Microglial Heparan Sulfate Proteoglycans Facilitate the Cluster-of-Differentiation 14 (CD14)/Toll-like Receptor 4 (TLR4)-Dependent Inflammatory Response

Background: Microglia, CNS-resident macrophages, release TNFα and IL1β following TLR4 activation by the bacterial endotoxin LPS.

Results: LPS-induction of TNFα, IL1β, and pro-CD14 is suppressed in heparanase-overexpressing primary microglia.

Conclusion: Microglial HSPGs facilitate a CD14-dependent pathway of TLR4 activation.

Significance: Heparanase remodeling of microglial HSPGs suppresses CD14-dependent TLR4 activation.

Abstract

Microglia rapidly mount an inflammatory response to pathogens in the central nervous system (CNS). Heparan sulfate proteoglycans (HSPGs) have been attributed various roles in inflammation. To elucidate the relevance of microglial HSPGs in a pro-inflammatory response we isolated microglia from mice overexpressing heparanase (Hpa-tg), the HS-degrading endoglucuronidase, and challenged them with lipopolysaccharide (LPS), a bacterial endotoxin. Prior to LPS-stimulation, the LPS-receptor cluster-of-differentiation 14 (CD14) and Toll-like receptor 4 (TLR4; essential for the LPS response) were similarly expressed in Ctrl and Hpa-tg microglia. However, compared with Ctrl microglia, Hpa-tg cells released significantly less tumor necrosis factor-α (TNFα), essentially failed to up-regulate interleukin-1β (IL1β) and did not initiate synthesis of proCD14. Isolated primary astroyctes expressed TLR4, but notably lacked CD14 and in contrast to microglia, LPS challenge induced a similar TNFα response in Ctrl and Hpa-tg astrocytes, while neither released IL1β. The astrocyte TNFα-induction was thus attributed to CD14-independent TLR4 activation and was unaffected by the cells HS status. Equally, the suppressed LPS-response in Hpa-tg microglia indicated a loss of CD14-dependent TLR4 activation, suggesting that microglial HSPGs facilitate this process. Indeed, confocal microscopy confirmed interactions between microglial HS and CD14 in LPS-stimulated microglia and a potential HS-binding motif in CD14 was identified. We conclude that microglial HSPGs facilitate CD14-dependent TLR4 activation and that heparanase can modulate this mechanism.

Introduction

Microglia, the resident macrophages in the central nervous system (CNS),³ are continually monitoring their surroundings for pathogens and tissue degeneration (1). Activated microglia assume a pro-inflammatory phenotype characterized by the release of cytokines, chemokines, and reactive oxygen species (2, 3). Many neurodegenerative disorders (e.g. Alzheimer disease) are accompanied by neuroinflammation, and it has been proposed that impaired regulation of pro-inflammatory microglia aggravates the severity of these conditions (4, 5). Astrocytes have multiple support functions in the CNS, but can also be triggered to produce pro-inflammatory cytokines (6, 7). However, the astrocyte inflammatory response often occurs after an initial microglial response, suggesting differences in the reaction potential of these cell types (8).

Heparan sulfate proteoglycans (HSPGs) are present in the extracellular matrix and as membrane spanning syndecans or glycophosphatidylinositol (GPI)-anchored glypicans on cell surfaces. They consist of a core protein with covalently attached heparan sulfate (HS) chains. HS is a linear polymer of alternating glucosamine and uronic (glucuronic or iduronic) acid residues that are sulfated at various positions (9). HSPGs often function as co-receptors to various cell-type specific receptors, for growth factors, morphogens, and chemokines (9). HS chains are fragmented by heparanase, a mammalian endoglucuronidase. The active enzyme is a heterodimer consisting of a 50- and an 8-kDa subunit (10), that are cleaved from 65 kDa proheparanase by cathepsin-L (11).

HS has been implicated in various mechanisms central to inflammatory events (12, 13). We recently demonstrated an essential role for HSPGs in macrophage-mediated CNS inflammation, using a transgenic mouse model that overexpresses heparanase (Hpa-tg) (14). The anti-inflammatory effect of heparanase overexpression was attributed to loss of vascular endothelial HS-chains, which are required to present chemokines from the inflamed tissue and in doing so direct transmigration of circulating monocytes into the CNS (see also Refs. 15, 16). Shedding of the HSPG syndecan-1 prevents amplification of endotoxin-induced TNFα and IL6 production by disconnecting HS-bound chemokines from the cell surface (17). These studies support the concept that cell-surface HSPGs are important targets of inflammatory mediators.

The Toll-like receptors (TLRs) are highly conserved pathogen detectors, with critical functions in innate immunity (18). The bacterial endotoxin lipopolysaccharide (LPS) is a well-established activator of TLR4. LPS first binds to cluster-of-differentiation 14 (CD14), which is GPI-anchored at the cell surface. LPS-bound CD14 interacts with TLR4, in complex with lymphocyte antigen 96 (MD2) (19). Subsequent NFκB-dependent signaling events induce IL1β and TNFα. CD14-dependent TLR4 activation is predicted in innate immune cells such as microglia. Importantly, LPS can also activate TLR4 independently of CD14 (20–22).

In the current study we explored whether LPS-driven inflammation in microglia and astrocytes is HSPG-dependent, using primary cell cultures established from Hpa-tg and Ctrl mice. Our results indicate that microglial HSPGs are integral to the CD14-dependent activation of TLR4, by which LPS induces up-regulation of TNFα and IL1β. These findings suggest a pro-inflammatory function for microglial HSPGs that can be modulated by heparanase.

Experimental Procedures

Heparanase Transgenic (Hpa-tg) Mice

The homozygous Hpa-tg mice express human heparanase preceded by a chicken heparanase signal peptide sequence, which promotes secretion of the enzyme (23). Constitutive expression of the heparanase construct is under control of the β-actin promoter (24). The chimeric enzyme construct was injected into fertilized eggs from C57BL/6 × Balb/c mice. The offspring were backcrossed at least 10 generations to a C57BL/6 background and C57BL/6 mice served as controls (Ctrl). All experiments involving mice were approved and carried out in accordance with the guidelines set by the local animal research ethics committee.

cDNA Preparation and Real-time Quantitative Polymerase Chain Reaction (qPCR)

RNA was isolated from mouse brain or primary cells using the E.Z.N.A. Total RNA kit (Omega bio-tek, Norcross, GA). cDNA was prepared using the iScript cDNA synthesis kit (Bio-Rad). Real-time quantitative PCR (qPCR) analysis of transcript expression was performed using SsoFast EvaGreen Supermix (Bio-Rad, Stockholm, Sweden) and the primer pairs listed in Table 1. Primers were designed using Primer Blast NCBI software and produced by Life technologies (Invitrogen, Stockhom, Sweden). Target transcript expression was quantified relative to the expression of β-actin (Actb) or glyceraldehyde 3-phosphate dehydrogenase (Gapdh) mRNA transcript expression in the same sample. Equal concentrations of microglial and astrocyte RNA were used to prepare cDNA and to subsequently perform qPCR. The average and standard deviation of the housekeeping gene's quantification cycle (Cq) for compared samples are stated in the figure legends.

TABLE 1.

Primers used for qPCR analysis of mRNA transcripts of interest

mRNA target (mouse)	5′-3′ forward strand	5′-3′ reverse strand
Interleukin 1β (Il1β)	TGCCACCTTTTGACAGTGATG	ATGTGCTGCTGCGAGATTTG
Tumor necrosis factor α (Tnfα)	AGCCGATGGGTTGTACCTTG	ATAGCAAATCGGCTGACGGT
Cluster of differentiation 14 (Cd14)	GAAGCAGATCTGGGGCAGTT	CGCAGGGCTCCGAATAGAAT
Toll-like receptor 4 (Tlr4)	TCTGGGGAGGCACATCTTCT	AGGTCCAAGTTGCCGTTTCT
Lymphocyte antigen 96 (Md2)	TTGCCGAAGCGTAAGGAAGT	TCACAGTCTCTCCTTTCAGAGC
β-Actin (Actb)	AAGAGCTATGAGCTGCCTGA	TACGGATGTCAACGTCACAC
Glyceraldehyde 3-phosphate dehydrogenase (Gapdh)	GGTGAAGGTCGGTGTGAAC	AATGAAGGGGTCGTTGATG

Purification of Primary Microglia and Astrocytes

Primary microglia were isolated as previously described (25, 26). Briefly, Hpa-tg and Ctrl neonates (<4 days old) were sacrificed by decapitation, and brains were isolated. The olfactory bulbs, hindbrain, and meninges were removed in chilled phosphate-buffered saline (PBS), under a dissecting microscope. One brain was triturated with a 1 ml pipette in 2 ml of 0.25% trypsin, EDTA (Sigma-Aldrich AB, Stockholm, Sweden) and then incubated (10 min) at 37 °C. Trypsin was inactivated with 5 ml of Dulbecco's modified Eagle's medium F12 (Sigma-Aldrich AB, Stockholm, Sweden), 10% fetal bovine serum (Cambrex AB, Karlskoga, Sweden), 1% Primocin (Sigma-Aldrich AB), referred to hereafter as culture medium (CM), and the suspension was homogenized by pipetting. A suspension of two brains was centrifuged at 1500 × g for 5 min at room temperature, and the resulting pellet was resuspended in CM, filtered through a 40 μm cell strainer and transferred to a 75 cm² culture flask (Corning Life Sciences, Stockholm, Sweden); CM was changed every 24 h. After 10–14 days in vitro, when bright amoeboid cells appeared on the surface of the confluent cell culture and in the medium, flasks were shaken overnight or tapped firmly. The cell suspension was collected and centrifuged at 1500 × g for 5 min; the resulting supernatant of glial-conditioned CM was filtered through a 0.2 μm syringe filter (VWR International AB, Stockholm, Sweden). The microglial pellet was resuspended and cultured in fresh CM substituted with 20% filtered glial-conditioned CM. These cultures were ∼100% positive for the microglial marker Iba-1 confirming high microglial purity. Primary astrocytes were isolated according to an adapted protocol (27). Briefly, the initial brain-derived glial mix was cultured in CM for 24 h after which the flasks were vigorously tapped (15–20 s) to remove easily detached cells and fresh CM was added. This was repeated every day for at least 5 days. On average, astrocytes isolated from 2 brains were confluent in a 75 cm² flask after 10 days in culture. Astrocytes were collected by trypsinization, counted and reseeded in CM. These cultures were >90% immunopositive for the astrocyte marker glial fibrillary acidic protein (GFAP; Abcam, Cambridge, UK).

Immunocytochemistry, Microscopy, and Image Analysis

Ctrl and Hpa-tg microglia or astrocytes were grown on poly-d-lysine coated (MatTek, Ashland, MA) slides for 24 h, washed, and then fixed (5 min) in 95% ethanol at −20 °C. Fixed cells were permeabilized in 1× PBS, 0.1% Tween, blocked (5–10 min) with Background sniper (Biocare Medical, Concord, CA) and then incubated overnight at 4 °C or at room temperature for 1 h with primary antibodies against ionized calcium-binding adapter molecule 1 of microglia (Iba-1; Abcam, Cambridge, UK), glial fibrillary acidic protein of astrocytes (GFAP; Abcam, Cambridge, UK), or heparanase (733, gift from Israel Vlodavsky). LPS-stimulated microglia were fixed in 4% paraformaldehyde, permeabilized in 1× PBS, 0.1% Tween, blocked in TNB blocking buffer (PerkinElmer Life Sciences, Waltham, MA) and immunostained with antibodies against CD14 (M305; Santa Cruz Biotechnology, Dallas, TX) and HS (10E4; Seikagaku Corporation, Kanagawa Japan). Antibodies were detected with species-specific fluorescently conjugated secondary antibodies (Alexafluor, Molecular Probes, Invitrogen). Fluorescence was visualized with a Nikon Eclipse 80i fluorescence microscope and captured with a Nikon DXM1200F digital camera operated with NIS 2.10 freeware. Confocal microscopy was carried out with a Zeiss LSM 710 instrument; images were acquired and analyzed using Zen imaging software. All images were processed with Photoshop software.

Characterization of Heparan Sulfate

Ctrl and Hpa-tg microglia (2 confluent 75 cm² flasks), and astrocytes (1 confluent 75 cm² flask) were cultured for 24 h in 8 ml CM/flask supplemented with 1000 μCi of Na₂³⁵SO₄. Cell medium was recovered and stored at −20 °C. Cells were lysed in 4 m urea, 50 mm Tris-HCl, pH 7.4, 0.25 m NaCl, 1% Triton-X100. The supernatant resulting from 20 min of 15,000-rpm centrifugation at 4 °C was used for HS chain analysis. Total protein concentration analysis of lysate samples confirmed that medium from Ctrl and Hpa-tg flasks had been conditioned by a similar number of cells. For purification of the ³⁵S-labeled HS, the lysates or medium were applied to a 1 ml diethylaminoethyl (DEAE)-Sephacel column (GE Healthcare, Uppsala, Sweden) that was equilibrated with 30 ml of 50 mm Tris HCl pH 7.4, 0.1% Triton X-100. The columns were then washed with 50 mm Tris-HCl pH 7.4 followed by 50 mm NaAc pH 4.5, 0.15 m NaCl, and subsequently eluted with 50 mm NaAc pH 4.5, 2 m NaCl. Eluted material was lyophilized, dissolved in a minimal volume of water and desalted on a PD10 column (GE Healthcare). Lyophilized samples were then treated overnight at 37 °C with 250 mU chondroitinase ABC (Seikagaku, Tokyo, Japan), followed by 2.5 h incubation with Benzonase nuclease (Novagen®, Merck, Stockholm, Sweden) to degrade chondroitin sulfate and DNA in the samples. Digests were next purified on a 0.5 ml of DEAE-Sephacel column as described above, desalted on a PD10 column and lyophilized. The products were dissolved in deionized water. Lysate digests were subjected to gel chromatography on a Superose 12 column (GE Healthcare), equilibrated with 50 mm Tris-HCl pH 7.4, 2 m NaCl, 0.1% Triton X-100. The eluates were collected as 0.5 ml fractions and ³⁵S-HS/³⁵S-HSPG radioactivity was analyzed by scintillation counting on a Beckman Coulter instrument. Medium digests were subjected to the same analysis, but first ³⁵S-HS was detached from core proteins by alkaline β-elimination. This was achieved by overnight incubation on ice in 0.5 m NaOH. Samples were neutralized with 0.5 m HCl prior to gel chromatography.

Inflammatory Stimulation of Microglia and Astrocytes

50 × 10³ microglia (Ctrl and Hpa-tg) or astrocytes (Ctrl and Hpa-tg) were seeded per well of 96-well cell culture plates (Invitrogen, Stockholm, Sweden). Before treatment, cells were washed in serum-free CM. The lipopolysaccharide (LPS) L2280, derived from O55:B5 Escherichia coli (Sigma-Aldrich), was diluted in serum-free CM to give the indicated final treatment concentrations. Cells were incubated with LPS for the indicated periods of time after which cell medium was collected and cells were lysed in CelLytic^TM M lysis reagent (Sigma-Aldrich) containing Complete Protease Inhibitor Mixture (Roche). Prior to analysis medium and lysates were stored at −20 °C in siliconized plasticware.

Enzyme-linked Immunosorbent Assay (ELISA) for Interleukin-1β (IL1β) and Tumor Necrosis Factor-α (TNFα)

IL1β and TNFα concentrations in CM were quantified by Quantikine® ELISA kits (R&D Systems, Minneapolis, MN) in accordance with the manufacturer's instructions. Optical density at 450 nm (minus optical density at 540 nm) was determined with a SpectraMAX 190 spectrophotometer and analyzed with SOFTMax Pro and Microsoft Excel software. Cytokine concentrations (pg/ml) were calculated from standard curves prepared with recombinant murine IL1β or TNFα (provided with the Quantikine® ELISA kits).

Western Blotting

Cell lysates or brain homogenates were denatured and reduced in Laemmli buffer (100 °C, 5 min) and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on Tris-Tricine gels (Invitrogen). Proteins were transferred to nitrocellulose or polyvinylidene difluoride membranes (Bio-Rad), which were then blocked (1 h) in 5% nonfat dry milk (Bio-Rad), 1× Tris-buffered saline (TBS), 0.1% Tween-20 (Sigma Aldrich) or Odyssey blocking buffer (LI-COR, Bioscience, Lincoln, NE). Membranes were incubated overnight at 4 °C with antibodies against heparanase (733 or 1453), IL1β (ab9722; Abcam, Cambridge England), CD14 (M305; Santa Cruz, Heidelberg, Germany), TLR4 (sc293072; Santa Cruz Biotechnology), MD2 (ab24182; Abcam), IkBα (sc-371; Santa Cruz Biotechnology), β-actin, or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and were then washed with TBS, 0.1% Tween-20. After subsequent incubation with the appropriate secondary antibody, immunosignals for 733 (heparanse), IL1β and CD14 (Fig. 3) were generated with SuperSignal West Dura Chemiluminescent Substrate (Thermo Scientific, Roskilde, Denmark), followed by exposure and development on ECL Hyperfilm (GE Healthcare, Uppsala, Sweden). CD14, TLR4, 1453 (heparanase), IκBα, and MD2 immunosignals were detected with IRDye secondary antibodies (LI-COR, Lincoln, NE) and visualized using an Odyssey two-channel infrared scanner (LI-COR). Fluorescent CD14 and IκBα Western blot images were processed with the ImageJ rolling-ball background subtraction application. Band intensities were determined using ImageJ. All Western blots were processed using Photoshop.

FIGURE 3.

Impaired inflammatory response in Hpa-tg microglia. A and E, Ctrl and Hpa-tg microglia (50 × 10³ cells) were challenged with LPS (50 ng/250 μl) for the indicated periods of time. IL1β (A) and TNFα (E) in cell medium were quantified by ELISA (pg/ml). Bars represent the mean and standard deviation from triplicate samples. B, pro-IL1β in cell lysates was detected by Western blot (B). Ratios of pro-IL1β band intensities, corrected for GAPDH levels, were included for comparison; the 24 h Ctrl value is set to 1.0. C, Ctrl and Hpa-tg microglia were incubated (24 h) with (+) or without (−) LPS (100 ng/250 μl), and lysates were subjected to IL1β Western blot (C). Band intensities relative to GAPDH levels from triplicate samples were quantified using ImageJ gel analysis software and presented as arbitrary units (AU). D, F, and G, Ctrl and Hpa-tg microglia (100 × 10³ cells) were incubated without (−) or with (+) LPS (50 ng/500 μl) for 24 h. qPCR analysis of Il1β (D) and Tnfα (F) mRNA transcript expression was corrected based on expression of the housekeeping gene Gapdh and presented relative to transcript expression in non-stimulated Ctrl microglia (set to 1.0). The average quantification cycle (Cq) for Gapdh in Fig. 4, D–F was 20.93 ± 0.36 (± S.D.). Bars represent the mean and S.D. of biological duplicates. IκBα degradation in Ctrl and Hpa-tg microglial lysates was assessed by Western blot. Ratios represent IκBα band intensities corrected for β-actin levels, relative to the non-stimulated Ctrl microglia (G). Significant differences between LPS-stimulated Ctrl and Hpa-tg microglia was determined by Student's t test; ***, p < 0.001; **, p < 0.01.

Sequence and Structure Analysis of CD14

The mouse CD14 sequence (P10810) was retrieved from the Uniprot database. Potential heparin/HS-binding motifs bearing relevant similarity to reported consensus sequences (13, 28) were manually identified and alignment with the human CD14 sequence (P08571) was performed with the ClustalW2 tool available at EMBL-EBI. Mouse CD14 three-dimensional conformation co-ordinates, deposited by (29) as 1WWL, were retrieved from the RCSB Protein Data Bank. The structure was viewed, and the relevant features annotated using PyMol (PyMOL Molecular Graphics System, Version 1.3 Schrödinger, LLC).

Statistical Analyses

Data points are presented with the mean and standard deviation. Significant differences between samples were determined using Student's t test and labeled as follows *, p < 0.05; **, p < 0.01; and ***, p < 0.001.

Results

We have previously reported that the recruitment of inflammatory macrophages across the blood-brain-barrier of LPS-stimulated Hpa-tg mice is impaired due to truncation of endothelial HS, which is required for transmigration (14). The LPS-induced cerebral IL1β response was also attenuated in Hpa-tg mice (14). The aim of the present study was to assess the involvement of Hpa-tg microglia and/or astrocytes in the impaired inflammatory response to LPS.

Heparanase Overexpression in Microglia Yields Fragmented HS Chains

Primary microglia were isolated from Ctrl and Hpa-tg brain. Lysates were probed by Western blot with the anti-heparanase polyclonal antibody 733, which confirmed higher levels of the 50 kDa active subunit of heparanase in primary Hpa-tg microglia compared with Ctrl cells (Fig. 1A). Size exclusion chromatography of metabolically ³⁵S-labeled HS/HSPGs isolated from microglia lysates and ³⁵S-labeled HS isolated from media revealed relative accumulation of HS fragments in Hpa-tg cultures (Fig. 1, B and C), indicative of truncated HS chains on cell-surface HSPGs.

FIGURE 1.

Characterization of primary Hpa-tg microglia. A, Western blotting for the active 50 kDa subunit of heparanase (antibody 733) in Ctrl and Hpa-tg microglial lysates (A). B and C, metabolically labeled ³⁵S-HS/HSPGs were purified from Ctrl and Hpa-tg microglia lysates (B), while ³⁵S-HS in the microglial medium was further dissociated from core proteins by alkaline treatment (C). Samples were analyzed by gel chromatography on Superose 12. Fractions were analyzed for ³⁵S by scintillation counting. Arrow indicates elution position of a heparin 18-mer.

Lipopolysaccharide Induction of TNFα and IL1β Is Impaired in Hpa-tg Microglia

Ctrl and Hpa-tg neonate brains showed similar basal mRNA expression of Cd14, Tlr4 and Md2, key components of the LPS-signaling complex (Fig. 2A). Western blotting similarly revealed equal expression of TLR4 and MD2 proteins in Ctrl and Hpa-tg neonate brain tissue (Fig. 2B). These data thus point to the availability of CD14, TLR4, and MD2 LPS-detectors in the Hpa-tg as well as the Ctrl mouse models.

FIGURE 2.

Characterization of key LPS-detectors in Ctrl and Hpa-tg neonatal brain. A, relative expression of Cd14, Tlr4, and Md2 mRNA transcripts, corrected for the housekeeping gene Actb (β-actin), was assessed by real-time quantitative PCR (qPCR) in Ctrl and Hpa-tg neonate brains (n = 5) (A). B, Western blotting of Ctrl and Hpa-tg neonate brain homogenates (n = 5) with antibodies against heparanase (1453), TLR4, MD2, and β-actin (B).

To test whether microglial HSPGs are involved in pro-inflammatory signaling events primary microglia from Ctrl and Hpa-tg neonates were challenged over a 24-h period with LPS in serum-free CM. LPS interacts with CD14, which activates the TLR4/MD2 inflammatory pathway, which culminates in the release of TNFα and IL1β (5, 22). LPS essentially failed to induce IL1β and induced significantly less TNFα in Hpa-tg compared with Ctrl microglia as determined by cytokine-specific ELISAs (Fig. 3, A and E). According to recent reports TNFα release from rat microglia peaks after 6 h of LPS stimulation (30), and after 8 h in mouse microglia isolated through a protocol that selects adherent microglia (31) rather than the amoeboid microglia collected here (25). In our experiments TNFα release continued to increase up to 24 h of LPS stimulation (Fig. 3E) suggesting that the dynamics of microglia cytokine release are influenced by species differences and cell isolation procedure. To address the mechanism behind the suppressed cytokine release in Hpa-tg microglia the expression of pro-IL1β in microglial lysates was assessed by Western blot. After 2.5 h LPS-stimulation, up-regulation of pro-IL1β (31 kDa band) was detected in Ctrl microglia, along with additional 28 kDa (32) and 17 kDa components. These additional proteins are caspase-1 cleavage products, indicative of inflammasome activity (33), the latter being the mature active form of IL1β (Fig. 3B). Pro-IL1β was also prominent after 8 and 24 h of LPS exposure in Ctrl microglia, whereas only a weak pro-IL1β band was detected in Hpa-tg microglia after 24 h (Fig. 3B). The impaired induction of pro-IL1β in Hpa-tg microglia was separately confirmed following 24 h exposure to a higher LPS dose (Fig. 3C). Additionally, qPCR analysis of Il1β and Tnfα mRNA expression confirmed that induction of cytokine transcripts was impaired in LPS-stimulated Hpa-tg microglia (Fig. 3, D and F). Together these results indicate that the diminished cytokine response in Hpa-tg microglia is due to a loss of function that precedes cytokine transcription.

LPS-stimulation of the TLR4 signaling pathway results in proteasomal degradation of IκBα (34). IκBα-bound NFκB is precluded from nuclear translocation, where it otherwise promotes cytokine expression. IκBα degradation was evident in LPS-stimulated (24 h) Ctrl as well as Hpa-tg microglia (Fig. 3G), suggesting that the NFκB pathway is at least partly viable in Hpa-tg microglia.

Lipopolysaccharide Fails to Induce CD14 Synthesis in Hpa-tg Microglia

The LPS-receptor CD14 is GPI-anchored to microglial membranes (mCD14) (35), but can also be released as a soluble form (36). Newly synthesized CD14 is evident on Western blot due to its increased molecular weight caused by the presence of a proCD14 sequence (Fig. 4A) (37, 38). A mCD14 immunosignal at 55 kDa (35) was observed in Ctrl and Hpa-tg microglia at all early time points of LPS stimulation (Fig. 4B, 0–2.5 h). After 8 and 24 h of LPS stimulation a strong 60 kDa pro-CD14 band appeared in Ctrl microglia, but not in Hpa-tg microglia blots (Fig. 4B). LPS-induction of microglial CD14 is partly attributed to paracrine/autocrine signaling of released TNFα (39). Therefore, the failure of Hpa-tg microglia to upregulate CD14 is presumably secondary to their diminished TNFα response to LPS challenge. The impaired induction of Cd14 mRNA in Hpa-tg microglia was confirmed in a separate experiment, where Cd14 expression by LPS-stimulated Hpa-tg cells remained as low as in unstimulated Ctrl or Hpa-tg microglia (Fig. 4D). Importantly, Ctrl and Hpa-tg microglial lysates showed similar levels of TLR4, with and without LPS stimulation (Fig. 4C). Tlr4 and Md2 mRNA expression in Ctrl and Hpa-tg microglia was similar in unstimulated cells (Fig. 4, E and F) and LPS induced the same changes in expression for both cell types, namely a slight decrease in Tlr4 mRNA and a slight increase in Md2 mRNA (Fig. 4, E and F). These findings suggest that the diminished LPS-induced TNFα response in Hpa-tg microglia results in a failure to up-regulate CD14. Importantly, CD14, TLR4, and MD2 are similarly expressed in unstimulated Ctrl and Hpa-tg cells, confirming that these LPS-detectors should be equally available to Ctrl and Hpa-tg microglia.

FIGURE 4.

Hpa-tg microglia fail to synthesize CD14. A and B, de novo CD14 synthesis is evident on Western blot as an increase in molecular weight due to the presence of a proCD14 sequence; this peptide is removed in the mature GPI-anchored (mCD14) and soluble (sCD14) forms of the protein, illustrated in A. CD14 expression in Ctrl and Hpa-tg microglia after LPS stimulation (50 ng/250 μl) for the indicated time periods (B). Ratios represent the CD14 band intensities, adjusted for GAPDH levels. C–F, Ctrl and Hpa-tg microglia (100 × 10³ cells) were incubated without (−) or with (+) LPS (50 ng/500 μl) for 24 h. TLR4 expression was detected by Western blot and ratios represent the TLR4 band intensities corrected for β-actin levels (C). qPCR analysis of Cd14 (D), Tlr4 (E), and Md2 (F). The mRNA transcript expression was corrected based on expression of the housekeeping gene Gapdh and presented relative to transcript expression in non-stimulated Ctrl microglia (set to 1.0). As in Fig. 3 the average quantification cycle (Cq) for Gapdh in Fig. 4, D–F was 20.93 ± 0.36 (± S.D.). Bars represent the mean and S.D. of biological duplicates.

Astrocyte TNFα Response Is Not Affected by Heparanase Overexpression

Astrogliosis is a prominent feature of numerous neurodegenerative disorders and reactive astrocytes are capable of a pro-inflammatory response (40, 41). To assess whether the impaired inflammatory response in Hpa-tg microglia was replicated in astrocytes we again isolated primary cells from Ctrl and Hpa-tg neonates. Immunocytochemical analysis using the antibody 733 confirmed heparanase overexpression in Hpa-tg relative to Ctrl astrocytes (Fig. 5A). Size-exclusion chromatography of metabolically ³⁵S-labeled HS/HSPGs isolated from astrocyte lysates and ³⁵S-labeled HS isolated from medium was indicative of an increased degradation of HS chains in Hpa-tg astrocytes (Fig. 5, B and C), similar to the findings for microglia (Fig. 1, B and C). Ctrl and Hpa-tg astrocytes were challenged with LPS for 24 h and TNFα and IL1β released into the medium were measured by ELISA. Unlike microglia, Ctrl and Hpa-tg astrocytes released comparable amounts of TNFα in response to LPS (Fig. 5D; microglia data from Fig. 3E inserted for comparison). LPS-stimulated astrocytes failed to produce any detectable IL1β (data not shown).

FIGURE 5.

The LPS-induced TNFα response in astrocytes is unaffected by heparanase overexpression. A–C, elevated heparanase expression is detected in Hpa-tg astrocytes, compared with Ctrl, as demonstrated by immunostaining with antibody 733 (A). Nuclei are counterstained with DAPI (original magnification 200×). Heparanase immunostaining presents as granular inclusions in Hpa-tg astrocytes (A; right panel, original magnification 400×). Metabolically labeled ³⁵S-HS/HSPGs were purified from Ctrl and Hpa-tg astrocyte lysates (B), while ³⁵S-HS in the astrocyte medium was further dissociated from core proteins by alkaline treatment (C). Samples were analyzed by gel chromatography on Superose 12. ³⁵S in effluent fractions was detected by scintillation counting. D, Ctrl and Hpa-tg astrocytes (50 × 10³ cells) were incubated with 0 or 50 ng/250 μl of LPS for 24 h. TNFα (pg/ml) in the astrocyte medium was quantified by ELISA and compared with the values previously presented for Ctrl and Hpa-tg microglia (Fig. 3D, 50 × 10³ cells). Bars represent the mean and S.D. of quadruplicate (astrocytes) and triplicate (microglia) samples; ***, p < 0.001. E–H, Ctrl astrocytes (100 × 10³ cells) were incubated with (+) or without (−) LPS (50 ng/500 μl) for 24 h. Lysates were immunoblotted for CD14 and TLR4; untreated Ctrl microglial lysates were included as positive controls, comparable loading was confirmed with β-actin (E). Medium from Ctrl astrocytes treated with (+) or without (−) LPS were subjected to CD14 Western blot; non-stimulated Ctrl microglial lysates (Mg Lys) were included as positive controls (F). The expression of Cd14, Tlr4, and Md2 mRNA transcripts in Ctrl astrocytes with (+) or without (−) LPS stimulation was assessed by qPCR. The expression levels were corrected for the expression of the housekeeping gene Actb (β-actin) (G) and presented relative to the equivalent expression in non-stimulated Ctrl microglia (Mg⁻). The average quantification cycle (Cq) for Actb (β-actin) in Fig. 5G was 20.23 ± 0.67 (± S.D.). Bars represent the mean and S.D. of biological duplicates. Levels of Cd14, Tlr4, and Md2 mRNA expression in astrocytes and microglia are also represented relative to Tlr4 expression in each cell-type (H).

mCD14 Is Not Detected in Astrocyte Lysates or Medium

Ctrl astrocytes were incubated with (+) or without (−) LPS for 24 h. CD14 Western blot failed to reveal any mCD14 in Ctrl astrocyte lysates. Untreated Ctrl microglial lysates were included as positive controls for mCD14 expression (Fig. 5E, CD14 blot). A weak proCD14 band was detected in one of the LPS-treated astrocyte samples, suggesting that given sufficient stimulation CD14 synthesis can be induced in these cells (Fig. 5E, CD14 blot, lane 4). In contrast to CD14, the expression of TLR4 in astrocytes was similar to that observed in an equal number of microglia and was unaltered by LPS stimulation (Fig. 5E, TLR4 blot). No CD14 immunosignals were detected in Ctrl astrocyte medium, irrespective of LPS administration (Fig. 5F). These findings imply that the LPS-induced TNFα expression in astrocytes, which is substantially lower than that observed in an equal number of Ctrl microglia (Fig. 5D), is the result of CD14-independent TLR4 activation.

To further explore the expression of the LPS detectors in Ctrl astrocytes we performed qPCR analysis of Cd14, Tlr4, and Md2 mRNA transcripts, with or without LPS stimulation. Compared with unstimulated Ctrl microglia (Mg⁻, set as 1.0), unstimulated Ctrl astrocytes expressed less Cd14, Tlr4 and Md2 (Fig. 5G); however, while Tlr4 and Md2 transcript expression was ∼3-fold lower, Cd14 was >20-fold lower. LPS stimulation increased Cd14 and to some extent Md2, but not Tlr4 expression in Ctrl astrocytes (Fig. 5G). Finally, we established that the Cd14:Tlr4 ratio differs distinctly between microglia and astrocytes, in that Cd14 expression is 2-fold lower than Tlr4 in astrocytes, but 4-fold higher than Tlr4 in microglia (Fig. 5H). Together these findings are consistent with the concept that LPS-signaling in astrocytes occurs through CD14-independent TLR4/MD2 activation, while in microglia a greater cytokine response is observed due to additional CD14-dependent TLR4/MD2 activation. Given that the LPS-response is impaired in Hpa-tg microglia, but not in Hpa-tg astrocytes, we propose that microglial HSPGs facilitate CD14-dependent TLR4/MD2 signaling.

Heparan Sulfate Colocalizes with CD14 on LPS-stimulated Microglia

Ctrl microglia were stimulated with LPS and immunostained with antibodies against HS (10E4) and CD14 (M305). Confocal microscopy revealed that HS associated with CD14-positive clusters (Fig. 6A). Plotting the profile of HS and CD14 immunosignal intensities from single z-scans along a line revealed HS and CD14 co-localization within these clusters (Fig. 6B, a and b).

FIGURE 6.

Microglial CD14 and HS co-localize. A and B, LPS-stimulated (50 ng/500 μl) Ctrl microglia were immunostained for CD14 and HS (antibody10E4) and imaged by confocal microscopy. HS accumulated in the periphery of the cell in areas that were positive for clusters of CD14 (A). The image in A represents a maximum intensity projection of eight z-planes collapsed together (A, z1–8). The spatial distribution of CD14 and HS signal intensities under the lines labeled a and b, from a single z-plane (z5, indicated in the illustration), are presented as profile plots in a and b. (Original magnification, 630×; optical thickness of each z-plane = 1.2 μm, interval between z-planes = 1 μm).

Identification of a Potential HS-binding Motif in CD14

The microglia HS/CD14 interaction imaged in Fig. 6 implies that CD14 carries a HS-binding site. However, CD14 was not detected in a recent bioinformatics screen for potential heparin/HS-binding motifs in murine immune proteins (13). Davis et al. applied the consensus heparin/HS-binding motifs XBBXXBX and XBBXBX (where X = aliphatic/aromatic amino acids and B = basic amino acids)(28), as well as truncated variants thereof, to detect potential HS-binding domains in 42 plasma membrane proteins, including TLR4, which has previously been reported from in vitro experiments (42). By manually analyzing the murine CD14 sequence we identified the potential heparin/HS-binding motif XXBNXNXBBX (where n = neutral serine residues) in the leucine-rich repeat 2 region (Fig. 7, A–C). This motif lies beneath the LPS-binding domain of CD14 (Fig. 7A) and, based on available structural data (PDB ID: 1WWL (29), the basic residues protrude in a manner that could facilitate HS/heparin interactions (Fig. 7, A and B). Through sequence alignment we determined that human CD14 also contains a HS-binding motif (XBXXBXBBX) in this region (Fig. 7C).

FIGURE 7.

A potential HS-binding motif of CD14. A–C, based on comparisons with heparin/HS-binding consensus sequences a potential HS-binding motif was identified below the LPS-binding domain of murine CD14 (A). This motif is found in the leucine rich repeat 2 (LRR2) region. The basic, potentially HS-interacting, residues (B) are presented in blue, the aliphatic/aromatic residues (X) are in green and neutral residues (N) in the motif are represented by gray sticks (B). Similar HS-binding potential of the corresponding region of human CD14 was confirmed by sequence alignment. The structural representations were prepared in PyMol using the murine CD14 structure deposited at the RCSB Protein Data bank as 1WWL.

Discussion

Our results demonstrate that microglial HSPGs are required for a robust response to LPS via the CD14-dependent TLR4 pathway. The proposed function of pro-inflammatory microglial HSPGs and the rational supporting this model are illustrated in Fig. 8. Briefly, LPS-induced IL1β and TNFα release were dramatically suppressed in Hpa-tg microglia (Fig. 3), which carry shorter HS-chains than Ctrl cells. CD14, TLR4, and MD2 were equally expressed by unstimulated Ctrl and Hpa-tg microglia (Fig. 4), and therefore not a limiting factor in the Hpa-tg response. However, LPS-induction of CD14 was completely abrogated in Hpa-tg microglia (Fig. 4). In contrast to microglia, Ctrl and Hpa-tg astrocytes responded equally to LPS, as judged by release of TNFα (Fig. 5D). Previous reports have offered conflicting conclusions regarding the expression of CD14 in astrocytes (42–45). In particular, Tarassishin et al. recently reported that mouse astrocytes express CD14 (46). Notably, however, Tarassishin et al. passaged primary mouse astrocytes for 6-weeks prior to experiment. In contrast we isolate astrocytes based on adherence, do not passage the cells and perform experiments typically within 10 days of isolation. It is conceivable that these different approaches result in astrocyte cultures that, despite being GFAP-positive, differ in their expression patterns for other proteins, including CD14. Interestingly, like Tarassishin et al. (46) and Lehnardt et al. (45) we observed that LPS increased astrocyte expression of CD14 mRNA (Fig. 5G), albeit at low level compared with microglia (Fig. 5G). It thus seems reasonable to assume that under specific circumstances astrocytes are capable of expressing CD14. However, at the time-points investigated in this study neither mCD14 nor soluble CD14 could be detected in astrocytes (Fig. 5, E and F). LPS can induce inflammatory cytokines via CD14-dependent and CD14-independent TLR4 activation (20–22). Therefore, we attribute the LPS-induced TNFα response in astrocytes largely to CD14-independent TLR4 activation. Considering that heparanase overexpression had no effect on TNFα induction in astrocytes (Fig. 5D), we propose that this CD14-independent TLR4 activation does not rely on cell surface HSPGs. Equally, due to the loss of cell surface HS in Hpa-tg microglia their CD14-dependent cytokine response is defective; therefore, as in astrocytes, the residual TNFα response detected in Hpa-tg microglia is attributed to CD14-independent TLR4 activation, which does not require HSPGs.

FIGURE 8.

Microglial HSPGs facilitate CD14-dependent TLR4 activation. LPS stimulates the maximal IL1β and TNFα response, and induces pro-CD14 via a combination of CD14-dependent and CD14-independent TLR4 signaling in Ctrl microglia. Heparanase overexpression truncates cell surface HS and impairs the ability of HSPGs to facilitate CD14-dependent TLR4 activation, thus suppressing the IL1β, TNFα, and proCD14 response to LPS (Hpa-tg microglia). Astrocytes lack CD14; therefore the LPS-induced TNFα release is attributed to CD14-independent TLR4 activation, which does not require astrocyte HSPGs (Hpa-tg/Ctrl astrocytes).

These considerations imply that HS on the microglial surface is required to facilitate CD14-dependent TLR4 activation, in particular, IL1β up-regulation. In accord with this conclusion, microglial HS was found co-localized with CD14 in LPS-stimulated microglia (Fig. 6), and a potential HS-binding motif was identified in the CD14 sequence (Fig. 7).

Details of the downstream signaling pathway affected by heparanase overexpression remain unclear. IκBα degradation was observed in Hpa-tg as well as in Ctrl microglia following 24 h of LPS treatment (Fig. 3G). However, the interpretation of these results is complicated by the fact that NFκB activation initiates cycles of IκBα degradation, NFκB translocation, followed by NFκB induced up-regulation of IκBα (47). The accumulated activation of the NFκB pathway is therefore not readily assessable. Moreover, the NFκB pathway may be intact in Hpa-tg microglia, while other transcription effectors (e.g. activator protein 1; AP1) may be affected by the loss of HS function. Further studies will be necessary to elucidate which specific signaling pathways are impaired in Hpa-tg microglia resulting in the diminished cytokine response.

Contrary to the anti-inflammatory effects of microglial heparanase overexpression presented here, several recent studies have implicated heparanase as a promoter of inflammation. In an experimental model of ulcerative colitis, the chronic inflammatory state was attributed to macrophage production of TNFα and cathepsin-L, which respectively induced and activated heparanase derived from the colon epithelium. Further, exogenous addition of recombinant heparanase to peritoneally derived macrophages increased their LPS-induced TNFα response (48). Importantly, this effect was dependent on catalytic activity of the enzyme, and Lerner et al. (48) proposed that heparanase relieves a HS-mediated inhibition of LPS-induced TLR4 signaling, supporting previous work by Brunn et al. (49). In accord with these findings Blich et al. reported loss of the pro-inflammatory activity of recombinant heparanase in peritoneal macrophages isolated from TLR2 and TLR4 knock-out mice (50). In a related context HS has also been considered an endogenous TLR4 agonist (51). Moreover, heparanase and soluble HS fragments were recently found to induce pro-inflammatory cytokines in human peripheral blood mononuclear cells and mouse splenocytes (52). It should be noted that contrary to the studies referenced above, where exogenous heparanase is added to cell cultures, we derived microglia from Hpa-tg mice that endogenously overexpress heparanase; therefore a direct comparison between the results is not readily made. Furthermore, the sulfation patterns of HS chains are cell-type specific (53) such that a microglial-specific HSPG may be specialized in supporting LPS-induced CD14-dependent TLR4 activation. Recently we reported lower levels of cerebral IL1β in Hpa-tg mice than in Ctrl mice in response to intraperitoneal LPS injection. The difference was primarily attributed to decreased recruitment of blood borne monocytes (14). The present study extends our understanding of these findings to include defective cytokine up-regulation in Hpa-tg microglia. Clearly, heparanase can promote pro- and anti-inflammatory states.

We have previously reported that the Alzheimer disease (AD) amyloid-β peptide elevates the levels of specific cell surface microglia HSPGs (26). In light of the present study, it is conceivable that amyloid-β induction of microglia HSPGs would enhance a pro-inflammatory phenotype and sustain an inflammatory state in the diseased brain. Elucidating the context- and cell-specific inflammatory effects of heparanase in relation to cell-surface and soluble HS will increase our understanding of the various pathological conditions in which heparanase overexpression has been reported (48, 54, 55). The present study highlights the ability of heparanase to attenuate inflammatory states relevant to the CNS through its capacity to fragment microglial HSPGs.