hPIs used in this study were isolated from six cadaveric multiorgan donors, in particular three women and three men, donor age 52.6 ± 3.6 years, BMI 25.5 ± 2.6 kg/m², islets purity 82.5 ± 5.2%, according to the procedure previously described (Ricordi et al., 1988; Petrelli et al., 2011). The overall protocol has been approved by the Niguarda Cà Granda Ethics Board. Islets were isolated using the automated method previously described (Ricordi et al., 1988). Pancreata were obtained from multiorgan cadaveric donors utilizing cold perfusion. Exclusion and inclusion criteria were applied based on the Italian Guidelines. Briefly, pancreata were digested by a cold enzymatic blend solution of collagenase and thermolysin (Liberase MTF GMP Grade kit, Roche Diagnostics, Mannheim, Germany) reconstituted in Hank’s Balanced Salt Solution (HBSS, Euroclone, Italy) with 25 mM of HEPES. Subsequently, islets were purified with discontinuous polysucrose solutions at decreasing density 1.132, 1.108, 1.096, 1.060 and 1.037 g/L (Mediatech-Cellgro, VA, United States). Islets were counted by dithizone staining islet equivalent (IEQ) method (see “Dithizone staining” section) and they were cultured at 24°C, 20% O₂, 5% CO₂ in a humidified atmosphere in MIAMI Medium #1A (Mediatech-Cellgro, VA, USA) supplemented with Ciprofloxacin (Fresenius Kabi, Verona, Italy), or in serum-free medium in the presence of basic fibroblast growth factor (bFGF, PeproTech) and epidermal growth factor (EGF, PeproTech) at final concentrations of 10 ng/ml and 20 ng/ml, respectively.

All reagents and solvents used for the peptide synthesis were purchased in highest quality commercially available and used without further purification. All peptides were synthesized with solid-phase Fmoc-based chemistry on Rink amide resin (0.19–0.56 mmol/g, 100–200 mesh) using a Liberty Blue System synthesizer (CEM Corp, Matthews, NC, Canada). Peptides were cleaved from resin by addition of a freshly prepared mixture containing 92.5% TFA, 2.5% H₂O, 2.5% DODt, 2.5% TIS. All synthesis were carried out on a 0.25 mmol in presence of a 0.2 M amino acid solution (in DMF), 1 M DIC (in DMF), and 1 M Oxyma (in DMF). The deprotection of Fmoc groups was determined by a 10% v/v of piperazine in 9:1 NMP/EtOH. The N-terminal acetylation (for CK₁ and pureHYDROSAP components) was performed using 20% v/v solution of Ac₂O (in DMF). The crude products were purified via reversed-phase chromatography by semi-preparative Waters binary HPLC (>96%) using a c18 RestekTM column and then lyophilized (Labconco, Kansas City, MO, USA). Purified peptides powder was subsequently dissolved in 0.1 M HCl to remove the presence of possible TFA salts. Three different SAPs were used for this study: pureHYDROSAP (Marchini et al., 2019; Marchini et al., 2020), FAQ (NH₂-FAQRVPP-GGG-LDLKLDLKLDLK-CONH₂) (Gelain et al., 2012) and CK₁ (Ac-CGGLKLKLKLKLKLKGGC-CONH₂) (Pugliese et al., 2018c; Ciulla et al., 2022). As previously described, pureHYDROSAP is composed by linear SAPs Ac-(LDLK)₃-CONH₂, Ac-KLPGWSGGGG-(LDLK)₃-CONH₂ (Caprini et al., 2013) and Ac-SSLSVNDGGG-(LDLK)₃-CONH₂ (Gelain et al., 2012) and branched SAP tris(LDLK)₃-CONH₂ (Pugliese et al., 2018a). For the experiments, pureHYDROSAP (abbrev. HYDROSAP), FAQ and CK₁ powders were dissolved respectively to a final concentration of 2% (w/v), 5% (w/v) and 5% (w/v) in distilled water (Gibco).

The rheological experiments were carried out on a rotational AR-2000 ex rheometer (TA Instruments, Waters Corp, Milford, CT, United States) equipped with an acrylic cone-plate geometry (diameter, 20 mm; angle, 1; truncation gap, 34 µm). All the tests were performed at 37°C using a Peltier plate in the lower plate of the instrument to assess a continuum control of the temperature during each test. All samples were dissolved in distilled water at the corresponding concentration and incubated overnight at 4°C. Thus, 584 mM sucrose solution were added to each peptide solution (ratio 1:1); then, MIAMI medium (instead of the volume of hPIs) were included to obtain a final concentration of 0.76% (w/v), 1.9% (w/v) and 1.9% (w/v) for HYDROSAP, FAQ and CK_1, respectively. Each sample (50 µL) was gently placed on the middle of the Peltier plate. A lid on the top serves to protect sample from the water evaporation. To evaluate the storage (G′) and loss (G″) moduli increments, frequency-sweep experiments were recorded as a function of angular frequency (0.1–100 Hz) at a fixed strain of 1% after a 3 h time-sweep experiment (performed at 1 Hz constant angular frequency) in presence of D-PBS (1X). Stress–strain sweeps were performed in the range 0.01%–1,000% of strain. Thixotropy of peptides was performed with shear-thinning tests by a series of peak holds at constant shear rates. All data were performed in triplicate and the results were processed with OriginPro 2019 (OriginLab Corporation, Northampton, MA, United States). Graphs were reported in the Supplementary Figure S1).

Cell cultures were prepared within 24/48 h after receiving islets. hPIs were incapsulated inside hydrogels or plated in suspension in 48-well with two different type of culture media: MIAMI medium #1A (Mediatech-Cellgro, VA, United States) (a CMRL-based culture medium commonly used in clinical trial for islet culture before transplantation) or GF medium (a serum-free medium with bFGF and EGF growth factors). 25 IEQ (low density) or 500 IEQ (high density) were incapsulated inside SAPs, previously dissolved in distilled water and diluted with 584 mM sucrose solution (ratio 1:1). A droplet of 25 µL was placed onto glass coverslip in 48-well, medium was added to start SAP gelation and to obtain free-floating samples. Same concentration of islets was used for samples in suspension in 48-well. In these conditions, the samples were maintained in culture up to 14-days (T14) and 28-days (T28) at 24°C, 20% O₂, 5% CO₂ in a humidified atmosphere. As positive control, suspensions of islets in MIAMI medium or GF medium were maintained in culture for 1 day (T1). hPIs incapsulated insides hydrogel were monitored individually during culture time and brightfield images from day 1 to day 28 were acquired weekly via Zeiss light microscope at ×5 magnification.

Dithizone (Sigma-Aldrich, St Louis, MO, United States) is a zinc chelating agent, well-known to selectively stain pancreatic islets in a brownish red color. During islets purification and after 1 (T1), 14 (T14) and 28 days (T28) post-isolation, an aliquot of suspension of hPIs was stained with dithizone to track the islet morphology during time and to observe the purity of preparation. 200 µL of Dithizone solution was added to 1 ml of islets suspension in culture media for 10 min at room temperature. Following incubation time, three washes in D-PBS 1X (Gibco) was performed and images were acquired using Zeiss light microscope at ×5 magnification.

Images acquired via Zeiss light microscope of hPIs encapsulated inside SAPs and hPIs in suspension after Dithizone staining were processed via NIH-ImageJ software and Axiovision software. Diameter of each islet at each timepoint were calculated and reported.

hPIs in suspension and encapsulated in hydrogel maintained in culture for 1, 14 and 28 days were fixed in paraformaldehyde (PFA) 2% and 4% (w/v in PBS) and cryosectioned at 50 µm-thick via Cryostat (Histo-Line Laboratories). Sections were permeabilized with 0.3% Triton X-100 for 10 min at 4°C and blocked with 10% normal goat serum (NGS, Gibco) for 1 h at room temperature. The following primary antibodies were used: rabbit anti-insulin (1:300, ThermoFisher), mouse anti-chromogranin (1:100, ThermoFisher), mouse anti-glucagon (1:8000, Sigma-Aldrich), rabbit anti-Ki67 (1:750. Novus Biologicals), rabbit anti-vWF (1:500, DakoCytomation), mouse anti-fibroblast (1:200, Acris Antibodies), rabbit anti-collagen IV (1:100, Cedarlane), mouse anti-collagen I (1:2000, Sigma-Aldrich), rabbit anti-laminin (1:30, Sigma-Aldrich). To reveal primary antibodies, the following secondary antibodies were used: goat anti-rabbit Cy3 (1:1,000, Jackson), goat anti-mouse Cy3 (1:1,000, Jackson), goat anti-rabbit Alexa 488 (1:1,000, Invitrogen) and goat anti-mouse Alexa 488 (1:1,000, Invitrogen). Cell nuclei are stained with HOECHST 33342 (1:500, Molecular Probes).

Tunel assay (In situ cell death detection kit fluorescein, Roche) was performed to detect and quantify apoptotic cells. The protocol was performed following the manufacturer’s instructions. Briefly, slices were permeabilized with 0.3% Triton X-100 for 10 min at 4°C and incubated with Tunel reaction mixture (1:10 enzyme in label solution) for 1 h at 37°C. Cell nuclei are stained with HOECHST 33342 (1:500, Molecular Probes).

Whole hPIs were imaged in fluorescence at × 20 and × 40 magnification via Zeiss Microscope connected to Apotome System and image analysis was performed with Fiji-ImageJ NIH-software. A minimum of three independent experiments were performed for each experimental condition, timepoint and marker. For insulin, chromogranin, glucagon, Ki67 and Tunel Assay markers, quantitative analysis were performed by counting manually positive cells for each marker, compared to total cells contained in a single islet and stained with Hoechst. For vWF, Fibroblast, Laminin and Collagen I and IV, fluorescent images were converted into binary images and reactivity area were quantified by measuring the number of positive pixels for each marker, compared to the total area of hPIs.

All the animal procedures and ethical revision were performed according to the current Italian Legislation (Legislative Decree 4 March 2014, n.26) enforcing the 2010/63/UE Directive on protection of animals used for biomedical research. In vivo protocol was approved by Institutional Animal Care and Use Committee (IACUC number 63/2022-PR). 18 CD-1 Nude Male Mice, weight 27/30 g (approx. 7 weeks) were divided into three groups: 1) six animals received freshly-isolated hPIs (positive control); 2) six animals received hPIs pre-cultured in SAPs for 2 weeks; 3) six animals received hPIs pre-cultured in SAPs for 4 weeks hPIs used for in vivo experimentations came from two different donors. After diabetes induction, 1,500 IEQ hPIs were transplanted under kidney capsule of diabetic mice, by inspiring a well-known surgical technique (Szot et al., 2007). Mice were monitored for 50 days post-surgery by studying non-fasting blood glucose level (non-fasting BGL) and body weight every other day, and intraperitoneal glucose tolerance test (IPGTT) at two- and 4-week post-transplantation. Normoglycemia was achieved when blood glucose levels scored values < 200 mg/dl. At the end of in vivo experiment, mice were sacrificed by cervical dislocation and engrafted kidneys were excised, fixed, and cryo-sectioned to visualized with immunohistochemical studies the presence of engrafted hPIs via human insulin markers. See Supplementary Figure S1 and Methods for more details.

Results are reported as means ± SD in all graphs. Data were processed using GraphPad Prism seven software for in vitro and in vivo experiments. In vitro tests were performed in triplicate: three different islets preparations, derived from three different donors, are used for each condition (time-points, type of medium and scaffold) for each marker. Statistical analyses were evaluated by Two-way ANOVA followed by Tukey’s multiple comparison. Comparison between low density and high density of hPIs were performed by two-way ANOVA followed by Bonferroni’s multiple comparison test. Statistical significance was set at p-value < 0.05.

Biomaterials are key structural elements in 3D cell cultures to tune cell attachment, growth, differentiation and function (Marchini and Gelain, 2022). The ECM composition and organization to be mimicked by the biomaterials can differ and depends on the tissue of origin. Pancreatic ECM, enriched in laminin, collagen IV and VI, and fibronectin, provides structural and biochemical support to β-cells (Townsend and Gannon, 2019). Islet-ECM interaction regulates survival, insulin secretion, β-cells proliferation, differentiation, and preservation of round islet morphology (Stendahl et al., 2009). Current islet isolation techniques disrupt the native islet ECM, leading to faster cell death and dysfunction. Biomaterials, and in particular SAPs, can act as efficient synthetic substitutes, since they can be tailored according to specific in vitro and/or in vivo applications by changing mechanical, physicochemical and biological parameters (Marchini and Gelain, 2022). SAPs used in this work were widely applied in the field of neural stem cells (Gelain et al., 2012; Pugliese et al., 2018c; Marchini et al., 2019; Marchini et al., 2020). It is well-known that brain tissue is very soft, with a stiffness ranging from 0.1kPa to 1 kPa (Ciulla et al., 2022): pancreas stiffness lays in the similar range (Guimarães et al., 2020), with healthy pancreas tissue ∼900 Pa and with increased ECM stiffness in pancreatic cancer (∼2,900 Pa) (LeSavage et al., 2022). Three different SAPs (HYDROSAP, FAQ and CK₁) were evaluated in terms of rheological analysis to investigate their viscoelastic and mechanical properties, as well as to compare their characteristics to that of pancreatic tissue. In order to mimic viscoelastic properties of normal human pancreatic tissue, thixotropy (Supplementary Figure S1, on the left), and stiffness, in terms of storage modulus (G′) and loss modulus (G″) (Supplementary Figure S1, on the right) were evaluated. The rheological characterization revealed that storage moduli in HYDROSAP, FAQ and CK₁ enclose G’ values ranging to 826 ± 75 Pa for FAQ, 1,583 ± 165 for HYDROSAP, and 5,440 ± 421 Pa (CK₁), showing a range of stiffness close to the ECM surrounding of human pancreatic cells. In other words, the SAPs scaffolds used in this study were satisfactorily similar (in term of rigidity) to the healthy pancreas tissue (HYDROSAP, FAQ), or to tumorigenic pancreas (CK₁). Lastly, thixotropy experiments described the capacity of these materials to recover their original behavior after reversible breaking and recovering of self-assembled network.

In vitro studies, standard hPIs cultures were maintained in free-floating condition as a control group (Figure 1A). In Figure 1B, hPIs morphology and integrity are monitored and captured in brightfield microscopy. At 1-day post-isolation (T1), hPIs appeared as distinct round clusters; after 14 (T14) and 28 (T28) days, hPIs clustered together, by losing their original morphology and integrity. At T28, the edge appeared completely jagged and islets loss their morphological integrity, due to ECM disruption during enzymatic isolation. Dithizone staining, routinely applied in free-floating condition for assessing islet quality, purity, and morphology, demonstrated a robust decrease of β-cells functionality (Khiatah et al., 2019), underlined by a steady switch of islets color from reddish to brown one (Supplementary Figure S2). To overcome these problems, functionalized SAPs were used as building blocks to mimic biofunctional and structural features of ECM (Marin and Marchesan, 2022) and support hPIs, by controlling numerous cellular processes, such as cell adhesion (Figure 1C). In this regard, islets morphology was monitored through the weeks via an inverted microscope. As an example, in Figure 1D are shown hPIs embedded in HYDROSAP, but similar observations also were found for FAQ and CK₁ (Supplementary Figure S2). As depicted in Figure 1D, hPIs were not subjected to morphological alteration during weeks. SAPs prevented cluster formation and hPIs edge remained clearly defined. To corroborate the above-mentioned qualitative data, the alteration in hPIs diameter was monitored overtime up to 28-days post-isolation (Figure 1E). As expected, hPIs embedded inside HYDROSAP kept stable their dimension, starting from 258.29 ± 46.36 μm at T1, to 256.17 ± 47.03 μm at T14, and 250.95 ± 46.42 μm at T28. Conversely, hPIs in suspension, in absence of a structural support, completely lost their initial morphology and their diameter decreased significantly from 234.49 ± 89.94 μm at T1, to 212.08 ± 70.51 μm at T14, and 197.5 ± 57.11 μm at T28. Indeed, at 14 and 28 DIV (days in vitro) statistical analysis showed significant differences in islets diameter between suspension of hPIs and hPIs inside HYDROSAP (***p < 0.001). Diameter conservation was also assessed for islets embedded inside FAQ and CK₁ scaffold, as depicted in Supplementary Figure S3 and Supplementary Table S1.

As demonstrated in the previous section, HYDROSAP, FAQ and CK₁ could act as ECM-like scaffolds, due to their mechanical properties similar to pancreatic tissue, and their ability to maintain overtime morphological conformation of hPIs. First, cellular content in all samples at 14 DIV was analyzed with immunofluorescence stainings against the main islet hormones (insulin and glucagon), endocrine cells, proliferating cells and apoptotic cells (Figure 2). Islets were 3D seeded in HYDROSAP, FAQ and CK₁ scaffolds, or plated in suspension, with two different cell culture media: MIAMI medium and GF medium. As a positive control, freshly-isolated suspensions of islets were cultured in MIAMI media for 1 day (T1). Islets maintained in suspension for 14-days in MIAMI or GF media (without mechanical support) were chosen as negative controls. Percentage of cells producing insulin (Figure 2A and Supplementary Table S2) decreased drastically (***p < 0.001) in suspension in MIAMI medium at T14, compared to freshly-isolated islets T1 (45.84 ± 12.59% and 12.18 ± 9.29%, respectively). Conversely, in free-floating condition, GF medium seems to maintain more efficiently insulin content for 14-days, thanks to bFGF and EGF growth factors that increases insulin secretion (Wang et al., 2001; Lee et al., 2008). The effect of 3D matrices was then investigated. Only HYDROSAP scaffold cultured in MIAMI medium was able to preserve percentage of insulin cells (32.35 ± 19.51%); indeed, HYDROSAP was the only hydrogel not showing significant differences with freshly-isolated islets at T1. Percentage of glucagon decreased in all conditions with ***p < 0.001 (Figure 2B; Supplementary Table S2) compared to freshly-isolated islets in MIAMI medium (28.76 ± 9.75%) and in GF medium (28.56 ± 5.34%). Glucagon component is not a crucial hormone for patients with Type 1 Diabetes: its decreased amount can be considered as less crucial to the success of islets transplantation. The percentage of chromogranin, a protein located in secretory vesicles of endocrine cells, was also performed (Figure 2C). Percentage of chromogranin-positive endocrine-cells is conserved in islets embedded in HYDROSAP in MIAMI medium with 46.22 ± 19.68% and in GF medium with 45.33 ± 23.64%, if compared with freshly isolated samples (61.73 ± 13.67% for MIAMI medium and 58.31 ± 12.40% for GF medium). Conversely, islets maintained in FAQ and CK₁ obtained results comparable to islets in suspension for 14-days (27.04 ± 17.99%) (Supplementary Table S2). The proliferative capacity of all endocrine cells in islets was also tested (Figure 2D). Compared to T1 (8.06 ± 4.49%), proliferation rate was preserved overtime in all substrates, without significant differences between samples (Supplementary Table S2). Quantitative analysis of apoptotic cells inside islets grown into different substrates was achieved (Figure 2E). The rate of cell apoptosis is relatively higher in all conditions overtime, compared to hPIs suspension at T1 (2.07 ± 0.86%), with the highest percentage of apoptotic cells reached by islets maintained in CK₁ in MIAMI medium for 14-days (7.96 ± 5.53%, **p < 0.01, Supplementary Table S2). Despite some detected apoptosis, the obtained results denoted a negligible degree of cytotoxicity.

In order to characterize different ECM components inside hPIs and to evaluate their preservation right after their isolation (T1) and at 14 DIV, immunofluorescence analyses were performed for the main ECM components of pancreas (laminin, collagen IV, and collagen I), for fibroblast cells and for endothelial cells (Figure 3). Isolation procedure disrupts internal vascularization, innervation of islets, intra-islet ECM and peripheral islet ECM, by interrupting fundamental cell-matrix signaling (Stendahl et al., 2009). Biomimetic scaffolds could act as supporting elements by mimicking ECM and preserve, or even induce production, of some ECM components. Pancreatic islets are extensively vascularized and vascular network is essential for their function and survival (Xiong et al., 2020). Moreover, islets re-vascularization is a key elements for graft success following islet transplantation in the treatment of Type 1 Diabetes (Brissova and Powers, 2008). Von Willebrand factor (vWF) is a key component of blood and is produced by endothelial cells (Figure 3A). At 14 DIV, vWF expression did not decrease if compared to freshly-isolated islets: on the contrary a significantly higher expression in HYDROSAP with GF media (3.75 ± 1.08%) was found, probably due to the presence of EGF, that plays an important role in angiogenesis and significantly influence differentiation and proliferation of vascular endothelial cells (Chen et al., 2020). The combination of the multi-functionalized microenvironments of HYDROSAP and the exposure to EGF probably induced this considerable increase. Indeed, percentage of reactivity area in hPIs embedded into HYDROSAP in GF medium is almost doubled compared to freshly-isolated islets (1.96 ± 0.48%, ***p < 0.001); in hPIs cultured in HYDROSAP and CK₁ in MIAMI medium (2.44 ± 1.25%, 2.58 ± 1.00%, respectively), vWF values were still higher than negative control (1.23 ± 0.36%) with **p < 0.01 and ***p < 0.001, respectively (Supplementary Table S3). Human islets are surrounded by a peri-islet capsule consisting of a layer of fibroblast and collagen fibers, produced by fibroblast. This capsule is closely associated with basement membrane which provides physical stability and anchorage to the islets (Banerjee, 2020). Fibroblasts are a cell population that participate on ECM and connective tissue formation, maintain tissue homeostasis, and preserve structural components. It was demonstrated that in a subcutaneous islet transplantation model, fibroblasts is associated with a faster graft re-vascularization, a higher insulin-positive area and a lower cell death (Perez-Basterrechea et al., 2017). Fibroblasts too are degraded during the enzymatic and mechanical isolation process of hPIs. As expected, bFGF, enclosed in GF media, enhanced fibroblast content in all hPIs encapsulated inside scaffolds (Figure 3B; Supplementary Table S3). The effect of bFGF was also observed in the short time in suspension of hPIs at T1 (4.30 ± 1.94%), compared to the counterpart in MIAMI medium (2.74 ± 1.50%, **p < 0.01). HYDROSAP scaffold showed significantly higher percentages of fibroblast positive cells (3.83 ± 1.34% in MIAMI medium, 4.63 ± 1.43% in GF medium). On the other hand, FAQ and CK₁ preserved fibroblast component just like in positive control.

Together with collagens, laminin plays an important role among the ECM components of hPIs. If collagens provide structural stiffness to tissue, laminins maintain integrity and conformation of tissue structure. In human islets, laminins are involved in cytoskeletal remodeling, contractility, and control β-cells differentiation and insulin secretion (Llacua et al., 2018). Laminin resulted well-conserved in hPIs seeded in all SAPs (Figure 3C). Lasty, it was evaluated the most abundant ECM molecule in hPIs, i.e., collagens. Collagen improves endocrine functions, survival, and proliferation, and it is also a principal target during digestion for islet isolation (Riopel and Wang, 2014; Sakata et al., 2020). Collagen IV is a major component of the peri- and intra-islet ECM, modulates ECM stiffness, promotes cell survival, and decreases insulin production at high concentrations (Kaido et al., 2006). hPIs cultured for 14-days inside CK₁ samples in GF medium (3.95 ± 2.33%) obtained an unexpected higher value in reactivity area compared to initial situation at T1 (1.41 ± 0.91%, ***p < 0.001) (Figure 3D and Supplementary Table S3). For the reason mentioned before, probably this result justified the lower values of β-cells in CK₁-seeded islets (Figure 1A). The other SAPs preserved Collagen IV very well, without significant differences compared to hPIs in suspension at T1. Similar results were obtained for Collagen I (Figure 3E), that is normally localized within and around islets such as collagen IV, promotes islet cellular survival and differentiation and also improves β cell function (Wieland et al., 2021). CK₁ scaffold supplemented with GF medium significantly enhanced Collagen I content (reactivity area) compared to freshly isolated islets (2.66 ± 1.76% and 1.01 ± 0.76%, respectively), while no decreased expression was observed for other SAPs (Supplementary Table S3). Some studies have demonstrated that abnormal collagen upregulation is associated with cancer cell proliferation. Such doubled production of Collagen I and IV in cells cultured in CK₁ scaffold for 14-days could be due to the peculiar rigidity of CK₁ that matches with tumorigenic pancreatic tissue. Additional analysis on collagen fibers orientation could be significant in predicting cancer (Angel and Zambrzycki, 2022). Tumor-ECM is more abundant, condensed, and stiffer than the ECM in the surrounding healthy tissue. Moreover, oriented collagen fibers around tumor cells, identification of specific collagen organization patterns and evaluation of collagen-associated biomarkers are indicators of tumor progression (Baldari et al., 2022; Song et al., 2022).

In summary, a stable expression of insulin, endocrine component, proliferative cells, apoptotic cells, vWF expression, fibroblast, laminin, collagen IV and I, between freshly-isolated islets (positive control) and islets embedded into HYDROSAP for 14-days was achieved. Moreover, if compared to islet in suspension for 14 days in MIAMI medium (negative control), a higher percentage of insulin, glucagon, proliferative cells, vWF, fibroblast and collagen IV and a lower percentage of apoptotic cells was observed. No relevant differences between the 2 cell culture media were detected, even if the GF medium looked like to increase the percentage of some ECM components. As previously described, FAQ and CK₁ have not fulfilled all requirements. Still, CK₁, with its high storage modulus (5,440 ± 421 Pa, Supplementary Figure S1), could be further investigated for application in the field of pancreatic tumors. Indeed, enhanced ECM stiffness and the upregulation of collagen production were demonstrated to be closely correlated to cancer progression (Zhang et al., 2021; Kpeglo et al., 2022). Alternatively, since changing SAPs concentration influence SAPs storage and loss moduli (Caprini et al., 2013; Gelain et al., 2021) it is very likely that a fine-tuning of CK concentration could allow to obtain a scaffold that better matches native pancreas biomechanics.

Encouraged by the obtained results, the culture time was extended till 28-days to evaluate the feasibility of long-term cultures of hPIs. Among the analyzed SAPs, HYDROSAP was chosen as the best substrate for 3D hPIs cultures and therefore additionally tested in this work.

All marker analyses were repeated at 28 DIV (T28) and previous data at T1 and T14 were reported in graphs as reference (Figure 4). Only HYDROSAP scaffold was chosen to encapsulate hPIs for 28-days, while FAQ and CK₁ were excluded from the study because of their sub-optimal performances. Figure 4 shows the overall in vitro characterization: endocrine markers (insulin, glucagon, and chromogranin), cells in proliferation, apoptotic cells and markers related to endothelial cells, fibroblast, and physiological ECM molecules (laminin, collagen I and IV). β-cells content (Figure 4A; Supplementary Table S4) in hPIs suspension in MIAMI medium at T1 scored 45.83 ± 12.59% of total cells; this amount decreased drastically in islets in suspension at T28 (23.25 ± 5.57%, ***p < 0.001), but it was better preserved in HYDROSAP-embedded hPIs in MIAMI medium (38.50 ± 12.01%), without significant differences with hPIs at T1. On the other hand, only 27.11 ± 11.41% is reached by hPIs in HYDROSAP in GF medium with a significant difference compared to positive control (***p < 0.001). Similarly, glucagon content is preserved over weeks in vitro compared to fresh-isolated islets (28.76 ± 9.75%) with values between 25.33 ± 7.00% for hPIs-embedded HYDROSAP in MIAMI medium and 25.18 ± 10.43% for hPIs-embedded HYDROSAP in GF medium at T28 (Figure 4B; Supplementary Table S4).

A similar trend observed for endocrine cells: the percentage of chromogranin-positive endocrine-cells was conserved in long-term cultures of HYDROSAP-embedded hPIs in MIAMI medium (49.33 ± 10.36%) (Figure 4C; Supplementary Table S4). In GF medium chromogranin reached a percentage of 40.47 ± 9.20%, significantly lower than in positive control (61.73 ± 13.67%, **p < 0.01). Cellular proliferative rate underwent a slow reduction in all groups: the percentage of 11.97 ± 4.01% (at T14 in HYDROSAP MIAMI) is halved at T28 in the same culture conditions (5.24 ± 2.73%, *p < 0.05); identically, hPIs cultured in HYDROSAP in GF medium decreased from 11.42 ± 5.47% at T14 to 6.66 ± 4.13% at T28 (Figure 4D; Supplementary Table S4).

On the other hand, the percentage of apoptotic cells (TUNEL assay) in all conditions remained stable (Figure 4E; Supplementary Table S4) if compared to results obtained at T14, suggesting a controlled apoptosis at 28 DIV (Figure 1E).

Immunofluorescence analysis of ECM components showed a vWF expression of HYDROSAP-embedded hPIs in MIAMI medium similar to freshly-isolated islets; conversely, it was found a significantly higher expression in HYDROSAP-embedded hPIs at T14 (3.75 ± 1.08%) and T28 (2.81 ± 0.86%) in GF medium if compared to hPIs in MIAMI medium at T1 (1.96 ± 0.48%) and T14 (1.23 ± 0.37%) and to hPIs in GF medium at T1 and T14 (1.80 ± 1.16% and 1.63 ± 0.38%, respectively) (Figure 4F and Supplementary Table S4). As mentioned before, this result is probably due to the combination of HYDROSAP scaffold and the presence of EGF in the GF culture medium (Figure 1F).

On the other hand, the high fibroblast content reported in HYDROSAP in GF media at T14 (4.63 ± 1.43%), decreased drastically after 28-days of culture (1.77 ± 0.75%, ***p < 0.001) (Figure 4G; Supplementary Table S4). Moreover, reactivity area for fibroblast marker is reduced also inside HYDROSAP maintained in MIAMI medium at T28 (1.31 ± 0.43%), compared to its T14 counterpart (3.83 ± 1.34%, ***p < 0.001).

Finally, extracellular matrix component (laminin, collagen IV, and collagen I) (Figure 4H–L; Supplementary Table S4) are well-conserved in HYDROSAP-embedded hPIs maintained in MIAMI medium for 28-days: statistical analysis did not show significant difference with freshly isolated hPIs and with same condition at T14. Conversely, the percentage of laminin (Figure 4H) and Collagen IV (Figure 4I) decreased drastically at 28 days in HYDROSAP with GF media (0.51 ± 0.52% and 0.61 ± 0.20% respectively) compared to the same conditions at T14 (3.36 ± 2.15% for laminin with **p < 0.01 and 2.93 ± 2.00% for Collagen IV with ***p < 0.001, Supplementary Table S4)

In conclusion, hPIs cultured in vitro have obtained excellent results in term of cellular components. When 3D cultured in HYDROSAP scaffold, all main cellular elements are preserved up to 28 DIV if compared to fresh-standard hPIs cultures. At 28-days vWF and collagen I remained stable inside HYDROSAP both in Miami medium and GF medium. The percentage of fibroblast, laminin, and collagen IV decreased drastically at 28-days in HYDROSAP with GF media and remained almost stable in HYDROSAP cultured in MIAMI medium. Accordingly, HYDROSAP in MIAMI medium significantly preserved functionality of α and β cells, controlled cell apoptosis, preserved ECM components, maintained a rounded hPIs morphology and islets diameter in vitro up to 4 weeks.

Abovementioned results were conducted with 25IEQ inside each 3D scaffold. Now, islets concentration was boosted to 500 IEQ to evaluate islets viability and functionality at high concentrations used for the in vivo preparation protocol (see methods). High concentration of hPIs allows to evaluate if an increased interaction between hPIs ameliorates (or worsens) the quality of long-term hPIs cultures. To distinguish these concentrations, 25IEQ are named Low-Density (LD) of hPIs and 500IEQ as High-Density (HD) of hPIs. Low-Density and High-Density of hPIs in suspension in MIAMI medium, hPIs in suspension in GF medium, hPIs-embedded HYDROSAP in MIAMI medium and hPIs-embedded HYDROSAP in GF medium were compared at 28-days of culture. hPIs in suspension in MIAMI and GF media cultured for 28-days are considered as negative controls. All markers considered in previous analysis were studied in Low-Density and High-Density conditions (Figure 5).

Percentage of insulin positive β-cells increased (Figure 5A) in HYDROSAP in GF medium (41.82 ± 9.01%, HD of hPIs), compared to its LD counterpart in GF medium (26.45 ± 11.61%, ***p < 0.001), obtaining values similar to HYDROSAP in MIAMI medium at HD (45.29 ± 4.91%) and to freshly-isolated islets at T1 (45.84 ± 12.59%, Figure 2A). α-cells secerning glucagon (Figure 5B) significantly enhanced glucagon expression in HD condition in hPIs suspended in GF medium (32.98 ± 5.87%), in hPIs-embedded HYDROSAP in MIAMI medium (38.16 ± 7.68%) and in hPIs-embedded HYDROSAP in GF medium (36.98 ± 7.66%). As mentioned before, glucagon content is poorly conserved in HYDROSAP scaffold overtime. However, maintaining hPIs in HD for 28-days into HYDROSAP allowed to reach the value of freshly-isolated hPIs at T1 (28.76 ± 9.75%, Figure 2B). Similar results were achieved for endocrine cells (Figure 5C).

Thus, β-cells, α-cells and endocrine component showed increased values in HD cultures (especially for hPIs maintained in GF medium, hPIs-embedded HYDROSAP in MIAMI medium and also in GF medium). Despite the increased hPIs concentration, proliferative cells and apoptotic cells percentages inside islets remained unaffected (Figure 5D, E).

However, HD conditions negatively affected vWF and Fibroblast expression, that decreased drastically in most conditions with ***p < 0.001 (Figure 5F, G). On the contrary, the major components of ECM (laminin, collagen IV and collagen I) (Figure 5H–L) remained stable, except for collagen IV in HYDROSAP in GF medium that significantly increased its expression in HD condition (1.42 ± 0.87%, compared to 0.61 ± 0.20% at LD, *p < 0.05), and for Collagen I in HYDROSAP in MIAMI medium that drastically reduced its expression (0.45 ± 0.23% in HD compared to 1.21 ± 0.74% at LD, **p < 0.01).

In conclusion, HD cultures of hPIs in HYDROSAP up to 28 days showed an increased percentage of α-cells and β-cells, a controlled apoptosis, and a lower content of ECM components.

Encouraged by excellent in vitro results in term of preservation cellular components, we tested the efficacy of transplants of hPIs previously cultured in vitro. Inside HYDROSAP scaffold, β-cells and endocrine component were well-preserved up to 14 DIV and 28 DIV, if compared to standard hPIs cultures (hPIs suspensions at 1 day).

Effective functionality of hPIs maintained in culture for 2 and 4 weeks was investigated in vivo in a pilot study: hPIs were transplanted into sub-renal capsule of diabetic mice. The subcapsular space is considered an ideal site for pancreatic islets implantation in diabetic mice: indeed, pancreatic islets engraftment in the subrenal capsule site is a well-established protocol (Szot et al., 2007). Researchers reported that human islets transplantation into mice kidney cured diabetes in 75%–80% of recipients; while transplantation of murine islets yielded 100% success (Stokes et al., 2017). Animals in vivo experiments were divided into three groups: 1) six mice receiving freshly-isolated hPIs (control group), 2) six animals transplanted with hPIs pre-cultured in HYDROSAP for 2 weeks, and 3) six animals receiving hPIs pre-cultured in HYDROSAP for 4 weeks. In these experiments, only hPIs cultured in MIAMI medium were considered. Freshly-isolated hPIs were immediately transplanted into the sub-renal capsule; conversely, hPIs pre-cultured in HYDROSAP were maintained in culture for two or 4 weeks at HD of hPIs and mechanically dissociated from HYDROSAP immediately before transplantation. As best examples, here we report two mice of the control group and one animal implanted with hPIs (pre-cultured for 2 weeks) that responded positively to islets transplantation, approaching normoglycemic values (Supplementary Figure S4) and an optimal responsiveness to intraperitoneal glucose tolerance tests (Supplementary Figure S4), especially 15 days after transplantation (Supplementary Figure S4). Also, a progressive increase of body weight through the course of the study (Supplementary Figure S4) and the presence of transplanted functional hPIs at the implant site (Supplementary Figure S5) confirmed the success of engraftment in all three animals. The poor success of the overall in vivo transplants (3/18 mice) and, in particular, of positive controls (2/6 mice), led us to consider this study as preliminary. A refinement of implantation technique is required. Indeed, the injection technique could have partially destroyed cells, or the amount of engrafted hPIs was not enough to restore normoglycemia (Estil Les et al., 2018) and hPIs resulted unable to secrete a sufficient amount of insulin to reduce blood glucose concentration at physiological levels (Perteghella et al., 2017). Moreover, most animals developed a severe hyperglycemia after STZ injection with secondary complication, leading to a high mortality rate and a low success of engraftment (Supplementary Figure S6) (Deeds et al., 2011). For example, a decrease in weight in all mice receiving pre-cultured hPIs was detected (Supplementary Figure S6).

Being a pilot study, it is not possible to draw specific conclusions about the efficacy of our 3D treatment for hPIs for now, and further investigation is required. However, one animal recovered normoglycemia after transplantation with hPIs pre-cultured for 2 weeks, and such return to physiological glucose level is very likely related to the success of hPIs pre-culturing and engraftment.