Tlr4 Deletion Modulates Cytokine and Extracellular Matrix Expression in Chronic Spinal Cord Injury, Leading to Improved Secondary Damage and Functional Recovery

Abstract

Toll-like receptors (TLRs) play an important role in the innate immune response after CNS injury. Although TLR4 is one of the best characterized, its role in chronic stages after spinal cord injury (SCI) is not well understood. We examined the role of TLR4 signaling in injury-induced responses at 1 d, 7 d, and 8 weeks after spinal cord contusion injury in adult female TLR4 null and wild-type mice. Analyses include secondary damage, a range of transcriptome and protein analyses of inflammatory, cell death, and extracellular matrix (ECM) molecules, as well as immune cell infiltration and changes in axonal sprouting and locomotor recovery. Lack of TLR4 signaling results in reduced neuronal and myelin loss, reduced activation of NFκB, and decreased expression of inflammatory cytokines and necroptotic cell death pathway at a late time point (8 weeks) after injury. TLR4 null mice also showed reduction of scar-related ECM molecules at 8 weeks after SCI, accompanied by increase in ECM molecules associated with perineuronal nets, increased sprouting of serotonergic fibers, and improved locomotor recovery. These findings reveal novel effects of TLR4 signaling in chronic SCI. We show that TLR4 influences inflammation, cell death, and ECM deposition at late-stage post-injury when secondary injury processes are normally considered to be over. This highlights the potential for late-stage targeting of TLR4 as a potential therapy for chronic SCI.

Keywords: Toll-like receptor; cell death; chronic inflammation; cytokine; extracellular matrix molecules; locomotor recovery; serotonergic fiber sprouting; spinal cord injury.

Figure 1.

Improved neuronal survival, reduced myelin and oligodendrocyte loss, and changes in cell death pathways in chronic SCI in TLR4 KO mice. A–C, Changes in mRNA (A; n = 6 for all groups) and protein (B,C; n = 4 for all groups) in RIPK3 at 7 d and 8 weeks post-SCI in wild-type and TLR4 null mice. Note the significantly lower mRNA and phospho-RIPK3 expression in TLR4 null mice compared with wild-type mice at 8 weeks. Panel C shows Western blot. D–F, Changes in mRNA (D; n = 5–6 for all groups) and protein (E,F; n = 3–4 for all groups expression of MLKL at 7 d and 8 weeks post-SCI in wild-type and TLR4 null mice). Note again the significantly lower mRNA and phospho-MLKL protein expression in TLR4 null mice compared with wild-type mice at 8 weeks. Each group normalized to its own controls. Panel F shows Western blot. G, Changes in mRNA expression of NLRP3 at 7 d and 8 weeks post-SCI in wild-type and TLR4 null mice. Note the significantly increased expression in wild-type mice at both time points, but no changes in TLR4 null mice (n = 6 for all groups). H, Quantification of Western blots shows no changes in expression of cleaved phospho-gasdermin D (GSDMD) at either time points in both genotypes (n = 3–4 for all groups). Panel I show Western blot. J, Micrographs showing improved survival of NeuN⁺ neurons (500 µm rostral to lesion) in TLR4 null mice compared with WT mice 8 weeks after SCI. K, Quantification shows greater neuronal survival on either side of the lesion epicenter in TLR4 KO mice at 8 weeks post-SCI as detected by NeuN staining. The red boxplots show the values obtained from uninjured TLR4 null and wild-type mice (n = 6 for all groups). L, Micrographs of LFB staining shows reduced myelin loss (400 µm caudal to lesion) in TLR4 null mice compared with WT mice 8 weeks after SCI. M, Quantification of LFB staining shows greater myelin on either side of the lesion epicenter in TLR4 KO mice. The red boxplots show the values obtained from uninjured TLR4 null and wild-type mice (n = 5 for all groups). N, CC1 staining of oligodendrocytes in the dorsal column of the spinal cord in WT and TLR4 null mice. O, Note increase in CC1⁺ oligodendrocytes in TLR4 null mice in the dorsal column white matter of the spinal cord compared with wild-type mice. The red boxplots show the values obtained from uninjured TLR4 null and wild-type mice (n = 5 for all groups). One-way ANOVA with post hoc Tukey’s multiple-comparisons test (A,B,D,E,G,H); p < 0.0001 (A,D,G); p = 0.0001 (B); p = 0.001 (E); p = 0.807 (H). Two-way RM-ANOVA; genotype effect, with post hoc Bonferroni’s multiple-comparisons test (K,M,O); p = 0.008 (K); p = 0.0017 (M); p = 0.002 (O); *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 compared with uninjured naive level. ^#p ≤ 0.05; ^##p ≤ 0.01; ^###p ≤ 0.001 comparing the two injured genotypes. Scale bars: J, L, 200 µm; N, 100 µm.

Figure 2.

Immunostaining showing TLR4 expression at 7 d and 8 weeks after SCI and reduced expression of necroptosis marker (p-MLKL) at 8 weeks post-SCI. A, TLR4 immunostaining of cross sections of the spinal cord of uninjured (naive; Ai), 7 d (Aii), and 8 weeks (Aiii) after SCI. Note the increase in expression at 7 d which is further increased at 8 weeks. Note also in the 7 d panel (Aii) the elongated profiles of astrocyte-like cells in the lateral and ventrolateral white matter and rounded profile of macrophage-like cells in the dorsal region that sustains more damage. No staining was seen in negative controls (data not shown). The graph (Aiv) shows quantification of densitometric data of TLR4 staining (n = 3–4 for all groups); one-way ANOVA with post hoc Tukey’s multiple-comparisons test; p < 0.0001.) B, Double immunofluorescence staining of spinal cord sections labeled with TLR4 (green) and either GFAP, CD11b, or NeuN (all red) for animals 7 d and 8 weeks after SCI. Note that the merged images show expression of TLR4 increases in astrocytes between 7 d (Biii) and 8 weeks (Bxii; ventrolateral white matter). TLR4 is expressed in CD11b⁺ macrophages at 7 d (Bv,vi) but not 8 weeks (Bxiii–xv). At the latter time point, note the TLR4 staining of CD11b-negative astrocyte-like profiles is seen (Bxv; dorsal region of the spinal cord). There is no expression of TLR4 in NeuN⁺ neurons (dorsal gray) at both time points (Bvii–ix,xvi–xviii). The rounded profiles outlined with weak TLR4 staining (arrows in Bvii) are NeuN negative (Bix) indicating that these are likely to be profiles of blood vessels showing weak labeling of endothelial cells or astrocytes. Note, however, in the merged image (Bxviii) that TLR4 staining (green) is seen in the white matter along the right side. Sections for GFAP and CD11b taken close to the epicenter of the injury while for NeuN sections taken 500 µm caudal to the injury. C, Double immunofluorescence staining of the dorsal horn region [Ci–iii (WT) and Civ–vi (KO)] and ventral horn region [Cvii–ix (WT) and Cx–xii (KO)] labeled with p-MLKL (green) and NeuN (red) and merge (yellow) at 8 weeks post-injury. Note that these merged images show increased expression of p-MLKL (yellow) in NeuN+ neurons in the dorsal horn (DH) and ventral horn (VH; arrows). This labeling is stronger in WT mice after SCI but shows sparse expression in NeuN⁺ neurons in TLR4 null (KO) mice. Scale bars: A, 100 µm; B, C, 50 µm.

Figure 3.

TLR4 null mice exhibit differential transcriptional responses at 7 d and 8 weeks post-SCI. 2D PCA (A) and 3D PCA (B) of bulk RNASeq filtered counts (17,878 sequences). PCA shows treatment separation along PC1, time point post intervention (PC3), and WT versus TLR4 KO samples (PC2). C, Summary of (sequence) pairwise differential expression analysis with six group design matrix between TLR4^−/− and WT mice. Only differentially expressed sequences with adjusted p value (BH-correction) <0.05 were considered for downstream analyses. D, Volcano plot showing statistical significance (adjusted p value) and log2 fold change (TLR4 KO vs WT) at 7 d post-SCI. An adjusted p value threshold of 0.01 (dotted blue line) and an abs(log2FC) >0.25 (dotted red line) were used to highlight the most relevant differentially expressed sequences. From this pool, E shows the top 10 upregulated (sorted by log2FC; top) and the top 10 downregulated (sorted by log2FC; bottom) sequences between TLR4 KO mice and WT at 7 d post-SCI. F, ORA of 2,463 differentially expressed unique genes (TLR4 KO vs WT) matching KEGG, Reactome, and Gene Ontology (Biological Process and Molecular Function) gene libraries. G, Volcano plot and H top 10 up- and downregulated genes (sorted by log2FC) and (I) ORA pathway analysis in TLR4 KO versus WT mice 8 weeks post-SCI. Significance scores are denoted as **p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

Figure 4.

Reduced expression of MyD88, NFκB, and cytokines in chronic SCI in TLR4 null mice. A–C, Changes in MyD88 mRNA (A; n = 6 for all groups) and protein expression (B,C; n = 4–5 for all groups) at 7 d and 8 weeks post-SCI in wild-type and TLR4 null mice. Note the increased expression at 8 weeks in wild-type but not TLR4 null mice. D–I, Western blots of nuclear (D,E) and cytoplasmic (F,G) NFκB and phospho-IκBα/IκB ratio (H,I). Note the increase at 8 weeks in nuclear NFκB (E) and phospho-IκBα/IκB ratio (I) in wild-type mice as compared with TLR4 null mice (n = 3–4 for all groups). J, Double immunofluorescence labeling of NFκB (green) and CD11b (red) and merge (yellow) 8 weeks after SCI in wild-type (WT; Ji–iii) and TLR4 KO (Jiv–vi) mice. Note strong NFκB expression in the cytoplasm and nucleus in CD11b+ macrophages in WT mice (arrows; Jiii) but lack of nuclear localization in TLR4 KO mice (arrows; Jvi). Also note the significant increase in expression in wild-type mice at 8 weeks compared with TLR4 null mice in IL-1β (K), TNFα (L), IL-6 (M), CXCL1 (N), CXCL2 (O), and CCL3 (P). K–P, n = 5–6 for all groups; one-way ANOVA with post hoc Tukey’s multiple-comparisons test; p < 0.0001 (A,B,I,K–P); p = 0.004 (E); p = 0.004 (G). *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 compared with uninjured naive level. ^#p ≤ 0.05; ^##p ≤ 0.01; ^###p ≤ 0.001 comparing the two injured genotypes. Scale bar: J, 50 µm.

Figure 5.

Reduced immune cell recruitment 1 d after SCI and reduced gliosis in chronic SCI in TLR4 null mice. A, (Ai) t-SNE flow cytometry analysis at 1 d post-SCI reveals the presence of 13 identified and 3 nonidentified CD45⁺ cell populations within the injured spinal cord. (Aii) Heat map showing the relative expression of extracellular markers in the 13 identified clusters. B, t-SNE plot in WT and TLR4 null mice to identify differences between them. C,D, Graphs showing quantification of innate (C) and adaptive (D) immune cell recruitment, 1 d following spinal cord injury (results were assessed for normality using the Shapiro–Wilk test and analyzed using a two-tailed unpaired t test. Data are shown as mean ± SEM; n = 5 in KO and n = 5 in WT groups. E, F, Graphs showing the balance between Ly6C^high, Ly6C^medium, and Ly6C^low populations within WT and TLR4 null mice 1 d after SCI. Note the reduction of Ly6C^inter in TLR4 null mice compared with WT controls. G, Bar graph showing changes in the expression of phenotypic markers in Ly6C^inter cell population. Note the significant reduction of iNOS expression in TLR4 null mice compared with WT. H–K, 8 weeks after SCI, there is significantly reduced immunoreactivity for GFAP (H,I) and CD11b (J,K) indicative of reduced of astrocyte and macrophage/microglial activation, respectively, in TLR4 KO mice compared with wild-type mice (500 µm from the lesion epicenter). I, n = 5 (TLR4 KO) and n = 6 (WT); K, n = 6 per group. Scale bar: H, J, 100 µm.

Figure 6.

Changes in expression of MMP9 and ECM molecules. A–C, At the mRNA level, MMP9 expression is significantly higher in wild-type mice than that in TLR4 KO mice at 8 weeks (A). A significantly higher expression of MMP9 protein was detected by Western blot in wild-type mice compared with TLR4 null mice at 7 d and 8 weeks after SCI (B,C; A, n = 6 for all groups; B,C, n = 3–4 for all groups). D, Double immunofluorescence labeling for MMP9 (green) and either GFAP (Dii,v) or CD11b (Dviii–xi; both red) and merge (yellow) at 8 weeks post-SCI. MMP9 labeling is strong in wild-type animals in GFAP⁺ astrocytes (Diii) and CD11b⁺ macrophages (Dix) but is reduced in TLR4 null mice. Arrows indicate double labeled cells. E–N, Changes in protein level expression detected by Western blot of various ECM molecules. Note the statistically significant increase in wild-type mice at 8 weeks in the expression of versican (E,F; n = 5–6 for all groups), phosphacan (G,H; n = 4 for all groups), and decorin (I,J; n = 5–6 for all groups), while TLR4 null mice were not significantly different from uninjured controls. Expression of lumican (K,L; n = 5–6 for all groups) and collagen 1a1 (M,N; n = 5 for all groups) was increased in both genotypes at 8 weeks; one-way ANOVA with post hoc Tukey’s multiple-comparisons test; p = 0.0002 (A); p < 0.0001 (B,J,L); p = 0.0007 (N); p = 0.003 (F); p = 0.0013 (H). *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001 compared with uninjured naive level; ^#p ≤ 0.05, ^##p ≤ 0.01, ^###p ≤ 0.001 comparing the two injured genotypes. Scale bar: D, 50 µm.

Figure 7.

Immunostaining showing localization of ECM molecules 8 weeks after SCI. A, B, Versican immunostaining. In the uninjured spinal cord, versican is detected at very low levels in discrete cells in the white and gray matter by immunofluorescence (Ai,ii). After injury, versican expression is increased in wild-type and TLR4 null mice in the central core of the lesion where it is localized to CD11b⁺ macrophages (Aiii–vi). Quantification shows significantly lower expression of versican in TLR4 null mice compared with wild-type mice (B; n = 4 for all groups). C, D, Phosphacan immunolabeling is detected at low levels in the uninjured spinal cord in the dorsal and ventral gray matter, which at higher magnification (Cii) appears as punctate staining surrounding neurons. After SCI, phosphacan expression is markedly increased in astrocytes in the white matter with strong labeling along the lesion border in both genotypes (Ciii–vi). Note the expression level is lower in TLR4 null mice compared with wild-type mice (D; n = 4–5 for all groups). E, F, Decorin staining. Immunostaining for decorin showed very weak labeling in the uninjured spinal cord (Ei,ii). After SCI, decorin staining in both genotypes is markedly increased but confined mainly to the central core of the lesion and in CNS tissue immediately dorsal (Eiii–vi). Some decorin staining colocalized with GFAP staining (Civ) but the majority of the staining is within the GFAP-negative central core of the lesion (Eiii–vi). Quantification shows a small but significant reduction in TLR4 null mice (F; n = 6 for all groups). G, H, Lumican labeling. Lumican is not detectable in the uninjured spinal cord (Gi,ii). Its expression increases after SCI and is localized to the same region as decorin. It does not show colocalization with CD11b⁺ macrophages (Giv,vi) but appears to be in the matrix surround the cells. Lumican staining is also significantly lower in TLR4 null mice compared with wild-type mice (H; n = 6 for all groups), *p ≤ 0.05. Two-tailed Mann–Whitney U test (B,D,F,H); p = 0.02 (B); p = 0.015 (D); p = 0.04 (F); p = 0.04 (H). Scale bar: for all panels, 100 µm.

Figure 8.

Expression of aggrecan and localization of WFA staining. Ai, Western blot analysis of aggrecan in TLR4 null and wild-type mice at 7 d and 8 weeks post-SCI. Aii, Quantification of Western blots. Note that expression is significantly reduced at 7 d in both genotypes as compared with uninjured levels (n = 5–6 for all groups). B, Immunofluorescence staining of aggrecan at 8 weeks after SCI shows loss of expression at and near the lesion epicenter in both groups. C, Quantification shows significantly greater expression of aggrecan on either side of the lesion epicenter in TRL4 null mice compared with wild-type mice (n = 5–6 for all groups). D, WFA staining shows similarities to aggrecan staining with more intense labeling in TLR4 null mice compared with wild-type mice, which is confirmed by quantification (E; n = 5–6 for all groups). F, G, Double labeling for aggrecan and NeuN (F) and WFA and NeuN (G) shows that aggrecan and WFA labeling is localized to regions surrounding NeuN⁺ neurons, a pattern that resembles PNNs. n = 5–6 for all groups; *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. Aii, One-way ANOVA with post hoc Tukey’s multiple-comparisons test, p = 0.006; (C,E), two-way RM-ANOVA; genotype effect, with post hoc Bonferroni’s multiple-comparisons test; p = 0.0016 (C); p < 0.0001 (E). Scale bar: for all panels, 100 µm.

Figure 9.

Reduction in CSPG GAG staining, increase in 5-HT sprouting, and improvement in locomotor recovery in TLR4 null mice 8 weeks after SCI. Ai, ii, CS-56 monoclonal antibody labeling of CSPG GAG chains labels wild-type injured spinal cord at 8 weeks more strongly than TLR4 null mice. Images taken at the same distance from the lesion epicenter (500 µm). Aiii, Quantification shows significant reduction in CS-56 labeling in TLR4 null mice compared with wild-type mice (n = 4–6 per all group). Bi, ii, 5-HT immunoreactivity in the ventral horn region 1 mm caudal to the lesion epicenter shows increased innervation in TLR4 null mice (Bii) compared with wild-type mice (Bi). Quantification shows significant increase in 5-HT labeling in TLR4 null mice compared with wild-type mice (Biii; n = 5–6 per all group). C, Locomotor recovery assessed by BMS analysis shows delayed improvement in locomotor recovery in TLR4 null mice by 8 weeks; n = 14 (TLR4 KO) and n = 15 (WT). Two-way RM-ANOVA; time × genotype effect with post hoc Bonferroni’s multiple-comparisons test; p = 0.02 (D). BMS subscores, which evaluate finer aspects of locomotor control, show significant improvement starting from 35 d post-SCI; n = 14 (TLR4 KO) and n = 15 (WT). Two-way RM-ANOVA; time × genotype effect with post hoc Bonferroni’s multiple-comparisons test; p < 0.0001. E, DigiGait analysis shows significant improvement is several parameters of gait in TLR4 null mice compared with wild-type mice: (i) stride length, p = 0.005; (ii) stride duration, p = 0.005; (iii) propulsion duration, p = 0.04; (iv) swing duration, p = 0.003; (v) stride length, p = 0.005; (vi) brake duration, p = 0.04; two-tailed Mann–Whitney U test; n = 12 (TLR4 KO) and n = 10 (WT); *p ≤ 0.05; **p ≤ 0.01. Scale bar: A, B, 100 µm.

Figure 10.

Schematic diagram illustrating the key findings of the role of TLR4 signaling after SCI. Except for the bottom of the figure which shows the early effects of TLR4 in the infiltration of leukocytes, the rest of the diagram focuses mainly on the chronic effects of TLR4, 8 weeks after SCI. TLR4 is expressed in astrocytes and macrophages at 7 d and in astrocytes at 8 weeks. It regulates NFκB signaling in macrophages and astrocytes and possibly downstream effects on chemokine and cytokines expression. TLR4 also regulates expression and deposition of PNN- and scar-related ECM molecules. At the chronic stage, TLR4 signaling via non-neuronal cells also appears to regulate the necroptosis (cell death) pathway (p-MLKL) in neurons and possibly also oligodendrocytes (data not shown). TLR4 and NFκB can also influence expression of MMP9 that may play a role in modulating the deposition of PNN- and scar-related ECM molecules in divergent ways. Three key components of the TLR4 response in chronic SCI (cell death, inflammation, and ECM deposition) are depicted in red. The reduction in necroptosis, inflammatory regulators, and modulation of scar- and PNN-related ECM molecules observed in TLR4 null mice contribute in varying degrees to improved sprouting of 5-HT axons and improvement in locomotor recovery during the chronic period after SCI.