Quercetin Targets Cysteine String Protein (CSPα) and Impairs Synaptic Transmission

Background Cysteine string protein (CSPα) is a synaptic vesicle protein that displays unique anti-neurodegenerative properties. CSPα is a member of the conserved J protein family, also called the Hsp40 (heat shock protein of 40 kDa) protein family, whose importance in protein folding has been recognized for many years. Deletion of the CSPα in mice results in knockout mice that are normal for the first 2–3 weeks of life followed by an unexplained presynaptic neurodegeneration and premature death. How CSPα prevents neurodegeneration is currently not known. As a neuroprotective synaptic vesicle protein, CSPα represents a promising therapeutic target for the prevention of neurodegenerative disorders. Methodology/Principal Findings Here, we demonstrate that the flavonoid quercetin promotes formation of stable CSPα-CSPα dimers and that quercetin-induced dimerization is dependent on the unique cysteine string region. Furthermore, in primary cultures of Lymnaea neurons, quercetin induction of CSPα dimers correlates with an inhibition of synapse formation and synaptic transmission suggesting that quercetin interfers with CSPα function. Quercetin's action on CSPα is concentration dependent and does not promote dimerization of other synaptic proteins or other J protein family members and reduces the assembly of CSPα:Hsc70 units (70kDa heat shock cognate protein). Conclusions/Significance Quercetin is a plant derived flavonoid and popular nutritional supplement proposed to prevent memory loss and altitude sickness among other ailments, although its precise mechanism(s) of action has been unclear. In view of the therapeutic promise of upregulation of CSPα and the undesired consequences of CSPα dysfunction, our data establish an essential proof of principle that pharmaceutical agents can selectively target the neuroprotective J protein CSPα.


Introduction
In neurons, there are significant demands on cellular folding events. Complex interactions between multiple cellular components underlie synaptic transmission, a process that occurs with speed, precision and plasticity for extended periods of time. Rigorous synaptic quality control mechanisms likely provide defense against the detrimental effects of functionally impaired synaptic proteins. Indeed, Huntington's, Alzheimer's, Parkinson's, Amyotrophic lateral sclerosis and prion diseases are caused by defects in protein folding, underlying the biological importance of the problem of aberrant protein folding in neurons. What synaptic mechanisms mediate the balance between protecting proteins and preventing accumulation of misfolded proteins remains a current biological question.
Cysteine string protein (CSPa) is a 34 kDa synaptic vesicle protein and molecular chaperone that is critical in the defense against neurodegeneration. CSPa, so called because it contains a 25 amino acid domain comprising a string of 13-15 cysteine residues, is a member of the conserved J protein family based on the presence of a 70 amino acid signature J domain [1]. CSPa is abundant in neural tissue and displays a characteristic localization to synaptic vesicles [2] as well as clathrin coated vesicles [3]. Furthermore, CSPa null-organisms display widespread neurodegeneration [4][5][6][7]. Deletion of the CSPa gene generates mice that are normal for the first 2-3 weeks of life followed by a progressive loss of muscle strength and motor coordination, neurodegeneration, blindness and premature death [5,6]. Although the underlying molecular mechanisms of neurodegeneration in CSPa-null mice have not yet been established, electron microscopic analysis indicates that degeneration begins presynaptically [5]. In Drosophila, CSPa knockout flies that survive to adulthood show paralytic uncoordinated sluggish movements, spasmic jumping, shaking and temperature sensitive paralysis. While the precise sequence of pathogenic events remains to be identified, the reported defects include a 50% reduction of nerve-evoked neurotransmitter release at 18-22uC, a drastic reduction in evoked release above 29uC, a reduced ability to maintain normal presynaptic Ca 2+ levels and reduction of synaptic boutons at neuromuscular junctions [8][9][10][11][12][13]. Reduction in synaptic transmission, temperature sensitive paralysis and premature lethality are reversed by the expression of normal CSPa [9,10,12].
The J domain of CSPa interacts with and activates the ATPase activity of Hsc70 (70 kDa heat shock cognate protein) [14,15] and Hsp70 (70 kDa heat shock protein) [16]. Together with Hsc70 and SGT (small glutamine-rich tetratricopeptide repeat domain protein), CSPa assembles into an enzymatically active chaperone complex [17,18]. Following activation of the heat shock response, another member of the J protein family, Hsp40 (heat shock protein of 40kDa) assembles with the CSPa complex [19]. The presence of this chaperone complex on secretory vesicles suggests that CSPa is a coordinating anchor in key conformation/activity change(s) of client protein(s) critical in synaptic transmission. A number of client proteins for the CSPa system have been proposed including: G proteins, SNAREs (soluble N-ethylmalemide-sensitive factor attachment receptor), synaptotagmin, rab3, voltage sensitive calcium channels and CFTR (cystic fibrosis transmembrane conductance regulator) [20] and it is likely that neurodegeneration in null-organisms is due to the progressive misfolding and accumulation of dysfunctional client protein(s).
It has been suggested that interference with CSPa function (eg toxic proteins, environmental toxins) may be an underlying mechanism leading to neurodegenerative diseases [5]. It follows that relatively small changes to CSPa's activity would be expected to significantly affect neural survival, however there is currently no direct support for this notion. In this study, we begin to address the hypothesis that inhibition of CSPa activity may be common to the pathological sequence of events that underlies neurodegenerative disease and that the neuroprotective synaptic vesicle protein CSPa represents a promising therapeutic target for the treatment or prevention of neurodegeneration. Here we identify CSPa as a target for quercetin, a naturally occurring flavonoid. The Western diet contains ,25 mg/day of mixed flavonoids (quercetin ,70%). Quercetin is particularly high in apple skins, green tea and red grapes. It is also a major component of the nutrient supplements Ginkgo Biloba and St. John's Wort. Ginkgo Biloba is widely heralded as a memory enhancer but is also commonly taken for altitude sickness and cancer. St. John's Wort is commonly taken for depression. Ginkgo biloba leaves have been used for many centuries in traditional Chinese medicine and justification for quercetin supplements is historical rather than mechanistic. In this study we show that quercetin promotes CSPa dimerization and inhibits synaptic transmission as well as synapse development. The identification of quercetin as an agent that targets CSPa is a first step in the identification of pharmaceuticals that target members of the large J protein family and as such serves as proof of principle that pharmaceutical tools can selectively target J proteins. Given CSPa's anti-neurodegenerative properties, upregulation of CSPa may hold therapeutic promise in protecting nerve terminals from misfolded or toxic proteins [21].

Results
Quercetin stimulates formation of CSPa-CSPa dimers 70 kDa detergent-resistant CSPa dimers have been extensively reported in rat brain [16,22,23], various cell lines [24,25] and purified preparations [26] however, the role this dimer plays in CSPa-mediated conformational work is not known. In order to investigate the possibility that the neuroprotective synaptic vesicle protein CSPa can be targeted by pharmaceutical agents, we screened for drugs which initiate changes in CSPa dimerization. Figure 1A shows that under control conditions, a small fraction of total CSPa is detectable as a dimer in CAD cells (CNS-derived catecholaminergic neuronal cells) in addition to monomeric CSPa (both unmodified and palmitoylated). Post translational modification of CSPa, involving extensive fatty acylation (*) results in its retarded migration upon SDS-PAGE [22,27]. Interestingly,  exposure to quercetin stimulates increased CSPa dimerization in CAD cells. The CSPa dimer was detected by Western analysis with either anti-myc monoclonal or anti-CSPa polyclonal antibodies. The structure of quercetin (3,39,49,5,7-pentapentahydroxyflavone), a common dietary flavonoid, is shown in Figure 1B. In contrast to quercetin, the Hsp90 inhibitor geldanamycin, the neurotoxin MPP + , the oxidizing agent H 2 O 2 and the proteasome inhibitor lactacystin, did not stimulate CSPa dimerization, indicating that quercetin's effect on CSPa is selective. The quercetin-induced CSPa dimer is resistant to disruption by incubation in SDS-containing sample buffer at either 37uC or 80uC ( Figure 1C). SDS separates the vast majority of cellular protein complexes to monomers by treatment 37uC and more stable protein complexes at 80uC. The SDS and temperature resistance of the CSPa-CSPa complex demonstrates the great stability of the dimer. In contrast, no oligomerization or changes in expression of Hsc70 were observed. Actin detection is shown as a loading control. Our data demonstrate that the extremely stable CSPa-CSPa complex is selectively increased by quercetin.
We then asked if quercetin has the same effect in rat cortical neuronal cultures. Figure 2A shows that quercetin promoted CSPa dimerization in a concentration (20 mM-100 mM) dependent manner in cortical neurons expressing endogenous levels of CSPa. Increases in dimer were detectable at 20 mM quercetin. The expression levels of Hsc70 and Hsp40 chaperones which are known to associate with CSPa [14,15,15,16,19], were not altered by quercetin in treated cortical cultures. Actin is shown as a loading control. No CSPa dimerization was initiated by the DMSO vehicle control. The concentration-dependent induction of the CSPa dimer by quercetin in CAD cells is shown in Figure 2B. No changes in either the expression or formation of dimers were observed for the chaperones Hsc70, Hsp40, Rdj2 or the putative CSPa client proteins Gas and syntaxin, consistent with findings in cortical neurons. Figure 2C shows that relative differences in CSPa monomer and dimer expression are further observed in different rat brain regions. The CSPa dimer was highest in the thalamus, midbrain, entorhinal cortex and pons and low in the spinal cord and medulla. Taken together, these results show that a) in intact brain, the expression of CSPa-CSPa dimers is region specific, b) in cortical neurons CSPa dimerization is induced in the presence of quercetin in a concentration dependent manner and c) quercetin does not cause a generalized oligomerization of J proteins.
We next established the time course of quercetin's induction of CSPa dimers. Transfected CAD cells were treated with 100 mM quercetin, lysed at 5 min, 8 hrs, 24 hrs, 48 hrs and the formation of CSPa dimers evaluated by Western analysis with an anti-myc antibody and quantified by Biorad multiimager and QuantityOne software. Figure 3A shows that in the presence of quercetin the CSPa-CSPa dimer is 4 fold higher at 24 hrs compared to control. At 5 min, no difference in CSPa dimer expression was observed between control and quercetin treated CAD cells transiently expressing CSPa, suggesting that quercetin is acting to stabilize the CSPa dimer. Consistent with these findings Figures 1 and 2 Rdj2, syntaxin and Ga s were detected by Western analysis. b-actin is shown as a loading control. Data are representative of three separate experiments. (C) Native CSPa was detected in adult rat brain by Western analysis with a monoclonal anti-CSPa antibody. Twenty-five micrograms of unfractionated tissue homogenate isolated from the indicated regions of rat brain were separated by SDS-PAGE, transferred to PVDF and probed. Arrows indicate the CSPa dimer at ,72 kDa; * indicates a palmitoylated CSPa monomer at ,34 kDa. Actin is shown as a loading control. doi:10.1371/journal.pone.0011045.g002 demonstrate that regardless of whether CSPa is expressed as a heterologous protein, (i.e. transient expression in CAD cells) or present under native conditions (i.e. cortical neurons), quercetin increases dimer expression and this increase is likely due to enhanced dimer stabilization. In addition, the proteosome inhibitor lactacystin was not found to promote CSPa dimerization indicating that quercetin is not working as a general proteosome inhibitor. While these data are consistent with a direct interaction between CSPa and quercetin, they do not permit us to rule out the possibility that quercetin acts indirectly to initiate CSPa dimers. To investigate this possibility, we tested the ability of quercetin to directly inititate dimerization of purified recombinant soluble CSPa. Figure 3B shows that quercetin triggered dimerization of recombinant CSPa in a concentration dependent manner; we further observed that quercetin promoted dimerization of CSPa in a crude rat brain homogenate (data not shown), demonstrating that quercetin action on CSPa is direct.
To identify which amino acids are important for quercetininduced CSPa-CSPa dimer formation, we transfected CAD cells with a CSPa truncation construct CSPa 1-82 that encodes only the J domain, a CSPa deletion construct lacking the cysteine string region amino acids 113-136, CSPa D , or the neural J protein Rdj2 (also called DnaJA2). The lack of quercetin-induced dimerization of CSPa 1-82, CSPa D or Rdj2 clearly identifies the cysteine string region (residues 113-138) to be critically involved in dimerization ( Figure 3C). Rdj2 has a J domain and although it contains 11 cysteines throughout its amino acid sequence, it has no cysteine string region. The cysteine string region is a unique region to CSPa absent in other J proteins, emphasizing the notion that drugs may selectively target unique regions, thereby targeting distinct members of the J protein family. Taken together Figures 1, 2 and 3 demonstrate that quercetin selectively and directly interacts with CSPa to promote formation of highly stable CSPa-CSPa dimers and that this process is dependent on the cysteine string region of CSPa.

Quercetin inhibits synapse formation and synaptic transmission
Deletion of CSPa is ultimately linked to a state where the integrity of synaptic terminals is compromised. Therefore, the ability of quercetin to target CSPa prompts the question: does quercetin promote or inhibit synaptic function? To discern between these possible senarios we evaluated synapse formation in the fresh water snail Lymnaea stagnalis primary neuronal culture model that is uniquely suitable for precise measurements of synaptic transmission at a resolution not achievable elsewhere. In these experiments, the functionally defined respiratory neurons VD4 (visceral dorsal 4; presynaptic, cholinergic) and LPeD1 (left pedal dorsal 1; postsynaptic) were plated with their somata juxtaposed onto poly-L-lysine-coated dishes and the excitatory synapses allowed to develop for 12-18 hrs either in the presence or absence of 25 or 100 mM quercetin ( Figure 4A). Prior to intracellular recordings, quercetin was washed off for 1 to 2 hrs and VD4 presynaptic action potentials and LPeD1 excitatory postsynaptic potentials (EPSP) were recorded. As expected, in control synapses, current injection-induced action potentials in VD4 generated 1:1 EPSPs with averaged amplitudes of 9.861.7 mV in LPeD1 (n = 8) ( Figure 4A and B) similar to that seen in vivo [28][29][30]. However, in the presence of 25 mM quercetin, the mean amplitude of evoked EPSPs was significantly reduced to 2.560.7 mV (n = 5, p,0.05) ( Figure 4A and 4B, insert). Five VD4/LPeD1 cultures failed to form synapses in the presence of 100 mM of quercetin, while synaptic transmission was detected in 3 neuron pairs but with greatly reduced amplitude (0.6+0.4 mV) (n = 5). Furthermore, quercetin caused clamping of tetanic action potential firing in 90% of presynaptic VD4 cells examined (n = 21) ( Figure 4A). We next measured the amplitude of potentiated EPSP (pEPSP) following a tetanus compared to the amplitude of the EPSPs before tetanus ( Figure 4B, insert). The ratio of pEPSP and EPSP was significantly reduced by 100 mM (n = 8) but not by 25 mM (n = 5) of quercetin. Overall, our data establish that quercetin impaired synapse formation and reduced synaptic plasticity in Lymnaea and thus support the hypothesis that quercetin targets and inhibits CSPa function leading to a loss of synaptic integrity.
As expected, Western blot analysis revealed both CSPa monomers and quercetin-induced CSPa-CSPa dimers in Lymnaea ( Figure 5A). Lymnaea were maintained in pond water containing 100 mM quercetin for 12-18 hrs and the VD4 ganglia were then harvested and neuronal proteins were resolved by SDS-PAGE. Native CSPa was detected by Western analysis with the CSPa  polyclonal antibody generated against the C terminus of rat CSPa. Quercetin-induced CSPa dimer formation in Lymnaea respiratory neurons ( Figure 5A) was similar to that found for transfected CAD cells and cortical neurons (Figures 1 and 2). Confocal microscopy confirmed that CSPa is abundant in paired Lymnaea neurons, with highest localization observed at the plasma membrane ( Figure 5B). No signal was obtained with either quercentin alone or secondary antibody alone (data not shown). In addition to Lymnaea, CSPa homologues have been reported in Torpedo, Xenopus, Drosophila and numerous mammals; this high degree of evolutionary conservation of the CSPa system makes a strong case for an important cellular function. Of course, J proteins are even more extensively conserved and have been reported in bacteria, plants and viruses [20] where they most certainly are involved in conformational work. Our data provide unequivocal evidence that the the sensitivity of the CSPa chaperone system to quercetin is conserved from mammals to Lymnaea.
To ask whether the functional defects observed following quercetin treatment of intact neurons were time dependent, we next tested the effects of acute application of quercetin on synaptic transmission. Strikingly, Figure 6 shows that bath application of 25 mM quercetin to functional VD4/LPeD1 cultured neurons rapidly reduced synaptic transmission within 10 minutes. Neurons were paired overnight and on day two, simultaneous pre-and postsynaptic intracellular recordings under control conditions clearly showed that induced action potentials in VD4 produced 1:1 EPSPs in LPeD1 and these were potentiated significantly after a brief presynaptic tetanus. The amplitude of EPSPs in LPeD1 was greatly diminished within minutes of perfusion with 25 mM quercetin ( Figure 6A) indicating that quercetin rapidly inhibited synaptic transmission. A typical action potential is shown before (red) and after (blue) exposure to quercetin. Quercetin prolonged the presynaptic repolarization phase of the action potential, thereby lengthening action potential duration and rendering the neuron incapable of rapid and repetitive firing. Washout of quercetin (10-20 min) did not reverse the reduction in EPSP ( Figure 6B) suggesting that quercetin-dependent alteration of CSPa is irreversible in the short term. In the absence of an action potential quercetin did not stimulate exocytosis and spontaneous EPSPs were not observed following exposure to quercetin (data not shown). Larger reductions in EPSP amplitude were observed with 100 mM compared to 25 mM quercetin. The ratio of the pretetanus EPSPs to that of post-tetanus was reduced following quercetin treatment ( Figure 6C) demonstrating a reduction in the ability of neurons to exhibit short-term synaptic plasticity.
Even more telling, quercetin did not alter the post synaptic response to acetylcholine ( Figure 7A). Postsynaptic LPeD1 cells were cultured overnight (12-18hrs) and 1 mM Acetylcholine (ACh), the transmitter released from presynaptic VD4, was exogenously applied before and after exposure to quercetin (25 or 100 mM) for 30 mins while the postsynaptic potential was monitored. ACh elicited excitatory postsynaptic potentials which triggered firing of action potentials both before and after the exposure to either 25 or 100 mM quercetin for 30 mins (n = 4). At higher concentrations of quercetin LPeD1 action potentials showed a slower rate of repolarization.
To further test for quercetin-induced changes in presynaptic function, action potentials were recorded simultaneously with intracellular Ca 2+ concentration [Ca 2+ ]i measurements before and after perfusion of quercetin (25 mM) ( Figure 7B). VD4 cells were cultured overnight and then loaded with the fluorescent Ca 2+ indicator, Fura-2 AM. Elevations in intracellular free [Ca 2+ ] in response to single action potential and action potential bursts were significantly reduced after exposure to quercetin for 20 mins (n = 4) compared to control neurons ( Figure 7B insert). Because 100 mM quercetin induced action potential clamping ( Figure 4A and 6A), we opted to directly measure presynaptic Ca 2+ currents (ICa) by whole cell voltage clamp with direct depolarizing steps from holding potential of 280 mV to +70 mV in 10 mV increments. Exposure to 100 mM quercetin for 15 min reduced depolarization-induced Ca 2+ current (Figure 7c). Figure 7C left panel shows raw traces of ICa induced by a depolarization step from 280 mV to +20 mV before and after exposure to quercetin 100 mM for 15 min. Normalized current-voltage relations of ICa, (Figure 7C right panel) shows that the peak current occurred at +20 mV and the normalized peak current was significantly reduced from a control level of 0.9060.10 to 0.4960.10 (n = 4, P,0.05) after 15 mins of 100 mM quercetin. These observations indeed indicate that quercetin indeed rapidly inhibits action potentialinduced Ca 2+ entry and reduces synaptic transmission in Lymnaea neurons, consistent with the hypothesis that quercetin binds to and inactivates CSPa, leading to a sequence of events that involves disabling ion channels (eg voltage sensitive Ca 2+ channels). We have previously shown that CSPa promotes G protein-mediated inhibition of N-type Ca 2+ channels [23,31] thereby directly regulating Ca 2+ channel activity. Of note, CSPa has been suggested to be a Ca 2+ channel chaperone, but this notion has been controversial [32][33][34][35][36][37][38]. In addition, previous studies have concluded that quercetin is both an L type Ca 2+ channel activator [39] and inhibitor [40,41] as well as a BKCa channel activator [42], Kir channel inhibitor [43] and Ca 2+ ATPase inhibitor [44,45]. Although our data in Lymnaea VD4/ LPeD1 respiratory neurons support the idea that quercetin's inhibition of CSPa activity leads to downstream inhibition of synaptic transmission, these findings together with the above stated reports could potentially also be explained by a direct block of presynaptic Ca 2+ channels by quercetin. Previous studies have demonstrated that lowering extracellular Ca 2+ also blocks synapse formation between cultured Lymnaea neurons [46] showing the importance of Ca 2+ in the activation of synaptogenesis.
To establish more definitely if channel blockers generally altered CSPa dimer levels, we again utilized the transiently transfected CAD cell model. No significant differences in CSPa dimer formation was observed following treatment of transiently transfected CAD cells with CdCl 2 , nifedipine, D-2-amino-5phosphonovalerate (D-APV) or tetrodotoxin (TTX) ( Figure 7C). Furthermore, lowering cytosolic free [Ca 2+ ] by incubation with the membrane permeable Ca 2+ chelator BAPTA-AM (20 mM), a Ca 2+ chelator, for 1 hour did not stimulate CSPa dimer formation or influence quercetin's induction of the dimer. Taken together, these data indicate that none of the channel blockers examined promoted CSPa-CSPa dimer formation, suggesting that ion channel block per se or reduction in cytosolic free Ca 2+ does not stimulate CSPa dimerization, while quercetin impaired presynaptic Ca 2+ influx and reduced neuronal ability to maintain normal Ca 2+ levels. Independent of whether quercetin directly targets CSPa and secondarily disables channels or directly targets channels as well as CSPa, the results shown in Figures 4, 5 and 6 demonstrate that quercetin promoted dimerization of CSPa, reduced the amplitude of EPSPs in LPeD1 Lymnaea neurons and rapidly triggered a prolongation of presynaptic repolarization associated with inhibition of synaptic transmission and reduced synapse formation.

Increases in the CSPa:CSPa dimer correlate with a decrease in active CSPa:Hsc70 units
CSPa-mediated conformational work relies on the assembly of a CSPa with the ATPase Hsc70 [14][15][16][17]. To analyze the influence of quercetin on the assembly of the CSPa active chaperone complex, CSPa was immunoprecipitated from CAD cells treated with quercetin and the co-association of Hsc70 was evaluated. The CSPa dimer, monomeric palmitoylated CSPa, and monomeric unpalmitoylated CSPa immunoprecipitated from CAD cell lysates. Following treatment with quercetin, higher levels of CSPa dimer immunoprecipitated as expected. Immunoprecipitation of the monomeric palmitoylated CSPa (but not the unpalmitoylated) species was relatively lower following quercetin. Figure 8 shows  that quercetin treatment reduced Hsc70 association with CSPa indicating that quercetin inhibits assembly of the active chaperone complex. These results suggest that the quercetin-induced changes in CSPa dimerization and synaptic transmission involve inhibition of CSPa chaperone activity.
Finally, we asked whether other flavonoids also elicit an increase in the abundance of CSPa-CSPa dimers. Exposure to epigallocatechin gallate (EGCG) increased CSPa dimerization in CAD cells like that observed for quercetin (Figure 9). The CSPa dimer was detected by Western analysis with anti-myc monoclonal antibody. Figure 9B shows that the ratio of CSPa dimer to monomeric palmitoylated CSPa was higher following quercetin treatment then EGCG treatment. Actin is shown as a loading control. We conclude that quercetin and EGCG are potent activators of CSPa dimerization and predict that the common polyphenol structure is a central structure involved in increasing CSPa dimerization.

Discussion
Our work fulfills the proof of principle that pharmacological compounds like quercetin can target select members of the J protein family to modulate their cellular behavior. Specifically, our data show that the flavonoid, quercetin, interferes with CSPa function to maintain synaptic integrity. A current working model of the CSPa complex is illustrated in Figure 10. In this model, CSPa is anchored to the secretory vesicle and on its own is inactive. At least two other CSPa complexes exist: the active Hsc70/SGT/CSPa complex, and the CSPa dimer. CSPa displays unique anti-neurodegenerative properties and impairments in CSPa activity lead to impairment in synaptic transmission. In our model, quercetin targets CSPa, increases the abundance of stable CSPa-CSPa dimers and reduces assembly of CSPa with Hsc70, thereby reducing the folding capacity of the CSPa complex. CSPa dimerization may be a cause for the inhibition of neurotransmission, synaptic plasticity and synapse formation, essential features of normal brain function but other possibilities exist. Mechanistic insight into neurodegeneration following CSPa-inoperation remains limited but ultimately a reduction in synaptic folding capacity and a progressive buildup of unfolded presynaptic client proteins most likely creates a situation where the integrity of synaptic terminals is compromised.
Quercetin does not initiate widespread oligomerization of J proteins (Figures 2 and 3). The human genome encodes for over 40 J proteins with specific cellular and subcellular distributions [1]. The importance of the J protein family in protein folding has been recognized for many years (Reviewed: [47,48]). Members of the J protein family have a modular architecture in which a signature J domain is grafted to other sequences that impose specific cellular functions. The J domain is a 70 amino acid signature region comprised of four helices with a highly conserved tripeptide of histidine, proline and aspartic acid (HPD motif) located between helices II and III that is critical for chaperone activity. Hsc70 is targeted to a particular substrate through binding to the ''J domain'' of its partner and regulates the conformation and activity of the target protein via mechanisms that appear to involve cycles of substrate binding and release, which are governed by ATP binding and hydrolysis. Since J proteins are thought to provide the basis for selective chaperone action in the cell they are also promising therapeutic targets for the manipulation of specific protein folding processes [21].
Many questions remain about the biochemical pathway(s) responsible for CSPa-mediated neuroprotection. It is obvious that the CSPa chaperone unit is highly conserved machinery built upon the universal J domain/Hsc70 association and contributes to presynaptic protection. Furthermore, it is known that other J proteins do not compensate for the absence of CSPa, consistent with the idea that CSPa is designed to facilitate specific synaptic folding events and that the unfolded CSPa client is toxic. That said, significant efforts towards establishing the identity of CSPa client proteins and the underlying molecular details of CSPa's essential synaptic quality control are ongoing. Regulators of the degeneration seen in CSPa knockout models have been identified. Chandra and colleagues have shown that a-synuclein, a small neural protein whose biological function is unclear, selectively modulates the CSPa neurodegeneration pathway [7]. CSPa deficient mice are rescued from presynaptic degeneration and lethality by the overexpression of a-synucein. Moreover, ablation of endogenous a-synuclein accelerates the degeneration of presynaptic terminals observed in mice lacking CSPa. Furthermore, asynuclein specifically rescues CSPa deletion, but transgenic asynuclein does not rescue the spinal cord degeneration in mice that express mutant superoxide dismutase. Precisely how asynuclein abolishes neurodegeneration triggered by the absence of CSPa is unclear. Results reported here show that the cellular events following exposure to quercetin include i) a reduction in presynaptic Ca 2+ and ii) an increase in the presynaptic repolarization phase rendering neurons incapable of firing continuously, diminishing the amplitude of EPSPs and preventing additional synapse formation. One possibility is that following reduction of CSPa folding activity, a downstream progressive misfolding of protein(s) central to Ca 2+ homeostasis leads to inhibition of synaptic transmission. Another possibility is that quercetin has cellular targets in addition to CSPa perhaps directly blocking voltage dependent Ca 2+ channels. Although the precise mechanism by which quercetin alters Ca 2+ homeostasis remains to be established in detail, it is notable that lowering of cytosolic free [Ca 2+ ] with either the Ca 2+ chelator BAPTA-AM or the Ca 2+ channel blocker CdCl 2 does not stimulate CSPa dimerization, demonstrating that CSPa dimerization is not a general cellular response to Ca 2+ channel blockade. Independent of which explanation is correct, our results illustrate that quercetin targets CSPa and impairs synaptic function.
If quercetin is so powerful in inhibiting synaptic transmission, why do so many choose to supplement their diet with flavonoid mixtures (eg Ginkgo biloba)? Flavonoids are a class of compounds with polyphenolic structures and in all likelihood a spectrum of physiological functions can be expected depending on the position of hydroxyls and side chains. Our results indicate that the common structural features of quercetin and EGCG are important for triggering CSPa dimerization (Figures 1 and 9). Such compromised CSPa activity could contribute to the rate of progression of neurodegenerative (misfolding) diseases (eg. Alzheimer's disease, Huntington's disease) however there is no direct evidence for this notion. Neurotoxicity would depend on the flavonoid mixture utilized, metabolites and final concentration in the CSF (cerebrospinal fluid) after oral intake. Not all related compounds would necessarily inhibit CSPa activity, and it remains to be established if select flavonoids would enhance rather than inhibit CSPa function. Direct evidence supporting a role for flavonoids in memory enhancement is currently not available.
In view of the crucial importance of CSPa in synaptic integrity, our data identify a key role for compounds that interfere with its specialized presynaptic function. We speculate that neuropathological abnormalities could be due to or exacerbated by toxins with inhibitory actions towards CSPa similar to that of quercetin. Conversely, compounds related to quercetin may be found to enhance rather than inhibit CSPa activity. Thus, the identification of quercetin as tool that selectively modulates CSPa's neuroprotective function is a promising lead towards the identification of agents that enhance CSPa's neuroprotective function and thereby have a high potential in therapy development for neurodegenerative diseases. Our study further predicts that compounds that selectively target J proteins may have considerable potential as novel therapeutic agents.

Reagents and chemicals
Anti-CSPa rabbit polyclonal was prepared as described previously [22]. Anti-Hsp40 rabbit polyclonal were from Stressgen. Anti-Hsc70 mouse monoclonal, anti-b-actin mouse monoclonal, anti-syntaxin mouse monoclonal, quercetin, forskolin, MPP + and lactacystin was from Sigma. Anti-c-myc mouse monoclonal was from Clontech. Anti-CSPa mouse monoclonal was from BD Biosciences. Anti-Ga s rabbit polyclonal and geldanamycin were from Calbiochem. Anti-Rdj2 mouse monoclonal was from Abnova. H 2 O 2 was from VWR. ACh was obtained from Research Biochemicals (Natick, MA; product A-112) CAD mouse neuroblastoma cells [49,50] were seeded into 6 well plates and grown in DMEM/F12 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin as previously described. For differentiation cells were grown in Opti-MEM for 3 days. Cells were lysed in 40 mM Tris (pH 7.4), 150 mM NaCl, 2 mM EDTA, 1 mM EGTA, 1 mM Na 3 VO 4 , 0.1% SDS, 1% T-X100, 0.5 mM PMSF and protease inhibitor (Sigma) at 4uC for 1 hour. Lysates were centrifuged at 150006g for 5 minutes at 4uC and the supernatant was collected. Protein concentration was determined using a Bradford style assay kit (BioRad). For transient transfection, CAD cells were washed in PBS and transiently transfected with c-myc tagged rat CSPa  DNA using Lipofectamine-2000 (Invitrogen) in Opti-MEM and maintained in culture for 24 hours prior to treatments.

Rat cortical neurons
Rat primary cortical neurons (Cryopreserved) were purchased from QBM Cell Science (Ottawa, Canada) and stored in liquid nitrogen. Prior to culture, cells were thawed and gently transferred into pre-warmed neurobasal medium (Invitrogen, No. 21103-049) supplemented with 2% B27 (Invitrogen, 17504-044). Cells were then plated onto poly-D-lysine and laminin coated cover slips and maintained in neurobasal medium for 7 days at 37uC in 5% CO 2 prior to treatment with quercetin for 24 hours. Cells were harvested with lysis buffer and equal volumes of protein lysate were resolved by SDS-PAGE.

Lymnaea ganglia
Equal numbers of whole ganglia were harvested and lysed at 4uC for 40 minutes and resolved by SDS-PAGE.
Fresh water snails, Lymnaea stagnalis, were maintained at room temperature (22-23uC) in a well-aerated aquarium containing filtered pond water. Neurons were isolated from 1-2 month old snails with a shell length of 20-22 mm and the Lymnaea brainconditioned medium was prepared using 3-6 month old animals with a shell length of 25-30 mm. The cell isolation and cell culture procedures have been described in detail elsewhere [51]. Briefly, the Lymnaea were dissected and central ring ganglia were removed. Following treatment with Trypsin (2 mg/ml) for 23 mins, the central ring ganglia were then treated with trypsin inhibitor (2 mg/ml) for another 15 mins. Identified presynaptic neuron visceral dorsal 4 (VD4, cholinergic) and postsynaptic neuron left pedal dorsal 1 (LPeD1) (for synaptogenesis and synaptic transmission experiments) were isolated by applying gentle suction through a fire-polished and Sigma-coat (Sigma, St. Louis, MO)treated pipette. Isolated cells were then plated onto poly-L-lysine coated glass dishes in the presence of either medium (L-15; Life Technologies, Gaithersburg, MD; special order) or conditioned medium which contains trophic factors. Soma-soma synapses were prepared by juxtaposing VD4 and LPeD1 cell bodies against each other. The synapses developed overnight as described previously and were tested through direct intracellular recordings [28].

Immunoblotting
Proteins were electrotransferred from polyacrylamide gels to 0.45 mm nitrocellulose membrane in 20 mM Tris, 150 mM glycine and 12% methanol. Membranes were blocked with 4% milk solution (prepared in PBS with 0.1% Tween 20) and incubated with primary antibody for 2 hours at room temperature or overnight at 4uC. The membranes were washed in blocking solution and incubated with horseradish peroxidase-coupled secondary antibody. Antigen was detected using West Pico reagent (Pierce Biotechnology Inc.). Immunoreactive bands were visualized following exposure of the membranes to Kodak film. Bound antisera were quantitated by Biorad Fluor-S MultiImager Max and QuantityOne 4.2.1 software.

Immunoflourescence
Lymnaea neurons were plated on coverslips and maintained in Lymnaea brain conditioned medium with or without quercetin overnight (12-18 hrs). Neurons were washed in PBS and fixed in 2% paraformaldehyde for 1 hour at room temperature then washed 3 times with PBS for 10 minutes each. After blocking in 10% goat serum solution containing 0.3% Triton-X100 for 30 minutes cells were incubated with primary antibody in the blocking solution for 2 hours at room temperature or overnight at 4uC, washed 3 times with PBS for 10 minutes each, and incubated with goat anti-mouse conjugated to Alexafluor 546 secondary antibody in the blocking solution for 1 hour at room temperature. Cells were then washed 3 times in PBS for 10 minutes each, mounted with DABCO (Sigma) and photographed with a confocal microscope (LSM 510 Meta, Zeiss, Germany) under a 63X oil immersion objective at 543 nm excitation wavelength and images were collected using a band-pass filter (560-615).

Electrophysiology
Neuronal activity was recorded using conventional intracellular recording techniques as described previously [30]. Glass microelectrodes (1.5 mm internal diameter; World Precision Instruments, Sarasota, FL) with tip resistances of 20-50 MV were filled with a saturated solution of K 2 SO 4 . Neurons were impaled using Narashige (Tokyo, Japan) micromanipulators (MM202 and MM 204) (Axiovert 135; Zeiss, Thronwood, NY) on an inverted microscope. To test for synaptic connections, current was injected into the presynaptic neuron VD4 via an intracellular microelectrode which induced action potentials in the VD4 cell, and postsynaptic responses in LPeD1. The recorded electrical signals were displayed on a digital oscilloscope (PM 3394; Philips, Eindhoven, Netherlands) and relayed through a digitizer (Digidata 1322A, MDS Inc, Toronto, Canada) and recorded on a computer using Axoscope 9.0 software (MDS Inc, Toronto, Canada). Acetylcholine (ACh, 1 mM) was pressure applied (10-20 psi, 0.5-1 s duration) directly to the somata through a glass pipette (2-4 mm tip diameter) connected to a PV800 pneumatic picopump (World Precision Instruments).
Whole-cell recordings of voltage-gated Ca 2+ currents (I Ca ) were performed using a Multiclamp 700B amplifier (Axon Instruments) connected to an analog-to-digital interface Digidata 1322 (Axon Instruments). Signals were acquired and stored on a personal computer equipped with pClamp 9.2 software (Axon Instruments). Borosilicate pipettes (A-M Systems, Inc, Sequim, WA) were pulled using a Sutter P-97 microelectrode puller (Sutter) and the pipette resistance was 3-7 MV after being filled with pipette solution containing: 35 mM CsCl; 1 mM CaCl 2 ; 2 mM MgATP; 10 mM EGTA; 10 mM HEPES; adjusted to pH 7.4 with CsOH. The external bath solutions used to isolate Ca 2+ currents contains: 10 mM CaCl 2 ; 1 mM MgCl 2 ; 45.7 mM TEA-Cl; 10 mM HEPES; 5 mM 4-AP; adjusted to pH 7.9 with TEA-OH. I Ca in the presynaptic VD4 cells were elicited by depolarizing the cells from a holding potential of 280 mV to +70 mV in 10 mV steps., The Ca 2+ current data were analyzed using Clampfit 9.0 software (Axon Instruments) and traces were plotted using OriginPro 8.0 SRO (Northampton, MA, USA).

Ca 2+ imaging
Fura-2 AM (Molecular Probes, Carlsbad, CA), a membrane permeable and ratiometric Ca 2+ sensor, was used to determine changes in the intracellular Ca 2+ levels. A detailed Ca 2+ measurement procedure has been described elsewhere [52]. In brief, neurons were loaded with Fura-2 AM (10 mM) at room temperature (21-22uC) for 45 min. and were then exposed alternately to excitation wavelengths 340 and 380 nm using a high-speed wavelength switcher LAMBDA DG4 (Sutter Instrument, Novato, CA). The emitted fluorescence signal was collected at 510 nm by a Regiga Exi camera. Images were acquired with Northern Eclipse software running ionwave program (Empix Imaging, Canada). The free intracellular Ca 2+ concentration ([Ca 2+ ] i ) was estimated based on values obtained with a fura-2 Ca 2+ imaging calibration kit (F-6774, Molecular Probes) according to [53].

Immunoprecipitation
Immunoprecipitation was achieved by incubating detergent solubilized cells with the myc-monoclonal overnight at 4uC, followed by protein A/G agarose for 2 hrs at 4uC. Samples were washed, resuspended in 30ul of sample buffer, separated by SDS-PAGE, transferred to nitrocellulose and probed with antibodies for Western blot analysis.