Microfluidic vascular formation model for assessing angiogenic capacities of single islets

Pancreatic islet transplantation presents a promising therapy for individuals suffering from type 1 diabetes. To maintain the function of transplanted islets in vivo, it is imperative to induce angiogenesis. However, the mechanisms underlying angiogenesis triggered by islets remain unclear. In this study, we introduced a microphysiological system to study the angiogenic capacity and dynamics of individual islets. The system, which features an open‐top structure, uniquely facilitates the inoculation of islets and the longitudinal observation of vascular formation in in vivo like microenvironment with islet‐endothelial cell communication. By leveraging our system, we discovered notable islet−islet heterogeneity in the angiogenic capacity. Transcriptomic analysis of the vascularized islets revealed that islets with high angiogenic capacity exhibited upregulation of genes related to insulin secretion and downregulation of genes related to angiogenesis and fibroblasts. In conclusion, our microfluidic approach is effective in characterizing the vascular formation of individual islets and holds great promise for elucidating the angiogenic mechanisms that enhance islet transplantation therapy.

islets in the pancreas possess a high degree of vascularization with fenestrated endothelium throughout the islet core, receiving 15%−20% of the pancreatic blood supply, while only comprising 1%−2% of the total mass (Lifson et al., 1985).Posttransplantation, the transplanted islets rely on reconnection with the vascular network in the recipient's body to obtain nourishment, typically requiring 7−14 days.To limit islet ischemia and improve the outcome of islet transplantation, a deeper understanding of the angiogenic process and the acceleration of vascularization after transplantation is necessary.
Various techniques for accelerating the angiogenic process have been proposed, such as prevascularization of the graft site (Pepper et al., 2015), cotransplantation of islets with other cells that support vascular formation, including endothelial cells, mesenchymal cells, and microvasculature (Nalbach et al., 2021;Phelps et al., 2013;Sakata, 2011;Takahashi et al., 2018) and growth factor-releasing devices (Uematsu et al., 2018).Although these supporting cells and devices have demonstrated favorable results in promoting angiogenesis toward the transplanted islets, it is evident that the angiogenic capacity of the individual islets that induces the surrounding endothelial cells to form a vascular network is still crucial.This is because an active interaction between the transplanted islets and endothelial cells of recipients is necessary to reform the native isletspecific vasculature, which is fenestrated and located adjacent to thecells across a double-layered basement membrane, to achieve rapid sensing of fluctuations in blood glucose and the outflow of secreted hormones.Therefore, vascular formation induced by islets is a critical indicator of their angiogenic capacity.
The angiogenic capacity of pancreatic islets has been explored using animal models (Speier et al., 2008;Wagner et al., 2022), which have revealed specific angiogenic secretions from the islets, such as vascular growth factor A (VEGF-A) (Brissova et al., 2006;Xiong et al., 2020), the significant contribution of the donor's endothelium to the formation of a vascular network after transplantation (Brissova et al., 2004;Linn et al., 2003;Nyqvist et al., 2005), and the effects of immunosuppressive drugs on vascular formation (Nishimura et al., 2013).However, analyzing individual islets for their molecular functions in animal models is technically difficult, making it challenging to unveil the detailed mechanisms of vascular formation induced by islets.The cellular composition of the islets is heterogeneous (Dybala & Hara, 2019) and partially affects vascular formation after transplantation (Lau et al., 2012).Additionally, the extra-islet environment is heterogeneous, such as the distance from the preexisting recipient vascular network and local islet density.
Therefore, there is a high demand for the development of experimental tools to simultaneously evaluate islet angiogenic capacity and cellular composition.
Microphysiological systems (MPSs), which include a subset of organ-on-a-chip (OoC) systems, are in vitro platforms that generate an in vivo-like microenvironment with physiologically relevant factors (Bhatia & Ingber, 2014;Park et al., 2019) and have been adapted to islet culture systems (Hori et al., 2019;Jun et al., 2019;Patel et al., 2021).Recently, various organ models were combined with perfusable vascular networks within the MPSs (Ewald et al., 2021;Lee et al., 2018;Zhang et al., 2021).For the application to islets, R-VECs exhibited tissue-adaptable features, and their vascular network extended deeply into human islets (Palikuqi et al., 2020).
Although these reports visualized islets integrated with a vascular network in vitro, the analysis of angiogenic capacity for individual islets is challenging because vascular formation was largely assisted by mesenchymal and gene-edited endothelial cells, implying that even without the islets, active vascular formation could be observed.
To the best of our knowledge, an MPS capable of analyzing the angiogenic capacity at the single-islet level has not yet been reported.
In this study, we developed a microfluidic platform that enabled the assessment of the angiogenic capacity of single islets (Figure 1).This microfluidic device features an open-top architecture that facilitates both the introduction of islets and the observation of the ensuing angiogenic response in a highly individualized manner.Furthermore, the open-top configuration provides a straightforward method for the extraction of islets after the assessment of angiogenic capacities for downstream analysis, such as RNA sequencing (RNA-seq).Using this platform, we stratified the angiogenic capacities of islets and evaluated their gene expression profiles using RNA-seq analysis.
F I G U R E 1 Evaluation of islet vascularization process in a microfluidic device.Islets derived from a rat were inoculated into a microfluidic device and the vascularization processes in an angiogenic manner were visualized in vitro.Following the device culture, the islet and surrounding vascular network were collected for RNA-seq analysis to evaluate gene expressions.RNA-seq, RNA sequencing.

| Fabrication of open-top microfluidic device
We employed a previously reported open-top microfluidic device (Nashimoto et al., 2023) for islet vascularization (Supporting Information Figure S1a).Briefly, the device consisted of two layers, with the lower layer featuring five channels for the cultivation of the vascular network, whereas the upper layer functioned as an islet culture (Supporting Information Figure S1b).A thin polydimethylsiloxane (PDMS) membrane with a thickness of 100 μm separated the upper and lower layers and enables communications between the islet and the vascular network through a small aperture (vascular access hole, ϕ = 0.5 mm).
The mold used to produce the lower layer was fabricated on a silicon wafer (thickness 380 μm; Canosis).A SU-8 3050 negative photoresist (KAYAKU Advanced Materials) with 100 μm thickness was patterned onto the silicon wafer using ultraviolet photolithography.A PDMS prepolymer (SILPOT 184W/C; Dow Toray), with a base-to-curing agent ratio of 10:1, was cast against the molds and subjected to a curing process of >8 h at 70°C.Subsequently, the cured PDMS was peeled off the mold to a thickness of approximately 1 mm.To fabricate a thin PDMS membrane for separating the upper and lower area, a spin-coating process was employed, using an 850 rpm (for 100 µm in thickness) or a 650 rpm (for 150 µm in thickness) speed for 30 s on a glass slide (Matsunami), followed by an overnight curing process at 70°C.The vascular access hole in the thin PDMS membrane was created using a 0.5 mm biopsy punch (KAI Corporation).The upper layer was fabricated a PDMS with 4-5 mm in thickness on plain glass.The culture area for an islet and outlets in the upper layer were fabricated using 8.0-, 4.0-, 3,0-, or 2.0-mm biopsy punches (KAI Corporation).After air-plasma treatment, each layer was permanently attached using a plasma modifier (PM100; Yamato Scientific).For sterilization, the microfluidic device was subjected to ultraviolet radiation for >8 h before cell seeding.To improve analytical throughput, a device with three access holes was used in a similar manner (Supporting Information Figure S1a).The detailed dimensions of each device component are presented in Supporting Information Figure S1c.

| Cell culture
HUVECs expressing the green fluorescent protein (GFP-HUVECs) were purchased from Angio-Proteomie.GFP-HUVECs were maintained in endothelial growth medium (ECGM-2; PromoCell) and passages 5−6 were used for the experiment.All cells were maintained in a humidified incubator at 37°C and 5% CO 2 .

| Animals and islet isolation
Male Lewis rats were purchased from Japan SLC Inc. and used as islet donors.All the animals had free access to standard food and were maintained in a pathogen-free environment.All surgeries were performed under anesthesia, and efforts were made to minimize suffering.
Briefly, Lewis rats were anesthetized by inhalation of isoflurane (Viatris Inc.).The bile duct was identified and clamped to the papilla of Vater.Ten milliliters of cold Hank's balanced salt solution (HBSS) supplemented with 0.8 mg/mL collagenase type V (Sigma Chemicals) was administered into the common bile duct, leading to the pancreas.
The pancreas was removed, incubated in a water bath at 37°C for 12 min, and then subjected to enzymatic digestion.The resulting cell suspension was washed three times in cold HBSS, centrifuged for 1 min, and subjected to density-gradient centrifugation using Histopaque-1119 (Sigma Diagnostics) and Lymphoprep™ (Axis-Shield) for 10 min to isolate the pancreatic islets.The islets were then cultured overnight in islet medium (RPMI-1640 medium, containing 5.5 mmol/L glucose, 10% fetal bovine serum [Thermo Fisher Scientific], and 1% penicillin−streptomycin [Thermo Fisher Scientific]), at 37°C, in a 5% CO 2 -humidified atmosphere before transferring to a microfluidic device.Three rats were used as donors in this study.These experiments were approved by the local ethics committee (protocol ID: 2020IrA-001) and conducted in compliance with national and institutional regulations.

| Islet vascularization in a microfluidic device
A blank fibrin-collagen gel was employed as an extracellular matrix surrounding the endothelial cells and islets to simulate the angiogenic process toward the islets.The gel was composed of 2.5 mg/mL fibrinogen (Merck), 0.2 mg/mL collagen (Corning), 0.15 U/mL aprotinin (Merck), and 0.5 U/mL thrombin (Merck), which were dissolved in Dulbecco's phosphate-buffered saline (D-PBS [−]; Nacalai Tesque).Channels 1, 3, and 5 were filled within the fibrin-collagen gel, whereas channels 2 and 4 were filled with endothelial medium (ECGM-2).The following day, GFP-HUVECs were introduced into channels 2 and 4 at a density of 5.0 × 10 6 cells/mL.Upon introducing the cells, the device was tilted at a 90°angle and incubated at 37°C in 5% CO 2 for 30 min to allow the GFP-HUVECs to adhere to both sides of the fibrin-collagen gel.In this study, we defined Day 0 as the time when the endothelial cells were seeded onto the devices.Islets of 150−200 μm were collected through 150-and 200-μm filters (PluriSelect Life Science) and used for device culture.The upper layer was filled with islet medium.For medium exchange, half of the medium in channels 2 and 4 and the upper layer was withdrawn, and an equal volume of new medium was added daily.In the culture setup, angiogenic sprouts from channels 2 and 4 can be induced by secretions from a target tissue cultured centrally in channel 3 (Nashimoto et al., 2017).

| Cell staining and imaging
For nuclei and CD31 staining in the microfluidic device, cells were fixed with 4% paraformaldehyde (Alfa Aesar) and stored for more than 8 h at 4°C.Subsequently, the samples were treated with 0.1% Tween-20 (Merck) for 120 min at 4°C.Then, the anti-CD31 antibody (1:300; BioLegend) with 1% bovine serum albumin (Merck) was dissolved in D-PBS (−).Subsequently, 10 μg/mL Hoechst 33342 (Thermo Fisher Scientific) and 4 µg/mL anti-mouse IgG conjugated with Alexa Fluor TM 647 (Thermo Fisher Scientific) were added to the device and preserved at 4°C for more than 24 h.
For histological and immunobiological analyses, the islets were fixed in the upper layer with 4% paraformaldehyde in D-PBS (−) at 4°C overnight.After three washes with D-PBS, iPGell (GenoStaff) was added to the upper layer to embed the samples.The sample was collected using a 2 mm biopsy punch, transferred into a 35 mm dish, and overlaid with another iPGell.The samples were stored in 70% ethanol and sectioned at the pathology section of the Biomedical Research Core of the Tohoku University Graduate School of Medicine.The islets were dehydrated, embedded in paraffin, and sectioned for hematoxylin and eosin (H&E) staining.For immunostaining, anti-CD31 antibody (Leica), anti-insulin antibody (Abcam), anti-glucagon antibody (Abcam), and a secondary detection system (3,3'-diaminobenzidine) were used.
The fluorescence and time-lapse images of the microfluidic device were acquired using a DMi8 inverted microscope (Leica).
Histological and immunohistological sections were imaged using an Eclipse Ts2 inverted microscope (Nikon) and images were recorded using a Michrome 20 color CCD camera (Tucsen Photonics).Optical sections were obtained using a confocal microscope (TCS SP8; Leica) at Tohoku University and Tokyo Medical and Dental University.
Image postprocessing was performed using the Fiji image processing software (National Institutes of Health).

| Synthesis of cDNA
After device culture, islets and HUVECs were lysed using the loading Buffer RLT (Qiagen), and RNAs in the lysate was collected from the device.The collected RNAs was purified using an RNeasy Micro Kit (Qiagen).Islets before device culture were lysed using lysis buffer (Takara Bio) containing an RNase inhibitor (Takara Bio).HUVECs cultured independently in the microfluidic device without any islet were also subjected to lysis by infusing Buffer RLT into channels 2 and 4. We constructed an RNA-seq library using a previously reported protocol (Hirano et al., 2023).
Briefly, mRNA was reverse-transcribed with SMARTScirbe (Takara Bio), barcoded oligo-dT primers, and template-switching oligos.The residual primers were digested using exonuclease I (Takara Bio).Full-length cDNA was amplified by 10−12 cycles of PCR cycles using SeqAmp DNA polymerase (Takara Bio).The sequencing library was prepared by amplifying cDNA fragments derived from 3' end of mRNA using TruSeq Read1 sequence in the barcoded oligo-dT primer and Read2 sequence inserted with the Nextera XT DNA Sample Preparation Kit (Illumina).Sequencing was performed using an Illumina HiSeqX platform (Azenta).
All oligonucleotide sequences used for RNA-seq are listed in Table S1.
Further data analysis was performed using the R-Studio environment.
Samples with less than 3000 detected genes were filtered out as lowquality samples.Uniform manifold approximation and projection (UMAP) was performed using the Seurat package (4.1.0).Gene set enrichment analysis (GSEA) was performed using ClusterProfiler (4.2.2).The islets were ordered as shown in Figures 3c and 4b using the DiffusionMap function in Destiny (3.8.1).

| Statistical analysis
All statistical analyses were performed using the R Studio software.Significance was assessed by implementing a nonparametric Brunner−Munzel test for single comparisons (Supporting Information Figure S2), and statistical significance was set at p < 0.05.

| Evaluation of the angiogenic capacity of islets
To establish an ex vivo islet transplantation model, we employed our open-top microfluidic device to induce angiogenesis toward in implanted islets (Supporting Information Figures S1a and S1b).
Vascular formation in channel 3 was facilitated by employing a fibrin gel as the substrate, whereas the islets were cultured on the gel in the upper layer using the islet medium.The endothelial medium was introduced into channels 2 and 4, and the GFP-HUVECs adhered to the sidewall of the fibrin gel in channel 3. Channels 1 and 5 were filled with a blank fibrin gel.
A few days (2−3 days) after the inoculation of an islet, we observed the active formation of angiogenic sprouts from both sides of the fibrin gel, which extended toward the islet.Conversely, in the microfluidic device without islets, although we observed noncontinuous (single-cell-like) migration of endothelial cells, sprout formation was infrequent (Figure 2a).The vascular area of the microfluidic device with islets was significantly larger than that of the device without islets (Figure 2b and Supporting Information Figure S2).We confirmed that the sprouts induced by the islets had a luminal structure and were partially attached to the islets that secreted insulin and glucagon (Figure 2c,d).
Interestingly, there was considerable variation in the vascular area among islets (Figure 2b).Using the vascular area on Day 8 in a microfluidic system without islets as a benchmark, we classified islets with >20% of the vascular area as "plus" islets, whereas those with lesser vascular area were classified as "minus" islets.We confirmed that the variation was not derived from the difference in islet quality depending on the isolation method (Supporting Information Figure S2a).To investigate the molecular mechanisms that differentiate the resulting vascularization, we performed transcriptome analysis (RNAseq) of the islets following device culture.

| Transcriptome analysis
By filtering low quality data, we acquired a total of eight plus samples and four minus samples, collected mRNA of islets and surrounding vascular network and analyzed the gene expression (Figure 3) along with the islets before device culture (defined as "naïve").UMAP on the islets revealed that the device culture altered the gene expression patterns of the islets from those of naïve islets.No further subpopulations were observed within either naïve or cultured islets, implying that the naïve islets had relatively uniform cellular compositions and no transcriptomic signature to differentiate angiogenic capacity (Figure 3a).Although we performed a comprehensive differential expression analysis, we could not detect any genes with significant differences between plus and minus islets.To uncover the difference in the transcriptomic state between the plus and minus islets further, we performed GSEA and identified upregulation of gene sets related to angiogenesis and fibroblast migration in the islet minus group and enrichment of a gene set related to insulin secretion in the islet plus group (Figure 3b).The genes included in each term are shown in Figure 3c.The comprehensive GSEA results for the islets are presented in Data S1.Next, we analyzed the gene expression patterns of the vascular network (endothelial cells).Among the four minus samples (Figure 3), only one remained after quality assessment, likely due to the insufficient vascular formation surrounding the minus islets.It was found that genes related to chemokines were enriched in endothelial cells cocultured with plus islets, consistent with active vascular formation and migration toward the plus islets (Figure 4).The gene expression profile of HUVECs with minus islets was more analogous to that of HUVECs without any islet coculture.The comprehensive GSEA results for the vascular network are shown in Data S2.
Collectively, our microfluidic platform enabled the visualization of the angiogenic process toward inoculated islets and revealed gene expression changes in angiogenesis, insulin secretion, and chemokines that occurred during the process.

| DISCUSSION
An angiogenic process toward a transplanted islet is a crucial step in sustaining its function in recipients.Our microfluidic device facilitated the visualization of the angiogenic capabilities of individual islets by partially mimicking the in vivo interactions between endothelial cells and islets after transplantation as well as enabling the collection of samples and gene expression analysis following vascular formation.
Real-time monitoring of the angiogenic process toward single islets revealed significant variability in vascular formation induced by individual islets (Figure 2b).For convenience, we classified the islets cultured in the device as plus or minus islets based on the vascular area (Figure 2b and Supporting Information Figure S2).This classification revealed differences in the gene expression profiles of angiogenesis, fibroblast migration, and insulin secretion between plus and minus islets (Figure 3b).
Insulin is a pivotal islet-secreted hormone that regulates blood glucose levels and has angiogenic effects (Escudero et al., 2017).
By activating the intracellular phosphorylation cascades in endothelial cells, insulin can regulate endothelial cell migration, proliferation, and in vitro tubular formation, subsequently leading to enhanced VEGF-A (Walker et al., 2021) and TGF-β (Budi et al., 2019) responsiveness in endothelial cells.The abundance of insulin secretion-related genes in the plus islets (Figure 3b) implied that insulin secretion induced the formation of vasculature surrounding the plus islets.On the other hand, this finding can also be interpreted as a consequence of enriched vascularization around the islets in plus group.Recent studies have demonstrated that the majority of insulin molecules within beta cells localize at the interface with the vascular network, suggesting positive effects of a vascular network on insulin secretion (Cottle et al., 2021;Low et al., 2014).Given that an extremely small amount of insulin is secreted from a single islet, the latter is more likely to occur in the present model.Angiogenesis-related genes were more abundant in minus islets, suggesting that islets lacking a vascular network upregulated angiogenic signals.
Fibroblast-like cells were notably abundant in close proximity to minus islets, especially those with the least vascular area on day 8.
Longitudinal microscopy revealed that the vascular network of the islets was transiently formed during the early culture period (by approximately Day 4 and Supporting Information Figure S3) and steadily disappeared as fibroblast-like cells started to migrate away from the islets and dispersed within the fibrin gel (Days 6 and 8, Supporting Information Figure S3), indicating a repulsive impact of fibroblast-like cells on the vascular network (Figure 2b and Supporting Information Figure S3).Because of the lack of GFP expression, we concluded that the fibroblast-like cells originated from the rat islets in the device culture.Gene expression analysis consistently showed that a gene set related to fibroblast migration was enriched in the minus islets (Figure 3b).Linn et al. reported that endothelial cells remaining in the islets developed outward cord-like structures when cultured in fibrin gel (Linn et al., 2003) and several studies have demonstrated the positive effects of the remaining endothelial cells on vascular formation after transplantation (Brissova et al., 2004;Nyqvist et al., 2005).In contrast, fibroblast-like cells surrounding the minus islets actively migrated in a single-cell manner and lacked cordlike structures.These results imply that fibroblast-like cells surrounding the islets may be a potent risk factor for successful vascular formation.
We observed that genes related to angiogenesis were predominantly enriched in the minus islets (Figure 3b).This observation implies that the minus islets, which are positioned farther away from the vascular network in the microfluidic device, require more vascular networks.Notably, we also observed a significant correlation between the vascular area and Vegfa, which is a canonical gene involved in islet vascularization (Supporting Information Figure S4).
These results suggest that although some angiogenesis-related genes were activated in the minus islets, key molecules, such as Vegfa were lacking, resulting in poor vascular formation.We envision that largescale screening of islet vascularization followed by RNA-seq would allow us to overview the landscape of single-islet heterogeneity and identify the link between gene expression and the vascularization phenotype.
In this study, we used human endothelial cells and rat islets, the particular differences of which should be addressed in future studies.
Although rat and human islets exhibit similarities, some differences should be considered.For example, human islets have been reported to be more heterogeneous in beta, alpha, and delta cells (Arrojo e Drigo et al., 2015), and the amino acid sequence of angiogenic factors (e.g., VEGF-A) in rats is slightly different from that in humans.The heterogeneity in angiogenic capacities should be evaluated using human islets for clinical application.Additionally, assaying islet function after vascular formation is crucial.Future research should explore glucose tolerance through the vascular network and real-time monitoring of islet-specific vascular fenestrations.We anticipate that this microphysiological platform will deepen our understanding of the angiogenic process and islet function after vascular formation and contribute to the progress of future islet transplantation therapy.
Rambøl et al. successfully reformed a vascular network derived from human umbilical vein endothelial cells (HUVECs) surrounding an islet, assisted by mesenchymal stem cells (Rambol et al., 2020).Palikuqi et al. used "reset" vascular endothelial cells (R-VEC) with transient ETV2 expression as the source of the engineered vascular network.

F
I G U R E 2 Angiogenic process with a rat islet in a microfluidic device.(a) Bright-field and fluorescence images of microfluidic devices with (w/) and without (w/o) islets during 8 days in the culture.Greens represent GFP-HUVECs.Scale bars are 200 µm.(b) Quantification of the vascular area in the device culture.Plus and minus represent the islets with high and low angiogenic capacities, respectively.(c) Orthogonal views of the vascular network induced by an islet after 8 days in the culture.Arrowheads indicate the luminal structures in x−z, y−z sections.Yellow: CD31, cyan: nuclei, scale bar is 250 µm.(d) Histological and immunohistological sections of the islet after device culture.The scale bar is 50 µm.GFP-HUVECs, green fluorescent protein-human umbilical vein endothelial cells.F I G U R E 3 Gene expression analysis of islets after the device culture.(a) UMAP analysis.(b) Enrichment plots of gene sets in plus and minus islets defined by the vascular area.The green line represents the enrichment score, and the red broken line indicates the position of the maximum enrichment score.Genes were ranked on the horizontal axis based on their fold changes of the gene expressions (plus/minus).(c) Heatmaps for the gene sets in (b).The sample information was shown at the lowest row as the colors.UMAP, uniform manifold approximation and projection.

F
I G U R E 4 Gene expression analysis of HUVECs after the device culture.(a) Enrichment plots of gene set related to response to chemokines.(b) Heatmap of gene expression related to the response to chemokines.The samples ordered by human gene expression are shown at the lowest row by the colors."Only HUVEC" means HUVECs cultured in the microfluidic device for the same culture period without islet coculture.HUVECs, human umbilical vein endothelial cells.