1. Cell Biology

Site-specific phosphorylation and caspase cleavage of GFAP are new markers of Alexander disease severity

  1. Rachel A Battaglia
  2. Adriana S Beltran
  3. Samed Delic
  4. Raluca Dumitru
  5. Jasmine A Robinson
  6. Parijat Kabiraj
  7. Laura E Herring
  8. Victoria J Madden
  9. Namritha Ravinder
  10. Erik Willems
  11. Rhonda A Newman
  12. Roy A Quinlan
  13. James E Goldman
  14. Ming-Der Perng
  15. Masaki Inagaki
  16. Natasha T Snider  Is a corresponding author
  1. University of North Carolina, United States
  2. University of Durham, United Kingdom
  3. Thermo Fisher Scientific, United States
  4. Columbia University, United States
  5. National Tsing Hua University, Taiwan, Republic of China
  6. Mie University Graduate School of Medicine, Japan
https://doi.org/10.7554/eLife.47789
Share this article
Cite this article
  1. Rachel A Battaglia
  2. Adriana S Beltran
  3. Samed Delic
  4. Raluca Dumitru
  5. Jasmine A Robinson
  6. Parijat Kabiraj
  7. Laura E Herring
  8. Victoria J Madden
  9. Namritha Ravinder
  10. Erik Willems
  11. Rhonda A Newman
  12. Roy A Quinlan
  13. James E Goldman
  14. Ming-Der Perng
  15. Masaki Inagaki
  16. Natasha T Snider
(2019)
Site-specific phosphorylation and caspase cleavage of GFAP are new markers of Alexander disease severity
eLife 8:e47789.
https://doi.org/10.7554/eLife.47789
  2. Download BibTeX
  3. Download .RIS

Abstract

Alexander disease (AxD) is a fatal neurodegenerative disorder caused by mutations in glial fibrillary acidic protein (GFAP), which supports the structural integrity of astrocytes. Over 70 GFAP missense mutations cause AxD, but the mechanism linking different mutations to disease-relevant phenotypes remains unknown. We used AxD patient brain tissue and induced pluripotent stem cell (iPSC)-derived astrocytes to investigate the hypothesis that AxD-causing mutations perturb key post-translational modifications (PTMs) on GFAP. Our findings reveal selective phosphorylation of GFAP-Ser13 in patients who died young, independently of the mutation they carried. AxD iPSC-astrocytes accumulated pSer13-GFAP in cytoplasmic aggregates within deep nuclear invaginations, resembling the hallmark Rosenthal fibers observed in vivo. Ser13 phosphorylation facilitated GFAP aggregation and was associated with increased GFAP proteolysis by caspase-6. Furthermore, caspase-6 was selectively expressed in young AxD patients, and correlated with the presence of cleaved GFAP. We reveal a novel PTM signature linking different GFAP mutations in infantile AxD.

eLife digest

Alexander disease is a rare and fatal neurodegenerative condition. It is caused by mutations in a gene that is crucial for the structure of astrocytes, a type of brain cell whose role is to support neurons. The gene codes for a protein called GFAP, which is made almost exclusively in astrocytes. In Alexander disease, mutated versions of the gene make GFAP collect in disordered clumps or aggregates, which interfere with the astrocytes’ normal activities.

All Alexander disease patients develop GFAP aggregates, but the type of mutation they have in the gene for GFAP does not predict how their illness will progress. The age of onset of disease, for example, can vary between less than one year old to more than 70 years old. Battaglia et al. sought to understand how GFAP aggregates form in the cells of Alexander disease patients. One way that GFAP can be altered in the cell is by a process called phosphorylation. Enzymes called kinases add phosphate groups to GFAP, which can regulate the protein’s activity, stability and interactions with other proteins.

Battaglia et al. found high levels of phosphorylation at one specific site in the GFAP protein in people who had very early onset of Alexander disease. This phosphorylation was not related to any particular mutation in the gene for GFAP. An added phosphate group at this location in the protein made GFAP more likely to be broken into two pieces by an enzyme called caspase-6. One of the breakdown products is already known to play a role in aggregation. Young patients with Alexander disease had high levels of GFAP breakdown products and caspase-6. The phosphorylated protein and this enzyme were found to accumulate in astrocyte aggregates.

The findings provide a basis for investigating new strategies to treat Alexander disease that target phosphorylation – as removing or preventing the addition of phosphate groups can be done with drugs. But before exploring how to do this, it will be necessary to find out which enzyme is responsible for phosphorylating GFAP at this particular position. These studies may also give a broader understanding of other GFAP-like proteins (called intermediate filament proteins), which are involved in over 70 human diseases.

Introduction

Alexander disease (AxD) is a rare and invariably fatal neurological disorder that affects primarily infants and small children, but can also manifest later in life (Alexander, 1949; Sosunov et al., 2018; Messing, 2018). Autosomal dominant gain-of-function mutations in GFAP, which encodes glial fibrillary acidic protein (GFAP), cause AxD (Messing, 2018; Brenner et al., 2001). GFAP is the major component of the intermediate filament (IF) cytoskeleton in astrocytes (Hol and Pekny, 2015). The accumulation and incorporation of mutant GFAP within cytoplasmic aggregates called Rosenthal fibers (RFs), causes reactive gliosis, leading to secondary injury to neurons and non-neuronal cells (Olabarria and Goldman, 2017; Wang et al., 2015; Li et al., 2018; Jones et al., 2018). Silencing GFAP via antisense oligonucleotide intervention in vivo eliminates RFs, reverses the stress responses in astrocytes and other cell types, and improves the clinical phenotype in a mouse model of AxD (Hagemann et al., 2018). While the utility of GFAP as a key therapeutic target in AxD is clear, the molecular mechanisms for how AxD-associated GFAP missense mutations (affecting over 70 different residues on GFAP) lead to defective GFAP proteostasis are not well understood. Deciphering these mechanisms may yield novel interventions, not only for AxD patients, but also for patients with other diseases where IF proteostasis is severely compromised.

Normal functioning IFs are stress-bearing structures that organize the cytoplasmic space, scaffold organelles, and orchestrate numerous signaling pathways. In contrast, dysfunctional IFs directly cause or predispose to over 70 tissue-specific or systemic diseases, including neuropathies, myopathies, skin fragility, metabolic dysfunctions, and premature aging (Omary, 2009; www.interfil.org). Disease-associated IF proteins share two key molecular features: abnormal post-translational modifications (PTMs) (Snider and Omary, 2014) and pathologic aggregation. The GFAP-rich RF aggregates that are hallmarks of AxD astrocytes bear strong similarities to pathologic aggregates of other IFs, including epidermal keratins (Coulombe et al., 1991), simple epithelial keratins (Nakamichi et al., 2005), desmin (Dalakas et al., 2000), vimentin (Müller et al., 2009), neurofilaments (Zhai et al., 2007) and the nuclear lamins (Goldman et al., 2004). There are unique advantages to studying IF proteostasis mechanisms in the context of GFAP because of its restricted cellular expression, homopolymeric assembly mechanism, and because GFAP is the sole genetic cause of AxD as a direct result of its toxic gain-of-function accumulation and aggregation.

Like all IF proteins, GFAP contains three functional domains: amino-terminal ‘head’ domain, central α-helical ‘rod’ domain and carboxy-terminal ‘tail’ domain (Eriksson et al., 2009). The globular head domain is essential for IF assembly and disassembly, which are regulated by various PTMs, in particular phosphorylation (Omary et al., 2006). It was shown previously that phosphorylation of multiple sites in the head domain of GFAP (Thr-7, Ser-8, Ser-13, Ser-17 and Ser-34) regulates filament disassembly during mitosis and GFAP turnover in non-mitotic cells (Inagaki et al., 1990; Takemura et al., 2002a; Inagaki et al., 1994; Inagaki et al., 1996). Additionally, phosphorylation of GFAP has been observed after various injuries of the central nervous system (CNS) including kainic acid-induced seizures, cold-injury, and hypoxic-ischemic models, where phosphorylated GFAP is expressed in reactive astrocytes (Valentim et al., 1999; Takemura et al., 2002b; Sullivan et al., 2012). These observations reveal that phosphorylation of GFAP is important for re-organization of the astrocyte IF cytoskeleton and plasticity in response to injury. However, it is not clear if, and how, abnormal GFAP phosphorylation compromises proteostasis and contributes to AxD pathogenesis.

Here, we identified a critical phosphorylation site in the GFAP head domain that is selectively and strongly upregulated in the brain tissues of AxD patients who died very young, independently of the position of the disease mutation that they carried. Further, we show that this site-specific phosphorylation promotes GFAP aggregation and is a marker of perinuclear GFAP aggregates associated with deep nuclear invaginations in AxD patient astrocytes, but not in isogenic control astrocytes. Finally, we demonstrate a correlation between site-specific GFAP phosphorylation and caspase cleavage in cells and in post-mortem brain tissue from AxD patients. Although our study does not establish a causal relationship between GFAP phosphorylation and caspase cleavage, we show that caspase-6 is a new marker for the most severe form of human AxD.

Collectively, our results reveal a new PTM signature that is associated with defective GFAP proteostasis in the most severe form of AxD. Future interventional studies targeting these PTMs will determine whether they contribute to, or are the consequence of, disease severity.

Results

Phosphorylation of Ser13 on GFAP is a marker of the most aggressive form of AxD

IFs undergo protein synthesis-independent turnover and re-organization to meet cellular demands (Robert et al., 2016). PTMs are key in that process, as they regulate filament polymerization and depolymerization, protein-protein interactions, and oligomerization properties of IF proteins (Snider and Omary, 2014). Of all known PTMs that regulate IFs, phosphorylation is the most ubiquitous and can facilitate or antagonize other types of PTMs via complex cross-talk mechanisms (Omary et al., 2006). We hypothesized that AxD-associated GFAP missense mutations (Figure 1A) promote GFAP accumulation and aggregation by dysregulating site-specific phosphorylation. We extracted GFAP from post-mortem brain cortex tissue of 13 AxD patients, representing 10 different mutations (Supplementary file 1) and three non-AxD controls (Supplementary file 2). GFAP from the insoluble high salt extracts (HSEs), prepared according to the procedure described in Figure 1—figure supplement 1, was used in phospho-proteomic analysis, revealing 12 unique phosphorylation sites on GFAP in AxD (Figure 1B–C). While the AxD-specific phospho-peptides localized to all three functional domains of GFAP (head, rod, tail), the most abundantly phosphorylated residue was a conserved serine (Ser13) in the head domain (Figure 1C–D).

Figure 1 with 1 supplement see all
Download asset Open asset
GFAP is phosphorylated on head domain Ser13 in human AxD brain.

(A) Schematic displays the frequency and location of AxD patient GFAP mutations. (B) Method used to identify GFAP phospho-peptides. (C) Graph of AxD-specific GFAP phospho-peptides identified by mass spectrometry and type/position of patient mutations. PSM = peptide spectrum match. Green diamonds represent GFAP mutations in young patients (median age at death = 1.7 years; range 0.5–14 years) and pink diamonds represent older patients (median age at death = 38 years; range 27–50 years). (D) Amino acid conservation at the N-terminus of human, rat and mouse GFAP. The green box indicates the serine corresponding to human Ser13, which is conserved in rat and mouse.

Strikingly, we found that the pSer13-GFAP peptide was selectively elevated in the cortex tissue from AxD patients who died very young (median age at death = 1.7 years; range 0.5–14 years) (Figure 2A). Overall, we did not observe significant phosphorylation of GFAP in the control subjects (Figure 2—source data 1), or in AxD patients who lived 27–50 years (median age at death = 38 years). Further, immunoblot analysis using a phospho-specific antibody (KT13) (Sekimata et al., 1996) against pSer13-GFAP validated the mass spectrometry results in the AxD patients (Figure 2B–C). Although there was one notable outlier in each age group (Figure 2B lanes 3 and 11), our results suggest that pSer13-GFAP is primarily associated with the more aggressive, infantile form of AxD. Furthermore, the differences in phosphorylation were not a result of age, since pSer13 GFAP was generally not present in the brain lysates from non-AxD control subjects, regardless of age (Figure 2D).

Figure 2
Download asset Open asset
GFAP is phosphorylated on head domain Ser13 primarily in AxD brain from young patients.

(A) Quantification of pSer13-GFAP abundance by mass spectrometry in young (green) vs. old (pink) AxD patients (*=p < 0.05 unpaired t-test). (B) Validation of pSer13-GFAP by western blot of HSE from AxD patients, using a phospho-specific antibody to pSer13-GFAP. The order of samples, by AxD donor ID number, is: 1482, 1070, 885, 5488, 1161, 2768, 338, 613, 5377, 5517, M3596, 5109, and 4858 (listed in Supplementary file 1). (C) Quantification of the relative intensity of pSer13-GFAP on western blot in young (green) and old (pink) AxD patients (*=p < 0.05 unpaired t-test). Signal intensity was normalized to total GFAP in each sample. (D) Western blot of pSer13-GFAP in non-AxD control brain lysates of different ages. The order of samples, by donor ID number, is: 1547, 5941, 103, 1791, 1670, 4898, 1706, 1711, 1011, 632, 4640, and 4915 (listed in Supplementary file 2).

Figure 2—source data 1

Raw data from mass spectrometry PTM profiling of GFAP extracted from AxD and control human brain.

https://cdn.elifesciences.org/articles/47789/elife-47789-fig2-data1-v2.xlsx

Phospho-mimic mutation at Ser13 promotes GFAP aggregation

To determine the functional significance of pSer13 on GFAP filament organization, we analyzed the filament properties of non-phosphorylatable (S13A) and phospho-mimic (S13D and S13E) GFAP mutants. We optimized a transient over-expression system in the SW13 vimentin-negative adrenocarcinoma cells (SW13vim-) for this assay, which resulted in primarily filamentous WT GFAP and insoluble aggregated forms of common AxD mutants of GFAP (Figure 3—figure supplement 1). Compared to wild-type (WT) GFAP, the S13D and S13E mutants assembled primarily into large aggregates, similar to the most common AxD-associated mutant R79H-GFAP (Figure 3A–B). S13A formed mostly filaments, although they appeared shorter compared to WT GFAP. To determine if the phospho-mimic mutation directly promotes aggregation, we compared the assembly properties of purified WT, S13A and S13D GFAP (Figure 3C). Consistent with the phenotype observed in the transfected cells, the S13A mutant formed abnormally short filaments in vitro. In contrast, S13D was completely incapable of filament assembly, forming globular structures that were homogeneous in size and not aggregation-prone. Our results with the phospho-deficient and phospho-mimic mutants reveal that S13 is a key site that regulates the assembly properties of GFAP and that its phosphorylation status may modulate the dynamics between filaments and aggregates.

Figure 3 with 1 supplement see all
Download asset Open asset
Effect of phospho-deficient and phospho-mimic S13 substitutions on GFAP filament assembly in cells and in vitro.

(A) Representative images of immunofluorescence staining of DNA (blue) and GFAP (green) in SW13vim- cells transfected with wild-type GFAP (WT), R79H mutant GFAP (R79H), non-phosphorylatable GFAP (S13A), and phospho-mimic GFAP (S13D and S13E) as single or double mutations, as noted in the images. Scale bar = 5 µm. (B) Quantification of percentage of cells containing GFAP filaments, aggregates or both (n = 41–103 cells per condition). RH = R79H; SA = S13A; SD = S13D; SE = S13E. (C) Electron micrographs showing the filament properties of in vitro assembled GFAP (WT, S13A and S13D). Bottom three panels represent magnified areas marked by the white boxes in the top panels. Scale bars = 500 nm.

Generation of AxD induced pluripotent stem cells (iPSCs) and isogenic controls

In order to explore the function of this phosphorylation event in a disease-relevant system, we used an in vitro human astrocyte model of AxD. We generated iPSCs using fibroblasts from a young AxD patient and characterized their pluripotency by immunofluorescence staining (Figure 4A). Karyotype analysis showed that there were no chromosomal abnormalities due to the reprogramming process (Figure 4—figure supplement 1). To generate isogenic control cells, we corrected the heterozygous point mutation in GFAP (c.715C > T, p.R239C) using CRISPR/Cas9 mediated gene editing (Figure 4B). Representative chromatograms are shown for the original patient cells and the isogenic controls (Figure 4C). We also isolated ‘CRISPR control’ clones, which were edited on the wild-type GFAP allele, thereby retaining the AxD-causing mutation and serving as an additional disease control for the gene editing procedure. Similar to the original patient cells, the edited cells were karyotyped and characterized for pluripotency (Figure 4—figure supplement 1). We confirmed that there were no off-target effects due to the editing procedure (Supplementary file 3).

Figure 4 with 2 supplements see all
Download asset Open asset
Generation and characterization of AxD patient iPSC-astrocytes and isogenic controls.

(A) Characterization of iPSC pluripotency. Bright field images of AxD patient fibroblasts (top left) and iPSCs (bottom left). Immunofluorescence staining for pluripotency markers in AxD iPSCs. (B) GFAP sequence for the AxD mutant allele and the corrected allele. Differences between the sequences are indicated by red text. The AxD-causing mutation is underlined, and all other changes are silent mutations. The area of gRNA recognition is indicated by the red line. (C) Chromatograms showing AxD heterozygous mutation in the original patient cells (top), correction of the mutant allele in the isogenic control (middle) and correction of the wild-type allele in the CRISPR control (bottom). Red arrows denote presence of the disease mutation and green check mark denote genetic correction and presence of silent mutations. (D) Schematic representation of astrocyte differentiation protocol. NIM, neural induction medium; NPM; neural progenitor medium; ADM, astrocyte differentiation medium; AMM; astrocyte maturation medium. (E) Immunoblot of GFAP in iPSC-astrocytes. Pan-actin blot and Coomassie stain serve as loading controls. (F) Quantification of band intensities for GFAP from panel E. ****p<0.0001 compared to isogenic control; one-way ANOVA.

GFAP accumulation and perinuclear aggregation into RF-like structures in AxD iPSC-astrocytes

AxD, CRISPR control, and isogenic control iPSCs were differentiated to astrocytes (iPSC-astrocytes) via neural progenitor cells (NPCs), as described in the Materials and methods and shown schematically in Figure 4D. After 54 days in culture, iPSC-astrocytes express classical astrocyte markers (Zhang et al., 2014), including alcohol dehydrogenase 1 family member L1 (ALDH1L1), solute carrier family 1 member 3 (SLC1A3), excitatory amino acid transporter 2 (EAAT2), Connexin 43 and GFAP (Figure 4—figure supplement 2). To assess if our model recapitulated key features of AxD, we analyzed total GFAP expression in the iPSC-astrocytes by immunoblot, and found that GFAP levels were significantly higher in the cells that carried the heterozygous GFAP point mutation (AxD patient and CRISPR control lines) relative to the isogenic controls (Figure 4E–F). This is consistent with in vivo observations of GFAP levels in AxD patients (Jany et al., 2015) and mouse models (Jany et al., 2013). In addition, high-molecular-mass (hmm) GFAP oligomers were present in the AxD iPSC-astrocytes, similar to what we observed when we ectopically expressed the R239C-GFAP mutant (Figure 5A). Finally, we observed by immunofluorescence staining that the AxD mutant iPSC-astrocytes formed both GFAP filaments and perinuclear aggregates (Figure 5B), whereas the isogenic control iPSC-astrocytes formed only GFAP filaments (Figure 5C). In vivo, GFAP antibodies stain the periphery, while DAPI stains the core of RFs (Der Perng et al., 2006; Sosunov et al., 2017). The in vitro-derived AxD iPSC-astrocytes displayed similar characteristics, with RF-like perinuclear aggregates staining positively for GFAP at their periphery and DAPI in the center (Figure 5B).

Figure 5
Download asset Open asset
Oligomerization and perinuclear aggregation of GFAP in AxD iPSC-astrocytes.

(A) GFAP blot of AxD iPSC-astrocytes (left) and SW13vim- cells transfected with R239C mutant GFAP (right) reveals GFAP monomer and high-molecular-mass GFAP oligomers. Immunoblots on the bottom are of the same membranes at lower exposure. (B) Immunofluorescence staining for GFAP (magenta) and DAPI (white) in AxD iPSC-astrocytes reveals presence of perinuclear GFAP aggregates, marked by the yellow arrows. Scale bars = 10 µm. Boxed area in the merged image is shown by the enlarged image on the right. (C) Immunofluorescence staining for GFAP (magenta) and DAPI (white) in isogenic control iPSC-astrocytes. Scale bars = 10 µm.

pSer13-GFAP marks the core of perinuclear GFAP aggregates localized within deep nuclear invaginations

Next, we determined if pSer13-GFAP was present in the AxD iPSC-astrocytes, similar to what we observed in the human brain tissues. As shown in Figure 6A, pSer13-GFAP signal was detected strongly within the core of the perinuclear GFAP aggregates of AxD iPSC-astrocytes. Somewhat surprisingly, we also observed pSer13-GFAP signal in the isogenic control cells, possibly triggered by the in vitro culture conditions. Nevertheless, unlike AxD astrocytes, in the isogenic control astrocytes pSer13-GFAP organization was filamentous and paralleled that of total GFAP. Therefore, the in vitro iPSC-astrocyte model revealed that, only in the presence of the AxD disease mutation, pSer13-GFAP is incorporated within the core of perinuclear inclusions. While in all AxD cells pSer13 signal was detected in the aggregates, we also observed cells with pSer13-positive diffuse cytoplasmic staining and filaments, likely reflecting different states of the GFAP network (Figure 6—figure supplement 1). Furthermore, the pSer13-positive GFAP aggregates appeared adjacent to prominent nuclear invaginations (Figure 6A). Nuclear deformations, similar to what we observed in the AxD iPSC-astrocytes, are also present in RF-bearing astrocytes in AxD human brain (Sosunov et al., 2017). To determine whether the perinuclear aggregates compromised the nuclear envelope, we examined the AxD iPSC-astrocytes by electron microscopy. While we observed filamentous bundles on the cytoplasmic side of the nuclear invaginations, the nuclear envelope appeared intact (Figure 6B). Thus, pSer13-GFAP marks cytoplasmic GFAP aggregates adjacent to nuclear invaginations. It should be noted that the perinuclear aggregates containing disorganized GFAP filaments are not identical to the electron-dense RFs that are seen in post-mortem patient brain, but that they may reflect an intermediate state of GFAP accumulation.

Figure 6 with 1 supplement see all
Download asset Open asset
pSer13 marks perinuclear accumulation of GFAP within nuclear invaginations in AxD iPSC-astrocytes.

(A) Immunofluorescence staining of total GFAP (magenta), pSer13-GFAP (green) and DAPI (blue) in isogenic control (top panels) and AxD mutant (bottom panels) iPSC-astrocytes. Perinuclear GFAP aggregates are indicated by the yellow arrows. Scale bars = 10 µm. (B) Electron microscopy images of AxD patient iPSC-astrocytes revealing large, juxtanuclear fibrous bundles (boxed area on left), shown at higher magnification on the right. Scale bar = 5 µm (left) and 0.5 µm (right).

Phosphorylation at Ser13 promotes caspase-mediated cleavage of GFAP

To understand the mechanism for how GFAP phosphorylation may promote GFAP aggregation, we conducted a biochemical analysis of the S13A, S13D and S13E GFAP mutants. In line with our immunofluorescence result (Figure 3A), we observed an increase in high-molecular-mass ~100 kDa GFAP oligomer in the phospho-mimic mutant by immunoblot analysis (Figure 7A). However, more strikingly, we observed increased levels of a cleaved GFAP fragment (24 kDa) in S13D and S13E, which was significantly lower in WT- and S13A-GFAP (Figure 7A–B). Cleavage of GFAP by caspase-6 in vitro generates two fragments of 24 and 26 kDa size (Chen et al., 2013). The 24 kDa C-terminal fragment is recognized by the monoclonal GA5 antibody, (Chen et al., 2013) which was used here. Therefore, we tested the effect of a peptide inhibitor of caspase-6 (Ac-VEID-CHO), and found that it significantly reduced the amount of cleaved S13D-GFAP (Figure 7C–D). Furthermore, we observed augmented cleavage of S13D-GFAP when combined with an AxD-causing mutation (S13D/R79H double mutant), and this was also blocked by the caspase-6 inhibitor (Figure 7C–D). Further analysis of the AxD mutant R79H in the transfection system revealed phosphorylation not only at S13, but also at nearby Y14, S16, and S17 (Figure 7—figure supplement 1 and Figure 7—source data 1). Of note, mutagenesis of S16 and S17 to non-phosphorylatable alanines reduced both the cleavage and oligomerization of R79H (Figure 7—figure supplement 1). Phospho-motif analysis revealed that S13, S16 and S17 are part of a segment in the GFAP head domain that is a potential target for several kinases (Supplementary file 4). Candidate kinases include casein kinase 2 (CK2), protein kinase A (PKA), PKC, MAP kinase activated protein kinase 2 (MAPKAP2), and glycogen synthase kinase 3 (GSK3). These data suggest that phosphorylation of Ser13 (and nearby S16/17) may promote caspase-6-mediated cleavage of GFAP in the context of AxD mutations. In line with that, we observed increased levels of cleaved GFAP (upon normalization for total GFAP) in the AxD iPSC-astrocytes compared to isogenic control astrocytes (Figure 7E), along with intense caspase-6 staining within perinuclear GFAP aggregates in AxD iPSC-astrocytes, but not isogenic control astrocytes (Figure 7F).

Figure 7 with 1 supplement see all
Download asset Open asset
Phosphorylation of Ser13 on GFAP promotes caspase-6 cleavage of GFAP.

(A) GFAP blot of SW13vim- cells transfected with vector, WT, S13A, S13D and S13E - GFAP. Full-length (fl) and cleaved fragment (cf) of GFAP are indicated by arrows. Immunoblot on the bottom shows GFAP monomer (fl) from the same membrane at a lower exposure. (B) Quantification of panel A by densitometry shows cleaved and full-length GFAP in phospho-mutants relative to WT GFAP (mean ± SD from three independent experiments; *p<0.05 two-way ANOVA). (C) GFAP blot in SW13vim- cells transfected with either S13D or S13D/R79H double mutant GFAP and treated for 48 hr with a caspase-6 inhibitor (Ac-VEID-CHO). (D) Quantification of GFAP bands in panel C by densitometry (mean ± SD from three biological replicates; **p<0.01; ****p<0.0001 two-way ANOVA). (E) Immunoblot for GFAP monomer (fl) and cleaved fragment (cf) in isogenic control and AxD iPSC-astrocytes. Different amounts of total protein were loaded to normalize GFAP monomer levels. (F) Immunofluorescence staining of caspase-6 (magenta), GFAP (green) and DAPI (blue) in human AxD and isogenic control iPSC-astrocytes showing caspase-6 co-localization within GFAP aggregates in the AxD cells, indicated by the arrowheads. Scale bars = 20 µm.

Figure 7—source data 1

Raw data from mass spectrometry PTM profiling of GFAP R79H extracted from transfected SW13vim-cells.

https://cdn.elifesciences.org/articles/47789/elife-47789-fig7-data1-v2.xlsx

Interference with GFAP cleavage by caspase-6 partially reduces aggregation of the phospho-mimic mutant S13D

To determine how blocking GFAP cleavage affects aggregation, we performed site-directed mutagenesis to block cleavage of GFAP at Asp225. As shown in Figure 8A–B, the D225E mutation reduced cleavage of S13D GFAP by >90%. This resulted in partial rescue of filament structure in S13D, although the D225E mutation on its own caused significant filament bundling and perinuclear structures that resembled large aggregates (Figure 8C–D). We also tested the effect of the caspase-6 inhibitor Ac-VEID-CHO, and found that it reduced both the size of the S13D aggregates (Figure 8E) and the presence of ~100 kDa hmm GFAP oligomers (Figure 8F–G). However, similar to the mutagenesis experiment, filament bundles were observed in WT and S13D GFAP treated with Ac-VEID-CHO, suggesting that caspase-6 regulates both aggregation and normal GFAP filament re-organization.

Figure 8
Download asset Open asset
Inhibition of GFAP cleavage by caspase-6 partially alleviates aggregation due to S13D phospho-mimic mutation.

(A) Western blot of GFAP total cell lysates from SW13vim- cells transfected with empty vector control, WT, S13D, D225E, and double S13D/D225E mutants. Shown are GFAP cleaved fragment (cf), full-length (fl) monomer and pan-actin (loading control). (B) Quantification of the abundance of cleaved GFAP in the three mutants shown in panel A relative to WT GFAP (mean ± SD from three biological replicates; ****p<0.0001 compared to S13D; one-way ANOVA). (C) Representative images of immunofluorescence staining of DNA (blue) and GFAP (green) in SW13vim- cells transfected with wild-type GFAP (WT), phospho-mimic GFAP (S13D), and non-cleavable GFAP (D225) as single or double mutations, as noted in the images. Scale bar = 10 µm. (D) Quantification of percentage of cells containing GFAP filaments, aggregates or both (n = 76–85 cells per condition). (E) Representative images of immunofluorescence staining of DNA (blue) and GFAP (green) in SW13vim- cells transfected with wild-type GFAP (WT) or phospho-mimic GFAP (S13D) and treated with vehicle (control) or the caspase-inhibitor Ac-VEID-CHO (10 µM, 48 hr). (F) Western blot analysis of SW13vim- total lysates transfected with S13D GFAP and treated with vehicle (control) or caspase-6 inhibitor Ac-VEID-CHO (10 µM, 24 hr), showing the 24 kDa caspase-cleaved fragment (cf), 50 kDa full-length (fl), and high-molecular-mass ~100 kDa GFAP. (G) Quantification of the relative abundance of hmm GFAP in control and Ac-VEID-CHO – treated cells. n = 3; **p<0.01; unpaired t-test.

Caspase-6 expression and GFAP cleavage are upregulated in AxD patients

Caspase-6 is not expressed highly in the normal human brain, especially after birth (Godefroy et al., 2013). Therefore, we wanted to examine its expression in the context of AxD. Using immunoblot analysis of total brain lysates, we found that caspase-6 is expressed in the brain tissue from all 8 AxD patients who died very young, but is essentially undetectable in the patients who survived longer (Figure 9A). To ensure caspase-6 expression is not simply more abundant in young individuals, we compared brain lysates from young and old AxD patients to non-AxD control brains from age-matched individuals, and observed a significant increase in caspase-6 expression selectively in young AxD patients, but not in the other groups (Figure 9B–C).

Figure 9 with 1 supplement see all
Download asset Open asset
High expression of caspase-6 in young AxD patient brain tissue correlates with increased levels of cleaved GFAP.

(A) Immunoblot for caspase-6 in total lysates from human AxD post mortem brain tissue shows that caspase-6 is upregulated in young AxD patients. Pan-actin is used as a loading control. (B) Immunoblot for caspase-6 in total lysates from young and old non-AxD control and AxD patient post-mortem brain tissue. Pan-actin blot serves as a loading control. (C) Quantification of band intensities in panel B by densitometry of caspase-6 normalized to actin. **p<0.01; two-way ANOVA. (D) Western blotting for full-length GFAP or cleaved GFAP (D225 antibody) in HSEs from human AxD post-mortem brain tissue. (E) Quantification of band intensities from panel D by densitometry of D225, normalized to total GFAP (**p<0.01, unpaired t-test). (F) Immunofluorescence staining showing widespread presence of cleaved GFAP (D225; magenta) in cerebral cortex and underlying white matter of 347 day-old child with AxD and low expression of cleaved GFAP in a 42 year old AxD patient. Wider fields of view and sections from additional patients are shown in Figure 7—figure supplement 1. DAPI nuclei are shown in white in bottom panels, and arrow highlights perinuclear aggregate containing cleaved GFAP and staining positively for DAPI in brain tissue from a child with AxD. Scale bar = 100 µm (top) and 10 µm (bottom).

Next we asked whether AxD patients, particularly young AxD patients that exhibit more pSer13-GFAP and caspase-6 expression, also displayed increased GFAP cleavage. To determine the extent of caspase-6-cleaved GFAP in AxD patient brains, we utilized an antibody that specifically recognizes N-terminally caspase-6-cleaved GFAP (D225) (Chen et al., 2013). We detected cleaved GFAP in extracts from AxD patient brains, and we observed a significant increase in the amount of D225 signal in young AxD patients, which paralleled the increased pSer13 signal in these samples (Figure 9D–E). In agreement with the biochemical evidence, brain tissues from young AxD patients stained intensely for cleaved GFAP, while the signal was significantly weaker in AxD patients who were older (Figure 9F; Figure 9—figure supplement 1). The signal was particularly strong around perinuclear areas and surrounded circular structures that stained positive for DAPI (Figure 9F, bottom panels), similar to what we observed in the AxD iPSC-astrocytes. Thus, our results show that caspase-6 expression in AxD patient brain tissue parallels the presence of cleaved GFAP, and both are selectively and significantly elevated in patients who succumbed to the disease very early in life.

Discussion

Our study reveals that missense mutations, affecting discrete domains on the GFAP molecule, share a common PTM signature that is associated with compromised GFAP proteostasis in the severe form of AxD. Using patient brain tissue and human iPSC-derived AxD astrocytes, we show that head domain phosphorylation promotes defective filament assembly and perinuclear accumulation and incorporation of mutant GFAP within nuclear invaginations. By taking an unbiased mass spectrometry proteomic approach, we were able to identify GFAP phospho-peptides that were selectively elevated in human AxD brain tissue, and subsequently validated these results using a phospho-specific antibody against the most abundant epitope (pSer13-GFAP). We demonstrate the importance of the Ser13 site for GFAP assembly in vitro and in cells. Phospho-mimetic mutation S13D completely abolished the ability of GFAP to form filaments in vitro, without leading to aggregation. In transfected SW13vim- cells, phospho mimic S13D-and S13E-GFAP mutants formed highly abnormal perinuclear aggregates that correlated with increased cleavage of GFAP by caspase-6. We detected a dramatic increase in caspase-6 expression, in association with Ser13 phosphorylation and cleavage of GFAP, in the brain tissue of AxD patients who succumbed to the disease very early in life. While the N-terminal caspase-6 fragment of GFAP promotes filament aggregation in vitro (Chen et al., 2013), presently we do not have direct evidence of cause and effect between caspase-6 cleavage and GFAP aggregation in AxD patient cells. Nevertheless, our current findings provide a basis for exploring PTM-based diagnostic and potential therapeutic strategies in AxD.

Our study does not address whether Ser13 phosphorylation directly promotes caspase cleavage of GFAP, or if these two PTMs are independent markers of an increased cellular stress response in AxD. One possibility is that Ser13 phosphorylation destabilizes the filament structure, thereby promoting access of caspase-6 to the rod domain Asp225 residue, where the cleavage occurs. Another likely possibility is that the increased cleavage of GFAP is an indirect result of stress-dependent caspase-6 activation in the more severe form of AxD. This is supported by previous studies showing that AxD mutations promote activation and nuclear accumulation of p53 (Wang et al., 2015), which can directly induce caspase-6 expression (MacLachlan and El-Deiry, 2002). Future studies in AxD iPSC-astrocytes and animal models will be required to determine the timing of GFAP phosphorylation and caspase-6 activation in relationship to GFAP cleavage and aggregation.

Given our findings that pSer13-GFAP is enriched in the most aggressive form of AxD, monitoring the levels of this phospho-epitope (in addition to total GFAP) in AxD patient cerebrospinal fluid or blood may provide added sensitivity for disease activity (Jany et al., 2015). Phosphorylation of Ser13 by protein kinase C and cAMP-dependent protein kinase was initially described in vitro using purified recombinant GFAP (Inagaki et al., 1990). In the presence of active kinases, Ser-13 phosphorylation occurred in conjunction with phosphorylation at three additional sites (Thr-7, Ser-8, and Ser-34). Phosphorylation of monomeric GFAP at these sites prevented filament assembly, while phosphorylation of in vitro assembled GFAP filaments led to their disassembly (Inagaki et al., 1990). Using the same antibody to pSer13-GFAP that we used in this paper (clone KT13) it was later shown that Aurora-B and Rho-associated kinase phosphorylate GFAP in cultured astrocytoma cells during mitosis (Inagaki et al., 1996). This may bear relevance to AxD, since human and mouse AxD astrocytes with RFs display mitotic abnormalities (Sosunov et al., 2017). However, it was also shown using knock-in mice with the human GFAP head domain that, in vivo, the distribution of pSer13 localization was not limited to mitotic astrocytes, but that select astrocyte populations within multiple regions were pSer13 positive, such as those in the olfactory bulb, subpial regions, and subventricular zone (Takemura et al., 2002b; Hagemann et al., 2005). Interestingly, the regional distribution of pSer13 largely overlaps with areas that are known to be most enriched in RFs in the AxD mouse model (Hagemann et al., 2005). Therefore, this particular phosphorylation event on GFAP may occur during mitosis, or in phenotypically distinct astrocyte populations. This remains to be addressed in the future using the appropriate model systems, as over-expression studies in cancer cell lines (such as the SW13vim- cells we used here) may not be truly reflective of the signaling that occurs in astrocytes. In particular, it remains to be resolved whether phosphorylation of GFAP on Ser13 is part of a sequentially priming phosphorylation cascade involving nearby Ser16/17 (as predicted by the kinase motif analysis) or if Ser16/17 phosphorylation is unique to the SW13 over-expression system. Importantly, identifying the relevant in vivo kinase(s) that phosphorylate GFAP in human AxD may lead to potential novel interventions via kinase inhibition.

Caspase-mediated proteolysis of IF proteins is an important mechanism by which the filament networks re-organize during apoptosis. Although multiple effector caspases are capable of cleaving IF proteins, caspase-6 is frequently implicated in cleavage at a conserved motif within the linker L12 region of the rod domain, which results in the generation of two fragments of similar sizes. This was initially demonstrated to be the case for the type I keratins (Caulín et al., 1997), and later shown to also occur on vimentin (Byun et al., 2001), desmin (Chen et al., 2003), A-type lamins (Ruchaud et al., 2002), and GFAP (Hol and Pekny, 2015). Caspase-6 cleavage of GFAP at 222VELD225 in vitro generates an N-terminal 26 kDa fragment and a C-terminal 24 kDa fragment. The N-terminal fragment directly impairs assembly of full-length GFAP and promotes aggregation in vitro (Chen et al., 2013). Using a specific antibody recognizing the N-terminal GFAP fragment (D225), we show here that GFAP cleavage is significantly increased in AxD tissues from patients presenting with an aggressive form of AxD, and that this parallels elevated expression of caspase-6. This could suggest that misregulation of caspase-6 may contribute to the severity of AxD. However, we were not able to demonstrate in cells that inhibition of caspase-6, or mutagenesis of the cleavage site on GFAP, can resolve aggregate formation. These results point to a more complex function for caspase-6, likely involving cytoskeletal remodeling in response to stress.

Indeed, caspase-6 upregulation has been reported in other neurodegenerative diseases involving protein aggregation, including Huntington’s Disease (HD) and Alzheimer’s Disease (AD) (Albrecht et al., 2007; Graham et al., 2010; Guo et al., 2004). Similar to GFAP, there is a caspase-6 cleavage site on the aggregation-prone proteins in both AD (amyloid precursor protein) and HD (huntingtin). Furthermore, in caspase-6 cleavage-resistant genetic mouse models of both HD and AD, neuronal dysfunction and degeneration are rescued (Graham et al., 2006; Galvan et al., 2006; Saganich et al., 2006). Caspase-6 can promote neurodegeneration via induction of neuronal apoptosis or axon pruning (Geden et al., 2019). However, the functions of caspase-6 in astrocytes are not clear. In the context of human AxD it still remains to be determined which astrocyte populations express caspase-6, and whether it promotes apoptosis or performs a non-apoptotic role, such as sculpting the cytoskeletal architecture in reactive astrocytes. Based on our demonstration that caspase-6 localizes within the perinuclear GFAP inclusions in the AxD iPSC-astrocytes, it is intriguing to speculate that, similar to keratin inclusions in epithelial cells (Dinsdale et al., 2004), RFs sequester active caspases away from other cellular substrates and may protect reactive astrocytes from apoptosis.

Recently, iPSC-derived patient astrocyte models have emerged as an important system for dissecting the cellular mechanisms in AxD. For example, these novel tools have revealed that AxD astrocytes have defects in the secretory pathway, impaired ATP release, and attenuated calcium waves (Jones et al., 2018); that they inhibit oligodendrocyte precursor cell proliferation (Li et al., 2018), providing a potential mechanistic explanation for the degeneration of white matter observed in patients; and that they have defects in mechanotransduction signaling pathways (Wang et al., 2018). A novel aspect of the AxD astrocyte cell model that we generated in our study is the perinuclear accumulation of pSer13-GFAP that was associated with prominent nuclear abnormalities. As such, these patient-derived cells replicate a key phenotypic characteristic of RF-bearing AxD astrocytes in vivo, since nuclear invaginations have been described in electron microscopy studies of AxD mouse models and AxD patient cortex (Sosunov et al., 2017). Another important parallel is that the GFAP inclusions we observe in the AxD patient astrocytes in vitro stain positive for DAPI, and it was shown that DAPI is a reliable and sensitive marker of RFs in human and mouse brain (Sosunov et al., 2017). Therefore patient-derived iPSC-astrocytes provide a unique model system to investigate cytoplasmic-nuclear mechanics in AxD.

Invaginations of the nucleus, such as those we observe here, have been described in physiological and pathological states (Malhas et al., 2011). Control of nuclear shape is critical for regulation of gene expression and response to mechanotransduction signals (Uhler and Shivashankar, 2017). The effects of impaired nuclear morphology can be very severe, as evidenced by mutations in lamin A that lead to defective nuclear morphology in Hutchinson-Gilford Progeria Syndrome (HGPS), where patients experience accelerated aging (Eriksson et al., 2003). An elegant study combining multiple 3D imaging strategies established a direct link between intermediate filaments, actin and the nuclear envelope within nuclear invaginations, and genetic evidence indicates that filamentous actin may play a role in generating these structures (Jorgens et al., 2017; Frost et al., 2016). It is hypothesized that nuclear invaginations provide localized control of gene expression and nuclear-cytoplasmic transport deep within the nucleus, since they have been found to contain calcium receptors and nuclear pores (Malhas et al., 2011). Our study provides the first link between abnormal cytoplasmic PTM processing and perinuclear accumulation of mutant GFAP with nuclear defects, setting the stage to address how nucleo-cytoskeletal coupling is adversely impacted by defective IF proteostasis in AxD and related human diseases.

Materials and methods

Key resources table
Reagent type
(species) or resource		DesignationSource or referenceIdentifiersAdditional
information
Gene (human)GFAPNAGene ID: 2670
Cell line (human)R239C-GFAP fibroblasts from AxD patientCoriell InstituteGM16825Brenner et al. (2001)
Cell line (human)R239C-GFAP induced pluripotent stem cellsGenerated in the studyGeneration of the AxD iPSCs (R239C-GFAP) is described in the
Materials and methods below. Cells can be obtained by contacting the corresponding author.
Cell line (human)R239R-GFAP isogenic control induced pluripotent stem cellsGenerated in the studyGeneration of the AxD iPSCs (R239C-GFAP) is described in the Materials and methods below. Cells can be obtained by contacting the corresponding author.
Cell line (human)SW13 vim-Sarria, A.J et al. J Cell Sci 1994
Biological sample (human)Human brain specimensNIH NeuroBioBankListed inSupplementary files 1 and 2
Antibodyrabbit anti-GFAPAgilent/DAKOClone Z0334Dilution = 1:10,000 immunoblot, 1:500 immunofluorescence
Antibodyrabbit anti-caspase-6Cell Signaling TechnologyCat# 9762Dilution = 1:1000 immunoblot
Antibodyrabbit anti-caspase-6abcamCat# ab185645Dilution = 1:100 immunofluorescence
Antibodyrabbit anti-D225PMID: 24102621gift from Dr Ming Der Perng, Dilution = 1:5000 immunoblot (overnight), 1:150 immunofluorescence
Antibodymouse anti-GFAPSigmaClone GA5, Cat# G3893Dilution = 1:3000 immunoblot, 1:300 immunofluorescence
Antibodymouse anti-pSer13 GFAPPMID: 8647894gift from Dr Masaki Inagaki, Dilution = 1:500 immunoblot (overnight), 1:20 immunofluorescence
Antibodymouse anti-pan ActinNeoMarkersCat# MS-1295Dilution = 1:3000 immunoblot
Antibodymouse anti-Tra-1–60ThermoFisherCat# 41–1000Dilution = 1:300
Antibodymouse anti-Tra-1–81ThermoFisherCat# 41–1100Dilution = 1:300
Antibodymouse anti-SSEA4ThermoFisherCat#41–4000Dilution = 1:300
Antibodyrabbit anti-Oct4abcamCat# ab19857Dilution = 1:40
Antibodyrabbit anti-Sox2ThermoFisherCat# 48–1400Dilution = 1:125
AntibodyAlexa 488-conjugated goat anti-mouseThermoFisherCat# A32723Dilution = 1:500
AntibodyAlexa 488-conjugated goat anti-rabbitThermoFisherCat# A32731Dilution = 1:500
AntibodyAlexa 594-conjugated goat anti-mouseThermoFisherCat# A32742Dilution = 1:500
AntibodyAlexa 594-conjugated goat anti-rabbitThermoFisherCat# A32740Dilution = 1:500
Recombinant DNA reagentpCMV6-XL6-GFAPOrigeneCat# SC118873
Peptide, recombinant proteinTrueCut Cas9 Protein v2ThermoFisherCat# A36499
Commercial assay or kitPrecision gRNA Synthesis KitThermoFisherCat# A29377
Commercial assay or kitAgilent Quikchange IIAgilentCat# 200524
Commercial assay or kitRneasy KitQiagenCat# 74104
Commercial assay or kitTaqman ScorecardThermoFisherCat# A15870
Chemical compound, drugECL ReagentsPerkin ElmerNEL103E001EA
Chemical compound, drugAc-VEID-CHOMillipore SigmaA6339
Software, algorithmCRISPR off-targetPMID: 27380939http://crispor.tefor.net/

Antibodies

The following antibodies were used: rabbit anti-GFAP (DAKO Agilent, Santa Clara, CA, Z0334), rabbit anti-caspase-6 (Cell Signaling Technologies, Danvers, MA, 9762), rabbit anti-Caspase-6 (abcam, Cambridge, UK, ab185645), rabbit anti-D225 (Chen et al., 2013), mouse anti-GFAP (Sigma, GA5), mouse anti-pSer13-GFAP (KT13 [Sekimata et al., 1996]), mouse anti-pan Actin, mouse anti-Tra-1–60, mouse anti-SSEA4, rabbit anti-Oct4, rabbit anti-Sox2, and Alexa 488- and Alexa 594-congujated goat anti mouse or rabbit antibodies (Thermo Fisher Scientific, Waltham, MA).

Cell lines

Request a detailed protocol

SW13vim- cells were provided by Dr Bishr Omary and cultured in DMEM with 10% fetal bovine serum and 1% penicillin-streptomycin. Authentication of the cell line was done by short tandem repeat (STR) profiling by ATCC. Fibroblasts from a male 6 year old type I AxD patient were obtained from the Coriell institute (Camden, NJ). Sanger sequencing was performed to confirm the AxD mutation was present in the cells (c.715C > T; p.Arg239Cys). The cell lines used tested negative for mycoplasma contamination, as assayed using the Universal Mycoplasma Detection Kit (ATCC 30–1012K).

Human brain tissues

Request a detailed protocol

De-identified post-mortem fresh-frozen and fixed AxD patient and control brain tissues were provided by the NIH NeuroBioBank and are described in Supplementary files 1 and 2.

Mass spectrometry

Request a detailed protocol

Sample Preparation: HSEs from AxD patient post-mortem brain cortex tissue were prepared as described previously (Battaglia et al., 2017; Snider and Omary, 2016) and in Figure 1—figure supplement 1, then subjected to SDS-PAGE followed by Coomassie stain. Bands corresponding to GFAP were excised and the proteins were reduced, alkylated, and in-gel digested with trypsin overnight at 37 °C. Peptides were extracted, desalted with C18 spin columns (Pierce – Thermo Fisher Scientific) and dried via vacuum centrifugation. Peptide samples were stored at −80°C until further analysis. LC-MS/MS Analysis: The peptide samples were analyzed by LC/MS/MS using an Easy nLC 1200 coupled to a QExactive HF mass spectrometer (Thermo Fisher Scientific). Samples were injected onto an Easy Spray PepMap C18 column (75 μm id ×25 cm, 2 μm particle size) (Thermo Fisher Scientific) and separated over a 1 hr method. The gradient for separation consisted of 5–40% mobile phase B at a 250 nl/min flow rate, where mobile phase A was 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in 80% ACN. The QExactive HF was operated in data-dependent mode where the 15 most intense precursors were selected for subsequent fragmentation. Resolution for the precursor scan (m/z 300–1600) was set to 120,000 with a target value of 3 × 106 ions. MS/MS scans resolution was set to 15,000 with a target value of 1 × 105 ions. The normalized collision energy was set to 27% for HCD. Dynamic exclusion was set to 30 s, peptide match was set to preferred, and precursors with unknown charge or a charge state of 1 and ≥7 were excluded. Data Analysis: Raw data files were processed using Proteome Discoverer version 2.1 (Thermo Fisher Scientific). Peak lists were searched against a reviewed Uniprot human database, appended with a common contaminants database, using Sequest. The following parameters were used to identify tryptic peptides for protein identification: 10 ppm precursor ion mass tolerance; 0.02 Da product ion mass tolerance; up to two missed trypsin cleavage sites; phosphorylation of Ser, Thr and Tyr were set as variable modifications. The ptmRS node was used to localize the sites of phosphorylation. Peptide false discovery rates (FDR) were calculated by the Percolator node using a decoy database search and data were filtered using a 5% FDR cutoff. The peak areas for the identified peptides were extracted and used for relative quantitation across samples.

Site directed mutagenesis, in vitro assembly, transfections, and immunofluorescence

Request a detailed protocol

Mutagenesis of GFAP (Origene, Rockville, MD, in vector CMV6-XL6) was performed using the QuikChange II mutagenesis kit (Agilent) to generate the designated point mutants. Sanger sequencing of the entire coding sequence of GFAP was performed to confirm the wild-type and mutant sequences. We used established procedures for the purification and in vitro assembly of GFAP (Perng et al., 2016). For transfections, lipofectamine 2000 was used according to the supplier instructions (Invitrogen, Thermo Fisher Scientific, Carlsbad, CA), and experiments were performed 20–24 hr after transfection. For immunofluorescence, cells were fixed in methanol at −20°C for 10 min, washed three times in PBS and incubated in blocking solution (2.5% bovine serum albumin, 2% normal goat serum in PBS) for 1 hr at room temperature. Primary antibodies were diluted into blocking buffer and incubated overnight at 4°C. The next day, cells were washed 3 times in PBS and incubated with Alexa Fluor-conjugated secondary antibodies diluted into blocking buffer for 1 hr at room temperature. Cells were washed 3 times in PBS, incubated in DAPI for 5 min, washed 3 times and mounted in Fluoromount-G (SouthernBiotech, Birmingham, AL) overnight. Cells were imaged on Zeiss 880 confocal laser scanning microscope using a 63x (1.4 NA) oil immersion objective (Zeiss, Jena, Germany).

Preparation of protein lysates and western blotting

Request a detailed protocol

High salt extracts (HSEs) and triton-X (TX) lysates were prepared as previously described (Battaglia et al., 2017). Total lysates were prepared by homogenizing 25 mg tissue directly into hot 2X Tris-Glycine SDS Sample Buffer (Thermo Fisher Scientific) and heating for 5 min at 95°C. Immunoblotting was performed as previously described (Trogden et al., 2018). Briefly, samples were resolved on 4–20% gradient SDS-PAGE gels transferred onto activated polyvinylidene difluoride membranes at 40V overnight. The transferred gels were routinely stained with Coomassie blue and the membranes were blocked in 5% non-fat milk in 0.1% tween 20/PBS (PBST). Post-transfer Coomassie-stained gels served as another loading control where the levels of housekeeping protein (actin) varied (Figure 4E). For immunoblotting, the membranes were incubated with the appropriate primary antibody diluted in 5% milk/PBST, with the exception of KT13, which was incubated in 5% bovine serum albumin/PBST for blocking, primary antibodies and secondary antibodies. Antibodies were detected using ECL reagents (PerkinElmer Life Sciences, Hopkinton, MA). For 2D gel analysis, HSEs were dissolved in 2-D starter kit rehydration/sample buffer (Biorad; 1632106) for separation by isoelectric focusing (IEF). Immobilized pH gradient (IPG) strips (Biorad; 11 cm; pH 4–7; 1632015) were passively rehydrated in 2-D starter kit rehydration/sample buffer overnight. Cup loading method was employed to load the protein samples in cathode side (as isoelectric point of GFAP is 5.2) of the Protean IEF cell tray (Biorad; 1654020). The IEF separation was done using 72000 vh. After IEF separation the protein samples were further separated based on molecular weight using SDS-PAGE gel by applying constant 90 volts.

Cellular reprogramming, characterization and karyotyping of iPSCs

View detailed protocol

Skin fibroblasts were reprogrammed under feeder free conditions using Cytotune –iPS 2.0 Sendai Reprogramming kit and individual iPSC clones were picked for propagation in culture for 10 passages. To confirm stemness and differentiation capabilities of reprogrammed and edited iPSCs, we used the qPCR based TaqMan human Pluripotent Stem Cell Scorecard Panel (Thermo Fisher Scientific). iPSCs were differentiated into all three germ layers using STEMdiff Trilineage Differentiation Kit (StemCell Technologies, Vancouver, Canada), and a monolayer-based protocol was used to directly differentiate hES cells in parallel into the three germ layers (~1 week). Non-differentiated and differentiated cells were lysed and total RNA purified using the RNeasy kit (QIAGEN). RNA reverse transcription was performed following the Taqman Scorecard’s manufacture guidelines and the qRT-PCR was carried out using the QuantStudio 7 Flex Real-Time PCR system. The TaqMan PCR assay combines DNA methylation mapping, gene expression profiling, and transcript counting of lineage marker genes (Bock et al., 2011). Reprogrammed and edited iPSCs were submitted to a standard G-band analysis consisting of 20 metaphase spreads. The analysis (carried out by Karyologic Inc) can identify gender, chromosome number, and detect aberrations that include trisomies, monosomies, deletions, insertions, translocations, duplications, breaks, polyploidy, among others. No abnormalities were found in our cell lines (Figure 4—figure supplement 2A).

CRISPR/Cas9 genome editing

Request a detailed protocol

We used the TrueCut Cas9 Protein V2, sgRNAs and the Neon Transfection system (Thermo Fisher Scientific) to edit iPSCs. The recombinant TrueCut Cas9 V2 was diluted in resuspension buffer R provided in the kit and mixed with 900 ng of sgRNA and 2700 ng of single-stranded donor oligonucleotide, incubated 15 min at room temperature and then a total of 3 × 105 iPSCs were electroporated with the ribonucleoprotein mix. Seventy-two hours after electroporation, cells were dissociated into single cells, diluted, and seeded on Matrigel-coated 96-well plates. Single-cell colonies were selected after two weeks and tested for gene correction. Genomic DNA of single clones was extracted and the gene of interest amplified by PCR using allele specific primers. Sanger sequencing of positive clones demonstrated single or double allele gene correction. Off-target sites within the exons of genes were predicted via selection of the top candidates using the MIT software (CRISPR.mit.edu). The analysis was performed via PCR of 400 bp fragments, which flanked the predicted off-target cut site followed by Sanger sequencing. The chromatograms for edited clones were compared to sequences from the original AxD patient cells.

iPSC culture and astrocyte differentiation

Request a detailed protocol

iPSCs were maintained on Matrigel in StemFlex medium (Thermo Fisher Scientific) and passaged every 3–4 days with 0.5 mM EDTA dissociation solution. iPSCs were differentiated into neural progenitor cells (NPC) using an embryoid body (EB) protocol. Briefly, iPSCs at 80% confluence were collected, resuspended in Neural Induction Medium (NIM, StemCell Technologies) and seeded on one well of an Aggrewell 800 plate (StemCell Technologies) at 3 × 106 cells per well. At day five, EBs were seeded on poly-ornithine and laminin (PLO/LAM)-coated dishes in NIM. Rosette selection was performed after 12 days using Rosette Selection Reagent (StemCell Technologies). NPCs were expanded for 7 days in Neural Progenitor Medium (StemCell Technologies). NPCs were then differentiated into astrocyte precursors by seeding dissociated single cells at 1 × 105 cells/cm2 density on PLO/LAM dishes in STEMdiff astrocyte differentiation medium (StemCell Technologies). Astrocyte precursors were maintained for 20 days with medium changes every 48 hr and splitting every week with Accutase (Millipore, Burlington, MA). Astrocytes were expanded for up to 120 days in STEMdiff astrocyte maturation medium (StemCell Technologies).

Transmission electron microscopy

Request a detailed protocol

AxD iPSC-astrocytes grown on a polystyrene dish were fixed in 2.5% glutaraldehyde in 0.1M sodium cacodylate buffer, pH 7.4, for one hour at room temperature and stored at 4°C. The cells were washed 3 times in 0.1M sodium cacodylate buffer followed by post-fixation in 1% buffered osmium tetroxide for 1 hr. After three washes in deionized water, the cells were dehydrated in ethanol, infiltrated and embedded in situ in PolyBed 812 epoxy resin (Polysciences, Inc, Warrington, PA). The cell monolayer was sectioned en face to the substrate with a diamond knife and Leica UCT Ultramicrotome (Leica Microsystems, Inc, Buffalo Grove, IL). Ultrathin sections (70 nm) were mounted on 200 mesh copper grids and stained with 4% uranyl acetate and lead citrate. The sections were observed and digital images were taken using a JEOL JEM-1230 transmission electron microscope operating at 80kV (JEOL USA, Inc, Peabody, MA) equipped with a Gatan Orius SC1000 CCD Digital Camera (Gatan, Inc, Pleasanton, CA).

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files.Source data files for mass spectrometry results in Figure 2 and Figure 7 are provided in Figure 2—source data 1 and Figure 7—source data 1, respectively.

References

    1. Inagaki M
    2. Gonda Y
    3. Nishizawa K
    4. Kitamura S
    5. Sato C
    6. Ando S
    7. Tanabe K
    8. Kikuchi K
    9. Tsuiki S
    10. Nishi Y
    (1990)
    Phosphorylation sites linked to glial filament disassembly in vitro locate in a Non-α-helical head domain
    The Journal of Biological Chemistry 265:4722–4729.

Decision letter

  1. Huda Y Zoghbi
    Senior Editor; Texas Children's Hospital, United States
  2. Kang Shen
    Reviewing Editor; Howard Hughes Medical Institute, Stanford University, United States
  3. W Christopher Risher
    Reviewer; Duke University Medical Center, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

We are particularly excited about the findings that Alexander Disease-causing mutations perturb key post-translational modifications (PTMs) on GFAP. The selective phosphorylation of GFAP-Ser13 in patients who died young raises the hypothesis that Ser13 phosphorylation and increased GFAP proteolysis by caspase-6 facilitated GFAP aggregation, which leads to disease state.

Decision letter after peer review:

Thank you for submitting your article "Phosphorylation-dependent proteolysis perturbs GFAP proteostasis, implicating caspase-6 in early onset Alexander Disease" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Huda Zoghbi as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: W Christopher Risher (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

Battaglia and co-authors have identified a posttranslational modification of the GFAP protein and potential mechanism involving caspase-6 that appear to correlate with disease severity in the rare but lethal disorder Alexander disease (AxD). In this report post-mortem samples from individuals with known GFAP mutations were analyzed for GFAP-PTM by mass spectrometry. Phospho-Ser13 was identified primarily in samples from AxD patients with early onset, regardless of the mutational variant. Subsequent transfection experiments with Ser13 phospho- and de-phospho mimics further suggest that Ser13 phosphorylation promotes protein aggregation. In addition, the authors show that pSer13-GFAP along with caspase-6 localize to aggregates found in astrocytes derived from AxD iPSCs, and that the phospho-mimic S13D leads to caspase-6 cleavage of GFAP, similar to AxD mutations. Finally, they show that pSer13, caspase-6 protein, and caspase-6 cleaved GFAP are present in brains from early onset AxD patients.

Essential revisions:

All three reviewers found the question interesting and the results potentially significant.

1) The major critique from all three reviewers is that the causality between phosphorylation of GFAP and cleavage of caspase-6 in relation to the pathology of AxD are not explored. For example, does blocking phosphorylation of this site on GFAP in the patient iPSC astrocytes prevent/reverse the GFAP aggregation and perinuclear accumulation of GFAP? Are there known effects on astrocyte health/reactivity in addition to GFAP aggregation that can be monitored, to see if they are also rescued? Experiments along these lines would greatly strengthen the study.

Specific comments along this line include "The key pieces of evidence suggesting cause and effect are the transfection experiments with phopho-mimetic constructs. Although N-terminal substitutions are rare in AxD, how do we know that the S13D mutation does not cause GFAP aggregation in itself (multiple AxD associated mutations have been identified at Ser385, 393 and 398)? Did the authors try other substitutions, e.g. S13V, S13E (or D225E to prevent caspase-6 cleavage)? Since it is known that over-expression of wild-type GFAP causes aggregation, how are GFAP expression levels controlled between transfections? If phosphorylation of GFAP-Ser13 promotes caspase-6 cleavage and GFAP aggregation, shouldn't elimination of pSer13 alleviate aggregate formation in the R79H/S13A transfection? Would caspase-6 inhibitors (Ac-VEID-CHO) prevent aggregation in either the SW13vim-cells or iPSC astrocytes? The authors claim in the Discussion that impairment of GFAP assembly by the S13D phospho-mimic is due to caspase-6 cleavage, but this link is not demonstrated by the data shown. "

2) The Western blots in the figures (Figure 3E, Figure 6B, Figure 7B) used Coomassie staining as the loading control, but no details were given. Given that antibody labeling of Coomassie-stained gels is notoriously difficult, did the authors use different gels for determining protein amounts vs. blotting? This would not be an appropriate normalization method for the actual blotted membranes, and would therefore need to be replaced. If these Coomassie controls were done on the same gels that went on to be blotted for antibodies, please provide technical details in the Materials and methods, as this is a non-standard approach.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Phosphorylation-dependent proteolysis perturbs GFAP proteostasis, implicating caspase-6 in early onset Alexander Disease" for further consideration at eLife. Your revised article has been favorably evaluated by Huda Zoghbi (Senior Editor), a Reviewing Editor, and three reviewers.

The reviewers agreed that the manuscript has been improved and is a strong candidate for eLife, but there are some remaining issues that need to be addressed before acceptance, as outlined below:

The main remaining issue is on how strong the conclusions can be drawn. Please make modifications to the text and potentially to the title to reflect the concerns of reviewer 3.

Reviewer #1:

The authors have added a number of new experiments to the paper that strengthen the conclusion linking phosphorylation of GFAP to caspase-6 cleavage of GFAP and to the pathology of Alexander disease. While major point 1 is not fully addressed (determining whether these changes are a cause or a consequence of more severe disease mutations), the paper does make significant new findings in the area of Alexander disease that warrant publication in eLife.

Reviewer #2:

The authors have satisfactorily addressed all of my major and minor concerns; I have no further issues. Additionally, the mechanistic insight provided by the new experiments and analyses has significantly elevated the impact of the study.

Reviewer #3:

The authors have made a significant effort in adding important data to address the reviewers' concerns and questions. These include in vitro assembly analysis for the GFAP S13D phospho-mimic, the addition of the S13E phospho-mimic and analysis of GFAP-R79H phosphorylation sites in transfection experiments (SW13vim-), and both genetic and pharmacological inhibition of caspase-6 to assess the effects of GFAP cleavage on protein aggregation in SW13vim- cells. Important controls were also added for the iPS cell and AxD patient experiments. Most of these additions have strengthened the manuscript and support a connection between GFAP phosphorylation and aggregation. However, the new data showing caspase-6 inhibition in itself disrupts GFAP filament organization and fails to rescue S13D aggregation complicate interpretation of the results, and evidence for direct cause and effect between caspase-6 GFAP cleavage and protein aggregation is still lacking. The findings in this report are important in identifying PTMs as new markers of disease severity and should be published, but as noted in the original review, the authors should be more conservative with their conclusions.

The authors indicate a direct link between GFAP-pSer13, caspase-6 cleavage, and protein aggregation in the title, Introduction and Discussion. A major question is still whether the observed phenomena contribute to disease severity or are downstream in the stress response and simply reflect the more severe pathology in early onset AxD. The following 3 areas contribute to this concern:

1) In agreement with a more severe phenotype and astrocyte response, GFAP levels are higher in early onset AxD cases compared to those with late onset (Figure 9D). GFAP has been shown to be phosphorylated in reactive gliosis associated with certain injury models (Takemura et al., 2002), and the overall damage response could have a broader effect in activating caspase-6; p53 for example is activated in AxD (Wang et al., 2015) and is known to transactivate caspase-6 (MacLachlan and El-Deiry, 2002). The authors hypothesize that GFAP phosphorylation directly affects caspase-6 accessibility, but aggregation of GFAP-S13D and S13E in SW13vim- cells likely causes a cellular stress response that could also indirectly activate caspase-6. While I agree with the authors that investigating these mechanisms is beyond the scope of the current manuscript, they should address this possibility in their discussion and conclusions.

2) New experiments to inhibit caspase-6 cleavage of GFAP produce ambiguous results. Inhibition of caspase-6 alone leads to GFAP bundling in SW13vim- cells and in combination with the S13D phospho-mimic, aggregates are not resolved. Western analysis shows clearance of a 100 kD (hmm) band, but a ~200 kD band remains and no western data is shown to demonstrate the level of WT-GFAP oligomerization induced by caspase-6 inhibition. As the authors suggest in the discussion, the role of caspase-6 activation is complex and may also be involved in cytoskeletal reorganization.

3) The supplemental data added to Figure 7 introduce important evidence for a link between GFAP phosphorylation, aggregation and caspase-6 cleavage. Immunoblotting shows de-phospho-mimics of serine residues (S to A) within the S13-S17 motif of mutant R79H GFAP eliminate high molecular weight oligomerization and significantly reduce caspase-6 cleavage fragments. However, these data are complicated by the limited contribution of pSer13 in GFAP-R79H aggregation in SW13vim- cells and the more dominant effects of pS16 and pS17, which have not been otherwise explored in this report. Although the SW13vim- cell line is commonly used to test the effects of GFAP mutation, these results demonstrate the limitations of the system for these experiments. Future work should include similar experiments in the iPSC derived astrocytes.

https://doi.org/10.7554/eLife.47789.sa1

Author response

Essential revisions:

All three reviewers found the question interesting and the results potentially significant.

1) The major critique from all three reviewers is that the causality between phosphorylation of GFAP and cleavage of caspase-6 in relation to the pathology of AxD are not explored. For example, does blocking phosphorylation of this site on GFAP in the patient iPSC astrocytes prevent/reverse the GFAP aggregation and perinuclear accumulation of GFAP? Are there known effects on astrocyte health/reactivity in addition to GFAP aggregation that can be monitored, to see if they are also rescued? Experiments along these lines would greatly strengthen the study.

We present new experimental evidence (Figure 7—figure supplement 1; Figure 7—source data 1) and kinase motif-based predictions (Supplementary file 4) revealing the complexity of this PTM mechanism on GFAP (e.g. additional effects of nearby phosphorylation sites S16/17). Although we are actively working to identify the critical kinase(s) involved in the phosphorylation of S13 (through candidate and unbiased approaches), we currently do not have strong and conclusive evidence for the specific involvement of a major kinase that we can selectively inhibit to measure the effects. However, we fully agree with the reviewers that this is of key importance and it is an avenue that we are actively exploring, using a combination of in vitro and in vivo approaches to find the most promising target for intervention.

Specific comments along this line include "The key pieces of evidence suggesting cause and effect are the transfection experiments with phopho-mimetic constructs. Although N-terminal substitutions are rare in AxD, how do we know that the S13D mutation does not cause GFAP aggregation in itself (multiple AxD associated mutations have been identified at Ser385, 393 and 398)?

To address this question, we purified WT-, S13A-, and S13D-GFAP and performedin vitro filament assembly assays. As shown in new Figure 3C, the S13D mutant is unable to assemble into filaments, and it does not aggregate on its own. This is in contrast to the phenotype of the N-terminal caspase-cleaved fragment, which does aggregate on its own (please see Figure 2B of Chen et al., 2013).

Did the authors try other substitutions, e.g. S13V, S13E (or D225E to prevent caspase-6 cleavage)?

We now include new data on the effects of an S13E substitution, which phenocopied S13D with respect to aggregation (Figure 3A-B) and caspase cleavage (Figure 7A-B). We also generated D225E mutant alone, and in combination with S13D. We present new data to show that D225E mutation essentially eliminates S13D-GFAP cleavage, and that this correlates with a decrease in aggregation (Figure 8A-D). We note that D225E on its own interfered with GFAP filament formation, which is not surprising as caspase-cleavage is an important mechanism used to re-organize IFs.

Since it is known that over-expression of wild-type GFAP causes aggregation, how are GFAP expression levels controlled between transfections?

For all transfection experiments, we initially carried out titration of WT-GFAP cDNA amounts (250ng-1µg) in a 24-well format using three different transfection reagents (Lipofectamine 2000, 3000, and LTX). We combined this with a time-course experiments (24, 48, 72h). Based on these initial optimization experiments, we selected to use Lipofectamine 2000/500ng cDNA/24hr conditions. Under these conditions, there is minimal aggregation of WT-GFAP (based on protein solubility in TritonX-100 buffer and filament organization). In contrast, under these same conditions, AxD-associated GFAP mutants are significantly less soluble and form the expected aggregates. We now show this as a supplement (Figure 3—figure supplement 1).

If phosphorylation of GFAP-Ser13 promotes caspase-6 cleavage and GFAP aggregation, shouldn't elimination of pSer13 alleviate aggregate formation in the R79H/S13A transfection?

We present new data that R79H in the over-expression system is highly phosphorylated at nearby serines 16/17 (Figure 7—figure supplement 1; Figure 7—source data 1). Mutagenesis of these residues to non-phosphorylatable alanines decreases GFAP cleavage and oligomerization of the R79H mutant. Therefore, our current hypothesis is that phosphorylation of S13 is a coordinated event that works in tandem with nearby phosphorylation sites. This is also predicted by the kinase motif analysis (Supplementary file 4). Another contributing factor may be that the S13A substitution leads to the formation of abnormal filaments, based on the in vitro assembly assay (Figure 3C). Given these results, it is not surprising that S13A does not alleviate R79H aggregation.

Would caspase-6 inhibitors (Ac-VEID-CHO) prevent aggregation in either the SW13vim-cells or iPSC astrocytes?

These data are shown in new Figure 8E-G. In the presence of the caspase-6 inhibitor, S13D aggregates in the SW13vim- cells were reduced in size, but we observed more filament bundles (consistent with previous studies on other IFs showing that caspase cleavage is required for filament re-organization). Additionally, the presence of a high-molecular-mass (100kDa) S13D-GFAP oligomer was significantly reduced by caspase-6 inhibition.

The authors claim in the Discussion that impairment of GFAP assembly by the S13D phospho-mimic is due to caspase-6 cleavage, but this link is not demonstrated by the data shown.

We modified this section of the Discussion to include the new results with the recombinant protein and propose a potential mechanism regarding the aggregation mechanism of the S13D mutant.

2) The Western blots in the figures (Figure 3E, Figure 6B, Figure 7B) used Coomassie staining as the loading control, but no details were given. Given that antibody labeling of Coomassie-stained gels is notoriously difficult, did the authors use different gels for determining protein amounts vs. blotting? This would not be an appropriate normalization method for the actual blotted membranes, and would therefore need to be replaced. If these Coomassie controls were done on the same gels that went on to be blotted for antibodies, please provide technical details in the Materials and methods, as this is a non-standard approach.

The stained gels are the same gels that were transferred onto the PVDF membranes. We routinely stain all gels after wet transfer overnight at 40V. Regardless of transfer technique, there is typically protein remaining on the gel after the transfer of total cell and total tissue lysates. We find that in most cases this is a more reliable loading control than a ‘housekeeping’ protein, which can change with conditions. However, to address the concern regarding equal protein loading, we re-ran the gels in question and included pan-actin blots for the membranes in new Figure 4E (old Figure 3E) and new Figure 9B (old Figure 7B) to show that our original conclusions remain. We also modified the Materials and methods section to include information about post-transfer gel staining.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The main remaining issue is on how strong the conclusions can be drawn. Please make modifications to the text and potentially to the title to reflect the concerns of reviewer 3.

Reviewer #1:

The authors have added a number of new experiments to the paper that strengthen the conclusion linking phosphorylation of GFAP to caspase-6 cleavage of GFAP and to the pathology of Alexander disease. While major point 1 is not fully addressed (determining whether these changes are a cause or a consequence of more severe disease mutations), the paper does make significant new findings in the area of Alexander disease that warrant publication in eLife.

We thank reviewer #1 for their careful and positive evaluation of our revised manuscript. We have made changes to the title, Introduction, and Discussion to clarify that causality was not established with regards to the PTMs and AxD disease severity, and also in the answers to reviewer 3 comments below.

Reviewer #2:

The authors have satisfactorily addressed all of my major and minor concerns; I have no further issues. Additionally, the mechanistic insight provided by the new experiments and analyses has significantly elevated the impact of the study.

We thank reviewer #2 for their careful and positive evaluation of our revised manuscript and the new data.

Reviewer #3:

The authors have made a significant effort in adding important data to address the reviewers' concerns and questions. These include in vitro assembly analysis for the GFAP S13D phospho-mimic, the addition of the S13E phospho-mimic and analysis of GFAP-R79H phosphorylation sites in transfection experiments (SW13vim-), and both genetic and pharmacological inhibition of caspase-6 to assess the effects of GFAP cleavage on protein aggregation in SW13vim- cells. Important controls were also added for the iPS cell and AxD patient experiments. Most of these additions have strengthened the manuscript and support a connection between GFAP phosphorylation and aggregation.

We thank reviewer #3 for their careful evaluation of our revised manuscript and our new data further linking GFAP phosphorylation and aggregation.

However, the new data showing caspase-6 inhibition in itself disrupts GFAP filament organization and fails to rescue S13D aggregation complicate interpretation of the results, and evidence for direct cause and effect between caspase-6 GFAP cleavage and protein aggregation is still lacking. The findings in this report are important in identifying PTMs as new markers of disease severity and should be published, but as noted in the original review, the authors should be more conservative with their conclusions.

We agree with the reviewer that additional experiments would be required to conclusively establish if there is a direct cause and effect between GFAP caspase-6 cleavage and aggregation in AxD patient cells. We modified the Introduction and Discussion to reflect that point, as noted in the responses to the specific comments below. We also agree that the PTMs highlighted in our report represent new markers of AxD severity, which is now reflected in the revised title of the manuscript.

The authors indicate a direct link between GFAP-pSer13, caspase-6 cleavage, and protein aggregation in the title, Introduction and Discussion. A major question is still whether the observed phenomena contribute to disease severity or are downstream in the stress response and simply reflect the more severe pathology in early onset AxD.

We agree with the reviewer that more work is needed to establish where in the disease process these PTMs occur. To address this concern, we revised the statements indicating a direct link and made the following additional changes:

1) Title was changed to: “Site-specific phosphorylation and caspase cleavage of GFAP are new markers of Alexander Disease severity”

2) We added the following statement to the Introduction: “Collectively, our results reveal a new PTM signature that is associated with defective GFAP proteostasis in the most severe form of AxD. Future interventional studies targeting these PTMs will determine whether they contribute to, or are the consequence of, disease severity.”

3) We added the following statement to the Discussion: “While the N-terminal caspase-6 fragment of GFAP promotes filament aggregation in vitro (Chen et al., 2013), presently we do not have direct evidence of cause and effect between caspase-6 cleavage and GFAP aggregation in AxD patient cells.”

The following 3 areas contribute to this concern:

1) In agreement with a more severe phenotype and astrocyte response, GFAP levels are higher in early onset AxD cases compared to those with late onset (Figure 9D). GFAP has been shown to be phosphorylated in reactive gliosis associated with certain injury models (Takemura et al., 2002), and the overall damage response could have a broader effect in activating caspase-6; p53 for example is activated in AxD (Wang et al., 2015) and is known to transactivate caspase-6 (MacLachlan and El-Deiry, 2002). The authors hypothesize that GFAP phosphorylation directly affects caspase-6 accessibility, but aggregation of GFAP-S13D and S13E in SW13vim- cells likely causes a cellular stress response that could also indirectly activate caspase-6. While I agree with the authors that investigating these mechanisms is beyond the scope of the current manuscript, they should address this possibility in their discussion and conclusions.

We agree that the PTMs could be upstream or downstream of the stress response, and have revised the text to reflect this point as follows:

1) We added the following statement to the Introduction: “Finally, we demonstrate a correlation between site-specific GFAP phosphorylation and caspase cleavage in cells and in post-mortem brain tissue from AxD patients. Although our study does not establish a causal relationship between GFAP phosphorylation and caspase cleavage, we show that caspase-6 is a new marker for the most severe form of human AxD.”

2) We added the following paragraph to the Discussion: “Our study does not address whether Ser13 phosphorylation directly promotes caspase cleavage of GFAP, or if these two PTMs are independent markers of an increased cellular stress response in AxD. One possibility is that Ser13 phosphorylation destabilizes the filament structure, thereby promoting access of caspase-6 to the rod domain Asp225 residue, where the cleavage occurs. Another likely possibility is that the increased cleavage of GFAP is an indirect result of stress-dependent caspase-6 activation in the more severe form of AxD. This is supported by previous studies showing that AxD mutations promote activation and nuclear accumulation of p53 (Wang et al., 2015), which can directly induce caspase-6 expression (MacLachlan and El-Deiry, 2002). Future studies in AxD iPSC-astrocytes and animal models will be required to determine the timing of GFAP phosphorylation and caspase-6 activation in relationship to GFAP cleavage and aggregation.”

2) New experiments to inhibit caspase-6 cleavage of GFAP produce ambiguous results. Inhibition of caspase-6 alone leads to GFAP bundling in SW13vim- cells and in combination with the S13D phospho-mimic, aggregates are not resolved. Western analysis shows clearance of a 100 kD (hmm) band, but a ~200 kD band remains and no western data is shown to demonstrate the level of WT-GFAP oligomerization induced by caspase-6 inhibition. As the authors suggest in the discussion, the role of caspase-6 activation is complex and may also be involved in cytoskeletal reorganization.

We agree that the role of caspase-6 is complex and requires thorough investigation. With respect to the hmm GFAP bands, we focused primarily on the 100kDa band in the S13D mutant because of the significant difference in the intensity of this band compared to WT GFAP (Figure 7A), and after caspase-6 inhibition (Figure 8F). We hypothesize that the 200kDa band likely reflects presence of normal GFAP tetramer, but we have not tested that specifically.

We also added the following statement to the Discussion:“However, we were not able to demonstrate in cells that inhibition of caspase-6, or mutagenesis of the cleavage site on GFAP, can resolve aggregate formation. These results point to a more complex function for caspase-6, likely involving cytoskeletal remodeling in response to stress.”

3) The supplemental data added to Figure 7 introduce important evidence for a link between GFAP phosphorylation, aggregation and caspase-6 cleavage. Immunoblotting shows de-phospho-mimics of serine residues (S to A) within the S13-S17 motif of mutant R79H GFAP eliminate high molecular weight oligomerization and significantly reduce caspase-6 cleavage fragments. However, these data are complicated by the limited contribution of pSer13 in GFAP-R79H aggregation in SW13vim- cells and the more dominant effects of pS16 and pS17, which have not been otherwise explored in this report. Although the SW13vim- cell line is commonly used to test the effects of GFAP mutation, these results demonstrate the limitations of the system for these experiments. Future work should include similar experiments in the iPSC derived astrocytes.

We agree that the limitations of the various systems need to be taken into account, and have emphasized this point in the Discussion: “This remains to be addressed in the future using the appropriate model systems, as over-expression studies in cancer cell lines (such as the SW13vim- cells we used here) may not be truly reflective of the signaling that occurs in astrocytes. In particular, it remains to be resolved whether phosphorylation of GFAP on Ser13 is part of a sequentially priming phosphorylation cascade involving nearby Ser16/17 (as predicted by the kinase motif analysis) or if Ser16/17 phosphorylation is unique to the SW13 over-expression system.”

https://doi.org/10.7554/eLife.47789.sa2

Article and author information

Author details

  1. Rachel A Battaglia

    Department of Cell Biology and Physiology, University of North Carolina, Chapel Hill, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Funding acquisition, Validation, Investigation, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared

  2. Adriana S Beltran

    1. Department of Pharmacology, University of North Carolina, Chapel Hill, United States
    2. Human Pluripotent Stem Cell Core, University of North Carolina, Chapel Hill, United States
    Contribution
    Formal analysis, Supervision, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared

  3. Samed Delic

    1. Department of Cell Biology and Physiology, University of North Carolina, Chapel Hill, United States
    2. Department of Biosciences, University of Durham, Durham, United Kingdom
    Contribution
    Formal analysis, Validation, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared

  4. Raluca Dumitru

    Human Pluripotent Stem Cell Core, University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation, Formal analysis, Validation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared

  5. Jasmine A Robinson

    Department of Cell Biology and Physiology, University of North Carolina, Chapel Hill, United States
    Contribution
    Formal analysis, Validation, Investigation, Writing—review and editing
    Competing interests
    No competing interests declared

  6. Parijat Kabiraj

    Department of Cell Biology and Physiology, University of North Carolina, Chapel Hill, United States
    Present address
    Department of Neurology, Mayo clinic, Rochester, United States
    Contribution
    Data curation, Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared

  7. Laura E Herring

    Department of Pharmacology, University of North Carolina, Chapel Hill, United States
    Contribution
    Data curation, Formal analysis, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared

  8. Victoria J Madden

    Department of Pathology, University of North Carolina, Chapel Hill, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7909-7743

  9. Namritha Ravinder

    Thermo Fisher Scientific, Carlsbad, United States
    Contribution
    Methodology, Writing—review and editing
    Competing interests
    are paid employees of ThermoFisher Scientific, whose products were used to complete parts of the study. ThermoFisher Scientific had no role in the study design, data analysis, decision to publish, or preparation of the manuscript.

  10. Erik Willems

    Thermo Fisher Scientific, Carlsbad, United States
    Contribution
    Formal analysis, Investigation, Methodology, Writing—review and editing
    Competing interests
    are paid employees of ThermoFisher Scientific, whose products were used to complete parts of the study. ThermoFisher Scientific had no role in the study design, data analysis, decision to publish, or preparation of the manuscript.

  11. Rhonda A Newman

    Thermo Fisher Scientific, Carlsbad, United States
    Contribution
    Methodology, Writing—review and editing
    Competing interests
    are paid employees of ThermoFisher Scientific, whose products were used to complete parts of the study. ThermoFisher Scientific had no role in the study design, data analysis, decision to publish, or preparation of the manuscript.

  12. Roy A Quinlan

    Department of Biosciences, University of Durham, Durham, United Kingdom
    Contribution
    Resources, Supervision, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0644-4123

  13. James E Goldman

    Department of Pathology, Columbia University, New York, United States
    Contribution
    Resources, Formal analysis, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared

  14. Ming-Der Perng

    Institute of Molecular Medicine, National Tsing Hua University, Hsinchu, Taiwan, Republic of China
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared

  15. Masaki Inagaki

    Department of Physiology, Mie University Graduate School of Medicine, Mie, Japan
    Contribution
    Resources, Writing—review and editing
    Competing interests
    No competing interests declared

  16. Natasha T Snider

    Department of Cell Biology and Physiology, University of North Carolina, Chapel Hill, United States
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    ntsnider@med.unc.edu
    Competing interests
    is a member of the Scientific Advisory Board for Elise's Corner Fund, which supported part of this work
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7663-4585

Funding

Elise's Corner Fund (Research Grant)

  • Natasha T Snider

United Leukodystrophy Foundation (Research Grant)

  • Natasha T Snider

National Science Foundation (Graduate Research Fellowship)

  • Rachel A Battaglia

University of North Carolina at Chapel Hill (Department of Cell Biology and Physiology)

  • Natasha T Snider

National Institutes of Health (P30 CA016086)

  • Victoria J Madden

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by grants from Elise’s Corner Fund and the United Leukodystrophy Foundation (to NTS), the National Science Foundation Graduate Research Fellowship (to RAB) and the Department of Cell Biology and Physiology at UNC-Chapel Hill. The authors thank Kristen White, Microscopy Services Laboratory, for her assistance with EM specimen preparation. The Microscopy Services Laboratory, Department of Pathology and Laboratory Medicine, is supported in part by P30 CA016086 Cancer Center Core Support Grant to the UNC Lineberger Comprehensive Cancer Center. The authors thank all the patients and their families for donating tissue that enabled this research.

Senior Editor

  1. Huda Y Zoghbi, Texas Children's Hospital, United States

Reviewing Editor

  1. Kang Shen, Howard Hughes Medical Institute, Stanford University, United States

Reviewer

  1. W Christopher Risher, Duke University Medical Center, United States

Version history

  1. Received: April 18, 2019
  2. Accepted: November 4, 2019
  3. Accepted Manuscript published: November 4, 2019 (version 1)
  4. Version of Record published: December 23, 2019 (version 2)

Copyright

© 2019, Battaglia et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 3,016
    Page views
  • 433
    Downloads
  • 22
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Rachel A Battaglia
  2. Adriana S Beltran
  3. Samed Delic
  4. Raluca Dumitru
  5. Jasmine A Robinson
  6. Parijat Kabiraj
  7. Laura E Herring
  8. Victoria J Madden
  9. Namritha Ravinder
  10. Erik Willems
  11. Rhonda A Newman
  12. Roy A Quinlan
  13. James E Goldman
  14. Ming-Der Perng
  15. Masaki Inagaki
  16. Natasha T Snider
(2019)
Site-specific phosphorylation and caspase cleavage of GFAP are new markers of Alexander disease severity
eLife 8:e47789.
https://doi.org/10.7554/eLife.47789

Categories and tags

Research organism

Further reading

    1. Cell Biology
    2. Microbiology and Infectious Disease

    Synchronization of oscillatory growth prepares fungal hyphae for fusion

    Valentin Wernet, Marius Kriegler ... Reinhard Fischer
    Research Article Updated

    Communication is crucial for organismic interactions, from bacteria, to fungi, to humans. Humans may use the visual sense to monitor the environment before starting acoustic interactions. In comparison, fungi, lacking a visual system, rely on a cell-to-cell dialogue based on secreted signaling molecules to coordinate cell fusion and establish hyphal networks. Within this dialogue, hyphae alternate between sending and receiving signals. This pattern can be visualized via the putative signaling protein Soft (SofT), and the mitogen-activated protein kinase MAK-2 (MakB) which are recruited in an alternating oscillatory manner to the respective cytoplasmic membrane or nuclei of interacting hyphae. Here, we show that signal oscillations already occur in single hyphae of Arthrobotrys flagrans in the absence of potential fusion partners (cell monologue). They were in the same phase as growth oscillations. In contrast to the anti-phasic oscillations observed during the cell dialogue, SofT and MakB displayed synchronized oscillations in phase during the monologue. Once two fusion partners came into each other’s vicinity, their oscillation frequencies slowed down (entrainment phase) and transit into anti-phasic synchronization of the two cells’ oscillations with frequencies of 104±28 s and 117±19 s, respectively. Single-cell oscillations, transient entrainment, and anti-phasic oscillations were reproduced by a mathematical model where nearby hyphae can absorb and secrete a limited molecular signaling component into a shared extracellular space. We show that intracellular Ca2+ concentrations oscillate in two approaching hyphae, and depletion of Ca2+ from the medium affected vesicle-driven extension of the hyphal tip, abolished the cell monologue and the anti-phasic synchronization of two hyphae. Our results suggest that single hyphae engage in a ‘monologue’ that may be used for exploration of the environment and can dynamically shift their extracellular signaling systems into a ‘dialogue’ to initiate hyphal fusion.

    1. Cell Biology
    2. Microbiology and Infectious Disease

    A CRISPR-based rapid DNA repositioning strategy and the early intranuclear life of HSV-1

    Juan Xiang, Chaoyang Fan ... Pei Xu
    Research Article Updated

    The relative positions of viral DNA genomes to the host intranuclear environment play critical roles in determining virus fate. Recent advances in the application of chromosome conformation capture-based sequencing analysis (3 C technologies) have revealed valuable aspects of the spatiotemporal interplay of viral genomes with host chromosomes. However, to elucidate the causal relationship between the subnuclear localization of viral genomes and the pathogenic outcome of an infection, manipulative tools are needed. Rapid repositioning of viral DNAs to specific subnuclear compartments amid infection is a powerful approach to synchronize and interrogate this dynamically changing process in space and time. Herein, we report an inducible CRISPR-based two-component platform that relocates extrachromosomal DNA pieces (5 kb to 170 kb) to the nuclear periphery in minutes (CRISPR-nuPin). Based on this strategy, investigations of herpes simplex virus 1 (HSV-1), a prototypical member of the human herpesvirus family, revealed unprecedently reported insights into the early intranuclear life of the pathogen: (I) Viral genomes tethered to the nuclear periphery upon entry, compared with those freely infecting the nucleus, were wrapped around histones with increased suppressive modifications and subjected to stronger transcriptional silencing and prominent growth inhibition. (II) Relocating HSV-1 genomes at 1 hr post infection significantly promoted the transcription of viral genes, termed an ‘Escaping’ effect. (III) Early accumulation of ICP0 was a sufficient but not necessary condition for ‘Escaping’. (IV) Subnuclear localization was only critical during early infection. Importantly, the CRISPR-nuPin tactic, in principle, is applicable to many other DNA viruses.

    1. Cell Biology

    Mitochondrial defects caused by PARL deficiency lead to arrested spermatogenesis and ferroptosis

    Enrico Radaelli, Charles-Antoine Assenmacher ... Marco Spinazzi
    Research Article Updated

    Impaired spermatogenesis and male infertility are common manifestations associated with mitochondrial diseases, yet the underlying mechanisms linking these conditions remain elusive. In this study, we demonstrate that mice deficient for the mitochondrial intra-membrane rhomboid protease PARL, a recently reported model of the mitochondrial encephalopathy Leigh syndrome, develop early testicular atrophy caused by a complete arrest of spermatogenesis during meiotic prophase I, followed by degeneration and death of arrested spermatocytes. This process is independent of neurodegeneration. Interestingly, genetic modifications of PINK1, PGAM5, and TTC19 – three major substrates of PARL with important roles in mitochondrial homeostasis – fail to reproduce or modify this severe phenotype, indicating that the spermatogenic arrest arises from distinct molecular pathways. We further observed severe abnormalities in mitochondrial ultrastructure in PARL-deficient spermatocytes, along with prominent electron transfer chain defects, disrupted coenzyme Q (CoQ) biosynthesis, and metabolic rewiring. These mitochondrial defects are associated with a germ cell-specific decrease in GPX4 expression leading arrested spermatocytes to ferroptosis – a regulated cell death modality characterized by uncontrolled lipid peroxidation. Our results suggest that mitochondrial defects induced by PARL depletion act as an initiating trigger for ferroptosis in primary spermatocytes through simultaneous effects on GPX4 and CoQ – two major inhibitors of ferroptosis. These findings shed new light on the potential role of ferroptosis in the pathogenesis of mitochondrial diseases and male infertility warranting further investigation.