The death of cholinergic neurons in the cerebral cortex and certain subcortical regions is linked to irreversible dementia relevant to AD (Alzheimer's disease). Although multiple studies have shown that expression of a FAD (familial AD)-linked APP (amyloid β precursor protein) or a PS (presenilin) mutant, but not that of wild-type APP or PS, induced neuronal death by activating intracellular death signals, it remains to be addressed how these signals are interrelated and what the key molecule involved in this process is. In the present study, we show that the PS1-mediated (or possibly the PS2-mediated) signal is essential for the APP-mediated death in a γ-secretase-independent manner and vice versa. MOCA (modifier of cell adhesion), which was originally identified as being a PS- and Rac1-binding protein, is a common downstream constituent of these neuronal death signals. Detailed molecular analysis indicates that MOCA is a key molecule of the AD-relevant neuronal death signals that links the PS-mediated death signal with the APP-mediated death signal at a point between Rac1 [or Cdc42 (cell division cycle 42)] and ASK1 (apoptosis signal-regulating kinase 1).
Neuronal dysfunction and death relevant to AD (Alzheimer's disease) are pathological abnormalities of AD that are directly linked to dementia .
Aβ (amyloid β) levels are generally up-regulated in AD brains. Sequential cleavage by β- and γ-secretase of APP (Aβ precursor protein) produces Aβ that aggregates into clumps called plaques . PS (presenilin) 1 is an essential subunit of γ-secretase . A FAD (familial AD)-linked mutation in APP, PS1 and PS2 has been shown to increase the production of Aβ in cultured cells and transgenic mice [1,3].
There is accumulating evidence which suggests that the AD-related increase in the Aβ oligomers may cause neuronal dysfunction in the central nervous system . In contrast, the mechanism underlying AD-linked neuronal death has been addressed insufficiently because a very limited number of in vitro and in vivo models for neuronal death relevant to AD have been established [1,3]. Transgenic rodents overexpressing a FAD-linked gene are unable to mimic AD-relevant neuronal death . As a consequence, although it has been generally hypothesized that the long-term incubation of neurons with increased levels of Aβ may result in the onset and the progression of neuronal death in human AD cases (the Aβ cascade hypothesis), there is no direct evidence supporting this notion. Since neuronal death plays a major role in inducing irreversible dementia in human AD cases, it is worthwhile addressing the mechanism underlying AD-relevant neuronal death using currently available in vitro death models.
Multiple groups have demonstrated that expression of a FAD-linked mutant APP and PS gene, but not expression of wt (wild-type) APP or PS, causes death in cultured neuronal cells [5–8,9] (for a review see [10,11]) and primary neurons [12–16], by activating neuronal death signals [6–9,12–16]. Although these in vitro death models may serve as a powerful tool to investigate the mechanism of neuronal death relevant to AD, a limited number of investigations have addressed it.
It has been reported previously that low expression of V642I-APP, a London-type FAD-linked mutant of APP, and K595N/M596L-APP, the Swedish-type FAD-linked mutant of APP, induced death in multiple cultured neuronal cells and primary cortical neurons by activating a signal pathway consisting of a heterotrimeric G protein Go, Rac1 [or Cdc42 (cell division cycle 42)], ASK1 (apoptosis signal-regulating kinase 1), JNK (c-Jun N-terminal kinase), NADPH oxidase and caspases [8,9,17,18], whereas low expression of wtAPP did not induce neuronal cell death. FAD-linked APP mutant-induced death occurs in a manner independent of Aβ production and the so-called APP intracellular domain as a transcription factor [6,8]. As APP structurally resembles a single transmembrane receptor , it has been hypothesized previously that APP is a receptor for an unknown death ligand(s) and that a FAD-linked mutation caused the constitutive activation of the death signal pathway without binding of the relevant ligand. Consistent with this hypothesis, we found that TGFβ (transforming growth factor β) 2, which is generally up-regulated in AD brains [20,21], could cause neuronal cell death by binding to APP and activating the same signal transduction pathway that is activated by expression of V642I-APP using cultured neuronal cells and primary cortical neurons [22,23]. It has been shown previously that death in primary cortical neurons, in vitro induced by superphysiological concentrations of Aβ, was in part mediated by autocrinally secreted TGFβ2 . Even in non-neuronal cells, APP may be a putative receptor for certain unknown ligands and may mediate some biological activities .
As suggested by the fact that a number of PS-binding proteins have been identified , it is highly likely that PSs may have functions unrelated to γ-secretase. The low expression of a FAD-linked PS1 mutant with its mutation located in the C-terminal portion and a FAD-linked PS2 mutant similarly causes neuronal death, which is dependent on Go, NADPH oxidase and caspases [9–11]. These findings suggest that the PS mutant-triggered death signal pathway overlaps with the APP-mediated death signal pathway. However, it remains unknown how the signal is related to the APP-mediated death signal and what the key molecule underlying the integration of these death signals is.
MOCA (modifier of cell adhesion) was originally identified as being one of the PS-binding proteins, is exclusively expressed in neuronal tissues and testis , and is involved in cell adhesion and neurite growth . MOCA was also shown to be lost from the soluble fraction of AD brain . MOCA accumulated in neurons with neurofibrillary tangles  and enforced expression of MOCA decreased Aβ secretion by regulating APP degradation . These findings suggest that MOCA may be linked to the pathogenesis of AD.
In the present study, we show that the PS-mediated signal is needed for APP-mediated neuronal death in a γ-secretase-independent manner and vice versa. We have also found that MOCA is essential for the V642I-APP-, C401Y-PS1- and N141I-PS2-induced death, and that MOCA is located at a point of the death signal transduction pathway between Rac1 (or Cdc42) and ASK1. Taken together, MOCA is a key molecule of the AD-relevant neuronal death that links the PS1- (or PS2-) mediated death signal with the APP-mediated death signal at a point between Rac1 (or Cdc42) and ASK1.
Cell lines and genes
Neurohybrid F11 cells were as described previously . F11 cells are the hybrids of rat embryonic day 13 primary cultured neurons with mouse neuroblastoma NTG18 cells. Most rat genes were lost from F11 cells. Human wtAPP, V642I-APP, wtPS1, C410Y-PS1, M146L-PS1, wtPS2 and N141I-PS2 cDNAs in the pcDNA3 vector (Invitrogen) were as described previously [6,8,10,11]. wtPS1 and wtPS2 cDNA were amplified by PCR with KOD DNA polymerase (Toyobo), and were subcloned into the pEF1/MycHis vector (Invitrogen). Mouse MOCA cDNA was a gift from Dr Hideo Kimura (National Institute of Neuroscience, Tokyo, Japan) [27,32]. For the construction of an N-terminally FLAG-tagged MOCA plasmid (pCIneo/FLAG–MOCA), a DNA fragment encoding FLAG, generated by annealing of sense (5′-CGACGCGTACCATGGATTACAAGGACGACGATGACAAGGAGTCGACACGC-3′) and antisense (5′-GCGTGTCGACTCCTTGTCATCGTCGTCCTTGTAATCCATGGTACGCGTCG-3′) oligonucleotides, was inserted into the 5′-end of mouse MOCA cDNA at the MluI/SalI site. For the construction of a N-terminally GST (glutathione transferase)-fused MOCA plasmid, the GST-encoding cDNA was inserted into the same site of pCIneo-MOCA. To delete the Lys1060–Glu1393 region from the MOCA cDNA, we PCR-amplified and self-ligated the deleted cDNA with a sense primer (5′-TTCCCACAGGCCGTTGCTATGCAGCACCCC-3′) and an antisense primer (5′-CCTCTTGGCTGAAGTAATGATTTCTAGCTG-3′) using the pCIneo/FLAG–MOCA plasmid as the template (KOD-Plus-Mutagenesis Kit, Toyobo). To change Thr1285 to an alanine in the MOCA cDNA, we PCR-amplified and self-ligated the mutant cDNA using the pCIneo/FLAG–MOCA plasmid as the template with a sense primer (5′-GCTGAGTGGCAGCGGAAAGAGGGACTGTGTAG-3′) and an antisense primer (5′-TTGGGATGGGTAGTGTAGGAACTCCCGTAATG-3′) using the KOD-Plus-Mutagenesis Kit. ca (constitutive active) JNK, caASK1, caRac1, caCdc42, HA (haemagglutinin)-tagged dn (dominant-negative) ASK1 and dnJNK were as described previously [17,18].
Recombinant proteins and antibodies
Synthetic Humanin was purchased from Peptide Institute. Recombinant human TGFβ2 was purchased from R&D Systems. Antibodies against the indicated peptides and proteins used in the present study were: anti-APP and anti-FLAG epitope (M2) (Sigma); anti-Myc (Invitrogen); anti-APP (22C11) and anti-PS1 (clone hPS1-NT produced by rat with the GST-fused N-terminus of human PS1 residues 21–80) (Chemicon); anti-PS2 (polyclonal antibody against the residues surrounding amino acid 330 of human PS2) (Cell Signaling Technology); anti-HA (Roche Diagnostics); and anti-PS1 (named H70), anti-GST (named B-14) and anti-Cdc42 (Santa Cruz Biotechnology). The anti-APP antibody (22C11) cross-reactively recognizes APLP2 (Aβ precursor-like protein 2), whereas the anti-APP antibody (A8717) only recognizes APP.
siRNA (short interfering RNA)-encoding plasmids
Transfection procedure and cell death assay
The transient transfection procedure was modified from one described previously [8,22]. Briefly, F11 cells, seeded at 7×104/well in six-well plates in HF (Ham's F12) medium plus 18% FBS (fetal bovine serum; HF-18%) for 12–16 h, were transfected with the indicated vectors for 3 h in the absence of serum and were then incubated with HF-18% for 2 h. At 5 h from the onset of transfection, the culture medium was replaced by HF-10%. At 24 h from the onset of transfection, they were replaced by HF plus N2 supplement (Gibco). In some experiments, an indicated inhibitor or TGFβ2 was added into the medium at this time point. At 72 h from transfection, representative microscopic views were taken and cell death was assessed using the Trypan Blue exclusion assay [8,22]. Dying cells become round, are detached from dishes and disappear from microscopical views.
Primary cortical neurons and cell death assays
Mouse primary cortical neurons were seeded in poly-L-lysine-coated six-well plates (Sumitomo Bakelite) at 5×105 cells/well in Neuron medium (Sumitomo Bakelite) . After incubation for 3 days (39°C), the culture medium was replaced with DMEM (Dulbecco's modified Eagle's medium) containing N2 supplement. After 4 days of incubation, transfection was performed with Lipofectamine™ 2000 (Invitrogen) according to the manufacturer's instructions. At 48 h after transfection, the cells were fixed with 4% paraformaldehyde for 30 min, followed by incubation at room temperature (22°C) with antibody against APP (A8717) for 2 h. The cells were stained with Texas Red-conjugated anti-rabbit IgG antibody (Jackson ImmunoResearch Laboratory). The nuclei were stained with Hoechst 33258 (Sigma).
F11 or COS7 cells co-transfected with the indicated vectors were harvested at 72 h from the onset of transfection for pull-down analysis with 15 μl of a 1:1 slurry of glutathione–Sepharose and immunoblot analysis with antibodies against GST, PS1, PS2 and FLAG.
Cell lysates (20 μg/lane) or pulled-down precipitates were subject to SDS/PAGE, and fractionated proteins were transferred on to PVDF membranes. Visualization of the immunoreactive bands was performed using ECL (enhanced chemiluminescence; Amersham Pharmacia Biotech).
All cell viability experiments were done with n=3. All values in the experiments are means±S.D. Statistical analyses were carried out by Student's t test. P<0.05 was considered as significant.
Co-expression of wtAPP and wtPS1 (or wtPS2) induces neuronal death
In a previous study using a neuronal cell system in which expression of a gene of interest in an ecdysone (an insect steroid hormone)-inducible plasmid is precisely controlled by ecdysone doses , we showed that low to moderate overexpression of wtAPP, wtPS1 or wtPS2 alone did not induce death in F11 neurohybrid cells [8–11]. This finding was confirmed using primary cortical neurons and an adenovirus-mediated expression system . Consistent with these findings, by transiently transfecting 0.5 μg of a pcDNA3-derived vector in each well on the six-well dishes and harvesting the cells for the Trypan Blue exclusion assay, we found that expression of wtAPP, wtPS1 or wtPS2 alone did not induce death in F11 neurohybrid cells (Figure 1). The co-expression of wtAPP and wtPS1 (or wtPS2) induced death in F11 cells, but the co-expression of wtPS1 and wtPS2 did not (Figure 1). These results suggest that both the APP-mediated and the PS1- (or the PS2-) mediated signals co-operate to induce neuronal death in F11 cells.
wtAPP and wtPS co-operate to induce neuronal death
Endogenous PS1 is indispensable for APP-mediated death and endogenous APP is indispensable for C410Y-PS1- (or N141I-PS2-) induced death
Using the ecdysone-inducible expression system, we showed in our previous studies that low to moderate expression of a FAD-linked mutant of APP, PS1, or PS2 induced the death in F11 cells [8–11]. To test the hypothesis that both the APP- and the PS1- (or PS2-) mediated signals are indispensable for the induction of neuronal death, we generated plasmid-based siRNAs for PS1, PS2 and nicastrin (Supplementary Figure S1 at http://www.BiochemJ.org/bj/442/bj4420413add.htm) [22,33]. PS1 (or PS2) and nicastrin are essential components of γ-secretase . The siRNA-mediated reduction of endogenous PS1 expression inhibited the V642I-APP-induced neuronal death (Figures 2A and 2B). In contrast, the siRNA-mediated reduction of PS2 or nicastrin did not inhibit the V642I-APP-induced death. The co-expression of siRNAs did not affect the expression of wtAPP (Figure 2C). These results suggest that PS1 is essential for the V642I-APP-induced death in a γ-secretase-independent manner.
V642I-APP-induced death was inhibited by the siRNA-mediated reduction of PS1 expression in F11 cells
Quantitative real-time PCR analysis suggested that the level of PS2 mRNA expression was lower than that of PS1 mRNA expression in F11 cells (Supplementary Figure S2 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). We therefore assumed that the siRNA-mediated reduction of PS2 expression did not inhibit the V642I-APP-induced death in F11 cells (Figure 2), because PS1 and PS2 were functionally redundant and the endogenous expression of PS2 was less than PS1 in F11 cells. In support for this hypothesis, we found that the overexpression of wtPS2 abolished the PS1 siRNA-induced inhibition of the V642I-APP-induced death (Supplementary Figure S3 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). The result indicates that PS2 was able to replace PS1 from the standpoint of its co-operation with APP to induce death in F11 cells.
We confirmed the effects of these siRNAs using another APP-mediated neuronal death model. As shown previously , the TGFβ2-induced death in F11 cells, which overexpress wtAPP, occurs via the death signal pathway identical with the V642I-APP-induced death signal pathway. The siRNA-mediated reduction of PS1 expression, but not that of PS2 or nicastrin expression, inhibited the death in F11 cells expressing wtAPP that was induced by the addition of TGFβ2 (Figure 3 and Supplementary Figure S4 at http://www.BiochemJ.org/bj/442/bj4420413add.htm).
TGFβ2-induced death via APP was inhibited by the siRNA-mediated reduction of PS1 expression
Conversely, we investigated whether the reduction of endogenous APP expression inhibited the C410Y-PS1- (and N141I-PS2-) induced death. We constructed two potential siRNAs for APP. We found that APP siRNA#1 was unable to reduce APP expression, whereas APP siRNA #2 was able to reduce expression (Supplementary Figure S5 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). As shown in Figure 4, the co-expression of APP siRNA #2, but not APP siRNA #1, inhibited C410Y-PS1- (and N141I-PS-) induced death (Figure 4A and Supplementary Figure S6 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). The reduction of APP expression by itself significantly reduced cell viability (compare the vector siNC and siAPP#2 columns of Figure 4A), possibly by disturbing the cell adhesion that is mediated by APP . Immunoblot analysis showed that the expression of the APP siRNA did not affect the expression of PSs (Figure 4B). Taken together, these results suggest that the endogenous-level APP-mediated signal is needed for the mutant PS-induced neuronal death and, reciprocally, that the endogenous-level PS1- (or PS2-) mediated signal is needed for APP-mediated neuronal death.
C410Y-PS1 and N141I-PS2-induced deaths were inhibited by the siRNA-mediated reduction of APP expression
MOCA, which interacts with PS, is involved in V642I-APP-induced neuronal death
MOCA was originally cloned as a PS-binding protein [26–28] and has been shown to be a putative GEF (guanine-nucleotide exchange factor) for the small G-protein Rac1 . Rac1 is an essential component of the APP-mediated death signal pathway [17,18,22]. To characterize MOCA from the standpoint of the PS involvement in the APP-mediated death signal pathway, we examined whether MOCA was involved in APP-mediated neuronal death. To this end, we constructed two plasmid-based siRNAs for mouse MOCA (Supplementary Figure S7 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). Expression of each of the siRNAs (siMOCA#1 and siMOCA#4) reduced the level of simultaneously co-expressed MOCA–FLAG at the protein level.
As shown in Figure 5, the siRNA-mediated reduction of endogenous MOCA expression attenuated the V642I-APP-induced death in F11 cells (Figures 5A and Supplementary Figure S8 at http://www.BiochemJ.org/bj/442/bj4420413add.htm) without a reduction of V642I-APP expression (Figure 5B), whereas the transfection of a mock siRNA did not reduce V642I-APP-induced death (Figure 5A and Supplementary Figure S8).
MOCA was essential for V642I-APP-induced death in F11 cells and PCNs
We confirmed this result using more physiological neurons, mouse PCNs (primary cortical neurons). Since transfection efficiency was very low in PCNs (2–5%), we counted the percentage of apoptotic cells in the total cells that expressed transfected genes by monitoring cell fluorescence [APP immunofluorescence and EGFP (enhanced green fluorescent protein) as markers of V642I-APP and siRNA expression respectively]. Nuclei were stained with Hoechst 33258 to monitor apoptotic changes (chromosomal condensation and fragmentation; see Supplementary Figure S9 as an example of the assay at http://www.BiochemJ.org/bj/442/bj4420413add.htm). PCNs, which overexpressed V642I-APP, became apoptotic at rates of approximately 75% (Figure 5C, V642I-APP siNC column), whereas control PCNs, which were transfected with the vector, became apoptotic at rates of approximately 30–45% (Figure 5C, vector columns). Co-expression of PS1 siRNA or MOCA siRNA reduced the rate of V642I-APP-induced apoptosis in PCNs to approximately 35–45% (Figure 5C, V642I-APP siMOCA and siPS1 columns). We also found that PCNs which were co-transfected with both the V642I-APP-encoding vector and the PS1 siRNA (or the MOCA siRNA), but expressed only V642I-APP, became apoptotic at rates of 70–75% (Figure 5D). Taken together these results indicate that the knockdown of PS1 or MOCA expression inhibited V642I-APP-induced death in PCNs.
MOCA is essential for C410Y-PS1- and N141I-PS2-induced death
In addition to FAD-linked APP mutants, a FAD-linked PS1 with its mutation localized in the C-terminal portion and a FAD-linked PS2 mutant were shown to cause neuronal death, which was blocked by inhibitors of NADPH oxidase and caspases . On the basis of the fact that MOCA was originally identified as an interactor with PS1 , we hypothesized that MOCA was a downstream effector of the PS1- (or PS2-) induced death signal. In accordance, we found that the reduction of endogenous MOCA expression inhibited neuronal death that was induced by expression of C410Y-PS1 and N141I-PS2 (Figure 6A and Supplementary Figure S10 at http://www.BiochemJ.org/bj/442/bj4420413add.htm), without affecting the expression of PS (Figure 6B).
MOCA, a PS interactor, was essential for C410Y-PS1 or N141I-PS2-induced death
Pull-down analysis using glutathione–Sepharose showed that GST-tagged MOCA was co-precipitated with wtPS1 and wtPS2 (Figures 6C and 6D). Interestingly, the FAD-linked mutations in PS1 increased the levels of PS1 that was co-precipitated with GST–MOCA (Figure 6C, vector columns), whereas the N141I mutation in PS2 did not increase the level of PS2 that was co-precipitated with GST–MOCA (Figure 6D). The increase in the affinity of MOCA to PS1 mutants may contribute to the FAD-linked mutation-induced enhancement of the PS1-induced cell death signal.
MOCA is located at a point between Rac1 (or Cdc42) and ASK1 in the APP-mediated signal pathway
APP-mediated neuronal death has been shown to be mediated by Go, Rac1 (or Cdc42), ASK1, JNK, NADPH oxidsase and caspases [8,17,18,22]. We also found that expression of a ca form of Rac1, Cdc42, ASK1 or JNK induced the death in F11 cells [17,18]. To determine the point where the MOCA signal joins the APP-mediated signal pathway, we examined whether the siRNA-mediated reduction of MOCA expression attenuated ca Rac1-, Cdc42-, ASK1- and JNK-induced death in F11 cells. As shown in Figure 7(A) and Supplementary Figure S11 (at http://www.BiochemJ.org/bj/442/bj4420413add.htm), the reduction of MOCA expression attenuated ca Rac1- or Cdc42-induced neuronal death without affecting the expression of each protein (Figure 7B), whereas it did not attenuate the ca ASK1- or JNK1-induced neuronal death. These results indicate that MOCA is located in the APP-mediated death signal pathway at a point between Rac1 (Cdc42) and ASK1.
The siRNA-mediated reduction of MOCA expression inhibited caRac1- or caCdc42-induced death, whereas it did not inhibit caASK1- or caJNK-induced death
Overexpression of wtMOCA and MOCA without the putative GEF domain induced neuronal death
To confirm the involvement of MOCA in the APP-mediated cell death signal pathway, we examined whether the overexpression of MOCA induced neuronal cell death and whether the MOCA-induced death signal merged with the APP-mediated death signal. As shown in Figure 8(A) and Supplementary Figure S12 (at http://www.BiochemJ.org/bj/442/bj4420413add.htm), the overexpression of MOCA induced death in F11 cells. The MOCA-induced death was inhibited by co-incubation with SP600125 (a JNK inhibitor), apocynin (a NADPH oxidase inhibitor) and Humanin (a general inhibitor of AD-relevant neuronal cell death) , whereas it was not inhibited by L-NMMA (NG-monomethyl-L-arginine; a nitric oxide synthase inhibitor) (Supplementary Figure S12) without the reduction of V642I-APP and MOCA expression (Figure 8B). The MOCA-induced neuronal death was also inhibited by the co-expression of dn ASK1 or JNK-1 (Figures 8C and 8D, and Supplementary Figure S13 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). These results together indicate that the MOCA-induced neuronal death signal pathway overlaps with the V642I-APP-induced neuronal death signal pathway, which are mediated by Go, Rac1/Cdc42, ASK1, JNK, NADPH oxidase and caspases [8,9,17,18], and support the idea that MOCA may be a component of the V642I-APP-induced neuronal death pathway.
Overexpression of MOCA induced neuronal death via ASK1, JNK1 and NADPH oxidase
MOCA was shown to be a GEF for Rac1 . Pull-down analysis showed that GST-tagged Rac1 co-precipitated with MOCA (Figure 9A). It was therefore possible that MOCA may present its pro-apoptotic activity by increasing the amount of the GTP-bound form of Rac1 (an active form of Rac1). Contrary to the expectation, we found that the expression of a GEF domain-lacking mutant of MOCA (Δ1060–1393MOCA) or a putative GEF activity null point mutant of MOCA (T1285A-MOCA) induced death in F11 cells in a manner identical with neuronal death, induced by overexpression of wtMOCA (Figures 9B and 9C, and Supplementary Figure S14 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). These results suggest that MOCA does not present its proapoptotic activity by activating Rac1 or Cdc42 with its GEF activity.
The GEF domain of MOCA was not necessary for neuronal death that was induced by overexpression of MOCA
In the present study, we have shown that both the PS1- (or PS2-) and APP-mediated signals are essential for the induction of neuronal death relevant to AD, and that these signals merge at a point between Rac1/Cdc42 and ASK1 via MOCA as a common downstream effector (Supplementary Figure S15 at http://www.BiochemJ.org/bj/442/bj4420413add.htm). PS1 or PS2 is an essential component of γ-secretase. Pen2 (PS enhancer protein 2), Aph1 (anterior pharynx defective 1 homologue) and nicastrin are other essential components of γ-secretase . Because the siRNA-mediated reduction of nicatrin expression did not affect the V642I-APP-induced neuronal death or the TGFβ2-induced neuronal death via APP (Figures 2 and 3), the APP-mediated neuronal death appears to be independent from γ-secretase activity. In accordance with this, γ-secretase inhibitors did not inhibit the V642I-APP-induced neuronal death (results not shown).
In the present study, we showed that the knockdown of PS1, but not that of PS2, inhibited the APP-mediated neuronal death (Figures 2 and 3). Regarding this issue, we suggested that PS1 and PS2 were functionally redundant from the standpoint of their co-operation with APP to induce apoptosis (Supplementary Figure S3) and the endogenous expression of PS2 mRNA appeared to be less than that of PS1 mRNA in F11 cells (Supplementary Figure S2). These facts may explain why the siRNA-mediated reduction of PS2 expression did not significantly inhibit V642I-APP-induced death, whereas the siRNA-mediated reduction of PS1 did in F11 cells (Figures 2 and 3). However, other possibilities have not been completely excluded in the present study.
A FAD-linked mutation of PS increases the production of Aβ, ultimately leading to the appearance of AD-related senile plaques. In addition, FAD-linked mutants of PS1 or PS2 have been shown to cause abnormal Ca2+ homoeostasis that may contribute to the onset and progression of neuronal death . Besides these normal and abnormal functions of PS, PS may have other unknown functions because a number of PS1-binding proteins, unrelated to γ-secretase activity and Ca2+ regulation, have been identified . In the present study, we have shown that MOCA connects PSs with the death signal transduction pathway that is mediated by ASK1, JNK, NADPH oxidase and caspases (Supplementary Figure S15).
JNK and JKK1 (JNK kinase 1), a MAPKK (mitogen-activated protein kinase kinase) for JNK, are activated in human AD neurons [36,37]. Oxidative stress is implicated in the early phase of AD [36,38]. Oxidative stress-induced activation of JNK resulted in the increase of BACE1(β-amyloid precursor protein cleaving enzyme 1) expression, dependent on γ-secretase activity . Therefore it is likely that the activation of the JNK-mediated signal leads to increases in BACE1 expression and Aβ production, However, because γ-secretase did not appear to involve APP-mediated death (Figures 2 and 3) and the V642I-APP-induced death in F11 cells was not affected by a deletion of the two amino acids corresponding to the 41st and 42nd amino acids of Aβ42 (amyloid β peptide 1–42) , it is highly likely that the APP-mediated death via JNK is Aβ-independent.
In our previous studies, we showed that TGFβ2 is generally up-regulated in AD brains [20,21] and could cause neuronal death by binding to APP and activating the same signal transduction pathway that was induced by the expression of V642I-APP, using cultured neuronal cells and PCNs [22,24]. In the present study, we further showed that the TGFβ2-induced neuronal death via APP was dependent on the PS signal (Figures 2 and 3), which was mediated by MOCA, in F11 cells (Figure 5). Lee et al.  found that phospho-Smad, a central intracellular effector of TGFβ family ligands, was mislocalized in the cytoplasm, suggesting that the TGFβ signals via the authentic TGFβ receptor, which are thought to be neuroprotective, do not work in AD. Thus both the TGFβ2-induced neurotoxicity via APP and the loss of neuroprotective activity of TGFβ via the authentic TGFβ receptor may contribute to the onset of AD.
MOCA is predominantly expressed in neuronal cells and may play an important role in various neuronal functions. Although a previous study indicated that MOCA is a GEF for Rac1 , the present study shows that GEF function is not essential for the MOCA-induced neuronal death (Figure 9B). As the siRNA-mediated reduction of MOCA expression inhibited ca Rac1-induced neuronal death (Figure 7), it is highly likely that MOCA is a downstream effector of Rac1 (or Cdc42) in the APP-mediated neuronal death pathway (Supplementary Figure S15).
The expression of MOCA has been shown to be up-regulated in neurons that bear neurofibrillary tangles in AD brains . As the overexpression of MOCA by itself induces neuronal death (Figure 9), this finding suggests that the increase of MOCA expression may contribute to the onset and the progression of AD-relevant neuronal death, especially neuronal death in some sporadic AD cases.
In contrast with a FAD-linked PS1 mutant with its mutation located in the C-terminal portion, the PS1 mutant with its mutation located in the N-terminal portion induces non-apoptotic neuronal death by activating an unknown pertussis toxin-sensitive protein and nitric oxide synthase . Because MOCA also binds to a PS1 mutant with its mutation located in the N-terminal portion (M146L-PS1; Figure 6D), it is possible that MOCA is also involved in the neuronal death pathway, triggered by PS1 mutant with its mutation located in the N-terminal portion. If this is the case, the underlying mechanism should be more complex.
Aβ could cause death of neuronal cells in vitro [40,41]. However, superphysiological concentrations of Aβ are required to induce neuronal death in vitro. Consequently, it remains unknown whether these death models are relevant to AD. The Aβ-induced death in PCNs is in part mediated by autocrinally secreted TGFβ2 . We found that p75 neurotrophin receptor (p75NTR) behaves as a receptor for Aβ and that the binding of Aβ to p75NTR induces neuronal death by activating Go, Rac1 (or Cdc42), ASK1, JNK, NADPH oxidase and caspases . On the basis of the fact that all of the components in the Aβ-induced neuronal death signal pathway via p75NTR are shared by the APP-mediated death signal pathway, it is highly likely that MOCA is also involved in Aβ-induced neuronal death via p75NTR.
It is generally hypothesized that increased levels of Aβs cause AD in sporadic AD and FAD . In support of this, it has been experimentally demonstrated that in rodent models, the oligomeric forms of soluble Aβ cause neuronal dysfunction, which leads to the progression of the memory impairment [3,4]. It has been also assumed in the Aβ cascade hypothesis that the long-term neuronal dysfunction, caused by increased levels of Aβs, may result in the onset and the progression of neuronal death. However, because there are no good rodent models that mimic human AD-relevant neuronal death , it remains speculative how prominent neuronal death, which is invariably observed in brains of human AD cases, is induced in vivo. Relying on the Aβ hypothesis, numerous clinical trials lowering brain Aβ burden are being performed. Reported results from a couple of clinical studies have failed in showing that this hypothesis is correct [42,43], although multiple clinical studies, currently being performed, will provide a clearer answer to this issue in the near future.
Multiple studies have indicated that FAD-linked APP and PS genes cause neuronal death by activating intracellular death signal pathways [5–16]. Such in vitro death seems to occur independently of Aβ [6,8]. These results suggest that the Aβ-independent mechanism may contribute to AD-related neuronal death.
In the present study, using in vitro death models, we have shown that MOCA is a unifying component of the APP- and PS-mediated death signals, and that MOCA is located at the point where both signals merge. The detailed examination of MOCA function and expression would provide a clue to the better understanding of the mechanism underlying Aβ-independent neuronal death relevant to AD.
Aβ precursor-like protein 2
Aβ precursor protein
apoptosis signal-regulating kinase 1
β-amyloid precursor protein cleaving enzyme 1
cell division cycle 42
c-Jun N-terminal kinase
modifier of cell adhesion
primary cortical neuron
short interfering RNA
transforming growth factor β
Nobuyuki Tachi and Yuichi Hashimoto performed the experiments. Yuichi Hashimoto and Masaaki Matsuoka designed the experiments. Masaaki Matsuoka conceived of and supervised the study. Masaaki Matsuoka wrote the paper.
We thank Dr Ikuo Nishimoto (KEIO University School of Medicine, Tokyo, Japan); Ms. Takako Hiraki, for essential assistance; and Dr Hideo Kimura (National Institute of Neuroscience, Tokyo, Japan) for providing us with the MOCA cDNA.
This work was supported by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO), the Grant-in-Aid for Scientific Research (B) [grant number 23390059 (to M.M.)], (C) [grant number 225902417 (to Y.H.)], and the Mitsui Life Social Welfare Foundation.
These authors contributed equally to this work.