Evaluation of biomaterial scaffold delivery of IL-33 as a localized immunomodulatory agent to support cell transplantation in adipose tissue

Abstract

Introduction

The development of novel immunomodulatory strategies that might decrease the need for systemic immune suppression would greatly enable the utility of cell-based therapies. Cell transplantation on biomaterial scaffolds offers a unique opportunity to engineer a site to locally polarize immunogenic antigen generation. Herein, we investigated the localized delivery of IL-33, which is a novel cytokine that has been shown to have beneficial immunomodulatory effects in certain transplant models as mediating anti-inflammatory properties in the adipose tissue, to determine its feasibility for use as an immunomodulatory agent.

Results

Localized IL-33 delivery from poly(lactide-co-glycolide) (PLG) scaffolds implanted into the epididymal fat specifically increased the Foxp3+ population of CD4+ T cells in both blank scaffold implants and scaffolds seeded with allogeneic islets. In allogeneic islet transplantation, we found IL-33 delivery results in a local upregulation of graft-protective T cells where 80% of the local CD4+ population is Foxp3+ and overall numbers of graft destructive CD8+ T cells are decreased, resulting in a prolonged graft survival. Interestingly, local IL-33 also delayed islet engraftment by primarily inducing a local upregulation of Th2 cytokines, including IL-4 and IL-5, leading to increased populations of ST2+ Type 2 innate lymphoid cells (ILC2s) and Siglec F+ eosinophils.

Conclusions

These results suggest that local IL-33 delivery from biomaterial scaffolds can be used to increase Tregs enriched in adipose tissue and reduce graft-destructive T cell populations but may also promote innate cell populations that can delay cell engraftment.

Graphical abstract

1. Introduction

The vast majority of human allogeneic islet transplantations for the treatment of Type 1 Diabetes Mellitus (T1DM) have been performed by infusion of donor islets into the hepatic portal vein. While these transplants have shown promise in restoring insulin independence, it is widely acknowledged that the hepatic vasculature is not an ideal transplant site due to initial loss of graft mass caused by the instant blood mediated inflammatory reaction (IBMIR), suboptimal engraftment related to poor local oxygenation and tissue revascularization, and the difficulties controlling the immune response to donor cells with systemic immunosuppressants that may also have islet toxicity.^1,2 In searching for alternative transplant sites, the omentum, a major adipose tissue depot, has been identified as a promising target due to its large transplant area, similarity in portal drainage to the native pancreas environment, and ease of surgical accessibility.^3-5 Omental transplants can be modeled in mice using the perigonadal fat pad (epididymal fat pad in male mice).⁶ As a non-immunoprivileged site, the usage of adipose tissue as a novel transplant site requires an immune intervention strategy to protect the graft from the host immune system. While CD4+ and CD8+ T cells are primary mediators of transplant rejection, innate immunity can also contribute significantly to graft survival.⁷ Adipose tissue constitutes a unique immune microenvironment that is maintained in homeostasis between an inflammatory and anti-inflammatory state by many of the same effector cells that mediate transplant acceptance and rejection.⁸

Current clinical trials for human omental islet transplantation use biologic scaffolds to support islet transplantation.³ While the aforementioned studies used scaffolds produced from autologous plasma and thrombin, a wide variety of materials including synthetic polymers like PLG can be used to create scaffolds with tunable mechanical and structural properties.⁹ Scaffolds can facilitate integration into native tissue and represent a customizable environment that can be designed to incorporate additional intervention strategies to modulate the local environment, such as co-transplantation of synergistic cell types, surface coupling of modifying extracellular matrix proteins and other ligands, and the incorporation of releasable factors.^10-13

Biomaterial scaffolds can be designed to incorporate protein for localized release. Local delivery of cytokines can be used to interact directly with infiltrating immune cell types that mediate graft failure and rejection at a primary site of interaction with donor tissue. We have previously reported on PLG scaffolds designed to support islet transplantation into the epididymal fat pad that also release TGF-β1 extend islet allograft survival through local depletion of leukocytes.¹⁴ In considering adipose tissue as a potential site for cell transplantation, the identification of site-specific immune factors could lead to novel therapeutic targets. IL-33 has become a topic of interest in adipose tissue immunology, where it has been shown to interact with a number of different locally enriched cell types expressing the IL-33 receptor ST2, including CD4+ Th2 and regulatory T cells (Tregs), macrophages, and innate lymphoid group 2 cells (ILC2s).^15-18 IL-33 is an IL-1 family cytokine that been shown to have both pro- and anti-inflammatory properties depending on the immunological context.^19,20 In the realm of allogeneic transplant studies, IL-33 has been shown to have graft-protective effect in murine cardiac and skin transplant models.^21-23

In this article, we sought to use our PLG scaffold platform to release IL-33 that would target the polarization of local immunoregulatory cell populations. Initial studies investigate the impact of IL-33 on the local tissue, with a focus on the expansion of ST2+ Tregs. We characterize the effects of local release of IL-33 on immune cells localized within an adipose tissue-biomaterial scaffold environment and explore the ramifications for future use of IL-33 as a tolerogenic factor for allogeneic islet transplant. The ability to provide localized immune interventions may lessen the need for systemic immune suppression and its associated complications and side effects.

2. Methods

2.1 PLG Scaffold Production

2% or 6% W/V 75:25 PLG (Lakeshore Biomaterials) was dissolved in dichloromethane (DCM) and homogenized in 1% poly-vinyl alcohol using a Polytron 3100 homogenizer to create the PLG microspheres. Particles were mixed for 3 hours to evaporate the organic solvent, then washed with water to remove excess PVA. After washing, the microspheres were frozen in liquid nitrogen and lyophilized.

Control and IL-33 loaded layered scaffolds were created as described previously with modifications. Control inner layers were created by mixing 2 mg 2% PLG with 1 mg BSA and 1 mg mannitol in a total volume of 100 μL water. The mixture was frozen in liquid nitrogen, lyophilized, and pressed into a disc (3mm diameter, 100 μm thick) using a handpress. For IL-33 scaffolds, recombinant murine IL-33 (Biolegend) was added to the above mixture. Full scaffolds were formed by sandwiching the pressed inner layer between two outer layers comprised of 1.25 mg 6% PLG and 37.5 mg NaCL (weights +/− 5%). Scaffolds were compressed in a 5mm dye under 1500 pounds per square inch (psi) using a Carver Press. Scaffolds were gas foamed overnight after equilibration to 800 psi under CO₂ gas and stored at −20°C.

2.2 Animals

Animals were obtained from Jackson Labs or Charles River Laboratories. 8-12 week-old C57BL/6 males were used for scaffold recipients. 8-12 week-old Balb/c or C57BL/6 males were used for allogeneic and syngeneic islet isolations respectively. For syngeneic and allogeneic islet transplant studies, diabetes was induced by a single i.p. injection of 180mg/kg body weight streptozotocin delivered after a 4-6 hour fast (Sigma Aldrich). Mice were considered diabetic and eligible for transplant after two consecutive non-fasting blood glucose measurements over 350mg/dL. All procedures were approved by the Northwestern University Animal Care and Use Committee (ACUC) or University of Michigan Unit for Laboratory Animal Medicine (ULAM).

2.3 IL-33 in vitro release assay

Scaffolds were leached in 10mL milliQ water for 1 hour to remove the porogen and incubated in 1mL PBS supplemented with 1% BSA and Pen-Strep. Scaffolds were placed on a shaker at 37°C. At the time of collection, supernatants were spun down, snap-frozen on liquid nitrogen and stored until analysis by ELISA. Scaffolds were then incubated in fresh PBS with BSA and Pen-Strep until the next collection time point.

2.4 IL-33 in vitro bioactivity assay

Naïve T cells were isolated from the spleens of 8-12 week old C57 males using the Miltenyi Biotec Naïve T cell isolation kit. Spleens were crushed between frosted glass slides and filtered through 70 μm cell strainers. Red blood cells were lysed using ACK buffer (Thermo Fisher Scientific). The cell pellet was incubated first with a biotinylated antibody cocktail containing antibodies against CD8a, CD11b, CD11c, CD19, CD25, CD45R (B220), CD49b, CD115, MHCII, TER119, and TCRγ/δ, then anti-biotin magnetic beads. The cells were then passed through a MACS LS column and untouched CD4+ naïve T cells were collected and resuspended in RPMI 1640 supplemented with 10% FBS and 1× Pen-Strep. Naive Cells were incubated for 1 hour at 37°C before being transferred to wells surface coated with 3 μg/mL anti-mouse CD3ε. 5*10⁵ CD4+ T cells were added to each well, supplemented with 2 μg/mL anti-mouse CD28. Blank or IL-33 scaffolds were added to wells and cells were incubated 72 hours at 37°C. Cell culture supernatants were spun down to remove cells, snap-frozen on liquid nitrogen, and stored at −80°C until analysis. Samples were sent to the University of Michigan ELISA core for analysis of IL-13.

2.5 Scaffold Implants

Prior to implantation, scaffolds were leached in 10mL of milliQ water per scaffold for 1 hour. Water changed after 30 minutes. Scaffolds were disinfected in 70% ethanol for 1 minute, then washed twice in media supplemented with 10% FBS. Mice were anesthetized using isoflurane (2% flow rate). The abdomen of each mouse was shaved and prepared in a sterile fashion with 3 successive administrations of Betadine and ethanol. The intraperitoneal space was exposed by a lower abdominal midline excision, the epididymal fat pad was exposed, and scaffolds were wrapped securely and returned to the cavity. The abdominal wall was closed with a running stitch using 5-0 sutures and the skin was secured with wound clips. Mice received a post-operative subcutaneous injection of Carprofen (5 mg/kg) 24 after implant and surgery sites were monitored until termination of the study or for 10 days until clips were removed.

2.6 Islet Isolation and Transplantation

Pancreata from euthanized mice were inflated with 0.51 mg/mL collagenase XI (Sigma) via bile duct cannulation and digested for 15 minutes in a 37°C water bath with periodic agitation. After filtration through a mesh screen, islets were separated from acinar tissue using a density gradient (Biochrom or Histopaque). Islets were picked from the gradient interface and washed thoroughly before being transplanted immediately. Islets were counted and seeded onto as single side of the scaffold using a customized glass pipette tool. Scaffolds were then placed directly onto the exposed epididymal fat pad and re-inserted into the abdominal cavity. To monitor graft function, non-fasting blood glucose readings were taken at least three times during the first week post-transplant then twice a week for the remainder of the study. For syngeneic transplants, engraftment was determined by establishment of stable normoglycemia (<200mg/dL blood glucose). For allogeneic transplants, graft rejection was confirmed by consecutive days of blood glucose readings exceeding 250mg/dL after stable normoglycemia was achieved.

2.7 Flow Cytometry

Mice were euthanized by cervical dislocation under isoflurane-induced anesthesia. Tissues were harvested immediately and stored in HBSS on ice. For scaffold implants, excess adipose tissue was trimmed away to isolate the immediate scaffold environment. Scaffolds were weighed to normalize for variable tissue collection. Spleen samples were mechanically disrupted by agitation between frosted glass slides. The resulting tissue homogenate was filtered through a 70 μm cell strainer and washed with MACS (PBS supplemented with 2mM EDTA and 0.5% BSA). For scaffold implants and adipose tissue, enzymatic digestion was used to create a single cell suspension. Tissues were weighed, shredded, and added to 1mg/mL collagenase Type II. Tissue was incubated in a 37° C water bath for 30 minutes with gentle shaking every 5 minutes. 100 μL of 0.5 M EDTA was added to each tube to a final concentration of 10 mM and incubated for an additional 5 minutes at 37° C. Tissue homogenate was strained through a 70 μm filter and washed with MACS. The resulting cell pellets were then incubated with 1mL ACK buffer on ice to lyse the red blood cells and washed with MACS. In preparation for staining with Live/Dead fixable stain, cells were washed with PBS.

Live Dead Fixable Violet stain (Thermo Fisher Scientific) was used for removal of dead cells from analysis. The Foxp3/Transcription Factor Staining Buffer (Ebioscience) was used for cells requiring intracellular staining. The following conjugated antibodies (clone) were purchased for analysis from Biolegend, Ebioscience, or Biorad: CD3ε (145-2c11), CD4 (RM4-5), CD5 (53-7.3), CD8a (53-6.7), CD11b (M1/70) CD11c (N418), CD25 (PC61), CD45 (30-F11), CD45R/B220R (RA3-6B2), CD127 (A7R34), CD301 (ER-MP23), F4/80 (BM8), FcεRIα (MAR-1), Foxp3 (FJK-16s), Ly6C (HK1.4), Ly6G (1A8), NK1.1 (PK136), Siglec F (1RNM44N), ST2 (RMST2-2). Isotype antibodies were used to establish gating. The full gating scheme used for CD4+ and CD8+ cells is shown in Supplementary Figure 1.

Samples were analyzed on the DAKO Cyan 5 ADP. In order to derive absolute cell numbers in each scaffold, 50 μL of Absolute Countbright Beads (Thermo Fisher Scientific) were added to each sample and gated in each sample to use as an internal control. Approximately 10,000 beads were counted per sample and overall cell numbers were adjusted based on the expected numbers of beads specified by each lot.

2.8 Gene Expression

RNA isolation was prepared from scaffold implants using the Qiagen RNeasy Kit with modifications for fatty tissue. Scaffold implants were extracted from mice and immediately snap frozen in isopentane on dry ice for storage at −80°C. Frozen grafts were homogenized in 1mL Trizol (Ambion) using a rotor-stator homogenizer at 10,000 rpm. Homogenates were centrifuged to remove adipocytes prior to RNA extraction. Choloroform was added to the clarified supernatant and the resulting mixture was shaken for 15 seconds, incubated at room temperature for 2-3 minutes, then centrifuged for 15 minutes 12000 g 4°C. The upper clear organic phase was then transferred to a new tube, mixed with an equal volume of 70% ethanol, and pipetted onto an RNeasy mini spin column. An on-column DNase digest was performed by applying Rnase free DNAse (Qiagen) to the membrane. RNA was eluted using 50uL of nuclease free water. RNA concentration and purity was assessed by Nanodrop 2000.

Two-step RT-PCR was used to assess changes in gene expression. RNA was converted to cDNA using the iScript cDNA converstion kit. Qiagen Sybr Green PCR master mix was used for PCR. Primers were designed using Mouse Primer Depot.²⁴ Gene expression was calculated using the 2^−ΔΔCq method. Hrpt1 was used as the housekeeping gene for normalization. Primers used for analysis are listed in Table 1.

Table 1.

qPCR Primers

Il10 For	ATCGATTTCTCCCCTGTGAA
Il10 Rev	TGTCAAATTCATTCATGGCCT
Foxp3 For	TGGCAGAGAGGTATTGAGGG
Foxp3 Rev	CTCGTCTGAAGGCAGAGTCA
Il2 For	AACTCCCCAGGATGCTCAC
Il2 Rev	CGCAGAGGTCCAAGTTCATC
Ifn-γ For	ACAGCAAGGCGAAAAAGGAT
Ifn-γ Rev	TGAGCTCATTGAATGCTTGG
Tnf-α For	CCACCACGCTCTTCTGTCTAC
Tnf-α Rev	AGGGTCTGGGCCATAGAACT
St2 For	CGTGTCCAACAATTGACCTG
St2 Rev	CAAGTAGGACCTGTGTGCCC
Ccl2 For	CCTGCTGTTCACAGTTGCC
Ccl2 Rev	ATTGGGATCATCTTGCTGGT
Arg1 For	AGAGATTATCGGAGCGCCTT
Arg1 Rev	TTTTTCCAGCAGACCAGCTT
Nos2 For	TGAAGAAAACCCCTTGTGCT
Nos2 Rev	TTCTGTGCTGTCCCAGTGAG
Il4 For	TGAACGAGGTCACAGGAGAA
Il4 Rev	CGAGCTCACTCTCTGTGGTG
Il13 For	TGTGTCTCTCCCTCTGACCC
Il13 Rev	CACACTCCATACCATGCTGC
Il6 For	TGATGCACTTGCAGAAAACA
Il6 Rev	ACCAGAGGAAATTTTCAATAGGC
Hprt1 For	TCCTCCTCAGACCGCTTTT
Hprt1 Rev	CATAACCTGGTTCATCATCGC

2.9 ELISA and Luminex analysis

Cell culture and scaffold wash supernatants were measured for cytokine concentration using R&D ELISA Duoset kits. Supernatants were collected and stored at −80°C before analysis. Tissue homogenate protein levels were assayed using the Milliplex MAP Mouse Cytokine/Chemokine Magnetic Bead Panel – Premixed 32 Plex (Millipore). To prepare samples for analysis, tissues were homogenized in ice-cold PBS supplemented with 1× protease inhibitor cocktail (Pierce) using a rotor-stator homogenizer. Supernatants were centrifuged to remove adipocytes prior to analysis. Protein concentration was determined by BCA (Pierce). All samples were diluted to 1 mg/mL before analysis.

2.10 Statistical analysis

Statistical analyses were performed using GraphPad Prism (La Jolla, Ca). Graphs depict mean and standard error of the mean (SEM). Unpaired student t-test were used to calculate statistical significance unless otherwise indicated.

3. Results

3.1 IL-33 incorporated into PLG scaffold design retains in vitro bioactivity

We first tested the inclusion of carrier protein into the inner layer of our protein-incorporating scaffold to increase protein loading and recovery. In vitro protein release from scaffolds loaded with 2 μg of IL-33 showed an average total recovery of 1000 ng of IL-33 with BSA incorporation compared to 200 ng using the previous BSA-free design, validating our inclusion of a carrier protein (Figure 1a). BSA incorporation also decreased the amount of IL-33 lost during the initial leaching steps. The scaffold shows a bolus release of protein where the majority of protein is recovered within a day of incubation. In vitro bioactivity of IL-33 scaffolds was determined by the induction of IL-13 from activated naïve T cells.²⁵ Incubation with IL-33 scaffolds significantly increased production of IL-13 from naive T cells undergoing anti-CD3/28 activation (Figure 1b).

Figure 1. IL-33 Scaffold Characterization.

(A) In vitro release profile collected at Day 1, 3, and 7 from layered scaffolds loaded with 2 μg IL33 with or without 1mg of BSA incorporated as a carrier protein (B) Supernatant concentration of IL-13 detected by ELISA from naïve T cells activated by platebound anti-CD3. Blank condition refers to a BSA scaffold without IL-33. N = 4 scaffolds per condition ***p < .001. Statistics determined by unpaired t test. (1 column)

3.2 Implantation of PLG scaffold alters local composition of resident innate immune cells

We initially investigated the impact of implantation of control scaffolds on the local immune cells in adipose tissue, which has a unique microenvironment that controls metabolic processes. In normal adipose tissue and day 7 post-implant scaffolds, close to 70% of recovered immune cells express CD11b+ and belong to the myeloid lineage (Figure 2A). The distinct population of F4/80 Hi CD11b+ macrophages previously identified as a resident population in normal adipose tissue is reduced in scaffolds recovered seven days post-implant.²⁶ In characterizing resident macrophage phenotype, 70% of tissue resident macrophages in adipose tissue had a surface expression of CD11c− CD301+ associated with the anti-inflammatory alternatively activated phenotype compared to 20% expressing the CD11c+ CD301− inflammatory macrophage phenotype (Figure 2B). In the scaffold, this ratio was reversed with 70% expressing the inflammatory phenotype compared to 10% expressing the alternatively activated phenotype. Close to 40% of the identified CD45+ cells expressed a SSC Hi CD11b+ F4/80 Int Ly6G− Siglec F+ phenotype consistent with eosinophils (Figure 2C). Other myeloid cell subtypes also residing within the scaffold include aforementioned tissue resident macrophages (F4/80 Hi Ly6G− Siglec F−), neutrophils (F4/80− Ly6C+ Ly6G+ Siglec F−), and monocytes (F4/80− Ly6G− Siglec F− Ly6C Hi), though all populations were significantly lower than the aforementioned eosinophils.

Figure 2. Innate immune cell environment of PLG scaffold implants.

(A) Representative flow cytometry plots comparing F4/80 expression on live CD45+ CD11b+ cells between unimplanted adipose tissue and PLG scaffolds seven days after implant (B) Representative flow plots and quantification of CD11c+ and CD301+ subsets of F4/80 hi cells. (C) Gating scheme and quantification of subsets of CD11b+ cells isolated from PLG scaffolds including (1) eosinophils (CD11b+ F4/80 Int Siglec F+), (2) macrophages (Cd11b+ F4/80 hi Siglec F− Ly6G−), (3) neutrophils (CD11b+ F4/80 intermediate Ly6G+ Ly6C Int), and (4) monocytes (Cd11b+ F4/80 Int Ly6C Hi). Subsets are quantified as percentage of total live CD45+ cells isolated from the scaffold. N = 4 scaffolds per condition. Graphs depict mean ± SEM. (1.5 column)

3.3 Scaffold-mediated IL-33 delivery expands local CD4+ Foxp3+ Tregs

Localized delivery of IL-33 was subsequently investigated for its ability to increase local populations of Foxp3+ Tregs, in particular the ST2+ subpopulation naturally enriched in adipose tissue. For in vivo studies, we increased the loading amount in the scaffold to 5 μg IL-33 to match the cumulative amounts of protein via i.p. injection administered in previously published allogeneic transplant models.^21,22 IL-33 release in the scaffold environment increased the proportion of Foxp3+ cells within the total CD4+ population from 25% to 45% (Figure 3A). We did not detect a statistically significant increase in the percentage of Tregs expressing ST2+, suggesting expansion of Tregs may be occurring by both a direct engagement between IL-33 and ST2+ Tregs in addition to another indirect IL-33 induced mechanism (Figure 3B). We also analyzed splenic populations of regulatory T cells to identify whether the effects of local IL-33 delivery could be detected systemically. We did not detect a significant change the in percentage of Foxp3+ Tregs present with the spleen nor an increase in ST2+ expression amongst Tregs, indicating the effects of IL-33 were concentrated within the localized scaffold environment (Figure 3C).

Figure 3. IL-33 increases local population of CD4+ Foxp3+ Tregs in blank scaffold implant.

(A) Representative flow cytometry plots and quantification of percentages and overall numbers of live CD3+ CD4+ Foxp3+ Tregs isolated from control and IL-33 scaffolds seven days after implant. Frequency of ST2+ Foxp3+ CD4+ T cells seven days after implant as quantified by flow cytometry (B) Representative flow cytometry plot and quantification of percentages and overall numbers of ST2+ Foxp3+ CD4+ T cells seven days after implant (C) Representative flow cytometry plots and quantification of percentages of CD4+ Foxp3+ cells and ST2+ Tregs isolated from spleens of animals receiving control or IL-33 scaffolds. Overall numbers are normalized to weight of tissue immediately after isolation. Graphs depict mean ± SEM. N = 5-6 scaffolds per condition. **p < 0.01 ***p < 0.001 ****p < 0.0001 Statistics determined by unpaired t-test. (1.5 column)

3.4 IL-33 decreases CD8+ T cells and increases ST2+ Tregs in the presence of allogeneic cells

We next focused on the impact of IL-33 on the local adaptive immune response by tracking CD4+ and CD8+ populations in scaffolds transplanted with allogeneic islets. Release of local IL- 33 led to a significant decrease in the number of CD3+ CD8+ cells recovered from the allograft environment (Figure 4A). Moreover, the ratio of CD4 to CD8 T cells increased from an average of 0.75 to 4 (p<0.01) (Figure 4B). Taken together, localized IL-33 release decreases the CD8+ T cell response to allogeneic cells within the scaffold at day 7. In the allograft model, the percentage of Foxp3+ cells in the total CD4+ population increases from 40% to 75% with the addition of IL-33 (Figure 4C). Unlike the blank implantation model, IL-33 specifically expanded the ST2+ population of CD4+ Foxp3+ Tregs, increasing the percentage of ST2+ Tregs from 35% to 60% (Figure 4D). A significant increase in ST2 expression was also detected in the ST2+ Treg population.

Figure 4. IL-33 decreases CD8+ T cells and expands ST2+ Tregs in the presence of allogeneic islets.

(A) Representative flow cytometry plots and quantification of CD4+ and CD8+ populations of live CD3+ cells isolated from control and IL-33 scaffolds seven days after transplant with allogeneic islets (B) Ratio of overall CD4:CD8 cells isolated from scaffolds (C) Representative flow cytometry plots and quantification of percentages and overall numbers of live CD3+ CD4+ Foxp3+ Tregs isolated from allografts (D) Representative flow cytometry plot and quantification of percentages and overall numbers of ST2+ Foxp3+ CD4+ T cells 7 days after transplant. Overall numbers are normalized to weight of tissue immediately after isolation. Graphs depict mean ± SEM. N = 5-6 scaffolds per condition. *p < 0.05 **p < 0.01 Statistics determined by unpaired t-test. (2 column)

3.5 IL-33 delivery extends allograft survival but delays engraftment

We then analyzed the functional effect of IL-33 delivery on long-term islet allograft survival. 250 islets isolated from Balb/c males were seeded onto BSA or IL-33 scaffolds and transplanted into streptozotocin-induced diabetic C57BL/6 mice. We observed that control scaffolds rapidly restored normoglycemia and had a similar survival time to our previous published control layered scaffolds, demonstrating incorporation of BSA into the inner layer does not produce a local immune response that compromises local function of islets (Figure 5A). Release of IL-33 signficantly delays early engraftment of allogeneic islets and diabetes reversal compared to control scaffolds (Figure 5B, C). However, all allogeneic IL-33 transplants recovered to record consecutive days of normoglycemia after Day 10 following early graft dysfunction, indicating hyperglycemia was not due to graft failure. The median survival time of allogeneic grafts on IL- 33 scaffolds significantly increases from 14 to 33 days compared to control scaffolds (Figure 5D). The delay in engraftment appears to be intrinsic to IL-33 release given a similar delay in engraftment was also noted when using a syngeneic transplant model where C57BL/6 islets were transplanted via the same IL-33 releasing scaffolds this is not observed with the control scaffolds (Figure 5E,F).

Figure 5. IL-33 extends islet allograft survival but delays engraftment.

Individual nonfasting blood glucose measurements for diabetic recipients receiving 250 allogeneic islets transplanted on (A) control scaffolds and (B) IL-33 scaffolds. Mice were electively euthanized after experiencing consecutive days of hyperglycemia after stable engraftment in addition to weight loss and other signs of declining health. Solid line indicates 250mg/dL used for assessing graft rejection. Kaplan-Meyer analysis of (C) diabetes reversal and (D) survival of islet allografts over time. Three consecutive readings under 200mg/dL blood glucose indicated diabetes reversal. Two consecutive readings above 250mg/dL with no recovery of consecutive days of normoglycemia indicated graft failure. N = 7 mice per condition. * p<0.05. Statistics determined by log-rank test. Average blood glucose readings of mice receiving 250 syngeneic islets on (E) control or (F) IL-33 scaffolds. Grafts were removed at day 64 (indicated by arrow) to confirm recipients revert to diabetes upon removal of transplanted islets. Solid line indicates 200mg/dL used to define normoglycemia. Graphs depict mean± SEM. N = 3 mice per condition. (2 column)

3.6 Scaffold delivery of IL-33 induces expression of a local Type 2 immune response

Gene expression in the scaffold following localized IL-33 delivery indicated that IL-33 was responsible for inducing a local Th2 polarizing response. RNA was isolated from scaffolds excised three days after implantation without the presence of islets, in order to better assess the direct impact of IL-33 release on the local immune microenvironment. Increased expression (100-300 fold) was detected for the cytokines IL-4, IL-5, and IL-13. (Figure 6A). We also identified a similar level of upregulation for Ym1, which is associated with an alternatively activated phenotype in macrophages but also can act as an eosinophil chemotactic factor. Nos2 and Arg1 showed increased RNA expression levels of 10-15 fold (Figure 6B). While Nos2 and Arg1 are normally used to distinguish M1 and M2 macrophages respectively, they are also expressed by a variety of other immune cell lineages that may be included in the total RNA isolated from the scaffold, including eosinophils and ILC2s. We did not detect any change in expression of IFN-γ or TNF-α, though an increase in IL-6 was observed (Figure 6C). Consistent with an increase in Tregs, we detected an increase in Foxp3 expression (Figure 6D). However, we observed a significant decrease in IL-10 expression and did not detect a change in TGF-β1 (data not shown).

Figure 6. IL-33 induces RNA expression of Th2 cytokines 3 days after implant.

(a) Gene expression measured by two-step RT-PCR of IL-4, IL-5, IL-13, and Ym1 from RNA isolated from scaffolds harvested 3 days. (b) Gene expression of Arg1 and Nos2 (c) Gene expression of IFN-γ, TNF-α, IL-2, and IL-6. (d) Gene expression IL-10 and FOXP3. Expression was calculated using 2^−ΔΔCq method. Experimental genes were normalized to the housekeeping gene HPRT1. N = 4 scaffolds per condition. *p<0.05 ** p<0.01 ***p<0.001 ****p<0.0001 Statistics determined by unpaired t-test. (2 column)

In order to follow the longer-term effects of the observed gene expression changes induced by IL-33, we performed a Luminex analysis on homogenized scaffold implants seven days after implant without islets, assessing the cytokine milieu closer to the time of graft failure. Significant increases in IL-4 and IL-5 were detected in the IL-33 condition though neither IL-10 nor IL-13 were detectable within the local environment (Figure 7A). While IL-6 was also significantly increased in the IL-33 scaffold, cytokines indicative of a Th1, Th17, or CD8 expanding response such as IFN-γ, TNF-α, IL-1β, IL-12p70, and IL-17 were under the detectable limit (Figure 7B). A significant decrease in T cell homeostasis cytokines IL-2 and IL-15 was noted in the IL-33 scaffold (Figure 7C). While a number of chemokines such as Eotaxin, KC, MCP-1 were expressed at detectable concentrations in both scaffold conditions, we noted a significant decrease in MIG and IP-10 expression in the IL-33 scaffold conditions, two chemokines typically induced by IFN-γ (Figure 7D).

Figure 7. IL-33 scaffold environment has higher protein expression of Th2 cytokines 7 days after implant.

Cytokines were detected by Luminex analysis of protein homogenates from scaffolds. (a) Th2-associated cytokines detected from the scaffold. (b) Th1 and Th17-associated cytokines detected from the scaffold. (c) T cell proliferation-associated cytokines detected from the scaffold. (d) Chemokines detected from the scaffold. 1mg of total protein was loaded for each sample in technical duplicates. Graphs depict mean ± SEM. N = 4 scaffolds per condition *p<0.05 **p<0.01. Statistics determined by unpaired t-test. (2 column)

3.7 Scaffold delivery of IL-33 expands eosinophils and ST2+ ILC2 cells

We next assessed the innate immune cell populations present within the scaffold seven days after implant without islets. Delivery of IL-33 tripled the number of live CD45+ cells recovered from the scaffold at day 7. As previously noted, the control scaffold is already enriched for eosinophils. With the local delivery of IL-33, the eosinophil population increased from 40% to 70% of the total CD45+ population in the SVF (Figure 8A). While the overall numbers of other myeloid lineages remained constant between BSA and IL-33 scaffolds, the number of eosinophils increased from 1500 to 10,000 cells/mg tissue. IL-33 has been shown to potently increase ST2+ ILC2s in adipose tissue.²⁷ ILC2s are detected as lineage negative (CD11b, CD11c, CD45R/B220, Ly6G, Ly6C, CD3, CD4, CD5, FcεRIα, NK1.1) SSC lo ST2+ CD127+ (Figure 8B). While essentially absent in the BSA scaffold, IL-33 delivery significantly expands ILC2s to 700 cells/mg of tissue, corresponding to roughly 0.05% of the total CD45+ population. We did not detect any increase in the CD11b+ F4/80 hi population of macrophages present within the scaffold, nor did we detect significant changes in the polarization state away from a classically inflammatory CD11c+ phenotype.

Figure 8. IL-33 delivery expands eosinophils and ILC2s.

(A) Representative flow cytometry plots and quantification of percentage and overall number of CD45+ innate cell subsets as defined in Figure 2, focusing on changes in the eosinophil population (Siglec F+ F4/80 med) (B) Representative flow cytometry plots and quantification of percentage and overall number of ILC2 amongst CD45+ SSC lo cells. Graphs depict mean ± SEM. N = 4 scaffolds per condition ***p<0.001 ****p<0.0001. Statistics determined by unpaired t-test. (1.5 colun)

4. Discussion

In this manuscript, we report the effects of the novel immunomodulatory cytokine IL-33 when delivered locally to the epididymal fat pad via release from a biomaterial scaffold. IL-33 is a pleiotropic cytokine with a number of putative targets among adipose tissue resident cell lineages. Like most cytokines, IL-33 can be found to contribute to divergent effects depending on the combinations of factors that are being co-expressed.^20,28 IL-33 monotherapy studies via systemic administration have generally found IL-33 to extend graft survival in allogeneic cell transplantation.^21-23

IL-33 expands CD4+ Foxp3+ Tregs both in a blank implant model as well as the allograft model. A number of groups have reported expansion of Foxp3+ Tregs through IL-33 administration, including specifically within the adipose tissue.^21,29,30 Our work confirms IL-33 delivery into adipose tissue biomaterial scaffold environment robustly increases CD4+ Foxp3+ cells present in the graft, with and without the presence of the allogeneic antigen. The ability to expand Tregs in the presence of alloantigen is crucial given the differences in T cell mediated responses when introducing allogeneic tissue, which can be seen through differences in the CD4+ and CD8+ populations between blank scaffolds and scaffolds containing allogeneic islets. Interestingly, the ST2+ population of Tregs that would theoretically be responsive to IL-33 was only significantly increased amongst the total Foxp3+ pool with the presence of allogeneic islets. This may suggest that IL-33 induced Treg proliferation is enhanced by TCR activation. IL-33 also has pro-inflammatory capabilities in certain disease models and can activate CD8+ cells, which are a key component of allograft rejection.^31,32 We did not observe an increase in CD8+ cells in the graft with the addition of IL-33, which re-emphasizes the importance that the presence of other cytokines plays in controlling immune responses within tissues.

While it is unclear from our experiment whether the Foxp3+ Tregs within the scaffold take an active role in immune suppression, IL-33 appears to polarize T cells towards non-destructive graft phenotypes. We did not observe an enhancement in gene expression of common immunosuppressive factors IL-10 and TGF-β1. In the case of IL-10, a decrease in both day 3 gene expression and day 7 protein expression was present, making immune suppression by soluble factor release unlikely. However, the fact that 70% of CD4+ T cells recovered from scaffolds transplanted with allogeneic islets expressed Foxp3+ is consistent with the notion that there likely was a decrease in the presence of Th1 CD4+ cells that are linked with acute cell- mediated rejection. While IL-10 and TGF-β1 were not upregulated even with the expansion of local Tregs, a known mechanism of Treg-mediated immune suppression is the ability to consume IL-2 through expression of the high-affinity IL-2 receptor subunit CD25 while not producing additional IL-2 helping activate additional T cells. Recent work has shown in vivo Treg-associated IL-2 depletion can suppress expansion of CD8+ cells, which is consistent with the allogeneic experiments.³³ Though IL-33 has been shown in some cases to promote a Th1 response that would be graft destructive, we do not detect the presence of any inflammatory cytokines associated with a Th1 or Th17 response besides IL-6. This lack of Th1 response is consistent the low levels of IL12p70 in the scaffold, which has been implicated as a factor that induces Th1 when coexpressed with IL-33 and signifies IL-33 alone is not sufficient for inducing Th1 responses.³⁴ Luminex analysis detected significant populations of chemokines 7 days after implant that could be sources of monocyte recruitment into the scaffold. However, IL-33’s effects seemed limited to decreasing the concentration of two IFN-γ induced chemokines CXCL9 and CXCL10.

Scaffold delivery of IL-33 induced a Th2 cytokine response that is consistent with previous literature.^25,28 Gene expression analysis of localized IL-33 scaffold delivery confirms the induction of a type 2 local immune response within the localized adipose scaffold environment. Day 3 RNA expression revealed close to 100-fold inductions of IL4, IL5, and IL13, all known to be induced by IL-33.²⁸ Curiously, while IL-4 and IL-5 remain upregulated within the IL-33 scaffold 7 days post implant, IL-13 was undetectable. While we did not stain the remaining CD4+ cells for the Th2 transcription factor GATA3 (a marker of the Th2 lineage) for flow analysis, the elevated presence of Th2 cytokines IL-4 and IL-5 in the IL-33 scaffold suggests that a significant portion of the remaining 30% of CD4+ T cells would be polarized towards the Th2 lineage. While there have been some reports of Th2-Treg plasticity, it does not appear that there is any loss of Tregs due to IL-4/IL-5 upregulation by IL-33 as the percentage of Foxp3+ CD4 cells increased in both blank scaffold implantation and allogeneic islet transplantation and no decrease in Foxp3 expression was detected.³⁵ While Th2 cells are not inherently tolerogenic, several studies have suggested that shifting the T cell polarization from Th1 to Th2 can prolong allogeneic islet engraftment.^36,37 However, alloreactive Th2 cells been shown to be sufficient to be able to mediate islet graft rejection.^38,39 While mechanisms exist for a Th2 mediated rejection of allografts, the prevention of a Th1 inflammatory response is still ultimately an important goal to extend allograft survival.

The primary cell population expanded by local IL-33 delivery was CD11b+ F4/80 Int Siglec F+ Side Scatter Hi eosinophils. Systemic IL-33 delivery has been shown to promote eosinophilia specifically due to the induction of Th2 cytokines.⁴⁰ Eosinophils have been shown to be an important cell lineage required to maintain immune homeostasis within adipose tissue through the release of the same type 2 cytokines that are responsible for allergy.²⁸ However, high local populations of eosinophils have been linked to transplant rejection in cardiac allograft models, typically induced by cytokines produced by Th2 cells like IL-5.^41,42 While Th1 cells are typically the most strongly associated helper subset associated with graft rejection, Th2 cells have also been shown to be sufficient to cause allo-immune rejection in the absence of a Th1 response, with eosinophils being proposed as a likely effector cell type that can cause graft damage through release of reactive oxidative species and granule cationic proteins.^42,43 While additional mechanistic studies are necessary to demonstrate that eosinophils are directly involved in the graft dysfunction observed, heavy eosinophilic infiltrate in the allograft is typically associated with negative transplant outcomes. While we do not observe a large population increase in the Th2 subset of graft-resident CD4+ T cells, local IL-33 delivery appears to temporarily create a cytokine microenvironment simulating a local increase in Th2 cells, creating activating conditions for eosinophils with and without the presence of an alloantigen stimulus. This highlights a feature of our scaffold delivery system: temporary delivery of factors that can avoid long term unregulated release of cytokines that can create permeant local imbalances in immunity. Additionally, even without the addition of IL-33, there is a dominant population of eosinophils that make up close to half of the observed live CD45+ cells in the BSA scaffold. This underscores the importance of local immune context in assessing the compatibility of a singular cytokine for use as a tolerogenic strategy. The enhanced presence of eosinophils in the control scaffold does not appear to have negative effects on engraftment and survival of transplanted islets. This may be an indication that activation of eosinophils by IL-33 is required for any engraftment disruptive characteristics to be manifested. The exposure of islets to high local concentrations of IL-33 may cause the delay in engraftment. However, no published literature indicating IL-33 has a direct effect on pancreatic islets was found, nor is there evidence that islets express the IL-33 receptor ST2. A single publication did show that IL-33 was able to induce ER stress in the mouse islet cell line MIN6N8, yet no further studies were performed to address the effects of IL-33 on endogenous mouse islets in vivo or in vitro.⁴⁴ Moreover, in observing the in vitro release kinetics of IL-33, over 90% of protein in the scaffold is expected to be leached out within the first day, whereas hyperglycemia appears to occur in IL-33 scaffold recipients starting at day 3 and ending around day 8.

We also looked at a number of other innate immune cell lineages that have been shown in the literature to be targets for modulation by IL-33, including ILC2s, adipose resident macrophages, and myeloid-derived suppressor cells (MDSCs). While only a few publications have focused on potential roles of ILC2s in transplant biology, its role in inducing a Th2 response has been well established in allergy and adipose homeostasis models.^27,45,46 As a cell type identified by expression of ST2, it is also available to respond directly to IL-33. We saw significant expansion of an ILC2 population that, while only comprising a small proportion of the total immune cell population, appears to contribute heavily to the local Th2 cytokine response induced by IL-33.^17,45 Adipose tissue-associated ILC-2s have been reported to produce high levels of type 2 cytokines like IL-5 and IL13, as well as the pro-inflammatory cytokine IL-6, which we observed to be one of the few Th1-associated cytokines upregulated in the graft.⁴⁷ This result strongly suggests that ILC2s may be one of the primary targets of localized IL-33 release into adipose tissue. Adipose resident macrophages, characterized as CD11b+ F4/80 Hi are important regulators of local inflammation. In lean mice, adipose tissue is enriched for alternatively activated CD301+ M2 macrophages over inflammatory CD11c+ M1 macrophages.^48,49 In contrast to published literature, scaffold delivery of IL-33 appeared to have limited effects on the macrophage population in the scaffold. While the control scaffold environment is enriched for F4/80 hi macrophages possessing a classically inflammatory CD11c+ phenotype, IL-33 does not significantly alter the number of these cells in the scaffold. A number of studies using IL-33 to promote transplant tolerance have reported IL-33 mediated immune suppression is mediated by a MDSCs population.^21,22 This heterogeneous population of cells is often characterized as CD11b+ Gr-1 intermediate but most accurately is defined by its namesake suppressive capacity. In the gene expression analysis of total RNA, we saw a similar rise in both Arg1 and Nos2, which are both upregulated in MDSCs.²¹ However, while flow cytometry analysis of the scaffold identified a population of CD11b+ F4/80 Int Ly6C Int cells, most of this population also co-expressed Siglec F and was identified as eosinophils rather than MDSCs. It does not appear as though MDSCs are prevalent in the control scaffold either and thus this lack of IL-33-induced MDSCs may reflect that protein is being delivered to an area where the cell of interest are not located or recruited.

5. Conclusions

In summary, we used a novel biomaterial platform to locally deliver a pleiotropic cytokine IL-33 into a unique adipose tissue transplant environment to try and reconcile its studied effects in an allotransplant setting. We altered previous designs of the scaffold to include a carrier protein to maximize IL-33 loading. Due to IL-33’s importance in adipose tissue immunity, we identified a number of immune cell lineages that were able to respond robustly to the delivered IL-33. IL-33 robustly increased the local Foxp3+ population with and without the presence of allogeneic tissue, thus confirming its ability to expand an important immunosuppressive phenotype. IL-33 also is able to decrease local CD8+ T cell proliferation in the allograft model. We then saw a significant increase in the allograft survival of islets, suggesting that there are long-term benefits to localized IL-33 delivery. However, we also found that IL-33 delayed islet engraftment in both allogeneic and syngeneic models, which may be explained by the induction of a type-2 cytokine response and concurrent expansion of eosinophils and ILC2s within the transplant environment. We suggest that IL-33 may be a compelling factor to deliver locally in combination with other factors that might be able to synergize with its ability to expand locally tolerogenic phenotypes such as Tregs while controlling its engraftment-delaying characteristics.

Highlights.

• PLG scaffolds releasing IL-33 are transplanted into murine epididymal fat pads

• IL-33 expands Tregs and decreases CD8+ T cells in presence of allogeneic islets

• Transplanted islet allografts show extended survival but delayed engraftment

• Local type 2 cytokine response observed with expansion of eosinophils and ILC2s