The location of sensing determines the pancreatic β-cell response to the viral mimetic dsRNA

Viral infection is an environmental trigger that has been suggested to initiate pancreatic β-cell damage, leading to the development of autoimmune diabetes. Viruses potently activate the immune system and can damage β cells by either directly infecting them or stimulating the production of secondary effector molecules (such as proinflammatory cytokines) during bystander activation. However, how and where β cells recognize viruses is unclear, and the antiviral responses that are initiated following virus recognition are incompletely understood. In this study, we show that the β-cell response to dsRNA, a viral replication intermediate known to activate antiviral responses, is determined by the cellular location of sensing (intracellular versus extracellular) and differs from the cellular response to cytokine treatment. Using biochemical and immunological methods, we show that β cells selectively respond to intracellular dsRNA by expressing type I interferons (IFNs) and inducing apoptosis, but that they do not respond to extracellular dsRNA. These responses differ from the activities of cytokines on β cells, which are mediated by inducible nitric oxide synthase expression and β-cell production of nitric oxide. These findings provide evidence that the antiviral activities of type I IFN production and apoptosis are elicited in β cells via the recognition of intracellular viral replication intermediates and that β cells lack the capacity to respond to extracellular viral intermediates known to activate innate immune responses.


INTRODUCTION
Type 1 diabetes (T1D) is characterized by the selective, autoimmune-mediated destruction of insulin-secreting β-cells found in the endocrine pancreas (1). Although genetic factors play a key role in determining the susceptibility towards the development of autoimmune diabetes (2,3), concordance rates of disease in monozygotic twins is low, estimated to be 39-65% (4)(5)(6)(7). These findings support environmental factors as contributors to T1D development and, because of their ability to activate the immune system, viral infections (and in particular, viruses belonging to the picornavirus family including coxsackievirus) are hypothesized to be one potential environmental trigger initiating autoimmune-mediated β-cell damage (8)(9)(10)(11). While a number of distinct viral classes are known to directly infect and lyse β-cells (9), the mechanisms leading to β-cell damage during viral infection and the mechanisms by which βcells and islets can recognize and respond to viral infection are incompletely understood.
Interleukin (IL)-1β is a proinflammatory cytokine that inhibits β-cell oxidative metabolism, inhibits insulin secretion, and induces DNA damage. If the magnitude of damage and time of exposure is sufficient, IL-1β can stimulate β-cell death (12). The destructive effects of IL-1β are dependent on β-cell expression of iNOS and production of micromolar levels of nitric oxide (13)(14)(15)(16). Resident macrophages are one potential source of IL-1 in islets. When activated, intraislet macrophages produce IL-1β to levels sufficient to stimulate β-cell expression of iNOS and nitric oxide-dependent inhibition of β-cell function (17)(18)(19). We have shown that intraislet macrophages can be activated in response to multiple pathogen associated molecular patterns (PAMPs) such as endotoxin (LPS; (19)), and viral structural components, including viral double stranded (ds)RNA or virus capsid protein (20,21).
The classically studied cellular response initiated to defend against viral infection is the expression of type I interferons (IFN-α and IFN-β). These key antiviral effectors are produced by cells in response to recognition of dsRNA that is produced intracellularly during virus replication or released extracellularly following the lysis of infected cells (22)(23)(24)(25). Multiple dsRNA sensors have been identified and they maintain distinct sub-cellular localizations and ligand specificity for dsRNA. Toll-like receptor 3 (TLR3) is a dsRNA sensor located in the endosome that can stimulate type I IFN expression in response to extracellular dsRNA (24,26-28), specifically the endocytosis of dsRNA released from neighboring cells rather than dsRNA accumulating within cells during infection (24,29). Melanoma differentiation antigen 5 (mda5) and retinoic acid inducible gene-I (RIG-I) are cytosolic helicases that induce type I IFN expression in response to dsRNA accumulating in the cytosol (30,31); however, mda5 preferentially recognizes long dsRNA (multiple kilobases in length, including polyI:C or RNA from Picornaviridae viruses such as coxsackievirus and encephalomyocarditis virus), whereas RIG-I recognizes shorter dsRNA (less than 1 kilobase) or uncapped ssRNA (32-35). The dsRNA-dependent protein kinase R (PKR) is another cytosolic dsRNA sensor that inhibits protein translation by phosphorylating eukaryotic initiation factor (eIF)2α (36,37).
Studying the mechanisms by which viral infection contributes to diabetes development is complicated by the diversity of RNA and DNA viruses from multiple viral families that have been implicated in disease development. Also, parsing the direct actions of the virus from the host cell antiviral responses adds to this complexity. Viruses such as enteroviruses, coxsackieviruses, and EMCV (a diabetogenic virus in mice) have been shown to inhibit host-cell translation (38)(39)(40), expression of type I IFNs (41,42), and cellular induction of apoptosis (43,44). In this report, the synthetic dsRNA mimetic, polyI:C, has been used to evaluate the effects of intracellular versus extracellular dsRNA recognition on the viability of insulinoma cells and primary rodent islet cells. Additionally, the actions of dsRNA were compared to the effects of cytokine treatment, as the bystander production of cytokines, such as IL-1β, have been implicated in the loss of functional β-cell mass following viral infection (20). We show that β-cells fail to respond to extracellular dsRNA, intracellular dsRNA recognition results in β-cell apoptosis, and the effects of dsRNA are distinct from the actions of IL-1β which stimulates nitric oxidedependent β-cell necrosis.

The response of β-cells to intracellular and extracellular polyI:C.
In response to polyI:C macrophages express a number of inflammatory genes including iNOS, IL-1, and COX-2; however, βcells fail to respond to extracellular polyI:C treatment (45). In previous studies, we have shown that macrophage expression of each of these inflammatory genes in response to polyI:C is controlled by nuclear factor (NF)-kB and at least one additional signaling cascade that is selective for the target gene: PKA and CREB for iNOS, JNK and p38 for COX-2 and 4 ERK for IL-1 (reviewed, (21)). Recently, we have identified CCR5 as the cell surface signaling receptor required for polyI:C-induced inflammatory gene expression by macrophages (46). β-cells fail to produce nitric oxide in response to polyI:C (45) and also do not express CCR5 (expression is restricted to cells of hematopoietic lineage, Fig. 1A, (47)). To determine if β-cells have the ability to respond to polyI:C, the expression of dsRNA sensors in β-cells was compared to macrophages. Like RAW 264.7 macrophages, INS832/13 β-cells express comparable levels of PKR, RIG-I, and mda5 mRNA but they not express the endosomal dsRNA sensor TLR3 (Fig. 1A). These findings suggest that β-cells have the ability to sense intracellular dsRNA but lack the ability to sense extracellular polyI:C. Neither extracellular nor intracellular polyI:C activate inflammatory signaling cascades such as NF-kB activation (IkB degradation) or the phosphorylation of JNK or p38 in INS832/13 cells. IL-1 is shown as a positive control for MAP kinases and NF-kB activation in INS832/13 ( Fig 1B). When delivered intracellularly (polyI:C transfected using lipofectamine 2000), polyI:C stimulates the mRNA accumulation of type 1 IFNs ( Fig 1C) and this correlates with basal expression of the cytosolic dsRNA sensors (RIG-I and mda5, Fig. 1A) that are known to activate type 1 IFN expression. Extracellular polyI:C fails to stimulate IFN-β mRNA accumulation and this correlates with the absence of detectable levels of the dsRNA sensor TLR3 (Fig. 1A), known to be activated by endocytosis of dsRNA. Also, the lack of inflammatory gene expression in response to extracellular polyI:C correlates with an absence of CCR5 expression in INS832/13 cells. To clarify whether the absence of response of β-cells to extracellular polyI:C was due to lack of extracellular or endosomal receptors such as CCR5 and TLR3 or due to lack of endocytosis, macrophages and β-cells were treated with extracellular rhodamine-labelled polyI:C, washed, and polyI:C uptake was determined by fluorescence spectroscopy. The uptake of polyI:C by INS832/13 cells is not statistically different than that observed for RAW264.7 macrophages (Fig. 1D). These data suggest that endocytosis can occur in β-cells but they lack the ability to sense extracellular dsRNA.

The effects of intracellular and extracellular polyI:C on β-cell expression of type I IFNs and ISGs.
dsRNA sensors are responsible for the induction of type 1 IFN, and type I IFNs go on to regulate the expression of hundreds of interferon-stimulated genes (ISGs) that mediate the antiviral responses of interferons (23). In a time-dependent manner, intracellular polyI:C stimulates the accumulation of type I IFNs (IFN-α and IFNβ) and the ISGs Mx2, RNase L, and viperin ( Fig. 2A-E). Extracellular polyI:C fails to enhance type 1 IFN or ISG expression by INS832/13 cells. Of the cytosolic dsRNA sensors that are detected basally (Fig. 1A), the expression of PKR is not modified by intracellular or extracellular polyI:C (Fig. 2F), while intracellular polyI:C increases mda5 and RIG-I mRNA accumulation in a time-dependent manner in INS832/13 cells (Fig. 2G, 2H). Although not detected under basal conditions, intracellular polyI:C also stimulates the time-dependent accumulation of TLR3 mRNA (Fig. 2I). Lipofectamine alone and extracellular polyI:C do not stimulate ISG or dsRNA sensor expression.
Since it is possible to stimulate β-cell expression of TLR3, we examined whether TLR3 expressing β-cells have the capacity to respond to extracellular polyI:C. Treatment of INS832/13 cells for 12 hours with IFN-β results in a > 40-fold increase in TLR3 mRNA accumulation; however, the addition of extracellular polyI:C to these cells fails to increase IFN-β mRNA accumulation (Fig. 3B). Additionally, IFN-β treatment fails to stimulate the accumulation of CCR5 mRNA (Fig. 3A) and extracellular polyI:C fails to stimulate the accumulation of iNOS mRNA following this IFN-β treatment (data not shown). These findings suggest that β-cells lack an ability to sense extracellular dsRNA either by TLR3 via endocytosis or extracellular recognition by CCR5.

Effects of intracellular and extracellular polyI:C on β-cell viability.
In addition to stimulating the production of type I IFNs, pathways leading to the induction of apoptosis are activated in cells following recognition of dsRNA (24,30,48). This response provides a means to prevent or limit the release and continued replication of viral progeny, as viral infection is enhanced in cells and animals lacking dsRNA sensors (37,49). Consistent with the ability of β-cells to sense intracellular polyI:C ( Fig. 1-3), in a timeand concentration-dependent manner, intracellular polyI:C decreases INS832/13 cell viability with maximal cell death observed following a 36 h incubation (Fig. 4A, 4B). Conversely, extracellular polyI:C at concentrations 10-fold higher than intracellular polyI:C levels does not modify INS832/13 cell viability (Fig. 4A, 4B). While β-cells produce nitric oxide and are sensitive to nitric oxidemediated damage following cytokine treatment (13)(14)(15)(16), nitric oxide does not contribute to INS832/13 cell death in response to intracellular polyI:C. (Fig 4C). Furthermore, the cellular recognition of dsRNA appears to be a far more potent inducer of cell death than exposure to cytokines, as approximately 70% of INS832/13 cells die following a 36 h treatment with intracellular polyI:C as compared to 18.5% of INS832/13 cells treated for 36 h with IL-1β (Compare Fig. 4A and 4D).

Intracellular polyI:C induces β-cell apoptosis.
We and others have shown that cytokines such as IL-1β stimulate β-cell necrosis in a nitric oxide-dependent manner (50,51), while the loss of β-cell viability in response to recognition of intracellular dsRNA is more consistent with apoptotic killing (24). In response to intracellular polyI:C, morphological changes include INS832/13 cell shrinkage and chromatin condensation (Sytox green staining, inset) much like the morphological changes observed following INS832/13 cell apoptosis following treatment with camptothecin (Fig 4). In contrast, extracellular polyI:C and lipofectamine alone do not modify the cellular morphology or stimulate Sytox nuclear accumulation. In response to IL-1β there are a low percentage of INS832/13 cells with nuclear Sytox accumulation and this DNA staining is prevented by the iNOS inhibitor NMMA. Overall, the morphology of nuclear Sytox staining indicates that intracellular polyI:C and camptothecin induce chromatin condensation with areas of punctate Sytox staining foci, while the response to IL-1β is morphologically distinct with Sytox staining encompasses the entire nucleus ( Fig 5B). The nuclear Sytox staining in response to intracellular polyI:C and camptothecin is consistent with the apoptotic killing of these cells, while nitric oxidemediated damage in response to cytokine treatment is consistent with DNA damage associated with cellular necrosis (Fig. 5).

Caspase activation in response to polyI:C.
Consistent with morphological changes shown in Fig 5, intracellular (but not extracellular) polyI:C stimulates a timedependent increase in caspase-3/7 activity that is first detected 3 h and maximal following a 12 h incubation (Fig. 6A). The pan-caspase inhibitor I (Z-VAD (OMe)-FMK, Fig. 6A) attenuates intracellular polyI:C-induced caspase 3/7 activity and β-cell death (Fig 6A  and 6C). IL-1β fails to stimulate caspase-3/7 activity (Fig. 6B) and the pan-caspase inhibitor does not modify the loss of β-cell viability in response to this cytokine (Fig. 6D). In agreement with nitric oxide-dependent necrosis (16,(50)(51)(52), the loss of INS832/13 cell viability in response to IL-1β is prevented by the iNOS inhibitor NMMA (Fig. 5, Fig. 6D and 6E). These findings provide evidence that intracellular polyI:C is far more toxic to β-cells than treatment with cytokines and that the type of cell death differs, where intracellular polyI:C induces apoptosis while cytokines induce nitric oxide-dependent necrosis of a limited number of β-cells.
Nitric oxide has been shown to attenuate apoptosis by inhibiting caspase activity (53)(54)(55) and by inhibiting activation of the DNA damage response (52,56). Consistent with our previous studies, exogenous addition of nitric oxide using the chemical donor DPTA/NO attenuates camptothecin-induced βdeath at 300 µM (52). Camptothecin-induced cell death is mediated by DNA damage and DDR signaling (52,56). Given that intracellular polyI:C stimulates β-cell death in a manner dependent on caspase activity, similar to camptothecin ( Fig. 5 & 6), we examined whether nitric oxide donors would also attenuate death in response to intracellular polyI:C. Surprisingly, DPTA/NO does not attenuate intracellular polyI:C induced apoptosis, rather it enhances intracellular polyI:C-induced INS832/13 cell death (Fig.  6F). These findings show that the mechanisms by which intracellular polyI:C induce β-cell apoptosis differ from the actions of camptothecin or DDR-directed apoptosis.

Dispersed rat islet cells produce type I IFNs and undergo apoptosis selectively in response to intracellular polyI:C.
Similar to INS832/13 cells, intracellular polyI:C induces the accumulation of type I IFN (Fig. 7A, 3B) and ISGs mRNA ( Fig. 7B-D) in rat islet cells. Extracellular polyI:C also stimulates the accumulation of Mx2, RNaseL, and viperin mRNA, albeit to lower levels than intracellular polyI:C, and this occurs without modifying type 1 IFN mRNA accumulation ( Fig. 7A-D). Pancreatic islets contain 10-20 resident islet macrophages that are capable producing type I IFNs, as well as other inflammatory genes, in response to extracellular polyI:C (18,19), suggesting that the induction of ISG mRNA accumulation in rat islets cells in response to extracellular polyI:C may reflect type 1 IFN production by tissue macrophages and type 1 IFN-mediated induction of ISG expression by islet cells. To explore this possibility, low temperature culture (7 days culture at 23 o C) was used to deplete resident macrophages found in islets, and this method removes over 90% of resident islet lymphoid cells (19,57). We have used this approach to show that activated intraislet macrophages produce IL-1β in the local environment of the islet to levels sufficient to induce β-cell expression of iNOS and nitric oxide-dependent damage to β-cells (17). In support of macrophage depletion, we show that extracellular polyI:C stimulates IL-1β mRNA accumulation only in islets containing macrophages ( Fig 8A). In response to extracellular polyI:C there is a greater than 90% decrease in the accumulation of Mx2 and viperin mRNA in macrophage depleted islets as compared to islets containing macrophages. Extracellular polyI:C does not increase RNaseL mRNA accumulation regardless of the presence of intraislet macrophages. These findings suggest that intraislet macrophages and islet cells are capable of producing type 1 IFN in response to intraislet polyI:C and this production correlates with ISG mRNA accumulation. While extracellular polyI:C can increase the mRNA of some ISGs (Mx2 and Viperin), it is dependent on intraislet macrophages and not islet endocrine cells.

Intracellular polyI:C induces islet cell death.
TUNEL staining, a marker of DNA damage that has been associated with induction of apoptosis (58,59), was used to examine the effects of polyI:C on the viability of rat islet cells. Intracellular polyI:C stimulates ~ 30% increase in insulin containing TUNEL positive cells, while lipofectamine and extracellular polyI:C (at concentrations 10-fold higher than intracellular polyI:C concentrations) do not increase TUNEL positivity in insulin containing cells (Fig. 9A and 9B). These findings indicate that intracellular dsRNA induces β-cell apoptosis, consistent with the effects observed using insulinoma cells.

Discussion
In this study, we show that pancreatic β-cells lack the capacity to respond to extracellular dsRNA; however, in response to intracellular dsRNA, they express antiviral type I IFNs and ISGs and die by caspasedependent apoptosis. The selective recognition of intracellular dsRNA contrasts with the response of macrophages, which have the capacity to respond to both intracellular and extracellular dsRNA (32,60). In macrophages, intracellular dsRNA recognition results in the expression of type 1 IFN and ISG (26,30,32,60), while extracellular signaling activated by dsRNA results in macrophage expression of inflammatory genes such as iNOS, IL-1β, and COX-2 in a manner dependent on CCR5 signaling (46). While βcells fail to respond to extracellular dsRNA, they express iNOS and COX-2 in response to the inflammatory cytokine IL-1β (17). Nitric oxide is both damaging and protective to βcells. It inhibits mitochondrial oxidative metabolism and insulin secretion and causes DNA damage, but also stimulates a protective adaptive unfolded protein response, induces the expression of DNA repair genes, and attenuates DNA-damage associated apoptosis (56). We now provide biochemical evidence showing that dsRNA elicits responses from pancreatic β-cells that differ from the responses elicited by cytokines. Cytokines activate an inflammatory cascade in β-cells that includes the production of nitric oxide. Conversely, β-cells fail to produce nitric oxide or express type I IFNs in response to extracellular polyI:C, while intracellular dsRNA stimulates the expression of type 1 IFN and causes caspase-dependent βcell apoptosis (56).
The differing responses of macrophages and β-cells are likely a consequence of the expression and signaling by multiple sensors that recognize viral replicative intermediates such as dsRNA. Once activated these sensors control the expression of type I IFNs during a viral infection (22). Type I IFNs act in an autocrine or paracrine manner to regulate the expression of hundreds of antiviral genes that (among other functions) promote the cell's ability to recognize viral PAMPs and inhibit different steps in viral replication, maturation, and release (23,24). Even though there are multiple dsRNA sensors, their capacity to identify and respond to dsRNA is not redundant. dsRNA sensors display selectivity for the structure of dsRNA as well as the cellular compartment in which the dsRNA is identified. Although less well studied, cell type selective expression of dsRNA sensors also limits the cellular response to dsRNA. For example, the dsRNA sensor TLR3 is localized to endosomes and recognizes extracellular dsRNA that is taken up during endocytosis in macrophages and dendritic cells (26-28), and is required for expression of type I IFN ((32,46)). Consistent with the lack of TLR3 expression, β-cells do not express type I IFN or ISG in response to extracellular polyI:C (Fig. 1, 2). While type I IFN treatment stimulates β-cell expression of TLR3 (Fig. 3A), this alone is not sufficient to confer insulinoma cells (Fig. 3B) or rat islet cells depleted of resident macrophages (Fig. 8) the capacity to produce type I IFNs in response to extracellular polyI:C. This is unlikely due to failure of efficient endocytosis given that similar amounts of extracellular polyI:C are taken up by β-cells and macrophages (Fig. 1D), whose sensing of extracellular polyI:C is dependent on endosomal TLR3 (46). Consistent with our findings, others have shown that insulinoma cells and primary β-cells (rat, human, and mouse) do not readily express TLR3 (61,62); however, intracellular polyI:C can induce the expression of type I IFN (62). In addition to triggering type 1 IFN production, intracellular polyI:C stimulates NF-κB nuclear localization to similar levels in islet cells isolated from wild-type and TLR3-deficient mice indicating that dsRNA sensors in addition to TLR3 contribute to the β-cell response to dsRNA.
The expression of type I IFNs and ISGs in β-cells in response to intracellular polyI:C (Fig. 2, 3) correlates with the basal expression of multiple cytosolic dsRNA sensors, including mda5, RIG-I, and PKR (Fig. 1). Each receptor is capable of recognizing dsRNA produced during viral replication (30,31); however, of the two RNA helicases, mda5 selectively respond to long dsRNA (multiple kilobases, such as diabetogenic RNA picornaviruses like EMCV and polyI:C), whereas RIG-I recognizes short dsRNA (<1 kilobase) and uncapped single stranded (ss)RNA (32-35). In support of this observation, macrophages lacking mda5 fail to express type I IFNs in response to EMCV infection (32,60) or intracellular polyI:C (32) and mda5-deficient mice rapidly become diabetic and die in response to EMCV infection due to impaired type I IFN responses (49). These findings illustrate the important role of dsRNA sensormediated type I IFN responses in β-cell in the protection against viral infection.
PKR is an eIF2α kinase that has been shown to attenuate cap-dependent translation following recognition of dsRNA (36,37) and to participate in the regulation of mda5-dependent type I IFN expression (63). In fibroblasts PKR participates in the activation of NF-κB via inhibitory kappa kinase (IKK) phosphorylation and subsequent IκBα degradation (48,64). In contrast, PKR is not required for dsRNAdependent activation of NF-κB or the expression of inflammatory genes in macrophages (65,66). The role of PKR in the response of β-cells to dsRNA has yet to be fully characterized. We have shown that PKR regulates the induction of apoptosis in dispersed islet cells in response to polyI:C (20), suggesting that PKR may be the primary dsRNA sensor controlling β-cell viability in response to recognition of intracellular dsRNA, and is consistent with PKR's well-described role in the regulation of apoptosis (20,48,67).
In response to IL-1β exposure, the function and viability of β-cells is decreased due to the expression of iNOS and β-cell production of nitric oxide (13)(14)(15), and this damage has been suggested to contribute to the development of autoimmune inflammatory diseases including diabetes (12,68,69). While reports suggest that cytokines kill β-cells by apoptosis, nitric oxide has been shown to attenuate caspase activation in multiple cell types including β-cells (53)(54)(55). Others have suggested that nitric oxide does not contribute to β-cell damage; however, cytokine treatment fails to decrease the viability of islets isolated from mice deficient in iNOS (16,20,50,51). Our studies support nitric oxide as the primary mediator of cytokine-induced inhibition of mitochondrial oxidative metabolism and insulin secretion (13)(14)(15)17). Although nitric oxide inhibits DDR-induced apoptosis (53-55), nitric oxide does not affect apoptosis in response to recognition of intracellular dsRNA ( Figure 6F). Rather it is the collateral bystander necrotic cell death caused by nitric oxide that may be an important determinant in tissue damage and potential auto-antigen release from β-cells during disease development. Therefore, β-cell induction of apoptosis in response to intracellular recognition of viral dsRNA may function to limit viral spread locally during infection and to limit inflammation by decreasing virus PAMP formation.
β-cell death in response to intracellular polyI:C is mechanistically distinct from the type of cell death induced by IL-1β. β-cell death in response to intracellular polyI:C is associated with caspase 3/7 activation and morphological changes that are consistent with apoptosis, including cell shrinkage and chromatin condensation ( Fig.  4-6). Interestingly, intracellular polyI:C stimulated death is not prevented by nitric oxide (Fig. 6F), even at nitric oxide concentrations sufficient to attenuate apoptotic cell death in response to DNA damaging agents like camptothecin (Fig.  6F, (52,56)). These data suggest that recognition of dsRNA within β-cells, which results in the rapid and potent induction of apoptosis (Figs. 4, 6), may be a protective mechanism to limit the spread of infectious viral progeny and to prevent necrosis of infected cells that could otherwise result in the release of auto-antigens or viral epitopes that structurally mimic self-epitopes, such as glutamic acid decarboxylase and the P2-C protein in coxsackievirus B4 (70).
Intraislet macrophages may also contribute to β-cell protection. Intracellular polyI:C stimulates IFN-β mRNA accumulation and the expression of the ISGs Mx2, Viperin, and RNaseL by rat islet cells (Figs. 7, 8). Depletion of intraislet macrophages via low temperature culture attenuates IFN-β and ISG mRNA accumulation (Fig. 8). In response to extracellular polyI:C, Mx2 and viperin mRNA accumulation is detectable (Fig. 7). The accumulation of ISG mRNA in response to extracellular polyI:C is likely due to intraislet macrophages, as islets devoid of macrophages no longer accumulate Mx2 and viperin mRNA (Fig. 8). These findings suggest that intraislet macrophage-generated responses including IFN-β mRNA accumulation and ISG expression may function as an early signal to βcells that an infection or potential damaging agent is present and increase defenses that limit β-cell damage during infection.
Macrophage depletion from islets: Resident islet macrophages were depleted from rat islets using the low temperature culture at 23 0 C for 7 days in complete CMRL-1066 media in an atmosphere of 5% CO2 and 95% air as previously described (19,57). Poly IC uptake was examined by incubating cells (50,000 cells/200 µL of DMEM) with a mixture of HMW and LMW rhodamine labeled polyI:C (5 µg/mL) (InvivoGen) and unlabeled polyI:C (45 µg/mL) for 16 h. The cells were then washed 3 three times DMEM and uptake was determined by fluorescence spectroscopy using an excitation of 546 and measuring the emission at 576 nm.
Nitrite determination: Nitrite production was determined by the addition of 50 µl of the Griess reagent to 50 µl culture supernatant (74). The absorbance was measured at 540 nm and nitrite concentrations were calculated from a sodium nitrite standard curve.
Western blot analysis: Protein samples were separated by SDS-PAGE, transferred to nitrocellulose membranes (Amersham Life Sciences, Pittsburgh, PA) under semi-dry transfer conditions, and blocked in either 3% BSA in TBST or 5% milk in TBST for 1 hour. Membranes were incubated overnight at 4 o C with the following antibody dilutions: mouse anti-GAPDH, 1:20,000; rabbit anti-phospho-JNK, 1:5,000; rabbit anti-phospho-p38, 1:2,000; and 1:1000 for all other primary antibodies. After three washes in TBST, membranes were incubated with HRP-conjugated secondary antibodies for 1 h at 1:10,000 dilutions. Antigen was detected by chemiluminescence as previously described (75).
Immunofluorescence: Following dispersion, islet cells were added to wells of a plastic Permanox chamber slide at a concentration of 50 dispersed islets per 200 µl of complete CMRL 1066. TUNEL stain was performed according to manufacturer's instructions (Roche) NS Hoechst 33342 (Molecular Probes) was used as a nuclear marker (5 µg/ml in PBS for 15 min, room temperature). Guinea pig anti-insulin (1:100 dilution) followed by Cy3-conjugated donkey anti-guinea pig (1:400) was used to identify βcells. Images were captured using a Nikon eclipse 90i confocal microscope. All images were deidentified and quantification performed by individuals blinded to the experimental conditions. Insulin-and TUNEL-positive cells were quantified by manual counted in a blinded using ImageJ (National Institutes of Health).
Cell viability: Cells were plated (5 x 10 4 / 100 µl media) in duplicate and following treatments, one replicate was incubated with 5 µM Sytox dye and the other replicate was incubated with both Sytox dye and 120 µM digitonin. After 30 min incubation with Sytox at 37 o C, cell viability was determined by fluorescence spectroscopy with an excitation/emission of 504/523 nm on a BioTek SynergyMx plate reader. Cell viability was determined by dividing the mean RFU of each by its corresponding average digitonin-treated control.
Real-time PCR analysis: Following lysis, total cellular RNA was isolated using the RNeasy RNA isolation kit according to manufacturer's instructions (Qiagen). Turbo DNA-free (Applied Biosystems) was used for DNase digestion. First-strand cDNA synthesis was performed using oligo(dT) and reverse transcriptase Superscript Preamplification System (Invitrogen). Semi-quantitative real-time PCR analysis was performed on cDNA samples using Quantitect SYBR Green reagent (Qiagen) and the MJ Research DNA Engine Opticon System or the SsoFast Evagreen Supermix (BioRad) and the BioRad CFX96 Real-Time detection system per manufacturer's instructions. Each sample was normalized to GAPDH (ΔCtT) and each condition was expressed as a fold-increase compared to each corresponding untreated control (2 -ΔΔCt ). cDNAs were amplified using the following primers purchased from Statistics: Statistical comparisons were made between two groups using students Ttest. Statistical comparisons made between three or more independent conditions were performed using one-way analysis of variance (ANOVA). Significant differences between groups (P < 0.05, *) were determined using the Tukey-Kramer post-hoc test.