CaMKII phosphorylates collapsin response mediator protein 2 and modulates axonal damage during glutamate excitotoxicity

Authors

  • Sheng T. Hou,

    1. Experimental NeuroTherapeutics Laboratory, Institute for Biological Sciences, National Research Council Canada, Ottawa, Ontario, Canada
    2. Department of Biochemistry, Microbiology and Immunology, Faculty of Medicine, University of Ottawa, Ontario, Canada
    Search for more papers by this author
  • Susan X. Jiang,

    1. Experimental NeuroTherapeutics Laboratory, Institute for Biological Sciences, National Research Council Canada, Ottawa, Ontario, Canada
    Search for more papers by this author
  • Amy Aylsworth,

    1. Experimental NeuroTherapeutics Laboratory, Institute for Biological Sciences, National Research Council Canada, Ottawa, Ontario, Canada
    2. Department of Biochemistry, Microbiology and Immunology, Faculty of Medicine, University of Ottawa, Ontario, Canada
    Search for more papers by this author
  • Graeme Ferguson,

    1. Experimental NeuroTherapeutics Laboratory, Institute for Biological Sciences, National Research Council Canada, Ottawa, Ontario, Canada
    Search for more papers by this author
  • Jacqueline Slinn,

    1. Experimental NeuroTherapeutics Laboratory, Institute for Biological Sciences, National Research Council Canada, Ottawa, Ontario, Canada
    Search for more papers by this author
  • Houwen Hu,

    1. Experimental NeuroTherapeutics Laboratory, Institute for Biological Sciences, National Research Council Canada, Ottawa, Ontario, Canada
    Search for more papers by this author
  • Thomas Leung,

    1. Institute of Molecular and Cell Biology, National University of Singapore, Singapore
    Search for more papers by this author
  • Joachim Kappler,

    1. Institut für Biochemie und Molekularbiologie, Universität Bonn, Bonn, Germany
    Search for more papers by this author
  • Kozo Kaibuchi

    1. Department of Cell Pharmacology, Graduate School of Medicine, Nagoya University, Japan
    Search for more papers by this author

Address correspondence and reprint requests to Sheng T. Hou, Institute for Biological Sciences, National Research Council Canada, 1200 Montreal Road, Bldg M54, Ottawa, ON, Canada, K1A 0R6. E-mail: sheng.hou@nrc-cnrc.gc.ca

Abstract

Intracellular calcium influx through NMDA receptors triggers a cascade of deleterious signaling events which lead to neuronal death in neurological conditions such as stroke. However, it is not clear as to the molecular mechanism underlying early damage response from axons and dendrites which are important in maintaining a network essential for the survival of neurons. Here, we examined changes of axons treated with glutamate and showed the appearance of βIII-tubulin positive varicosities on axons before the appearance of neuronal death. Dizocilpine blocked the occurrence of varicosities on axons suggesting that these microstructures were mediated by NMDA receptor activities. Despite early increased expression of pCaMKII and pMAPK after just 10 min of glutamate treatment, only inhibitors to Ca2+/calmodulin-dependent protein kinase II (CaMKII) and calpain prevented the occurrence of axonal varicosities. In contrast, inhibitors to Rho kinase, mitogen-activated protein kinase and phosphoinositide 3-kinase were not effective, nor were they able to rescue neurons from death, suggesting CaMKII and calpain are important in axon survival. Activated CaMKII directly phosphorylates collapsin response mediator protein (CRMP) 2 which is independent of calpain-mediated cleavage of CRMP2. Over-expression of CRMP2, but not the phosphorylation-resistant mutant CRMP2-T555A, increased axonal resistance to glutamate toxicity with reduced numbers of varicosities. The levels of both pCRMP2 and pCaMKII were also increased robustly within early time points in ischemic brains and which correlated with the appearance of axonal varicosities in the ischemic neurons. Collectively, these studies demonstrated an important role for CaMKII in modulating the integrity of axons through CRMP2 during excitotoxicity-induced neuronal death.

Abbreviations used:
AIPII

autocamtide-2-related inhibitory peptide

AKT

protein kinase B

CaMKII

Ca2+/calmodulin-dependent protein kinase II

CRMP

collapsin response mediator protein

DAPI

4′,6-diamidino-2-phenylindole

EGFP

enhanced green fluorescent protein

GSK

glycogen synthase kinase

MAP2

microtubule-associated protein 2

MAPK

mitogen-activated protein kinase

MCAO

middle cerebral artery occlusion

MK801

dizocilpine

PI3K

phosphoinositide 3-kinase

PSD

post-synaptic density

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

In the context of stroke-induced brain damage, the molecular and biochemical mechanisms involving retraction and collapse of the axonal network remain unclear. One of the early morphological changes accompanying excitotoxicity-induced neuronal death in cultured neurons is the retraction/collapse of the neurite network, which indicates that axonal damage occurs before the emergence of typical morphological hallmarks of neuronal death (Deckwerth and Johnson 1994; Raff et al. 2002). Typically, axonal degeneration is manifested by irregular blebbing of axons with thinning and fragmentation, followed by retraction and collapse of the axonal network. While axonal damage was regarded as an outcome of the death process occurring within the cell body, more importantly, it may, in and of itself, be a trigger for death of the whole neuron (Raff et al. 2002). In studies of white matter damage, axonal injury in response to ischemia is associated with increased axonal membrane permeability with excess Na+ and/ or Ca2+ influx into axons (Stys and Jiang 2002; Stys 2004). This imbalanced Ca2+ influx activates deleterious cascades of locally localized intracellular protein kinases and proteases which subsequently triggers the breakdown of cytoskeletons and disturbance of axon transport leading to degeneration and neuronal death (Hara and Snyder 2007).

Ca2+/calmodulin-dependent protein kinase II (CaMKII) functions as a link between Ca2+ stimuli and neuronal death caused by NMDA receptor activities (Waxham et al. 1996; Tang et al. 2004). When calcium increases in the post-synaptic component, CaMKII is autophosphorylated and activated. Activated pCaMKII is translocated to the post-synaptic density (PSD) to target NR2B, a major component of the extrasynaptic NMDA receptors (Fink and Meyer 2002; Yoshimura et al. 2002). The importance of CaMKII in neural functions is underscored by the fact that mice lacking CaMKII show numerous deficiencies in learning and neuronal plasticity (Silva et al. 1992a,b). Loss of CaMKII activity also results in decreased damage to neurons in response to both focal and global ischemia in mice and inhibitors to calmodulin and CaMKII are also potently neuroprotective (Hajimohammadreza et al. 1995; Waxham et al. 1996; Takano et al. 2003). These studies strongly suggest an important role CaMKII plays in modulating neuronal viability. CaMKII is also known to influence neurite extension, arborization, and dynamics (Cammarota et al. 2002; Fink and Meyer 2002; Wen et al. 2004). Local calcium concentration apparently is critical in modulating these events. However, the role and mechanism of CaMKII in axon damage in response to excitotoxicity remain unclear.

Recent studies showed that collapsin response mediator protein (CRMP) 2 promotes axonal outgrowth through regulating microtubule assembly and Numb-mediated endocytosis (Inagaki et al. 2001; Fukata et al. 2002). CRMP2, which has also been independently identified as Ulip2/CRMP62/Turned On After Division, 64 kDa (TOAD-64)/Dihydropyrimidinase-related protein 2 (DRP-2), is one of at least five isoforms (Goshima et al. 1995; Arimura et al. 2004). CRMP2 is expressed in growth cones and distal parts of the growing axons to modulate axonal length by binding to tubulins thereby promoting extension of microtubules and axons (Inagaki et al. 2001; Fukata et al. 2002). Over-expression of CRMP2 causes the development of multiple axons (Inagaki et al. 2001). Although how CRMP2 precisely modulates cytoskeleton to induce axonal changes remains an active research area, it is known that CRMP2 acts as a common intracellular target by integrating both positive and negative effects on axon extension possibly through distinct upstream activators (Arimura et al. 2000, 2005; Inagaki et al. 2001; Fukata et al. 2002; Yoshimura et al. 2006). Identification of modulators of CRMP2 is important in understanding axonal outgrowth and stress response. Recent mass spectrometry studies have identified more than 30 proteins at PSD to be substrates of CaMKII (Yoshimura et al. 2002) and interestingly CRMP2 is one of them. Previous studies, including those of our own, showed that CRMP2 was present in relatively high concentrations in the PSD/synaptosome fraction (Jiang et al. 2007) and that ischemia increased the expression of CRMP2 (Chung et al. 2005). However, it remains unknown whether CaMK II targets CRMP2 to maintain axonal integrity during excitotoxicity.

In the present study, we examined axonal changes following glutamate toxicity. Several calcium responsive signaling kinases were examined. Blocking the early induction of CaMKII was effective in protecting axons and neurons. Evidence was also obtained showing that both calpain and CaMKII target CRMP2 through cleavage and phosphorylation, respectively. The fact that pCaMKII and pCRMP2 levels were also increased in ischemic brains strongly suggests that CaMKII-CRMP2 signaling cascade may be a target for therapeutic modulation in stroke.

Materials and methods

Neuronal cultures and treatment

Primary cortical neurons were prepared from embryonic E15–16 CD1 mice and cultured in neurobasal media for 5–14 days as previously described (O’Hare et al. 2000) and in Appendix S1.

Cerebral ischemia produced by middle cerebral artery occlusion

All procedures using animals were approved by the IBS Animal Care Committee following the guidelines established by the Canadian Council on Animal Care. C57B/6 mice (20–23 g) were obtained from Charles River Laboratories Inc., Senneville, QC, Canada and bred locally. Under temporary isofluorane anesthesia, mice were subjected to middle cerebral artery occlusion (MCAO) using an intraluminal filament as previously described (Jiang et al. 2005, 2007). After 1 h of MCAO, the filament was withdrawn, blood flow restored to normal and wounds sutured. Sham-operated mice were subjected to the same surgery without MCAO, which were used as controls. Brain tissue was collected after 24 h reperfusion. The brain tissue sectioning and protein extraction were then performed following the standard procedure.

Subcellular protein fractionation

Synaptosomes were extracted using the protocol as described (Jiang et al. 2007). Details are presented in Appendix S1.

Plasmid preparation and transfection

To over-express CRMP2, CRMP2 cDNA was sub-cloned into an expression vector pBK-cytomegalovirus immediate early gene enhancer/promoter (CMV) (Stratagene, La Jolla, CA, USA) as previously reported (Franken et al. 2003). A pXJ40-FLAG-tagged CRMP2 mutant CRMP2-T555A, which can not be phosphorylated at T555 site, was made as previously described (Leung et al. 2002; Brown et al. 2004). Plasmid DNAs were purified using a QIAGEN plasmid midi-purification kit (Valencia, CA, USA). In total, 10 μg DNA was used per transfection assay with the Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA, USA). Briefly, cortical neurons plated in a 24-well plate were transfected at 7 days in vitro (DIV) as previously described (Huang et al. 2005). The transfection efficiency is about 10% of the total number of cells in the dish. After 2 days of transfection, neurons were treated with glutamate (100 μM) for 2–4 h before fixation with freshly prepared formalin. Fixed neurons were subjected to double immunostaining to detect co-localization of βIII-tubulin and CRMP2.

In vitro CaMKII phosphorylation assay

Pure CaMKII was purchased from New England Biolabs (P6060S) (Pickering, Ontario, Canada) and the assay was performed according to the manufacturer’s protocol. To activate the recombinant CaMKII, CaMKII was pre-incubated in the kinase buffer containing 50 mM Tris, pH 7.5, 10 mM MgCl2, 2 mM DTT, 0.1 mM EDTA, 2 mM CaCl2, 200 mM ATP, 1.2 mM calmodulin. The incubation was for 10 min at 30°C. Immunopurified CRMP2 was used as a substrate for CaMKII assay. Purified CRMP2 protein was incubated with the kinase buffer in the presence or absence of active CaMKII (250 U) for 10 min at 30°C. The reaction was stopped by adding sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) loading buffer. Samples were boiled for 5 min and loaded on an 8% sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Western blotting was performed following the standard protocol and probed with a specific antibody to pCRMP2 (Thr555).

Results

Axonal changes in response to glutamate toxicity

Our previous studies have shown that incubation of neurons with high dose of glutamate (100 μM) for more than 4 h will cause extensive neuronal death (Hou et al. 2000; Smith et al. 2003). However, numerous structural and biochemical changes occur in neurons within a few minutes of treatment. Especially, axons and dendrites begin to show visible signs of stress after 2 h incubation with 100 μM of glutamate under phase contrast microscope. As shown in Fig. 1, immunostaining for βIII-tubulin revealed the appearance of positive protrusions from axons (varicosities). Axons also appeared to become thinner compared with those untreated controls (Fig. 1 panels a′–c′) and became fragmented after 4 h treatment with glutamate. The numbers of varicosities per 100 μm of axons were counted and plotted in Fig. 1(d) and the thickness of axons was also measured and shown in Fig. 1(e). Blocking NMDA receptors using NMDA receptor antagonist, dizocilpine (MK801), completely prevented the appearance of varicosities on axons treated by glutamate (100 μM for 2 h) and the thickness of axons was also maintained in comparison with axons treated with glutamate alone (Fig. 1c, c′, d and e), suggesting that the appearance of varicosities is in response to NMDA receptors.

Figure 1.

 NMDA receptor-dependent damage to axons. The appearance of varicosities and the reduced thickness of axons of cortical neurons treated with glutamate were ameliorated by NMDA receptor antagonist MK801. Mouse cortical neurons cultured in neural basal media for 8 DIV were treated with 100 μM glutamate for 2 h in the presence or absence of prior incubation with MK801 (10 μM). Immunostaining using a primary antibody to βIII-tubulin was performed (red color). High resolution images were taken using an oil immersion objective (×63). The length and thickness of axons were measured using Image J and the numbers of axonal varicosities were counted and plotted in panel (d) and (e) (= 5, with more than 100 cells measured per experiment). Enlarged images in a’–c’ show axons corresponding to the lower power images derived from (a) to (c), respectively. Arrows indicate the appearance of varicosities. Error bars are SE, ** indicates statistical significant with < 0.01 compared with the glutamate treated neurons by Student’s t-test. Scale bar = 100 μm.

Early induction of pCaMKII in response to glutamate toxicity

Activation of NMDA receptors causes calcium influx which triggers a cascade of signaling events to cause damage to axons and neurons. To determine temporal changes of calcium-activated intracellular kinases and proteases in response to glutamate toxicity through NMDA receptors, western blotting was performed on proteins collected from cortical neurons treated with 100 μM glutamate for 0, 10, 30, 60, 120 and 240 min (Fig. 2a). As shown in Fig. 2(a), the level of pMAPK and pCaMKII, were increased after 10 min of exposure to glutamate treatment indicating activation of CaMK and mitogen-activated protein kinase (MAPK) pathways. Interestingly, pCRMP2 level was also increased after 10 min of treatment. The level of pCRMP2 peaked at 60 min which was later compared with those of pAKT, pCaMKII and pMAPK. In contrast, pGSK3β appeared decreased after 10 min of glutamate treatment while the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) did not change its expression level (Fig. 2a).

Figure 2.

 CaMKII inhibitors protected axons from glutamate toxicity. Cultured cortical neurons were treated with 100 μM glutamate for 10, 30, 60, 120 or 240 min. Cells were collected at the time indicated and cell lysates were subjected to SDS–PAGE (10%) followed by western blotting to detect changes in the expression of pMAPK, total MAPK, pCaMKII, CaMKII, pCRMP2, dephospho-CRMP2, pGSK3β, GSK3β and GAPDH (Panel a). Inhibitors were added to cultured cortical neurons 15 min before the addition of glutamate to the cultures. The inhibitors used and their concentrations were as indicated in each figure. After 2 h treatment with glutamate, cells were fixed with freshly prepared formalin and followed by immunostaining for βIII-tubulin. Arrows indicate the appearance of varicosities. (b′)–(g′) are enlarged views of corresponding panel (b)–(g), respectively. The length and thickness of axons were measured using Image J and the numbers of axonal varicosities were counted and plotted in panel H and I (= 13). ** indicates statistical significant with < 0.01 compared with the glutamate treated neurons by anova with Tukey’s post hoc analysis to identify significant groups. Error bars are SE. Scale bar = 100 μm.

CaMKII inhibitors protected axonal damage caused by glutamate treatment

To determine which of these activated pathways may be more important in modulating axonal damage in response to glutamate, selective inhibitors were used targeting the following pathways shown to be activated in response to NMDA receptor activation, phosphoinositide 3-kinase (PI3K), CaMKII, MAPK and Rho kinase. As shown in Fig. 1(c), blocking NMDAR using the NMDAR antagonist, MK801, completely prevented the appearance of βIII-tubulin positive varicosities on neurites (Fig. 1c, d and e) and protected neurons [previously published data (Smith et al. 2003)]. Interestingly, inhibitors to CaMKII [KN62, KN93 and autocamtide-2-related inhibitory peptide (AIPII)], and calpain (calpeptin), also completely prevented the appearance of βIII-tubulin positive varicosities (Fig. 2b–d, b′–d′). CaMK inhibitors have no known adverse effect on neuronal survival. KN62 can inhibit several CaM kinases, while AIPII is a specific inhibitor for CaMKII.

In contrast, inhibitors to MAPK (PD98059), Rho kinase (Y27632) and PI3K (LY294002) were not effective in eliminating the appearance of βIII-tubulin varicosities (Fig. 2e–g, e′–g′). The numbers of varicosities and the thickness of axons were measured and quantified as shown in Fig. 2(h) and (i). Inhibitors to CaMKII and calpain were effective in significantly reducing the appearance of varicosities and the thickness of axons. Together, these studies showed that calpain and CaMKII activations were critical for inducing axonal changes in response to glutamate toxicity.

CaMKII directly targets CRMP2 through phosphorylation

To understand how CaMKII and calpain affect axonal changes, we examined changes of CRMP2 which we have previously shown to be a target of calpain in response to glutamate toxicity and cerebral ischemia (Jiang et al. 2007; Hou et al. 2006). Calpain activation in response to glutamate toxicity affects all CRMPs including CRMP1, 2, 3, 4, and 5 by creating internal proteolytic cleavage generating a smaller fragment with a molecular weight of 54 kDa (Jiang et al. 2007; Hou et al. 2006). As CRMP2 is important in mediating axonal outgrowth through modulating tubulin polymerization (Inagaki et al. 2001; Fukata et al. 2002; Arimura et al. 2005), we hypothesized that CaMKII may target CRMP2 through phosphorylation to affect axon morphology during glutamate treatment. To test this hypothesis, we first determined whether CaMKII targets CRMP2 during glutamate toxicity. Temporal changes in pCRMP2 and internal cleavage by calpain were determined using western blotting. As shown in Fig. 2(a), CRMP2 phosphorylation increased almost immediately following the activation of CaMKII (10–30 min after glutamate treatment), while the cleavage of CRMP2 occurred at 30 min and peaked at 60 min after glutamate treatment, suggesting that phosphorylation of CRMP2 occurred early and prior to CRMP2 cleavage. Second, and importantly, CaMKII inhibitors were effective in suppressing phosphorylation of CRMP2 in response to glutamate treatment (Fig. 3a and b). In contrast, inhibitors to MAPK, PI3K and calpain had no effect (Fig. 3a and b). Finally, to determine if CaMKII directly targets CRMP2, in vitro CaMKII kinase assay was performed using His-tagged (×6) CRMP2 as a substrate. His-CRMP2 was over-expressed in HEK293 cells and purified using a Ni chromatography column. As shown in Fig. 3(c), purified CRMP2 was negative to pCRMP2 probed by the antibody to pCRMP2(T555) on a western blot. Addition of reaction buffer to the CaMK assay did not produce pCRMP2 (Fig. 3c). In contrast, addition of active CaMKII to the reaction mixture clearly increased the level of pCRMP2 confirming that CaMKII directly targets CRMP2 through phosphorylation. Total dephospho-CRMP2 was detected to show equal protein input loading (Fig. 3c).

Figure 3.

 CaMKII targets CRMP2 through phosphorylation. Cortical neurons treated with glutamate (100 μM) for 2 h in the presence of several inhibitors (as indicated in panel a) to known calcium-activated intracellular pathways were subjected to western blotting to detect changes in pCRMP2 level. GAPDH level was used as a control for equal protein loading. The band intensities were measured using Image J and the ratio to that of GAPDH was calculated as shown in panel (b) (= 3). To demonstrate that CaMKII directly targets CRMP2 through phosphorylation, immunopurified CRMP2 was mixed with the activated CaMKII (250 U) using a method as described in the Methods section. Western blotting was performed to detect the level of pCRMP2 (Thr555) as shown in panel (c). Western blotting using the antibody against dephospho-CRMP2 was also performed to show equal input of proteins (panel c, lower panel). Error bars represent SE. ** indicates statistical significant (< 0.01) using Student’s t-test by comparing with the glutamate treatment only sample.

CaMKII phosphorylation of CRMP2 prevents calpain from cleaving CRMP2

Works from our laboratory and others have shown that CRMP2 protein is a target of activated calpain and CRMP2 is cleaved towards the C-terminal end (Jiang et al. 2007; Bretin et al. 2006; Zhang et al. 2007; Rogemond et al. 2008). To determine whether phosphorylation of CRMP2 affects calpain cleavage of CRMP2, normal brain extractions were treated with 5 mM calcium, which we have previously shown to activate calpain and induce CRMP2 cleavage (Jiang et al. 2007; Hou et al. 2008). Here, as shown in Fig. 4(a), calcium treatment evoked a robust induction of pCaMKII and pCRMP2. However, adding various concentrations of inhibitors to CaMKII (AIPII), MAPK (PD98059) and PI3K (LY294002) did not prevent CRMP2 from cleavage (Fig. 4b). Interestingly, inhibition of CaMKII using AIPII increased the amount of calpain-cleaved CRMP2 (54 kDa band in Fig. 4b) as shown by densitometry quantification (Fig. 4c). In contrast, inhibitors to MAPK and PI3K produced a similar amount of cleaved CRMP2 (54 kDa band) relative to the calcium (5 mM) only treated brain sample. Only calpain inhibitors were effective in completely attenuating CRMP2 cleavage (Fig. 4d). The level of pCRMP2 was reduced by about 50% (= 2) in brain samples treated with inhibitors to CaMKII as shown in Fig. 4(e) and (f) in comparison with samples treated with inhibitors to MAPK, PI3K and Rho kinase which showed a similar level of pCRMP2 compared to calcium (5 mM) treated only brain sample. Collectively, these studies suggest that CRMP2 is independently targeted by calpain and CaMKII in response to calcium activation, but phosphorylation of CRMP2 at T555 by CaMKII increases its resistance to calpain-induced breakdown.

Figure 4.

 CaMKII inhibitor increased calpain-mediated cleavage of CRMP2. Normal mouse brain extracts were treated with 5 mM CaCl2 to elicit the activation of calcium-dependent kinases and proteases including CaMKII and calpain, respectively. Total proteins were separated on a 10% SDS–PAGE and followed by western blotting using primary antibody to pCRMP2, pCaMKII and GAPDH (panel a). In panel (b), normal brain lysate (NB) were pre-treated with a range of doses of inhibitors, as indicated, for 30 min before adding CaCl2 (5 mM). After 24 h incubation, cell lysates were separated on 10% SDS–PAGE and followed by western blotting to detect dephosphorylated CRMP2 (b and d). The intact CRMP2 has a molecular weight of 63 kDa, while the cleaved CRMP2 is at 54 kDa (b and d). The intensities of the p63 kDa and p54 kDa bands were measured using NIH Image J software and the ratio of p54 kDa band versus p63 kDa band was calculated. The fold changes of p54/p63 kDa band ratio against CaCl2 treated sample were plotted in panel (c) (= 3). Corresponding level of pCRMP2 was also detected using western blotting as shown in panel (e). The band intensities were also measured and the fold change was normalized against normal brain (NB) as shown in panel (f) (= 2). Error bars are SE. ** indicates statistical significant comparing with the CaCl2 only treated brain sample by Student’s t-test with < 0.01.

Over-expression of CRMP2 reduces varicosities on axons

In order to determine if CRMP2 is important in maintaining axonal integrity during glutamate toxicity, the following two experiments were performed: first, co-localization of βIII-tubulin with CRMP2 was performed on cortical neurons treated with glutamate and/or CaMKII inhibitor, AIPII. As shown in Fig. 5(a–c), CRMP2 was distributed throughout axons and co-localized with βIII-tubulin in untreated mature neurons (Fig. 5a). In contrast, CRMP2 expression was absent in the varicosities of damaged axons (Fig. 5b and inset). In cortical neurons treated with CaMKII inhibitor AIPII, CRMP2 expression was maintained in axons and showed co-localization with βIII-tubulin (Fig. 5c and inset). Together, these studies suggest that CRMP2 is important in maintaining the integrity of axons in response to glutamate toxicity.

Figure 5.

 Over-expression of CRMP2 increased axonal resistance to produce varicosities in response to glutamate. Panels (a–c) Double immunostaining for βIII-tubulin (green color) with CRMP2 (red color) of cortical neurons showed that CRMP2 was abundantly expressed in axons and co-localized with βIII-tubulin (yellow color derived from the merged image in panel a, arrows). CRMP2 expression was drastically reduced in glutamate treated axons (panel b), and appeared absent in varicosities (arrows in b and inset). In contrast, the distribution of CRMP2 did not change in axons pre-treated with CaMKII inhibitor, AIPII (panel c and inset; arrows indicate axons expressing both βIII-tubulin and CRMP2). Panels (d–m) Cortical neurons were transfected with plasmids encoding EGFP, CRMP2 or CRMP2-T555A. After 1 day of transfection, neurons were either collected for western blotting to detect increased expression of CRMP2 (d), or were treated with 100 μM glutamate for 2–4 h. Cells were fixed with formalin and double-immunostained for βIII-tubulin (red) with CRMP2 (green) in panels (i–k) (anti-his-tag staining). For panel (l–m), anti-FLAG staining (green) was performed to show the expression of CRMP2-T555A. Panels (e–h) showed neurons only transfected with EGFP and the co-localization of EGFP with βIII-tubulin positive varicosities (panel h). The number of varicosities per 100 μm of axons were counted and plotted in panel (n) (= 10). The average thickness of axons was also measured and plotted in panel (o) (= 10). Scale bars = 100 μm; Error bars are SEM. ** indicate statistical significant with anova and post hoc analysis using Tukey’s test when compared with EGFP and CRNP2-T555A transfected neurons (< 0.01).

Second, CRMP2 and its mutant (CRMP2-T555A) were over-expressed in cortical neurons using plasmid transfection (as shown by western blotting in Fig. 5d), followed by treatment with glutamate (100 μM) for 2–4 h (Fig. 5e–o). The CRMP2(T555A) mutant and enhanced green fluorescent protein (EGFP) plasmid-transfected cortical neurons were used as controls. As shown in Fig. 5(e–h) and (l–m), glutamate treatment caused the appearance of varicosities on neurites of EGFP and CRMP2-T555A expressing neurons, repectively. These varicosities were positive to both EGFP and βIII-tubulin (Fig. 5h and m). In contrast, CRMP2 over-expression co-localized with βIII-tubulin staining on axons (Fig. 5i–k). The numbers of varicosities were significantly reduced following glutamate treatment (Fig. 5n; < 0.01, = 4). The thickness of the βIII-tubulin positive axons was also significantly higher in CRMP2 over-expressing neurons in comparison with the EGFP and CRMP2-T555A expressing axons treated with glutamate (Fig. 5o; < 0.01, = 4), suggesting that CRMP2 is important in maintaining axonal integrity in response to glutamate toxicity.

Early induction in pCaMKII expression and axonal damage in stroke mouse brain

Cerebral ischemia-induced neuronal damage is largely caused by glutamate-mediated excitotoxicity. Focal cerebral ischemia (MCAO) indeed evoked a marked increase in pCaMKII on the ipsilateral side of the ischemic brain after just 1 h reperfusion in comparison with that from the contralateral side of the brain (Fig. 6a; 500% increase; = 3). While, at this time point, the level of total CaMKII did not change. The level of CaMKII decreased and remained very low in the ischemic side of the brain after 2 h reperfusion in comparison with both the sham-operated brain and the contralateral side of the brain. The level of pCRMP2 expression was also increased (150%; = 3), but after 2 h of reperfusion which was later than that of pCaMKII, suggesting a potential sequential correlation. Most importantly, CRMP2 cleavage only begins to appear in the ischemic side of the brain after 4 h reperfusion (panel a). The induction of pCaMKII was by far the most robust and the earliest in the ischemic brain in comparison with those of pMAPK and pAKT. The level of expression pGSK and total glycogen synthase kinase (GSK) did not show clear changes between the ipsilateral and contralateral side of the brain (Fig. 6a). Synaptosomes were isolated from ischemic brain using a sucrose gradient. Western blotting showed that synaptosomes were enriched with CaMKII and that the level of pCaMKII and pCRMP2 increased significantly in the ischemic side of the brain after 1 h reperfusion in comparison with that in the contralateral side of the brain (Fig. 6b and c) confirming that CaMKII may target CRMP2 to modulate synaptic function.

Figure 6.

 Early and robust induction of pCaMKII in ischemic mouse brain and synaptosomes. Mice were subjected to 1 h MCAO and reperfusion as described in the Methods section. At the specified reperfusion time as shown in panel (a), mice were killed and their brains collected. Brains were separated into the contralateral side (right hemisphere, R) and ischemic side (left hemisphere, L). Total protein lysate from both sides of the brain (at 10 μg) were separated on a 10% SDS–PAGE and followed by western blotting using specific antibodies to pAKT, pCaMKII, CaMKII, pCRMP2, dephospho-CRMP2, pMAPK, pGSK3b, total GSK and GAPDH. GAPDH was used to indicate equal protein loading (panel a). Synaptosomes were isolated from the contralateral side of the brain (R = right) and ischemic side of the brain (L = left) using a sucrose gradient as described in the Methods section (= 3). Synaptosome proteins at 10 μg were separated on a SDS–PAGE and followed by western blotting to detect the expression of pCaMKII, CaMKII, pCRMP2 and GAPDH. The protein band intensities were measured using Image J and the ratio to that of GAPDH was calculated as shown in panel (c) (= 3). ** in panel (c) indicates statistical significance with < 0.01 by Student’s t-test. Error bars represent SE. Panels (d–g) Serial sections of ischemic mouse brains were also cut and subjected to triple staining to detect MAP2 (empty arrows; red color), cell death (TUNEL staining, green color, arrowheads) and nuclei (DAPI, blue color). The less organized microtubules and varicosities (solid arrows) are visible in ischemic side of the brain. Panel (h) shows a sketch of brain section to indicate the locations of images (d–g). Scale bar = 100 μm.

To determine if cerebral ischemia also induces the appearance of varicosities on axons, serial brain sections from ischemic mouse were cut and were subjected to triple staining with an antibody to microtubule-associated protein 2 (MAP2) (red color in Fig. 6d–g), terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining (green color in Fig. 6d–g) and 4′,6-diamidino-2-phenylindole (DAPI) (blue color in Fig. 6d–g). The locations of the images taken on a coronal brain section are shown in panel (h) (Fig. 6). MAP2 positive microtubules on the contralateral side of the ischemic brain appeared as bundles of smooth straight fibers oriented perpendicular to the pial (empty arrows in Fig. 6e). In contrast, MAP2 staining of the ischemic side of the brain showed a less organized axonal network than those from the ischemic side of the brain. Large numbers of varicosities appeared on MAP2 positive axons (Fig. 6f and g) confirming in vitro glutamate toxicity studies shown in Figs 1 and 2. Most importantly, ischemic neurons with extensive varicosities were not positive to TUNEL staining (solid arrows in panel f and g), while neurites appeared to be absent in TUNEL positive neurons (arrowheads), strongly suggesting that changes in microtubule networks occurred prior to neuronal death following cerebral ischemia.

Discussion

In the present study, we demonstrate that axons produce numerous varicosities following glutamate treatment. This process was NMDA receptor activity dependent, as MK801 effectively prevented the production of varicosities on axons and protected neurons. The appearance of varicosities occurred prior to the onset of expression of markers for neuronal death, such as changes in membrane permeability (propidium iodite positive staining of the nucleus and TUNEL positivity) as previously shown (Hou et al. 2000), suggesting that axonal changes occurred prior to the death of the soma. More importantly, robust activation of CaMKII occurred early (10 min) after glutamate treatment. Blocking CaMKII using several selective inhibitors was effective in preventing the appearance of varicosities on axons and inhibited phosphorylation of CRMP2. Calpain mediated cleavage of CRMP2 occurred later than CaMKII-mediated phosphorylation of CRMP2 during glutamate toxicity in vitro and cerebral ischemia in vivo. Evidence that inhibition of CaMKII-induced phosphorylation of CRMP2 increased calpain-mediated cleavage of CRMP2 suggests that CaMKII and calpain both target dephospho-CRMP2 and that CaMKII-mediated phosphorylation of CRMP2 (T555) increased CRMP2 resistance to calpain cleavage. Together, these studies strongly argue for a role of CaMKII in mediating early changes of axonal response to glutamate toxicity.

Previous studies showed that CaMKII was highly enriched in the synaptic terminus. It was believed that CaMKII promotes axonal outgrowth as well as axonal branching, but CaMKII is less important for axonal maintenance (Cammarota et al. 2002; Nishiyama et al. 2003; Wen et al. 2004; Yamauchi 2007). CaMKII functions to detect calcium spike frequencies, such as the ones caused by stimulation with netrin-1, which evokes high levels of Ca2+ transients on restricted regions of an axon coincident with rapid development of branches (Tang and Kalil 2005). It is not yet known whether CaMK also plays a role in axon maintenance in adult mature neurons and especially during excitotoxicity injury. The present study for the first time demonstrated that CaMKII may play an important role in maintaining the integrity of axons through targeting CRMP2. CaMKII was robustly activated during glutamate toxicity. Inhibitors to CaMKII were able to ameliorate the production of varicosities on axons and protected neurons.

Mechanistically, CaMKII-induced varicosities were produced by targeting CRMP2. Phosphorylation of CRMP2 plays an important role in maintaining the integrity of neurites. For example, during development, CRMP2 is phosphorylated in vitro and in vivo by Rho kinase (Arimura et al. 2000, 2005; Fukata et al. 2002; Yoshimura et al. 2006), GSK3β (Yoshimura et al. 2006) and cyclin-dependent kinase 5 (Cole et al. 2006, 2008) which modulate CRMP2 binding to tubulins, thereby affecting axonal outgrowth and branching. However, different kinases prime CRMP2 for differential regulation of CRMP2 (Cole et al. 2006). In addition, Rho kinase dependent and independent pathways also exist for growth cone collapse through CRMP2 phosphorylation (Arimura et al. 2000). The antibody to pCRMP2 we used detects the phosphorylation state of the extremely c-terminal amino acid Ile550-Thr555 (p)-Ala560 (Arimura et al. 2000, 2005). Evidence suggests that CaMKII phosphorylation of CRMP2 at Thr555 may represent a Rho-independent pathway in response to glutamate toxicity. Rho activation is known to mediate excitotoxic neuronal death through the p38α pathway (Semenova et al. 2007). It appears, however, that Rho kinase pathway may not be important in early axonal changes seen in the present study as Rho inhibitor failed to prevent the appearance of varicosities. In fact, the roles of Rho downstream effectors (more than 20 of them) in excitotoxicity are not yet clear (Semenova et al. 2007). Importantly, that blocking Rho kinase using Y27632 also failed to show any effect on glutamate-induced neuronal death even at 100 μM concentration (Semenova et al. 2007) lends further support to the view that Rho kinases are dispensable for glutamate-induced axonal changes and neuronal death.

The idea that phosphorylation of CRPM2 by CaMKII may serve as a signal for photolytic cleavage was investigated. Temporal changes of pCRMP2 were examined which showed that CRMP2 phosphorylation occurred prior to the onset of large scale cleavage of CRMP2 (Fig. 2a). In vitro experiments using calcium stimulation of CaMKII and calpain showed that inhibition of CaMKII using several specific inhibitors did not prevent CRMP2 cleavage (Fig. 4). Only inhibitors specific to calpain were able to completely inhibit CRMP2 cleavage. These studies demonstrated that CaMKII and calpain both target CRMP2 during glutamate toxicity to axons.

Both CaMKII-mediated phosphorylation of CRMP2 and calpain-induced cleavage of CRMP2 can weaken CRMP2 function in neurite outgrowth. However, the two events may either be linked, or occur independently. Quantitative analysis of CRMP2 breakdown by calpain (Fig. 4) showed a positive correlation between the reduced phosphorylation at Thr555 and the increased calpain cleavage of CRMP2. These data suggest that CaMKII phosphorylation at Thr555 prevented pCRMP2 from degradation and contributed to neuronal survival. However, reduction of pCRMP2 at T555 through inhibiting CaMKII using CaMKII inhibitor was also protective to neurites. This apparently contradicts with the finding that over-expressing CRMP2 increased axonal resistance to produce varicosities in response to glutamate toxicity. Possible explanations to this difference are that (i) CaMKII has additional targets for neurite protection; (ii) CaMKII and calpain both independently target CRMP2 and that these processes are spatially and temporally regulated, and (iii) pCRMP2 may have unknown detrimental effects to neurites. Indeed, as shown in Fig. 2(a), pCRMP2 shows the strongest band at 240 min after glutamate treatment, and yet, the level of uncleaved CRMP2 is much reduced. At this time point, about 40% of neurons were in fact dead. It remains to be determined whether the high level pCRMP2 may contribute to the death of neurons in response to glutamate treatment. Despite these, the present study provided evidence to argue that CRMP2 is important in maintaining axonal integrity and which acts as a common target of both CaMKII and calpain. Although it is still unclear as to the consequence of CRMP2 cleavage by calpain, our previous studies on CRMP3 has shown that the smaller fragment of CRMP3 undergoes nuclear translocation to modulate death of the soma of neurons (Hou et al. 2006). Indeed, recent studies showed that processed CRMP2 underwent nuclear translocation during development to inhibit neurite outgrowth (Rogemond et al. 2008).

Previous studies demonstrated that CRMP2 was proteolytically cleaved during ischemic neuronal injury (Jiang et al. 2007; Bretin et al. 2006; Zhang et al. 2007). The present study provided additional evidence to corroborate this finding and further showed that both CRMP2 cleavage and phosphorylation at T555 increased in the ischemic side of the brain (Fig. 6). It is particularly interesting to see the increased activation of CaMKII in the synaptosomal fraction from the ischemic side of the brain just 1 h after reperfusion (Fig. 6b). Based on our previous finding that CRMP2 is also enriched in the synaptosomal fractions of the ischemic brains (Jiang et al. 2007), it is highly possible that activated CaMKII may target CRMP2 to modulate synaptic structure and function, which remains an exciting area for further exploration. The demonstration of the appearance of varicosities on axons in ischemic brains is one of the most interesting findings of the present report. The fact that neurons with axonal varicosities are not positive to TUNEL staining is highly suggestive that changes in microtubules in axons occurred early and before genomic degradation in the nucleus.

In summary, we have demonstrated that NMDA receptor mediated excitotoxicity activates CaMKII and calpain. Through convergence on targeting of CRMP2, CaMKII and calpain play very important roles in early axonal response to glutamate toxicity. Modulating CaMKII and calpain may therefore represent a novel venue for designing effective therapeutics to stroke-induced axonal damage.

Acknowledgements

We thank IBS animal facility for timely supply of animals. The project was supported by grants from the Heart and Stroke Foundation of Canada (NA5393; T5760) and Canadian Institutes of Health Research (CCI85680) to STH.

Ancillary