Background
Protein misfolding is a common feature of many neurodegenerative diseases. In some of these diseases, such as the synucleinopathies and the tauopathies, cytoplasmic proteins aggregate to form intracellular deposits. However, in the amyloidoses, which include Alzheimer's disease (AD), prion diseases and the British and Danish familial dementias, proteinaceous aggregates are observed extracellularly [
1‐
4]. There is increasing evidence that the mechanism of neurotoxicity in these amyloidoses is similar and that it is the conformation of the aggregated protein, rather than its specific amino acid sequence which results in altered membrane permeability to calcium [
5]. Therefore, studies on the mechanism of neurotoxicity in one disease may provide insights into the mechanisms involved in other diseases.
Familial amyloidotic polyneuropathy (FAP) is a rare autosomal dominant disease characterised by the deposition of transthyretin (TTR) protein in peripheral nerves. The early clinical manifestations of FAP include progressively aberrant thermosensation and nociception in the lower extremities followed by profound autonomic dysfunction [
6‐
9]. TTR is a 55 kD homotetrameric protein that has been well characterised for its role in the transport of thyroxine and retinol [
8]. More than 100 TTR mutations are known, and most have been shown to be amyloidogenic [
10]. Many studies have shown that mutant TTR aggregates to form oligomers more readily than wild-type TTR, and that further aggregation leads to the formation of amyloid fibrils [
11]. There is a correlation between the rate of aggregation of TTR in vitro and the extent or severity of the disease phenotype. For example, the rare L55P mutation produces a more aggressive amyloidosis than the more common V30M mutation, and
in-vitro studies show that L55P TTR aggregates more much readily than V30M TTR [
12‐
16].
The mechanism by which TTR forms fibrils is not entirely understood. Some studies suggest that amyloid deposition involves the formation of low molecular weight "nuclei" that must reach a critical concentration before fibril elongation [
17]. However, other studies suggest that amyloid aggregation may be a nucleation-independent process [
18,
19]. More specifically, and consistent with this latter view, Hammarström
et al [
20] and Hurshman Babbes
et al [
14] have shown that TTR aggregation may be a nucleation-independent process. Mutant TTR has been shown to be toxic to cells in culture [
12,
21]. It has been reported that TTR-induced toxicity is mediated by the receptor for advanced glycation end-products (RAGE) and that activation of RAGE leads to endoplasmic reticulum stress, activation of ERK1/2 and caspase-dependent apoptosis [
22]. There is also evidence to suggest that misfolded proteins like TTR mediate their toxic effects by binding directly to lipid-rich areas of the plasma membrane [
13,
23]. Also, the toxicity of TTR aggregates is correlated with membrane binding affinity, destabilisation of cell membrane fluidity and subsequent decrease in cell viability [
13].
There is ample evidence suggesting that some of the toxic effects of amyloid proteins are mediated via an increase in calcium permeability. For example, the β-amyloid protein (Aβ) of AD is known to induce calcium influx into cells [
24,
25]. This disruption of calcium homeostasis is likely to cause abnormal neuronal function since calcium is an important mediator of synaptic plasticity and excitotoxicity. However, the mechanism by which amyloid proteins induce calcium entry in cells is poorly understood.
Previously, we have shown that in SH-SY5Y neuroblastoma cells, TTR induces an influx of extracellular calcium across the plasma membrane. This TTR-induced increase in calcium permeability is primarily mediated by voltage-gated calcium channels (VGCC), with a small proportion (~20%) of the calcium influx through voltage-independent channels [
12]. However, neuroblastoma cells are not a physiologically relevant cell type for studying FAP. FAP is a peripheral polyneuropathy involving amyloid deposition affecting peripheral neurons including sensory neurons of dorsal root ganglia (DRG).
In the present study, we examined the effect of amyloidogenic forms of TTR on calcium levels in cultured DRG neurons. We demonstrate that TTR induces calcium influx into DRG neurons, similar to that observed using SH-SY5Y cells. Importantly, we demonstrate that calcium influx into the growth cones of small-diameter TrkA-positive DRG neurons requires the presence of Nav1.8 voltage-gated sodium channels and transient receptor potential (TRP) M8 channels. The results suggest that activation of TRPM8 channels by L55P TTR results in the subsequent opening of voltage-sensitive sodium and calcium channels. Our study suggests that TRP channels may be an important therapeutic target, not only for the development of drugs for the treatment of FAP, but also for many important neurodegenerative diseases.
Discussion
In this study we report that the mutant TTR variant L55P, induces calcium influx in growth cones of DRG sensory neurons. The extracellular calcium influx seen in DRG growth cones occurred within minutes of exposure to L55P and was sustained for at least 500 sec. The observation that L55P-mediated increases in growth cone calcium originated exclusively from extracellular sources confirms previous work in SH-SY5Y cells that showed TTR-induced calcium influx is mediated primarily by ion channels at the plasma membrane [
12]. In FAP, deposition of aggregated TTR protein is a pathological hallmark of a polyneuropathy and subsequent peripheral neurodegeneration. These results strongly suggest that in FAP, distal sensory neurons are exposed to prolonged calcium dysregulation, leading to progressively aberrant signalling in small-diameter sensory neurons.
Our data show that a significant portion of the calcium influx was due to the opening of VGCCs. For example, approximately 50% of the calcium influx was blocked by the L-type channel blocker nifedipine. We considered whether voltage-gated sodium channel activity was responsible for the opening of VGCCs. To answer this question, we examined the effect of voltage-gated sodium channel blockers on TTR-induced calcium influx. Nociceptive sensory neurons express a subset of Na
v channels that are resistant to TTX blockade [
30,
37‐
40]. Na
v1.8 channels in DRGs are activated by protein kinase A (PKA) and protein kinase C (PKC) dependent mechanisms and they are involved in diabetes-induced hyperalgesia and allodynia where they exhibit significantly increased slow and fast ramp activation profiles and left-shifted voltage-dependent activation [
41,
42]. Consistent with the possibility that voltage-gated sodium channels mediate the TTR-induced opening of VGCC, we found that while TTX had no significant effect on calcium influx, ambroxol and carbamazepine, blockers of TTX
R Na
v1.8 channels [
43,
44], inhibited up to 80% of the calcium influx. As a small percentage of the calcium influx was not inhibited by ambroxol or carbamazepine, this suggested that some influx may have occurred via a voltage-insensitive mechanism. This latter possibility was supported by the finding that SKF96365, a relatively non-specific inhibitor of many types of TRP channels (including TRPM8 channels) [
33,
34], and BCTC, an inhibitor of TRPM8 and TRPV1 channels [
35] significantly inhibited calcium influx. The involvement of TRPM8 channels in TTR-induced calcium influx was confirmed by experiments which showed that the TRPM8 agonist icilin induced similar calcium influx in DRG growth cones and that the effect of TTR was strongly inhibited by three different TRPM8 siRNA oligonucleotides. As TRPM8 channels are permeable to both sodium and calcium [
45,
46], it seems likely that they are both directly responsible for some of the calcium influx, but they may also contribute to the changes in membrane potential which trigger opening of voltage-gated sodium and calcium channels.
It was surprising to observe that TRPM8 channels were responsible for calcium influx in most growth cones. Previous studies have shown that TRPM8-positive cells account for only a small fraction of the total population of neurons in DRG tissue [
45,
46]. However, in our DRG cultures, TRPM8-positive cells accounted for >50% of all neurons. The greater percentage of TRPM8-positive cells in the cultures compared with the number
in vivo is likely to be due to the positive selection of TrkA-positive small diameter neurons since the cells were cultured in the presence of NGF.
TRPM8 channels have been described as the prototypic thermosensitive ion channel [
45,
46]. They are activated by the compounds menthol, icilin and mild cold stimuli and are a prominent receptor subtype on small diameter (C and Aδ fibre) DRG neurons [
47]. While FAP exhibits a variety of clinical manifestations, one of the most common symptoms is a progressive parasthesia involving the extremities, especially the lower limbs and affecting thermosensation and nociception [
6,
7,
48‐
50]. While the current study strongly suggests that TRPM8 and Na
v1.8 channel activation occurs in response to L55P attack on DRG neurons
in vitro, the involvement of other ion channels in TTR toxicity cannot be excluded. However, it is likely the parasthesia seen in FAP is a clinical correlate of a small diameter sensory neuron phenotype undergoing chronic excitotoxic insult.
The precise TTR species that mediate the effect on calcium influx are yet to be elucidated. Preparations of L55P contained significant amounts of oligomeric aggregates (~1800 kDa) that were not seen in preparations of wild-type TTR. Urea-denaturation studies suggest that unlike WT TTR, L55P aggregates through a complicated pathway involving the formation of dimers and trimers. However the precise mechanism responsible for the formation of soluble oligomers at physiological pH has yet to be determined [
14]. Soluble oligomeric TTR aggregates are more toxic to cells than fibrillar or monomeric species [
12,
16]. We have previously shown that the aggregation state of TTR variants is correlated with the magnitude of calcium entry and cytotoxicity in SH-SY5Y cells [
12]. In the present study, freshly prepared L55P elicited a greater calcium influx in DRG growth cones than either V30M or WT proteins. Reixach
et al. [
51] have reported that while monomers and small oligomeric (<100 kDa) forms of the V30M variant induced cell death in a human neuroblastoma cell line, the formation of these smaller species proceeded at a relatively slow rate under physiological conditions.
It remains unclear how aggregated TTR can activate TRPM8 channels. One possibility is that TTR binds directly to the channels and causes a conformational change which leads to channel opening. However, there is no direct evidence for this. Another possibility is that TTR binds to a membrane component which triggers the opening of TRPM8 channels. This membrane component could be a lipid. TTR binds preferentially to phospholipids in cholesterol-rich regions of membranes and alters the membrane fluidity [
13]. There is some evidence of preferential localisation of TRPM8 activity at detergent-insoluble areas of the plasma membrane or lipid rafts [
52,
53]. As TRPM8 channels are mechanosensitive [
54‐
58], it is possible that disruption of the lipid membrane by TTR could contribute to channel opening. Also, like other TRPM family members, TRPM8 is activated by a major phospholipid component of lipid rafts, phosphatidylinositol 4,5-bisphosphate (PIP
2) [
59‐
61]. It is likely, therefore, that L55P-mediated calcium influx at the cell membrane is the culmination of multiple interactions and further work will be required to fully explore this possibility.
The present study identifies TRPM8 channels as a possible molecular target for pharmacological intervention in FAP. Currently, liver transplantation is an effective treatment for FAP, and several pharmacological interventions such as clonazepam, doxepine and significantly, carbamazepine, have been shown to be effective in the symptomatic treatment of the polyneuropathy and parasthesia associated with FAP [
48]. Recently, small molecule stabilisers which inhibit the aggregation of TTR have been undergoing promising clinical trials [
20],.
As amyloidoses may share common mechanisms of neurotoxicity, this study has implications for understanding the cause of other neurodegenerative diseases. Neuronal calcium dysregulation has also been implicated in the pathogenesis of a variety of amyloidoses, including AD [
62,
63]. Accumulation of aggregated Aβ in the AD brain has been reported to disrupt calcium metabolism with concomitant downstream cytotoxic effects such as mitochondrial dysfunction, activation of caspases [
64,
65] microtubule destabilisation and tau phosphorylation [
66]. While there is some evidence implicating TRP channels in the pathogenesis of AD [
67], it is unclear whether aggregated Aβ interacts directly with TRP channels in the AD brain and therefore future experiments in this area would add considerably to our understanding of the early neurotoxic events in AD.
Methods
Materials
Wild-type (WT), V30M and L55P transthyretin were expressed in BL21(DE3)- RIG CodonPlus E.coli and purified as described previously [
13]. Proteins were incubated for2-4 hr at 22-24°C prior to their use in calcium imaging experiments. Pharmacological compounds used in experiments were obtained from the following sources: nifedipine, ω-agatoxin IVA, ω-conotoxin GIVA (Alomone Labs, Israel), SKF-96365 (Tocris, Bristol, UK), [N-(4-tert-butylphenyl)-4-(3-chloropyridin-2-yl)piperazine-1-carboxamide] (BCTC), (ENZO, NY, USA), ambroxol and carbamazepine (Sigma-Aldrich, MO, USA).
Cell culture
All animal experimentation was performed in accordance with the guidelines and protocols of the Animal Ethics Committee, University of Tasmania. Dissected thoracic dorsal root ganglia from embryonic day 16-18 (E16-18) hooded Wistar rats were mechanically dissociated into sensory neuron medium (SNM) composed of Dulbecco's modified Eagle's medium/Ham's F-12 medium 1:1, (Sigma-Aldrich, MO, USA) containing foetal bovine serum (5% v/v), penicillin G (100 U/ml), streptomycin (100 μg/ml), nerve growth factor (NGF, 50 ng/ml, Sigma-Aldrich, MO, USA) and N2 neural medium supplement (Invitrogen, CA, USA). Cells were plated onto poly-ornithine (100 μg/ml, Sigma, MO, USA) and laminin (50 μg/ml, Invitrogen, CA, USA) coated glass coverslips.
Calcium imaging
Between 12 and 24 hr after plating, sensory neurons, grown on coverslips, were loaded with Fluo-4 AM calcium indicator (1 μM, Invitrogen, CA, USA) in Hanks balanced salt solution (HBSS) without phenol red (Sigma-Aldrich, MO, USA) for 7 min, rinsed with fresh HBSS and incubated for a further 15 min at 37°C to allow complete de-esterification. All imaging experiments were carried out at 24-26°C. Calcium-free experiments were performed in calcium-free HBSS (Sigma-Aldrich, MO, USA) supplemented with 300 μM EGTA (glycol-bis(2-aminoethylether)-N,N,N',N'-tetra-acetic acid). Growth cones from small neurons (<15 μm in diameter) were selected for analysis. Freshly prepared TTR proteins were added to culture dishes to a final concentration of 0.5 mg/ml for all experiments and remained in contact with growth cones for the duration of the imaging period. Calcium experiments on aged (36 hr) protein were only performed using the V30M variant. Fluorescence images were acquired at 1-5 Hz using a confocal microscope (Zeiss LSM 510, Jena, Germany) using a 20× (1.0 NA) water immersion objective. Fluorescence images and pixel intensities were analysed using ImageJ software [
68] and custom software (Matlab, MathWorks, MA, USA). ΔF for each image frame was calculated by subtracting integrated pixel intensity of a region of interest encompassing the entire growth cone from the average integrated pixel intensity or F
0, of the first 20-30 frames of the imaging session for each cell. Maximal ΔF/F
0 was calculated for each growth cone over the entire 7 min imaging period. Prism 4 (GraphPad Software, CA, USA) and Adobe Illustrator CS3 (Adobe Systems, CA, USA) were used for graphical and statistical analysis of data.
Atomic force microscopy (AFM)
AFM was performed as previously described [
12]. Briefly, TTR solutions (5-50 μg/mL), were deposited on a surface of freshly cleaved, highly oriented pyrolytic graphite and incubated at 37°C for 30 min. AFM analysis was performed on a Digital Instruments Nano-Scope IV, multimode scanning probe microscope equipped with a 15 μm E scanner (Veeco Instruments, NY, USA). Images were obtained in tapping mode at oscillation frequencies of 200-300 kHz and analyzed using WSxM 4.0 software (Nanotec Electronica S.L., Madrid, Spain).
Analysis of TTR aggregation by dynamic light scattering (DLS)
Aggregation of wild-type (WT) and L55P TTR (1 mg/mL in 20 mM phosphate buffer (pH 7.4) containing 150 mM NaCl) was measured at 37°C for 6 hr using DLS. Measurements at 30 min intervals were acquired on a Zetasizer Nano S (Malvern Instruments, UK), and analyzed using Zetasizer Nano software 5.0.2. The hydrodynamic diameter and predicted molecular masses of protein species were calculated using Zetasizer Nano software 5.0.2, utilising the Stokes-Einstein equation and the Perrin factor to derive an estimate of molecular mass for a globular protein [
69].
siRNA knockdown of TRPM8 expression
Fluorescently (Cy-3) labelled TRPM8 siRNA (GGCCAUGGAGAGCAUAUGC), (CGAGAAUGCGUCUUCUUUA), (GCACAAAAAUGUAUGGAAA) and negative control siRNA oligonucleotides were obtained from (Applied Biosystems, CA, USA). Oligonucleotides were loaded into sensory neurons using previously established methods [
70‐
72]. Briefly, oligonucleotides (5 nM) were added to dissected DRGs in SNM. DRGs were then gently triturated and maintained in SNM (with 5 nM oligonucleotide) throughout the culture period (24 hr). Incorporation of fluorescently labelled oligonucleotides was confirmed by confocal microscopy (data not shown).
Immunocytochemistry
DRG cultures were fixed in 4% (w/v) paraformaldehyde and blocked with 10% (v/v) goat serum for 4 hr at 4°C. The primary antibodies, a polyclonal rabbit anti-TRPM8 (extracellular) (1:1000, Alomone Labs, Israel) or polyclonal rabbit anti-TrkA (1:1000, Abcam, Cambridge, UK) were added to coverslips and incubated for 4 hr at 22°C. Detection of primary antibodies was performed using fluorescently labelled goat anti-rabbit antibodies (Invitrogen, CA, USA). Nuclear staining was performed with 4',6-diamidino-2-phenylindole (DAPI) (Invitrogen, CA, USA). Images were acquired on a Zeiss LSM 510 confocal microscope and processed using ImageJ, Adobe Photoshop CS3 and Adobe Illustrator CS3 (Adobe Systems, CA, USA).
Immunoblotting
Primary DRG cultures treated with either control or specific TRPM8 siRNA oligo (5 nM) were established at high density in 6-well polystyrene culture plates (BD Falcon, NSW, Australia). Following incubation for 24 hr, cells were rinsed 3 times in ice-cold PBS. Ice-cold cell lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1%, v/v, Triton X-100, 1%, w/v, Na deoxycholate, 0.1%, w/v, SDS) supplemented with Complete® Mini protease inhibitor cocktail (Roche Diagnostics, NSW, Australia) was added to dishes and slowly agitated for 15 min at 4°C. Cell lysates (10 μg total protein) were separated on 8% SDS-PAGE, transferred onto PVDF membranes and then blocked overnight in blocking solution [5% (w/v), non-fat skim milk in 20 mM Tris-HCl and 150 mM NaCl (TBS)]. Primary antibodies, a rabbit anti-TRPM8 (1:200 in blocking solution, Alomone Labs, Israel) and mouse anti-GAPDH (1:2000 in blocking solution, Sigma-Aldrich, MO, USA) were added to the membrane, incubated overnight at 4°C then rinsed thoroughly in TBS. Membranes were subsequently probed with goat anti-mouse-HRP or goat anti-rabbit-HRP secondary antibodies for 1 hr at room temperature. Conjugates were detected using Immobilon chemiluminescent substrate reagent (Millipore, MA, USA) and a Chemi-Smart 5000 CCD image acquisition system (Vilber Lourmat, Marne-la-Vallee, France). CCD image analysis was performed with ImageJ software and statistical analysis was performed with Prism 4 (GraphPad Software, CA, USA).
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
RG carried out the imaging, DLS and knockdown experiments, analysed and interpreted the data, participated in overall direction of the study and preparation of the manuscript. XH carried out the AFM experiments. DK participated in the analysis and interpretation of the DLS experiments. HC, AV, LF and HP were involved in the analysis and interpretation of data and critical revision of the manuscript. DS conceived and participated in overall direction of the study and preparation of the manuscript. All authors have read and approved the final manuscript.