Molecular pathway connecting amyloid and mitochondrial pathology in Alzheimer's disease CHCHD6/APP

The mechanistic relationship between amyloid β precursor protein (APP) processing and mitochondrial dysfunction in Alzheimer's disease (AD) has long been unstudied
。 Recently, Acta Neuropathol published a research paper entitled "A CHCHD6–APP axis connects amyloid and mitochondrial pathology in Alzheimer's disease", in which the authors' team found that CHCHD6, the core protein of mammalian mitochondrial contact sites and crest tissue systems (coil-coil-helix-coil- helix - helix domain containing 6) connects AD features
by reducing CHCHD6 and improving APP processing through a circular feedback loop mechanism.

In cellular and animal AD models, as well as in human AD brains, APP intracellular domain fragments inhibit their transcription
by binding to the CHCHD6 promoter.
CHCHD6 is combined with the APP and is stable
.
The decrease of CHCHD6 enhanced the accumulation of APP on mitochondria-associated ER membranes, accelerated the processing of APP, and induced mitochondrial dysfunction and neuronal cholesterol accumulation, promoting amyloid lesions
.
Compensation for CHCHD6 deletion in AD mouse models can reduce AD-related neuropathological and cognitive impairment
.
Therefore, the authors believe that CHCHD6 connects APP processing in AD with mitochondrial dysfunction, which provides a potential new therapeutic target
for patients.

Reduction of CHCHD6 in AD models

Associated with the dissolution of the MICOS core complex

In mammals, CHCHD6 forms a MICOS (mitochondrial contact site and cristae organizing system) core component of about 700 kDa with mitofilin and CHCHD3 complexes, which control the integrity of mitochondrial membrane structure and biogenesis
。 The authors' team first detected changes in the core components of MICOS in AD cell models by conventional western blotting using antibodies against CHCHD6, mitofilin, or CHCHD3 by non-denaturing polyacrylamide gel electrophoresis (BN-PAGE
).
The results showed that CHCHD6 was significantly reduced in mitochondrial MICOS isolated from wild-type Neuro2a cells and Neuro2a cells that stably expressed wild-type APP or AD associated APP Swedish mutations (Figure 1a).

WB analysis of total protein lysate also confirmed selective loss of CHCHD6 (mitofilin and CHCHD3 unaffected, Figure 1b).

Real-time quantitative PCR results also showed an approximately 70% reduction in mRNA for CHCHD6 in APPwt and APPswe cells compared to the control group (Figure 1c).

In contrast, there was no significant difference in mRNA levels for other components of the MICOS complex, including mitofilin, CHCHD3, MINOS1, APOO, APOOL, CHCHD10, and QIL1 (Figure 1c).

Taken together, these results suggest that the decrease in CHCHD6 may play an important role
in the solubilization of the MICOS core complex in AD.

To see if the composition of the MICOS core complex also underwent similar changes in AD model mice and human AD tissues, the authors' team started with APPNL-G-F and APPNL-F knock-in (KI) mice
.
These mice carry the humanized mouse APP gene driven by their endogenous promoter and contain human AD mutations that are able to develop amyloid plaques and develop neurocognitive deficits
.
The APPNL−G−F KI structure contains a humanized Aβ region and three pathogenic mutations (Swedish "NL", Iberian "F" and Arctic mutation "G"), while the APPNL−F KI structure has only NL and F mutations
.
Both mice expressed APP at wild-type levels, which avoided the illusion that APP overexpression might introduce and also produced a high amount of pathogenic Aβ
.

In general, three-mutant APPNL−G−F KI mice were able to rapidly develop AD pathology and cognitive impairment, while APPNL−F KI mice showed slower chronic disease progression
.
The authors' team isolated hippocampal mitochondria from APP KI or WT mice at 3, 6, and 9 months of age and subjected their protein extractables to western blotting
of BN-PAGE and MICOS core proteins.
The results showed that levels of CHCHD6 on hippocampal MICOS were age-dependent in AD mice (this phenomenon was not present in WT mice) and were most severe in APPNL-G-F KI MICE with rapid disease progression (Figure 1d).

While CHCHD6 is the only MICOS core complex protein that downregulated with age in APP KI, levels of mitofilin and CHCHD3 remained similar in APP KI and WT mice of all ages (Figure 1e).

The decline in CHCHD6 in APPNL-G-F KI mice with rapid disease progression occurred earlier and was also larger than in APPNL-F KI mice with chronic disease development (Figure 1e).

CHCHD6 mRNA was also significantly reduced in the hippocampus of APP KI mice (Figure 1f).

In addition, a reduction in CHCHD6 was also found in postmortem hippocampal testing in AD patients (Figure 1g), and mRNA levels of mitofilin and CHCHD3 did not change (Figure 1g).

Next, the authors' team observed hippocampal CHCHD6 expression in brain tissues of AD mice and human AD patients by immunofluorescence and observed an age-dependent decrease in CHCHD6 expression in APPNL−G−F mice compared to age-matched WT mice (Figures 2a and b).

Importantly, a decrease in CHCHD6 immune density is accompanied by an increase in Aβ plaque accumulation (Figures 2a and b).

CHCHD6 is mainly expressed in NeuN+ neurons, and its expression decreases when the number of NeuN+ cells in the hippocampal region of WT and AD mice remains unchanged (Figures 2c and e).

Similarly, the intensity of CHCHD6 in NeuN+ neurons in the hippocampus and cortex of AD patients was significantly reduced compared to normal subjects (Figures 2d and f).

The expression of CHCHD6 in Iba1+ microglia and GFAP+ astrocytes was significantly lower, indicating that the expression of CHCHD6 was mainly enriched in neurons
.
Figure 1: CHCHD6 selectively decreased at transcription level in the AD model

CHCHD6 and APP are physically stable to each other,

The absence of CHCHD6 will cause abnormal processing of the APP

The authors' team exposed neuroblastoma cells to toxic Aβ1-42 (APP cleavage products) without altering the protein or mRNA levels of CHCHD6, mitofilin, and CHCHD3, suggesting that amyloid may have affected MICOS integrity
in other aspects of biology.
To further investigate the molecular role of CHCHD6 in AD, the authors' team used computational analysis to identify the genes, pathways, phenotypes, and diseases
associated with CHCHD6 at the genetic, functional, phenotypic, and disease levels.
Prioritizing biomedical entities based on the previously developed context-sensitive network ordering algorithm revealed many MICOS genes
significantly associated with CHCHD6.

The top predicted pathways for CHCHD6 are "protein metabolism," "mitochondrial protein delivery," "immune response," "lipid and lipoprotein metabolism," and "Alzheimer's disease
.
" These data strongly suggest that CHCHD6 may have a biological crossover with the APP, and thus may be involved in the pathophysiological process
of AD.

Since the app had localized to mitochondria, the authors' team next examined whether CHCHD6 and the app might interact
directly.
Immunoprecipitation APP results from mouse hippocampal HT-22 cells can show a physical interaction between CHCHD6 and APP, which is confirmed by CHCHD6 CRISPR-cas9 knockout (KO) (Figure 3a).

In vitro binding assays of the recombinant protein also indicate a direct and selective interaction between CHCHD6 and APP, as no binding was observed between APP and mittoillin or CHCHD3 (Figure 3b).

Further detection of APP binding to CHCHD6 by in situ proximity assay (PLA), PLA-positive spots were observed in control group cells, but not in CHCHD6 KO cells (Figure 3c).

PLA-positive spots between anti-APP and anti-CHCHD6 antibodies were also observed in the hippocampus of WT and APPNL−G−F KI mice, and the interaction in AD mice was significantly reduced (Figure 3d).

The authors' team further demonstrated a strong direct interaction between CHCHD6 and APP in the hippocampus after death in normal people, while the interaction was weakened in the hippocampus in AD patients (Figure 3e).

In addition, the authors' team observed a significant increase in CHCHD6 levels after downregulating the app using shRNAs (Figure 3f).

This suggests that CHCHD6 is negatively correlated
with APP levels.

Interestingly, the team found that the reduction in CHCHD6 enhanced the accumulation
of APP on MAMs (mitochondria-associated endoplasmic reticulum membranes).
。 CHCHD6 KO cells produced more positive PLA fluorescent spots between APP and MAM-labeled fatty acid coenzyme a ligase 4 (FACL4) compared to WT (Figure 3g); Mitochondrial sorting of WB also confirmed enhanced
levels of APP and C99 on APP and C99 on CHCHD6 KO cells.
The number of PLA-positive spots between APP and FACL4 in hippocampal regions of APPNL-G-F KI mice was also significantly increased compared to WT (Figure 3h).

MAMs are APP-treated intracellular bits, and the accumulation of C99 cleavage products (Aβ amylprotosome) at MAMs can impair mitochondrial bioenergy, disrupt cellular lipid homeostasis, and disrupt membrane lipid components
.

The authors' team observed a significant decrease in APP total protein levels and C99 fragment enhancement in CHCHD6 KO cells (Figure 3i).

APP levels recovered after reexpression of CHCHD6 and C99 levels decreased (Figure 3i).

In summary, these results show that CHCHD6 interacts with APP and stabilizes each other, and the loss of CHCHD6 promotes the accumulation and abnormal processing
of APP on MAMs.
CHCHD6 is reduced
in brain neurons in AD mice and AD patients.

APP processing product AICD combined

CHCHD6 promoter and inhibit CHCHD6 transcription

In addition to C99, another lysate processed by APP is AICD, which regulates gene transcription
by binding directly to gene promoters via the AICD-Fe65-Tip60 complex.
APP proteolyzed AICD translocates to the nucleus and regulates gene transcription
by binding to Fe65.
Therefore, the team further investigated whether AICD may be involved in transcriptional inhibition
of CHCHD6 associated with elevated APP levels.
It was found that the intensity of AICD in the nucleus was greatly enhanced after knocking out CHCHD6 in HT-22 cells (Figure 3j).

This result shows that the loss of CHCHD6 promotes the processing of APP and also leads to the accumulation
of AICD.

Next, the authors' team overexpressed APP, GFP-tagged AICD, and MYC-Fe65 (adaptor proteins that stabilize AICD and enhance its nuclear translocation) in HEK293 cells with or without APP shRNA (Figure 3k).

The results showed that APP overexpression consistently reduced CHCHD6 protein levels, and that expression of IC alone or AICD and Fe65 co-expression resulted in a greater reduction (Figure 3k).

The co-expression of AICD and Fe65 also inhibited the elevation of CHCHD6 protein caused by APP knockdown (Figure 3k).

These results indicate an inverse correlation
between AICD and Fe65 levels and CHCHD6 levels.
Histone acetyltransferase Tip60 can form a complex with the cytoplasmic tail and nuclear adaptor Fe65 of the APP, enhancing the binding ability
of the complex to the targeted gene promoter.

To further investigate whether AICD and Fe65 can directly bind to the CHCHD6 promoter, the authors' team overexpressed AICD, Fe65, and Tip60 in HEK293 cells and subcloned the CHCHD6 gene 5' region fragment into the
luciferase reporter vector pGL3.
It was found that the co-expression of AICD and Fe65 significantly reduced the ability of the CHCHD6 promoter to drive luciferase expression (Figure 3l).

Chromatin immunoprecipitation (ChIP) assays in cells co-expressing AICD, Fe65, and Tip60 also found that they bind to the CHCHD6 promoter (Figure 3m).

These results indicate that the AICD/Fe65/Tip60 complex directly binds to the CHCHD6 promoter to inhibit its activity and supports APP in mediating transcriptional inhibition of CHCHD6 in a highly expressed form (Figure 1).

Figure III.
CHCHD6 and APP are interdependent and mutually regulated

CHCHD6 deficiency can be caused

Mitochondrial damage and cell death

Next, the team evaluated the effects
of CHCHD6 deficiency on mitochondrial integrity and neuronal survival.
CHCHD6 KO in HT-22 cells leads to mitochondrial depolarization, exhibiting a significant decrease in mitochondrial membrane potential (MMP) (Figure 4a).

The downregulation of CHCHD6 increases the proportion of cells with Tom20+ punctate or short rod-like mitochondria, indicating mitochondrial fragmentation (Figure 4b).

And CHCHD6 KO enhances sensitivity to a β-induced cell death (Figure 4c).

Consistent with previous studies, CHCHD6 deficiency inhibits mitochondrial respiration activity, manifested by a decrease in mitochondrial basal oxygen consumption rate, a decrease in maximum oxygen consumption rate, and a decrease in ATP production in intact cells (Figure 4d).

Decreased ATP production was also observed in stable APPWT and APPSWE-overexpressed Neuro2a cells (Figure 4e).

Lower levels of mitochondrial cytochrome C were also detected in CHCHD6 KO HT-22 cells (Figure 4f
).

Overexpressing the CHCHD6-FLAG construct to compensate for ATP production in stable APP Neuro2a cells (Figure 4e) was able to reduce CHCHD6-induced cytochrome c release in hippocampal neurons in HT-22 mice (Figure 4f) and attenuate aβ-induced cell death (Figure 4g).

Taken together, these results suggest that CHCHD6 is necessary to maintain mitochondrial quality control and bioenergetics, and that the loss of CHCHD6 can disrupt the integrity of the mitochondrial crest, leading to bioenergetic deficiencies and cell death
.

Figure IV.
CHCHD6 deficiency can cause mitochondrial damage and cell damage

CHCHD6 is missing in the AD model

Induces neuronal cholesterol accumulation

To further evaluate the effect of CHCHD6 deficiency on AD-associated neuronal damage, the authors' team performed whole-transcriptome RNA-Seq analysis
of CHCHD6 KO HT-22 mouse hippocampal neurons.
Of the 5336 RNA transcripts identified in HT-22 cells, 1827 RNAs (Figure 5a, green and red dots) significantly altered in CHCHD6 KO HT-22 cells were screened for pathway enrichment analysis
.
The results of the KEGG analysis showed significant alterations in genes involved in cholesterol metabolism in CHCHD6 KO cells (Figure 5a).

Filipin, a fluorescent probe used to monitor cholesterol deposition, also showed significantly more filipin-bound cholesterol in CHCHD6 KO HT-22 cells compared to control cells (Figure 5b).

The accumulation of cholesterol in AD is synaptic toxic, and these findings suggest that deletion of CHCHD6 may promote the accumulation
of cholesterol in neurons by regulating the expression of cholesterol metabolism genes.

To test the effect of CHCHD6 loss on the accumulation of cholesterol levels in neurons in vivo, the authors' team specifically downregulated CHCHD6
in APPNL−F mice and age-matched WT mice using AAV-expressing shRNA.
and under the control of the human synaptophytin (hSyn) promoter, the virus is limited to selective expression in neurons (Figure 5c).

Six months after stereotactic injection of the virus, the authors' team observed that mCherry-labeled AAVs containing CHCHD6 shRNA were successfully expressed (Figure 5c).

And the content of CHCHD6 in mouse hippocampal protein extracts injected with AAV5-CHCHD6 shRNA was significantly reduced compared to the control group (Figure 5d).

Filipin staining results in the hippocampus of mice showed that inhibition of CHCHD6 caused significant cholesterol accumulation (Figures 5e and 5f
).
ELISA also confirmed that inhibition of CHCHD6 increased total cholesterol levels in the hippocampus in mice (Figure 5g).

The authors' team further analyzed genes involved in cholesterol biosynthesis, degradation, transport, and modification and found that hmgcr, cyp46a1, and lrp1 showed consistent changes in hmgcr, cyp46a1, and lrp1 after CHCHD6 inhibition in cells and APPNL-F KI mice (Figure 5h
).
Hmgcr encodes HMG-CoA reductase (HMGCR), a rate-limiting enzyme for cholesterol synthesis; Cyp46a1 encodes cytochrome P450 family 46 subfamily A1 (CYP46A1), which is specifically expressed in neurons to catalyze the conversion of 24-hydroxycholesterol from the brain
.
Thus, downregulation of CHCHD6 may promote cholesterol accumulation
in APPNL-F KI mice by enhancing cholesterol biosynthesis and inhibiting cholesterol elimination.
lrp1 encodes LDL receptor-associated protein (LRP1), which mediates the uptake
of cholesterol-containing ApoE by astrocytes.
The transcription level of LRP1 is also inhibited by the AICD-Fe65-Tip60 complex binding to its promoter, so its decreased expression may be similarly regulated
to the decline in CHCHD6.

Figure V.
In the AD model, CHCHD6 deficiency induces neuronal cholesterol accumulation

CHCHD6 reduction accelerates APPNL−F KI

Cognitive deficits and AD pathology in mice

Next, the authors' team investigated whether downregulation of CHCHD6 affects AD-like neuropathological and behavioral deficits
in APPNL−F KI mice with chronic pathology.
Neither WT nor APPNL−F mice at 12 months of age (6 months after AAV injection) were injected with AAV control shRNA, and cognitive impairment
was not shown on the Y maze cognitive test.
However, the rate of spontaneous change decreased in the WT and APPNL−F mouse Y maze assays after AAV5-CHCHD6 shRNA injection (Figure 6a).

While no difference was shown in the Barnes maze cognitive test between 12-month-old WT and APPNL−F mice injected with control shRNA, APPNL−F KI mice injected with AAV5-CHCHD6 shRNA spent longer and produced more errors in finding target escape regions (Figure 6b).

Taken together, these data suggest that chronic CHCHD6 deficiency in APPNL−F AD mice promotes spatial learning and long-term memory impairment
.

To investigate whether pathological markers of AD were also affected by CHCHD6 knockout, the authors' team first examined the levels of
APP-treated products in 12-month-old hippocampal protein extracts.
WB analysis showed that injection of AAV-CHCHD6 shRNA enhanced APP treatment in WT and APPNL−F mice, showing elevated levels of C99 fragments (Figure 6c).

In APPNL−F mice injected with AAV-CHCHD6 shRNA, levels of C99 product were much higher than in mice injected with control shRNA (Figure 6c).

In addition, AAV-CHCHD6 shRNA injection increased the intensity of AICD in NeuN+ cells of APPNL-F mice compared to mice injected with control shRNA group (Figure 6d).

Staining the brain fractions of 12-month-old mice with 6E10 antibody to mark amyloid aggregation revealed more 6E10+ amyloid deposits and a larger amyloid plaque coverage area in the hippocampus of APPNL−F mice injected with AAV-CHCHD6 shRNA (Figure 6e).

These data are consistent with the findings found in HT-22 cells (Figures 3i and j) that the deletion of CHCHD6 promotes APP processing, which in turn accelerates the pathology
of amyloid.

Since synaptic loss in AD is strongly associated with cognitive decline, the authors' team stained
the mouse brain fraction with antibodies against synaptophysin (a presynaptic marker) and postsynaptic density 95 (PSD95: a postsynaptic marker).
The results showed that after AAV-CHCHD6 shRNA injection, the number of synaptophycin and PSD95 immune activity points in the CA3 region of APPNL-F mice decreased significantly, indicating a decrease in synaptic density (Figures 6f and g).

Next, the authors' team examined neuroinflammation
.
Immunofluorescence staining showed significantly higher intensities in the hippocampal IBA1 (marker of microglia) and GFAP (marker of astrocytes) injected with AAV-CHCHD6 shRNA than in mice injected with shRNA control (Figures 6 g and i).

This suggests that downregulation of CHCHD6 enhances AD-related neuroinflammation
.
Overall, these data suggest that downregulation of CHCHD6 accelerates neuropathological and cognitive deficits
in AD mice.

Figure VI.
Viral vector-mediated downregulation of CHCHD6 accelerates cognitive deficit and AD pathology in APPNL−F KI mice

Compensation for CHCHD6 deletion is mitigated

Neuropathological and cognitive deficits in APPNL-G-F KI mice

Finally, the authors' team explored whether compensating for CHCHD6 deficiency would alleviate AD-like neuropathological and cognitive deficits
in APPNL−f-g AD mice.
CHCHD6 is inserted into the AAV vector under the control of the hSyn promoter (AAV5-hSyn-CHCHD6-eGFP) to achieve selective expression of GFP-labeled CHCHD6 in neurons (Figure 7a).

AAV5-hSyn-eGFP was used as the empty vector control
.
AAV5-hSyn-CHCHD6-eGFP or AAV control construct is injected stereotactically into the bilateral hippocampus of 3-month-old APPNL-G-F mice and age-matched WT sibble mice
.

After 6 months, western blot analysis showed that CHCHD6 protein levels in hippocampal protein extracts from AAV5-eGFP-CHCHD6-injected APPNL-F-G mice were significantly up-regulated, similar to WT mice injected with AAV5-eGFP control (Figure 7b).

。 APPNL-G-F MICE INJECTED WITH THE AAV5-eGFP control group showed short-term cognitive decline through Y-maze assessment at 9 months of age; In contrast, age-matched APPNL−G−F mice injected with AAV5-eGFP-CHCHD6 had an improved rate of spontaneous change in the Y-maze test, reaching levels similar to WT mice (Figure 7c).

These results suggest that CHCHD6 neuronal overexpression improves spatial working memory
in APPNL−F−G AD mice.
In addition, CHCHD6 overexpression significantly improved the performance of 9-month-old APPNL−G−F mice in the Barnes maze test (Figure 7d).

APPNL-G-F MICE INJECTED WITH AAV-CHCHD6 HAVE LESS TIME TO FIND TARGET ESCAPE REGIONS AND FEWER ERRORS THAN MICE INJECTED WITH CONTROL VIRUS (FIGURES 7c and d).

The authors' team also evaluated whether compensation for CHCHD6 loss could attenuate AD-related cholesterol metabolism disorders
.
It was found that AAV5-eGFP-CHCHD6 injection eliminated the elevated cholesterol synthesis gene HMGCR and inhibited the cholesterol elimination gene CYP46A1 (Figure 7e).

A significant increase in filipin+ cholesterol in the hippocampus of APPNL-G-F mice injected with control virus was observed at 9 months of age, and AAV-eGFP-CHCHD6 virus expression decreased from 3 months of age (Figures 7f and G).

In addition, ELISA analysis showed that AAV-eGFP-CHCHD6 expression eliminated increased cholesterol content in the hippocampus of APPNL-G-F mice (Figure 7h).

AAV-eGFP-CHCHD6 had no effect on cholesterol content in WT mice (Figure 7h).

CHCHD6 overexpression also reduced AICD intensity and hippocampal Aβ coverage in 9-month-old APPNL−G−F mouse NeuN+ cells compared to mice injected with control virus (Figures 7j and 8a).

In addition, imaging analysis showed that the hippocampal synaptophysin and PSD95 decreased significantly in mice injected with the control virus at 9 months of age and returned to normal by AAV-mediated CHCHD6 overexpression (Figure 8b).

In addition, AAV5-eGFP-CHCHD6 overexpression significantly reduced IBA1 immune density in APPNL−G−F mice, indicating inhibition of neuroinflammation (Figure 8c).

Thus, this finding suggests that compensating for the loss of CHCHD6 in the hippocampus of rapidly developing APPNL-G-F mice can reduce AD-related pathology and cognitive impairment
.

Figure VII.
Compensation for CHCHD6 deletion can reduce neuropathological and cognitive deficits in APPNL-G-F KI MICE

Figure VIII.
Compensation for CHCHD6 deletion can reduce amyloid accumulation, synaptic loss, and glial hyperplasia in APPNL-G-F KI MICE

In this study, the team revealed the unique function
of the MICOS subunit CHCHD6 in regulating the APP amyloid process (one of the markers of AD) and neuronal cholesterol accumulation (a known risk factor for AD).
The pathological deletion of CHCHD6 acts as an induction condition upstream of APP amyloid processing and neural cholesterol accumulation, which provides a direct mechanistic explanation
for the effect of mitochondrial damage on AD neuropathology.
At the same time, in the AD model, compensation for CHCHD6 deletion can restore cholesterol homeostasis, inhibit amyloid production, and reduce neurodegeneration and cognitive deficits
.