Islet-expressed TLR2 and TLR4 sense injury and mediate early graft failure after transplantation

Abstract

While islet transplantation is an effective treatment for Type 1 diabetes, primary engraftment failure contributes to suboptimal outcomes. We tested the hypothesis that islet isolation and transplantation activate innate immunity through Toll-like receptors (TLR) expressed on islets. Murine islets constitutively express TLR2 and TLR4, and TLR activation with peptidoglycan or lipopolysaccharide upregulates islet production of cytokines and chemokines. Following transplantation into streptozotocin-induced diabetic, syngeneic mice, islets exposed to LPS or PGN had primary graft failure with intra- and peri-islet mononuclear cell inflammation. Use of knockout mice showed that recipient CD8 T cells caused engraftment failure and did so in the absence of islet-derived DCs. To mimic physiological islet injury islets were transplanted with exocrine debris. Transplanation of TLR2/4^−/− islets reduced proinflammatory cytokine production and improved islet survival. Stressed islets released the alarmin HMGB1 and rHMGB1 induced NFkB activation and the effect was prevented in the absence of both islet TLR2 and TLR4. rHMGB1 pretreatment also prevented primary engraftment through a TLR2/4 dependent pathway. Our results show that islet graft failure can be initiated by TLR2 and TLR4 signaling and suggest that HMGB1 is one likely early mediator. Subsequent downstream signaling results in intra-islet inflammation followed by T cell-mediated graft destruction.

Introduction

Islet transplantation holds promise as a definitive therapy for Type 1 diabetes, but the long-term results have been disappointing, as progressive loss of graft function is observed in the majority of patients [1]. The islet mass is already marginal shortly after transplantation and thus susceptible to become insufficient when subsequently exposed to negative local influences. Recent estimates indicate that less than 30% of islets stably engraft, a result that explains the requirement for infusing large numbers of islets and for repeat islet infusions to maintain insulin-free euglycemia [2]. Mechanisms underlying early islet loss following transplantation remain poorly defined but apoptotic cell islet cell death associated with peri- and intra-islet graft inflammation have been described [3, 4].

Toll-like receptors are a family of pattern recognition receptors that bind to pathogen-associated molecular patterns (PAMPs) or to endogenous ligands released by damaged cells (damage associated molecular patterns, DAMPs). Among the latter group are heat shock proteins (HSPs), high mobility group box protein 1 (HMGB1), heparan sulfate, hyaluronan fragments, and fibronectin [5]. Regardless of the source of the specific ligand, TLR-transmitted signals activate innate immunity by inducing chemokine and cytokine release and through upregulating costimulatory molecule expression, among a multitude of other effects [6]. Recent studies revealed the importance of islet expressed TLRs, particularly TLR2 and TLR4, participating in the pathogenesis of autoimmune diabetes and allogeneic islet transplant rejection [7-9]. Whether TLR transmitted signals in the islets impact early islet engraftment has not been studied.

Our group, among others, showed that following physical manipulation, prolonged cell culture, ischemia/reperfusion injury, or virus-mediated gene transduction, islets can produce cytokines and chemokines in patterns reminiscent of those induced by TLR stimulation [10-15]. Upon transplantation, such manipulations amplify peri-islet inflammation and result in impaired islet graft function, further supporting the concept that early islet injury is in part mediated through TLR signals.

To define the mechanisms of early graft dysfunction, we studied the impact of TLR stimulation on graft survival following transplantation. Our data provide the first direct evidence that islet-expressed TLR2 and TLR4 are relevant mediators of the posttransplant inflammation associated with early graft dysfunction. These effects require recipient T cells, occur in the absence of islet DCs, and are fully reproduced by stimulation with HMGB1, an endogenous TLR2/4 ligand that is released by pancreatic tissue after sterile injury. In addition to providing insight into mechanisms underlying early graft loss, our findings indicate that TLR2 and TLR4 are potential targets for novel therapies aimed at preserving islet mass.

Results

TLR2/4 stimulation induces chemokine and cytokine release without altering in vitro viability or function

Using RT-PCR we found that RNA from a pancreatic β cell line and from purified C57BL/6 islets expressed message for TLR2 and TLR4 (Fig 1A). To test whether the expressed TLRs were functional, we cultured the islets overnight with or without the specific TLR2 ligand peptidoglycan (PGN) and the TLR4 ligand lipopolysaccharide (LPS), and measured chemokine and cytokine gene expression as a response (Fig. 1B). PGN stimulation induced significant increases in islet production of CCL2/MCP-1, TNFα and IL-6 RNA. LPS stimulation similarly increased expression of these genes, and also upregulated CXCL10/IP-10 mRNA (Fig. 1B).

Figure 1. The effect of islet-associated TLR stimulation on viability and apoptosis.

(A) RT-PCR analysis of TLR2 and TLR4 mRNA of isolated islets, β cell line (β-TC3), and splenocytes (amplified with 30 cycles). (B, C) Islets were cultured overnight without (Unstim) or with the TLR2 specific ligand PGN (10 μg/ml) or the TLR4 ligand LPS (100 ng/ml) (n=3 per group). mRNA expression of (B) cytokines and chemokines and (C) key components for glucose-inducible insulin release were assessed by qRT-PCR (n=3 per group; *p<0.05 compared with unstimulated controls). Positive control consisted of 0.5 mM streptozotocin-treated islets. (D) Islets were cultured in RPMI-1640 medium overnight in the presence/absence of LPS or PGN and then the glucose concentration of the culture media was increased from 5 to 20 mmol/. Insulin concentrations in the culture supernatants were measured by ELISA after 1 hr of high glucose stimulation (n=3 per group; *p<0.05 compared with basal condition). (E) Islet cells (unstimulated or stimulated with LPS or PGN) were lysed and apoptosis was assessed by measuring caspase 3 activity (left panel) or using the colorimetric ApoPercentage assay (right panel) (n=2 per group and assay). Data are mean + SEM; p-values were determined by Student's t-test.

To assess whether engagement of TLR2/4 directly affects islet function, we evaluated GLUT2 and glucokinase RNA, and determined glucose-induced insulin secretion following stimulation with LPS or PGN (Fig. 1C-D). Despite the above noted alterations in chemokine gene expression, we found that TLR stimulation had no acute significant effect on any of these measurements. We tested whether the LPS or PGN affected islet in vitro viability, and found neither a significant increase in caspase 3 activity nor in the percentage of apoptotic cells compared with untreated controls (Fig. 1E).

TLR stimulate intragraft inflammation and apoptosis and prevent islet engraftment

To determine the impact of TLR stimulation on islet engraftment in the absence of alloimmunity, we transplanted a marginal mass of 250 islets of untreated or TLR ligand-stimulated C57BL/6 islets into syngeneic diabetic mice (Fig. 2A). Transplantation of unstimulated WT islets rapidly reversed diabetes, whereas transplantation of islets pre-treated with either LPS or PGN prevented the restoration of euglycemia. Transplantation of TLR2^−/− or TLR4^−/− islets reversed diabetes despite treatment with their specific ligand, demonstrating the specificity of the TLR-mediated effects.

Figure 2. Islet isograft function, intragraft inflammation and apoptosis after TLR activation.

(A) Wild type (WT), TLR2^−/− or TLR4^−/− islets were cultured overnight without (Unstim) or with PGN (10 μg/ml) or LPS (100 ng/ml). For marginal mass syngeneic transplantation, 250 handpicked cultured islets were transplanted beneath the renal capsule of STZ-induced syngeneic recipients. Tail-vein glucose was measured daily. Glucose reduction day 0 vs. day 7 within each group for unstimulated WT p=0.001; LPS WT p=NS; PGN WT p=NS; PGN TLR2^−/− p=0.003, and LPS TLR4^−/− p=0.03 (n=3-7 each, NS: not significant). (B) Islets were cultured overnight without (Unstim) or with the TLR2 specific ligand PGN (10 μg/ml) and TLR4 ligand LPS (100 ng/ml) and then transplanted. Grafts were harvested on day 2 post-transplant (B) Expression of cytokines, chemokines and cell surface markers analyzed by qRT-PCR (n=3 each). (C) Islet cross-sections were examined for TUNEL positivity. Representative images showing TUNEL+ cells (left). The average number of TUNEL+ cells in 4-6 different cross-sections from each group and animal (right). (D) Caspase 3 mRNA analyzed by qRT-PCR (n=3 each). (*p<0.05 vs. unstimulated islets). Data are mean + SEM; p-values were determined by Student's t-test.

To assess mechanisms underlying early graft loss, intragraft inflammation was characterized by qRT-PCR on day 2 after transplantation (Fig. 2B). Although the effects of the two TLR ligands were distinct, pre-culture with LPS or PGN resulted in higher in vivo gene expression of CCL2/MCP-1, CCL3/MIP-1α, TNFα, IL-6, and/or IL-1β when compared with unstimulated islets. Higher expression of CD68 (monocyte/macrophage marker) and CD3 (T cell marker) mRNA in LPS- or PGN-stimulated graft tissue was also noted on day 2 compared with controls. Differences in these gene expression profiles were not observed on day 7 posttransplant, suggesting that TLR-induced inflammation is transient (data not shown).

The extent of intra-islet apoptosis was measured using TUNEL staining (Fig. 2C) and caspase 3 mRNA expression (Fig. 2D). Pretreatment with either LPS or PGN resulted in marked and significant increases in the percentage of apoptotic cells on day 2. Because the in vitro studies (Fig. 1) revealed no direct effect of LPS or PGN on islet viability, these in vivo findings suggest that TLR-induced islets produce chemokines and cytokines, leading to inflammation, which secondarily resulted in early islet apoptosis and dysfunction.

In clinical transplantation it is technically inevitable that exocrine and ductal cells will be present as residual debris in isolated islets and we previously demonstrated that these constituents stimulated islet chemokine expression [12]. We created a physiological model of islet injury by transplanting islet preparations with 50% purity (by adding exocrine debris). Of note, our standard islet purity after isolation is >90%. We observed that WT islets of 50% purity did not restore euglycemia whereas transplantation of TLR2/4^−/− islets cured diabetes despite the presence of exocrine debris (Fig. 3A). WT islets of 50% purity expressed more intragraft proinflammatory cytokines, macrophages and T cells compared with TLR2/4^−/− islets (Fig. 3B), showing that debris activated islet TLR2/4, and exaggerated the inflammatory response synergistically. By day 7 posttransplant, the inflammatory response had subsided.

Figure 3. Islet isograft function, intragraft inflammation and apoptosis in the absence of TLR2/4.

(A) 250 WT or TLR2/4^−/− islets were co-transplanted with WT exocrine tissue into diabetic syngeneic recipients and tail-vein glucose was measured daily. Glucose reduction day 0 vs. day 7 within each group for WT p=NS; TLR2/4^−/− p=0.001 (n=4 each, NS: not significant). (B) Day 2 and day 7 intragraft expression of cytokines, chemokines and cell surface markers analyzed by qRT-PCR (n=3-4 each). *p<0.05 within each time point. Data are mean +SEM; p-values were determined by Student's t-test.

We and others have recently shown that purified islets deficient in TLR2, TLR4, or MyD88 were rejected at the same tempo as WT controls when transplanted into untreated allogeneic recipients [16, 17]. We also found increased endogenous TLR ligands in allografts, including HMGB1 [16]. Thus, we determined whether TLR2/4^−/− islets allografts resulted in improved glucose reduction and lower intragraft inflammation. A marginal mass of untreated allogeneic TLR2/4^−/− islets produced only a modestly better glucose reduction in contrast to WT islets (Fig. 4A) but the absence of TLR2/4 signaling was linked with lower levels of TNF-α, IP-10, and IL-1β, and decreased macrophage and T cell recruitment (Fig. 4B). These experiments support a role for TLR2/4 in sensing islet injury. It is currently unknown whether the reduced inflammatory state affects allograft survival in the context of subtherapeutic immunosuppression.

Figure 4. Marginal mass islet allograft function and intragraft inflammation.

(A) Wild type B6 (WT) and TLR2/4^−/− B6 islets were cultured overnight. For marginal mass allogeneic transplantation, 250 handpicked islets were transplanted beneath the renal capsule of STZ-induced BALB/c recipients and tail-vein glucose was measured daily. Isografts (B6->B6) are shown as controls. Glucose reduction day 0 vs. day 7 within each group for WT B6->BALB/c p=NS; TLR2/4^−/− B6->BALB/c p=0.03; B6->B6 p=0.001 (n=3-7 each, NS: not significant). (B) Day 7 intragraft expression of cytokines, chemokines and cell surface markers analyzed by qRT-PCR (n=3 each). Data are mean + SEM; p-values were determined by Student's t-test.

Recipient T cells prevent engraftment of TLR-stimulated islets

Since early islet dysfunction is associated with mononuclear cell chemo-attractants and mononuclear cell infiltrates, we tested whether after TLR stimulation T cells are requisite pathogenic mediators of impaired islet engraftment. Syngeneic transplants were placed into T cell-deficient nude mice. In striking contrast to the observed effects of TLR stimulation on engraftment in WT recipients, LPS- or PGN-stimulated islets engrafted in all nude recipients, rapidly normalizing serum glucose levels (Fig. 5A).

Figure 5. Islet isograft function in wild type mice, nude mice, CD4- or CD8-deficient mice.

(A) WT islets were treated and transplanted as described in Fig 3. The diabetic recipients were T-cell deficient (nude) mice or their genetically matched WT littermates controls; LPS->WT (p=NS); PGN->WT (p=NS); LPS->nude p=<0.001; PGN->nude p=0.003 (n=3-5 each). (B) Diabetic recipients were CD8^−/− or CD4^−/− mice. LPS->CD8^−/− p<0.0001; PGN->CD8^−/− p=0.0009; LPS->CD4^−/− p=NS; PGN->CD4^−/− p=NS. The p-values given represents glucose reduction day 0 vs. day 7 within each group (n=3-6 each, NS: not significant). (C) AUC was calculated for euglycemic recipients (glucose <200 mg/dl prior i.p. glucose infusion) on day 7. AUC was significantly higher for CD8^−/− recipients compared with nude recipients after LPS (p=0.026) and PGN (p=0.002) treatment (n=3-6 each). Data are mean + SEM; p-values were determined by Student's t-test.

To identify which T cell subset was responsible for preventing engraftment, additional transplants into CD4- or CD8-deficient recipients were performed. TLR-stimulated islets did not engraft in CD4^−/− mice (all animals remained hyperglycemic) indicating that CD8 T cells were sufficient to prevent engraftment. In contrast, TLR-stimulated islets normalized serum glucose values following transplantation into diabetic CD8^−/− recipients, albeit with slightly delayed kinetics (Fig. 5B). Both, TLR2 and TLR4 stimulated islets resulted in euglycemia when transplanted into CD8 deficient mice, but had higher AUC on day 7 compared with nude mice, indicating some effects of CD4+ T cells (Fig. 5C). Together, these results indicated that CD8 T cells were the dominant pathogenic mediators of primary graft dysfunction after TLR stimulation.

Engraftment failure is mediated by TLRs on parenchymal cells of the islets

To determine if TLR-expressing DCs within the islets were required for early graft dysfunction, DTR-CD11c^GFP mice, in which the diphtheria toxin (DT) receptor is exclusively expressed on murine DCs and all CD11c+ DCs express green fluorescent protein (GFP) were used [18]. As shown in Fig. 6A-C, when isolated islets were treated with DT fluorescent microscopy and flow cytometric analysis showed more than 99% reduction in the number of islet-derived CD11c+ cells. Nonetheless, CD11c-depleted islets still expressed TLR2 and TLR4 (Fig 6D). The non-DC TLRs were functional because treatment of DC-depleted islets with PGN or LPS still upregulated proinflammatory cytokines (Fig 6E) and prevented engraftment (Fig. 6F). In control experiments, DT treatment did not functionally impair the islets, because transplantation of unstimulated but DT-treated islets restored euglycemia with similar kinetics as untreated control islets (Fig. 6F). These results indicated that TLR expression on islet intra-islet CD11c+ cells, including DC, were not the principal mediators of inflammatory effects.

Figure 6. In vitro depletion of intra-islet dendritic cells (DC) and isograft function.

Isolated islets from CD11c-DTR transgenic mice were treated overnight with (DT⁺) or without (DT⁻) 30 ng/ml diphtheria toxin . (A) Treated islets were visualized under normal and inverted fluorescent microscopy. (B, C) Treated islets were dispersed into a single cell suspension and (B) analyzed by flow cytometry for CD11c (cells were gated for CD45+ cells) or (C) were sorted according to GFP expression. (D) qRT-PCR analysis of TLR2 and TLR4 mRNA of the treated islets (n=3). (E) qRT-PCR analysis of cytokine and chemokine mRNA expression of isolated CD11c-DTR islets with DT after overnight culture without (Unstim) or with the TLR2 specific ligand PGN (10 μg/ml) or the TLR4 ligand LPS (100 ng/ml) (n=3). (F) CD11c-DTR transgenic islets were LPS- or PGN-stimulated in the presence of DT, or were left unstimulated in the presence/absence of DT and transplanted into diabetic WT syngeneic recipients. Unstimulated->B6DTR DT⁻ p=0.004; Unstimulated->B6DTR DT⁺ p=0.002; LPS->B6DTR DT⁺ p=NS; PGN->B6DTR DT⁺ p=NS (n=3-4 each). The p-values given represent glucose reduction day 0 vs. day 7 within each group. NS: not significant. Data are mean + SEM; p-values were determined by Student's t-test.

Islets release endogenous HMGB1 that stimulates TLR2 and TLR4 and prevents engraftment

The data indicated that islet-expressed TLR2- or TLR4-transmitted signals prevented engraftment following transplantation. It remains unclear whether experimental protocols in which islets were stimulated with LPS and/or PGN have physiological relevance to transplantation of sterile islets. HMGB1 is released by pancreatic β-cells treated with IL-1, and can be found early in islets after intrahepatic transfusion [19, 20]. We and others have shown that HMGB1 can bind to and activate TLR2 and/or TLR4 in vitro [21-24], raising the possibility that HMGB1 could act as a sterile DAMP that contributes to engraftment failure following transplantation into the renal subcapsular space.

When islets were exposed to 3% O2 for 24 h, a hypoxic state that closely mimics the microenvironment of subcapsular transplanted islets [25], we found that morphologically intact islets released significant amounts of HMGB1 into culture supernatants (Fig. 7A). Consistent with this data HMGB1 was up-regulated in recently transplanted and untreated syngeneic islets (Fig. 7B). In addition, exocrine cells excreted HMGB1 (8.1 ± 1.2 ng/mg protein) when cultured for 24 hours. To determine if HMGB1 signals through TLRs, WT islets were stimulated with rHMGB1 (5 μg/ml) and NF-κB nuclear translocation was assessed as a measure of TLR engagement [26]. As showwn in Fig. 7C, stimulation with rHMGB1 induced NF-κB translocation. LPS stimulation (100 ng/ml) and PGN stimulation (10 μg/ml) also induced translocation of NF-κB, and the effects were prevented in the absence of their specific TLR. rHMGB1-induced only modelstly lower NF-κB activation in either TLR2^−/− or TLR4^−/− islets. In contrast, islets deficient in both TLR2 and TLR4 had a greater than 60% reduction in NF-κB activation (Fig. 7C), indicating that HMGB1 signaled via both receptors.

Figure 7. Endogenous HMGB1 release and islet graft function after rHMGB1 stimulation.

(A) HMGB1 release by islets exposed to 3% hypoxia was measured by ELISA in the supernatant (n=4-5) and (B) HMGB1 mRNA levels post transplantation were measured by qRT-PCR in isografts (n=3). *p<0.05 compared with day 1. (C) WT, TLR2^−/−, TLR4^−/−, TLR2/4^−/− islets were stimulated with LPS, PGN, or rHMGB1 and NF-κB activation was determined by staining for p65 and DAPI. P65 translocation was visualized by confocal microscopy and the percentage of cells with translocated p65 per islet cross-section determined (n=4). (D) WT and TLR2/4^−/− islets were cultured with rHMGB1 and transplanted into diabetic syngeneic recipients and tail-vein glucose was measured daily. 1 μg/ml HMGB1 WT p<0.001; 5 μg/ml HMGB1 WT p=NS; 5μg/ml HMGB1 TLR2/4^−/− p<0.001 (n=3 each). The p-values represent glucose reduction day 0 vs. day 7 within each group. Data are mean + SEM; p-values were determined by Student's t-test.

The effects of HMGB1 on islet graft function were assessed by transplanting HMGB1-stimulated WT or TLR2/4^−/− islets into syngeneic recipients. While HMGB1 stimulation prevented engraftment of WT islets, TLR2/4^−/− islets engrafted in all animals, normalizing serum glucose levels with similar kinetics to untreated WT islets (Fig. 7D).

DISCUSSION

Our results delineate several new insights into the pathogenesis of early islet graft failure, including the notable result that TLR2 and TLR4 are key participants in this process. We demonstrated that stimulation via either TLR2 or TLR4 initiated a proinflammatory milieu, likely via chemokines and cytokine release at the graft site, associated with graft apoptosis and early graft failure (Fig. 2), but did not directly affect islet viability or function in vitro (Fig. 1). In experiments mimicking physiological islet injury by adding exocrine debris (Fig. 3) or by alloimmune response (Fig.4). TLR2/4^−/− islets reduced proinflammatory cytokine production and/or improved islet survival. Recipient T cells and principally CD8 T cells mediated the graft destruction, because TLR-stimulated islets restored euglycemia in CD8^−/− mice (Fig. 5). While the specific T cell targets are not known, our data demonstrate that the CD8 T cells did not require DCs (Fig. 6). The data newly revealed that HMGB1, a highly conserved chromosomal protein, could be released from islets in response to hypoxic stress or transplantation and that through signaling via TLR2 and TLR4 this endogenous DAMP prevented primary engraftment (Fig. 7).

These studies extend our previous report in mice [10] and of others in humans [13] that isolated pancreatic islets produce chemokines following short-term culture, and high pre-transplant CCL2 concentrations correlated with poor islet graft function. Our previous data showed the damage to the islets could not be completely accounted for by the interaction of CCL2 with its receptor CCR2, suggesting a role for other cytokines or chemokines [10]. Our current findings explain this previous work by implicating islet-expressed TLRs as the mechanistic link between pre- and peritransplant events and increased expression of pro-inflammatory genes, attracting macrophages and T cells.

While we demonstrated that early islet graft loss occurred in CD4^−/− but not in CD8^−/− recipients (Fig.5), indicating a pathogenic role for CD8 T cells, the specific mechanisms underlying this observation remain to be elucidated. We speculate that the local inflammation associated with the transplant procedure, compounded by the absence of CD4 Treg in CD4^−/− animals facilitates activation of autoreactive CD8 T cells. The primed CD8 cells are attracted to the inflamed graft, where they elicit effector functions that mediate injury and amplify the local inflammation. This hypothesis is supported by previous work from our group in which we demonstrated that cotransfer of Treg protected inflamed islet grafts from innate immune injury and led to enhanced islet function.

We extended previous studies on the role of TLRs in transplant models by studying potential ligands. HMGB1 is a chromatin-binding protein that regulates transcription and chromosome architecture. Its release from the cell nucleus into the extracellular environment can occur passively as cells undergo necrotic death, or actively in response to stressors, when it functions as a proinflammatory danger signal in a TLR2 and/or TLR4 dependent manner [21, 22, 24, 27]. HMGB1 is an attractive DAMP candidate as a significant proportion of islets is necrotic or undergoes apoptosis at the end of the isolation process [28, 29]. A recent paper confirmed that islets contain abundant HMGB1 [20]. These authors found that recipients receiving anti-HMGB1 treatment after intraportal islets transfusion had improved islet function. In contrast to TLR4, mice lacking TLR2 and receptor for advanced glycation end products (RAGE) had improved islet function, suggesting that locally produced HMGB1 targets intahepatic immune cells, e.g. DC, expressing these receptors [20]. It is important to note that in contrast to our study Matsuoka et al. did not investigate the role of islets in sensing alarmins. In addition, the difference in HMGB1-mediated effects on TLR4 might be due to the different model (transplant site) and cell type (islet cells versus bone marrow derived immune cells). While our and Matsuoka et al. [20] observations support the hypothesis that HMGB1 is one relevant candidate for TLR-mediated islet injury, other endogenous ligands released from dead cells such as hyaluran, heat shock proteins, uric acid, fibronectin or DNA-RNA protein complexes [5, 6]. With the expression of a functional LPS receptor, even a very low amount of endotoxin might activate islet-associated TLR4 and may be clinically significant, as suggested by data that endotoxin contaminated enzymes used for islet isolation were detrimental to islet function [30].

In the clinical context, TLR antagonists are in clinical development and blockade of their common signaling pathways is more likely to be successful than targeting individual ligands or receptors which often serve redundant functions. Together with previous studies demonstrating the beneficial effects of TLR inhibition on IR injury, acute rejection, and tolerance, our study sets the stage for future work aimed at inhibiting TLR activation in a clinical setting [6]. There is extensive evidence that the innate immune system interacts with the adaptive immune system and targeting these receptors may have value both for improving early engraftment and for long-term maintenance of graft function and survival.

Materials and Methods

Animals and cell lines

C57BL/6 (H-2^b), BALB/c (H-2^d), athymic male mice (CBy.Cg-Foxn1^nu, nu/nu), their genetically matched WT male littermates, CD8^−/− (B6.129S2-Cd8a^tm1Mak), CD4^−/− (B6.129S2-Cd4^tm1Mak), TLR2^−/− (TLR4^−/−B6, H-2^b), B6.FVB-Tg(Itgax-DTR/EGFP) 57Lan/J (DTR-CD11c^GFP, H-2^b) were purchased from The Jackson Laboratory (Bar Harbor, Maine). TLR4^−/− (TLR4^−/−B6, H-2^b) were provided by Dr. Maria Abreu [31]. TLR2 and TLR4 double knockout (TLR2/4^−/−) were generated by crossing the individual knockouts. Mice were used at 8 to 12 weeks of age, housed under specific-pathogen-free conditions, and treated in strict compliance with regulations established by the Institutional Animal Care and Use Committee. The β-cell line (β TC3) was provided by Dr. Teresa P. DiLorenzo.

Reagents

Collagenase P was purchased from Roche Diagnostics (Mannheim, Germany). Streptozotocin (Sigma, St. Louis, MO). The following reagents were used: Anti-CD3 mAb (BD Pharmingen, San Jose, CA, USA), anti-CD68 mAb (Serotec, Raleigh, NC), anti-IgG (Jackson Immunoresearch, West Grove, PA), anti-IFN-γ and biotinylated anti-IFN-γ mAb (BD Pharmingen, San Jose, CA), alkaline phosphatase-conjugated anti-biotin antibody (Vector Laboratories, Burlingame, CA), anti-human HMGB-1 monoclonal antibody (capture antibody, Upstate Biotechnology, Lake Placid, NY)), anti-HMGB1 antibody (detection antibody, R&D Systems, Minneapolis, MN), EZ-Link Sulfo-NHS-LC-biotin reagent (Pierce Biotechnology, Rockford, IL), streptavidin-alkaline phosphatase conjugate (Amersham Biosciences, Freiburg, Germany), 4-nitrophenyl phosphate (Serva Electrophoresis, Heidelberg, Germany), p65 (clone C22B4, Cell Signaling Technology, Danvers, MA), Cy5 (Jackson Immunoresearch, West Grove, PA), purified LPS (E. coli 0111:B4), peptidoglycan (InvivoGene, San Diego, CA), diphtheria toxin (List Biological Laboratories, Campbell, CA), polymyxin B (Fluka Chemie GmbH, Buchs, Switzerland), rHMGB1 (Sigma, St. Louis, MO).

Diabetic model, islet isolation, and transplantation

Islet recipients were rendered diabetic by a single intraperitoneal injection of 180 mg/kg streptozotocin and considered diabetic when the tail vein blood glucose concentration was more than 300 mg/dl for 2 consecutive days. Islet isolation and transplantation were previously described in detail [32]. For marginal mass syngeneic or allogeneic transplantation, 250 handpicked islets were transplanted, with or without prior stimulation, in serum-free medium beneath the renal capsule, and tail-vein glucose was measured daily [10]. To mimic physiological injury, 250 handpicked islets were co-transplanted with exocrine debris at a 1:1 ratio. Intraperitoneal glucose tolerance testing was performed on day 7 as described [33], and for groups with a post-transplant glucose concentration of less than 250 mg/dl the area under the curve (AUC) was calculated.

TLR stimulation and simulation of hypoxia

Islets (500 islets/ml) were stimulated at 37°C for 5 hours in 1 ml of fresh serum-free medium containing 0.5 % fetal calf serum in the presence or absence of purified lipopolysaccharide (LPS, 100 ng/ml) and peptidoglycan (PGN, 10 μg/ml). The ultra-pure LPS used only activates the TLR4 pathway [34]. Except for LPS treated samples, polymyxin B (10 μg/ml) was added to prevent the possible effect of contaminating endotoxin. rHMGB1 was endotoxin tested and contained <0.01 EU/μg. Hypoxic conditions were simulated using a hypoxia chamber. Cells were seeded in 6-well plates and placed into the chamber for 24 hrs. The chamber was flushed with hypoxic air (3% O₂, 5% CO₂, 92% air) twice (time point 0 and 1 hour) to establish stable hypoxic conditions.

HMGB1 ELISA

HMGB-1 in cell supernatants was measured by ELISA using monoclonal antibodies as previously described [35]. In brief, 96-well plates (Nunc, Roskilde, Denmark) were coated overnight at 4 °C with an anti-human HMGB-1 monoclonal antibody (1.5 μg/ml). After blocking (1% BSA), samples were added, incubated, washed and biotinylated anti-HMGB1 antibody (0.75 μg/ml) was added. Visualization was done by using streptavidin-alkaline phosphatase conjugate and 4-nitrophenyl phosphate on a microplate spectrophotometer at 405 nm. Serial dilutions of rHMGB1 (0.41 to 300 ng/ml) were used as internal standards and all samples were run in duplicates.

Depletion of intra-islet DCs

Islets were isolated from DTR-CD11c^GFP mice, plated in 12-well plates and cultured in 1.5 ml of RPMI containing 0.5% FCS and 1% penicillin/streptomycin with diphtheria toxin (30 ng/ml) at 37° C in a 5% CO₂ incubator for 24 hours. Medium was then aspirated, the islets were washed with PBS, and the presence of GFP⁺ cells was analyzed using a Leica DMRA2 fluorescence microscope. For flow cytometric analysis, single cell suspension of islets were stained for CD11b, CD11c, MHC class II, and CD45. For cell sorting, islets were dispersed using 0.5% Trypsin-EDTA (1 min) and GFP⁺ cells were sorted using FACS Vantage (Becton Dickinson). FACS data were analyzed using FlowJo software (Tree Star Inc, Ashland, OR).

Quantitative RT-PCR (qRT-PCR)

Total RNA was extracted from isolated islets or grafts with 1 ml of phenol/guanidine isothiocyanate containing TriZol solution (Life Technologies BRL, Grand Island, NY). For cDNA synthesis, total RNA was primed with oligo(dT) and PCR was performed on a LightCycler (Roche Applied Science, Indianapolis, IN) with the FastStart QuantiTect SYBR Green PCR kit (Qiagen, Valencia, CA) as described previously [10, 32].

Glucose-stimulated insulin secretion

Groups of 30 isolated islets or 3 × 10⁴ β-cells were cultured in complete RPMI 1640 medium (10% FBS, 100 U/ml penicillin, 0.1 mg/ml streptomycin, and 2 mM l-glutamine) that contained 1.1 mM glucose in 24-well tissue culture plates. After resting for 1 hour, additional glucose was added to a final concentration of 4.4 mM. Supernatants were analyzed for insulin content using a rodent-specific insulin ELISA kit (Crystal Chem, Chicago, IL).

Apoptosis assays

In vitro apoptosis was assessed using the fluorometric, immunosorbent enzyme assay for the specific, quantitative determination of caspase 3 activity (Roche Applied Science, Indianapolis, IN) and by using APOPercentage Dye (Biocolor, Irland, UK) as recommended by the manufacturer. In vivo apoptotic cell death in the islets grafts was evaluated on day 2 after transplantation using the terminal deoxynucleotidyl transferase enzyme for nick end labeling (TUNEL) method using the APOPTAG kit (Oncor, Gaithesburg, MD). Staining was performed as per manufacturer's instructions and apoptotic cells were quantified as the number of TUNEL positive cells per islet cross-section. Four to 6 different islet cross-sections per graft were analyzed.

P65 translocation assay

After 5 hours of stimulation, islets were fixed in 2% PFA, permeabilized with 0.1% Triton X-100, then stained with a monoclonal antibody specific for p65. Cy5-labeled secondary antibody was used for visualization. Imaging was done by confocal microscopy using DAPI as a nuclear counter stain [26]. A total of 4 islets per group and culture condition were analyzed. For each islet cross-section, which contains an average of 250 cells, p65 translocation and DAPI nuclear stained cells were counted.

Statistics

Results are expressed as mean ± SEM. Differences between groups were compared by Student's t-test. P-values <0.05 were considered statistically significant.