Ganglioside deficiency causes inflammation and neurodegeneration via the activation of complement system in the spinal cord

The generation of KO mice of GM2/GD2 synthase [5] and GD3 synthase [8] was previously reported. To generate DKO mice efficiently, we also mated female homozygotes of the GD3 synthase gene and male heterozygotes of the GM2/GD2 synthase gene [9]. DKO mice with the two genes were designated as DKO (or GM3-only) mice, and wild-type mice for both genes were presented as WT mice. C3-KO mice (B6.129S4-C3 ^tm1Crr /J) were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). To generate triple KO (TKO) mice lacking GM2/GD2 synthase, GD3 synthase, and C3, DKO mice of GM2/GD2 synthase and GD3 synthase genes were mated with C3 KO mice, and genotypes of the offspring were screened for the three genes. Genotypes in C3-KO mice were screened as described [19]. WT and homozygous mutant mice of 4, 8, 15, 28 to 32, 42 to 50 weeks, and over 60 weeks after birth were used in this study. All experimental protocols were approved by the animal experimental committee of the Graduate School of Medicine in Nagoya University in accordance with the guidelines of Japanese government and were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals (1966). The number of animals used and their suffering were minimized.

Antibodies used for Western immunoblotting were as follows: monoclonal anti-mouse C1q (rat IgG1) (Hycult Biotech, Uden, The Netherlands), monoclonal anti-β-actin (mouse IgG) (Sigma-Aldrich, St. Louis, MO, USA), or monoclonal anti-porcine glial fibrillary acidic protein (GFAP) (mouse IgG1) (Chemicon, Temecula, CA, USA). Chemiluminescence detection was performed by using horseradish peroxidase (HRP)-conjugated rabbit anti-rat IgG (H + L) (Zymed Laboratories, now part of Invitrogen, Carlsbad, CA, USA) and sheep anti-mouse IgG (Amersham Biosciences, now part of GE Healthcare, Little Chalfont, Buckinghamshire, UK). Antibodies used for immunohistochemistry were as follows: monoclonal anti-porcine GFAP (mouse IgG1) (Chemicon) and monoclonal anti-mouse F4/80 (rat IgG2b) (Serotec, Oxford, UK). Immunofluorescence detection was performed by using Alexa Fluor 488-goat anti-rat IgG2b (Invitrogen) or Alexa Fluor 555-goat anti-mouse IgG1 (Invitrogen).

Primers used for real-time reverse transcription-polymerase chain reaction (RT-PCR) were designed according to Primer 3 Input™ [20] as shown in Additional file 1: Table S1.

Glycolipid extraction and thin-layer chromatography (TLC) were performed as described previously [21]. Briefly, lipids were extracted by chloroform/methanol at ratios of 2:1, 1:1, and then 1:2 sequentially. Glycolipids were isolated by a Florisil column after acetylation, and then neutral and acidic fractions were separated by DEAE-Sephadex (A-50) column chromatography.

The footprint test is performed by applying Chromacryl (Chroma Acrylics, Lititz, PA, USA) to mouse feet. Usually, a red color is used on the forelimb and a blue color for the hind limb. Mice were then placed within a restricted cardboard tunnel (80 cm long × 10 cm wide × 10 cm high) with a white paper-covered floor, and footprints were made while the animal walked.

Response to mechanical stimuli was examined by the von Frey test. To determine the paw withdrawal threshold to mechanical stimulation, the plantar surface of the hind paws in the WT and DKO mice was stimulated with a set of calibrated nylon monofilaments (von Frey hairs) with increasing force until the mice withdrew the limb. The minimum intensity of mechanical stimuli was taken as the force at which the mouse withdrew the paw.

WT and DKO mice were perfused with phosphate-buffered saline (PBS) and then with 10% neutral buffered formalin (Wako, Osaka, Japan). Then the spinal cord was removed and was post-fixed with 10% neutral buffered formalin followed by Klüver-Barrera (KB) staining. The spinal cord was cut into lumber spinal cord segments L3-L4 and embedded in paraffin after dehydration and paraffin penetration. Blocked spinal cord was cut into 5 μm of cross-sections, and slices of spinal cord were stained with luxol fast blue solution and cresyl violet solution (KB staining). KB-stained slices of the spinal cord were observed by light microscopy.

WT and DKO mice were perfused with PBS. The spinal cords were isolated from mice and homogenized in Trizol™. Total RNA was extracted from tissues by Trizol™ in accordance with the protocol of the manufacturer. After the quantity and purity of RNAs were checked by determining absorbance at 260/280 nm by using a spectrophotometer, total RNA was reverse-transcribed into cDNA by using M-MLV Reverse Transcriptase™ (Invitrogen) and oligo dT primer (Sigma-Aldrich). cDNA was frozen at −80°C until use.

Real-time RT-PCR was performed by using 4 ng cDNA per well, with F-400 or F-410 in a SYBR green qPCR kit™ (Finnzymes, Espoo, Finland) and a Thermal Cycler PTC-200™ (MJ Research, now part of Bio-Rad Laboratories, Inc., Hercules, CA, USA). The PCR conditions were as follows: preheating at 95°C for 10 minutes, 40 cycles of 95°C (10 seconds), 60°C (20 seconds), and 72°C (20 seconds). The plate reader was set at 78°C or 75°C (2 seconds) depending on individual primer pairs. For the internal control in the quantitative analysis, mGAPDH was used. Every sample was measured in duplicate, and the gene expression levels were analyzed by using Opticon Moniter3™ software (Bio-Rad Laboratories, Inc.).

The spinal cords were isolated after perfusion with PBS from mice and were homogenized in lysis buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 5 mM MgCl₂, 50 mM NaF, 1 mM NaVO₄, 1% Triton X-100, 200 mM PMSF, 0.01-0.02 TIU/mL aprotinin). The lysates were pelleted by centrifugation at 8,000 g for 60 minutes at 4°C. The supernatant was centrifuged at 18,000 g for 90 minutes at 4°C, and clarified lysates were used for immunoblotting. Lysates were frozen at −80°C until use.

Lysates from the spinal cord of WT and DKO mice were mixed with SDS sample buffer, heated at 100°C for 5 minutes, and resolved by a 10% to approximately 12% SDS-PAGE. Proteins were transferred onto polyvinylidene difluoride membranes by semi-dry electrophoresis for 60 minutes at 15 V. Membranes were incubated in PBST (PBS containing 0.1% Tween 20) with 5% skim milk for 2 hours at room temperature or overnight at 4°C, then incubated with primary antibodies for 1 hour at room temperature or overnight at 4°C, followed by HRP-conjugated second antibodies for 45 minutes at room temperature. Bands were visualized by ECL™ (Western Lightning Chemiluminescence Reagent; PerkinElmer Life Sciences, Waltham, MA, USA).

WT and DKO mice were perfused with PBS and then 4% paraformaldehyde in 0.1 M phosphate buffer. The spinal cord was removed, post-fixed with 4% paraformaldehyde in 0.1 M phosphate buffer overnight at 4°C, then replaced with 10% sucrose for 6 hours, 15% sucrose for 6 hours, 20% sucrose for 6 hours, and embedded in OCT compound (Sakura Finetechnical, Nagoya, Japan) and frozen in liquid nitrogen. Frozen sections were set at longitudinal orientation for the spinal cord and were cut into 7 μm-thick sections on a cryostat (CM3050S; Leica, Wetzlar, Germany) at −20°C. The sections were placed on MAS-coat grids (Matsunami Glass, Osaka, Japan), and tissue slices were stored at −80°C until used.

For immunohistochemistry, tissue slices were washed twice with PBS for 5 minutes at room temperature, blocked with 10% normal goat serum for 1 hour at room temperature, and incubated with primary antibodies for 1 hour at room temperature or overnight at 4°C. Then slices were washed 3 times with PBS for 5 minutes at room temperature, and incubated with Alexa Fluor-conjugated second antibodies for 45 minutes at room temperature. Then they were washed 3 times with PBS for 5 minutes at room temperature, and were sealed with PermaFlour™ aqueous mounting medium (Thermo Fisher Scientific Inc., Waltham, MA, USA). Cover slips were mounted and observed by fluorescence microscopy.

After perfusion with PBS, the spinal cords were isolated from mice and homogenized and sonicated in lysis buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 5 mM MgCl₂, 50 mM NaF, 1 mM NaVO₄, 1% Triton X-100, 1 mM PMSF, 0.01-0.02 TIU/mL aprotinin) (spinal cord 100 mg tissue/300 μL lysis buffer). The lysates were centrifuged at 8,000 g for 60 minutes at 4°C, and the supernatants were centrifuged at 18,000 g for 90 minutes at 4°C. Five hundred μg/50 μL of tissue lysates was used for the assay. Levels of interleukin-1-alpha (IL-1α), IL-1β, and tumor necrosis factor-alpha (TNFα) proteins were determined by using assay kits for mouse IL-1α/IL1F1™, mouse IL-1β/IL-1 F2™ enzyme-linked immunosorbent assay (ELISA), or mouse TNFα/TNFF1A™ (R&D Systems, Minneapolis, MN, USA). The assay was performed in accordance with the instructions of the manufacturer.

All results were presented as the mean ± standard deviation and were initially subjected to the Bartlett, Hartley, and Levene test for homogeneity of variance. The Shapiro-Wilk test was used to verify that the data followed a normal distribution. Statistical significance was calculated by using Student t test in cases of two comparison groups and one-way analysis of variance (ANOVA) followed by Tukey-Kramer post hoc test in cases of more than two comparison groups. Two-way ANOVA followed by Bonferroni post hoc test was used comparing mouse group and time course. All statistical significances were set at *P <0.05, **P <0.01, ***P <0.001.

We generated DKO mice lacking the GM2/GD2 synthase and the GD3 synthase genes (Figure 1A) and investigated phenotypes of the DKO mice. Ganglioside composition in the spinal cord as analyzed by TLC exhibited only GM3 in DKO mice (Figure 1B). In the footprint test, DKO mice walked in a zigzag line (Figure 1C) and with abnormal stride (Figure 1D) whereas WT mice showed straight walking. These abnormal gates were exacerbated with aging in DKO mice (Figure 1D). Interaction of mouse group and time course did not turn out to be statistically significant (P = 0.2624). The differences among time courses in DKO mice (but not WT mice) turned out to be statistically significant (WT, P = 0.674; DKO, P = 0.0167) (Figure 1D). Sensory nerve functions in the 7-, 14-, 25-, 32-, and 42-week-old mice were examined with the von Frey test. Sensitivity to mechanical pain stimuli in the hind limbs apparently reduced at 25 weeks after birth or later (Figure 1E). Interaction of mouse groups and time courses showed P = 0.0311. The results of comparison among time courses in mice were as follows: WT, P = 0.00226; DKO, P = 0.00173 (Figure 1E). These results suggested that abnormal nerve functions occurred because of the lack of gangliosides in DKO mice. Furthermore, the spinal cords of 44-week-old DKO mice were thinner than those in WT mice (Figure 1F) and weighed less in the DKO mice both at 30 weeks (WT: 120.6 ± 7.8 mg, n = 11 versus DKO: 112.6 ± 4.9 mg, n = 10) and at 50 to 65 weeks after birth (WT: 143.6. ± 10.5 mg, n = 13 versus DKO: 120.0 ± 10.1 mg, n = 12) (Figure 1G).

To examine whether histopathological changes related to abnormal gait and reduced sensitivity to mechanical pain stimuli occur in the spinal cord of DKO mice, we performed KB staining for lumber spinal cord 3 (L3) of spinal cord from 44-week-old WT and DKO mice (Figure 2A). There was an apparent loss of motor neurons in the DKO mice as shown in spinal lamina IX (Figure 2A and B). The thickness of spinal lamina II and III decreased in DKO mice compared with WT mice (Figure 2C and 2D). The number of neuronal cells in spinal lamina II and that of marginal cells (in spinal lamina I) also decreased in DKO mice compared with WT mice (Figure 2D). These results suggested that abnormal phenotypes found in gait and sensitivity to mechanical pain stimuli were caused by neurodegeneration in the spinal cord of DKO mice.

To investigate the mechanisms by which neurodegeneration was caused in DKO mice, we performed a DNA microarray analysis with mRNAs from the spinal cords of 28- and 48-week-old mice. Among genes differentially expressed between WT and DKO mice, complement genes (C1qα, C4, and C3aR) were found to be generally upregulated in the spinal cord (Table 1). The expression level of C4 was markedly upregulated in both the 28- and 48-week-old DKO mice (30-week-old WT: 4.07 versus DKO: 10.27 and 48-week-old WT: 6.71.07 versus DKO: 14.09). The expression level of C3aR, receptor for anaphylatoxin C3a, was most strongly upregulated in the 48-week-old DKO mice (WT: 0.29 versus DKO: 1.44). To precisely determine the gene expression profiles, we performed real-time reverse transcription-PCR (RT-PCR) with mRNAs from individual mice. Consequently, the expression level of C4 was upregulated in DKO mice compared with WT at all ages examined (Additional file 1: Figure S1A). The expression level of C3aR was upregulated in all DKO male mice and many DKO female mice (Additional file 1: Figure S1B). These results indicated that C4 and C3aR were upregulated regardless of the presence of skin injury and more definitely in male DKO mice than in female. Since complement system functions in innate immunity and inflammatory reactions with sequential activation of diverse components, we performed real-time RT-PCR for complement system-related genes by using mRNAs from the spinal cord and liver of 28-week-old mice. The expression levels of C1qα, C4, C5, C3aR, and C5aR were significantly upregulated in the spinal cord of DKO mice (Figure 3A). The expression levels of C1qβ, C1qγ, and C3 in the spinal cord tended to increase in the DKO mice. The expression of C6, C7, C8β, and C9 could not be detected. These results suggested that the majority of complement-related genes were upregulated in the spinal cord of DKO mice. On the other hand, the expression levels of complement-related genes were generally higher in the liver than in the spinal cord, and C4 was persistently low in the liver of DKO mice compared with WT mice (Figure 3B). Thus, the expression of these genes in the central nervous system (CNS) seemed to be regulated in a distinct manner from that in the liver (Additional file 1: Figure S2).

Because the phenotypes of the DKO mice were exacerbated with aging, we performed real-time RT-PCR to examine chronological changes in the expression levels of C1qα, C3, C4, C3aR, and C5aR by using mRNAs from the spinal cord of 4-, 15-, 28-, and 48-week-old WT and DKO mice (Figure 4A). All of these genes showed no or minimal differences at 4 weeks after birth between WT and DKO mice. The expression levels of complement genes were then upregulated in DKO mice with aging. These results suggested that upregulation of complement genes took place along with neurodegeneration in the nervous tissues of DKO mice. Furthermore, we performed immunoblotting to investigate the protein levels of C1q in the spinal cord of the DKO mice. Since C1q proteins are essential in the complement system, we examined the C1q protein in the spinal cord of 15-week-old mice, showing equivalent levels between WT and DKO mice. But it increased 2.3-fold in 30-week-old DKO mice and just tended to be higher in 60-week-old DKO mice (Figure 4B), whereas its mRNAs were upregulated with aging. There results suggested that complement components were consumed with complement activation.

To investigate whether inflammation occurs in the spinal cord of DKO mice, we performed immunohistostaining of astrocytes (GFAP) and microglia (F4/80). We found no definite difference in GFAP staining of white matter between WT and DKO mice at 8 to about 60 weeks after birth (Figure 5A). On the other hand, GFAP-active astrocytes in gray matter was found more prominently in 8-, 15-, and 30-week-old DKO mice than in WT mice (Figure 5B), whereas no difference was found between WT and DKO mice in 60-week-old mice. The increased GFAP expression in DKO mice with aging was confirmed by immunoblotting (Figure 5C-E). GFAP bands increased at 15 to about 30 weeks after birth in DKO mice compared with WT mice but were almost equivalent at 60 weeks after birth between WT and DKO mice. Accumulation of microglia tended to increase at 30 weeks after birth and significantly accumulated hereafter, whereas no difference could be found at 8 to approximately 15 weeks after birth between DKO and WT mice (Figure 5F and G).

To investigate whether abnormal proliferation of astrocytes and accumulation of microglia are associated with actual inflammation, we measured mRNA expression levels of inflammatory cytokines (IL-1α, IL-1β, and TNFα) in the spinal cord of WT and DKO mice with real-time RT-PCR. Expression levels of IL-1α and IL-1β were significantly upregulated in 4- and 15-week-old DKO mice, and those of IL-1α and TNFα were significantly upregulated in 48-week-old DKO mice (Figure 6A). Expression levels of IL-6 and INFγ could not be detected in either type of mice. We then performed ELISA for the inflammatory cytokines. None of these inflammatory cytokines was detected at 30 weeks after birth, but they tended to increase at 60 weeks (Figure 6B). These results suggested that the spinal cord of DKO mice underwent inflammatory reaction, and it was exacerbated with aging.

To investigate whether the inflammatory reaction and neurodegeneration in DKO mice are due to complement activation and can be rescued by disruption of complement systems, we generated TKO mice by mating the DKO mice of GM2/GD2 synthase and GD3 synthase genes with C3 KO mice (Figure 7A). As expected, expression levels of C3 were null in the TKO mice (Figure 7B), and C1qα in TKO also reduced to levels almost equivalent to that in WT mice. By measuring mRNA expression levels of inflammatory cytokines (IL-1β and TNFα), expression of TNFα was downregulated in the TKO mice, whereas IL-1β tended to decrease when compared with that in DKO mice (Figure 7B). As for F4/80 immunohistostaining, accumulation of microglia completely became equivalent between TKO and WT mice (Figure 7C). These results indicated that inflammatory reactions observed in DKO mice almost subsided in TKO mice.

To clarify whether the neurodegeneration of the spinal cord in DKO mice could be restored by disruption of complement systems, L3 of the spinal cord in 44- to 50-week-old mice was stained by KB staining. As described in Figure 2, there was a clear loss of neurons in DKO mice compared with WT mice. The number of motor neurons (spinal lamina IX cells) and neuronal cells in the spinal lamina II was fairly restored in TKO mice, but thicknesses of spinal lamina II + III and marginal cells (in spinal lamina I) were almost equivalent between DKO and TKO mice (Figure 8A and B). These results suggested that the inflammatory reaction and neurodegeneration observed in DKO mice were definitely alleviated in the TKO mice. However, neurodegeneration in the spinal cord could not be completely restored in the TKO mice compared with the WT mice.