Background
Inflammation, neuronal injury, and demyelination are common hallmarks of most central nervous system (CNS) diseases. Innate immunity activated through Toll-like receptors (TLRs) contributes to various forms of CNS injury. TLRs are a highly conserved class of type 1 transmembrane receptors that are involved in regulation of both innate and adaptive immunity [
1]. Thirteen members of the TLR family (TLR1-13) have been identified so far in humans, eleven of which are expressed in mice. Whereas some members of the TLR family including TLR3, TLR7, and TLR9 locate to intracellular compartments, other TLRs such as TLR4 are membrane-bound. TLRs detect conserved pathogen-associated molecular patterns expressed by various infectious agents including bacteria, viruses, fungi, and parasites. However, TLRs are not only activated by pathogen-associated molecules, but also by host-derived factors that are supposedly released during tissue injury [
1]. For example, TLR4 is activated by both lipopolysaccharide (LPS) derived from gram-negative bacteria and host-derived molecules such as heat shock proteins (HSPs) including HSP60 and HSP70, extracellular matrix proteins, such as fibronectin extra domain A and soluble heparin sulphate, and other proteins including β-defensin and high mobility group box 1 protein [
2-
8].
Upon binding to their respective ligands, TLRs initiate signaling through their intracellular Toll/interleukin-1 (IL-1) receptor (TIR) domains. These interact with other TIR domains such that, upon activation, each TLR binds to a specific set of adaptor proteins that also contain TIR domains. Myeloid differentiation factor 88 (MyD88) is the universal intracellular adaptor recruited by all known TLRs except TLR3. In general, triggering of the well-described complex TLR-associated signaling pathway results in activation of various downstream effectors including NF-κB, in turn leading to an inflammatory response [
1] or, occasionally, to cell death [
9]. TLRs are widely expressed in innate immune cells such as macrophages and microglia in the CNS and dendritic cells, but also in non-immune cells such as neurons and epithelial cells [
10,
11].
Injured tissues release factors including HSPs, which represent a collection of evolutionarily conserved proteins induced in response to cellular stress such as heat shock, nutrient deprivation, or mechanical damage, and which are considered endogenous danger signals to the immune system [
12-
14]. In addition to serving as a cellular chaperone, extracellular HSP60 directly activates immune cells including macrophages and dendritic cells through binding to TLRs and several other membrane receptors such as CD14 [
15,
16]. In line with this finding, we observed in previous work that HSP60 provides a signal to microglia, the major immune cell of the brain, alerting to the presence of CNS injury through activation of a TLR4-dependent pathway in cultured CNS cells [
17].
In this study we systematically analyze the impact of extracellular HSP60 on the brain. Intrathecal HSP60 results in CNS injury including neurodegeneration and demyelination. These effects require both TLR4 and MyD88. In contrast to intrathecal application of HSP60, intrathecal injection of the established pathogen-derived TLR4 ligand LPS does not affect neuronal or oligodendrocyte survival in vivo. However, whereas intrathecal LPS leads to robust production of proinflammatory molecules in the brain, CNS injury induced by HSP60 is not accompanied by such an inflammatory response. HSP60 is predominantly expressed in injured neurons and oligodendrocytes and is released during CNS injury such as focal cerebral ischemia in vivo. Based on our findings we speculate that activation of the TLR signaling pathway triggered by endogenous ligands may serve as a damage-specific danger signal in the CNS.
Discussion
Heat shock proteins released from damaged cells are key molecules during injury and inflammation and are capable of activating immunity. Accordingly, we showed in previous work that dying CNS cells release HSP60
in vitro, thereby leading to activation of microglia and subsequent neuronal injury [
17]. In contrast to the ongoing intensive research on the HSP60’s effects on cells of the peripheral immune system, the biological impact of extracellular HSP60 on the CNS
in vivo was undetermined so far. In the present study we demonstrate that intrathecal HSP60 mediates neurodegeneration and demyelination
in vivo.
In cerebral ischemia, TLRs contribute to CNS injury [
37,
38] and their expression was detected mainly in activated glia [
38]. These findings suggest that TLRs expressed in glia may recognize endogenous TLR ligands released during ischemic brain injury. However, the identity of such endogenous ligands is unknown. On the basis of our data it is tempting to speculate that HSP60 might be at least one of them. Our finding that HSP60 is overexpressed in injured neurons and oligodendrocytes and can be detected extracellularly in response to CNS injury match with the results from several other studies. For example, the amount of HSP60 in the serum of patients with a cerebral infarct is increased [
39]. Also, soluble HSP60 was detected in the CSF of children suffering from traumatic brain injury [
40].
At this stage, it remains unknown how HSP60 applied intrathecally exerts the observed degenerative effects in the brain. However, as described for other large proteins, intrathecal HSP60 may enter the brain tissue via receptor-mediated cellular uptake and intercellular transfer in principle [
41]. Several studies have reported intrathecal delivery of large proteins to the brain parenchyma. Moreover, in CNS diseases including stroke and neurodegenerative lysosomal storage disorders intrathecal application of such proteins can result in deep penetration into the brain tissue and is functionally effective [
42]. Also, we cannot rule out that intrathecal HSP60 leads to an inflammatory response that in turn causes degenerative effects in the brain paremchyma. Furthermore, it is unclear how much HSP60 is effectively released within the brain parenchyma at the site of injury in the context of CNS disorders and what the local concentration maxima are. The concentration of HSP60 used in the current study was based on our previous work
in vitro and on our experience with a ligand for TLR2
in vivo [
43,
44], and must be considered supraphysiologic. Our study is a proof of principle regarding the effects of HSP60 in the brain, and future work will be required to determine key properties of HSP60 as a signaling molecule in the brain and the pathophysiological concentration of HSP60 that is involved in CNS damage.
Glial activation regularly occurs in the setting of neuronal injury and demyelination [
24]. Microglia expressing TLR4 and releasing nitric oxide are required in neuronal injury induced by extracellular HSP60
in vitro [
17]. However, although HSP60 activates the immune response in the periphery [
45,
46], results of our present study indicate that intrathecal HSP60 does not result in major activation or proliferation of glia and does not induce a considerable inflammatory response, as provoked by LPS, in the cerebral cortex
in vivo. Notwithstanding, subtle activation of glial cells
in vivo likely occurs, as these cells constantly monitor their microenvironment and respond to injury in their proximity [
24]. We cannot rule out that specific inflammatory molecules different from the ones tested accumulate in response to intrathecal HSP60 or that the period of observation in our set-up was too short. Also, the expression of distinct proinflammatory mediators in the brain may vary depending on the concentrations of intrathecal HSP60, which we did not test. Although the release of inflammatory molecules from microglia activated by soluble danger signals has been demonstrated
in vitro [
17,
47], the situation in the CNS
in vivo is likely a lot more complex due to interaction of different cell types and further regulatory mechanisms. Indeed, recent studies on peripheral immune cells suggest that distinct inflammatory responses induced by HSP60 via TLR4 and TLR2 are dependent on the respective pathological context
in vivo [
48,
49]. Also, mechanisms that operate directly through neurons and oligodendrocytes have to be taken into account. TLR3, TLR8, and TLR7 mediate cell-autonomous neuronal and axonal injury upon activation [
44,
50,
51]. Although the expression of TLR4 in neurons is discussed controversely [
52,
53] and so far no significant expression of TLR4 in oligodendrocytes was detected [
23], TLR4 may be up-regulated and activated in these cells in the presence of HSP60
in vivo. However, in this case yet unidentified factors that have an impact on TLR activation must exist, since neurons and oligodendrocytes alone are not affected by HSP60
in vitro. Also, it cannot be ruled out that the observed degenerative effects
in vivo are not mediated by direct interaction between HSP60 and TLRs, but rather are caused by unknown secondary effects following the intrathecal application of HSP60.
It seems intriguing that activation of the same receptor and signaling molecule in the brain, namely TLR4 and MyD88, results in a different outcome regarding injury and inflammation depending on the ligand used. Intrathecal injection of the exogenous ligand LPS caused a robust inflammatory response but no CNS injury, while intrathecal injection of the endogenous ligand HSP60 led to neurodegeneration as well as demyelination but failed to induce a comparable inflammatory response in the brain as provoked by LPS. Although it is unknown, how inflammation and neuronal damage mediated by TLRs are related in the brain different pathways downstream of TLR4 and MyD88 may be activated in the CNS, depending on the ligand. Furthermore, taking into account the observed trend towards enhanced TUNEL+ positivity in brains of MyD88-deficient mice, but not in brains of TLR4-deficient animals after HSP60 treatment (see Figure
2D), we cannot rule out at this stage that other molecules downstream of TLR4 besides MyD88 contribute - at least in part - to HSP60-induced CNS injury. SARM, which is preferentially expressed in neurons and may play a crucial role in neuronal and axonal injury [
54], is a possible candidate molecule.
At least in the periphery HSP60 triggers a multifaceted process, dependent among others, on its interaction with other receptors [
55,
56]. Also, TLR4 exerts not only detrimental but also regenerative effects upon injury [
57]. The interaction between HSP60 and TLR4 may therefore not only cause damage in the CNS, but may also induce anti-inflammatory and neuroprotective effects [
58,
59].
Signaling by HSPs has long been the subject of considerable debate [
45]. In particular it was discussed that the effects of HSP60 on macrophages were likely to be caused by LPS contaminating recombinant HSP60 preparations [
60]. We showed previously that neurotoxic effects mediated by HSP60 are not caused by LPS [
17]. Notably, we used a low-endotoxin HSP60 batch in our studies, and this preparation was used after dilution to such an extent that the theoretical maximal content of LPS was by far not sufficient to induce neuronal injury
in vitro [
17]. We show here that intrathecal LPS itself does not affect neuronal and oligodendrocyte survival in the cerebral cortex
- in contrast to recombinant HSP60, which leads to degenerative processes
in vivo.
Besides TLR4, other receptors are involved in HSP60-induced inflammatory effects. LOX-1 serves as a receptor for HSP60 in dendritic cells in lung inflammation [
61]. The microglial TREM2 receptor was identified as an HSP60-binding protein, whose binding resulted in an increase in phagocytosis [
62]. Finally, not only TLR4 but also TLR2 in human fibroblasts responds to HSP60 [
2]. It is unknown whether the receptors named above are involved in CNS injury
in vivo. Also, other endogenous ligands besides HSP60 are released from both injured peripheral and CNS cells. For example, β-defensin, and HSP70 activating TLR2 and TLR4 [
3,
6], but also ligands of other receptors, such as let-7 miRNA activating TLR7 [
44] and HMGB1 binding to RAGE [
63], are host-derived molecules involved in inflammation and tissue damage. It is conceivable that many more yet unidentified molecules are released during tissue injury. However, since the identity of such molecules and the associated biological processes are still poorly defined, one can only estimate at this stage to which extent their interaction is biologically relevant in a given CNS disorder.
Material and methods
Animals
C57BL/6 J mice were purchased from Charles River, Sulzbach, Germany. TLR4−/− and MyD88−/− mice were generously provided by Dr. S. Akira (Osaka University, Department of Host Defense, Osaka, Japan). All animals were maintained according to the guidelines of the committee for animal care. In detail, mice were housed in groups with chip bedding and environmental enrichment on a 12 h light/dark cycle with ad libitum access to water and food (standard chow). All experiments were conducted in accordance with the European directive on the protection of animals used for scientific purposes and approved by the institutional review committee Landesamt für Gesundheit und Soziales, Berlin, Germany (LAGeSo registration TVA G 0175/07).
Primary culture of neurons and microglia
Purified microglia were generated from forebrains of neonatal P0-3 mice, as previously described [
18]. Briefly, brain tissue was dissociated with trypsin (Invitrogen, Darmstadt, Germany) for 20 min at 37°C. After mechanical dissociation, cells were plated in 75-cm
2 culture flasks in DMEM (Invitrogen, Darmstadt, Germany) supplemented with 10% FBS and penicillin/streptomycin. After 1 week in culture, mixed glial cultures were shaken for 30 min at 180 rpm. The supernatant containing >95% microglia was plated on poly-d-lysine-coated (BD Biosciences, San Jose, USA) glass coverslips. Microglia were maintained in DMEM with 10% FBS.
Primary cultures of cortical neurons were generated from forebrains of E17 mice. Cortices were dissociated by trituration with papain (Worthington, Lakewood, USA) in EBSS (Invitrogen, Darmstadt, Germany) for 5 min at 37°C. Subsequently, cells were resuspended in 0.25% trypsin inhibitor and 0.25% BSA (both obtained from Sigma-Aldrich, Munich, Germany) in EBSS, and incubated at 37°C for 5 min. Cells were pelleted by centrifugation at 1000 × g for 5 min. The cell concentration was adjusted to 1 × 106 cells/ml in MEM with GlutaMAX medium (Invitrogen, Darmstadt, Germany) supplemented with 10% FBS and penicillin/streptomycin. A total of 5 × 105 cells/24-well were plated onto poly-d-lysine-coated glass slides (BD Biosciences, San Jose, USA) and were maintained in humidified 5% CO2/95% air at 37°C. Immediate immunostaining revealed 90–95% purity for neurons.
At day 3 after plating of neurons half of the media per well was replaced by 60.000 microglia in DMEM. Co-cultures of microglia and neurons were incubated with the respective agents after a further 12 h.
Oli-neu precursor cell line and co-cultures containing Oli-neu cells and microglia
Oli-neu cells were generously provided by Dr. J. Trotter (Institute of Molecular Biology, Johannes Gutenberg-University, Mainz, Germany) [
64] and were cultured in DMEM (Invitrogen, Darmstadt, Germany) supplemented with 10% FBS and penicillin/streptomycin. At 12 h after plating of 6.000 Oli-neu cells half of the media per 24-well was replaced by 60.000 microglia in DMEM. Co-cultures of microglia and Oli-neu cells were incubated with the respective agents after a further 12 h.
Intrathecal injection into mice
Intrathecal injection into mice was performed as described previously [
43]. In brief, male C57BL/6, TLR4
−/−, or MyD88
−/− mice (~20 g) were anesthetized with i.p. ketamine (100 mg/kg; AllemanPharma, Pfullingen, Germany) and xylazine (16 mg/kg; Bayer, Leverkusen, Germany). A skin incision was made to expose the lumbar spine. Using a 30-gauge needle and a microliter syringe, 40 μl of water alone, LPS, HSP60, or SA solution, as indicated, were slowly injected into the spinal canal at vertebrae L2 or L3. The skin incision was closed using dermal clips. Animals were then allowed to wake up and given free access to food and water. Absence of paresis and adequate waking were verified. For toxicity studies, 40 μg recombinant human HSP60-Low Endotoxin (Enzo Life Sciences, Lörrach, Germany) was injected. The contamination with LPS of the HSP60 preparation as declared by the manufacturer was <50 EU/mg and ≤1.67 EU/mg, as determined by an independent laboratory specializing in endotoxin testing and published in previous work (Mikrobiologisches Labor, Münster, Germany; see also [
17]). Human serum albumin (SA, Ultra-Low endotoxin, HumanZyme, Chicago, USA, 40 μg) was used as control protein with a similar molecular weight as HSP60 (60 kDa). LPS (0111:B4) was purchased from Enzo Life Sciences, Lörrach, Germany.
Model of experimental murine stroke
Stroke was induced by middle cerebral artery occlusion (MCAo) in C57BL/6 J male mice at 10–12 weeks of age, as described previously [
65]. Briefly, under isoflurane (Abott, Wiesbaden, Germany) anaesthesia in a 1:2 mixture of oxygen/nitrous oxide a monofilament was inserted into the common carotid artery, advanced to the origin of the MCA, and left in place for 60 min until reperfusion. In sham-operated animals, the monofilament was also inserted and advanced to the MCA, but was withdrawn immediately, thereby avoiding ischemia. Mice were kept in heated cages for the next 2 h, and rectal temperature was measured regularly. Cerebrospinal fluid (CSF) samples were obtained, as previously described [
43], after one day. In brief, a skin incision was made over the head and neck. After dissection of the suboccipital muscles, the cisterna magna was punctured and CSF withdrawn using a 27-gauge butterfly cannula connected to a microliter syringe. Brains were removed, snap frozen in methylbutane, and stored at −80°C until further processing.
Immunocytochemistry and immunohistochemistry
Seventy-two hours after intrathecal injection or 24 h after MCAo, mice were deeply anesthetized with i.p. ketamine and xylazine and then fixed by transcardial perfusion with PBS followed by 4% paraformaldehyde (PFA) in phosphate buffered saline (PBS). Forebrains were post-fixed in 4% paraformaldehyde in PBS overnight, then cryoprotected in a row of 10%, 20%, and 30% sucrose. Cryostat coronal sections (15 μm) were thaw-mounted on coated glass slides. Representative brain sections (level 1: interaural 6.60 mm; level 2: 5.34 mm; level 3: 3.94 mm; level 4: 1.86 mm; level 5: −0.08 mm) or cell cultures were fixed with 4% PFA, washed with PBS, and treated with blocking solution (5% normal goat serum) for 1 h. They were then incubated with the respective primary antibody: anti-NeuN, anti-neurofilament (200 kDa, clone RT97), anti-myelin basic protein (MBP), anti-glial fibrillary acidic protein (GFAP), or anti-adenomatous polyposis coli (APC) (all purchased from Merck Millipore, Darmstadt, Germany), anti-HSP60 (Enzo Life Sciences, Lörrach, Germany), anti-Ki67 (ab 16667, Abcam, Cambridge, UK), anti-Iba1 (WAKO, Neuss, Germany), anti-CD11b (eBioscience, Frankfurt, Germany) or the isolectin IB4 (Invitrogen, Carlsbad, USA) overnight at 4°C. Subsequently, sections or cell cultures were incubated with the relevant secondary antibody (all purchased from Jackson Immuno Research, West Grove, USA) for 1 h at room temperature.
TUNEL staining of co-cultures and the CNS cortex was conducted after 3 d of incubation and 5 d after intrathecal injection, respectively, using the Apoptaq Plus Fluorescein In Situ Apoptosis Detection Kit, following the instruction manual (Merck Millipore, Darmstadt, Germany). DAPI staining was performed as previously described [
18]. Immunofluorescence images were obtained using an Olympus BX51 microscope.
Quantification of CNS cells in brain sections
Viability of neurons and oligodendrocytes in the brain’s cortex was analyzed by quantifying NeuN-positive and APC-positive cells, respectively, in six fields (x600) at level four of five representative sections of each brain. The mean was calculated, which is expressed as NeuN-positive or APC-positive cells per mm2, and each group is displayed with the median.
For analysis of apoptotic cells, brain sections were stained by TUNEL and DAPI. TUNEL-positive and DAPI-positive nuclei were counted in the cerebral cortex from representative levels one to five after visual verification of apoptotic hallmarks such as shrinkage and fragmentation, and the sum per field was calculated. Each group is displayed with the median.
Microglia and astrocytes were quantified by staining brain sections with anti-Iba1 and anti-GFAP antibodies, respectively. Iba1-positive or GFAP-positive cells were quantified in three fields per hemisphere at level four of five representative sections of each brain, and the mean was calculated, which is expressed as Iba1-positive or GFAP-positive cells per mm2. Each group is displayed with the median.
Stereological analysis of brain sections
Stereological quantification of neurons and oligodendroglia in mice was performed using the CellSenseTM software provided by Olympus. For experiments involving intrathecal injections, six fields (×60) in the cortex of five representative brain sections stained with NeuN or APC antibody per animal were analyzed by quantifying NeuN-positive or APC-positive objects and determining the percentage area of NeuN-positive or APC-positive region of interest (ROI), as indicated. The threshold was set to 400 dpi (NeuN+) or 150 dpi (APC+). For analysis of TUNEL-stained cells, whole cortices of five representative brain sections per animal were analyzed by quantifying TUNEL-positive objects and the percentage of TUNEL-positive region of interest (×20, threshold 20 dpi).
Measurement of fluorescence intensity in immunostained brain sections
For Iba1, GFAP, and MBP reactivity measures (mean fluorescence intensity per area) DAPI staining was used as reference. Mean fluorescence intensity of Iba1, GFAP, or MBP as assessed by Image J software was divided by the mean fluorescence intensity of DAPI from the same scan to assure technical accuracy. Representative brain slices containing the cortical area (20 μm) per animal were scanned using a confocal laser scanning microscope (Leica TCS SPE, Wetzlar, Germany). Z-stacks were analysed using Image J software by a person blind to the sample identity.
Real-time PCR
One brain hemisphere was homogenized in 1 ml TRIzol® with an Ultra-Turrax® at 21500 rpm for 30 sec. The homogenate was centrifuged at 12000 x g and 4°C for 15 min. DNA was removed with RQ1 RNase-free DNase and UltraPure™ phenol:chloroform:isoamyl alcohol. For synthesis of cDNA from 1 μg RNA, random hexamers were used with MMLV-RT (Promega, Mannheim, Germany). SYBR® Green-based quantitative real-time PCR was performed with the RT
2 qPCR Primer Assays (SABiosciences Corporation, Frederick, USA) according to the manufacturer’s manual with the RT
2 Real-Time™ PCR protocol (#1) and the ABI7500 default dissociation stage. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a housekeeping gene for control. For statistics dCT values (CT
GOI-CT
HKG) were log2-transformed according to [
66]. Fold exchange (2
-(ddCT)) relative to the control was calculated using the median of each group. P values and stars indicate statistical significance, fold change values <0.5 and >2 indicate biological significance.
Cytokine bead assay
Cytokine bead assay was performed with the FlowCytomix System (eBioscience, Frankfurt, Germany) according to the manufacturer’s manual. One half of a brain per mouse was lysed in 50 mM Tris pH 7.4 with cOmplete Ultra Tablet (Roche, Basel, Switzerland) with an Ultra-Turrax (5 section at 17500 rpm) on ice. Lysates were centrifuged at 10000 x g, 4°C for 30 min. Protein levels in supernatants were measured by BCA Protein Assay (Pierce, Rockford, USA).
Analysis of NO content
The amount of nitric oxide in one brain hemisphere was determined by the standard Griess’ reaction, as previously described [
17].
HSP60 and Il-1β ELISA
Amounts of HSP60 in the CSF of mice and Il-1β in murine brain lysates were determined by ELISA according to the manufacturer’s respective manuals (Enzo Life Sciences, Lörrach, Germany, and BD Biosciences, Heidelberg, Germany, respectively).
Western blot
Lysates (10 μg of protein) of primary neurons, microglia, and Oli-neu cells or lysates of brain hemispheres of MCAo-treated and sham-operated mice, as indicated, were applied to a 10% SDS-PAGE gel. HSP60 and neuronal nuclei were detected by immunoblotting using a mouse monoclonal antibody against HSP60 (Enzo Life Sciences, Lörrach, Germany) and neuronal nuclei (NeuN, Merck Millipore, Darmstadt, Germany), respectively. The mouse monoclonal antibody against synaptophysin was generously provided by Dr. M. Hoeltje (Charité-Universitaetsmedizin Berlin, Germany). Protein signals were visualized by enhanced chemiluminescence detection, as described previously [
17], and intensities of the signals were quantified by Image J software.
Statistical analysis
Data are expressed as indicated in the figure legends. Statistical differences between treatment groups were determined using Mann–Whitney U test, one-way and two-way ANOVA, or Kruskal Wallis test, as indicated. Differences were considered statistically significant when p < 0.05.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SL designed research, analyzed data, and wrote the manuscript. ES, AM, and HK designed research and wrote the manuscript. KR and SAW designed research, performed research, and analyzed data. KD, PD, OE, and CK performed research and analyzed data. All authors read and approved the final manuscript.