Neurodegeneration in an Aβ-induced model of Alzheimer’s disease: the role of Cdk5

Summary

Cdk5 dysregulation is a major event in the neurodegenerative process of Alzheimer’s disease (AD). In vitro studies using differentiated neurons exposed to Aβ exhibit Cdk5-mediated tau hyperphosphorylation, cell cycle re-entry and neuronal loss. In this study we aimed to determine the role of Cdk5 in neuronal injury occurring in an AD mouse model obtained through the intracerebroventricular (icv) injection of the Aβ_1–40 synthetic peptide. In mice icv-injected with Aβ, Cdk5 activator p35 is cleaved by calpains, leading to p25 formation and Cdk5 overactivation. Subsequently, there was an increase in tau hyperphosphorylation, as well as decreased levels of synaptic markers. Cell cycle reactivation and a significant neuronal loss were also observed. These neurotoxic events in Aβ-injected mice were prevented by blocking calpain activation with MDL28170, which was administered intraperitoneally (ip). As MDL prevents p35 cleavage and subsequent Cdk5 overactivation, it is likely that this kinase is involved in tau hyperphosphorylation, cell cycle re-entry, synaptic loss and neuronal death triggered by Aβ. Altogether, these data demonstrate that Cdk5 plays a pivotal role in tau phosphorylation, cell cycle induction, synaptotoxicity, and apoptotic death in postmitotic neurons exposed to Aβ peptides in vivo, acting as a link between diverse neurotoxic pathways of AD.

Introduction

Alzheimer’s disease (AD) is the most common form of dementia (Rodgers, 2003). The key agents of AD pathogenesis are the amyloid-beta peptides (Aβ) (Hardy & Selkoe, 2002; Nathalie & Jean-Noel, 2008), which tend to aggregate in fibrils, leading to the formation of amyloid plaques (Masters et al., 1985).

One of the proteins show to be involved in AD pathology is cyclin-dependent kinase 5 (Cdk5) (Dhavan & Tsai, 2001; Tsai et al., 2004). In fact, we previously demonstrated that Aβ peptides cause a dysregulation of this proline-directed serine-threonine kinase via the activation of calpains, a group of cytosolic proteases (Lopes et al., 2007). Calpain activation, triggered by the increase in intracellular calcium levels induced by Aβ exposure (Ferreiro et al., 2006), will lead to the cleavage of Cdk5 regulatory subunit p35 to p25. This pathogenic subunit associates with the kinase, forming a hyperactive and mislocalized p25/Cdk5 complex (Patrick et al., 1999). In AD, Cdk5 overactivation is associated with tau hyperphosphorylation, which culminates with the formation of neurofibrillary tangles (Hardy, 2003; Lee & Tsai, 2003) and can also be correlated to phenomena such as synaptic loss or neuronal death (Arriagada et al., 1992; Hamdane et al., 2003; Eckermann et al., 2007).

Interestingly, in several brain pathologies where Cdk5 dysregulation was described (Nguyen & Julien, 2003; Tsai et al., 2004; Qu et al., 2007; Wen et al., 2007), there was also the re-expression of different cell cycle markers (Nguyen et al., 2003; Kuan et al., 2004; Hernandez-Ortega et al., 2007; Hoglinger et al., 2007). Indeed, cell cycle reactivation is considered an important neuropathological feature of AD (Nagy et al., 1997; Yang et al., 2006; Hernandez-Ortega et al., 2007; Ahn et al., 2008) and is known to occur in the brain of patients, as well as in AD transgenic mouse models and in neuronal cell cultures exposed to Aβ peptide (Busser et al., 1998; Wu et al., 2000; Yang et al., 2006; McShea et al., 2007; Lopes et al., 2009a). In affected postmitotic neurons, there is a passage from the steady G0 state to an active cycling situation (Arendt, 2000). Among the re-expressed cell cycle proteins, Cdks 2, 4, and 6 are of vital importance, being intimately related with the G1/S checkpoint transition (Copani et al., 1999; Nguyen et al., 2003; Kuan et al., 2004). Activation of these Cdks by association with cyclin regulatory subunits, will lead to the phosphorylation of the retinoblastoma protein (Rb) (Weinberg, 1995; Tannoch et al., 2000), causing it to dissociate from a transcription-repressor complex and thus leading to the transcription of multiple cell cycle proteins that promote the progression past the G1/S checkpoint (Copani et al., 1999; Park et al., 2000; Nguyen et al., 2003; Kuan et al., 2004). Although re-cycling neurons advance to the S phase and DNA replication occurs, they do not pass the G2/M checkpoint and degenerate somewhere between the S and the G2 phase, prior to mitosis (Hernandez-Ortega et al., 2007). Even though Cdk5 does not participate directly in cell cycle re-entry (Dhavan & Tsai, 2001), it was shown in cultured neurons exposed to Aβ peptide that the dysregulation of this kinase influences the levels and localization of Cdk4. Furthermore, the blockage of Cdk5 overactivation was shown to prevent Aβ-induced cell cycle re-entry, indicating that the action of Cdk5 occurs prior to the G1/S checkpoint transition (Lopes et al., 2009a).

This study aimed to elucidate in vivo the role of Cdk5 in the neurodegenerative mechanisms triggered by the Aβ_1–40 peptide, a key agent in AD. Thus, in mice intracerebroventricularly (icv) injected with Aβ, we evaluated the levels and subcellular localization of proteins related with different neurotoxic pathways involved in AD pathology. We assessed tau phosphorylation, by using antibodies for different stages of tau phosphorylation, as well as synaptic loss. To study cell cycle reactivation, we analyzed the levels and localization of Cdk4, cyclin D1, phospho-Rb (pRb) and proliferating cell nuclear antigen (PCNA). In addition, to verify if cell cycle re-entering neurons complete a full cycle, the levels of phospho-histone H3 (phH3), a marker for the M phase, were also assessed. The link of these events to Cdk5 dysregulation was obtained by evaluating the levels of this kinase and its activators p35 and p25, and further supported by the use of the calpain inhibitor MDL28170 administered intraperitoneally (ip) in mice injected icv with Aβ_1–40. Our results suggest that neuronal Cdk5 dysregulation induced in vivo by the Aβ peptide is mediated by calpain activation and leads to tau hyperphosphorylation, cell cycle reactivation and synaptic and neuronal loss. These findings implicate Cdk5 as a possible link between different processes that underlie the neurodegeneration in AD.

Results

Intracerebroventricular administration of Aβ_1–40 induces Cdk5 overactivation

In a previous in vitro study we demonstrated that Aβ_1–40 peptide can induce Cdk5 dysregulation, mainly due to an increase in calpain activity (Lopes et al., 2007). To evaluate the effects of the same peptide in vivo, we developed an icv Aβ-injection model, based in the AD mouse model established by Medeiros and Predinger (Medeiros et al., 2007). We started by evaluating the levels of Cdk5 in brain extracts of mice injected icv with Aβ_1–40 (AB mice) or PBS (CTR mice). Figure 1A shows that the icv administration of Aβ did not alter the levels of Cdk5, as compared with CTR mice. Furthermore, no changes were observed in Cdk5 levels following ip injection of the calpain inhibitor MDL28170 (MDL and ABMDL mice).

Figure 1.  Exposure to Aβ causes Cdk5 overactivation via an increase in the levels of p25. Brain lysates of CTR, AB, MDL, and ABMDL mice were analyzed by Western blot for (A) Cdk5, (B) p35 and p25. GAPDH was used as a loading control. Quantitative analysis of the levels of these proteins relative to GAPDH reveals that there are no alterations in the levels of Cdk5 in any of the studied animal groups. Although the levels of the p35 activator also remained unaltered in all experimental conditions, p25 was significantly augmented in animals icv-exposed to Aβ (AB), when compared with CTR and MDL mice. Co-treatment with Aβ (icv) and MDL28170 (ip) (ABMDL mice) successfully prevented the increase in p25 levels. The data in each bar of the graph represents mean ± SEM of five experiments (1 animal per experiment) and are expressed as percentage of control. **P < 0.01 compared with CTR animals (control); #P < 0.05 compared with AB animals.

In normal conditions, Cdk5 requires the association to the regulatory subunit p35 to be activated (Tsai et al., 1994). However, calpain cleavage of p35 leads to the formation of the pathogenic activator p25, which, when associated with Cdk5, will cause the dysregulation of this kinase (Patrick et al., 1999; Tsai et al., 2004). Thus, we evaluated the levels of both activators that impact on Cdk5 overactivation. Although p35 remained unchanged in all animal groups (Fig. 1B, higher bands), the levels of p25 were significantly (P < 0.01) altered in AB mice, when compared with CTR animals (Fig. 1B, lower bands). Upon co-treatment with MDL28170, the levels of p25 were reduced to values similar to those of the control. This decrease indicates that MDL28170 blocked calpain activation, thus preventing a p25-mediated Cdk5 overactivation.

Calpain inhibition prevents tau hyperphosphorylation and synaptic dysfunction in Aβ-injected mice

Cdk5 dysregulation is known to be related with exacerbated tau phosphorylation (Ferreiro et al., 2006; Lopes et al., 2007) that, in the case of AD, leads to the formation of neurofibrillary tangles (Grundke-Iqbal et al., 1986; Hardy, 2003), as well as to synaptic deficits (Arriagada et al., 1992; Eckermann et al., 2007). Therefore, we used antibodies which allowed us to distinguish different states of tau phosphorylation: CP13, which labels for early phosphorylated tau and AT8 and PHF-1 that correspond to later forms of tau phosphorylation (Fig. 2). Our results showed increases in the levels of all tau forms in the brain of AB mice (Fig. 2A,B), although only CP-13 and AT8 were statistically significant (P < 0.05). The augment in tau phosphorylation induced by the Aβ peptide was prevented upon administration of MDL28170 (Fig. 2A,B). We also assessed the levels of normal tau (nonphosphorylated at serine 202) using the BT2 antibody, which yielded no alterations when comparing all the studied conditions (Fig. 2A). Immunofluorescence results also confirmed that CP-13 and AT8 immunoreactivity were higher in AB animals than in CTR mice, especially in the dentate gyrus, as can be seen in Fig. 2C,D – A,B. In MDL and AΒMDL animals, the levels of both forms of tau display an immunoreactivity pattern identical to those of the controls (Fig. 2C,D – C,D).

Figure 2.  Calpain inhibition prevents Aβ-induced tau hyperphosphorylation and synaptic loss. (A) Lysates from brains of CTR, AB, MDL, and ABMDL mice were analyzed by Western blot for different tau phosphorylation stages: CP13, AT8, and PHF-1, as well as for nonphosphorylated tau (BT2). (B) Quantitative analysis of the levels of the three proteins relative to GAPDH (loading control) revealed an increment in the levels of all phospho-tau forms in mice exposed to Aβ (AB animals), as compared with CTR mice. Calpain inhibition by MDL28170 ip administration in co-treatment with icv Aβ (ABMDL mice) successfully blocked tau hyperphosphorylation. (C) Immunofluorescence images representative of CP13 tau in dentate gyrus of the hippocampus from mice of the different experimental conditions. In AB animals (B and b) there is a significant increase in CP13 immunoreactivity, as compared with CTR (A) and MDL (C) mice. Co-treatment with calpain inhibitor (ABMDL) leads to a marked prevention of tau phosphorylation. (d) Immunofluorescence analysis of AT8 tau in the dentate gyrus of hippocampi from mice of the different experimental conditions. AB mice (B and b) display a significant increase in AT8 immunoreactivity, when compared with CTR (A) and MDL (C) animals. Calpain inhibitor co-treatment (ABMDL) leads to marked prevention of tau phosphorylation. *P < 0.05, compared with CTR animals (control); #P < 0.05, ##P < 0.01 compared with AB animals. Images were collected by fluorescence (A–D) and confocal (b) microscopy using 50× and 400× magnification, respectively.

Hyperphosphorylation of tau is associated with a disruption of the microtubules, which can lead to synaptic dysfunction and loss (Arriagada et al., 1992; Eckermann et al., 2007). In fact, we observed a significant (P < 0.01, data not shown) decrease in the levels of the synaptic marker synaptophysin in Aβ mice, when compared with the control littermates. This synaptotoxic effect of Aβ peptide was successfully prevented (P < 0.05) in ABMDL animals. MDL28170 alone did not affect the levels of synaptophysin.

Cdk5 is relocalized in Aβ-injected mice

Recently we showed that Cdk5 plays a significant role in cell cycle reactivation induced by Aβ and PrP peptides (Lopes et al., 2009a). Therefore, as some studies suggest that Cdk5 is translocated from the nucleus when cell cycle re-entry occurs (Zhang et al., 2008), we further determined the subcellular localization of this kinase in neurons from brain slices of animals from the different groups. Immunofluorescence images show some degree of co-localization between Cdk5 and the nuclear marker Hoechst 33342 in CTR, MDL, and ABMDL mice (Fig. 3A, A–C and G–L, arrows). However, in AB mice the neurons exhibited a mainly cytoplasmic localization for this kinase (Fig. 3A, D–F, arrows). Furthermore, the number of neurons labeled for Cdk5 that did not display nuclear co-localization with Hoechst was significantly increased (P < 0.01) in AB animals, an effect prevented by ip treatment with MDL28170 (Fig. 3B).

Figure 3.  Cdk5 in relocalized in mice icv-injected with Aβ. (A) Confocal microscopy images representative of Cdk5 and Hoechst 33342 labels in hippocampal neurons of CTR, AB, MDL, and ABMDL mice. Cdk5 labeling is shown in red, whereas the Hoechst dye, used to label the nuclei, is blue. The overlay of these images (Merge) shows that in neurons of control animals (A, B, C), as well as in MDL mice (G, H, I), the distribution of Cdk5 is both nuclear and cytoplasmic (arrows), whereas in neurons treated with Aβ (D, E, F) there is a relocalization from the nuclear area to the cytoplasm (arrows). In ABMDL mice (JKL), calpain inhibitor treatment leads to a prevention of nuclear translocation. Merge detail corresponds to a magnification of the area outlined in the Merge column. (B) Quantification of neurons with nonnuclear Cdk5 labeling shows a significant elevation in AB mice, whereas CTR, MDL, and ABMDL animals showed no differences to the control. The data in each bar of the graph are represented as mean + SEM of four experiments. **P < 0.01 compared with CTR animals (control); #P < 0.05 compared with AB animals.

Aβ exposure promotes Cdk4 increase without alterations in cyclin D1 levels

It has also been shown that upon Cdk5 dysregulation, the levels of cell cycle proteins in cultured cortical neurons become altered (Lopes et al., 2009b). Therefore, we decided to study different proteins involved in the two main checkpoints of the cell cycle: G1/S and the G2/M, in animals icv-injected with Aβ. Using Western blot and immunofluorescence techniques we evaluated the levels and cellular distribution of Cdk4, a kinase directly involved in the transition between the G1 and the S phases (Tannoch et al., 2000). In Fig. 4A it can be seen that AB animals exhibited a significant (P < 0.05) increase in Cdk4 levels, when comparing with CTR mice. This effect was prevented in animals co-treated with MDL. Furthermore, exposure to Aβ induced a change in the localization pattern of this kinase (Fig. 5B). In hippocampal neurons from brain slices of AB mice, partial overlapping between Cdk4 labeling and the nuclear dye Hoechst was observed (Fig. 4B, D–F), whereas in CTR, MDL, and ABMDL animals the localization of this kinase remained exclusively cytoplasmic (Fig. 4B, A–C and G–L). Indeed, the analysis of the percentage of neurons exhibiting co-localization between Cdk4 labeling and Hoechst displayed a significant increase (P < 0.01) for in animals icv-injected with Aβ (AB mice), which was successfully prevented by MDL28170 (ABMDL animals) (Fig. 4C).

Figure 4.  Aβ exposure induces Cdk4 increase and relocalization. (A) Brain lysates of CTR, AB, MDL, and ABMDL mice were analyzed by Western blot for Cdk4 levels. GAPDH was used as a loading control. Quantitative analysis reveals an increase of Cdk4 levels upon Aβ exposure, when compared with CTR and MDL animals. Co-treatment with Aβ (icv) and MDL28170 (ip) successfully prevented the alterations in Cdk4 levels. (B) Immunocytochemistry images representative of Cdk4 and Hoechst 33342 labels in hippocampal neurons of CTR, AB, MDL, and ABMDL mice. Cdk4 labeling is shown in red, whereas the Hoechst dye, used to label the nuclei, is blue. The overlay of these images (Merge) shows that in neurons from CTR (A, B, C) and MDL (G, H, I) mice, the distribution of Cdk4 is mainly perinuclear, whereas in neurons of Aβ-treated mice (D, E, F) there is a relocalization to the nuclear area. In ABMDL animals, calpain inhibition successfully prevented the alterations in Cdk4 localization. (C) Quantification of neurons with co-localization between Cdk4 and the nuclear dye Hoechst displays a significant increase for AB mice, as compared with CTR, MDL, and ABMDL animals. The data in each bar of the graph are represented as mean + SEM of four (c) or five (a) experiments. Compared with CTR animals (control); #P < 0.05, compared with AB animals. Images were collected by confocal microscopy using 630× magnification.

Figure 5.  Cyclin D1 localization, but not levels, is altered in icv Aβ-injected mice. (A) The levels of cyclin D1 in hippocampal neurons of CTR, AB, MDL, and ABMDL were determined by Western blot analysis, using GAPDH as a loading control. Quantitative analysis of cyclin D1 expression showed no changes in any of the experimental groups. (B) Immunofluorescent analysis of cyclin D1 showed a mainly cytoplasmic localization in CTR (A, B, C) and MDL (G, H, I) animals, which changes to an overall nuclear localization in neurons of AB animals (D, E, F – arrows). Calpain inhibition by ip administration of MDL28170 simultaneously to icv Aβ, prevented the changes in the localization of cyclin D1. Cyclin D1 labeling is shown in green, whereas the Hoechst dye, used to label the nuclei, is blue. (C) Quantitative assessment of the percentage of neurons with co-localization between cyclin D1 and the nuclear dye Hoechst yielded a significant augment in AB animals, when compared with the remaining experimental mouse groups. data in each bar of the graph represents mean ± SEM of four (c) or five (a) experiments (1 animal per experiment) and are expressed as percentage of control. **P < 0.01 compared with CTR animals (control); #P < 0.05 compared with AB animals. Images were collected by confocal microscopy using 630× magnification. The images were collected by confocal microscopy using 630× magnification.

As this kinase alone does not exhibit enzymatic activity, thus requiring the association to a regulatory subunit for activation, we also studied the levels of cyclin D1, the main activator of Cdk4 (Weinberg, 1995; Tannoch et al., 2000). No changes were detected in the levels of this protein in any of the studied animal groups (Fig. 5A). However, immunofluorescence images showed a different subcellular distribution of cyclin D1 in AB animals (Fig. 5B, D–F), as compared with the CTR group (Fig. 5B, A–C). Whereas control mice displayed a cytoplasmic distribution of this protein, characteristic of terminally differentiated neurons (Sumrejkanchanakij et al., 2003), hippocampal neurons of AB animals exhibited cyclin D1 immunoreactivity in the nucleus, similarly to the one observed for Cdk4 localization. In mice treated with MDL, no nuclear labeling for cyclin D1 was observed (Fig. 5B, G–L). In accordance, as can be seen in Fig. 5C, there is a significant augment (P < 0.01) in the percentage of cells yielding cyclin D1 nuclear localization in AB mice, when compared with CTR animals, and that is prevented in ABMDL mice.

Aβ increases pRb and PCNA levels through a Cdk5-mediated mechanism

The activation of Cdk4 can lead to the phosphorylation of the Rb, one of the most important steps in cell cycle re-entry at the G1/S checkpoint (Weinberg, 1995; Tannoch et al., 2000). So, the levels of pRb were also evaluated. In accordance with the results of Cdk4, pRb levels were also significantly (P < 0.05) augmented due to Aβ peptide exposure, as compared with the controls (Fig. 6A). However, when Aβ-injected animals were co-treated with MDL28170, a significant (P < 0.05) reduction in pRb was observed. Moreover, while in CTR animals pRb was mainly localized in the cytoplasm (Fig. 6D, A–C), in AB mice there was a change to a nuclear allocation (Fig. 6D, D–F). This change in the distribution pattern was prevented in ABMDL animals (Fig. 6D, J–L), being also identical in MDL mice (Fig. 6D, G–I).

Figure 6.  pRb and PCNA, but not phH3 levels are increased in animals injected icv with Aβ. Lysates from brains of CTR, AB, MDL, and ABMDL mice were analyzed by Western Blot for the levels of: (A) pRb, (B) PCNA, and (C) phH3. GAPDH was used as a loading control. Quantitative analysis shows that AB mice have increased levels of pRb and PCNA, when compared with CTR and MDL animals. Treatment with MDL28170 (ip) simultaneously to Aβ (icv) administration (ABMDL animals) successfully prevented the alterations in both proteins. The assessment of phH3 showed no alterations in the levels of this protein in any of the experimental animal groups. (D) Immunocytochemistry images representative of pRb and Hoechst 33342 labels in hippocampal neurons of CTR, AB, MDL, and ABMDL mice. pRb labeling is shown in red, whereas the Hoechst dye, used to label the nuclei, is blue. The overlay of these images (Merge) shows that Aβ exposure causes a change in the subcellular distribution of pRb, from a perinuclear (A, B, C and G, H, I – arrows) to a nuclear localization (D, E, F – arrows). In ABMDL animals, calpain inhibition successfully prevented the alterations in pRb localization (J, K, L – arrows). The data in each bar of the graph represents mean ± SEM of five experiments (1 animal per experiment) and are expressed as percentage of control. *P < 0.05, **P < 0.01 compared with CTR animals (control); #P < 0.05, ##P < 0.01 compared with AB animals. Images were collected by confocal microscopy using 630× magnification.

Upon Rb phosphorylation, the cell cycle overcomes the G1/S checkpoint and advances to the S phase (Park et al., 2000). Therefore, it was relevant to evaluate a marker for this cell cycle stage. Figure 6B shows that PCNA levels were increased (P < 0.01) in AB animals, when compared with CTR mice, whereas ABMDL mice exhibited values similar to the controls.

Re-cycling postmitotic neurons degenerate before the G2/M checkpoint

To further follow the progress of re-cycling neurons along the cell cycle, we evaluated the levels of the M phase marker, phH3. No alterations were observed in the levels of this protein in any of the studied animal groups (Fig. 6C), indicating that the neurons that re-enter the cell cycle do not progress past the G2/M point.

Using the dye Fluoro-Jade C (FJC) we evaluated if the exposure to Aβ would lead to neuronal loss. FJC staining of brain slices from AB mice revealed a marked neurodegeneration in the dentate gyrus area (Fig. 7B, arrows), as compared with CTR and MDL animals (Fig. 7A,C). In icv Aβ-injected animals that received ip MDL28170, neuronal death was considerably prevented (Fig. 7D).

Figure 7.  MDL28170 ip co-administration prevents Aβ-induced neuronal death in icv-injected mice. Confocal microscopy images representative of degenerating neurons labeled with the Fluoro-Jade C dye in the dentate gyrus of hippocampi from CTR, AB, MDL, and ABMDL mice. AB mice (B) display a significant increase in the immunoreactivity for Fluoro-Jade C (arrows), when compared with CTR (A) and MDL (C) animals, thus representing a marked augment in the levels of degenerating hippocampal neurons. Calpain inhibitor co-treatment (ABMDL mice) leads to a significant decrease in the levels of this dye. All images were collected using 200× magnification.

Discussion

Using a mouse model of AD, we obtained data that suggests the involvement of Cdk5 dysregulation in the neuronal demise that occurs in AD. This mouse model that consisted on the icv injection of Aβ_1–40, was based on a previously used and validated model by Medeiros et al., in which the animals display significant impairments of learning and memory functions, as well as marked synaptotoxicity, after only 7 days of exposure to the peptide (Medeiros et al., 2007).

Our data show that ‘aged’ Aβ, which is mainly constituted by fibrils (Schmid et al., 2008), induced in vivo Cdk5 dysregulation through a mechanism involving calpain activation. The calcium dependence of calpains renders these cytosolic proteases vulnerable to changes in the homeostasis of this ion. In fact, Aβ peptides promote an imbalance in intracellular calcium levels, both due to Ca²⁺ influx via voltage-sensitive channels (Sjogren et al., 2001), and through its release from intracellular compartments, namely the endoplasmic reticulum (Agostinho & Oliveira, 2003; Ferreiro et al., 2006). In cultured cortical neurons, these Ca²⁺ alterations were shown to occur after only 1 h of incubation with Aβ (Ferreiro et al., 2006), which can indicate that calpains will be in the first group of proteins to be affected by the dysregulation of calcium homeostasis in AD.

Calpain activation has been described in several pathologies (Li et al., 1998; Chen et al., 2006; Raynaud & Marcilhac, 2006), including many where Cdk5 dysregulation also occurs (Tsai et al., 2004; Wen et al., 2007; Alvira et al., 2008). Recently we demonstrated that Aβ and PrP peptides can trigger calpain activation, leading to Cdk5 dysfunction in cultured rat cortical neurons (Lopes et al., 2007). Activated calpains cleave the normal regulatory subunit p35 to p25, thus forming a p25/Cdk5 complex with an activity profile substantially higher than when p35 is associated with the kinase (Patrick et al., 1999). Cdk5 overactivation has been demonstrated in neuronal cultures exposed to Aβ and transgenic models of AD (Cruz et al., 2003; Kitazawa et al., 2005; Lopes et al., 2007). In our model, a significant increase in the level of p25 was observed in mice icv-injected with Aβ_1–40 (Fig. 1). This effect was successfully prevented by the intraperitoneal (ip) co-treatment with a calpain inhibitor capable of crossing the blood-brain barrier, MDL28170 (Li et al., 1998). The subcellular localization of the p25/Cdk5 complex is also altered, changing substrate specificity and leading to the hyperphosphorylation of substrates not normally phosphorylated by this kinase, like the cytoskeleton protein tau (Patrick et al., 1999). Tau hyperphosphorylation in AD is associated with the formation of neurofibrillary tangles, one of the markers for this pathology. Upon phosphorylation, tau molecules are released into the cytoplasm and polymerization starts (Grundke-Iqbal et al., 1986; Alonso et al., 2001; Hardy, 2003). Indeed, it has been reported that in AD both phosphorylated and total tau are augmented (Sjogren et al., 2001). In the current work we showed that exposure to Aβ triggered the increase of three different phospho-tau forms: CP13, AT8, and PHF-1 although the levels of normal tau were not altered (Fig. 2). Similarly to the observed by Medeiros et al. (2007), synaptic loss was registered in the mice injected icv with Aβ_1–40. This decrease of synaptic markers is known to be correlated with the occurrence of cognitive deficits and is in accordance with a recent study in another rodent model that demonstrated the deleterious effect of Aβ peptide injection in memory and learning (Cunha et al., 2008). Co-treatment of Aβ-injected animals with the calpain inhibitor MDL28170 successfully prevented the alterations in both tau phosphorylation and synaptic loss. This result is in line with that obtained for p25 generation and indicates that Cdk5 overactivation is involved and underlies the two events.

Apart from tau hyperphosphorylation, cell cycle re-entry has also been suggested as a possible neuropathological feature of AD (Nagy et al., 1997; Hernandez-Ortega et al., 2007; Lopes et al., 2009b). Cell cycle reactivation has been observed in the brain of AD patients (Busser et al., 1998), AD mouse models (Yang et al., 2006) as well as in cultured neurons exposed to Aβ peptides (Biswas et al., 2007; Majd et al., 2008; Lopes et al., 2009a). In both situations, cell cycle re-entry is associated with an increase in the levels of Cdk4 (Busser et al., 1998; Lopes et al., 2009a), a kinase involved in the phosphorylation of the Rb, one of the most relevant steps of cell cycle re-entry in postmitotic neurons (Weinberg, 1995; Tannoch et al., 2000). Furthermore, upon activation Cdk4 translocates to the nucleus (Nguyen et al., 2003; Sumrejkanchanakij et al., 2003) where it can access its phosphorylative target Rb. Similarly to that registered in AD brain and in Aβ-treated neuronal cultures, in this icv Aβ injection mouse model, no changes were observed in cyclin D1 (Cdk4 activator in the G1 phase) levels. However, we verified that Aβ exposure also led to a change in the subcellular localization of cyclin D1, moving from the cytoplasm to the nucleus. Thus, the increase in Rb phosphorylation depends essentially of augmented Cdk4 levels and not of changes in its activity, requiring, however, a translocation of the cyclin D1/Cdk4 complex to the nucleus. These results are in accordance with others obtained in a mouse model of ALS and in cultured cortical neurons (Nguyen et al., 2003; Lopes et al., 2009a).

Interestingly, Cdk5 overactivation has also been described in several neurodegenerative conditions (Nguyen & Julien, 2003; Tsai et al., 2004; Qu et al., 2007; Wen et al., 2007) in which ectopic cell cycle events were also reported (Busser et al., 1998; Nguyen et al., 2003; Hoglinger et al., 2007; Wen et al., 2007). A recent work by Zhang et al. (2008) described changes in the subcellular localization of Cdk5 when cell cycle re-entry occurs. This observation was also validated in our neurodegeneration model, since we observed that in Aβ-treated mice, Cdk5 localization is exclusively cytoplasmic whereas in the remaining experimental conditions, as like occurs in normal postmitotic neurons, it has a distribution through the whole cell (Fig. 3). Furthermore, the involvement of Cdk5 in neuronal cell cycle re-entry was demonstrated by the blockage of the increases in the levels of Cdk4, pRb, and PCNA, as well as the prevention of nuclear translocation of the cyclin D1/Cdk4 complex, achieved by intraperitoneal administration of MDL28170 in AB mice (Figs. 4–6). Considering that the alterations produced by this calpain inhibitor not only affected Cdk5, but also the cell cycle markers, we can suggest that Cdk5 overactivation occurs upstream from cell cycle re-entry. Although the link between the overactivation of Cdk5 and alterations in Cdk4 expression/activity (Nguyen et al., 2003; Lopes et al., 2009a) remains unknown, we can hypothesize how this regulatory action is made. The normal priming phosphorylation required by the cyclin D1/CDK4 complex to act on Rb is made by the CDK-activating kinase (CAK) (Diehl & Sherr, 1997; Gladden & Diehl, 2005). Assuming that in postmitotic neurons the expression/activity of CAK is reduced, overactivated Cdk5 can eventually phosphorylate the D1/Cdk4 complex, leading to its activation and, consequently, to abortive cell cycle re-entry.

Although the re-active cell cycle progresses past the G1/S checkpoint, reaching the S or even the G2 phase, the lack of a normal cell cycling machinery in these postmitotic neurons, when compared with normal cycling cells, will cause them to degenerate prior to the G2/M checkpoint (Hernandez-Ortega et al., 2007). Besides the absence of changes in the M phase marker phH3 in the brains of icv Aβ-injected animals, which labeled positive for markers of previous cell cycle phases, in the dentate gyrus of these mice there is a significant amount of neurons labeled positive for the FJC dye, an indicator of degenerating neurons (Bian et al., 2007). Indeed, in vitro studies with Aβ-treated cells demonstrated that Rb phosphorylation precedes the activation of Bax, a proapoptotic Bcl-2 family member, which leads to caspase-3 activation and apoptotic neuronal death (Giovanni et al., 2000; Ramalho et al., 2004).

Overall, the data indicate that Aβ triggers neuronal cell cycle re-entry through a mechanism involving Cdk5. This kinase is also associated with tau hyperphosphorylation and synaptic loss. The use of MDL28170, which inhibits calpain activation and, indirectly, blocks Cdk5 overactivation, was shown to prevent the deleterious effects induced by Aβ peptide in the brain. The possibility of using MDL28170 as a possible therapeutic agent is also an interesting question to address, especially due to its ability to cross the blood-brain barrier (Li et al., 1998). Several studies have already demonstrated the efficacy of this inhibitor in the prevention of neuronal and motor deficits in neurodegenerative conditions (Crocker et al., 2003), as well as in the prevention of the damage caused by ischemia, an event known to cause Cdk5 overactivation (Li et al., 1998). A major concern with the use of calpain inhibitors, especially when addressing their administration to neurons, is related with possible harmful effects, as calpains are involved in diverse important cellular mechanisms. Therefore, the dosage and exposure time to these compounds need to be closely monitored. In our study we tested MDL28170 alone (MDL) and, as the results obtained did not differ from the ones in the control condition (CTR), this allowed us to conclude that the administered amount of this calpain inhibitor was sufficient to prevent calpain overactivation caused by the Aβ peptide without causing any significant toxicity. However, the absence of specificity toward the neuronal population, and, ultimately, to degenerating neurons is still an issue, as there is a risk of disturbing other cell types in which the cell cycle remains active, such as astrocytes and glial cells, as well as subventricular zone and hippocampal neuronal precursor cells. Therefore, further investigation is needed, namely concerning the processes underlying pathological cell cycle reactivation in mature neurons and the pathways connecting it to Cdk5 dysregulation. The design of new studies aiming to identify the mechanisms of AD-related cell cycle reactivation cascades will be valuable in the identification of possible targets for the development of therapeutic approaches on AD.

Experimental procedures

Materials

Synthetic peptide Aβ_1–40 was from Bachem (Bubendorf, Switzerland). Alexafluor® IgG conjugate secondary antibodies and Hoechst 33342 were acquired from Molecular Probes (Leiden, the Netherlands). The fluorescent mounting medium was from DakoCytomation (Glostrup, Denmark) and the nonfluorescent mounting medium DPX was purchased from Fluka Chemie AG (Buchs, Switzerland). Reagents and apparatus used in immunoblotting assays were obtained from Bio-Rad (Hercules, CA, USA), whereas polyvinylidene fluoride (PVDF) membranes, alkaline-phosphatase-linked anti-mouse secondary antibody and the enhanced chemifluorescence (ECF) reagent were from Amersham Biosciences (Buckinghamshire, UK). Antibodies used include: monoclonal anti-Cdk5 (C-8), anti-p35/p25 (C-19), anti-PCNA (PC10), anti-cyclin D1 (A12) and polyclonal anti-Cdk4 (C-22) antibodies, from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Rabbit polyclonal antibodies against the phosphorylated form of Rb (Ser807/811) and α-tubulin were from Cell Signaling (Denvers, MA, USA). Mouse monoclonal antibody against phH3 (RR002) was purchased from Upstate (Lake Placid, NY, USA). Mouse monoclonal antibodies against human total tau (BT2), human PHF-tau (AT8) and synaptophysin (SVP-38) were purchased from Pierce Endogen (Rockford, IL, USA) and Sigma-Aldrich, respectively. Mouse monoclonals anti-CP-13 tau and anti-PHF-1 tau were a generous gift from Dr Peter Davies (Albert Einstein College of Medicine, Bronx, NY, USA). The monoclonal antibody anti-GAPDH (6C5) was from Chemicon-Millipore (Temecula, CA, USA). FJC was obtained from Chemicon (Billerica, MA, USA). MDL28170 and all other reagents were from Sigma-Aldrich (St. Louis, MO, USA).

Mouse treatments

Experiments were conducted using female C56BL/6 mice with approximately 8 weeks. The mice were divided in four groups, according to the established on Fig. 8B.

Figure 8.  Schematic representation of time course and experimental manipulations applied in this study. (A) The horizontal line represents the duration of the experiment. The numerals below the horizontal line indicate different time points. The vertical arrow above the horizontal line indicate the timing of Aβ_1–40 (1.16 nmol per animal) or PBS icv administration. The arrows below correspond to ip administration of MDL28170 (2 mg kg⁻¹ administration at day 0 and 1 mg kg⁻¹ in the remaining administrations) or PBS. All animals were sacrificed at day 15. See Experimental procedures for details. (B) The experimental animals were divided in four groups, depending on the combinations of Aβ/PBS icv administration and MDL28170/PBS ip administration: CTR, AB, MDL, and ABMDL.

Aβ_1–40 administration

Aβ_1–40 was prepared according to the manufacturer’s instructions (Bachem) and aged in PBS buffer for 1 days at 33 °C, in a stock concentration of 1 mg mL⁻¹. The mice were anesthetized with avertin (1.3% tribromoethanol, 0.8% amylalcohol) and placed in a stereotactic apparatus (Stoelting, Wood Dale, IL, USA). A single dose of 5 μL (1.16 nmol) of Aβ_1–40 was administered icv to 11 mice, corresponding to groups AB and ABMDL, at a rate of 0.5 μL min⁻¹. Mice used as controls (CTR and MDL) received icv the same volume in PBS (Fig. 8).

MDL treatment

MDL28170, a selective calpain inhibitor, was dissolved in PBS and administered intraperitoneally (ip) to half of the animals (groups MDL and ABMDL, n = 11). The remaining mice (groups CTR and AB, n = 11) received injections in the same manner with PBS only. Following a first injection of MDL28170 (2 mg kg⁻¹ of animal weight) on the same day as the Aβ_1–40 icv administration, the animals were injected (1 mg kg⁻¹ of animal weight) in alternate days for 2 weeks, according to the scheme in Fig. 8A. Mice were killed 24 h after the last injection. The brain was isolated, one of the brain hemispheres was fixed in 4% paraformaldehyde and the other hemisphere was used to make brain extracts.

All experimental work involving the use of these animals was conducted in a way to minimize animal discomfort, according to the ethically approved institutional guidelines that are in compliance with national and international laws and policies (EEC Council Directive 86/609). Previously and after injection all animals were housed at constant temperature (22 °C) and humidity (55%), with a fixed 12 h light-dark cycle and free access to food and water. After icv surgery, the animals were kept on a warming pad until they had fully recovered from anesthesia. All animals were killed by rapid decapitation after anesthesia.

Western blot

For the preparation of total and soluble brain extracts, one brain hemisphere was placed in ice-cold lysis buffer containing (in mm): HEPES-Na 25, MgCl₂ 2, EDTA 1, EGTA 1, supplemented with 100 μm PMSF, 2 mm DTT, 2 mm orthovanadate, 50 mm sodium fluoride and protease inhibitor cocktail (containing 1 μg mL⁻¹ leupeptin, pepstatin A, chymostatin, and antipain). Following mechanical dissociation and sonication, one half of the brain lysate was collected as total brain extract. The remaining half was frozen three times in liquid N₂ and centrifuged at 14 000 g for 10 min at 4 °C. The supernatant was collected as soluble extract. Protein concentration in both extracts was measured using the Bio-Rad protein dye assay reagent. Samples were denaturated at 95 °C for 5 min in a sample buffer, containing (in mm): Tris 83.3, DTT 100, 10.3% SDS, 30% glycerol, and 0.012% bromophenol blue. Equal amount of each sample of protein was separated by electrophoresis on a 10% SDS–PAGE and electroblotted onto PVDF membranes. The identification of proteins of interest was facilitated by the usage of a prestained precision protein standard (Bio-Rad), which was run simultaneously. The proteins in gel were electrophoretically transferred to membranes that were incubated for 1 h at room temperature (RT) in Tris buffer (TBS-T (in mm) NaCl 150, Tris–HCl 25, pH 7.6, with 0.1% Tween 20), containing 5% bovine serum albumin (BSA) to block nonspecific binding. Then the membranes incubated with the primary antibodies overnight at 4 °C in TBS-T containing 1% BSA. The primary antibodies used were: (i) rabbit monoclonal anti-Cdk5 (1:500 dilution), (ii) rabbit monoclonal anti-p35 (1:500 dilution), (iii) mouse monoclonal anti-AT8 tau (1:250 dilution), (iv) mouse monoclonal anti-nonphosphorylated tau (1:250 dilution), (v) mouse monoclonal anti-CP13 tau (1:10 dilution), (vi) mouse monoclonal anti-PHF-1 tau (1:20 dilution), (vii) rabbit polyclonal anti-Cdk4 (1:500), (viii) mouse monoclonal anti-cyclin D1 (1:1000), (ix) rabbit polyclonal anti-phosphorylated Rb (1:500), (x) mouse monoclonal anti-PCNA (1:1000), (xi) mouse monoclonal anti-phH3 (1:15 000) and (xii) mouse monoclonal anti-synaptophysin (1:40 000). After this incubation, the membranes were washed and incubated in TBS-T with 1% BSA, for 2 h at RT, with the appropriate alkaline-phosphatase-conjugated anti-rabbit or anti-mouse secondary antibody at a dilution of 1:25 000 or 1:20 000, respectively. GAPDH (1:2500) and α-tubulin (1:30 000) were used as loading controls. Immunoreactive bands were detected after incubation of membranes with ECF reagent for 5–10 min, on a Bio-Rad Versadoc 3000 Imaging System.

Immunofluorescence

Free-floating sections (50 μm thick) were sectioned on a Leica CM3050 S cryostat (Leica Microsystems, Wetzlar, Germany), and stored in 1% sodium azide in PBS. Fluorescent immunohistochemistry followed standard protocols as previously described (Blurton-Jones & Tuszynski, 2006). Briefly, sections were permeabilized with 0.2% Triton X-100/PBS for 15 min at RT, and blocked with 5% goat serum before incubation with primary antibody overnight at 4 °C. Afterwards, sections were incubated with appropriate Alexa Fluor secondary antibody for 1 h at RT. Subsequently, sections were washed in PBS, mounted and cover slipped with DakoCytomation fluorescent mounting medium (Carpinteria, CA, USA). Representative images obtained randomly from the neocortical area of brain sections were visualized with a Zeiss LSM 510Meta confocal microscope (Göttingen, Germany) or a Zeiss Axiovert 200 m fluorescence microscope. All images were obtained using the same exposure time and qualitative analysis was performed in a double blind manner.

Fluoro-Jade staining

Fluoro-Jade C stain procedure was carried out according to the standard protocol. First the sections were immersed in a solution containing 1% sodium hydroxide in 80% alcohol for 5 min, followed by 70% alcohol and distilled water each for 2 min. The sections were transferred into a solution of 0.06% potassium permanganate for 10 min, and rinsed in distilled water for 2 min. For FJC staining, the sections were immersed into 0.0001% solution of FJC dye dissolved in 0.1% acetic acid vehicle (pH 3.5) and stained for 10 min. After incubation in the FJC solution, the slides were washed three times in distilled water (each for 1 min) and left to dry overnight in darkness at RT. The sections were then air-dried, dehydrated in ethanol, cleared in xylene and coversliped with DPX. Finally, the FJC-stained sections were analyzed using a Zeiss LSM 510Meta confocal microscope. The FJC-positive stain exhibited strong green color when using a 488 nm excitation wavelength and a barrier filter that allows passage of all wavelengths longer than 505 nm. All images were obtained using the same exposure time and qualitative analysis was performed in a double blind manner.

Statistical analysis

Results are expressed as mean ± SEM. Statistical analysis was performed with Graphpad Prism software (Graphpad Software, La, Jolla, CA, USA). Significance was determined using an analysis of variance (anova), followed by Dunnet’s post hoc tests for comparisons between control and treated animals. For comparisons between mice groups with different treatments, a two-tailed Students’t-test was used.

Acknowledgments

We thank Paula Canas for the technical advising concerning the stereotaxic surgery technique, Prof. Rodrigo Cunha for providing the Fluoro-Jade C dye and Dr. Peter Davies for the CP13 and PHF-1 tau antibodies.