J. Neurochem. (2011) 10.1111/j.1471-4159.2011.07521.x
HspB8, a small heat-shock protein implicated in autophagy, is mutated in patients with distal hereditary motor neuropathy type II (dHMNII). Autophagy is essential for maintaining protein homeostasis in the central nervous system, but its role has not been investigated in peripheral motor neurons. We used a novel, multispectral-imaging flow cytometry assay to measure autophagy in cells. This assay revealed that over-expression of wild-type HspB8 in motor neuron-like NSC34 cells led to an increased co-localisation of autophagosomes with the lysosomes. By contrast, over-expression of mutant HspB8 resulted in autophagosomes that co-localised with protein aggregates but failed to co-localise with the lysosomes. A similar impairment of autophagy could also be demonstrated in peripheral blood mononuclear cells from two dHMNII patients with the HspB8K141E mutation. We conclude that defects in HspB8-mediated autophagy are likely to contribute to dHMNII pathology and their detection in peripheral blood mononuclear cells could be a useful, accessible biomarker for the disease.
bright detail similarity
Charcot Marie tooth disease
distal hereditary motor neuropathy type II
neuroblastoma × spinal cord hybrid cells
peripheral blood mononuclear cells
small heat-shock protein
HspB8 is a stress-responsive member of the small heat-shock protein family (sHsp) that includes nine other members (HspB1-10) in the human genome. sHsps possess an evolutionarily conserved α-crystallin domain that confers molecular chaperone activity, allowing sHsps to recognise and bind misfolded proteins to suppress their aggregation. Autosomal dominant missense mutations in HSPB8 (Hsp22/HspB8) and HSPB1 (Hsp27/HspB1) have been shown to cause type II distal hereditary motor neuropathies (dHMNII) and closely related axonal forms of Charcot Marie Tooth disease (CMT2L and CMT2F) (Irobi et al. 2004). dHMNs are a group of clinically and genetically heterogeneous disorders characterised by motor neuron degeneration in the peripheral nervous system. Patients display progressive weakness and wasting of extensor muscles of the toes and feet that may spread to proximal muscles of the lower limb and distal upper limbs (James and Talbot 2006).
The mechanism underlying neurodegeneration in dHMN remains unclear. Increased propensity of the dominant HspB8K141N and HspB8K141E mutants to aggregate and form hetero-oligomers with HspB1 was proposed to be a potential source of pathology (Irobi et al. 2004) and recently, these mutations have also been implicated in motor neuron-specific neurite degeneration (Irobi et al. 2010). In addition to its canonical function as a molecular chaperone, HspB8 acts in association with the co-chaperone Bag-3 to target protein aggregates for destruction via autophagy in neurons (Crippa et al. 2010) and muscles (Arndt et al. 2010; Carra et al. 2010). Decreased association of the HspB8K141N and HspB8K141E mutants with Bag-3 has been reported and could contribute to dHMNII pathology (Carra et al. 2010). The HspB8 and Bag-3 driven autophagy pathway has recently been called chaperone-assisted selective autophagy (Arndt et al. 2010). Autophagy, a pathway for cellular self-digestion, is traditionally classified into three main categories: macroautophagy (formation of autophagosomes and their eventual fusion with lysosomes), microautophagy (engulfment via invaginations formed in the lysosomal membrane) and chaperone-mediated autophagy (direct transport of proteins to the lysosomal lumen with chaperone assistance) (Mizushima et al. 2008; Wong and Cuervo 2010). The newly described chaperone-assisted selective autophagy is a type of macroautophagy because it adheres to the canonical pathway of autophagosome formation; followed by fusion with the lysosomes.
Constitutive, basal macroautophagy (called autophagy hereafter) is critical for maintaining protein homeostasis in the CNS and a growing number of neurodegenerative disorders of the CNS have been attributed to defects in autophagy (Wong and Cuervo 2010). However, neuronal autophagy defects have not been demonstrated as a cause of neurodegeneration in the PNS. Here, we asked how the dominant dHMNII causing mutations in HspB8 might impair autophagy in motor neurons and if an autophagic deficit could be demonstrated in accessible cells derived from the affected patients.
Materials and methods
Patients and preparation of cells for autophagy detection
Peripheral blood mononuclear cells (PBMCs) were obtained with informed consent from two related dHMNII patients harbouring the pathogenic HspB8K141E mutation and two age-matched, healthy individuals. The patients were sisters who developed distal weakness of the lower limbs in their early 30s and at the time of study, approximately 10 years later, required a cane for walking and had moderate weakness and wasting of the intrinsic muscles of the hands. PBMCs were prepared by gradient centrifugation and incubated with the lysosome-specific Lyso-ID Red dye (Enzo Life Sciences, Exeter, UK) for 15 min at 37°C in R10 medium [RPMI-1640 medium (Gibco, Paisley, UK), 100 U/mL penicillin, 10 μg/mL streptomycin, 2 mM l-glutamine and 10% v/v foetal calf serum], washed, then fixed and permeabilised with BD Cytofix buffer and BD Perm/Wash buffer (BD Biosciences, Franklin Lakes, NJ, USA), respectively. These PBMCs were sequentially incubated for 1 h at 22°C with a rabbit anti-Map1LC3A antibody (Epitomics, Burlingame, CA, USA; 1 : 200) and Alexafluor 405 dye-conjugated goat anti-rabbit IgG (Invitrogen, Carlsbad, CA, USA; 1 : 1000) to detect LC3, an autophagosomal marker. LC3 detection by immunoblotting was performed using rabbit anti-Map1LC3B (Sigma, St Louis, MO, USA; 1 : 1000) and horse radish peroxidise-conjugated donkey anti-rabbit IgG (GE Healthcare, Bucks, UK; 1 : 10 000).
To investigate the selective targeting of misfolded proteins to the autophagic pathway by HspB8, neuroblastoma x spinal cord hybrid cells (NSC34) cells were co-transfected with human wild-type superoxide dismutase 1 (SOD1) or its known, misfolding mutants (SOD1G93A and SOD1G85R) and haemagglutinin (HA)-tagged HspB8 constructs using Lipofectamine 2000 (Invitrogen), trypsinised, labelled with Lyso-ID, fixed and then permeabilised as before. These cells were sequentially labelled with the following antibodies: sheep anti-SOD1 (Calbiochem, San Diego, CA, USA; 1 : 1000; recognises both endogenous mouse-SOD1 and transfected human-SOD1) and Alexafluor 488 dye-conjugated donkey anti-sheep IgG (Invitrogen; 1 : 1000) to label SOD1, mouse phycoerythrin-conjugated anti-HA (Santa Cruz Biotechnology, Santa Cruz, CA, USA; 10 μg) to label HA-tagged, over-expressed HspB8 and LC3 was labelled as described for PBMCs before.
Measurement of autophagy using multispectral-imaging flow cytometry
The ImageStream IS100 (Amnis Corp., Seattle, WA, USA) instrument was used to image 25,000 cells from each sample after excitation with 405 and 488 nm laser light. LC3, SOD1, HA-HspB8 and LysoID signals, also, side scatter and bright field images were each collected via separate channels. Post-acquisition spectral compensation and data analysis was performed using ImageStream Data Exploration and Analysis Software (IDEAS). Co-localisation between two fluorescent probes was measured using the Bright Detail Similarity (BDS) algorithm designed by Amnis Corp. (Santer et al. 2010). Based on the Pearson’s correlation coefficient, BDS measures the degree of similarity between two fluorescent images e.g. the similarity between LysoID and LC3 images. The mean BDS value increases in step with increasing levels of similarity between the two images. In cases where co-localisation of more than two fluorescent probes was required e.g. to assess the autophagy of SOD1, spot counts were applied on single-cell populations. Specifically, spot masks were generated to accurately identify the spots observed for each fluorochrome. These masks were then merged under the Boolean operator to identify cells in which SOD1, LC3 and LysoID spots were co-localised. The number of co-localised spots in each cell was counted and cells displaying at least one co-localised spot were taken to be undertaking the autophagy of SOD1. Thus, autophagy levels were calculated by gating on the population of cells in which cells were displaying one or more co-localised spots, and the percentage of cells within this region was then calculated for each experiment.
Immunofluorescence confocal microscopy
Coverslip-grown NSC34 cells were transfected with enhanced green fluorescent protein-tagged human SOD1 and/or HA-tagged HspB8 constructs. 72 h later, they were washed with phosphate-buffered saline (PBS), fixed with 4% para-formaldehyde for 15 min, washed again with PBS, and permeabilised with 0.2% Triton X-100 for 10 min. Coverslips were then blocked with 5% (v/v) milk in PBS for 30 min before incubation with polyclonal rabbit anti-HA (Sigma; 1 : 1000) at 4°C overnight. HA-tagged HspB8 was then visualised by subsequent incubation with Alexafluor 546 dye-conjugated donkey anti-rabbit IgG (Invitrogen; 1 : 1000). Images were collected using a LSM 510 Meta confocal laser scanning microscope (Carl Zeiss Inc., Oberkochen, Germany) with a 40× objective.
Preparation of soluble and insoluble cell fractions for imuunoblotting
Transfected NSC34 cells were lysed on ice in Tris-EDTA-NaCl buffer [10 mM Tris–HCl, pH 8.0, 1 mM EDTA, 100 mM NaCl, 0.5% (v/v) nonidet P-40 and 1% (v/v) protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN, USA)]. Lysates were centrifuged (15 000 g, 10 min) to obtain nonidet P-40-soluble (supernatant) and -insoluble (pellet) fractions. Washed pellets were solubilised by 1% sodium dodecyl sulphate in Tris–EDTA–NaCl buffer for electrophoresis.
Anti-SOD1 antibody was added to 100 μg of total protein overnight at 4°C with rotation. Antibody was recovered by adding Protein G sepharose beads (Invitrogen) for 1 h and centrifugation (7000 g for 2 min). After washing with PBS, immunoprecipitates were released for western blotting by adding sodium dodecyl sulphate sample loading buffer.
Determination of cell viability
Transiently transfected NSC34 cells were treated with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (5 mg/mL) in PBS, 72 h after transfection. The reaction was terminated with absolute dimethyl sulfoxide and absorbance measured at 544 nm in a microplate reader (FluoStar; BMG, Ortenberg, Germany).
In situ hybridisation
cDNA encoding the first exon of mouse HspB8 was sub-cloned into pCR4-TOPO (Invitrogen), linearised and used as a template for in vitro transcription to generate digoxigenin-labelled RNA (riboprobe) using a commercial kit (Roche). Hybridisation was carried out according to the manufacturer’s protocol.
All data were analysed and plotted using GraphPad Prism 5 software (GraphPad Software, San Diego, CA, USA). p-values for all datasets were determined by two-way analysis of variance (anova) and Bonferroni post-test or Student’s t-test where appropriate.
Wild-type but not mutant HspB8 enhances autophagy of a misfolded protein in motor neuron-like NSC34 cells
We used the motor neuron-like NSC34 cells as an in vitro model to test whether HspB8 and its mutant HspB8K141N could play a role in the autophagic disposal of misfolded protein substrates. In an indirect assay, the transient over-expression of HspB8 in NSC34 cells has previously been shown to increase the turnover of an archetypal, misfolding mutant of SOD1 fused to green fluorescent protein (GFP-SOD1G93A) (Crippa et al. 2010). In contrast, we used multispectral-imaging flow cytometry technology to directly measure the levels of HspB8-driven autophagy in NSC34 cells with or without the co-expression of native, untagged human SOD1 or its misfolding mutants (SOD1G85R and SOD1G93A). The multispectral-imaging flow cytometry data was analysed with spot counts (see Methods) to quantify the extent of co-localisation between Lyso-ID (lysosomal marker), endogenous LC3 (autophagosomal membrane marker) and SOD1 in each cell as an index of the prevalent level of autophagy. This demonstrated that the over-expression of HspB8 alone in NSC34 cells was sufficient to increase cellular autophagy levels by ∼2-fold compared with cells transfected with vectors (pCAGG and pcDNA3) alone (Fig. 1a). The anti-SOD1 antibody recognises the endogenous mouse-SOD1 in vector-transfected control cells so was included in the co-localisation analysis to obtain a true background value for autophagy levels. However, over-expression of the HspB8K141N mutant failed to stimulate autophagy (Fig. 1a), demonstrating that the HspB8K141N mutation, found in a majority of dHMNII patients, impairs the ability of HspB8 to drive autophagy in motor neuron-like cells. Co-expression of wild-type HspB8 with misfolding SOD1 mutants (SOD1G93A and SOD1G85R) revealed a further stimulation of autophagy levels compared with cells expressing HspB8 alone or co-transfected with folding-competent, wild-type SOD1 (Fig. 1a). By contrast, the HspB8K141N mutant failed to promote any significant increase in autophagy levels when co-transfected with SOD1 mutants, indicating persistent autophagic dysfunction in the context of motor neuronal protein misfolding (Fig. 1a). These findings indicate that HspB8 can recognise and target misfolded proteins for degradation via autophagy in motor neuron-like cells, and that the K141 residue of HspB8 is critical for this process.
Mutant HspB8 can initiate autophagy but cannot bring it to completion
The lack of autophagic stimulation by HspB8K141N suggested that the K141N mutation compromises at least one step in the autophagy pathway which includes induction, autophagosome formation, autophagosome-lysosome fusion and recycling of autophagosomal components (He and Klionsky 2009; Ravikumar et al. 2010). Images from individual cells used to generate the histograms in Fig. 1a reveal that HspB8K141N over-expressing cells possess LC3-containing structures that co-localise with SOD1G93A aggregates (Fig. 1b lower panels) but, importantly, unlike wild-type HspB8 (Fig. 1b upper panels), they do not co-localise with the lysosomes. This suggests that the K141N mutation does not affect the ability of HspB8 to recognise aggregated proteins or induce autophagosome formation but potentially impairs a later step in this pathway. Consistent with this suggestion, LC3 levels in HspB8K141N expressing cells were ∼2-fold higher when compared to cells expressing wild-type HspB8 (Fig. 1c), where autophagic flux is maximal and lysosomal degradation of LC3 proceeds normally. Cells transfected with the HspB8 vector (pCAGG) showed intermediate levels of LC3, consistent with an autophagic flux that is neither driven by wild-type HspB8 nor impaired by HspB8K141N over-expression (Fig. 1c).
Next, by measuring the co-localisation of HspB8, SOD1 and Lyso-ID, we examined whether HspB8-driven autophagy involves physical delivery of the substrate protein (mutant SOD1) by HspB8 to the lysosome. Over-expression of HspB8 led to the selective delivery of mutant SOD1 aggregates to the lysosomal compartment without affecting the properly folded, wild-type SOD1 (Fig. 1d). By contrast, HspB8K141N over-expression failed to deliver mutant SOD1 to the lysosomes since the extent of co-localisation was reduced to background levels (Fig. 1d).
Consistent with an ability to recognise aggregated protein substrates, HspB8K141N was found to co-localise with SOD1G93A aggregates by confocal microscopy (Fig. 2a) and could also be co-immunoprecipitated with SOD1G93A using the anti-SOD1 antibody (Fig. 2b). It is clear that unlike HspB8K141N, wild-type HspB8-expressing NSC34 cells do not display high levels of visible SOD1G93A aggregates (Fig. 2a). Similarly, following biochemical fractionation, wild-type HspB8 was found to significantly reduce the amount of SOD1G93A in the insoluble fraction (Fig. 2c; compare lanes 2 and 4), but no such reduction was seen in cells expressing HspB8K141N (Fig. 2c; lane 6). Expectedly, wild-type HspB8 also improves the solubility of SOD1G93A while HspB8K141N fails to do so (Fig. 2c; compare lanes 8, 10 and 12). We conclude that the higher levels of autophagic flux driven by wild-type HspB8 facilitate the selective removal of SOD1G93A aggregates in NSC34 cells. The failure of HspB8K141N to reduce the prevalence of SOD1G93A aggregates is therefore likely due to the impaired autophagic flux seen in HspB8K141N-expressing cells (Fig. 1). A substantial proportion of HspB8K141N itself was in the insoluble fraction (Fig. 2c) consistent with previous studies (Irobi et al. 2004). LC3 immunoblotting analysis (Fig. 2d) further corroborated the autophagic impairment seen in cells expressing HspB8K141N. Higher steady state levels of LC3 were found in HspB8K141N-expressing cells than those expressing wild-type HspB8 or the vector alone (Fig. 2d). Similarly, the proportion of LC3-II, the lipidated form of soluble LC3 (i.e. LC3-I) present in autophagosomal membranes, was higher in cells expressing HspB8K141N compared to those expressing wild-type HspB8 (Fig. 2d). Collectively, these findings (Figs 1 and 2) further confirm that autophagy disruption in HspB8K141N mutants occurs after the initial steps of substrate recognition and autophagosome formation, either during lysosomal delivery or fusion.
To test whether HspB8-driven autophagy is beneficial to cell survival in the face of mutant SOD1 aggregation, cell viability was compared between differentially transfected cells (Fig. 2e). It is evident that wild-type HspB8 improves the viability of cells expressing mutant SOD1 while HspB8K141N is not protective. Interestingly, this effect was more pronounced in cells expressing SOD1G85R in comparison to those expressing SOD1G93A (Fig. 2e). Notably, the lack of dramatic effects on viability in cells expressing HspB8K141N is consistent with human dHMNII patients developing degeneration of peripheral motor neurons several decades after birth.
Defective autophagy in PBMCs derived from dHMNII patients
RNA in situ hybridisation performed on mouse spinal cord showed that HspB8 expression is relatively specific to motor neurons located in the region of Rexed’s lamina IX (Fig. 3a and b). However, sHsp expression is widespread and it was possible that the K141 mutations would result in an autophagic deficit in patient-derived non-neuronal cells. We tested this possibility by comparing basal autophagy levels in PBMCs derived from two dHMNII patients with the K141E mutation (Irobi et al. 2004) and two healthy controls. Using mean Bright Detail Similarity (see Methods) to detect the co-localisation of LC3 and Lyso-ID, as an index of autophagy, co-localisation levels in the two patients were lower than in the two healthy controls (Fig. 3c). The decreased co-localisation of LC3 and Lyso-ID is also illustrated by images of individual PBMCs (Fig. 3d). In addition, the decreased autophagy levels in patient PBMCs were accompanied by a higher level of LC3 compared to control PBMCs (Fig. 3e); suggesting that patient PBMCs are impaired in the lysosomal proteolysis of LC3. These data show that the highly conserved K141 residue in the α-crystallin domain of HspB8 is critical for its participation in the autophagic process and the K141E mutation causes an autophagic-deficit that is clearly detectable in PBMCs from the affected individuals.
We demonstrate, for the first time, that a mutation in HspB8 associated with an important group of inherited peripheral neuropathies results in impaired autophagy in patient-derived PBMCs. This indicates that although degeneration of peripheral motor neurons is the principal feature of dHMNs, HspB8 function is likely to be important for basal autophagy in many non-neuronal tissues as suggested by previous studies (Arndt et al. 2010; Carra et al. 2010). Given the inaccessibility of the target tissue (peripheral nerves), our assay using PBMCs has the potential to serve as a biomarker of disease activity and response to treatments targeting the autophagocytic process.
Hitherto the effect of dHMNII-causing mutations of HspB8 has been investigated using indirect assays of autophagy in a variety of cell lines. Uniquely, combining high-throughput imaging with flow cytometry, we show that in motor neuron-like NSC34 cells, mutant HspB8 generates autophagosomes that co-localise with mutant SOD1 aggregates but fail to complete lysosomal delivery or fusion. The results of this novel autophagy assay were corroborated by traditional biochemistry and low-throughput confocal microscopy. HspB8-driven macroautophagy in non-neuronal cells was reported to be strictly Bag-3-dependent, suggesting that HspB8 is responsible for substrate recognition while Bag-3 steers the autophagic process (Carra et al. 2008). The K141E and K141N mutations are reported to reduce the HspB8-Bag-3 complex formation (Carra et al. 2010). Our observation that the HspB8K141N mutant co-localises with protein aggregates and supports the assembly of autophagosomes, is consistent with the HspB8-Bag-3 complex formation being either partially affected or unaffected by the K141N mutation. Evidently, the K141 residue is proximal to but not a part of the HspB8 sequence (148–169) required for binding to Bag-3 (Fuchs et al. 2010). Consequently, the K141N mutation potentially interferes with downstream steps that ensure the co-localisation and eventual fusion of the autophagosomes with lysosomes. The molecular identity and temporal sequence of these downstream steps remains to be determined.
Autophagy is particularly important for protection against axonal degeneration and its induction is proposed to be an early response to stress and injury at the axonal terminals (Komatsu et al. 2007; Yue et al. 2008). Thus, the higher basal levels of HspB8 expression in motor neurons in vivo may reflect a constitutively high requirement for basal autophagy in these cells. Plausibly a greater reliance of axonal protein homeostasis on autophagy makes dHMNII patients with HspB8 mutations more vulnerable to the degeneration of long motor neurons in the PNS. Although, the primary trigger for this neurodegeneration remains to be identified, consistent with this proposal, ablation of the essential autophagy gene Atg7 causes the death of axonal terminals, which precedes cell death in Purkinje cells (Komatsu et al. 2007).
We would like to express our gratitude to the patients for participating in this research. VRA received a fellowship from the University of Oxford and Harris Manchester College. This research was funded by grants from the John Fell Oxford University Press Fund and the Greendale Foundation (Geneva) to VRA and from the Motor Neurone Disease Association to KT. Additional, supporting funds to VRA from the Neurosciences Group and the Nuffield Department of Clinical Neurosciences, University of Oxford are gratefully acknowledged. The authors declare no conflict of interests.