Department of Pathology and Microbiology, University of Nebraska Medical Center, Omaha, Nebraska, USA
Address correspondence and reprint requests to Tammy Kielian, University of Nebraska Medical Center, Department of Pathology and Microbiology, 985900 Nebraska Medical Center, Omaha, NE 68198-5900, USA. E-mail: firstname.lastname@example.org
Juvenile neuronal ceroid lipofuscinosis (JNCL) is a lysosomal storage disease caused by an autosomal recessive mutation in CLN3. Regions of microglial activation precede and predict areas of neuronal loss in JNCL; however, the functional role of activated microglia remains to be defined. The inflammasome is a key molecular pathway for activating pro-IL-1β in microglia, and IL-1β is elevated in the brains of JNCL patients and can induce neuronal cell death. Here, we utilized primary microglia isolated from CLN3Δex7/8 mutant and wild-type (WT) mice to examine the impact of CLN3 mutation on microglial activation and inflammasome function. Treatment with neuronal lysates and ceramide, a lipid intermediate elevated in the JNCL brain, led to inflammasome activation and IL-1β release in CLN3Δex7/8 microglia but not WT cells, as well as increased expression of additional pro-inflammatory mediators. Similar effects were observed following either TNF-α or IL-1β treatment, suggesting that CLN3Δex7/8 microglia exist in primed state and hyper-respond to several inflammatory stimuli compared to WT cells. CLN3Δex7/8 microglia displayed constitutive caspase-1 activity that when blocked led to increased glutamate release that coincided with hemichannel opening. Conditioned medium from activated CLN3Δex7/8 or WT microglia induced significant cell death in CLN3Δex7/8 but not WT neurons, demonstrating that intrinsically diseased CLN3Δex7/8 neurons are less equipped to withstand cytotoxic insults generated by activated microglia. Collectively, aberrant microglial activation may contribute to the pathological chain of events leading to neurodegeneration during later stages of JNCL.
Juvenile neuronal ceroid lipofuscinosis (JNCL) is a lysosomal storage disease caused by an autosomal recessive mutation in CLN3. Regions of microglial activation precede and predict areas of neuronal loss in JNCL; however, the functional role of activated microglia remains to be defined. In this report, primary microglia from CLN3Δex7/8 mutant mice over-produced numerous inflammatory cytokines in response to stimuli that are present in the JNCL brain, whereas wild-type microglia were relatively non-responsive. In addition, activated microglia induced significant cell death in CLN3Δex7/8 but not wild-type neurons, demonstrating that intrinsically diseased CLN3Δex7/8 neurons are less equipped to withstand cytotoxic insults. Collectively, aberrant microglial activation may contribute to the pathological chain of events leading to neurodegeneration during later stages of JNCL.
Neuronal ceroid lipofuscinosis (NCL; also known as Batten disease) encompasses a group of autosomal recessive lysosomal storage disorders characterized by the abnormal accumulation of proteins and lipopigments in all cell types (Palmer et al. 1993). Juvenile neuronal ceroid lipofuscinosis (JNCL), the most common inherited neurodegenerative disease in childhood, is caused by a mutation in the CLN3 gene (International Batten Disease Consortium 1995). Approximately, 80–85% of affected children harbor a 1.02 kb deletion on chromosome 16p12, although other point mutations have also been described (International Batten Disease Consortium 1995; Munroe et al. 1997). Despite identifying the causative gene, the normal function of CLN3 remains elusive, which has been complicated by low protein abundance and lack of reliable antibodies (Getty and Pearce 2011). JNCL typically presents in children between the ages of 5–10 years, initiating as blindness and progressing to seizures, advancing cognitive and motor decline, and dementia, with a life expectancy into the late teens to early twenties (Rakheja et al. 2007). Currently, there is no treatment for JNCL, which highlights the need to understand pathologic mechanisms contributing to disease progression for the development of novel therapeutics to extend the survival and quality-of-life of afflicted children.
Microglia are the resident immune cells within the central nervous system (CNS) parenchyma and exert multiple functions, including phagocytosis of cellular debris and pro-inflammatory mediator release (Kettenmann et al. 2011). Abnormal neuron-microglia interactions have been implicated in the pathogenesis of several neurodegenerative diseases, including Alzheimer's and Parkinson's disease (Huh et al. 2011; Byun et al. 2012). Activated microglia are also observed in the brains of JNCL patients, as well as associated mouse models, and predict areas that will undergo neurodegeneration (Pontikis et al. 2004, 2005). However, the secretory phenotype of activated microglia in the context of CLN3 mutation remains to be defined. This is a critical issue, since activated microglia may either positively or negatively influence neuronal survival via the production of growth factors or pro-inflammatory mediators, respectively (Arai et al. 2004).
Previous work from our laboratory and others has defined the inflammasome as a key molecular pathway responsible for processing the pro-inflammatory cytokine interleukin-1 beta (IL-1β) into its active form in microglia (Halle et al. 2008; Hanamsagar et al. 2011). IL-1β expression is also elevated in the brains of JNCL patients (Dr. Jonathan Cooper, personal communication), which suggests that microglial activation may contribute to the development and progression of neurodegeneration in JNCL. Depending on the initiating stimulus, activated microglia also produce reactive oxygen species (ROS) (Bal-Price and Brown 2001; Block et al. 2007), which have recently been shown to trigger inflammasome activation (Zhou et al. 2011), linking the two processes. Prior studies have revealed oxidative imbalance in the brains of CLN3 knockout (KO) mice (Benedict et al. 2007) and increased sensitivity of CLN3 mutant Drosophila to oxidative stress (Tuxworth et al. 2011), and IL-1β has long been recognized for its neurotoxic properties (Arai et al. 2004; Thornton et al. 2006). Collectively, these observations strongly suggest that these pathways may intersect, which could represent a pathological target in the context of CLN3 mutation.
In addition to pro-inflammatory cytokines, brains of JNCL patients and CLN3 KO mice are characterized by increased levels of ceramide, an important lipid metabolism intermediate (Puranam et al. 1997; Mencarelli and Martinez-Martinez 2013). Elevated ceramide has been shown to induce neuronal apoptosis, neuroinflammation, and demyelination (Cutler et al. 2004; Kim et al. 2012; Wang et al. 2012) and dysregulated ceramide biosynthesis has been implicated in microglial activation in mouse models of neuropathic pain and Alzheimer's disease (Filippov et al. 2012; Kobayashi et al. 2012). Here, we report that treatment of primary CLN3Δex7/8 microglia with C6 ceramide and neuronal lysates induced inflammasome/caspase-1 activation and IL-1β release, whereas wild-type (WT) microglia were largely non-responsive. Low levels of constitutive caspase-1 activity were detected in CLN3Δex7/8 microglia, which when inhibited resulted in significant glutamate release, presumably via hemichannel opening, which was also elevated in CLN3Δex7/8 microglia. CLN3Δex7/8 microglia also displayed increased expression of other pro-inflammatory cytokines and chemokines in response to C6 ceramide and neuronal lysates, indicating a broader effect extending beyond IL-1β release. Similar findings were observed when CLN3Δex7/8 microglia were stimulated with either TNF-α or IL-1β, demonstrating the existence of a conserved core pathway for eliciting heightened microglial activation in the context of CLN3 mutation. Finally, CLN3Δex7/8 neurons were more sensitive to microglial-induced cytotoxicity compared with WT cells. Collectively, these results suggest that CLN3Δex7/8 microglia exist in a primed state to produce inflammatory mediators with known neurotoxic effects, which over time, may contribute to JNCL progression.
Materials and methods
CLN3Δex7/8 mice (C57BL/6 background) that lack a 1.08 kb segment spanning exons 7 and 8 of CLN3 (Cotman et al. 2002) were provided by Dr. David Pearce (Sanford Children's Hospital, Sioux Falls, SD, USA). This represents the most common mutation in ~ 80–85% JNCL patients and CLN3Δex7/8 mice phenocopy several aspects of JNCL, including neuronal loss, glial activation, metabolic disturbances, and progressive deposition of autofluorescent storage material (Cotman et al. 2002; Pontikis et al. 2005; Herrmann et al. 2008; Osorio et al. 2009). Age- and sex-matched C57BL/6 mice were used as WT controls (National Cancer Institute, Frederick, MD, USA). The animal use protocol, approved by the University of Nebraska Medical Center Animal Care and Use Committee, is in accord with the National Institutes of Health guidelines for the use of rodents.
Primary microglia and neuron cultures
Mixed glial cultures were prepared from 1 to 2 day-old CLN3Δex7/8 and WT pups as previously described (Esen et al. 2004). Microglia were harvested weekly from mixed glial cultures over a 3-week period using a differential shaking technique, which resulted in purity of approximately > 98%. All experiments were performed with microglia after overnight culture to allow cells to return to a quiescent state after shaking and plating. Although some expansion of residual astrocytes was possible during this period, we and others have demonstrated that astrocytes do not produce pro-inflammatory cytokines (i.e. IL-1β or TNF-α) (Liu and Kielian 2011; Holm et al. 2012). Therefore, the inflammatory cytokines measured here are attributable to microglia and not a minor fraction of contaminating astrocytes. Primary microglia were exposed to various stimuli including C6 ceramide (Sigma-Aldrich, St. Louis, MO, USA), primary cortical neuron lysates, and recombinant mouse TNF-α or IL-1β (1–10 ng/mL; Invitrogen, San Diego, CA, USA). None of the stimuli utilized in these studies exhibited any toxicity to CLN3Δex7/8 or WT microglia, since pilot experiments were conducted to identify optimal doses that triggered microglial responses without any adverse effects on cell viability (data not shown).
Primary neurons were prepared from the cortices of E16 WT embryos as previously described with minor modifications (Bledi et al. 2000). Briefly, cerebral cortices were dissected and subjected to trypsinization, whereupon single cell suspensions were prepared in Neurobasal medium containing serum-free B27 supplement (2%) and 1% gentamycin and seeded in polyethyleneimine (PEI)-coated six-well plates at 2 × 106 cells/well. On day 2 in vitro (DIV 2), culture medium was supplemented with 5 μM cytosine arabinofuranoside (AraC) to limit glial proliferation. At DIV 5, medium was replenished with fresh Neurobasal medium containing B27 supplement and gentamycin. At DIV 10, primary neuron cultures were washed 3X with ice-cold phosphate-buffered saline (PBS), whereupon cell pellets were resuspended in 500 μL PBS and subjected to four freeze-thaw cycles to lyse cells. Soluble extracts were collected following centrifugation at 21 000 g for 10 min at 4°C. Neuron lysates were added to microglia at a 1 : 5 dilution in all experiments. This concentration was optimized in pilot studies by examining the ability of neuronal lysates to trigger cytokine production (data not shown).
For evaluating the effects of microglial-conditioned medium on neuronal survival, primary cortical and cerebellar neurons were prepared from E16 CLN3Δex7/8 and WT mice as described above and cultured in PEI-coated 96-well plates at 5 × 103 cells/well. At day 10 in culture, neurons were exposed to cell-free conditioned medium collected from unstimulated or CLN3Δex7/8 and WT microglia treated with C6 ceramide + neuronal lysates for 24 h. After a 24 h incubation period, neuronal viability was assessed by quantitating lactate dehydrogenase release according to the manufacturer's instructions (CytoTox 96®; Promega, Madison, WI, USA).
Preparation of bone marrow-derived macrophages
Bone marrow-derived macrophages were prepared as previously described (Kigerl et al. 2009). Briefly, marrow was flushed from the long bones of CLN3Δex7/8 and WT mice and after red blood cell lysis, cells were cultured for 7 days in RPMI-1640 medium supplemented with conditioned medium from the fibroblast cell line L929 as a source of macrophage colony-stimulating factor.
Caspase-1 activity assays
Primary CLN3Δex7/8 and WT microglia were exposed to C6 ceramide, neuronal lysates, or a combination of both stimuli for either 6 or 24 h. Cells were treated with lipopolysaccharide (LPS) + ATP as a positive control, since this combination is a well-characterized inducer of caspase-1 activity (Ferrari et al. 1997). Caspase-1 activation was analyzed using the FLICA reagent (Immunochemistry Technologies, Bloomington, MN, USA) according to the manufacturer's instructions and microglial viability following a 6- to 24-h treatment period was assessed by evaluating propidium iodide (PI) uptake. A total of 105 microglia were analyzed by flow cytometry using a BD LSRII cytometer (BD, Franklin Lakes, NJ, USA) with autofluorescence correction using unstained cells. Results are presented as the % caspase-1+ microglia that were viable (i.e. PI−). To demonstrate the requirement for caspase-1 activation in CLN3Δex7/8 microglia, cells were treated with various concentrations of the caspase-1-specific inhibitor Z-WEHD-FMK (R&D Systems, Minneapolis, MN, USA).
Quantification of cytokine and chemokine expression
To compare cytokine and chemokine expression patterns between CLN3Δex7/8 and WT microglia following C6 ceramide and neuronal lysate treatment, a multi-analyte microbead array was utilized according to the manufacturer's instructions (MILLIPLEX; Millipore, Billerica, MA, USA). This assay allows for the simultaneous detection of several inflammatory mediators, including IL-1α, IL-1β, TNF-α, IFN-γ, IL-6, IL-9, IL-10, IL-12p40, IL-12p70, IL-15, IL-17, CXCL1, CXCL2, CXCL9, CXCL10, CCL2, CCL3, CCL4, and CCL5. Results were analyzed using a Bio-Plex workstation (Bio-Rad, Hercules, CA, USA). In some experiments, standard sandwich ELISA kits were utilized to measure the production of IL-1β, TNF-α, and IL-6 (BD Biosciences, San Jose, CA, USA).
Determination of extracellular glutamate levels
Glutamate release from primary microglia was measured using an Amplex Red Glutamic Acid Assay Kit (Molecular Probes, San Diego, CA, USA), where glutamate is oxidized by glutamate oxidase to produce α-ketoglutarate, NH4, and H2O2. H2O2 reacts with the Amplex red reagent, which generates the highly fluorescent product resorufin. Conditioned supernatants from CLN3Δex7/8 and WT microglia were incubated with the Amplex Red reagent and signals were measured at 590 nm using a fluorescence microplate reader (VICTOR, Perkin Elmer, Waltham, MA, USA), with glutamate concentrations determined using a standard curve with l-glutamic acid.
Electron paramagnetic resonance (EPR) spectroscopy
Intracellular superoxide levels were assessed in real-time by EPR spectroscopy, as previously described (Rosenbaugh et al. 2010). Briefly, CLN3Δex7/8 and WT microglia were seeded in 6-well plates at 106 cells/well and treated with C6 ceramide, neuronal lysates, or a combination of both stimuli for 2 h, whereupon cells were washed twice with 0.5 mL of electron spin resonance (ESR) buffer (0.025 mM deferoxamine methanesulfonate salt and 0.005 mM diethyldithiocarbamic acid sodium salt, pH 7.4) and incubated with 1 mL of ESR buffer containing cyclic hydroxylamine (CMH; 200 μM) as the intracellular spin probe for 1 h at 37°C (all reagents were purchased from Noxygen Science Transfer and Diagnostics, Elzach, Germany). After incubation, 0.9 mL of ESR buffer was aspirated from each well and microglia were gently scraped using a rubber policeman and re-suspended in residual ESR buffer, whereupon 50 μL of each sample was loaded into a glass capillary tube for analysis in an EPR spectrometer (Bruker e-scan M, Billerica, MA, USA). The CMH radical signal was recorded and EPR spectrum amplitude quantified. EPR amplitudes for each sample are expressed as arbitrary units (AU) per 106 microglia to correct for potential differences in cell numbers between each treatment.
CLN3Δex7/8 and WT microglia were cultured on Lab-Tek glass chamber slides (Nunc, Rochester, NY, USA) for 24 h prior to experiments. Microglia were treated with C6 ceramide and neuronal lysates for 2 h, whereupon mitochondrial ROS production and total respiring mitochondria were assessed using Mitotracker Deep Red™ and Mitotracker Green™, respectively (both from Invitrogen). Microglia were imaged using a Zeiss 510 META laser scanning confocal microscope (Carl Zeiss, Oberkochen, Germany) with excitation/emission wavelengths of 490 nm/516 nm for Mitotracker Green and 644 nm/655 nm for Mitotracker Deep Red.
Hemichannel activity assays
Hemichannel activity in CLN3Δex7/8 and WT microglia was evaluated by monitoring the uptake of ethidium bromide (EtBr), a small molecular weight dye (394 Da) widely used for evaluating hemichannel function (Karpuk et al. 2011). Briefly, microglia were seeded in 35-mm dishes at 106 cells/dish for 24 h prior to experiments. Cells were treated with C6 ceramide, neuronal lysates, or both stimuli for either 6 or 24 h, whereupon hemichannel activity was assessed after a 1-h incubation period with EtBr (20 μM) and Hoechst dye (1 μM), the latter of which is internalized by all cells. Since both dyes accumulate in the nucleus, the total number of EtBr and Hoechst double-positive microglia were quantitated in each field of view using a Zeiss Axio Examiner microscope (Carl Zeiss), with results normalized to values obtained from WT unstimulated microglia. A minimum of six random 20X fields of view were acquired for each sample and analyzed using Axio Vision software (Carl Zeiss).
Significant differences between experimental groups were determined by a Student's t-test or one-way anova followed by the Holm–Sidak method using the SigmaStat program (Systat Software, San Jose, CA, USA). For all analyses, a p-value < 0.05 was considered statistically significant.
Reactive microglia have been implicated as key contributors in several neurodegenerative disorders through their production of pro-inflammatory cytokines and ROS (Glass et al. 2010). Previous studies in CLN3 KO and CLN3Δex7/8 mouse models have demonstrated evidence of early microglial activation, which predicts regions where neuronal loss occurs later in the disease process (Pontikis et al. 2004, 2005). Additional evidence to suggest a link between microglial activation, pro-inflammatory activity, and JNCL is provided by the fact that the brains of JNCL patients display immunoreactivity for IL-1β (Dr. Jonathan Cooper, personal communication), which is a pivotal mediator of neurodegeneration and is predominantly expressed by microglia. However, it remains to be determined what triggers microglial activation in the context of JNCL and the functional consequences of microglial activity are not known. When considering the most relevant stimuli that microglia would encounter during the course of JNCL, we initially evaluated a panel of molecules whose expression is altered in JNCL patient brains, including glutamate, ceramides, ATP, and primary neuron lysates to model neuronal loss (Puranam et al. 1997; Chattopadhyay et al. 2002; Persaud-Sawin and Boustany 2005; Cao et al. 2011). Our pilot studies examining caspase-1 activity revealed that a combination of neuronal lysates and ceramide induced significant increases in caspase-1 activation, whereas the other stimuli were poor inducers (data not shown). Therefore, we utilized a combination of C6 ceramide and neuronal lysates for all subsequent experiments to evaluate potential functional differences between CLN3Δex7/8 and WT microglia.
Prior studies have shown that ceramide accumulates in the brains of JNCL patients and is known to induce neuronal apoptosis (Brugg et al. 1996; Puranam et al. 1997). In addition, ceramide is a key inflammatory signaling pathway and recent evidence has linked ceramide to inflammasome activation in macrophages (Vandanmagsar et al. 2011). Since IL-1β release from macrophages/microglia requires two signals, namely transcriptional activation of the IL-1β gene and production of pro-IL-1β (‘signal 1’), and cytoplasmic cleavage of pro-IL-1β by the inflammasome (‘signal 2’) (Hanamsagar et al. 2012), we first investigated whether inflammasome activity was dysregulated in CLN3Δex7/8 microglia. In these studies, neuronal lysates and C6 ceramide were used as ‘signal 1’ and ‘signal 2’ to trigger pro-IL-1β production and inflammasome activation, respectively, leading to mature IL-1β release. We began by comparing inflammasome activation in CLN3Δex7/8 and WT microglia utilizing the FLICA reagent, which upon cleavage by active caspase-1 emits a fluorescence signal (Meissner et al. 2010). Caspase-1 activity was significantly increased in CLN3Δex7/8 microglia in response to C6 ceramide and neuronal lysates compared to WT cells (Fig. 1a and b). Since caspase-1 activation leads to the release of mature IL-1β, we next measured IL-1β levels in culture supernatants. Similar to what was observed for caspase-1 activation, treatment of CLN3Δex7/8 microglia with C6 ceramide and neuronal lysates induced significant IL-1β release, whereas WT cells were non-responsive (Fig. 1c). Both signal 1 and signal 2 afforded by neuronal lysates and C6 ceramide, respectively, were required for IL-1β secretion, since neither stimulus alone was capable of eliciting cytokine release (Fig. 1c). Surprisingly, the opposite response was observed when CLN3Δex7/8 microglia were treated with LPS plus ATP, which was included as a positive control, since this represents a well-characterized ‘signal 1-signal 2’ combination for inflammasome activation (Ferrari et al. 1997). Specifically, both caspase-1 activation and IL-1β release were reduced in CLN3Δex7/8 microglia compared to WT cells in response to LPS + ATP stimulation (Fig. 1a–c). Although interesting, the LPS plus ATP combination was not pursued further in these studies, since LPS is not a stimulus that microglia would encounter in situ during the course of JNCL.
CLN3Δex7/8 microglia are primed to overproduce a wide range of pro-inflammatory mediators
Microglia have the potential to influence neuronal survival by releasing a myriad of inflammatory mediators and reactive oxygen/nitrogen intermediates (Block et al. 2007). Indeed, microglial dysfunction has been implicated in many neurodegenerative disorders, including Alzheimer's and Parkinson's disease (Hensley 2010; Fellner et al. 2011). Therefore, we next evaluated whether the hyperactive IL-1β response observed in CLN3Δex7/8 microglia extended to other inflammatory mediators. Indeed, CLN3Δex7/8 microglia produced significantly higher levels of numerous pro-inflammatory cytokines/chemokines following C6 ceramide and neuronal lysate treatment, including TNF-α, IL-1α, IL-9, and IL-15 (Fig. 2). Several other mediators were also elevated in CLN3Δex7/8 microglia (i.e. CXCL2, IL-12, and IL-17), although these differences did not reach statistical significance. The anti-inflammatory cytokine IL-10 was also significantly increased in CLN3Δex7/8 microglia, suggesting that the net balance of pro- versus anti-inflammatory mediators released in the context of CLN3 mutation may dictate the functional impact of microglia on CNS homeostasis during JNCL.
Since a recent report demonstrated changes in peripheral immune cells in CLN3Δex7/8 mice (Staropoli et al. 2012), we next examined whether the pro-inflammatory phenotype observed in CLN3Δex7/8 microglia was unique to this population or represented an intrinsic signaling defect in other mononuclear phagocytes, namely macrophages. This cell type was examined, since blood–brain barrier compromise has been reported in Batten disease (Lim et al. 2007; Saha et al. 2012), conceivably allowing macrophage infiltrates to encounter ceramides and neuronal debris upon CNS entry and contribute to pathology, particularly at later stages of disease. Similar to microglia, CLN3Δex7/8 bone marrow-derived macrophages produced significantly more IL-1β in response to C6 ceramide + neuronal lysate (Figure S1). TNF-α levels were also elevated in CLN3Δex7/8 bone marrow-derived macrophages, although this did not reach statistical significance. These findings suggest that CLN3Δex7/8 mononuclear phagocyte populations respond similarly and possess an intrinsic capacity to be primed for inflammatory mediator release.
Prior studies have revealed oxidative imbalance in the brains of CLN3 KO mice (Benedict et al. 2007) and increased sensitivity of CLN3 mutant Drosophila to oxidative stress (Tuxworth et al. 2011). To evaluate whether CLN3Δex7/8 microglia displayed aberrant ROS production, we utilized intracellular EPR spectroscopy to quantitate superoxide levels in real time. There were no significant differences in ROS production between CLN3Δex7/8 and WT microglia following stimulation with C6 ceramide and neuronal lysates (Figure S2). Similar results were obtained with CM-H2DCFDA, which measures total cellular ROS and Mitotracker Deep Red that measures mitochondrial ROS production (data not shown and Figure S3, respectively). In addition, CLN3Δex7/8 and WT microglia displayed similar sensitivities to several oxidative stressors as measured by cellular survival, including H2O2 that generates hydroxyl radicals via the Fenton reaction (Akaike et al. 1998); diethylmaleate that depletes cellular glutathione (Mitchell et al. 1983); and paraquat, which generates superoxide anions (data not shown) (McCormack et al. 2005). Collectively, these data indicate that aberrant pro-inflammatory mediator secretion by CLN3Δex7/8 microglia proceeds independently of ROS and that CLN3 mutation does not render microglia more sensitive to oxidative stress.
Caspase-1 inhibition attenuates inflammatory mediator production in CLN3Δex7/8 microglia
To confirm that the exaggerated IL-1β response observed in CLN3Δex7/8 microglia was caspase-1-dependent, cells were treated with the caspase-1-specific inhibitor Z-WEHD-FMK (Hanamsagar et al. 2011). This was important to demonstrate, since alternative enzymes have recently been shown to cleave pro-IL-1β, including elastase and some lysosomal enzymes (i.e. cathepsin B) that may be liberated following the loss of lysosomal membrane integrity, which may be relevant to JNCL (Halle et al. 2008; Codolo et al. 2013). As predicted, IL-1β release from CLN3Δex7/8 microglia was inhibited by Z-WEHD-FMK in a dose-dependent manner (Fig. 3a). Somewhat unexpectedly, other inflammatory mediators that do not require caspase-1 activation for secretion (i.e. TNF-α and IL-6) were also reduced in CLN3Δex7/8 microglia following Z-WEHD-FMK treatment (Fig. 3b and data not shown). The fact that caspase-1 inhibition also attenuated TNF-α release in WT microglia yet mature IL-1β was not detected, suggests that either IL-1β levels fell below the ELISA detection limit but were still biologically active or alternative actions of Z-WEHD-FMK. Collectively, these findings imply an autocrine/paracrine role for IL-1β in augmenting cytokine release in CLN3Δex7/8 microglia, suggesting that inhibiting caspase-1 activation may have broader implications to prevent inappropriate pro-inflammatory responses in microglia during JNCL.
CLN3Δex7/8 microglia display constitutive caspase-1 activity that prevents glutamate release
Chronic exposure to high concentrations of extracellular glutamate is a main cause of neuronal excitotoxicity (Ankarcrona et al. 1995). Indeed, microglial-derived glutamate release has been implicated in neuronal death during Alzheimer's disease and Rett syndrome (Noda et al. 1999; Maezawa and Jin 2010). Recent studies have shown that TNF-α triggers glutamate release from microglia, in part, via hemichannel activity (Maezawa and Jin 2010). Hemichannels span the plasma membrane and when open, facilitate the bi-directional trafficking of small molecules (i.e. < 1000 Da), which is thought to disrupt homeostatic pH and ion gradients during CNS pathological conditions (Kielian 2008; Eugenin et al. 2012). Since TNF-α production was significantly elevated in CLN3Δex7/8 microglia and could be attenuated indirectly by inhibiting caspase-1 activation, we examined whether CLN3 mutation would lead to enhanced glutamate release. There were no differences in extracellular glutamate levels between CLN3Δex7/8 and WT microglia at the resting state; however, unexpectedly, caspase-1 inhibition in CLN3Δex7/8 microglia resulted in exaggerated glutamate release in the absence of any additional stimuli (Fig. 4a). Treatment of CLN3Δex7/8 microglia with C6 ceramide and neuron lysate alone had no impact on glutamate levels, which were only significantly elevated in CLN3Δex7/8 microglia when caspase-1 activity was inhibited (Fig. 4a). The ability of caspase-1 inhibition to augment glutamate release in CLN3Δex7/8 microglia is in agreement with the trend toward elevated caspase-1 activity in these cells under resting conditions (Fig. 1b).
We next determined whether elevated glutamate release in CLN3Δex7/8 microglia following caspase-1 inhibition occurred through hemichannels. Indeed, hemichannel activity was significantly higher in CLN3Δex7/8 microglia pre-treated with the caspase-1 inhibitor only (Fig. 4b), in agreement with enhanced extracellular glutamate levels (Fig. 4a). Collectively, these results indicate that constitutive caspase-1 activity is critical for preventing hemichannel opening and subsequent glutamate release in CLN3Δex7/8 microglia, suggesting an altered baseline state in the context of CLN3 mutation.
CLN3Δex7/8 microglia are primed to respond inappropriately to pro-inflammatory cytokines
Although our results demonstrated that CLN3Δex7/8 microglia produce exaggerated levels of numerous cytokines/chemokines in response to C6 ceramide and neuronal lysates, it was important to determine whether this would extend to additional stimuli that could be encountered in the brain during the course of JNCL. In particular, we recognize that significant neuronal loss does not occur until later stages of JNCL; therefore, we evaluated the effects two pro-inflammatory cytokines (i.e. TNF-α and IL-1β), since they are likely induced in affected brain regions earlier in the disease process. Similar to what was observed with C6 ceramide and neuronal lysates, both TNF-α and IL-1β led to exaggerated IL-6 release in CLN3Δex7/8 microglia compared to WT cells (Fig. 5). This finding indicates that the heightened responsiveness of CLN3Δex7/8 microglia to stimuli that would be encountered during JNCL progression is a conserved response, at least in terms of the treatments evaluated in this study.
CLN3Δex7/8 neurons are more sensitive to microglial-induced cytotoxicity
On the basis of the exaggerated inflammatory profile of CLN3Δex7/8 microglia after exposure to JNCL-relevant stimuli, we next determined whether CLN3Δex7/8 microglial-conditioned medium would confer worse neuronal damage than WT microglia. Surprisingly, conditioned medium collected from either activated CLN3Δex7/8 or WT microglia induced significant cell death in only primary CLN3Δex7/8 cortical and cerebellar neurons, whereas WT neurons were resistant to cytotoxicity (Fig. 6). These findings demonstrate that intrinsically diseased CLN3Δex7/8 neurons are less equipped to withstand inflammatory insults generated by activated microglia. In the context of JNCL, it is envisioned that chronic microglial activation and exposure of CLN3Δex7/8 neurons to inflammatory mediators could be one mechanism responsible for neuronal loss as the disease progresses.
Activated microglia are observed in the brains of CLN3 mutant mouse models (Pontikis et al. 2004, 2005); however, relatively little information is known about the biological consequences of activated microglia in the context of disease progression. Here, we report that CLN3Δex7/8 microglia exist in a primed state, where exposure to stimuli encountered in the JNCL brain elicit exaggerated cytokine and chemokine production, inflammasome/caspase-1 activation, and hemichannel activity, leading to glutamate release. In contrast, WT microglia were largely non-responsive, emphasizing the importance of mutant CLN3 in controlling the microglial pro-inflammatory state. The precise signaling pathways and molecular mechanisms responsible for these exaggerated responses remain to be completely defined; however, our studies have established a central role for inflammasome/caspase-1 activation (Fig. 7). In particular, CLN3Δex7/8 microglia displayed enhanced caspase-1 activity, which translated into elevated IL-1β secretion, a cytokine that requires proteolytic processing by caspase-1 for its release (Franchi et al. 2009). In addition, caspase-1 inhibition significantly reduced the expression of additional inflammatory mediators that do not rely on enzymatic processing, suggesting an important role for autocrine/paracrine actions of IL-1β in promoting the pro-inflammatory phenotype of CLN3Δex7/8 microglia. Surprisingly, treatment of CLN3Δex7/8 microglia with a caspase-1 inhibitor alone led to heightened glutamate release, implying that constitutive caspase-1 activity is critical for maintaining glutamate homeostasis. This possibility was strengthened by the finding that inhibition of constitutive caspase-1 activity led to increased hemichannel opening, which is a known mechanism for glutamate release (Bennett et al. 2012; Mika and Prochnow 2012). The molecular identity of the hemichannel that is affected in CLN3Δex7/8 microglia remains to be determined, but could involve traditional proteins (i.e. Cx43 or pannexin 1) or other channels that are capable of transporting small molecular weight compounds (i.e. P2X7R)(Mika and Prochnow 2012). Interestingly, CLN3Δex7/8 microglia displayed increased caspase-1 activation without any perturbation, which supports an adaptive role for caspase-1 activity in the context of CLN3 mutation. The degree of caspase-1 activation becomes more evident when CLN3Δex7/8 microglia are challenged with inflammatory stimuli, where the enzyme contributes to exaggerated cytokine release. Interestingly, caspase-1 activation was also observed in WT microglia following treatment with C6 ceramide and neuronal lysate, yet no IL-1β release could be detected. This finding suggests that the threshold to trigger cytokine release was not achieved or IL-1β levels were below the limit of detection, whereas this was clearly not the case for CLN3Δex7/8 microglia that produced high levels of IL-1β and other pro-inflammatory mediators. We predicted that lysates prepared from primary CLN3Δex7/8 neurons would elicit more inflammation in CLN3Δex7/8 microglia compared to WT lysates. Instead, we found that lysates from both CLN3Δex7/8 and WT neurons were equally capable of triggering similar inflammatory responses in CLN3Δex7/8 microglia (data not shown). However, it is important to note that we utilized embryonic neurons that harbor few lysosomal inclusions compared to adult neurons in vivo; therefore, this may be one reason why we did not observe a striking difference between the neuronal lysate preparations. It is technically challenging to isolate neurons from adult mice; therefore, we are not able to address this issue in greater detail.
Currently, the factor(s) responsible for triggering inflammasome activation in CLN3Δex7/8 microglia remain unknown. We treated CLN3Δex7/8 microglia with inhibitors for both ROS and cathepsin B, but neither had an effect on IL-1β release (data not shown), suggesting that inflammasome/caspase-1 activation proceeds independently of both pathways that have been reported to elicit caspase-1 activation in macrophages (Lamkanfi 2011). Alternative possibilities include cytoplasmic mitochondrial DNA release or cellular osmotic disturbances (Petrilli et al. 2007; Shimada et al. 2012), which remain to be investigated. Recently, Ca2+ and cAMP have been shown to control inflammasome activation in macrophages (Lee et al. 2012; Murakami et al. 2012; Rossol et al. 2012) and we are currently evaluating these signaling intermediates as potential molecular triggers for inflammasome activity in CLN3Δex7/8 microglia. A prevailing concept in the field is that inflammasome activation results from a generalized ‘stress signal’ elicited following lysosomal membrane damage (Hornung and Latz 2010). The connection with lysosome dysfunction in JNCL is intriguing and may explain why inflammasome activation is heightened in CLN3Δex7/8 microglia and why it occurs in response to diverse stimuli. Another outstanding question is what type of inflammasome is triggered in CLN3Δex7/8 microglia, since we only evaluated the last step in the pathway (i.e. caspase-1). Based on its widespread involvement in eliciting caspase-1 activation in numerous cell types, the NLRP3 inflammasome is a likely candidate (Franchi et al. 2009), although other inflammasomes (i.e. NLRP1) cannot be disregarded at the present time.
Our studies revealed a broader involvement of caspase-1 activation in CLN3Δex7/8 microglia. Specifically, not only was IL-1β processing affected, but other pro-inflammatory cytokines and glutamate release were modulated. It is important to note that caspase-1 is capable of cleaving a wide range of substrates besides IL-1β, including the vATPase that acidifies lysosomes (Dix et al. 2008). It is intriguing to speculate that caspase-1-mediated inactivation of the vATPase may be responsible, in part, for the lysosome acidification defect in JNCL, although this remains to be determined.
Several studies have demonstrated immune system involvement during the course of JNCL. For example, autoantibodies to brain and retinal proteins are present in JNCL patients and CLN3 KO mice, and attenuation of autoantibody production leads to improvements in motor function (Pearce et al. 2004; Seehafer et al. 2011). Currently, the mechanism(s) responsible for triggering autoantibody formation are unknown; however, it was interesting to note that IL-15 production was dramatically elevated in CLN3Δex7/8 microglia, whereas little cytokine was released from WT cells. IL-15 is a key B cell cytokine that facilitates plasma cell differentiation and antibody secretion (Fehniger and Caligiuri 2001; Steel et al., 2012). Therefore, it is possible that activated CLN3Δex7/8 microglia may contribute to autoantibody formation during JNCL via IL-15 release, although this remains speculative.
Numerous studies have established that pro-inflammatory mediators, such as TNF-α and IL-1β, can induce glutamate release and neuronal death (Stoll et al. 2000; Rothwell 2003; Allan et al. 2005; McCoy and Tansey 2008). In addition, TNF-α regulates synaptic strength by modulating AMPA receptors and excessive TNF-α release contributes to hippocampal seizure activity (Beattie et al. 2002; Stellwagen and Malenka 2006; Huie et al. 2012; Santello et al. 2012). Importantly, defects in AMPA receptor activity have been implicated in early JNCL (Kovacs and Pearce 2008; Kovacs et al. 2011). Our results have demonstrated exaggerated TNF-α release in addition to other pro-inflammatory cytokines known to impact neuronal homoeostasis, including IL-1β and IL-6 in CLN3Δex7/8 microglia. We propose that CLN3 mutation leads to a progressive pro-inflammatory state within the JNCL brain, which over time contributes to heightened glutamate levels and neuronal death. Accordingly, we found that conditioned medium collected from either activated CLN3Δex7/8 or WT microglia induced cell death in primary CLN3Δex7/8 neurons, whereas WT neurons were resistant to cytotoxicity. These findings demonstrate that intrinsically diseased CLN3Δex7/8 neurons are less equipped to withstand inflammatory insults generated by activated microglia. In addition, elevated glutamate has been demonstrated by MR spectroscopy in the brains of JNCL patients as well as CLN3 KO mice (Salek et al. 2011) and neuronal loss during the disease has been attributed, in part, to glutamate-induced excitotoxicity (Kovacs et al. 2006, 2011; Finn et al. 2011). Therefore, a better understanding of the inflammatory feedback loop and interactions between these factors may reveal novel therapeutic mechanisms to delay JNCL disease progression.
Interestingly, unlike the heightened pro-inflammatory responses observed in CLN3Δex7/8 microglia following treatment with C6 ceramide and neuronal lysates, TNF-α, or IL-1β, this was not the case when CLN3Δex7/8 microglia were exposed to the gram-negative cell wall component LPS, where cells were significantly less responsive compared to WT microglia (Fig. 1 and data not shown). Microglia were treated with a combination of LPS and ATP to provide the requisite ‘signal 1 and signal 2’, respectively, for inflammasome activation (Ferrari et al. 1997). It is currently not clear why LPS and ATP induced opposite results compared to the other stimuli tested here; however, it is important to note that LPS does not represent a stimulus that microglia would encounter during the course of JNCL. We are planning to exploit these differences in future studies to identify disparities in signal transduction pathways that may shed new light on novel disease mechanisms. Interestingly, there is correlative evidence to suggest that JNCL patients are more susceptible to infections (Castaneda et al. 2008); however, no studies have been performed in CLN3 mouse models to determine whether peripheral inflammation leads to perturbations within the CNS during disease.
Based on the currently available data in mouse models of JNCL, it is envisioned that microglial activation precedes neuronal deficits because activated microglia are observed as early as post-natal day 7 in CLN3 mouse models, yet neuronal death is not evident until 5–7 months of age (Pontikis et al. 2004, 2005). Therefore, it is likely that this early microglial response affects the CNS homeostatic milieu and, by extension, neuronal viability. The precise mechanism whereby CLN3Δex7/8 microglia affect neuronal integrity/function remains unknown; however, the primed pro-inflammatory state of CLN3Δex7/8 microglia identified in this study, along with the proposed opening of microglial hemichannels by cytokines, such as TNF-α, may disturb neurons via excessive glutamate release. Indeed, we show here that constitutive caspase-1 activity in CLN3Δex7/8 microglia is involved in preventing glutamate release, and glutamate excitotoxicity is a leading mechanism to account for neuronal loss in JNCL (Kovacs and Pearce 2008; Finn et al. 2011; Kovacs et al. 2011). Therefore, although we cannot conclude at the present time that the primed inflammatory state of CLN3Δex7/8 microglia described in this report is a causal factor leading to neuronal demise in vivo, this possibility is supported by our in vitro studies where conditioned medium from CLN3Δex7/8 microglia induced the death of CLN3Δex7/8 neurons.
Collectively, these studies suggest that CLN3Δex7/8 microglia exist in a primed state as evident by constitutive caspase-1 activation and exaggerated pro-inflammatory responses to stimuli that microglia encounter in the JNCL brain. Caspase-1 appears to play a central role in these responses, as inhibiting the enzyme has broad-reaching implications on cytokine/chemokine production, glutamate release, and hemichannel activity. Therefore, it is intriguing to consider the potential of targeting caspase-1 activity as a potential novel JNCL therapeutic; however, the complex nature of this disease will likely require intervention with compounds that can simultaneously affect multiple aberrant pathways to achieve maximal improvements in survival and quality-of-life for children suffering from this fatal neurodegenerative disease.
The authors thank Debbie Vidlak for performing the MILLIPLEX assays, Dr. Matthew Zimmerman for assistance with EPR spectroscopy, and Dr. Nikolay Karpuk and Maria Burkovetskaya for critical review of the manuscript. This study was supported by the Batten Disease Support and Research Administration (BDSRA) and the UNMC Edna Ittner Pediatric Research Fund (to T.K.). J.X. is supported by a UNMC-Chinese Scholarship Council fellowship. The authors have no conflicts of interest to disclose.