Harnessing inter-spheroid spacing and structural connectivity to direct collective cell migration and host vessel integration in 3D engineered tissue Authors: Sangmin Lee, SeongHeon Bae, Hyunseok Kwon, Ji Min Seok, Su A Park, Heungsoo Shin Published: 2026 ABSTRACT AbstractEngineering functional vascular networks within three-dimensional (3D) tissues remains a major challenge in regenerative medicine. Here, we introduce a hybrid platform that integrates spheroids with gelatin methacryloyl (GelMA) hydrogels and 3D-printed poly(ε-caprolactone) (PCL) scaffolds to generate pre-vascularized 3D constructs. The scaffold architecture was designed with spatially defined chambers and interconnected pores incorporating catching strands that anchor spheroid positioning and direct cell migration from embedded spheroids. Additionally, the GelMA matrix establishes a permissive microenvironment conducive to cell adhesion and matrix remodeling, and the PCL framework preserves macroscopic structural integrity. Co-culture spheroids composed of human umbilical vein endothelial cells (HUVEC) and human adipose-derived stem cells (hADSC) enhanced sprouting and collective migration compared with suspended single cells, and also provided significant upregulation of extracellular matrix (ECM) remodeling genes and pro-angiogenic markers. Spatially distributed spheroids with 3D structure strongly modulated vascular network morphogenesis, whereby the inter-spheroid distance finely tuned cell migration dynamics and ECM remodeling. Scaffold microarchitecture further governed host tissue integration in vivo, with higher pore interconnectivity facilitating superior vascular infiltration and increased co-localization between host and engineered vessels. Notably, host vessels aligned preferentially along the inferior catching strands, suggesting geometry-associated vascular integration. Collectively, these findings demonstrate that precise spatial organization of spheroids combined with tailored 3D-printed scaffold architecture supports robust pre-vascularization of engineered tissues, offering a promising strategy for a variety of biomedical applications. Introduction A hierarchically organized three-dimensional (3D) vascular network with different diameters is essential for the delivery of oxygen and nutrients as well as the removal of metabolic waste [1]. Thus, cells located beyond ∼200–250 μm from functional vasculature often experience hypoxia and eventual necrosis [2]. Consequently, the induction of vascular structures represents a fundamental requirement for ensuring long-term structural and functional stability in engineered 3D tissues. In addition, vascularization promotes rapid integration with host vasculature post-implantation, thereby, improving engraftment efficacy [3,4]. Insufficient vascularization, conversely, leads to hypoxic microenvironments, central necrosis, and inflammatory responses that impair tissue integration, leading to implant failure [5]. Several strategies have been explored to promote vascularization in engineered tissue, including sustained delivery of pro-angiogenic factors [6], endothelial–stromal co-culture systems [7], and bioengineering approaches such as microfabricated vascular architectures or host-mediated pre-vascularization [8]. Despite these advances, generating stable vascular networks with controlled spatial organization and heterogeneous vessel diameters remains challenging, highlighting the need for engineering strategies that integrate cellular interactions with precisely controlled structural and biochemical cues. Co-culture systems facilitating communication between multiple cell types have been widely used to generate functional pre-vascularized 3D tissues [8]. Endothelial cells (ECs) and mesenchymal stromal cells such as fibroblast-like stromal cells, mesenchymal stem cells (MSCs), and pericytes cooperatively regulate microvascular network formation and stabilization through paracrine signaling and direct physical interactions [9]. MSCs promote vascular maturation by secreting pro-angiogenic factors, differentiating into pericyte-like cells, and remodeling the extracellular matrix (ECM), while cell–cell contacts further support vascular stability [10]. However, conventional co-culture systems within 3D hydrogel environments often suffer from limited cell–cell contact, non-uniform cell distribution, and insufficient control over the cellular microenvironment, thereby restricting stable vascular network morphogenesis [8,11]. 3D co-culture spheroid systems have emerged as promising alternatives that enhance cell–cell and cell–ECM interactions [12,13]. For example, spheroids composed of endothelial colony-forming cells and MSCs have enhanced endothelial sprouting in collagen matrices and promoted vascularization after in vivo implantation [14,15], while spheroids of umbilical vein endothelial cells (HUVECs) and human adipose-derived stem cells (hADSCs) encapsulated in gelatin methacryloyl (GelMA) hydrogels enhanced host vessel infiltration [16]. Nevertheless, robust and reproducible formation of stable vascular networks within 3D hydrogel environments remains challenging. Engineered matrices supporting both biological and mechanical prerequisites for angiogenesis are essential for pre-vascularized tissue formation [17]. ECM-mimetic hydrogels such as polyethylene glycol-, collagen-, and gelatin-based matrices have enabled matrix metalloproteinase (MMP)-mediated remodeling that supports endothelial sprouting and lumen formation [[18], [19], [20]]. MMP-degradable hydrogels, however, often lack sufficient mechanical stability for extended culture. This limitation is particularly pronounced in spheroid-based systems, where inadequate mechanical support resulted in positional instability, matrix contraction, and heterogeneous spheroid distribution, thereby impairing vascular alignment [21,22]. Hybrid architectures combining compliant hydrogels with mechanically supportive scaffolds have been developed to maintain matrix remodeling while improving structural stability and vascular invasion [23]. Pore architecture played a critical role in vascularization efficiency; interconnected pores of ∼150–300 μm facilitated cell infiltration and endothelial sprouting, whereas small or poorly connected pores restricted oxygen transport and lumen formation [[24], [25], [26]]. Accordingly, hybrid scaffold systems integrating spheroids with mechanically supportive architectures represented a promising strategy for stable and functional pre-vascularized tissue formation [27]. Geometry-guided migration plays an important role in vascular morphogenesis by directing cellular orientation and organized tissue formation [28]. Micropatterned hydrogels and spatially controlled matrices can guide migration paths for endothelial cells and their sprouting orientation, resulting in aligned microvascular structures [22]. In addition, spheroid spacing has been shown to regulate inter-spheroid sprouting interactions and cellular bridge formation in 3D matrices. Appropriate spacing can provide sufficient physical separation for outward sprout extension while maintaining proximity for cellular interaction between neighboring spheroids, thereby contributing to aligned pre-vascular pattern formation [29]. Geometry-associated vascularization has also been implemented through engineered architectures such as aligned fibers, microchannels, and controlled pore networks that facilitated organized vascular branching and perfusable lumen formation [25,26]. Although these studies have highlighted the importance of geometric cues, most approaches have addressed individual factors such as contact guidance or spheroid spacing independently. Integrated platforms simultaneously incorporating matrix remodeling, controlled spheroid positioning, and geometry-associated organization within 3D scaffold architecture remain scarce. Given that, the objective of this study is to establish a hybrid platform integrating co-culture spheroids, GelMA hydrogels, and 3D-printed poly(ε-caprolactone) (PCL) scaffolds to create an engineered framework wherein mechanical support coexists with biological remodeling. This system adheres to the following design principles: (i) spheroid positioning within 3D printed scaffolds to provide mechanical stabilization and long-term culture viability; (ii) tailored spheroid spacing and scaffold pore architecture, including pore size and interconnectivity, to regulate migration paths via spatial cell migration and cellular bridging; (iii) the use of GelMA hydrogels containing intrinsic MMP-sensitive sites to provide a cell-responsive matrix that enables cell-driven ECM remodeling and sprout invasion; and (iv) lateral entry points along scaffold sidewalls and internal pores to enhance convective–diffusive transport and rapid inosculation with endogenous vessels. Through engineering of the spheroid layout (orientation and spacing) and the scaffold microarchitecture within the PCL/GelMA constructs, this platform provides reproducible ECM remodeling and pre-vascularization, addressing several limitations of existing strategies, such as poor reproducibility, inadequate structural stability, and insufficient spatial control, to improve early cell survival and functional maintenance in thick implantable tissues. Materials and methods Poly(caprolactone) (PCL) pellets were purchased from Sigma-Aldrich (Sigma, St. Louis, MO, USA), distilled water (DW) was obtained from an Elix Advantage system (Millipore, MA, USA). Tris-HCl was purchased from Shelton Scientific (Peosta, IA, USA). Phosphate-buffered saline (PBS), Dulbecco's phosphate-buffered saline (DPBS), and trypsin-ethylenediaminetetraacetic acid (EDTA) were purchased from Welgene (Gyeongsan-si, Korea). hADSCs, MSC growth medium, fetal bovine serum (FBS), and penicillin/streptomycin (P/S) were purchased from CEFOBIO (Seoul, Korea). HUVECs and endothelial growth medium-2 (EGM-2) were purchased from LONZA (Walkersville, MD, USA). A 96-well cell floater plate was obtained from SPL Life Sciences (Gyeonggi-do, Korea). GelMA was obtained from 3D Materials (Busan, Korea; type A, porcine skin, gel strength ∼300 bloom, degree of methacrylation >90%) and 2-hydroxy-2-methylpropiophenone (Darocure, photo-initiator) was purchased from Sigma-Aldrich. Antifade Mounting Medium with 4′,6-diamidino-2-phenylindole (DAPI) was obtained from Vectashield® (Burlingame, CA, USA), Hoechst 33342 and DiI/DiO dyes were purchased from Invitrogen (Carlsbad, CA, USA). A RNeasy mini kit, Maxime RT premix, and SYBR premix Ex-Taq were purchased from Qiagen (Valencia, CA, USA), Intron (Seoul, Korea), and TAKARA (Otsu, Japan), respectively. Preparation and characterization of GelMA hydrogel with PCL scaffold A scaffold of PCL (Mn = 45,000) was fabricated using an extrusion-based printing approach employing a 3D bioprinter (constructed in-house at the Korea Institute of Machinery and Materials). The instrument and the fabrication process were described previously [30,31]. Briefly, the 3D printing system featured a heating jacket, a 200-μm-diameter nozzle, a pressure pump with an x-y-z stage, and computer software. PCL pellets were melted in a heating cylinder at 100 °C for 30 min. The strand sizes and between-strand distances were varied using a computer-aided design (CAD) model followed by computer-aided manufacturing. This allowed layer-by-layer printing of a complex, interconnected 3D structure. The scaffold size was a 5 mm (x-axis) × 5 mm (y-axis) × 2 mm (z-axis) cuboid structure with a circumscribed circle composed of 10 layers. The strand thickness and between-strand distance were 200 and 800 μm, respectively. The printing speed was 300 mm/min, and the pneumatic pressure was 420 kPa. The 3D-printed PCL scaffolds were dispersed in a dopamine hydrochloride solution (2 mg/mL in 10 mM Tris-HCl buffer, pH 8.5) at a concentration of 2 mg/mL. Subsequently, the samples were incubated at 37 °C with gentle agitation (50 rpm) for 30 min and washed twice with DW. The prepared PCL and polydopamine (PD)-coated PCL scaffolds were then dried overnight using the built-in UV light of a clean bench at room temperature (R.T.), and their surfaces were observed with scanning electron microscopy (SEM, Hitachi S-4800, Hitachi, Ltd, Tokyo, Japan). In addition, the PD-coated surface was analyzed using attenuated total reflection (ATR) (Nicolet 6700, Thermo Fisher Scientific, MA, US) to identify the peaks corresponding to each bond. The GelMA solution (7.5% w/v in DPBS) was filtered using a syringe filter and mixed with 2-hydroxy-2-methylpropiophenone (0.1% v/v). A PD-coated PCL scaffold was positioned in a custom-made polytetrafluoroethylene (PTFE) mold (3 cm width x 3 cm height x 0.1 cm depth) and encapsulated with the prepared GelMA solution (10 μL). Then the samples were crosslinked using a UV spot-curing system (300 mW; UV SMT, Bucheon, Korea) for 10, 30, or 60 s. Bulk hydrogels were prepared by casting 100 μL of hydrogel solution into a cylindrical PTFE mold (8 mm diameter, 2 mm deep), and their storage modulus was measured using a rheometer (HR10, TA Instruments, New Castle, DE, USA) under an axial force of 0.5 N and 1% strain. The blue-colored hydrogel inside the PCL scaffold was obtained by mixing GelMA solution with 0.1% (v/v) trypan blue and encapsulating it within the PCL scaffold. The hADSCs were cultured in DMEM (with 1% P/S and 10% FBS) under standard culture conditions (5% CO2 and 37 °C). HUVECs were cultured in EGM-2 under standard culture conditions. The culture medium was refreshed every two days, and cells were passaged upon reaching over 80% confluency. Cells with passage numbers between 5 and 7 were used for all experiments. The hADSCs and HUVECs were harvested using Trypsin-EDTA and adjusted to a cell density of 1 × 106 cells/mL. Subsequently, hADSCs and HUVECs were combined to obtain a mixed cell suspension at a 3:1 ratio (ADSCs 3 × 106 cells: HUVECs 1 × 106 cells). Subsequently, the mixed cell suspension was dispensed into each well of an ultra-low attachment 96-well plate, such that each well contained 3 × 104 hADSCs and 1 × 104 HUVECs. Centrifugation was performed at 1000 rpm for 5 min, and the cells were allowed to incubate overnight at 37 °C in a 5% CO2 incubator. Co-culture spheroids were cultured in a co-medium consisting of DMEM and EGM-2 at a ratio of 1:1, following previously reported hADSC/HUVEC co-culture conditions [32,33], and the medium was refreshed every two days. The distribution of cells within the co-culture spheroids was evaluated by labeling the hADSCs and HUVECs with DiO and DiI dyes, respectively. Briefly, 1 × 106 hADSCs and 1 × 106 HUVECs detached from the culture dish were washed to remove residual medium, followed by staining with DiO or DiI dyes diluted to a final concentration of 5 μM in fresh medium. The cells were then incubated at 37 °C in a 5% CO2 incubator for 20 min. After overnight incubation, which led to spheroid formation, the following day was defined as Day 1, and the formed spheroids were imaged using a fluorescence microscope (Eclipse Ti2, Nikon, Tokyo, Japan). The viability of the fabricated co-culture spheroids on Day 1 was confirmed by staining the samples with a live/dead solution (Invitrogen, Carlsbad, CA, USA), consisting of 2 μM calcein-AM and 4 μM ethidium homodimer-1 (EthD-1) working solutions. The calcein-AM solution was added to the samples and incubated for 1 h at 37 °C. Additionally, co-culture spheroids composed of either 2 × 104 or 4 × 104 cells at a 3:1 ratio were fabricated using the same method described, and their diameters were quantified by outlining the outer boundary of each spheroid using ImageJ software. Measurements were performed on 10 individual spheroids. The two different sizes of spheroids were placed within the scaffold chamber, and images were acquired using a phase contrast microscope. Fabrication of co-culture cell/spheroid-laden GelMA hydrogel with PCL scaffold In the suspension group, a cell suspension containing 160,000 cells (equivalent to four 40,000-cell spheroids) was encapsulated within 10 μL of GelMA hydrogel and dispensed onto a PD-coated PCL scaffold. In the spheroid group, individual spheroids were manually transferred into the PD-coated PCL scaffold chamber using a pipette, followed by addition of 10 μL of the GelMA solution the next day. Samples were crosslinked for 10 s using a UV spot-curing system, and UV intensity (300 mW) was measured using a UV meter (IEI-Fastcure 250, Awellcure, Shanghai, China). After fabrication, the prepared samples were cultured for 7 days under standard culture conditions. Cell viability was determined by incubating each scaffold with 500 μL of PBS solution containing 0.5 μM calcein-AM and 4 μM EthD-1 at 37 °C for 30 min, followed by fluorescence imaging from the bottom of the scaffold using a fluorescence microscope. The relative gene expression of ECM remodeling markers [Matrix Metalloproteinase-9 (MMP-9), Matrix Metalloproteinase-14 (MMP-14)] and angiogenic markers [Endomucin (EMCN), Slit Guidance Ligand 3 (SLIT-3), Granulocyte Colony-Stimulating Factor (G-CSF)] in co-culture spheroids were analyzed on Day 7 through real-time reverse transcription polymerase chain reaction (RT-PCR). For the analysis, GelMA/PCL scaffolds were dissolved in collagenase solution (100 CDU/mL, CDU = collagen digestion units) for 1 h at 37 °C. The supernatant was discarded after centrifugation at 1200 rpm for 3 min, and mRNA of the co-culture spheroid was extracted using an RNA extraction kit. The total RNA concentration was measured using a nano-spectrometer (Nanodrop 2000; Thermo Scientific, Wilmington, DE, USA), and cDNA was synthesized using a 1 μg of RNA solution and a Maxtime RT PreMix kit. The cDNA was diluted with diethyl pyrocarbonate-treated water (1:1), mixed with SYBR Green PCR Mastermix, and RT-PCR was performed using a StepOnePlus thermocycler (Life Technology, Carlsbad, CA, USA). The primers used for RT-PCR had the following sequences: GAPDH (Fw: 5′- GTC AGT GGT GGA CCT GAC CT -3′, Rv: 5′- TGC TGT AGC CAA ATT CGT TG -3′), MMP-9 (Fw: 5′ - CGT CGT GAT CCC CAC TTA CT -3′, Rv: 5′- AAC ACA CAG GGT TTG CCT TC -3′), MMP-14 (Fw: 5′- AGT CAG GGT CAC CCA CAA AG -3′, Rv: 5′- GCA TTG GGT ATC CAT CCA TC -3′), EMCN (Fw: 5′- CCG GTG GTT ATT GCT TTG AT -3′, Rv: 5′- TGC AGA GTG CTC ACC AGA CT -3′), SLIT-3 (Fw: 5′- CTT GTC TCC CTG TCC ACCC AT -3′, Rv: CTG GAT GGG AAT CTC CTT GA -3′), G-CSF(Fw: 5′- GCT TGA GCC AAC TCC ATA GC -3′, Rv: 5′- TCC CAG TTC TTC CAT CTG CT -3′). For hematoxylin and eosin (H&E) staining, the sectioned specimens were serially hydrated by reversing the dehydration steps (xylene, 100%-70% ethanol, and water), stained with hematoxylin for 2 min and eosin for 6 min, dehydrated, and then mounted. The H&E image was captured using a phase contrast microscope. Effect of spheroid positioning on cell migration within GelMA hydrogel-PCL scaffolds Two experimental groups were established to evaluate the impact of catching strand orientation on cells migrated from spheroids. In Group 1, two spheroids were positioned parallel to the direction of the catching strands, while in Group 2, spheroids were positioned perpendicular to the catching strands. Phase contrast images were captured under a 10× objective lens. On days 3 and 7 using a phase contrast microscope under standard culture conditions, focusing on regions containing the two spheroids and the intermediate area between them to assess the migration of cells originating from the spheroids. Samples were fixed with 4% paraformaldehyde and embedded in paraffin blocks. Cross-sections (10 μm thickness) were prepared from the paraffin blocks using a microtome (Leica Biosystems GmbH, Wetzlar, Germany) to obtain specimens from both the upper regions that contained spheroids and catching strands. Sectioned specimens were collected sequentially. The “Upper” region was defined as the area ∼200 μm from the top surface of the scaffold, corresponding to the first 20 sections (each 10 μm thick) obtained at the beginning of the sectioning process. Similarly, the “Lower” region was defined as the area ∼200 μm from the bottom of the scaffold, corresponding to the last 20 sections collected at the end of sectioning. The specimens underwent deparaffinization by immersion in xylene, followed by gradual dehydration through sequential immersions in ethanol solutions of decreasing concentration (100%, 90%, 80%, and 70%). After dehydration, specimens were treated with a blocking buffer (5% donkey serum and 0.1% Tween-20 in PBS), mounted using a mounting medium with DAPI, and then observed with a fluorescence microscope. The region of interest (ROI) was defined as the scaffold's inner chamber area between the two spheroid-containing chambers, with dimensions of ∼1 × 1 mm2. The number of DAPI-positive nuclei within this ROI was counted using ImageJ software. The experimental groups, suspension cells, spheroid (0 mm), and spheroid (1 mm) were cultured under standard conditions for 7 days. The suspension cell group consisted of two types of cells (hADSCs and HUVECs) encapsulated within the hydrogel, while in the 0 mm group, two spheroids were positioned adjacent to each other within the chamber, and in the 1 mm group, the spheroids were placed 1 mm apart within the chamber. Subsequently, the orientation angles of cells either attached to or located around the catching strands were quantified using ImageJ software. The angle of each catching strand was set as 0°, and the orientation of individual cells was represented as a line to measure the angle between the cell alignment and the strand. Cells with orientation angles between −30° and +30° were classified as parallel, those between −90° to −80° and +80° to +90° as perpendicular, and the remaining regions were evenly divided and classified as random. Results were expressed as the percentage of cells within each angular range. Influence of spheroid positioning distance on ECM remodeling and vascularization Spheroids were positioned within scaffold chambers measuring 800 × 800 μm2. The spheroid group without spacing between adjacent spheroids was defined as 0 mm, while additional groups were defined with incremental spacing intervals of ∼1 mm, designated as 1 mm, 2 mm, and 3 mm based on the standard chamber spacing. Each group was cultured under standard conditions for 7 days, after which cell migration and viability were assessed. On day 7, the samples were treated with calcein-AM solution and incubated at 37 °C for 1 h. Subsequently, samples were carefully flipped using forceps, and both the upper side without catching strands (Top) and the lower side containing catching strands (Bottom) were imaged using a fluorescence microscope under a 10× objective lens. The percentage area occupied by calcein-AM positive cell area within the intermediate scaffold area (ROI, 1 cm2) between the two spheroids was measured. Transcript expression levels of ECM remodeling markers [MMP-2, MMP-9, MMP-14), Tissue Inhibitor of Metalloproteinases-4 (TIMP-4)] and angiogenic markers [CD31, EMCN, SLIT-3, and G-CSF] in co-culture spheroids were analyzed through RT-PCR. For the analysis, PCL/GelMA constructs were dissolved in collagenase solution (100 CDU/mL, CDU = collagen digestion units) for 1 h at 37 °C. The supernatant was discarded after centrifugation at 1200 rpm for 3 min. The mRNA of co-culture spheroids was extracted using an RNA extraction kit. The total RNA concentration was measured using a nano-spectrometer (Nanodrop 2000; Thermo Scientific, Wilmington, DE, USA), and cDNA was synthesized using 1 μg of RNA solution and the Maxime RT PreMix kit. The cDNA was diluted with diethyl pyrocarbonate-treated water (1:1), mixed with SYBR Green PCR Mastermix, and PCR was performed using a StepOnePlus thermocycler (Life Technology, Carlsbad, CA, USA). The primers used for RT-PCR had the following sequences: TIMP4 (Fw: 5′- TGG CCA GAT TCT CAG TGA TG -3′, Rv: 5′- ATA GAG CTT CCG TTC CAG CA -3′), CD31 (Fw: 5′- GCA AAA TGG GAA GAA CCT GA -3′, Rv: 5′- CAC TCC TCC CAC CAA CAC CT -3′). Effect of pore size in spheroid-laden GelMA hydrogel-PCL scaffolds on collective cell migration in mouse subcutaneous implantation Two types of scaffolds with different pore sizes, referred to as “Single” and “Double”, were prepared to examine their influence on cell migration and vascularization. Both were fabricated with dimensions of 3 × 5 × 2 mm3. The Single scaffold consisted of 10 alternating layers, each layer composed of strands 200 μm in diameter, while the Double scaffold was composed of 10 layers stacked in pairs, with two consecutive layers oriented in the same direction before alternating orientation. The difference in pore sizes between the two scaffold designs were observed by SEM. Pore areas located on the external surfaces of fabricated scaffolds were defined as “External”, whereas pore areas measured from cross-sectional images inside the scaffolds were defined as “Internal” as shown in Fig. 6a and b. Characterization of two types of scaffolds with different pore sizes, Single (smaller pores) and Double (larger pores) scaffolds. (a) Schematic of fabrication of single-layer (Single scaffold) and double-layer (Double scaffold) structure. (b) Conceptual illustration of pore size differences. (c) SEM images comparing single- and double-layer scaffolds. (d) Quantification of internal and (e) internal pores sizes (n = 3, samples). Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. Animal management procedures were approved by the Institutional Animal Care and Use Committee of Hanyang University (HY-IACUC-25-0054) and conducted according to IACUC guidelines (approval number: 2024-0042A). The following experimental groups of scaffolds were used: 1) Single, Parallel (Single scaffold, catching strand direction is parallel with the position of spheroids), 2) Single, Perpendicular (Single scaffold, catching strand direction is perpendicular with the position of spheroids), 3) Double, Parallel (Double scaffold, catching strand direction is parallel with the position of spheroids), and 4) Double, Perpendicular (Double scaffold, catching strand direction is perpendicular with the position of spheroids). Initially, co-culture spheroids composed of hADSCs and HUVECs at a 3:1 ratio were positioned 1 mm apart at the center of Single and Double scaffolds, each measuring 3 × 5 × 2 mm3, under standard culture conditions. On the following day, GelMA encapsulation was conducted, and scaffolds were pre-cultured in vitro for two weeks to ensure sufficient pre-vascularization. Before surgery, each mouse received an intraperitoneal injection of 150 μL of anesthetic solution composed of zoletil (60 mg/kg) and Rompun (6 mg/kg). Nude, 8-week-old female mice weighing 20.0 ± 5.0 g (Narabiotech; Seoul, Korea) were used for subcutaneous implantation, with 5 mice per experimental group. Following implantation, the incision sites were sutured and covered with Tegaderm™ (3M, Saint Paul, MN, USA). After four weeks, the mice were euthanized by CO2 inhalation, and skin tissues containing scaffolds were harvested, fixed in 10% formalin for at least 48 h rinsed in PBS, and subsequently stored in PBS at 4 °C until further analysis. For histological analyses, two sectioning methods were applied as described in Fig. 7b: longitudinal and transverse sections of scaffolds. As described in Section 2.3, samples underwent paraffin embedding, molding, deparaffinization, and dehydration processes. Sections were stained with hematoxylin for 5 min and eosin (BBC Biochemical, Mount Vernon, MA, USA) for 8 min and subsequently imaged using a phase-contrast microscope under a 10× objective lens. Cell migration through pores and interconnected regions between scaffold chambers was measured. The area (x 104 μm2) was calculated by normalizing the pixel values obtained from the eosin-positive region segmented using Adobe Photoshop (Adobe Systems, San Jose, CA, USA). Effect of pore size in spheroid-laden PCL/GelMA constructs on cell migration in mouse subcutaneous implantation. (a) Schematic timeline of in vivo transplantation and analysis. (b) Schematic representation of histological sectioning. (c) H&E stained longitudinal and transverse sections of single parallel and double parallel scaffolds. (d) Quantification of interconnected area in longitudinal section (n = 15, samples). (e) Quantification of interconnected area in transverse section (n = 12, samples). (f) Fluorescent images of HNA/DAPI staining in longitudinal sections. (g) Quantification of the number of cells in ROI (yellow box) (n = 10, samples). (h) Quantification of HNA positive cells in ROI (yellow box) (n = 5, samples). Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) Immunofluorescence staining was performed for human nuclear antigen (HNA) and DAPI to further evaluate cell migration and human cell distribution. Following deparaffinization, dehydration, and blocking as described previously (Section 2.3), samples were incubated for 1 h at R.T. with goat F(ab) anti-mouse IgG H&L antibody (1:10 dilution in blocking buffer; washed three times with PBS; ab6668, Abcam, Cambridge, UK) to block mouse IgG binding sites and minimize background signal from the secondary antibody. Subsequently, samples were incubated overnight at 4 °C with a mouse monoclonal anti-human nuclei antibody (1:100 dilution in blocking buffer, washed three times in PBS; MAB1281, Millipore, Burlington, USA). The following day, samples were incubated for 1 h at R.T. with fluorescein-conjugated goat anti-mouse IgG (H&L) secondary antibody (1:100 dilution in blocking buffer, washed three times in PBS) and counterstained with Hoechst 33342 solution (5 μg/mL). Images were captured using a fluorescence microscope under a 10× objective lens. The ROI was defined as the interconnected scaffold area located between the two spheroids. To define the ROI, the outer boundary of the DAPI-positive nuclei within this inter-spheroid region was manually traced, and the enclosed area was designated as the ROI. Within this region, the total number of DAPI-positive nuclei was counted, and the number of HNA-positive cells was subsequently quantified. The percentage of HNA-positive cells was then calculated by normalizing to the total number of DAPI-positive nuclei. All quantifications were performed using ImageJ software. Effect of spheroid-laden GelMA hydrogel-PCL scaffolds on vascularization in mouse subcutaneous implantation The effect of scaffold pore size differences on in vitro pre-vascularization and integration with host vessels was assessed. Dual staining for human CD31 (engineered vessel) and mouse CD31 (host vessel) was performed to assess host vessel integration at both the top and bottom surfaces. Paraffin sections (240-μm thickness) were prepared from paraffin blocks using a microtome following the methods described in Section 2.3. The first two sections obtained ∼500 μm from the top of the scaffold were designated as “Top (open)”, and the last two sections obtained ∼500 μm from the bottom were designated as “Bottom (strand)”. Following deparaffinization and dehydration (as described in Section 2.3), sections underwent blocking and overnight incubation at 4 °C with rabbit anti-human CD31 (PECAM) antibody (1:200 dilution in blocking buffer, washed three times in PBS; ab32457, Abcam, Cambridge, UK). Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibody (1:100 dilution in PBS, washed three times in PBS; ab150061, Abcam, Cambridge, UK) was then incubated for 1 h at R.T. Subsequently, rat anti-mouse CD31 Alexa Fluor 647-conjugated antibody (1:100 dilution in PBS, washed three times in PBS; ab307262, Abcam, Cambridge, UK) was incubated for 1 h at R.T. Finally, sections were stained with Hoechst 33342 solution (5 μg/mL) and imaged using a confocal laser microscope (Stellaris 8 STED FALCON, Leica). Imaging was conducted with an HC PL APO CS2 10×/0.40 DRY objective, laser lines at 405 nm (3.2% intensity), 499 nm (12.2% intensity), and 653 nm (21.6% intensity), a pinhole size of 53.1 μm, and a Z-stack interval of 3 μm. Images were processed using LAS X software (Leica Microsystems, Germany). The orientation of host vessels was quantified by measuring the angles between vessel segments and the catching strands (defined as 0° reference axis) using ImageJ software. For each section, individual vessel structures (stained positive for mouse CD31) were manually segmented by tracing a straight line along their major protrusion axis. Each line was treated as an independent data point (n = 1). The angle formed between this vessel line and the horizontal catching strand was measured using the angle tool in ImageJ. In the Bottom region, where vessels were more elongated and interconnected, major linear branches that could be clearly distinguished were selected for measurement, and care was taken to exclude regions where segmentation was ambiguous. The collected angles were then categorized and visualized in dot plots. Immunofluorescence staining for CD31 and alpha-smooth muscle actin (α-SMA) was performed as described in Section 2.3. Paraffin sections (10 μm thick) were subjected to deparaffinization, dehydration, and blocking. Sections were incubated overnight at 4 °C with rabbit anti-human CD31 (PECAM) antibody (1:200 dilution), followed by incubation with Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibody (1:100 dilution). Hoechst 33342 solution (5 μg/mL) was used for counterstaining, and samples were imaged using a fluorescence microscope. For α-SMA staining, sections underwent overnight incubation at 4 °C with rabbit anti-human α-SMA antibody (1:100 dilution; PA5-85070, Thermo Fisher Scientific, Waltham, MA, USA), followed by incubation with Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibody (1:100 dilution). Samples were counterstained and observed with the same method. Samples were divided into Top and Bottom regions based on a defined midpoint set at 1 mm from the bottom of the scaffold, corresponding to half the total scaffold height (2 mm). This midpoint was used to separate the sample into two equal regions, and quantification was performed only within areas containing positive signals in each respective region. The length of vessels marked by linear CD31-positive signals was measured, while curved signals were quantified using the segmented line tool in ImageJ. Each dot in the dot plot represented the length of a single CD31-positive signal fragment. In the case of curved signals, the total length was calculated by summing the lengths of each linear segment along the curve, and this cumulative value was represented as a single data point in the plot. Vessel areas marked by α-SMA-positive signals were quantified by calculating the percentage of positive pixels relative to the total area using Adobe Photoshop. Specifically, for each well-defined circular region, the α-SMA-positive area within the central hollow (inner region) was measured in pixels and normalized to the total region size. The quantified area of a single α-SMA-positive hollow region was represented by a dot in the dot plot. Finally, the co-localization between engineered and host vessels was evaluated. Fixed scaffolds were incubated in permeable blocking buffer (5% donkey serum and 0.3% Triton X-100 in PBS) for at least 3 days. Subsequently, samples were incubated for 2 days at 4 °C with rabbit anti-human CD31 antibody (PECAM; 1:200 dilution in blocking buffer, washed three times in PBS), followed by 4 h incubation at R.T. with Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibody (1:100 dilution). Next, rat anti-mouse CD31 Alexa Fluor 647-conjugated antibody (1:100 dilution in PBS, washed three times in PBS) was incubated for 2 days at 4 °C. After counterstaining with Hoechst 33342 solution (5 μg/mL), samples were imaged using a confocal laser microscope. Imaging parameters included an HC PL APO CS2 10×/0.40 DRY objective, laser lines at 405, 499, 655 nm (4.0%, 25.4%, and 16.4% intensity, respectively), a pinhole size of 53.1 μm, and a Z-stack interval of 5 μm. Images were processed using LAS X software. Co-localized vessel areas, identified as yellow signals resulting from the overlap of red and green fluorescence channels, were quantified using Adobe Photoshop. Yellow regions were masked to isolate the co-localized signals, and the area occupied by these regions was measured in pixels and summed for quantification, such that each dot in the dot plot represents the total co-localized area within a single image. Vessel thickness was performed using ImageJ software by manually selecting representative co-localized vessels from the image. The thickness of each individual vessel was measured perpendicularly across its width, and each dot in the dot plot corresponds to the thickness of a single vessel. Statistical analysis Quantitative data are presented as mean ± standard deviation. Statistical analyses were performed using GraphPad Prism 11 software (La Jolla, CA, USA). Comparisons between two groups were analyzed using Student's t-test. Comparisons involving more than two groups with a single independent variable were analyzed using one-way analysis of variance (ANOVA) followed by Tukey's honest significant difference post hoc test. Experiments involving two independent variables were analyzed using two-way ANOVA. A p value < 0.05 was considered statistically significant. Results In this study, we developed a hybrid strategy for engineering pre-vascularized tissues using endothelial–stromal co-culture spheroids encapsulated within hybrid PCL/GelMA scaffolds as described in Fig. 1. This platform integrated a mechanically robust 3D-printed PCL framework, which was filled with a crosslinked GelMA matrix. The 3D-printed PCL scaffold preserved construct geometry and generated interconnected porosity that spatially localized spheroids while preventing bulk contraction during culture. Complementarily, the GelMA matrix supported cell adhesion, migration, and matrix MMP-mediated remodeling, enabling ECM reorganization during vascular morphogenesis. To recapitulate vasculogenic hierarchy, HUVEC:hADSC (1:3 ratio) co-culture spheroids were prepared to promote endothelial–supporting cell paracrine signaling and coordinated vascular assembly. The effect of spatial distribution of spheroid within 3D printed constructs and their structural parameters such as pore size, interconnectivity, and strand geometry on vascular organization and host vessel infiltration have been systematically evaluated via in vitro–in vivo experimental processes. Schematic representation of the co-culture spheroids-laden PCL/GelMA constructs. The co-culture spheroids, consisting of hADSCs and HUVECs at a 3:1 ratio, are manually translocated within the system. By adjusting spheroid distance, spheroid direction, and scaffold side porosity, cell migration, ECM remodeling, and angiogenesis can be regulated. This approach induces pre-vascularization in vitro and confirms vascular anastomosis between engineered vessel and host vessel after in vivo implantation. Characterization of GelMA hydrogel with PCL scaffolds Scaffolds (5 × 5 × 2 mm3) featuring internal chambers (800 × 800 μm2) and 200 μm diameter strands were designed using CAD software (Fig. 2a). Catching strands spanning chamber openings were incorporated to prevent spheroid displacement during encapsulation and culture. Fabricated scaffolds faithfully reproduced the design, including catching strands (Fig. 2b). Top and bottom layers exhibited consistent pore alignment and uniform strand thickness, indicating high printing fidelity. Uniform PD coating on the scaffold surface was confirmed with SEM, revealing a continuous coverage across the PCL strands (Fig. 2c). Surface characterization with ATR spectroscopy corroborated the PD deposition, revealing characteristic peaks corresponding to O–H (3200–3600 cm−1), C=O (1726 cm−1), and C=C (1500–1600 cm−1) stretching (Fig. 2d). PD coating was essential for the fabrication of spheroid-laden GelMA/PCL constructs, as unmodified PCL scaffolds remained at the air–water interface whereas PD-coated scaffolds became submerged in water (Supplementary Fig. S1). GelMA hydrogel stiffness was tuned by crosslinking time, resulting in moduli from 0.61 ± 0.03 kPa (10 s) to 1.03 ± 0.10 kPa (60 s) (Fig.g 2e). A 10 s crosslinking time was selected because previous studies using the same GelMA formulation demonstrated that compliant hydrogels (∼0.5 kPa) promoted greater cellular sprouting, migration, and matrix remodeling from spheroid-based constructs than stiffer hydrogels (∼1.2 kPa) generated by longer UV exposure times [34]. Under the GelMA concentration we used in our study, slight change in crosslinking affected mechanical properties of resulting hydrogels. Trypan blue–labeled GelMA solution confirmed the uniform filling of the PD-coated scaffold, forming the GelMA–PCL construct (Fig. 2f). Confocal microscopy was used to confirm a homogeneous cell distribution in the co-culture spheroids (hADSCs(DiO):HUVECs(DiI) = 3:1) (Fig. 2g). Live/dead staining revealed high cell viability with predominant green fluorescence (Fig. 2h). Spheroids of 40,000 cells averaged 575.8 ± 24.0 μm in diameter (∼1.5-fold larger than the 20,000-cell spheroid). Within scaffold chambers, 40,000-cell spheroids were stably retained by the catching strands, whereas the 20,000-cell spheroids showed positional instability (Fig. 2i). Additional representative images of 20,000- and 40,000-cell spheroids positioned within scaffold chambers are provided in Supplementary Fig. S2. Characterization of GelMA hydrogel with PCL scaffold and co-culture spheroids. (a) CAD image. (b) Optical images of PCL scaffold (Top and Bottom views) (c) SEM images comparing pristine and PD-coated PCL surfaces. (d) Fourier-transform infrared spectra (measured using ATR) of pristine and PD-coated PCL. (e) Storage modulus of GelMA hydrogels crosslinked under different UV exposure times (n = 3, samples). (f) Optical images of Trypan-blue-stained GelMA hydrogel encapsulated within the PCL scaffold. (g) Confocal images of co-culture spheroid (DiO-labeled hADSCs and DiI-labeled HUVECs). (h) Fluorescent live/dead staining of co-culture spheroids using calcein-AM and EthD-1. (i) Analysis of spheroid diameter based on cell number (n = 10, samples). Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. Effect of spheroid-laden GelMA hydrogel-PCL scaffold on cell migration compared to suspension cells Co-culture spheroids were positioned within scaffold chambers, encapsulated in 7.5% (w/v) GelMA, and crosslinked by localized UV irradiation to generate PCL/GelMA constructs (Fig. 3a). The detailed procedure for manual spheroid transfer, positioning, GelMA encapsulation, and UV crosslinking is illustrated in Supplementary Fig. S3. After 7 days of culture, live/dead staining confirmed high viability in both suspension and spheroid conditions. However, the suspension cells retained rounded morphologies, whereas the spheroids displayed pronounced sprouting from their periphery (Fig. 3b). hADSCs and HUVECs in the suspension condition were homogeneously distributed and maintained a circular morphology throughout the chamber. In contrast, in the spheroid condition, cells were observed at the spheroid boundary and extending outward into the surrounding PCL/GelMA constructs (Fig. 3c). Whole-well fluorescence images of the spheroid group at Days 1, 3, and 7 are provided in Supplementary Fig. S4. Analysis of DiO-labeled hADSCs and DiI-labeled HUVECs further revealed that hADSC-derived signals occupied significantly greater volume and projected area than HUVEC-derived signals within the migrating regions at Days 1 and 3 (Supplementary Fig. S5), although both cell types contributed to the sprouting structures by collective migration. ECM remodeling genes were significantly upregulated in the spheroid vs. suspension cells: MMP-9 (2.6 ± 1.2-fold) and MMP-14 (51.4 ± 17.8-fold) (Fig. 3d). Pro-angiogenic markers (EMCN, SLIT3, and G-CSF) were also significantly elevated by 2.2 ± 0.1, 2.4 ± 0.6, and 6.1 ± 1.5-fold in spheroid cultures, respectively (Fig. 3e). H&E staining at Day 4 showed active cell proliferation and migration from spheroids into the chambers (Fig. 3f). Representative H&E-stained images of the suspension cell condition at Day 4 are provided in Supplementary Fig. S6. A schematic illustration summarizing the spheroid-induced vascularization within the GelMA matrix is presented in Fig. 3g. Comparison of suspension cells and spheroids within a GelMA hydrogel with a PCL scaffold. (a) Schematic procedure of fabricating PCL/GelMA constructs. (b) Fluorescent live/dead staining of suspension cells and spheroid on days 1 and 7 using calcein-AM and EthD-1. (c) Representative fluorescence images at Day 1 showing the spatial distribution of hADSCs (green) and HUVECs (red) under suspension and spheroid conditions within the PCL/GelMA constructs. (d) Gene expression analysis of ECM-remodeling markers (n = 3, independent experiments). (e) Gene expression analysis of endothelial and pro-angiogenic markers (n = 3, independent experiments). (f) Histological analysis of spheroid-laden GelMA hydrogel with PCL scaffold after 4 days of culture. (g) The appearance of suspension cells and spheroids within the GelMA hydrogel. Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. Effects of spheroid orientation on cell migration within GelMA hydrogel-PCL scaffolds Spheroids were positioned, being oriented parallel (Group 1) or perpendicular (Group 2) relative to catching strands, with spheroids placed one chamber apart (approximately 1 mm spacing) as depicted in Fig. 4a. Phase-contrast imaging showed cell migration within each chamber in both groups at Day 3. Notably, by Day 7, in Group 2, middle chambers remained unoccupied, whereas Group 1 cells migrated along strands into the middle chamber filled with GelMA hydrogel (Fig. 4b). Fluorescence imaging of the parallel orientation (Group 1) at Day 3 showed that hADSCs and HUVECs extended from spheroids and followed the catching strands (Fig. 4c). Bottom-view images of the parallel orientation (Group 1), stained for CD31, F-actin, and DAPI, showed that CD31-positive cells were distributed between spheroids positioned with 1 mm spacing, with cell coverage observed not only along the catching strands but also extending into the surrounding hydrogel regions (Fig. 4d). DAPI-positive nuclei within the defined region of interest (ROI) increased from 67 ± 8 (Upper) and 153 ± 39 (Lower) at Day 3 to 160 ± 30 (Upper) and 322 ± 73 (Lower) at Day 7 (Fig. 4e and f). Corresponding DNA quantification showed no substantial increase in total DNA content from Day 1 to Day 7 (Supplementary Fig. S7), suggesting that the increased cellular density within the ROI was associated primarily with cell redistribution and migration rather than proliferation-driven expansion. For orientation analysis, spheroids located in chambers immediately adjacent to the parallel-aligned catching strands were defined as 0 mm, whereas those positioned one chamber away (approximately 1 mm distance) were defined as 1 mm. Analysis of cell orientation in the lower regions revealed strand-attached cells migrated parallel across conditions (suspension, spheroid 0 mm, and spheroid 1 mm). Additionally, the surrounding cells showed random patterns in suspension/Spheroid 0 mm, but parallel migration in Spheroid 1 mm (Fig. 4g). Effect of spheroid positioning within PCL/GelMA constructs on cell migration. (a) Schematic representation of spheroid positioning in different groups. (b) Phase contrast images showing cell migration within the scaffold on days 3 and 7. (c) Representative fluorescence image at Day 3 showing hADSCs (green) and HUVECs (red) extending from spheroids and following the catching strands. (d) Representative fluorescence image at Day 7 showing CD31 (green), F-actin (red), and DAPI (blue) within regions between spheroids positioned with 1 mm spacing. Insets (blue dashed boxes) indicate higher magnification views of the region between spheroids. (e) DAPI-stained fluorescent images of Upper and Lower regions of the scaffolds on days 3 and 7, displayed in grayscale. (f) Quantification of DAPI-positive nuclei per area (n = 4, samples). (g) Spheroid orientation and migration patterns within the scaffold (n = 3, samples). Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. Influence of spheroid spacing on ECM remodeling and vascularization Four experimental groups with spacings of 0 (directly adjacent), 1, 2, and 3 mm were prepared and tested (Fig. 5a). After 7 days of culture, live/dead staining and fluorescence imaging showed strong calcein-AM signals and minimal dead-cell signals across all groups at both the top and bottom scaffold surfaces with stronger bottom-surface signals in the middle chambers (indicated by the yellow box, Fig. 5b). The middle-chamber ROI exhibited cytoskeletal coverage varied by spacing: top (39.3 ± 16% at 1 mm; 14 ± 2.4% at 2 mm; 2.2 ± 0.5% at 3 mm) and bottom (68.1 ± 3.7% at 1 mm; 28.4 ± 20% at 2 mm; 2.8 ± 1.1% at 3 mm) (Fig. 5c). RT-PCR analysis of ECM remodeling genes (MMP-2, MMP-9, MMP-14, TIMP-4) showed the highest expression at 1 mm spacing (Fig. 5d). MMP-2 and MMP-14 were upregulated ∼ threefold and 1.5-fold, respectively, at 1 mm compared with 0 mm, whereas their expression decreased at 2 mm and 3 mm. TIMP-4 exhibited the strongest upregulation (∼10-fold) at 1 mm. Expression of pro-angiogenic markers (CD31, EMCN, G-CSF, and SLIT3) showed a similar trend, with the highest expression at 1 mm spacing (Fig. 5e). In particular, SLIT3 increased by more than 10-fold and CD31 by ∼ two-fold compared with 0 mm, followed by decreased expression at larger spheroid separations. Corresponding quantitative analyses with individual data points and SD values are provided in Supplementary Fig. S8. Effect of spheroid spacing within 3D printed scaffold filled with GelMA hydrogels on ECM remodeling and vascularization. (a) Schematic representation of spheroid positioning within the scaffold. (b) Fluorescent live/dead staining of spheroids in the top and bottom layers on days 1 and 7 using calcein-AM and EthD-1. (c) Quantification of spheroid area in the top and bottom layers (n = 3, samples). (d) Heatmap of ECM remodeling-related gene expression (n = 3, samples). (e) Heatmap of vascularization-related gene expression (n = 3, samples). Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. Effect of pore size in spheroid-laden GelMA hydrogel-PCL scaffolds on cell migration in mouse subcutaneous implantation Single (smaller pores) and Double (larger pores) scaffolds (3 x 5 × 2 mm3) were fabricated by 3D printing as shown in Fig. 6a. Representative macroscopic images of the fabricated scaffolds are provided in Supplementary Fig. S9. SEM images confirmed distinct porosity, with Double exhibiting a more open architecture (Fig. 6b and c). The internal and external pore areas were 0.5 ± 0.3 and 1.6 ± 0.6 (x104 μm2) for the Single scaffold, and 4.3 ± 1.1 and 10.1 ± 2.4 × 104 μm2 for the Double scaffold (Fig. 6d and e). Additional cross-sectional SEM images provided detailed visualization of the interconnected pore structures in both Single and Double scaffolds. Furthermore, EDS elemental mapping identified the presence of nitrogen-containing regions on both external and internal scaffold surfaces, suggesting homogenous polydopamine coating throughout the scaffold architecture (Supplementary Fig. S10). Subcutaneous implantation of spheroid-positioned scaffolds was performed to evaluate cell migration in vivo (Fig. 7a), followed by longitudinal and transverse histological analysis (Fig. 7b). H&E staining revealed cell infiltration between adjacent chambers in both scaffolds, with more extensive interconnections in the Double group (Fig. 7c). In longitudinal sections, inter-pore regions appeared as narrow, partially filled tissue bridges in the Single scaffolds, whereas broader and more continuous tissue connections were observed in the Double scaffolds. Transverse sections similarly showed greater tissue occupation within pore spaces and along scaffold walls in the Double scaffolds, whereas larger unoccupied scaffold-derived spaces with limited tissue infiltration were observed in the Single scaffolds. These differences were reflected in increased longitudinal (1.6 ± 0.5 to 3.0 ± 1.4 × 104 μm2) and transverse interconnectivity (2.5 ± 0.6 to 4.4 ± 1.0 × 104 μm2) (Fig. 7d and e). HNA and DAPI staining further showed broader distributions of HNA-positive human cells in the Double scaffolds (Fig. 7f and g), whereas in the Single scaffolds, these cells were primarily localized near the chambers where spheroids were initially positioned. Accordingly, the total number of DAPI-positive nuclei was significantly greater in Double scaffolds (241 ± 74) than in Single scaffolds (118 ± 61) (Fig. 7h). Effect of spheroid-laden GelMA hydrogel-PCL scaffolds on vascularization in mouse subcutaneous implantation Post-implantation confocal imaging (hCD31/mCD31) revealed scattered mCD31 signals at the top-surface without linear hCD31 vessels, appearing as discontinuous, punctuate signals with no clear directional organization. The bottom regions showed mCD31 vessels parallel to catching strands, overlapping the hCD31 structures (Fig. 8a and b), forming elongated structures along the strand axis. Quantification of the angle between infiltrated mCD31-positive vessels and catching strands showed a broad distribution at the top surfaces (62.4 ± 51.8°) and a highly aligned orientation at the bottom surfaces (11.0 ± 11.0°) (Fig. 8c). Elongated hCD31-positive vessel-like structures were observed in longitudinal sections of both Single and Double scaffolds (Fig. 8d), with greater continuity and extension observed in the bottom regions and in the Double scaffolds. Vessel length was longer in the Bottom than in the Top regions in both scaffold types (Single: Top 0.3 ± 0.1 mm, Bottom 2.4 ± 0.3 mm; Double: Top 0.5 ± 0.2 mm, Bottom 0.8 ± 0.3 mm), with overall greater vessel lengths in the Double scaffolds compared to the Single scaffolds as measured from longitudinal sections (Fig. 8e). α-SMA-positive green fluorescence was predominantly observed around the bottom catching strands in both types, with broader distributions in the Bottom regions versus the Top regions in transverse sections (Fig. 8f), surrounding hCD31-positive regions, whereas in the Top regions, α-SMA signals were sparse and discontinuous, α-SMA-positive luminal areas were significantly larger in the Bottom regions for both Single and Double scaffolds from transverse sections (Fig. 8g), with greater α-SMA-positive luminal areas observed in the Double scaffolds compared to the Single scaffolds. Confocal imaging further showed co-localization of engineered vessels (hCD31) and host vessels (mCD31) at the scaffold interfaces at lateral entry points along scaffold sidewalls and internal pore regions (Fig. 8h), and co-localized vascular areas were significantly greater in the Double scaffolds (4.7 ± 0.7 × 104 μm2) than in the Single scaffolds (2.0 ± 0.4 × 104 μm2) (Fig. 8i). In addition, the vessel thickness at sites of co-localization followed the same trend (Fig. 8j). Vascularization of spheroid-laden PCL/GelMA constructs. (a) Schematic illustration of host vessel infiltration at the top and bottom regions. (b) Confocal images of vascularization at the top and bottom regions. (c) Quantification of vessel orientation angle (n = 20, samples). (d) Fluorescent images of CD31/DAPI stained longitudinal sections in the Single and Double scaffolds. (e) Quantification of vessel length (n = 15, samples). (f) Fluorescent images of α-SMA/DAPI stained transverse sections in the Single and Double scaffolds. (g) Quantification of vessel area (n = 10, samples). (h) Co-localization of engineered vessels and host vessels in the Single and Double scaffolds at lateral entry points along scaffold sidewalls and internal pore regions. (i) Quantification of the co-localized vessel area within these regions (n = 3, samples). (j) Thickness of co-localized vessels at the same lateral entry and internal pore regions (n = 20, samples). Statistical analyses used two-way ANOVA for tests of two or more variables from each group, and one-way ANOVA for tests with single variables from each group. All p-values are denoted in the graphs. Discussion The hybrid PCL/GelMA scaffold developed in this study provided a structural and biochemical microenvironment conducive to pre-vascularized tissue formation, which resulted in rapid inosculation upon implantation. The zigzag stacking configuration generated side-pore interconnectivity (800 × 200 μm), which maintained spheroid-housing chamber stability while preserving lateral microchannels connecting adjacent pores. This design established multidirectional pathways for cellular migration, nutrient diffusion, and paracrine factor exchange across the scaffold volume. Scaffold interconnectivity represented a critical determinant of vascularization efficiency. Enhanced pore connectivity promoted vascular density and host vessel penetration via improved convective–diffusive transport and endothelial migration pathways [26]. For example, sphere-templated hydrogel scaffolds with high pore interconnectivity exhibited ∼ two-fold increases in internal vascular density and significantly improved host vessel infiltration. The present architecture integrates vertical and lateral connectivity, thereby facilitating multidirectional diffusion routes that support cell migration and neovascular invasion while preserving mechanical stability [35]. Surface modification with polydopamine enhances scaffold bioactivity. The catechol and amine groups present in the PD-promoted protein adsorption and integrin-mediated cell adhesion, thereby facilitating adsorption of extracellular matrix proteins such as fibronectin and vitronectin. PD coatings also increase surface polarity and hydrophilicity through catechol oxidation, which improves protein adsorption capacity and cell–substrate interactions. In addition, nanoscale roughness generated during PD deposition further supported filopodia extension and focal adhesion formation. ATR spectroscopy confirmed successful PD polymerization, with characteristic O–H stretching (3200–3600 cm−1), quinone-related C=O (∼1726 cm−1), and aromatic C=C (1500–1600 cm−1) peaks, which are typical spectral signatures of dopamine oxidative self-polymerization [36]. Together, these results indicate that PD modification provided a chemically favorable interface that supported early cell attachment and spreading, thereby facilitating subsequent spheroid migration and matrix remodeling. The GelMA hydrogel (modulus of ∼0.5–1 kPa) provided a permissive mechanical environment supporting both cell sprouting and ECM remodeling. Previous studies have shown that hADSC/HUVEC spheroids within this stiffness range exhibited significantly increased sprout length and branching compared with stiffer matrices (≥2 kPa), accompanied by elevated MMP-2 and MMP-9 expression and improved host vessel integration following implantation [32]. Softer matrices maintain degradable crosslink density while minimizing mechanical restriction, thereby enabling MMP-mediated matrix reorganization and sprouting-driven network formation. Conversely, increased stiffness may suppress cellular invasion due to rapid crosslink reformation relative to enzymatic degradation rates [37]. Thus, the selected crosslinking condition provides a balanced mechanical regime that preserves scaffold integrity while allowing dynamic cell migration and sprouting within the GelMA–PCL hybrid system. In this study, GelMA stiffness was characterized by using separately prepared bulk hydrogels under the same formulation and UV exposure conditions; thus, the reported modulus represents the nominal bulk GelMA stiffness rather than a depth-resolved local modulus within the PCL/GelMA scaffold. Although local crosslinking heterogeneity caused by UV attenuation or PCL shadowing cannot be completely excluded, this limitation does not affect the main comparative conclusions of the study, as all experimental groups were fabricated under identical GelMA composition, scaffold geometry, and UV exposure conditions. Therefore, the observed differences in cell migration, ECM remodeling, and vascularization are primarily attributable to spheroid positioning and scaffold architecture rather than variations in the crosslinking condition. The hADSC/HUVEC co-culture spheroids at a 3:1 ratio enhanced angiogenic potential through multicellular endothelial–stromal crosstalk. This ratio was selected based on our previous study, in which hADSC/HUVEC co-culture ratios of 1:0, 1:1, and 3:1 were compared and the 3:1 condition produced the most favorable vasculogenic responses, including enhanced sprouting and vascular network formation [32]. Spheroid diameter plays an important role in regulating paracrine signaling and nutrient diffusion. Spheroids within the 300–400 μm range have been reported to exhibit optimal sprouting efficiency, whereas smaller spheroids failed to establish sufficient signaling gradients and larger spheroids (>500 μm) developed hypoxic cores [38,39]. The size regime used in this study likely amplifies growth factor retention and the coordinated sprouting responses. Aligned with this approach, spheroid cultures demonstrated pronounced multicellular sprouting and invasion, whereas suspension cells maintained rounded morphologies with limited migration. Previous studies have similarly reported that endothelial spheroids exhibited organized collective invasion in 3D matrices, whereas dispersed cells failed to generate coordinated sprouting structures [40]. This difference arise from enhanced cell–cell communication within spheroids, which established localized paracrine gradients of angiogenic factors such as VEGF, PDGF, and Ang-1 [8]. In addition, spheroid organization promotes ECM deposition and MMP-mediated remodeling, enabling collective migration rather than independent cell movement [41]. The strong upregulation of MMP-9 and MMP-14 observed in this study further supports the role of spheroids as dynamic remodeling units. MMP-9 releases matrix-bound angiogenic factors such as VEGF and FGF-2, thereby reinforcing angiogenic signaling gradients [42], while membrane-bound MMP-14 locally degrades collagen and facilitates sprouting invasion through VEGF signaling amplification [43]. Similar proteolytic remodeling has been reported in MSC–endothelial co-culture systems where increased MMP expression accelerated capillary morphogenesis [44]. Concurrent upregulation of pro-angiogenic genes suggests that ECM remodeling and angiogenic signaling are tightly coupled during spheroid-mediated vascular formation [[45], [46], [47]]. Together, these processes indicate that spheroid organization facilitates coordinated angiogenic signaling and ECM remodeling, thereby accelerating pre-vascularized network formation [48,49]. The radial expansion of spheroid-derived cells and progressive occupation of scaffold chambers reflect coordinated sprouting and migration rather than proliferation alone, which may have been attributed to spheroid cell–cell interactions and paracrine signaling, accompanied by ECM remodeling. Comparable dynamic processes have been reported in GelMA hydrogels, where ECM-mimetic architecture and tunable crosslinking have been shown to facilitate spheroid sprouting. A lower stiffness (∼1 kPa) maximized cellular invasion and MMP-mediated remodeling, whereas a stiffer matrix suppressed sprouting efficiency [50]. The chamber-wide cell expansion observed in Fig. 3e represents early-stage vascular pathway morphogenesis within the permissive GelMA microenvironment. Collective migration at the spheroid surface may be due to endothelial-stromal coordination where leader cells initiate ECM remodeling and pathfinding via MMP-2 and MMP-9 upregulation and cytoskeletal polarization, thereby guiding follower cells [51]. N-cadherin-mediated cell–cell adhesion maintained migration coherence because disruption of N-cadherin significantly reduces migration speed and capillary-like network formation [52]. In addition, traction forces generated by leader cells propagated through cadherin junctions to follower cells, enabling mechanically coupled collective migration across the cell cluster [53]. Collectively, suspension cells lacked such paracrine/mechanical coordination, yielding disorganized morphologies. Directional migration of spheroid-derived cells along catching strands suggests that scaffold microgeometry provided strong contact guidance cues. Previously, cells were reported to detect substrate topography, reorganizing focal adhesions and F-actin for alignment along features, as observed in ridge–groove or fibrous substrates [54,55]. In the present scaffold, catching strands appeared to template sprouting directionality and spatial organization. Greater upper-region DAPI-positive density suggests upward migration over proliferation, mirroring conch-like scaffolds with threefold upper-density increases after 7 days in the previous study [56]. Such upward migration may have been facilitated by continuous adhesion pathways provided by the catching strands and interconnected pores, allowing cells to generate traction forces and migrate along the scaffold architecture. In addition, ECM remodeling enzymes, chemokine gradients, and hydrogel stiffness gradients likely can reinforce directionality. Interestingly, directional alignment was also observed in cells not directly contacting the strands. Such behavior is consistent with “guidance at a distance,” reported in biomimetic nerve models, where aligned fibers orient adjacent cells via matrix remodeling or paracrine diffusion [57]. Notably, 1 mm inter-spheroid spacing resulted in the greatest cellular coverage within the middle chamber and the highest expression of ECM remodeling and angiogenesis-related genes, including MMP-2, MMP-14, TIMP-4, CD31, EMCN, G-CSF, and SLIT3. These findings suggest that 1 mm spacing provides a favorable geometric condition for coordinated outward sprouting, inter-spheroid cellular bridging, and scaffold-guided collective migration within the PCL/GelMA construct. In contrast, directly adjacent spheroids may reduce the available inter-spheroid migration space, whereas larger separations of 2–3 mm may limit the formation of continuous cellular bridges within the 7-day culture period. Importantly, soluble factor gradients, local protein distributions, and diffusion–consumption dynamics were not directly measured in this study. Therefore, we have avoided attributing the spacing-dependent response to a definitive diffusion-based mechanism. Rather, our data supports a spacing-dependent association between spheroid positioning, collective cell migration, ECM remodeling, and angiogenic gene expression. Together, these results indicate that spheroid spacing functions as a key geometric regulator of vascular continuity within pre-vascularized constructs. Our results indicate that the pore architecture of the scaffold is an important player in regulating host–tissue integration. The Double scaffold pore architecture (larger/interconnected) enhances host integration via expanded nutrient/oxygen transport and ECM–cell interfaces, promoting adhesion/remodeling, which was similar to previous results where increased pore size and connectivity in porous ceramic and polymer scaffolds have been shown to significantly improve angiogenesis and tissue integration in vivo [[58], [59], [60], [61]]. Although increased scaffold openness promotes cell infiltration and spatial distribution, the survival of transplanted human cells is primarily determined by interactions with the host tissue and the surrounding microenvironment. This suggests that while structural design facilitates overall cellular infiltration, long-term retention of transplanted cells remains limited. Previous studies have similarly reported that highly interconnected scaffolds improved host cell infiltration and nutrient transport, yet human cell persistence was still constrained by immune responses, local inflammatory conditions, and ECM–cell compatibility [62]. Therefore, scaffold architecture should be viewed not merely as a factor that enhances cell survival, but as a structural platform that facilitates ECM remodeling and host–construct integration. Future studies employing humanized mouse models or more advanced immunodeficient systems are necessary to evaluate long-term engraftment and vascular integration of transplanted cells. Finally, the scaffold geometry influenced integration of engineered vessels with the host after implantation. Open pores in the upper region permitted host vessel entry, whereas lateral interconnectivity enabled multidirectional infiltration. In contrast, aligned catching strands in the lower region provided guidance cues that promoted alignment between host vessels and engineered vascular structures. Similar geometry-associated vascular alignment has been observed in melt electro-written scaffolds where strand orientation regulates vessel organization and tissue infiltration [63]. The preferential localization of α-SMA-positive structures near the bottom strands further suggests that mechanical cues within this region promoted perivascular cell differentiation and vessel stabilization [64,65]. Moreover, the greater interconnectivity of the Double scaffold increased the spatial interface between host and engineered vessels, facilitating their co-localization. Comparable structural effects have been reported in porous Mg and silk scaffolds where interconnected microchannels significantly enhanced vascular penetration and host–construct integration [66,67]. Despite morphological integration, future studies including perfusion analysis and long-term in vivo imaging are needed to confirm functional vascular connectivity. Nevertheless, our findings demonstrate that scaffold geometry can actively regulate vascular alignment and host–engineered vessel integration, highlighting the importance of geometry-associated design strategies in engineering pre-vascularized 3D tissue. The present study demonstrates that spheroid spatial organization and scaffold microarchitecture can collectively regulate cell migration, ECM remodeling, angiogenic responses, and vascular tissue formation, thereby providing a structural framework for pre-vascularized tissue engineering. Although spheroid-based constructs exhibited enhanced sprouting, ECM remodeling, and angiogenic gene expression compared with suspension cells in vitro, direct in vivo comparison between suspension-cell and spheroid-laden scaffolds will be necessary to further define the contribution of spheroid organization to host integration. Building upon these findings, future studies should investigate a broader range of scaffold architectures, including variations in pore size, pore density, and interconnectivity, to establish predictive structure–function relationships and identify design parameters that maximize vascularization outcomes. In parallel, although geometry-dependent changes in ECM remodeling and angiogenic gene expression were observed, the molecular mechanisms linking scaffold architecture to these biological responses remain to be elucidated. Further investigation of signaling pathways involved in vascular morphogenesis and matrix remodeling, including VEGF/VEGFR-, Notch-, and PDGF-related pathways, may provide additional mechanistic insight into scaffold-guided vascularization. Moreover, while enhanced host vessel infiltration and integration were observed following implantation, the long-term fate of engineered human-cell-derived vessels remains unclear. Future lineage-tracing and functional perfusion studies will be important to determine whether these vessels persist as components of the mature vascular network or primarily function as transient templates that facilitate host vascular remodeling. Finally, validation in immunocompetent and large-animal models will be necessary to assess long-term vascular stability, tissue integration, and translational potential. Together, these studies will further advance the development of rationally designed pre-vascularized tissue constructs by integrating architectural control with biological function. Conclusions In this study, we have developed a spheroid-laden hybrid PCL/GelMA scaffold platform, enabling pre-vascularized tissue formation via precise spatiotemporal regulation of cell–cell interactions within a defined mechanical and biochemical microenvironment. Spheroid positioning and scaffold pore interconnectivity were systematically modulated to elucidate their influence on mesenchymal collective migration, ECM remodeling, and angiogenic signaling. Our results demonstrate that an inter-spheroid distance of 1 mm maximized the balance between MMP and TIMP expression while significantly enhancing pro-angiogenic signaling, including CD31 and SLIT3 expression, thereby promoting efficient intercellular communication and vascular network morphogenesis. In addition, the double scaffold configuration, which features increased pore interconnectivity, significantly improved in vivo cell infiltration, ECM integration, and host–graft connectivity. Engineered spatial organization of spheroids and pore-guided migration thus emerge as pivotal design parameters for vascularized tissue formation. Integration of multicellular spheroid assembly with geometry-controlled scaffold architecture leads to a biomimetic cascade of collective migration, ECM remodeling, and angiogenesis, establishing a rational framework for scaffold geometry optimization and highlighting geometry-driven multicellular strategy for engineering functional pre-vascularized tissues. CRediT authorship contribution statement Sangmin Lee: Conceptualization, Data curation, Formal analysis, Methodology, Validation, Visualization, Writing – original draft. SeongHeon Bae: Investigation, Methodology, Visualization, Writing – original draft. Hyunseok Kwon: Formal analysis, Writing – review & editing. Ji Min Seok: Methodology. Su A Park: Methodology, Resources, Supervision. Heungsoo Shin: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review & editing. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgement This work was supported by the Engineering Research Center (RS-2025-02263336) and by a grant (RS-2023-00207983) from the National Research Foundation of Korea (NRF), funded by the Korean government. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.mtbio.2026.103358. Contributor Information Su A Park, Email: psa@kimm.re.kr. Heungsoo Shin, Email: hshin@hanyang.ac.kr. Supplementary data The following is the Supplementary data to this article. Multimedia component 1 Data availability Data will be made available on request. References Associated Data Supplementary Materials Multimedia component 1 Data Availability Statement Data will be made available on request.