering at Ege University, Izmir, Turkey. HUVEC and HDF cells were adapted to and maintained in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% fetal bovine serum (FBS), 1% L‐glutamine (200 mm), 1% non‐essential amino acids, 0.5% gentamicin (10 mg/mL), and 0.5% penicillin/streptomycin (10,000 U–10 mg/mL). They were cultured in filtered cap T‐flasks under standard conditions in a humidified incubator at 37°C with 5% CO₂ and 95% air. MDA‐MB231 cells were maintained in supplemented Leibovitz's 15 (L‐15) medium and cultured in plugged‐cap T‐flasks, as CO₂ supplementation was not required. Subculturing of all cells at passage numbers 10–20 was performed using 0.05% trypsin‐EDTA upon reaching approximately 80%–90% confluence.

4.2.2.2. Spheroid Formation

MDA‐MB231 and HUVECs were seeded simultaneously but at a defined and fine‐tuned ratio for spheroid generation. Cells were introduced into a 55 mL RCCS‐Slow turning lateral vessel (STLV) (Synthecon Inc., USA) for dynamic self‐assembly conditions at a final density of 2 × 10⁵cells/mL. Both homotypic (MDA‐MB231 monoculture) and heterotypic (MDA‐MB231/HUVEC co‐culture) spheroids were generated, with heterotypic spheroids formed using a 1:1 MDA‐MB231:HUVEC ratio, as established in our previous study [67]. The vessel was completely filled with a 1:1 ratio of RPMI 1640 and L‐15 cell culture media, and all air bubbles were carefully removed using a syringe. The RCCS was operated at 10 rpm, gradually increased to 11 rpm as spheroids enlarged, and maintained in a humidified incubator at 37°C with 5% CO₂ and 95% air. The culture medium was refreshed twice a week by replacing three‐quarters of the total volume, and the system was maintained for 14 days.

4.2.3. Characterization of Spheroids

4.2.3.1. Live&Dead Staining

In addition to brightfield microscope (Motic AE31E, China) images, RCCS‐derived MDA‐MB231 and MDA‐MB231/HUVEC spheroids were freshly harvested and incubated in phosphate‐buffered saline (PBS) containing 2 µmol/L calcein AM and 4 µmol/L ethidium homodimer‐1 (Invitrogen, USA) for 45 min at RT in the dark. Following incubation, the spheroids were washed twice with PBS to remove excess dye. Live (green) and dead (red) cells were visualized using an inverted fluorescence microscope (Axio Vert.A1; Zeiss, Oberkochen, Germany).

4.2.3.2. Immunostaining

Spheroids were fixed with 4% paraformaldehyde (PFA; Sigma–Aldrich, Germany) at 4°C for 30 min, washed with PBS, and stored at 4°C. For staining, spheroids were transferred to V‐bottom plates, permeabilized with 0.5% Triton X‐100 for 15 min at RT, and blocked for 1 h at RT using a blocking solution containing 10% FBS, 1.15% glycine, 0.5% BSA, and 0.1% Tween‐20 in PBS. Primary antibodies against Ki‐67 and HIF‐1α (1:250, diluted in blocking solution) were applied and incubated overnight at 4°C on a shaker. After PBS washes, the spheroids were incubated with Alexa Fluor 488‐ and 647‐conjugated secondary antibodies (1:1000) for 1 h at RT, followed by DAPI staining (1.5 µg/mL) for 10 min. Final washes were performed and imaging was subsequently assessed using a confocal microscope (Zeiss LSM 880; Germany).

4.2.3.3. Spheroid Diameter Calculation

Spheroid diameters were measured using the ImageJ software (version 1.53; NIH, USA). The average diameter (l) of each spheroid was calculated as the geometric mean of two orthogonal diameters (a and b, respectively) using the following equation (Equation 5):

$$\mathrm{l}=\surd (\mathrm{a}\times \mathrm{b})$$ (5)

4.2.4. Integration of Vessel‐Like Structure Into the Platform and Creation of Tumor Microenvironment

GelMA (5% w/v) and photoinitiator Irgacure 2959 (1% w/v) were separately dissolved in the co‐culture medium at 70°C (70:30 volume ratio) and then combined at 40°C to prepare the prepolymer solution. The solution was filtered (0.22 µm) and mixed with stromal HDF cells at a density of 1 × 10⁵ cells/mL for ECM recapitulation. A thin layer of cellularized GelMA was applied to a PDMS‐based cellular chamber. Prefabricated vessel‐like structures (∼3 cm in length, corresponding to the platform dimensions), which were carefully UV crosslinked and sectioned prior to integration, were inserted and fixed with GelMA in place via UV crosslinking (30 s, 800 mW/cm², 4 cm) (Omnicure, Canada). Pre‐formed RCCS‐derived MDA‐MB231/HUVEC spheroids (approximately 20 per platform), previously stained with blue fluorescent Hoechst dye (H6024, Sigma–Aldrich), were positioned bilaterally around the vasculature and embedded in an additional thin layer of cellularized GelMA, followed by photocrosslinking. The chamber was sealed with a PDMS top layer and sandwiched between upper and lower PMMA plates measuring 5.5 cm × 4 cm, which were secured with screws and nuts for a leak‐proof assembly. Separate media reservoirs were filled with either co‐culture medium (control), THP1 cells (prelabeled with 10 µM red CellTracker fluorescent dye, at a density of 1 × 10⁵ cells/mL)‐seeded medium, or THP1‐seeded medium supplemented with either LPS (10 ng/mL) or IL‐4 (10 ng/mL) to simulate anti‐inflammatory‐like microenvironments. Continuous perfusion was maintained using peristaltic pumps at a flow rate of 7.3 µL/min, a condition previously fine‐tuned in our earlier study to ensure continuous flow while preventing cell washout from the vascular lumen [67]. In our BCoC model, wall shear stress within the microfluidic vascular‐like structure was estimated by assuming steady, incompressible, and fully developed laminar flow governed by the Hagen‐Poiseuille equation. Thus, the theoretical shear stress (τ) at the vessel‐like channel wall was calculated using Equation (6).

$$\tau =4\mu \mathrm{Q}/(\pi {\mathrm{r}}^3)$$ (6)

where µ is the dynamic viscosity of the culture medium (RPMI at 37°C, approximately 0.7 cP = 0.007 dyne · s/cm²) [85], Q is the volumetric flow rate (7.3 µL/min = 1.22 × 10⁻⁴ cm³/s), and r is the inner radius of the vessel‐like structure (r = 0.0325 cm from the vessel‐like structure diameter, which is approximately 0.65 mm). This approach is widely used for characterizing microfluidic vascular models under laminar flow conditions [70, 73]. The platforms were then incubated at 37°C with 5% CO₂ for 14 days.

4.2.5. Characterization of the Breast Cancer‐on‐a‐Chip Model

4.2.5.1. CTCs and Immune Cell Tracking

On days 7 and 14, samples were collected from the reservoirs of the microchip platform using a serological pipette, homogenized, and loaded onto a Neubauer slide for quantification of CTCs and immune cells. Cells shed from the blue Hoechst‐stained spheroids and appearing as CTCs were counted under a fluorescent microscope with a blue filter, whereas red‐labeled THP‐1 cells in the reservoirs were counted using a red filter. The cell concentration (cells/mL) at each time point was calculated using a standard cell counting formula.

4.2.5.2. ELISA Assay

Culture supernatants (n = 3) were collected from the chip platforms, and ELISA was performed to evaluate MMP2 (for EMT and matrix degradation) and IL‐6/IL‐10 (for inflammatory responses). Samples were analyzed using commercial human ELISA kits according to the manufacturer's instructions (Roche, Basel, Switzerland). All samples were centrifuged at 3000 rpm for 20 min to remove cellular debris, and the resulting supernatants were used for subsequent analyses. Absorbance was measured at 450 nm using a spectrophotometric microplate reader (BioTek ELx800, USA). The concentrations of MMP2 (ng/mL), IL‐6 (pg/mL), and IL‐10 (pg/mL) were calculated from the standard curves.

4.2.5.3. qRT‐PCR Analysis

Total RNA was isolated using the RNeasy Mini Kit (Qiagen, 74104, Germany). To assess RNA purity and quality, the concentration of each isolated sample was measured, and purity ratios (A260/A280 and A260/A230) were determined using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, USA). Measurements were performed using 2 µL of each RNA sample. cDNA synthesis was performed using the RT² First Strand Kit (Qiagen, 330404, Germany) [86] following the manufacturer's instructions. RNA samples were adjusted to a concentration of 100 ng/µL before the reaction. The two‐step protocol provided by the kit was followed. Genomic DNA elimination mixture (10 µL) was added to each sample, followed by incubation at 42°C for 5 min, and then placed on ice for at least 1 min. Subsequently, 10 µL of reverse transcription mixture was pipetted into each tube containing 10 µL of genomic DNA elimination mixture, followed by incubation at 42°C for 10 min. The reaction was then terminated by incubation at 95°C for 5 min. PCR amplification was performed under the following thermal cycling conditions: the reaction was initiated with one denaturation cycle at 95°C for 2 min, followed by 45 amplification cycles consisting of denaturation at 95°C for 15 s and annealing/extension at 60°C for 1 min. Gene expression changes were evaluated based on the obtained Ct values, and fold change and fold regulation values were calculated using the 2^−ΔΔCt method.

4.3. Statistical Analysis

The results are expressed as the mean ± standard deviation. Two‐way ANOVA with Tukey's multiple comparison tests was performed using GraphPad Prism analysis software (GraphPad Software, La Jolla, CA, USA) to assess the statistical differences between data populations. Differences were considered statistically significant at p < 0.05. Statistically significant values are presented as ns, ^* p ≤ 0.05, ^** p ≤ 0.01, ^*** p ≤ 0.001, and ^**** p ≤ 0.0001.

Author Contributions

Alessia Paradiso: conceptualization, methodology, investigation, data curation, visualization, validation, formal analysis, writing – original draft, writing – review & editing. Pelin Saglam‐Metiner: conceptualization, methodology, investigation, data curation, visualization, validation, formal analysis, writing – original draft, writing – review & editing, project administration. Ewa Walejewska: conceptualization, methodology, investigation, formal analysis, project administration. Marina Volpi: conceptualization, methodology, investigation. Basar Dogan: methodology, investigation, data curation, visualization, writing – original draft. Yagmur Filiz: conceptualization, methodology, investigation, data curation, visualization. Ipek Sarier: investigation, formal analysis, writing – original draft. Diana C. Martinez: investigation, data curation. writing – original draft. Reyhan Coban: methodology, investigation. Leila Sabour‐Takanlou: methodology, investigation. Cigir Biray‐Avci: methodology, investigation, project administration. Ozlem Yesil‐Celiktas: conceptualization, writing – review & editing, supervision, project administration, resources, funding acquisition. Wojciech Swieszkowski: conceptualization, writing – review & editing, supervision, project administration, resources, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adhm71317‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.