Stem Cell Therapy Improves Human Islet Graft Survival in Mice via Regulation of Macrophages

Islet or stem cell–derived β-cell transplantation offers great hope for patients with type 1 diabetes or other diseases (1–3). However, the effectiveness of islet/β-cell transplantation is still low. One cause is early islet cell death caused by the stress-induced innate immune response immediately after transplantation. Therefore, approaches that can enhance early islet cell survival may improve the effectiveness of islet/β-cell transplantation.

Mesenchymal stromal/stem cells (MSCs) are multipotent adult stem cells with immunomodulatory and tissue-protective properties (4). MSCs can be isolated from bone marrow, adipose tissue, umbilical cords, and many other tissues. MSCs from different sources show similar protective properties despite this heterogeneity. They attach to a standard cell culture plate; express cellular markers including CD90, CD105, and CD29; and do not express CD45, CD11b, and HLA-DR. They can be differentiated into adipocytes, chondrocytes, osteocytes, and other cell types (5). MSCs have the inherent capacity to secrete immunoregulatory, anti-inflammatory, and proangiogenic factors. MSC therapy has been tested in >1,000 clinical trials to treat autoimmune and other diseases (clincialtrials.gov). Our laboratory performed the first clinical trial testing the safety of autologous MSC and islet cotransplantation in chronic pancreatitis patients undergoing total pancreatectomy and islet autotransplantation (6).

Increasing evidence has shown higher survival rates and function of transplanted islets when cocultured or cotransplanted with MSCs in rodents (7). The hallmarks of such studies show early normoglycemia, improved glucose tolerance, and increased insulin secretion in the syngeneic (8,9), allogeneic (10–13), and xenogeneic islet transplantation models (14–16). In vitro coculture of islets with MSCs improves islet function through MSC-mediated mitochondrial transfer (17) and modifies the oxidative imbalance caused by cytokines (18). MSCs also create a favorable environment for pancreatic β-cell survival (19) and longevity (20,21) by protecting against endoplasmic reticulum–induced apoptosis (22), altering immune cell trafficking (14,23), and improving graft vascularization (17,24–26). MSCs can also differentiate into insulin-secreting cells (27). Coinfusion of MSCs and islets also improves islet graft morphology and function in the intrahepatic islet transplantation model (25). The anti-inflammatory, antiapoptotic, and regenerative properties of MSCs not only benefit transplanted islets but also may protect islets during pretransplant culture and shipment from islet-processing facilities to more distant islet transplantation sites (7,28–32).

In various studies investigators have attempted to enhance the natural therapeutic effects of MSCs by induction of protective genes or by treating MSCs with hypoxia or other approaches. For example, inducible expression of vascular endothelial growth factor (VEGF) by human embryonic stem cell–derived MSCs reduced the minimal islet mass required to reverse diabetes in mice when islets were transplanted into the omental pouch of diabetic nude mice (33). In a different study, induction of VEGF and human interleukin-1 receptor antagonist (hil-1Ra) in bone marrow–derived MSCs (BM-MSCs) improved outcomes of islet transplantation when human islets were transplanted under the kidney capsule of NOD-SCID γ (NSG) mice (34). Furthermore, cotransplantation of islets with MSCs cultured in hypoxic conditions accelerated glycemic utilization and improved islet tissue regeneration (35).

We engineered human α-1 antitrypsin (hAAT)-overexpressing MSCs to enhance the therapeutic effects of native MSCs (24). AAT is a protein belonging to the serpin superfamily. It is encoded by the SERPINA1 gene. The primary function of AAT is to inhibit various proteases, including neutrophil elastase. AAT also has anti-inflammatory effects and protects islet survival in type 1 diabetes with/without islet transplantation, as shown by other researchers in addition to our team (24,36).

In this study, we tested the beneficial effects of gene and stem cell therapy by cotransplanting hAAT-MSCs with human islets into the liver of diabetic NOD-SCID mice, a model mimicking clinical human islet transplantation. To further understand the mechanisms of MSC protection, we transplanted islets from luciferase (Luc) transgenic mice or Luc-transfected MSCs to follow the longitudinal presence of islets and MSCs after transplantation. Moreover, we assessed the effects of hAAT-MSCs on the migration, polarization, and distribution of macrophages in the in vitro setting and in livers bearing human islet grafts to explain the mechanisms of the protective effects of those cells.

Male NOD-SCID, C57BL/6, Luc transgenic (on C57BL/6 background) mice at 8–10 weeks of age (The Jackson Laboratory, Bar Harbor, ME) were used in the study. All procedures were conducted with the approval of the Institutional Animal Care and Use Committees at the Medical University of South Carolina (protocol no. 200972).

Diabetes was induced in NOD-SCID mice with injection of a single dose of streptozotocin (125 mg/kg body wt i.p. per mouse) (37). Mice with two consecutive blood glucose (BG) levels measuring >350 mg/dL were used for transplantation experiments. Human islets were obtained from Georgetown University. Donor information and islet characteristics can be found in Supplementary Table 1. The human islets were transplanted at a dose of 17 islets/g body wt per mouse (323–460 islets/mouse) via portal vein infusion into the liver of the NOD-SCID mice. Nonfasting BG levels were measured every 2 or 3 days until 60 days posttransplantation (PT), and mice with BG levels <200 mg/dL were considered normoglycemic. Blood samples were collected before (day 0) and at 1, 3, and 7 days PT. Serum was separated from whole blood, and human C-peptide was measured with the Human C-peptide ELISA kit (ALPCO, Salem, NH). In cell tracing experiments, 320–330 handpicked Luc⁺ islets isolated from Luc transgenic mice were transplanted into the liver of diabetic NOD-SCID mice (38).

Human BM-MSCs were separated from bone marrow aspirates from two healthy male donors (Lonza, Walkersville, MD). hAAT-MSCs were generated by lentiviral infection (24). The pHAGE CMV-a1aT-UBC-GFP-W lentiviral vector encoding hAAT with the green fluorescent protein (GFP) reporter was used for lentiviral production. Cells were purified through cell sorting based on GFP expression. Cells infected with control viruses were used as MSC controls.

Serum samples were collected from the tail vein before (day 0), and at 1, 3, and 7 days post–islet transplantation. The presence of 23 mouse serum cytokines was quantified with the Bio-Plex Pro Mouse Cytokine 23-plex Assay kit (cat. no. B6009RDPD; Bio-Rad Laboratories, Hercules, CA), following the manufacturer’s instructions using the Luminex system.

Recipient mice randomly selected from each group were fasted for 4 h and injected with d-glucose solution at 1 g/kg via tail vein injection. BG levels were measured at 0, 15, 30, 60, and 90 min after injection. A small aliquot of blood/serum was obtained at each time point. Human C-peptide levels were measured with the C-peptide ELISA kit (cat. no. 80-CPTHU-CH01; ALPCO) according to the manufacturer’s recommendations.

Livers bearing islet grafts were first perfused with 10% formalin as previously described (37). Serial sections of 5 µm thickness were collected. Tissue sections from at least 150 μm apart were collected from each animal and were stained with anti-insulin (cat. no. PA1-26938, 1:100; Invitrogen) and anti-glucagon (cat. no. 2760S, 1:100; Cell Signaling Technology) antibodies. Images were collected with use of a ZEISS Axio fluorescence microscope. Insulin⁺ area, glucagon⁺ area, number of insulin⁺ and glucagon⁺ cells per islet, and the total number of islets per section, were quantified with ImageJ software.

For generation of Luc virus, 293 T cells were transfected with the pMD2.G, psPAX2, and pLX304 Luc-V5 plasmids (nos. 12259, 12260, and 98580; Addgene). Viruses were collected and used to infect human MSCs. The luminescence intensity was measured with a BioTek spectrometer. The lentiviral-infected Luc-expressing human MSCs (Luc⁺MSCs) were cotransplanted with islets into the recipient’s liver. The infused Luc⁺ cells were traced with bioluminescent imaging using a Xenogen IVIS 200 preclinical in vivo imaging system (PerkinElmer) after administration of 150 mg/kg i.p. d-luciferin to each mouse.

For detection of Luc signals in individual organs, liver, intestine, lung, pancreas, and kidney from mice transplanted with human islets or human islets and Luc⁺MSCs were collected within 3 min after d-luciferin injection. Each organ was placed in an individual well of a sixwell plate. Luc intensity was read with a BioTek Synergy HT spectrometer. Organs harvested from Luc transgenic mice were used as positive controls.

MSCs or hAAT-MSCs were seeded into a 6.5-mm Transwell insert with a polycarbonate membrane and 0.4-μm pore (cat. no. 3470; Corning) and nested into a cell culture well containing Raw264.7 cells for 24 h. Raw264.7 cells were then treated with either IFN-γ (100 ng/mL) or IL-4 (10 ng/mL). The presence of M1 or M2 macrophages was analyzed with flow cytometry and RT-PCR.

Raw264.7 cells (4 × 10⁵) were seeded in the insert of a 6.5-mm Transwell plate with a polycarbonate membrane and 5.0-μm pore (cat. no. 3421; Corning). MSCs or hAAT-MSCs (4 × 10⁴) were seeded in the lower wells of the Transwell system in the presence or absence of human islets. At 24 h after coculture at 37°C in 5% CO₂, all cells in the lower wells were collected and percentages of F4/80⁺ macrophages were analyzed by flow cytometry.

Macrophages were harvested by scraping with a cell scraper, washed with PBS, and fixed with 2% paraformaldehyde. Cells were incubated with respective antibodies or corresponding isotype controls for 1 h in the dark on ice. Expression of cellular markers was analyzed on the BD LSRFortessa X20 Flow Cytometer. Data were analyzed with the FlowJo software.

hAAT concentration in the serum was measured with a hAAT ELISA kit following the protocol of the manufacturer (cat. no. 108799; Abcam).

Total RNA was extracted and reverse transcribed into cDNA with a Reverse Transcription kit (QIAGEN). SsoAdvanced Universal SYBR Green Supermix (Bio-Rad Laboratories) was used for quantitative RT-PCR in a CFX96 Real-Time Thermocycler (Bio-Rad Laboratories). Fold changes in gene expression normalized to β-actin expression were plotted and compared between groups.

Paraffin sections of liver tissue bearing islet grafts were deparaffinized and stained with the Roche Ventana Discovery Ultra Automated Research Stainer (Roche Diagnostics, Indianapolis, IN). Heat-induced epitope retrieval was performed in EDTA buffer, pH 9 (cat. no. S2367; Agilent/Dako, Santa Clara, CA) for 32 min at 100 °C and endogenous peroxidase was blocked with a hydrogen peroxide solution after incubation of the first primary antibody. Optimized multiplex immunofluorescence was performed with use of the Opal multiplexing method based on tyramide signal amplification. Antibodies used included CD11b (cat. no. 133357, 1:1,500; Abcam), CD11c (cat. no. 97585, 1:50; Cell Signaling Technology), F4/80 (cat. no. 70076, 1:100; Cell Signaling Technology), CD206 (cat. no. NBP-90020, 1:200; Novus Biologicals), and insulin (cat. no. 15848-1-AP, 1:1,000; Proteintech). After incubation with primary and secondary antibodies, fluorescence signals were generated with use of the following Akoya Biosciences Opal tyramide signal amplification fluorophores: Opal 480, Opal 520, Opal 570, Opal 620, and Opal 780 (Akoya Biosciences, Marlborough, MA). DAPI was used for nuclear counterstaining. Between each sequential antibody staining step, slides were incubated in citrate buffer, pH 6 [Cell Conditioning Solution (CC2), cat. no. 980-223; Roche Diagnostics] at 90°C to remove the previous primary and secondary antibody complexes. Multiplex-stained slides were mounted with ProLong Gold Antifade Reagent (cat. no. P36934, Thermo Fisher Scientific) and scanned at ×20 magnification with the Vectra Polaris Automated Imaging System (Akoya Biosciences). Whole slide scans were reviewed and images were captured with use of Phenochart whole slide contextual viewer software (Akoya Biosciences). Spectral unmixing and removal of autofluorescence were performed with use of the inForm software v2.4.10 (Akoya Biosciences), and the resulting images were exported in TIFF format for further analysis. The bioimage analysis and quantitation were performed with QuPath 0.3.0 (39).

Percentages of mice reaching normoglycemia were plotted with Kaplan-Meier curves, and differences in graft survival were compared with the log-rank test. Differences between groups were compared for statistical significance with one-way or two-way ANOVA with Tukey-Kramer post hoc test or Student t test using GraphPad Prism 8.0 (GraphPad, San Diego, CA). Other specific tests used are referenced in figure legends, with P < 0.05 denoting significance. Data are expressed as mean ± SEM.

The data sets generated during or analyzed during the current study are available from the corresponding author on reasonable request.

Substantial islet dysfunction and death occur 2–3 days PT, even in optimal islet transplantation settings (40). We assessed whether coinfusion of hAAT-MSCs with human islets would protect islets from stress-induced cell dysfunction and death. Islets transplanted alone (CTR) or cotransplanted with MSCs infected with control viruses were used as controls. In our model human islets were transplanted into NOD-SCID mice that had been rendered diabetic with streptozotocin injection. Mice with two consecutive BG levels >350 mg/dL were given 1) islets alone (CTR) (n = 6), 2) islets with MSCs (MSCs) (0.4 × 10⁶/kg body wt, n = 21), or 3) islets with hAAT-MSCs (hAAT-MSCs) (0.4 × 10⁶/kg body wt, n = 12). The dose of MSCs used was based on the dosage used in our previous human clinical trial (6). BG levels were measured as a surrogate marker for islet function. Similar to our previous study (38), transplantation of a suboptimal number of islets alone (17 islets/gram body wt, 340 islets for a mouse with body weight 20 g) led to 33%–35.7% (2 of 6 in this study and 10 of 28 in a previous work (38)) of recipients reaching and retaining normoglycemia (BG < 200 mg/dL) until the end of the study, at 60 days PT (Fig. 1A). In contrast, 66.67% (14 of 21, P = 0.052 vs. CTR, log-rank test) of mice receiving islets cotransplanted with MSCs reached and retained normoglycemia. Most notably, 91.67% (11 of 12, P = 0.007 vs. CTR, log-rank test) of mice receiving islets with hAAT-MSCs reached and remained normoglycemia at 60 days PT (Fig. 1A). In addition, hAAT-MSCs groups showed a trend of lower nonfasting BG levels throughout the study (P = 0.056 vs. control) (Fig. 1B and C). MSCs and hAAT-MSCs groups showed higher basal serum human C-peptide levels at 3 and 7 day PT, with the hAAT-MSCs group showing the highest basal human C-peptide levels at 7 days PT (Fig. 1D). At 7 and 28 days PT, better islet function in the MSCs and hAAT-MSCs groups was further demonstrated by lower serum BG levels and areas under the curve and higher stimulated human C-peptide at 0, 15, and 30 min after glucose stimulation during an intravenous BG tolerance test (IVGTT) (Fig. 1E and F, day 7, and Supplementary Fig. 1A and B, day 28). These data suggest that cotransplantation with hAAT-MSCs improves short-term and long-term islet survival and function after transplantation.

To directly visualize transplanted islets, we transplanted 320–330 mouse islets isolated from the firefly Luc transgenic mice (on a C57BL/6 background [Luc⁺ islets]), alone or with MSCs, or hAAT-MSCs (Luc⁻) into diabetic NOD-SCID mice and assessed Luc density by longitudinal in vivo bioluminescence imaging at 1, 3, 7, 14, and 28 days PT. As shown in Fig. 2A and B, a weak Luc signal was observed in control mice receiving Luc⁺ islets alone at 1 day PT. The signal was attenuated at 3 and 7 days, likely because of the detection limit of the Luc assay (41). In contrast, strong Luc signals were detected in mice receiving Luc⁺ islets cotransplanted with MSCs or hAAT-MSCs until day 7 PT (Fig. 2A and B). Luc signal was not detectable in MSCs mice at 14 days PT. Still, it remained detectable until 28 days PT in the hAAT-MSCs mice (Fig. 1A and B), suggesting that cotransplantation with hAAT-MSCs protected early islet cell death and helped retain the islet mass at a detectable level. These results were confirmed ex vivo when human islets were cocultured with MSCs or hAAT-MSCs. Human islets only had limited viability when cultured alone under standard islet culture conditions and showed dramatic cell death at 14 days postculture. Conversely, islets cocultured with MSCs or hAAT-MSCs had significantly lower cell death as identified with the SytoGreen and ethidium bromide live/dead cell staining (Supplementary Fig. 2A) and apoptosis ELISA (Supplementary Fig. 2B), confirming the prosurvival effects of MSCs and hAAT-MSCs.

The route of MSC infusion has been suspected to be related to their protective properties, and cell-to-cell interaction likely plays a vital role in islet survival when MSCs are cotransplanted with islets (42). To track the distribution and the destination of MSCs coinfused with islets, we cotransplanted human islets with 0.5 or 1 × 10⁶ fresh, Luc⁺MSCs into the livers of NOD-SCID mice via intrahepatic infusion and assessed Luc activity at 1, 3, and 7 days PT. Luminescence intensity before infusion confirmed Luc expression level in Luc⁺MSCs (not shown). Low doses of Luc⁺MSCs (0.5 × 10⁶) cotransplanted with human islets were detectable until day 3 PT and high doses of Luc⁺MSCs (1 × 10⁶) were detectable for up to 7 days PT (Fig. 2C and D). By measuring Luc signaling in individual organs, we confirmed that most Luc signaling was restricted to the liver (Fig. 2E). In addition, staining of GFP⁺ and hAAT⁺ cells showed the presence of hAAT-MSCs in the islets within the liver (Fig. 2F and G). A significant increase in serum hAAT level was also detected in hAAT-MSCs mice compared with that of the MSCs group at day 7 PT (Fig. 2H). Nevertheless, these results show for the first time the in vivo distribution and persistence of MSCs cotransplanted with islets by portal vein infusion.

To determine whether sustained islet survival observed in the Luc assay corresponds with better islet survival/mass long-term, we stained serial whole liver sections bearing islet grafts in mice from the different treatment groups with anti-insulin and anti-glucagon antibodies at 28 days PT. We quantified the number of islets and β- and α-cell–positive areas. We found fewer islets from the control liver receiving only islets (mean ± SEM 21.2 ± 3.6 islets/section). In contrast, significantly more islets were observed in the livers of MSCs or hAAT-MSCs groups than in controls (51.6 ± 9.2 islets/section in MSCs and 51.4 ± 7.9 islets per section in hAAT-MSCs groups, P < 0.05 for each) (Fig. 3A–C), further confirming better islet survival in MSCs or hAAT-MSCs–coinfused mice after transplantation. In addition, control islets were mostly damaged with sporadic insulin⁺ cells. In contrast, MSCs or hAAT-MSCs islets showed significantly better shape and preservation of islet structure (Fig. 3B). There was also a higher number of insulin⁺ and glucagon⁺ cells per islet (Supplementary Fig. 2C and D). Therefore, islets cotransplanted with hAAT-MSCs or MSCs showed better preserved islets with stronger insulin staining and larger insulin⁺ and glucagon⁺ areas than controls (Fig. 3D and E), suggesting better preserved islet mass long-term.

We showed in our previous studies that inflammatory responses in the liver after islet transplantation contribute to increased serum proinflammatory cytokine levels associated with islet death and treatment with AAT suppressed proinflammatory cytokine levels (38). To further explore the potential mechanisms of hAAT-MSCs protection, we assessed 23 proinflammatory cytokine/chemokines in serum from control, MSCs, or hAAT-MSCs mice collected on days 1, 3, and 7 PT. Mice receiving islets alone had increased expressions of interleukin-6 (IL-6), IFN-γ, IL-13, granulocyte colony-stimulating factor (G-CSF), MCP-1, IL-9, macrophage inflammatory protein-1α (MIP-1α), and keratinocyte-derived chemokine (KC) at 1–7 days PT (Fig. 4 and Supplementary Fig. 3). In contrast, mice receiving MSCs and islets showed significantly reduced levels of IL-6, KC, G-CSF, MCP-1, IL-13, and MIP-1α. Mice receiving hAAT-MSCs and islets exhibited lower levels of cytokines including IFN-γ, IL-9, G-CSF, and MIP-1α at day 1 PT (Fig. 4A–F and Supplementary Fig. 3). Most of the elevated cytokine levels declined to basal levels by day 7 in all groups (Supplementary Fig. 3). These data suggest that MSCs and hAAT-MSCs cotransplantation suppress systemic inflammation in the recipients, which might consequently reduce PT islet death.

Macrophages play critical roles during the cell-mediated host inflammatory response, leading to islet injury following intrahepatic islet transplantation (43–45). Transient macrophage inhibition, or depletion, prolongs islet allograft survival (38,46,47). MSCs actively attract immune cells, including macrophages, and modulate recruited cells to carry out their anti-inflammatory effects (48,49). We, therefore, tested whether hAAT-MSCs affect the migration of Raw264.7 macrophages seeded in the inset of a Transwell system with the membrane of the insert large enough to allow cells to pass through, in the presence or absence of MSCs and/or human islets (Fig. 5A). Flow cytometry analysis shows mean ± SEM 2.9 ± 0.4% of F4/80⁺ macrophages migrated to the bottom when Raw264.7 cells were cultured alone. Coculture of Raw264.7 cells with only MSCs or hAAT-MSCs attracted slightly more macrophages, with 13.1 ± 2.5% of F4/80⁺ macrophages in the lower well containing MSCs and 6.4 ± 1.5% in the wells containing hAAT-MSCs (Fig. 5B–E). When human islets were included at the bottom well alone in the coculture system, 69.7 ± 1.4% of F4/80⁺ macrophages were found in the bottom wells compared with 33.1 ± 7.1% in the MSCs and islets and 20.5 ± 6.3% in the hAAT-MSCs and islets groups (Fig. 5B and F–H). These data show that when cocultured with macrophages, both MSCs and islets could attract macrophages to migrate. Islets alone induced a dramatic macrophage migration, likely because of factors secreted by dying cells in cultured islets. Coculture with MSCs or hAAT-MSCs suppressed this migratory activity of macrophages toward islets.

We assessed the impact of hAAT-MSCs on M1 and M2 macrophage polarization. We seeded 2 × 10⁵ MSCs or hAAT-MSCs inside the upper chamber and 1 × 10⁵ Raw264.7 cells inside the lower chamber in a Transwell system (Fig. 6A). We then stimulated the cells with IFN-γ to induce M1 or IL-4 to induce M2 macrophage phenotypes. IFN-γ stimulation leads to M1-like macrophage generation featured by a dramatic increase of iNOS⁺ cells analyzed with flow cytometry and mRNA expression of M1 macrophage–related genes including iNOS, TNF-α, and CD11c (Fig. 6B–F). In contrast, the presence of M1-like macrophages and expression of M1-related genes were significantly reduced in cells cocultured with MSCs or hAAT-MSCs (Fig. 6B–F). In addition, treatment of Raw264.7 cells with IL-4 induced an M2-like (CD206⁺) phenotype with increased gene expression of M2-specific marker genes including CD206, arginase 1 (Arg1), and IL-10. Presence of MSCs or hAAT-MSCs groups leads to more M2 generation than control or IL-4 alone groups (Fig. 6G–K). Furthermore, hAAT-MSCs group showed significant suppression of M1 macrophage activation and increase of M2 macrophage activation compared with the MSCs group (Fig. 6C, D, and H–J). These data show that MSCs and hAAT-MSCs can suppress IFN-γ–induced M1 polarization while promoting IL-4–induced M2 macrophage polarization.

Macrophages from the recipient mice play critical roles in mediating the inflammatory response that contributes to graft death or survival following intrahepatic islet transplantation (43–45). In other models, activated (M1) macrophages promoted islet death, while M2 macrophages inhibited islet death (38). We therefore explored the mechanisms of MSCs protection by testing the hypothesis that hAAT-MSCs cotransplantation leads to islet survival via regulation of macrophages in vivo. We first measured the numbers of islets in a liver section and the presence of different subtypes of mouse (recipient) macrophages in livers bearing control or MSCs– or hAAT-MSCs–cotransplanted human islets at 3 days PT using multiplex immunofluorescent staining. Whole liver sections were stained with antibodies against mouse CD11b, F4/80, CD206, CD11c, and insulin. Consistent with data from day 28 (Fig. 3), control mouse liver had the least number of islets, with most of them damaged with only sporadic insulin⁺ cells remaining (Fig. 7A and D). In contrast, mice receiving MSCs or hAAT-MSCs cotransplantation showed significantly more insulin⁺ islets than those receiving islets alone, with hAAT-MSCs mice having the most islet numbers per liver section and the highest number of insulin⁺ cells per islet cluster. However, the difference between hAAT-MSCs and MSCs did not reach significance (Fig. 7B–E). The human and mouse pancreatic islets were stained and used as controls; none of the multiplex-stained antibodies had species cross-reaction (Supplementary Fig. 4). GFP⁺ MSCs were observed in MSCs and hAAT-MSCs islets and hAAT⁺ cells were also observed in hAAT-MSC islets (Fig. 2G), further supporting the presence of MSCs in livers bearing islet grafts.

We then identified the distribution of different subtypes of recipient macrophage within 25 μm of each islet in NOD-SCID liver sections bearing islets at day 3 PT. We found that CD11b⁺ and F4/80⁺ cells were mainly at the periphery of islets, while CD11c⁺ and CD206⁺ cells appeared inside the islets (Fig. 7A and Supplementary Fig. 5). We further counted the numbers of each subtype of macrophages in liver sections from each group. A large number of F4/80⁺, CD11c⁺, and CD11b⁺ cells surround control islets (Fig. 7A–C and F, G, and I). In contrast, MSC mice showed reduced CD11b⁺ and CD11c⁺ and F4/80⁺ cells compared with controls (P < 0.05). hAAT-MSCs islets had reduced CD11c⁺ and F4/80⁺ compared with controls. In addition, the hAAT-MSCs group also had significantly increased CD206⁺ M2-like macrophages and a higher ratio of CD206⁺ to CD11c⁺ cells compared with both the control and MSC group (Fig. 7J), suggesting that hAAT-MSCs mice had both reduced M1 and increased M2 macrophages, which favors islet survival.

One of the major hurdles in clinical islet or β-cell transplantation is early β-cell death PT (2,40,50). Cotransplantation with MSCs or AAT augmentation therapy could improve islet survival and function via different mechanisms (19,38). In this study, we tested the effect of novel engineered human MSCs that overexpressed hAAT, in protecting human islets from death after transplantation into diabetic NOD-SCID mice, and explored mechanisms of function.

Our findings show that coinfusion of hAAT-MSCs significantly preserved/increased the mass and function of cotransplanted human islets, likely via cell-cell interaction and modulation of recipient macrophage migration and activation. Using either Luc⁺ islets or Luc⁺MSCs, we show evidence that MSCs cotransplanted with islets remain in the liver for at least 7 days and mice receiving cotransplantation with MSCs or hAAT-MSCs had larger islet mass at day 3 and 28 PT compared with islets transplanted alone. Using multiplex immunofluorescence staining, we localized for the first time the distribution of different subtypes of recipient macrophage around an islet graft on day 3 PT. The impact of MSCs on macrophage activation was observed in vitro with use of Raw264.7 cells and verified in vivo after islet transplantation. hAAT-MSCs skew the M1-like phenotype toward an M2 phenotype, which might have favored islet survival. In addition to applications in the prevention of type 1 diabetes (24), chronic pancreatitis (51), and graft verse host disease (49), this study offers direct evidence and mechanistic insights that hAAT-MSCs can be used as a novel cellular therapy tool to preserve islet mass and function in the intrahepatic islet transplantation model mimicking the clinical setting.

Direct cell-cell contact or interaction was suggested to be imperative for MSCs to carry out their protective properties and effects (52–54). It allows MSCs to adhere to islets in suspension and even triggers the differentiation of MSCs into insulin-secreting cells. At the same time, indirect coculture promoted islet survival, likely by secretion of trophic factors such as VEGF, Von Willebrand factor, and IL-6 (55). We show that MSCs cotransplanted with islets remained in the liver for at least 7 days PT. The physical presence of MSCs in the liver after portal vein infusion made direct contact with islets possible and consequently protected islets from damage.

Many factors, including instant blood-mediated inflammatory responses, proinflammatory cytokines, and hypoxia, contribute to early islet graft death (56). One of the mechanisms of hAAT-MSC protection might contribute to their anti-inflammatory properties. Mice receiving islets and MSCs or hAAT-MSCs had lower serum IL-6, IL-9, IFN-γ, MCP-1, and other inflammatory cytokines. Consistent with previous studies, we observed that most cytokine levels were elevated in control mice on day 1 PT and their levels reduced on day 7. These data suggest islets cotransplanted with MSCs induce less cytokine production in serum and cause less inflammation, which may offer a positive environment that favors islet survival (38).

Another mechanism by which hAAT-MSCs protected islets from death might have been exerted via modulating recipient macrophage migration and polarization. Macrophages are versatile plastic cells that participate in many critical tasks in the body. They communicate with pancreatic β-cells and play homeostatic roles in type 1 and type 2 diabetes. For example, in NOD mice, monocyte-originated CD11c⁺ cells were increased in inflamed islets and contributed to the destruction of pancreatic β-cells caused by autoimmune responses (57). In obesity, bone-marrow–derived CD11c⁺ macrophages infiltrate adipose tissue in response to tissue damage. However, unresolved activation of these inflammatory macrophages exacerbates tissue injury and is associated with systemic insulin resistance and ablation of CD11c⁺ cells improves insulin sensitivity (58). In this study, staining of mouse islets within a pancreas using multiplex immunofluorescence staining showed that the major islet-resident macrophages were F4/80⁺ within each islet, with a few CD11b⁺ and CD11c⁺ cells at the periphery (Supplementary Fig. 4). Staining of liver tissue sections bearing islets showed that islet grafts were surrounded by different subtypes of recipient macrophages at day 3 PT. In addition to the reduced frequency of each subtype of macrophages, we found that there was a unique distributing pattern of those cells among control, MSC, and hAAT-MSC grafts, i.e., CD11b⁺ and F4/80⁺ macrophages are at the periphery, while CD11c⁺ and CD206⁺ are inside the islet. It is, therefore, reasonable to postulate that they may play different roles in graft survival or damage. Further studies outside the scope of this one are needed to confirm this hypothesis.

Macrophage activation plays a critical role in early islet death PT in both the autologous and allogeneic islet transplantation settings (38,59). Alternatively, activated M2 peritoneal macrophages prevented islet allograft rejection, prolonged allograft survival, and enhanced tolerance (59). In a macrophage depletion mouse model using liposomal clodronate, we found that the presence of M1 macrophages in the liver contributed to graft death after intrahepatic islet transplantation and injection of AAT suppressed macrophage activation (38). In contrast, MSCs promoted islet graft survival mainly by suppressing M1 macrophage, while hAAT-MSCs not only suppress M1 but also increase M2 macrophages (59). This was true both in vivo and in vitro. Although the exact roles of different subtype macrophages in islet graft death are still unclear and worthy of investigation, this study shows for the first time the distribution of different subtypes of recipient macrophages that suggest a distinct function in the islet graft after intraportal human islet transplantation.

hAAT-MSCs seemed to outcompete MSCs in multiple respects. First, mice receiving hAAT-MSCs have the best function, as indicated by better BG control and insulin/C-peptide secretion among the three treatment groups. Secondly, after transplantation, Luc⁺ islets cotransplanted with hAAT-MSCs showed a stronger and more sustained Luc signal. Thirdly, individual islet grafts seemed larger, with more insulin⁺ cells detected in the livers of the hAAT-MSCs group. Fourth, lower serum IFN-γ and IL-9 concentrations were found in the hAAT-MSCs group. Lastly, in addition to suppressing M1 macrophage polarization, hAAT-MSCs promote the generation of M2-like macrophages in vitro and in vivo. Since we observed an increased number of M2 macrophages in the hAAT-MSCs compared with control MSCs, and AAT itself was able to promote M2 macrophage formation (38), it is reasonable to postulate that this effect was mainly due to the effect of AAT. Nevertheless, it is possible that factors secreted from hAAT-MSCs also contributed to such effects. Our data also support that multiple anti-inflammatory effects exerted by hAAT-MSCs contributed to the final positive outcome when islets were cotransplanted into the liver.

In comparison with previous studies using hAAT (Prolastin-C; Grifols) augmentation therapy in the same islet transplantation setting (38,60), the novelty of this study lies in the applications of the engineered hAAT-MSC gene and cell therapy in the islet transplantation model that exceed the beneficial effects of AAT augmentation therapies. We also show for the first time the distribution and longevity of MSCs or hAAT-MSCs cotransplanted with islets via intrahepatic infusion. We further identified the unique immune profiling of macrophage subtypes in transplanted islets using multiplex immunofluorescence staining. In addition, unlike AAT that only suppressed M1 macrophage activation, treatment with hAAT-MSCs not only suppressed M1 but also promoted M2 polarization both ex vivo and in vivo.

Some limitations exist in our study. Although we know MSCs and hAAT-MSCs regulate macrophage phenotypes, we don’t know the gene expression profiles of those macrophages. We did not compare hAAT-MSCs with AAT and MSCs in combination. We only tested a single dose of MSCs and hAAT-MSCs that mirrored our clinical trial (6). Further, our control MSCs had been transfected with lentivirus, and there was no control MSC without this intervention. We also did not study the involvement of donor macrophages in this process, as there were very few donor macrophages detected in transplantation islets, and understanding of the distinct role of different subtypes of macrophages in islet graft death is still incomplete. More studies are needed to elucidate how different macrophages participate in islet graft survival or death. We also do not know whether MSCs or hAAT-MSCs directly impact macrophages or whether MSC reduces islet cell death and inflammation, consequently decreasing macrophage activation. Another limitation of the islet tracing experiments was that mouse islets harvested from the Luc transgenic mice had to be used to visualize islets longitudinally, which may not fully mimic the human islet transplantation situation.

We designed this study by closely mimicking our clinical trial setting (6) by using the same infusion sequence and the same ratio of MSCs and islets. We hope these findings and the related mechanistic insights will shed light on our clinical trial in which islet mass cannot be measured directly. In summary, this study demonstrates that hAAT-MSCs promote islet survival and function via suppression of β-cell death, reduction of inflammation, and regulation of macrophage polarization. hAAT-MSCs that combine the benefits of MSC and AAT therapies may be used in future cellular therapy. These findings facilitate mechanistic insight regarding results of clinical trials using similar strategies.

Acknowledgments. The authors thank Kevin Nguyen for critical language editing.

Funding. This study was supported in part by the National Institutes of Health (DK105183, DK120394, DK118529, and DK125464) and the Department of Veterans Affairs (VA-ORD BLR&D Merit I01BX004536). This study is partially supported by the Translational Science Shared Resource, Hollings Cancer Center, Medical University of South Carolina (P30 CA138313).

Duality of Interest. C.S. has grants from Grifols (manufacturer of Prolastin-C), Arrowhead Pharmaceuticals, and Vertex Pharmaceuticals paid to Medical University of South Carolina on the subject of α-1 antitrypsin deficiency (AATD). He consults for CSL Behring, Inhibrx, Takeda, and Vertex Pharmaceuticals on the topic of AATD. He is a medical director for AlphaNet, a disease management company for AATD. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. W.G. participated in the design of the study, performed experiments, analyzed data, and wrote part of the manuscript. W.H., L.S., and E.G. conducted some experiments. W.C. participated in the human islet study. K.M. participated in the study design. C.S. participated in study design and data discussion and revised the manuscript. H.W. designed the study, analyzed data, and wrote the manuscript. All authors read and approved the final manuscript. H.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.