Human traumatic brain injury induces autoantibody response against glial fibrillary acidic protein and its breakdown products

Abstract

The role of systemic autoimmunity in human traumatic brain injury (TBI) and other forms of brain injuries is recognized but not well understood. In this study, a systematic investigation was performed to identify serum autoantibody responses to brain-specific proteins after TBI in humans. TBI autoantibodies showed predominant immunoreactivity against a cluster of bands from 38-50 kDa on human brain immunoblots, which were identified as GFAP and GFAP breakdown products. GFAP autoantibody levels increased by 7 days after injury, and were of the IgG subtype predominantly. Results from in vitro tests and rat TBI experiments also indicated that calpain was responsible for removing the amino and carboxyl termini of GFAP to yield a 38 kDa fragment. Additionally, TBI autoantibody staining co-localized with GFAP in injured rat brain and in primary rat astrocytes. These results suggest that GFAP breakdown products persist within degenerating astrocytes in the brain. Anti-GFAP autoantibody also can enter living astroglia cells in culture and its presence appears to compromise glial cell health. TBI patients showed an average 3.77 fold increase in anti-GFAP autoantibody levels from early (0-1 days) to late (7-10 days) times post injury. Changes in autoantibody levels were negatively correlated with outcome as measured by GOS-E score at 6 months, suggesting that TBI patients with greater anti-GFAP immune-responses had worse outcomes. Due to the long lasting nature of IgG, a test to detect anti-GFAP autoantibodies is likely to prolong the temporal window for assessment of brain damage in human patients.

Figure 1. Human severe TBI patients developed circulating IgG autoantibodies against brain proteins within the 50-38 kDa range after injury.

A. An immunoblot of human brain lysate probed with sera from 5 normal controls (C1–C5) and 5 TBI patients (P1–P5), the latter at two time points post injury (Day 1 and Day 10). Sera were used at 1∶100 except TBI P4, which was 1∶2000. B. An immunoblot of human brain lysate probed with sera from 2 normal controls (Ctrl), or daily serum samples from a strongly immunoreactive TBI patient (Days 0–10 post TBI). C. An immunoblot of human brain lysate probed with control or TBI serum (Day 9), and then developed with secondary antibodies against human IgG or IgM. D. Lysates from a panel of human organs probed with pooled TBI sera (Day 10; n = 4). E. Sera from 53 TBI patients (Day 4–10) blotted individually against human brain lysate. Lanes on blots were arbitrarily subdivided into 5 kDa increments using vision works LS image acquisition software. The total number of bands in all lanes were counted, summed and plotted according to MW.

Figure 2. A proteomics approach identified GFAP as the predominant TBI-associated autoantigen.

Workflow of the proteomic analysis is shown at left. A. Human brain lysate was subjected to anion-based chromatographic separation. A graph of absorbance at 280 nm of fractions 11–26 with elution time is shown. B. Fractions were subjected to SDS-PAGE separation, and gels were stained with Coomassie blue to visualize proteins (fractions 17–25 are shown). C. Duplicate gels were blotted and probed with pooled TBI sera (Day 10; n = 4). D, E. The Coomassie gels in B were overlaid by TBI blots in C, immunoreactive bands from fractions 20–22 were excised for LC-MS/MS analysis (arrowheads), and results are shown in the table. F. A duplicate blot to that in C was probed with anti-GFAP antibody (Abcam) to confirm that GFAP was the autoantigen.

Figure 3. Calpain-associated GFAP breakdown occurred in vitro and in primary rat cortical cultures.

A. Rat brain (Rt brn) lysate was subjected to calpain-2 or caspase-3 digestion in vitro, blotted, and probed with antibodies against GFAP (Abcam; top panel) or αII-spectrin (Enzo; bottom panel). Calpain-2 produced a banding pattern similar to the human brain lysate, while caspase-3 did not. Enzymes produced their characteristic spectrin BDP patterns (SBDP150 for calpain, SBDP150i and SBDP120 for caspase). B. Rat CTX mixed cultures were untreated (Cntl) or subjected to the indicated drug treatments, then blotted and probed with GFAP antibody (Abcam). The calpain activator maitotoxin (MTX), but not the caspase activator EDTA, induced strong GFAP breakdown to the 38 kDa GFAP band. The calpain inhibitor SNJ1945 (SNJ), but not the caspase inhibitor Z-VAD-fmk (VAD), blocked GFAP breakdown by MTX. In A and B, vertical hash marks indicate where intervening lanes were removed. An asterisk marks the 38 kDa GFAP band throughout.

Figure 4. Calpain-associated GFAP breakdown to 38 kDa occurred in vivo.

A. Rats experienced penetrating ballistic-like brain injury (PBBI). Control (naïve, sham) or PBBI brain lysates at 1 day (1D) after injury were immunoblotted and probed with GFAP antibody (Banyan). The 38 kDa GFAP bands were quantified by densitometry. This band was significantly increased compared to controls post PBBI (mean ± SEM; n = 5 per group). Vertical hash marks indicate where intervening lanes were removed. B. Rats were injured by controlled cortical impact (CCI), and injected either with vehicle (n = 3) or the calpain inhibitor SNJ1945 (n = 5). Then brain lysates were immunoblotted and probed with antibodies against αII-spectrin (Enzo) or GFAP (Banyan). Quantification of the SBDP145 and 38 kDa GFAP bands showed that SNJ1945 significantly inhibited formation of SBDP145 and 38 kDa GFAP (mean ± SEM). An asterisk marks the 38 kDa GFAP band throughout.

Figure 5. Calpain removed the termini of GFAP to generate the 38-BDP.

A. Human GFAP was undigested or digested with calpain (Capn), resolved on SDS-PAGE, stained with Coomassie blue R-250, and the indicated bands were excised. An asterisk marks the 38 kDa GFAP band. B. The N-terminal sequence of each band is shown in the table. C. Calpain cleavage sites (scissors) near the N- and C-termini of GFAP are shown in a schematic drawing of the GFAP. Subdomains (head, rod, tail) and identified/putative phosphorylation sites (P) are indicated, and the predicted 38 kDa calpain fragment is shown. D. Three control and four post-TBI human serum samples (Day 4–10) were probed against human brain lysate (top panel) and cell lysate from HEK cells overexpressing the 38 kDa GFAP-BDP (lower panel).

Figure 6. TBI autoantibodies labeled GFAP in injured rat brain.

A. In sections of cortex from rats injured by CCI, anti-GFAP antibody (Cell Signaling) or human Day 10 TBI serum showed similar patterns of immunoreactivity (brown), while human normal serum did not. Nuclei are blue. B. Sections of hippocampus from naïve or CCI-injured rats were incubated with anti-GFAP antibody (Cell Signaling) plus human Day 10 TBI serum, and bound antibodies were visualized with fluorescent secondary antibodies (anti-GFAP in red, TBI serum in green). Image contrast was adjusted in the panels individually to best display the fluorescence. In naïve hippocampus, TBI autoantibodies showed little binding, while in CCI hippocampus, anti-GFAP antibody and TBI autoantibodies showed increased immunostaining, which partially colocalized (orange, arrows).

Figure 7. TBI autoantibodies colocalized with GFAP in primary rat astrocytes.

Single confocal images (A–E) and flattened z-stacks (F) are shown of astrocytes stained with anti-GFAP antibody (red), human pooled TBI sera (green, Day 9–10, n = 3) or human normal control serum (green). Nuclei are blue. A. GFAP antibody stained GFAP fibrils in primary rat astrocytes (left), while human control serum showed diffuse green minor staining (middle) not colocalizing with GFAP (right). B. Astrocytes digested with calpain-2 after fixation labeled with GFAP (left) but not normal human serum (middle) as in A. C. Immunostaining by GFAP antibody (left) and human TBI sera (middle) strongly colocalized, as evidenced by yellow color (right). D. After calpain digestion, staining by the GFAP antibody (left) and human TBI sera (middle) colocalized (right) as in C, without changes in signal intensities. E. GFAP antibody (left) and human TBI sera (middle) brightly co-stained (right) rounded cells with condensed nuclear material. F. Flattened z-stacks of multiple confocal images are shown, to illustrate the greater staining intensities of GFAP (left) and TBI sera (middle) in a rounded cell compared to healthy cells. Exposure times were shorter in E, F compared to A–D. Scale bars 10 microns.

Figure 8. Effects of human TBI serum anti-GFAP autoantibody primary rat glial cells.

(A–E) Primary rat glia cells were plated on cover slip and cell chamber and incubated in media that included 1/50 serum collected from either human TBI serum with no detectable anti-GFAP autoantibody (Control serum) (A,B) or human TBI serum with strong anti-GFAP autoantibody (C,D). Cells were treated with either vehicle (DMSO) (A,C) or 20 μM A23187 (calcium ionophore) (B,D) for 24 hours, After the 24 h incubation, media was removed and the cells were fixed, stained with Hoechst and developed with anti-human IgG secondary antibody, Scale bars 10 microns. Yellow arrowheads indicate glial cell body staining with IgG while blue arrows indicate glial processes staining. (E) To test of effects of glial health, we used pooled 4 control samples and 9 TBI samples with strong autoantibody response. Cells were co-treated with either vehicle (DMSO) or 20 μM A23187 for 24 hours. Cell death (by LDH release) was measured using CytoTox-Glo™ Cytotoxicity Assay. Luminescence was read and media from each of the different conditions was used to measure the background, which was subtracted from the individual readings. * (P<0.0001) Represents significant increase of glial death in cells treated with TBI serum as compared to cells treated with control serum (control or with A23187, respectively), using an ANOVA analysis with Bonferroni multiple comparisons.

Figure 9. Clinical analyses of GFAP autoantibodies in severe TBI and control subjects.

(A) Day 4–10 serum GFAP-BDP (38 kDa) autoantibody levels from severe TBI patients were semi-quantified and found to be significantly higher than those from normal controls. (B) Receiver Operating characteristic (ROC) curve plotting serum autoantibody levels to 38 kDa GFAP-BDP revealed that autoantibody level was a good discriminator between TBI subjects and healthy normal controls with AUC = 0.78. The average serum autoantibody levels at a cutoff of >8.49 arbitrary unit has a 64% sensitivity, 85% specificity, to distinguish between TBI and controls. (95% CI 0.70–0.87. AUC = area under receiver operating characteristic curve). (C) Frequency of autoab against 38 kDa GFAP-BDP in TBI serum (Day 4–10; n = 53) as compared to normal controls (n = 96). *Autoantibody positive represents a discrete immunoreactivity band signal of at least 1.50 densitometric units after background subtraction. (D) GFAP-BDP autoantibody correlated significantly to serum GFAP levels at 24 h post-TBI measured by sandwich ELISA. (E) A significant correlation between the best GCS score (index of severity) during the first 24 h after injury and GFAP-BDP autoantibody levels (GCS 3–8 are severe, GCS 9–13 moderate). (F) GFAP-BDP autoantibody levels at Day 4–10 correlated to outcome measurement GOS-E at 6 months.