Critical loss of CBP/p300 histone acetylase activity by caspase-6 during neurodegeneration

Histones are deacetylated during the onset of neuronal apoptosis

Acetylated histone levels were followed by western blot analysis during neuronal apoptosis. Acetylated H3 (Ac-H3) and H4 (Ac-H4) histone levels progressively decreased during K⁺ deprivation of CGNs (LK) (Figure 1A, upper panels), whereas total amounts of cellular histones were not modified (Figure 1A, lower panel). This acetylation decrease was confirmed further by immunostaining and Hoechst labeling (Figure 1B). In survival conditions (HK), nuclei that appeared large and round shaped displayed a bright Ac-H3 and Ac-H4 immunoreactivity. In apoptotic conditions (LK), condensed or fragmented nuclei displayed little or no Ac-H3 and Ac-H4 immunostaining. To investigate whether histone deacetylation was correlated with apoptosis entry, we performed a more precise time course of K⁺ deprivation on H4 histone acetylation status coupled to an evaluation of apoptotic nuclei. Figure 1C shows that histone deacetylation occurred prior to nuclear condensation, since Ac-H4 levels clearly decreased at 6 h of LK treatment, a time at which no significant changes in the percentage of apoptotic nuclei were detectable. It is noteworthy that acetylated H4 levels are even slightly lower than control at 2 and 4 h of LK treatment. Interestingly, Ac-H4 deacetylation was partly reversed by the general caspase inhibitor z-VAD-fmk (Figure 1D). Altogether, these results show that histone deacetylation is an early event within the apoptotic pathway, that occurs through caspase-dependent mechanisms.

CBP/p300 HAT activity specifically decreases during apoptosis

To investigate the possible cause of histone deacetylation during neuronal apoptosis, we investigated the fate of different HAT families by western blot analyses. Four HATs were tested: CBP, p300, TAF II p250 and PCAF. We found a strong decrease of CBP levels as soon as 6 h of LK treatment (Figure 2A). p300 levels also dropped between 6 and 12 h of K⁺ deprivation (Figure 2A). In contrast, neither TAF II p250 nor PCAF levels were changed upon K⁺ deprivation (Figure 2A), suggesting a selective decrease of the CBP/p300 HAT family of proteins during apoptosis. We then evaluated the consequences of CBP/p300 loss on either CBP/p300 HAT activity or global HAT activity (Figure 2B). The assay was tuned using a recombinant HAT protein, CBP (a generous gift from Dr J.R.Lundblad, Portland, OR), as positive control. CBP/p300 HAT activity decreased by 73% after 12 h of K⁺ deprivation and by 88% after 24 h, a decrease partly reversible by z-VAD-fmk (Figure 2B). Interestingly, global HAT activity was not altered by LK treatment (Figure 2B). This could probably be reflected by the fact that some HATs such as TAF II p250 can acetylate histones in an in vitro assay (Kuo and Allis, 1998), whereas they would have less affinity in living cells, when nucleosomal histones are in the chromatin context. Altogether, these results show a specific loss of CBP/p300 during neuronal apoptosis by a caspase-dependent mechanism, resulting in a drop in their HAT activity which could be responsible for histone deacetylation.

CBP is degraded during apoptosis

Further investigations were performed on CBP, that is potentially involved in neuroprotection. Semi-quantitative RT–PCR analysis revealed that cbp expression was not modified during K⁺ deprivation relative to the housekeeping gene g3ph (Figure 3A), suggesting that the decrease of CBP levels was likely to be due to a proteolytic processing. We therefore tested different protease inhibitors on CGN for their ability to reverse CBP degradation (Figure 3B). The proteasome inhibitor PSI had no effect on CBP levels, whereas z-VAD-fmk partly reversed the CBP decrease. The calpain inhibitor I, ALLN, did not reverse CBP degradation but allowed us to show a cleavage fragment migrating at an apparent mol. wt of ∼175 kDa (CBP-I, Figure 3B). The use of a CBP N-terminus-directed antibody in these conditions allowed the detection of another cleavage fragment migrating at ∼130 kDa (CBP-II, data not shown). These results suggest that CBP might be cleaved by caspases during apoptosis, giving two cleavage fragments further processed by calpains.

CBP is a caspase-6 substrate

To identify further the caspase(s) involved in CBP proteolysis during apoptosis, we investigated caspase activities using fluorogenic substrates. Caspase-3 is the main executioner caspase of the CNS and more particularly of CGN apoptosis (D’Mello, 1998; Boutillier et al., 2000). Caspase-6 was described as an initiator caspase, responsible for caspase-3 activation in K⁺/serum-deprived CGNs (Allsopp et al., 2000). We found that caspase-6 was moderately but rapidly activated (1 h) after K⁺ deprivation (Figure 4A). As expected, caspase-3 was strongly activated at later time points. Specific inhibitors for these caspases were then tested on CGN for their ability to reverse CBP degradation (Figure 4B). We found that z-DEVD-fmk, a caspase-3 inhibitor, partly reversed the decrease in CBP levels, whereas the caspase-6 inhibitor z-VEID-fmk was much more efficient (Figure 4B), suggesting that caspase-6 could be responsible for CBP degradation. To investigate this hypothesis further, we tested the effects of recombinant caspase-3 and -6 on CBP cleavage in an in vitro assay (Figure 4C). We used human double minute 1 (HDM1), the human MDM2 homolog, as a control for caspase-3 cleavage (Chen et al., 1997). Recombinant [³⁵S]methionine-labeled CBP and HDM1 were synthesized in a rabbit reticulocyte lysate and incubated with increasing amounts of caspase-3. Caspase-3 was not able to cleave CBP (Figure 4C, left panel), even when 10 U of recombinant protein were used (data not shown). In the same assay, caspase-3 cleaved HDM1 in a dose-dependent manner, giving an expected fragment migrating at 68 kDa (Figure 4C, right panel, HDM1-I) (Chen et al., 1997) and being reversed by both z-DEVD-fmk and Ac-VAD-CHO. Recombinant caspase-6 cleaved [³⁵S]methionine-labeled CBP in a dose-dependent manner, producing two cleavage fragments migrating at apparent mol. wts of 175 and 130 kDa, respectively (Figure 4C, left panel). This cleavage could be reversed by both z-VEID-fmk and Ac-VAD-CHO. It is noteworthy that cleavage fragments observed by western blot in CGNs undergoing apoptosis (CBP-I, Figure 3B, and CBP-II, data not shown) and those obtained with the caspase cleavage assay (Figure 4C, left panel) seem to migrate at identical molecular weights, suggesting that CBP could indeed be targeted by endogenous caspase-6 in CGNs. Collectively, these data suggest that CBP is preferentially cleaved by caspase-6 in the early events of CGN apoptosis.

Acetylated histone and CBP levels decrease in two other models of neuronal cell death

To determine whether histone deacetylation and CBP degradation were common traits of neurodegeneration, we investigated their levels in an in vitro model of embryonic cortical neuron cell death (Mbebi et al., 2002) and in an in vivo model of ALS (Ripps et al., 1995; Gonzalez de Aguilar et al., 2000).

Embryonic cortical neuron death was induced by activation of the APP signaling pathway with an antibody directed against the extracellular domain of APP (APP-Ab) (Mbebi et al., 2002). Western blot analyses revealed that Ac-H3 and CBP levels decreased during cortical neuron death (Figure 5A). CBP and Ac-H3 levels were quantified relative to actin levels and are represented as histograms (right panel). These results were confirmed by immunohistochemistry (Figure 5B): Ac-H3 and CBP immunoreactivities were easily detected in healthy neurons displaying large nuclei (Ct) and strongly decreased in dying neurons treated with APP-Ab. The use of z-VEID-fmk further suggested the implication of caspase-6 in this model as it was able to reverse CBP degradation, as did z-VAD-fmk (Figure 5C). Moreover, both z-VEID-fmk and z-VAD-fmk were also able to prevent cortical neuronal death induced by APP-Ab (Figure 5D). Altogether, these results show that histone acetylation and CBP levels also decreased in this cell death model, in a caspase-6-dependent manner.

The in vivo mice model of ALS, G86R mutant SOD-1 mice (Ripps et al., 1995; Gonzalez de Aguilar et al., 2000), represents a progressive neurodegenerative disorder inducing motoneuron death (Hand and Rouleau, 2002). CBP and H3 histone acetylation levels were monitored by western blots in the lumbar spinal cord of 3.5-month-old G86R mice, a time point at which motoneurons die (Figure 6A, left panel). Wild-type (WT) littermates were used as control. Quantification is shown on the right. We found that the CBP protein level decreased >70% in G86R compared with WT mice, but failed to detect any changes in total Ac-H3 amounts. Nevertheless, immunohistochemistry performed on lumbar spinal cord slides revealed some differences (Figure 6B). Nuclei of motoneurons were easily detected, as they appeared larger than those of other cells such as glial cells and interneurons (Casanovas et al., 2001). We found that Ac-H3 immunoreactivity was present in all cell types, but strongly decreased in motoneuron nuclei of G86R mice compared with WT (Figure 6B). This could explain why the decrease between G86R and WT mice could not be detected by western blot since the global amount of Ac-H3 was not that affected relative to the whole tissue. In contrast, CBP immunoreactivity was detected mainly in motoneurons and strongly decreased in G86R mice (Figure 6B), confirming results readily obtained by western blot analyses.

Altogether, the results obtained in these different models suggest that the decrease in histone acetylation and CBP levels may be a common trait of neurodegeneration.

Fine-tuning of HAT activity is necessary for neuronal survival

Proteins involved in neuroprotection are often degraded at the onset of apoptosis. To understand whether CBP loss of function has an important role during neuronal apoptosis, we tested the ability of CBP overexpression to counteract LK-induced CGN apoptosis (Figure 7B). Typical photographs of the results obtained are shown in Figure 7A–C, and quantification is represented at the bottom of each panel (Figure 7B and 7C). In control conditions (p2N3T empty vector, Figure 7A), neuronal survival in the transfected neuronal population is evaluated at 74% in survival conditions (HK, left panel) and only 30% in apoptotic conditions (LK, right panel). We found that CBP overexpression significantly protected neurons from apoptosis: survival was increased (65%) compared with the control (30%), as depicted in Figure 7B (upper panel). Interestingly, overexpression of p300 produced effects similar to CBP (Figure 7B), suggesting that protection occurs through similar functions. We repeated this experiment with a mutant of CBP, in which the HAT domain was deleted (CBPΔHAT). The percentage of healthy nuclei obtained with CBPΔHAT was significantly reduced when compared with CBP (Figure 7B), suggesting that the neuroprotective effects induced by CBP overexpression were indeed mediated in part through its HAT activity.

However, in survival conditions (HK), neurons transfected with CBP presented apoptotic nuclei, sometimes condensed but mainly fragmented (Figure 7C, upper panel). Quantitated results showed that CBP overexpression allowed only 33% survival in HK, which was significantly reduced compared with that obtained with the empty vector (74%) (Figure 7C, lower panel). This suggests that CBP overexpression is lethal for neurons maintained in survival conditions. p300 overexpression also induced cell death and chromatin fragmentation. We found that CBP overexpression-induced cell death was likely to be driven by its HAT function, as CBPΔHAT was without effect on neuronal survival (72% compared with 74% obtained with the empty vector).

Taken together, these results argue for a major role for CBP in neuroprotection by finely tuning specific acetylations through its HAT function.

Peptides, recombinant proteins and antibodies

Protease inhibitors z-DEVD-fmk, z-VEID-fmk, z-VAD-FMK, Ac-VAD-CHO, ALLN, PSI and recombinant caspase-6 were purchased from Calbiochem (La Jolla, CA), and recombinant caspase-1 from Alexis Biochemical (San Diego, CA). Anti-CBP Ct, anti-Ac-H3, anti-Ac-H4 antibodies, horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG and recombinant caspase-3 were obtained from Upstate Biotechnology (Lake Placid, NY). Anti-p300, anti-TAF II p250 and anti-PCAF antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-β-actin was a kind gift from Dr D.Aunis (Strasbourg, France). HRP-conjugated secondary goat anti-rabbit IgG was purchased from Pierce (Rockford, IL).

Neuronal primary cultures

CGNs from 7-day-old mice (fvb strain) were cultured as previously described (Boutillier et al., 1999). Experiments were performed after 6 days in vitro (DIV), except transfection experiments (3 DIV). HK is defined medium with final [K⁺] = 30 mM and LK, [K⁺] = 5 mM.

Cortical neurons from fvb mice embryos (E18) were cultured as described previously (Mbebi et al., 2002). Experiments were performed after 4 DIV. To induce cell death, the APP-Ab antibody was used at a final concentration of 8 µg/ml.

SOD G86R mice

Transgenic mice with the missense mutation G86R (human G85R equivalent) in SOD-1 (Ripps et al., 1995), showing a very early stage of hind leg paralysis (between 100 and 105 days old), were used as an in vivo model of ALS. WT littermates were used as control. All experiments were carried out according to current EC regulations. For immunocytochemistry, animals were anesthetized intraperitoneally with urethane (2 mg/g of body weight) and perfused with 4% paraformaldehyde in phosphate-buffered saline (PBS). Lumbar spinal cords were dissected and fixed for 24 h at 4°C. Tissues were then dehydrated in graded ethanol and embedded in araldite. Semi-thin sections were cut and collected on gelatin-coated slides. After araldite removal, sections were stained by immunofluorescence methods. For western blot analysis, animals were killed by decapitation and lumbar spinal cords were rapidly dissected and stored in liquid nitrogen. After tissue homogenization, lysates were centrifuged, protein determination was performed on supernatants and 70 µg of total proteins were analyzed.

Cellular histone extraction

Total cellular histones were acid extracted according to Nakajima et al. (1998) and 10 µg were separated on 13% polyacrylamide gels. Gels were either stained with Coomassie brilliant blue R-250 to visualize total amounts of cellular histones or processed further by western blot analyses.

Western blot analysis

Western blots were performed as previously described (Boutillier et al., 1999) with typically 50–70 µg of total cell extracts run on 7, 10 or 13% SDS–acrylamide gels for CBP, actin or histone detection, respectively. Specific bands were detected by enhanced chemiluminescence (ECL; Amersham).

HAT assay

HAT assay was performed using the kit purchased from Upstate Biotechnology either on CBP/p300-immunoprecipitated extracts or not, according to the manufacturer’s procedures. HAT activity was read in an LS6500 Multi-Purpose Scintillation Counter (Beckman Coulter, Fullerton, CA).

Morphological evaluation of apoptosis

Cells were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature and incubated for 30 min with the Hoechst dye 33342 (1 µg/ml). Stained nuclei were scored by blind analysis, those presenting condensed or fragmented chromatin being counted as apoptotic.

Immunocytochemistry

Immunocytochemistry was performed as previously described. Cy3 and Hoechst staining were visualized with a Leitz microscope. Photographs were taken with Nikon Digital Camera Dxm1200.

RT–PCR analysis

Total RNA was extracted from neurons by the guanidinium thiocyanate–phenol–chloroform method. A 2 µg aliquot was reverse-transcribed using 200 U of MMLV reverse transcriptase, and PCR analysis was performed (Promega, Madison, WI) as previously described (Boutillier et al., 1999). Cycle parameters were 1 min at 94°C, 1 min at 60°C and 1 min at 72°C for 22 cycles. Ethidium bromide-stained PCR products were separated on 2% agarose gels, visualized and quantified using a Gel Doc 2000 (Bio-Rad, Hercules, CA).

Mouse CBP primers: reverse: 5′-TCAAGTTTGGTGGCTGTTGA-3′, forward: 5′-TTGCCTCCCTTTGAAAAATG-3.

Caspase activity measurements

Caspase activities were determined by measuring the release of fluorescent cleavage products from the caspase tetrapeptide substrates as described previously (Boutillier et al., 2000). The caspase-specific substrates Ac-DEVD-AMC and Ac-VEID-AMC were purchased from Sigma (St Louis, MO). Fluorescence was monitored at excitation 360 nm, emission 465 nm, using a Perkin-Elmer HTS 7000 Bioassay reader (Foster City, CA). A standard curve of fluorescence versus free AMC was used to calculate the amount of substrate cleaved from fluorescent units.

In vitro synthesis of recombinant proteins and cleavage assay

In vitro synthesis of recombinant [³⁵S]methionine-labeled CBP and HDM1 was performed using the TNT® Coupled Reticulocyte Lysate Systems kit (L4611) purchased from Promega, according to the manufacturer’s procedures. [³⁵S]Methionine was purchased from Perkin Elmer (Boston, MA). The in vitro cleavage assay was performed on 2 µl of the recombinant protein mixture. Neo-synthesized proteins were incubated for 20 min at 37°C with recombinant caspase-1, -3 or -6 (1, 2.5, 5 or 10 U) with or without caspase inhibitor (z-DEVD-FMK, z-VEID-FMK, Ac-VAD-CHO, 50 µM). Reactions were stopped with 5× Laemmli buffer, boiled, and products were analyzed by 7% SDS–PAGE.

Colorimetric MTT assay

After 4 DIV, cortical neurons cultured in 48-well plates were treated with an anti-APP antibody as noted, in the absence or presence of z-VAD-fmk (50 µM) or z-VEID-fmk (100 µM). After 36 h, MTT assays were performed as described in Boutillier et al. (1999).

Co-transfections

Neuronal gene transfections were performed as reported previously (Boutillier et al., 2000). Neurons were grown on glass coverslips and co-transfected at 3 DIV for 30 min with 1 µg/ml of expression vector: p2N3T, p2N3T-CBP, p2N3T-CBPΔHAT (generous gifts from Drs A.Harel-Bellan and S.Ait-Si-Ali, Villejuif, France) or pcDNA3-p300 (generous gift from Dr J.R.Lundblad, Portland, OR) and 0.2 µg of pEGFP vector. Plates were spun for 5 min at 1500 r.p.m. Apoptotic nuclei were revealed by Hoechst staining and transfected cells were visualized with a Leitz microscope using a fluorescein isothiocyanate (FITC) filter set.

Statistics

Data were expressed as the mean ± SEM. Statistical significance was determined by one-way ANOVA followed by Newman–Keuls’ multiple comparison test. Differences were considered as significant at P < 0.05.