|Home | About | Journals | Submit | Contact Us | Français|
Ischemic tolerance is an endogenous neuroprotective mechanism in brain and other organs, whereby prior exposure to brief ischemia produces resilience to subsequent normally injurious ischemia. While many molecular mechanisms mediate delayed (gene-mediated) ischemic tolerance, the mechanisms underlying rapid (protein synthesis independent) ischemic tolerance are relatively unknown. Here we describe a novel mechanism for the induction of rapid ischemic tolerance mediated by the ubiquitin-proteasome system. Rapid ischemic tolerance is blocked by multiple proteasome inhibitors (MG132, MG115 and clasto-lactacystin β-lactone). A proteomics strategy was used to identify ubiquitinated proteins following preconditioning ischemia. We focused our studies on two actin binding proteins of the post synaptic density that were ubiquitinated following rapid preconditioning: myristoylated, alanine-rich C-kinase substrate (MARCKS) and fascin. Immunoblots confirm the degradation of MARCKS and fascin following preconditioning ischemia. The loss of actin binding proteins promotes actin reorganization in the post synaptic density and transient retraction of dendritic spines. This rapid and reversible synaptic remodeling reduces NMDA-mediated electrophysiological responses and renders the cells refractory to NMDA receptor-mediated toxicity. The dendritic spine retraction and NMDA-neuroprotection following preconditioning ischemia are blocked by actin stabilization with jasplakinolide, as well as proteasome inhibition with MG132. Taken together these data suggest that rapid tolerance results from changes to the post-synaptic density mediated by the ubiquitin-proteasome system, rendering neurons resistant to excitotoxicity.
Stroke is the third leading cause of death in the US today (AHA, 2003). It is a disorder without effective acute neuroprotective treatment. For this reason, attention has recently been brought to defining the brain’s own evolutionarily conserved endogenous neuroprotective mechanisms, that occur in ischemic tolerance or following ischemic preconditioning. In these settings, prior exposure to a non-harmful duration of ischemia renders the brain resilient to a subsequent harmful ischemic episode (Kitagawa et al., 1990; Dirnagl et al., 2003). Prolonged ischemia results in excitotoxic/ programmed cell death (Choi, 1992) and mechanisms of ischemic tolerance would be expected to inhibit these cellular processes.
Two mechanisms of ischemic tolerance have been described. Classic, or delayed, ischemic tolerance requires new protein synthesis and results in protection 24-72 hours following the preconditioning stimulus (Barone et al., 1998; Dirnagl et al., 2003; Meller et al., 2005). In contrast, rapid ischemic tolerance does not require new protein synthesis and produces neuroprotection within one hour of the preconditioning event (Perez-Pinzon et al., 1996; Reshef et al., 1996; Perez-Pinzon and Born, 1999; Meller et al., 2006). The relatively short time required for induction of rapid ischemic tolerance suggests that it is regulated by post translational neurochemical events. In addition, rapid ischemic tolerance reduces the activation of programmed cell death associated caspases (caspase3) following normally harmful ischemia (Meller et al., 2006).
The ubiquitin-proteasome system is the cell’s major extra-lysosomal system for protein degradation (Verma and Deshaies, 2000; Herrmann et al., 2007). The addition of poly-ubiquitin chains to a protein results in its translocation to the proteasome, a large multi-subunit protease, for degradation. We have recently identified the ubiquitin-proteasome system as playing an essential role in rapid tolerance to ischemia. Tolerance to ischemia can occur via the selective ubiquitination and degradation of a cell death associated protein, Bim (Bcl-2–interacting mediator of cell death). Rapid ischemic tolerance and Bim degradation occurs independent of new protein synthesis and is blocked by the proteasome inhibitor MG132 (Meller et al., 2006).
Here we further pursue the role of the ubiquitin/ proteasome system in mediating rapid molecular and cellular changes following preconditioning ischemia, which may underlie the neuroprotective phenotype of rapid ischemic tolerance. Accordingly a proteomic analysis of preconditioned neuronal cultures was performed to identify proteins ubiquitinated following preconditioning ischemia. In our study the majority of identified proteins have a role in the form and function of the post-synaptic density. We therefore focused our study of this proteomic analysis on two ubiquitinated proteins that had a structural relationship to the post-synaptic density: MARCKS and fascin. Here we show that following a preconditioning duration of ischemia/reperfusion, the ubiquitination and degradation of cellular proteins results in the reorganization of the actin cytoskeleton, remodeling of post-synaptic dendrite architecture, with a resultant selective attenuation of toxic NMDA receptor-mediated signaling at the time when tolerance to ischemia is acquired.
Cortical neuronal cultures were prepared from 1-2 day old Sprague Dawley rat pups using previously described methods (Meller et al., 2006). All experiments were performed in accordance with the American animal protection legislation and approved by the Legacy Health System Institutional Animal Care and Use Committee. For all experiments cells were used after 10-14 days in culture, when cultures consist of approx 80-90 % neurons. For western blotting, P62 pulldown and immunoprecipitation experiments, cells were plated out onto 3.5 or 6 cm poly D lysine-coated culture dishes (Primera; Becton Dickinson, San Jose, CA). For oxygen and glucose deprivation (OGD) experiments, cells were washed twice in PBS (0.5 mM CaCl2, 1 mM MgCl2; pH 7.4) and then incubated for various times in an anerobic chamber (Forma Scientific, Marietta, OH)(85 % N2, 5% H2, 10 % CO2; 35 °C) in PBS at 37 °C. Following ischemia, cells were replenished with Neurobasal A media and placed into a normoxic incubator.
Some cells were incubated with pharmacological compounds for one hour during the recovery phase following 30 min of tolerance inducing OGD and prior to 120 min harmful ischemia (Fig 1A). MG132 and jasplakinolide were purchased from Calbiochem (San Diego, CA), MG115 and clasto-lactacystin-β-lactone were purchased from Biomol (Plymouth Meeting, PA).
Cell death was assessed by propidium iodide exclusion as previously described (Meller et al., 2005; Meller et al., 2006). Briefly cells are incubated with propidium iodide (1.5 μg/ ml in PBS: Sigma St Louis, MO) for 3 min, then fixed in 4 % formalin at room temperature, permeabilized in 0.1 % Triton X-100/ phosphate buffered saline and mounted in DAPI containing media (Vector Labs, Burlingame, CA). Cells were visualized using a Leica microscope under Ex/Em wavelengths of 340/425 nm (blue), and 580/630 nm (red) respectively. Images were collected using an Optronics DEI-750 3-chip camera equipped with a BQ 8000 sVGA frame grabber and analyzed using Bioquant (Nashville, TN). PI positive cells are expressed as a % of the total number of cells in the view (DAPI positive). The average of three random views of a coverslip were assessed and counted for each data point, and experiments repeated on (n) independent sets of cultures.
Cell lysates (10 mg) were prepared in a RIPA buffer (Meller et al., 2005), cleared and then incubated for 2 hours at 4 °C with P62 UBA beads (Biomol, Plymouth Meeting, PA). MG132 was included during the recovery to prevent degradation of ubiquitinated proteins. Precipitated proteins were subject to PAGE (poly acrylamide gel electrophoresis) and gel sections (approximate 10-30 kDa gel fragments) were analyzed at the proteomics facility at the Fred Hutchinson Cancer Research Center (Seattle, WA). Unstained gel slices were washed twice with water and “in-gel” protein digestion was carried out as described (Shevchenko et al., 1996), but omitting the reduction and alkylation steps. Samples were desalted using micro C18 ZipTips (Millipore) and dried, and resuspended in 5 microliters of 0.1% trifluoroacetic acid and analyzed by liquid chromatography electrospray ionization tandem mass spectrometry (LC/ESI MS/MS) with a LCQ “classic” mass spectrometer (ThermoElectron) using a previously described instrument configuration (Gatlin et al., 1998). Data were collected in a data dependent mode in which a MS scan was followed by MS/MS scans of the three most abundant ions from the preceding MS scan. Mass spectrometry data were searched against rat or human protein databases (subsets of proteins from the NCBI nonredundant protein database) using the software search algorithm COMET (Institute for Systems Biology). Comet results were considered valid if at least 2 unique peptides identified back to a protein and if the peptide matches had raw scores greater than 200 for +1 ions, 300 for +2 ions, and 300 for +3 ions, Z-scores greater than 4, dN greater than 0.1, and %ions of greater than 30% and the identification did not appear in a control sample from a blank portion of the gel. Proteins present in both control and OGD-treated samples were omitted from further analysis (Supplementary Table 1).
To determine the Z score, the dot product between the experimental spectrum and a theoretical spectrum is determined, with the resulting score scaled to 1000. Z score is the number of standard deviations away from the mean of the top scoring peptide compared to the top 500 clustered peptide scores. The difference between normalized scores is the dN value. The % ionic is the percentage of matched peptide fragment ions over the total number of expected fragments ions for the best matching peptide.
Immunoblotting was performed as previously described (Meller et al., 2002). Protein samples (50 μg) were denatured, subject to polyacrylamide gel electrophoresis and transferred to polyvinylodene difluoride membranes (Biorad), prior to incubation with primary antibodies at 4 °C overnight; MARCKS, fascin and actin (Santa Cruz Biotechnology, Santa Cruz CA), MAP2B (Sigma, St Louis MO) NR1, NR2A, NR2B and PSD-95 (BD Biosciences, San Jose, CA) and phospho-CREB (Cell Signaling, Beverly MA). Membranes were incubated with anti-rabbit or anti-mouse IgG conjugated to horseradish peroxidase (Cell Signaling Technology, Beverly, MA, USA). Chemiluminesence (Visualizer™ Upstate, Charlottesville, VA) was captured and quantified using a Kodak Imagestation 2000RT.
Capture antibodies (5 μg) were cross-linked to protein A agarose using sulfo-SMCC (5 μg: Pierce, Rockford IL) washed and blocked with 1% BSA for 30 min at room temp, prior to incubating with cell lysates (750 μg) overnight at 4 °C. Precipitated proteins were washed using spin columns (Sigma), eluted by boiling in gel loading buffer, then subject to PAGE as for immunoblotting.
Cells were fixed with 4 % formalin, permeabilized with TX-100 (0.01%) and incubated with either antibodies specific to Fascin and MARCKS, or phalloidin (Oregon green 488) according to manufacturer’s directions (Molecular Probes, Eugene, OR). Primary antibodies were detected with specific secondary antibodies conjugated to CY3 or AlexaFluor 488. Immunostaining was studied using a Carl Zeiss Axioimager fitted with EC Plan Neo-Fluar (40x) or Plan Apochromat (100x oil) objectives under Ex/Em wavelengths of 359/461 nm (blue), 470/509 nm (green) and 550/570 (red). Images were captured using a Zeiss MRM camera and analyzed using Axiovision v4.6 software. Phalloidin staining was studied using a Leica microscope fitted with an objective under Ex/Em wavelengths of 359/461 nm (blue) and 500/550 nm (green). To obtain 100x images of Phalloidin staining we used the above Zeiss microscope fitted with an Apotome™. Phalloidin images were quantified using ImageJ (http://rsb.info.nih.gov/ij/). Briefly, images were converted to grayscale 8-bit images, auto thresholded to obtain a mask of objects, and objects with an area between 1 and 3.5 μm2 were counted and the density of objects determined.
Cultures were pre-incubated with 3,3’-dioctadecyloxacarbocyanine perchlorate (DiO) (Molecular Probes, Eugene, OR) (25 μg/ coverslip suspended in PBS) overnight. Cells were then subject to ischemia and then imaged using a 63x water immersion lens (HCX APO L 63x0.90 W) in a PBS solution containing glucose (4.5 g/l). Z-series stacks were obtained at 0.2 μm thickness using the Leica software. Image series were imported into Image J (NIH) and a Z-projection stack compiled of the max intensity of staining. The brightness and contrast were adjusted to visualize the spines and other dendritic features. Some cells were recovered in the presence of MG132 (1 μM) or jasplakinolide (1 μM) for 1 hour before viewing. Numbers of spines were determined per 10 μm length of dendrite from multiple dendrite sections from multiple neurons in (n) independent experiments. Primary dendrites were defined as those directly abutting the cell soma, secondary dendrites contacted a primary dendrite and tertiary dendrites contacted a secondary dendrite. Secondary and tertiary dendrites were only counted if their respective primary dendrite was also assessed.
Patch-clamp recordings were performed as described (Xiong et al., 2004). Patch electrodes, with resistances of 2 ~ 3 MΩ when filled with intracellular solution, were constructed from thin-walled borosilicate glass (1.5 mm diameter, WPI, Sarasota, FL). A multi-barrel perfusion system (SF-77, Warner Instrument Co., Hamden, CT) was employed to achieve a rapid exchange of extracellular solutions. Whole-cell currents were recorded using Axopatch 200B amplifiers (Axon Instruments, Foster City, CA). Data were filtered at 2 kHz and digitized at 5 Hz using a Digidata 1320 DAC unit (Axon Instruments). The on-line acquisition was done using pCLAMP software (Version 8, Axon Instruments). NMDA currents were activated at −60 mV by perfusion of neurons with a solution containing 100 μM NMDA plus 5 μM glycine. For each cell, a voltage step of -5 mV from the holding potential was applied to monitor the cell capacitance and access resistance. Recordings with access resistance larger than 15 MΩ were excluded from data analysis.
Data from cell death experiments are reported as mean ± sem of (n) independent experiments. Statistical analysis of cell death and morphological data was performed using one way analysis of variance (ANOVA) or two-way analysis of variance (ANOVA) followed by Bonferroni’s multiple comparison test using Graphpad Prism v 4.0 (GraphPad Software, San Diego CA). Electrophysiology recordings and phalloidin puncta quantification data were analyzed using unpaired two-tailed Student’s t test. Statistical significance was accepted at P<0.05.
We recently described a role for the ubiquitin-proteasome system in mediating rapid ischemic tolerance (Meller et al., 2006). The rapid ischemic tolerance paradigm is outlined in Fig 1A. Preconditioning was induced in primary rat neuronal cultures (10-14 DIV) by a non-harmful duration (30 minutes) of oxygen and glucose deprivation (OGD) (Fig 1B). A longer (120 min) duration of OGD resulted in an increase in PI uptake into cells (approximately 55% cell death) (Fig 1B and 1C). Preconditioned rat neuronal cultures were protected against a harmful ischemia (120 minutes duration) (Fig 1A-C).
To investigate the role of the ubiquitin-proteasome in rapid ischemic tolerance some cells were incubated with the proteasome inhibitor MG132 (Carbobenzoxy-L-leucyl-L-leucyl-L-leucinal). MG132 blocked the neuroprotection acquired in rapid ischemic tolerance when incubated with neurons following 30 min OGD preconditioning, but prior to harmful ischemic challenge (Fig 1B and 1C). Similarly, blocking the proteasome with MG115 (Carbobenzoxy-L-leucyl-L-leucyl-L-norvalinal) and clasto-lactacystin-β-lactone, also blocked rapid ischemic tolerance (Fig 1B and 1C). Neither MG 132, MG115 nor clasto-lactacystin-β-lactone significantly increased cell death in the cultures under basal conditions, when incubated with neuronal cultures for one hour (Fig 1B and 1C).
We initiated a proteomic strategy to further identify proteins ubiquitinated following preconditioning ischemia using an unbiased approach. Utilizing a ubiquitin-binding pull-down assay (Meller et al., 2006), we determined changes in protein ubiquitination following preconditioning ischemia. P62 (sequestome-1) shuttles ubiquitinated proteins to the proteasome (Babu et al., 2005). The pull-down of ubiquitinated proteins by P62 ubiquitin binding domain conjugated agarose beads was confirmed in an ex-vivo assay (Sup Fig 1A) and by mass spectrometry (Sup Fig. 1B). Following pull-down of proteins from control or preconditioned tissue, we performed one-dimensional gel electrophoresis and subjected gel sections to mass spectrometric analysis. Candidate proteins were considered for further analysis if the identifying peptides met the mass spectrometry criteria for high-confidence (see methods). Analysis of proteins identified by mass spectrometry yielded a list of 7 proteins exclusive to control samples and 17 proteins in OGD-treated (preconditioned) samples (Table 1). In addition, 34 proteins were identified as being present in both samples; these were excluded from further analysis (Sup Table 1). When we subdivided the candidate proteins by potential function, we found that 16 out of the 24 proteins exclusive to control or preconditioned tissue resided in the post-synaptic density, or interacted with proteins and signaling pathways in the post-synaptic density (see Table 1)(Husi et al., 2000; Husi, 2004).
The post-synaptic density is regulated by the ubiquitin-proteasome system (Ehlers, 2003). Actin is the major cytoskeletal protein regulating the highly dynamic structure of post-synaptic sites, especially dendritic spines. Accordingly, we focused our studies on two candidate proteins among those associated with the PSD which have actin binding functions, fascin and myristoylated, alanine-rich C-kinase substrate (MARCKS). Fascin, a 55 kDa protein, contains two actin binding domains and cross links actin filaments. Fascin plays a role in development and polarity of filopodia (early neurites) as well as growth cones (Kureishy et al., 2002). MARCKS (Myristilated alanine-rich C kinase substrate) is a 30 kDa protein (Wu et al., 1982) (but migrates at 80 kDa on SDS PAGE). MARCKS links and anchors the plasma membrane to the actin cytoskeleton (Hartwig et al., 1992). We further analyzed each of these proteins for PEST sequences (a predictor of proteolytic degradation (Rechsteiner and Rogers, 1996)) using the Pestfind algorithm (http://www.at.embnet.org/embnet/tools/bio/PESTfind/). Both MARCKS and fascin contain PEST sequences (3 and 5 PEST sites, respectively) and thus were subject to further investigation to identify their role in rapid ischemic tolerance.
Immunoblot analysis shows that levels of MARCKS and fascin are decreased one hour following 30 minutes preconditioning ischemia (approx. 77% and 67% of control, respectively) (Fig 2A and Supp Fig 3A). The degradation of both MARCKS and fascin was blocked by the proteasome inhibitor MG132 (Fig 2A and Supp Fig 3A), which also blocks rapid ischemic tolerance (Fig 1B)(Meller et al., 2006). In contrast, actin levels did not change following preconditioning ischemia (PC) or treatment with MG132 (Fig 2A). Following preconditioning ischemia the binding of actin to both fascin and MARCKS was reduced to 40% of control values (Fig 2B and quantified in Sup Fig 3B). The immunoprecipitation of MARCKS and fascin with actin was 93 and 62 % of control values, when cells were preconditioned and then treated with the proteasome inhibitor MG132 (Fig 2B and Supp fig 3). We did note that the immunoprecipitation of both MARCKS and fascin with actin was also reduced in cells treated with MG132 for 1 hour. However MG132 did not affect basal MARCKS or fascin protein levels (Fig 2A) and ICC staining patterns for MARCKS were similar between control and MG132 treated cells (data not shown): the significance of this observation is not clear, but is under further investigation. Taken together, these data suggest that MARCKS and fascin are degraded by the ubiquitin proteasome system following preconditioning ischemia, and this accounts for their reduced binding to actin.
We further investigated the localization of fascin and MARCKS following preconditioning using immunocytochemistry. Immunocytochemical studies of MARCKS (Fig 2 C-D) confirm our immunoblot data; fewer cells show MARCKS protein staining following ischemia (40 % vs. 20 % of DAPI positive cells). In control cells, fascin staining appears localized with neuronal processes, with only modest amounts present in the soma (Fig 2C). Following preconditioning ischemia the fascin staining in the dendritic processes appear reduced, but was more pronounced in the cell soma (Fig 2D).
To determine the effect of preconditioning ischemia-induced MARCKS and fascin degradation on the actin cytoskeleton, we used phalloidin to stain for filamentous actin. In control cells, actin staining revealed a punctate pattern along dendritic processes (Fig 3A and 3C). The punctate staining of phalloidin was reduced to 68 % of control levels one hour following preconditioning ischemia (Fig 3E). Furthermore, one hour following preconditioning ischemia we observed more filamentous staining pattern in the dendritic shaft of the cell and in the cell soma (Fig 3B and 3D). The phalloidin-staining pattern returned to normal at 4 hours following preconditioning ischemia (data not shown).
To support the phalloidin data showing changes in actin localization, cell lysates were prepared and centrifuged to separate globular actin (G-actin) remaining in the supernatant from filamentous actin (F-actin) in the pellet. The ratio of G-actin to F-actin following preconditioning ischemia was measured (Fig 3F & 3G). Immediately following preconditioning ischemia, the ratio of G-actin/ F-actin did not change compared to control 0 hrs (Fig 3F). Rather, the solubilization of F-actin occurred 0.5 and 1.0 hrs following preconditioning ischemia (Fig 3F).
The functional significance of actin reorganization in rapid ischemic tolerance was demonstrated in cells incubated with the actin stabilizer jasplakinolide (Halpain et al., 1998). Jasplakinolide blocked 30 min OGD induced rapid ischemic tolerance, but had no significant harmful effect on basal or 120 min OGD induced cell death (data not shown)(Fig 3H). The concentration of jasplakinolide used in this study stabilized F-actin content of the cells of the cells following ischemia (data not shown). Taken together these data show that preconditioning ischemia induces a reorganization of actin and that increased stabilization of filamentous actin using jasplakinolide prevents preconditioning-induced rapid ischemic tolerance.
To establish the morphological consequence of preconditioning ischemia-induced actin reorganization, we used the lipophilic membrane dye 3,3’-dioctadecyloxacarbocyanine perchlorate (DiO) to stain and assess dendritic spines in neurons. In control neurons, spines were visible on dendrites at a density of 4.3 ± 0.3 spines per 10 μm length (mean ± sem, n=8) (Fig 4A & 4B). The density of dendritic spines on our cultured neurons was similar to other studies using DiO and GFP transfected neurons (4.5 spine/ 10 μm length) (Hasbani et al., 2001b; Hasbani et al., 2001a; Kirov et al., 2004; Zha et al., 2006). The density of spines was not significantly different across primary (abutting the cell soma), secondary (abutting primary dendrites) or tertiary processes (abutting secondary processes) (4.3 ± 0.22, 4.3 ±0.3 and 4.4 ±0.3 spines per 10 μm dendrite length, respectively). In cells preconditioned with 30 min OGD (PC) subject to one hour recovery, dendritic spine density was reduced to 0.90 ± 0.3 spines per 10 μm length (Fig 4C & 4D) uniformly over primary, secondary and tertiary dendrites (1.02 ±0.3, 0.91 ±0.3 and 0.77±0.3 spines per 10 μm dendrite length, respectively) (Fig 4F). The number of dendritic spines was not significantly different from control levels 4 hours following preconditioning ischemia (Fig 4E), the time point when neuroprotection is lost (Meller et al., 2006). The proteasome inhibitor MG132, and the actin stabilizer jasplakinolide, block preconditioning induced neuroprotection (Fig 1B and Fig 3H). Both the actin stabilizer jasplakinolide and MG132 significantly blocked the loss of dendritic spines on all processes following preconditioning ischemia (Fig 4G + Supp Fig. 2A). However, neither jasplakinolide nor MG132 alone had a significant effect on spine density in non-ischemia treated cells (Fig 4G and Supp 2A).
Following preconditioning ischemia the dendrites also exhibit swelling and the appearance of varicosities (Fig 4C and 4D). The varicosity formations were reduced 4 hours following exposure to preconditioning ischemia (Fig 5A). Unlike the effect of ischemia on spine loss, more varicosities were formed on tertiary dendrites than on primary dendrites abutting the cell soma (Fig 5B). We also visualized dendrite varicosity formation using microtubule associated protein (MAP2) immunocytochemistry. MAP2 was uniformly distributed in the dendrites of control cells (Fig 5C), but following preconditioning ischemia the MAP2 staining appeared beaded on dendrites (Fig 5D). MG132 and jasplakinolide reduced the formation of varicosities following preconditioning ischemia (Fig 5E). When analyzed separately, jasplakinolide was less effective at reducing primary and secondary dendrite varicosity formation (Supp Fig 2B). These data suggest that following preconditioning ischemia the ubiquitin-proteasome system mediates dendritic spine loss and varicosity formation that results in changes in neuronal morphology, which may serve as a protective phenotype.
Dendritic spines constitute the predominant site of synaptic input into neurons. To determine if there was a functional correlate of dendritic spine changes, we investigated NMDA receptor-mediated responses in neurons following preconditioning ischemia. No change in NMDA receptor subunit expression was observed following preconditioning ischemia (Fig 6A and quantified in Supp Fig 3C). However, the precipitation of both the NMDA 2B receptor subunit and the post-synaptic density associated structural protein PSD-95 with actin, was decreased by 40%, in cells following preconditioning,an effect partially reversed by the proteasome inhibitor MG132 (Fig 6B and Sup. Fig 3D). These findings suggest that preconditioning ischemia results in the reconfiguration of post-synaptic receptors and associated signaling scaffolds with the actin cytoskeleton, both of which would impact postsynaptic signaling.
Prolonged NMDA receptor activation is toxic to neurons (Choi et al., 1988). To test whether post-synaptic responses to NMDA were changed following preconditioning ischemia, we measured NMDA-stimulated electrophysiological responses using whole cell patch configuration. Peak currents elicited by NMDA application (100 μM: 2 sec) were reduced by 30 % in cells following preconditioning when compared to control (Fig 6C and 6D). We also examined NMDA-mediated cell death in cortical neurons following preconditioning. NMDA (200 μM: 60 min) resulted in 44 % cell death in cortical cultures (Fig 6E). NMDA-induced cell death was reduced by 50 % in cells following preconditioning ischemia one hour prior to application of NMDA (Fig 6E). The rapid tolerance to NMDA was blocked by proteasome inhibition with MG132 and actin stabilization with jasplakinolide (Fig 6E). Taken together, these data are compatible with the morphologic alterations associated with preconditioning changing the structural relationship of NMDA receptors to the post-synaptic density resulting ni an attenuation of excitotoxic signaling mechanisms.
The mechanism by which the brain acquires rapid ischemic tolerance is unknown. Here we reveal a novel mechanism by which neurons become rapidly tolerant to ischemia in a protein synthesis independent manner and describe the tolerant phenotype acquired by the cell following the preconditioning process. We show that the downregulation of actin-binding structural proteins, mediated by the ubiquitin-proteasome system, results in actin re-organization, the retraction of dendritic spines and changes in post-synaptic NMDA receptor mediated signaling. These transient structural changes result in neuroprotection to ischemia and represent a novel approach to regulate cell death following harmful ischemia.
Protein modification by ubiquitination is a central feature of brain ischemia. Protein ubiquitination increases following harmful levels of ischemia, which may result in accumulation of protein aggregates and contribute to cell stress (Hu et al., 2000). Hypothermia or ischemic preconditioning prevent protein ubiquitination following harmful levels of ischemia (Liu et al., 2005; Liu et al., 2006). In addition, proteasome inhibitors have been shown to block the activation of the transcription factor NF-κB following harmful ischemia (Williams et al., 2004), resulting in protection. However, protein ubiquitination can also result in the selective and rapid downregulation of cell death associated proteins, such as Bim, to induce the neuroprotective state of rapid ischemic tolerance (Meller et al., 2006). In this current study we show that multiple inhibitors of the proteasome (MG132, MG115 and clasto-lactacystin-β-lactone) block rapid ischemic tolerance induced neuroprotection (Fig 1B), hence we investigated the global ubiquitination of proteins following preconditioning using a proteomics approach to identify mechanisms that may be relevant for the generation of a protective phenotype.
We began with a targeted proteomic analysis of ubiquitinated proteins in neuronal tissue following the induction of rapid ischemic tolerance modeled in vitro using an ubiquitin binding pull-down assay (Fig 1A) (Meller et al., 2006). Among the 24 proteins identified by mass spectrometry (Table 1), 16 were associated with the post synaptic density (Husi et al, 2000; Husi, 2004) (Collins et al., 2004). It has previously been shown that proteins in the post synaptic density are regulated by the ubiquitin-proteasome system and that these changes are activity related and reversible (Colledge et al., 2003; Ehlers, 2003; Bingol and Schuman, 2004; Joch et al., 2007). We show a structural component, evident during rapid tolerance, dendritic spine loss, which has been reported previously with sub-lethal doses of ischemia (Park et al., 1996; Hasbani et al., 2001a). The cytoskeleton protein actin regulates the structure and morphology of post synaptic sites (dendritic spines) (Wu et al., 1982; Hartwig et al., 1992; Kureishy et al., 2002). Therefore we focused our investigation on two of our candidate proteins associated with the post-synaptic density that could affect both synaptic structure and function: the actin binding proteins MARCKS and fascin.
Our data suggest that the degradation of MARCKS and fascin result in their reduced interaction with the actin cytoskeleton and relocation in the cell (Fig. 2A-B)Translocation of MARCKS from membrane to cytosol in muscle cells is mediated by protein kinase C (Disatnik et al., 2004). Interestingly in our model we observe a blockade of rapid ischemic tolerance with the protein kinase C inhibitor calphostin C (unpublished observation: RM). In a previous study using non-toxic doses of NMDA, Graber et al reported that cathepsin B mediated proteolysis of MARCKS results in dendritic spine collapse, but that the decrease in MARCKS was not blocked by the proteasome inhibitor MG132 (Graber et al., 2004). Hence two preconditioning stimuli (NMDA and ischemia) may induce the same net result in the neuron to induce a tolerant phenotype (loss of MARCKS and dendritic spine collapse), but via different biochemical mechanisms.
Our data strongly support the view that actin re-organization is involved in the generation of the protective phenotype following preconditioning ischemia. Neurochemically, actin is found in two forms, G-actin (globular) polymerizes into F-actin (filamentous) which forms the cytoskeleton associated with the postsynaptic membrane (Zigmond, 2004). F-actin is involved in the cytoskeleton within dendritic spines including those associated with glutamate clustering where the cytoskeleton may be in a dynamic state (Dai et al., 2000). Our data suggest that actin reorganization occurs following preconditioning ischemia and blocking such re-organization (with jasplakinolide) blocks rapid ischemic tolerance (Fig. 3). The effect of preconditioning ischemia on actin has not been studied, however some studies report actin mRNA are constant following harmful ischemia, up to 7 days(Onodera et al., 1989; Minami et al., 1992), whereas other studies show increases in actin mRNA levels, which may be due to gliosis in these regions (Kondo et al., 1994). Recently it has been reported that following harmful ischemia, the actin regulating protein WAVE-1 forms a complex with Bcl-XL, sequestering the anti-apoptotic function of Bcl-XL(Cheng et al., 2007), although it is not clear whether this complex is formed or inhibited by preconditioning ischemia. Interestingly we observed a decrease of WAVE-1 ubiquitination in our preconditioned cells (table 1) and this is currently under further investigation.
Spine remodelling is thought to underlie synaptic plasticity as well as being a consequence of excitotoxicity. Our data show that dendritic spines undergo transient retraction following preconditioning ischemia, coincident with actin remodelling (Fig 4). While an active role of actin in regulating spine morphology has previously been described (Matus, 2000; Rao and Craig, 2000), here we show a novel role of the ubiquitin-proteasome system in mediating morphological changes to the synapse, rather than at the molecular level (Colledge et al., 2003; Ehlers, 2003). Previous studies have suggested that harmful ischemia results in permanent spine retraction, suggesting that spine retraction may be a marker underlying excitotoxic damage to the cell (Park et al., 1996; Hasbani et al., 2001b; Hasbani et al., 2001a). Our data, are in agreement with other studies identifying spine retraction as a protective response of the cell to ischemia and excitotoxicity (Hasbani et al., 2001b; Hasbani et al., 2001a). Our view of spine retraction as a neuroprotective phenotype is further supported by other studies: hibernating arctic ground squirrels and hamsters have reduced numbers of dendritic spines associated with the ischemia tolerant state of torpor (Popov and Bocharova, 1992; Magarinos et al., 2006; von der Ohe et al., 2006). Further, a loss of dendritic spines has been reported in long-term depression (Zhou et al., 2004), which has many similar features to ischemic tolerance (brief stimulation resulting in a loss of response to a given stimuli). Hence dendritic remodeling may induce a protected phenotype to cells rendering them refractory to excitotoxic signaling.
A second morphologic feature associated with the tolerant state was the formation of varicosities along the dendritic arbor (Fig 5). Such varicosities, were originally considered cell death indicators resulting from excitotoxicity (Olney, 1971). More recent data show that varicosity formation and excitotoxic cell death occur through independent mechanisms; death is Ca+2 mediated while varicosities are mediated by Na+2 flux and AMPA internalization (Ikegaya et al., 2001). NMDA-induced varicosity formation was found to induce transient, reversible neuroprotection by attenuating excitatory neurotransmission (Ikegaya et al., 2001). Thus the varicosity formation as well as spine loss may play a combined role in the protective phenotype of rapid ischemic tolerance.
NMDA receptors are anchored to the actin cytoskeleton via a number of intermediate proteins (Husi et al., 2000). Following tolerance induction NMDA receptor anchoring to the cytoskeleton is reduced (Fig. 6B). The loss of NMDA receptor subunit interactions with the post-synaptic scaffold has been shown to block ischemia induced toxicity (Aarts et al., 2002) but exerts no short term effect on basal synaptic transmission or LTP formation in hippocampal cultures (Lim et al., 2003). Previous studies have shown that actin depolymerization reduces NMDA channel activity (Rosenmund and Westbrook, 1993). In agreement with these, we observe a reduction in NMDA-mediated currents and excitotoxicity following preconditioning (Fig 6C and 6E). Since tolerance to NMDA excitotoxicity was blocked by a proteasome inhibitor and an actin stabilizer (Fig 6E), this suggests the mechanism that results in protection against harmful ischemia also blocks NMDA excitotoxicity.
Recent studies suggest that NMDA receptor mediated signaling can be split into pathological and physiological mechanisms (Biegon et al., 2004). For example Hardingham et al reported that NR2B receptors in extra-synaptic locations mediate cell death, via abrogated CREB signaling (Hardingham and Bading, 2002; Hardingham et al., 2002). In addition, Liu et al reported that NR2B subunit containing NMDA receptors, whether synaptic or extra-synaptic, mediate harmful cell signaling where as NR2A subunit containing NMDA receptors appear protective (Liu et al., 2007). Our data and that of Sattler (Sattler et al., 2000) suggest that toxic ischemia induced NMDA signaling is inhibited following loss of dendritic spines, however the effect of dendritic spine collapse on NMDA receptor function is not yet fully clear and requires further investigation. However, it has been demonstrated that some NMDA signaling may be protective (Biegon et al., 2004) (Hardingham and Bading, 2002; Hardingham et al., 2002). To this end, we have observed that NMDA-induced CREB phosphorylation is maintained in rapid tolerance (data not shown). Therefore at the time point when cells display refractory tolerance to toxic stimuli, selective loss of toxic NMDA receptor signaling occurs, which is associated with the disassociation of the post synaptic intracellular receptor complex.
In summary, we have described the biology of rapid ischemic tolerance, which induces protein synthesis independent protection within one hour of the preconditioning stimuli. Furthermore we offer a mechanism for the protection, the proteasomal degradation of structural proteins following their ubiquitination that results in a morphological phenotype (dendritic spine loss), and a functional effect (reduced NMDA excitotoxicity). Hence our findings of an endogenous pathway that selectively reduces excitotoxic mechanisms of cell death may have potential for devising rapid acting strategies to protect against ischemia-induced brain injury.
This work was supported by NIH grants NS050669 & NS054023 (Meller), NS024728 (Simon) and the American Heart Association 0465430Z (Meller).