Expression of one important chaperone protein, heat shock protein 27, in neurodegenerative diseases

Many neurodegenerative diseases, also termed as protein conformational diseases[1], are characterised by accumulations of misfolded proteins that often share morphological and biochemical features and can colocalise with several other proteins, including various chaperone proteins. The accumulation of misfolded proteins may adversely affect neuronal connectivity and plasticity and trigger cell death signalling pathways [2]. These misfolded proteins include amyloid-β protein (Aβ) and tau in Alzheimer’s disease (AD); α-synuclein and synphilin 1 in Parkinson’s disease (PD); polyglutamine (polyQ)-expanded huntingtin in Huntington’s disease; transactive response DNA binding protein 43 (TDP-43) and copper-zinc superoxide dismutase 1 in motor neuron disease (MND); and tau, TDP-43 and fused in sarcoma in frontotemporal lobar degeneration (FTLD). These aggregates may consist of oligomeric complexes of non-native secondary structures and demonstrate poor solubility in aqueous or detergent solvents [3]. Chaperone proteins, such as heat shock proteins (HSPs), play a prime role in protein homeostasis by binding to substrates at risk, thereby keeping them in a state competent for either refolding or degradation [4]. Chaperone proteins have therefore been implicated as potent modulators of protein conformational disorders, suppressing toxicity of misfolding proteins and modifying early events in the aggregation process in a cooperative and sequential manner reminiscent of their functions in de novo protein folding [5],[6].

HSP27, also known as HSPB1, is one of the best studied members of the small HSP family. It functions as a molecular chaperone, aiding the refolding of non-native proteins, and plays a critical role in stabilisation of the cytoskeleton through interactions with several cytoskeletal components, such as actin, intermediate filaments and microtubules [7]. Normally, when functioning as a molecular chaperone, HSP27 can bind non-native substrates in an ATP-independent manner, which is quite different from most HSPs [8]. However, for substrate release, it requires the help of ATP-dependent chaperone proteins, such as HSP70 and HSP104 [9]. Additionally, HSP27 can exhibit a number of cytoprotective properties in cell culture, attenuating apoptosis through interaction with proapoptotic proteins such as Daxx, cytosolic cytochrome c and caspase 3, as well as suppressing aggregate formation [10],[11].

HSP27 has been implicated in various neurodegenerative diseases. For example, a highly induced expression of HSP27 has been reported in the brains of aged persons and of patients with AD, being present in proliferating astrocytes, especially in AD in those areas rich in senile (Aβ) plaques. Neurofibrillary tangles, Hirano bodies and some hippocampal neurons have also been reported to be HSP27-immunoreactive. However, in control brains, HSP27 immunoreactivity was restricted to blood vessels and to occasional astrocytes in the white matter. Similarly, patients with other types of dementia (Parkinson–dementia complex, multi-infarct dementia and normal pressure hydrocephalus) showed less expression of HSP27 in reactive astrocytes than that in AD, but more than that in controls. Occasional immunoreactive HSP27 astrocytes have also been reported to be present in Parkinson disease [12].

Although the physiological relationship between HSP27 and misfolded proteins in neurodegenerative diseases has been studied intensively in vivo and in vitro, any potential role of HSP27 in the pathogenesis of neurodegenerative diseases, and especially FTLD, is still incompletely understood. The principal aim of this study was to investigate the possible role of HSP27 in the pathogenesis of FTLD. This was achieved by comparing the distribution and amount of HSP27 in FTLD, AD, MND and control groups by immunohistochemistry, and by characterising biochemical changes using Western blotting.

Although the physical and physiological relationships between chaperone proteins, such as HSP27, and misfolded proteins has been studied intensively, a role for HSP27 in the pathogenesis of neurodegenerative disease is still incompletely understood. In the present study, the expression of HSP27 in FTLD, AD, MND and normal controls was analysed through immunohistochemical staining with HSP27 antibody and Western blotting. HSP27 is usually induced in response to cellular stress, but it can also stabilise the actin cytoskeleton, attenuate apoptosis, participate in cell differentiation and survival when phosphorylated [23], and promote cell cycle reentry into the S-phase by facilitating ubiquitination and degradation of the cell cycle inhibitor p27^Kip1[24]. The affinity of HSP27 for proteins to be chaperoned is modulated by its phosphorylation and oligomerisation status [25].

Our present results show that the expression patterns of HSP27 in these neurodegenerative diseases are quite different. Patients with FTLD and AD showed numerous intensely immunostained, HSP27-positive granules in the cytoplasm of pyramidal neurons of the cerebral cortex, along with diffuse clusters of similar granules within glial cells. Such changes were particularly severe in the CA4 region of the hippocampus and in cells of the dentate gyrus. Subgroup analysis showed that there were no differences in the patterns of immunostaining between FTLD-tau and FTLD-TDP cases. However, in cases of MND and normal controls, we found less extensive accumulation of HSP27 immunoreactive granules in these same brain areas. These observations probably reflect the greater ongoing neurodegeneration in these brain areas in AD and FTLD compared to MND.

There have been few previous studies in which investigators have evaluated HSP27 reactivity in neurodegenerative disease, and then only in relationship to AD. To our knowledge, there have been no such studies in FTLD. In general, these studies have shown there to be an increase in expression of HSP27 in AD brain tissue or in immunolabelling of pathological structures [12],[25]-[27]. Using a rabbit anti-HSP27 antiserum raised against a hybrid protein containing the N-terminal part of the murine HSP27 (amino acids 1 to 110) and the C-terminal part of the human HSP27 (amino acids 111 to 208), Renkawek and co-workers showed a highly induced expression of HSP27 in many proliferating astrocytes in affected regions of cerebral cortex, especially in those areas rich in senile plaques [12]. They also reported neurofibrillary tangles, Hirano bodies and some hippocampal neurons to be immunopositive [12]. The expression of HSP27 correlated with the severity of AD-specific morphological changes and the duration of dementia [12]. Shimura and co-workers partially confirmed these findings, demonstrating a coimmunoprecipitation of HSP27 with hyperphosphorylated tau protein from AD brain, but not with nonphosphorylated tau from normal brain homogenates [25]. However, in another study, a diffuse, rather than granular, HSP27 immunostaining was observed in a subpopulation of astrocytes and microglial cells associated with senile plaques and cerebral amyloid angiopathy, but this was not seen in neurofibrillary tangles in either the neocortex or hippocampus [27]. In a biochemical study, Bjorkdahl and colleagues reported the expression of HSP27 in AD brain to be increased (by approximately 20%) in homogenates from the medial temporal cortex [26]. Moreover, this increase in HSP27 correlated significantly with the expression levels of the majority of the phosphorylated tau epitopes, but not with AT270, Alz-50 or AT100 [26].

In the present study, we observed numerous intensely immunostained, HSP27-positive granules in the cytoplasm of pyramidal neurons of the cerebral cortex and as diffuse clusters of similar appearing granules within glial cells, in both FTLD and AD. However, in AD, there was no HSP27 immunostaining of senile plaques or neurofibrillary tangles within the frontal and temporal cortices or within the hippocampus. The failure to confirm previous findings [12] (but see [27]) may be partly due to the different HSP27 antibodies used. In the present study, the anti-HSP27 antibody used (ab5579; Abcam) is a rabbit polyclonal raised against a synthetic peptide, LLRGPSWDPFRC, corresponding to amino acids 10 to 21 of human HSP27. It is much more specific than the rabbit anti-HSP27 antiserum raised against a hybrid protein containing the N-terminal part of the murine Hsp25 (amino acids 1 to 110) and the C-terminal part of the human HSP27 (amino acids 111 to 208) used by Renkawek et al. [12]. Hence, the staining of plaques (amyloid) or tangles with this latter antibody may represent nonspecific cross-reactivity, an interpretation consistent with findings reported by Wilhelmus and colleagues [27], who likewise failed demonstrate any immunostaining of these structures with HSP27. Our presently reported observations that no HSP27 immunostaining was seen in Pick bodies in patients with FTLD-tau, or in TDP-43 immunoreactive neuronal cytoplasmic inclusions in patients with FTLD-TDP or MND, would support this conclusion. Interestingly, in previous studies where HSP27 immunostaining of astrocytes has been reported [12],[27], this has taken the form of a homogeneous, diffuse cytoplasmic staining rather than one with pronounced granularity as demonstrated here. Again, different antibody specificities may be responsible, picking up HSP27 proteins either in a different conformational state or in aggregated forms.

The pattern of HSP27 immunoreactivity in cells may reflect an activation of protein quality control systems consequent upon damage to neurons or a response to the presence of accumulated tau/TDP-43 proteins. It is increasingly clear that the cooperation of proteolytic machineries, such as the proteasome and lysosome, and chaperone systems is required to prevent the aggregation of potentially toxic misfolded proteins. If such misfolded proteins escape appropriate refolding by chaperone systems, the proteasome and lysosome systems will be recruited to degrade them. However, if the production of such misfolded proteins overwhelms the capacity of the protein quality control systems, protein aggregates will form.

Within our control group, there was variability in the expression of HSP27. In some control subjects similar (to AD and FTLD), although less intense, HSP27 changes were present in the cytoplasm of pyramidal neurons of the cerebral cortex, and again as diffuse clusters of similar granules within glial cells. Such changes probably reflect the mild AD-type pathological changes present in such control subjects. In a recent study of patients with amnestic mild cognitive impairment (aMCI), a transitional stage between normal ageing and AD, there was a significant increase (200%) in the expression of HSP27 in the hippocampus in the aMCI group compared to controls [28], which would be consistent with these ‘intermediate’ histological findings.

It is presently unclear which cellular compartment or substructure might be associated with HSP27 immunostaining. The granularity of the staining suggests a phagocytotic location. Consequently, we stained sections in which there was an abundance of HSP27-immunoreactive granules with LAMP1 (a lysosomal marker) and LCA3 (a marker of autophagy) antibodies. LAMP1 detected granular structures in neurons and glia similar to those detected by HSP27, but none such were demonstrated in LCA3 immunostaining. These observations suggest that HSP27 immunoreaction is possibly associated with the cells’ lysosomal compartments and may represent accumulated protein undergoing or awaiting degradation or the product of a failed degradation process. Nonetheless, a proteasomal location is also possible, and the pattern of immunoreactivity could represent proteasomes clogged or choked with accumulated protein. However, the fact that the ‘granules’ were not immunostained with either p62 or ubiquitin antibodies suggests that the HSP27 immunolabelled material does not contain either of these proteins, which act as molecular ‘flags’ for proteasomal degradation. Therefore, the subcellular localisation of HSP27 immunostaining, and its role in the neurodegenerative cascade, remains uncertain though may possibly be lysosomal.

In this study, we show that there are distinct patterns of HSP27 immunostaining with respect to different types of neurons or glial cells in different brain regions in patients with neurodegenerative disease. In general, patients with AD and FTLD were more severely affected than were patients with MND and control subjects; however, there was no association with FTLD histological subtype. Moreover, the ‘classic’ neuropathological changes in FTLD, AD and MND were not immunoreactive to HSP27, with the lack of HSP27 immunostaining of ‘classic’ pathology suggesting that HSP27 does not play a direct role in the formation of these structures. Therefore, HSP27 immunostaining in neurodegenerative disease likely reflects the extent of ongoing neurodegeneration in affected brain areas and is not specific to FTLD, AD or MND. It may represent an accumulation of misfolded, damaged or unwanted proteins, waiting for degradation via proteolytic mechanisms, though the precise pathway that this might enter, or fail to enter, for such a purpose remains unclear.