Amyloid Precursor Protein (APP), PSEN1, and Gamma-Secretase

Amyloid Precursor Protein (APP) is a single-pass Type I transmembrane glycoprotein encoded on human chromosome 21q21.3 (290 kb, 18 exons) that undergoes regulated intramembrane proteolysis (RIP) via two mutually exclusive enzymatic pathways: the constitutive non-amyloidogenic α-secretase (ADAM10) pathway at the plasma membrane and the endosomal amyloidogenic β-secretase (BACE1) plus γ-secretase (PSEN1/PSEN2) pathway. In the central nervous system, alternative splicing of Exons 7 and 8 generates three major isoforms—APP695 (lacking the Exon 7 Kunitz protease inhibitor KPI domain and expressed predominantly in neurons), APP751, and APP770 (expressed in glia, endothelial cells, and platelets)—which function physiologically in synaptic adhesion, axonal transport, copper/zinc ion homeostasis, and ferroportin-mediated iron export.

During amyloidogenic processing inside acidic early endosomes (pH 4.5–5.5), ectodomain shedding by BACE1 leaves a 99-amino-acid membrane-tethered C-terminal fragment (C99 or CTFβ), which is subsequently cleaved within the hydrophobic lipid bilayer by γ-secretase, a four-subunit intramembrane aspartyl protease whose catalytic core is formed by Presenilin-1 (PSEN1 on Chr14q24.2) or Presenilin-2 (PSEN2 on Chr1q42.13). Because γ-secretase acts as a processive carboxypeptidase that trims C99 in steps of three to four residues along two parallel product lines (Aβ49 → Aβ46 → Aβ43 → Aβ40 versus Aβ48 → Aβ45 → Aβ42), autosomal dominant mutations in APP, PSEN1, or PSEN2 destabilize the enzyme–substrate complex (ΔG‡), causing premature release of aggregation-prone Aβ₄₂ and Aβ₄₃ peptides that nucleate autosomal dominant early-onset Alzheimer disease (EOAD).

Non-Amyloidogenic Alpha-Secretase vs. Amyloidogenic Beta-Secretase Cleavage

Whether APP yields neuroprotective soluble ectodomains or pathogenic amyloid-β (Aβ) peptides depends on subcellular trafficking and competition between membrane-bound sheddases:

1. Non-Amyloidogenic Pathway (ADAM10 / α-Secretase): Roughly 85–90% of neuronal APP695 at the cell surface is cleaved by the zinc metalloprotease ADAM10 precisely between Lys16 and Leu17 (Lys687–Leu688 in APP770 numbering) inside the middle of the Aβ sequence, precluding Aβ formation. This cleavage releases the neurotrophic ectodomain sAPPα into the extracellular space—enhancing long-term potentiation (LTP) and nicotinic acetylcholine receptor signaling—and leaves an 83-residue C-terminal stub (C83 or CTFα). Subsequent γ-secretase cleavage of C83 yields the benign p3 peptide (Aβ17–40/42) and the APP intracellular domain (AICD).
2. Amyloidogenic Pathway (BACE1 / β-Secretase): Uncleaved APP is internalized via clathrin-mediated endocytosis (driven by its cytoplasmic YENPTY motif) into acidic early endosomes, where β-site APP-cleaving enzyme 1 (BACE1, encoded on Chr11q23.3) exhibits maximal catalytic activity (optimum pH ≈ 4.5). BACE1 cleaves APP between Met671 and Asp672 (Asp1 of Aβ), shedding sAPPβ and generating the 99-residue transmembrane stub C99 (CTFβ), the immediate substrate for γ-secretase.

Swedish Pathogenic vs. Icelandic Protective APP Mutations

Human genetic variation at the BACE1 cleavage site (codons 670–673 of APP770) provides a bidirectional dose-response demonstration of the amyloid hypothesis:

• The Swedish Double Mutation (APP KM670/671NL): Identified in two large Swedish families with autosomal dominant EOAD (mean onset ≈ 52 years), this dinucleotide substitution replaces Lys670–Met671 immediately N-terminal to the β-secretase site with Asn670–Leu671. The Asn–Leu dipeptide packs into the S1 and S2 substrate-binding pockets of BACE1 with a 10- to 50-fold higher catalytic efficiency (kcat/Km) than wild-type APP, shifting cleavage into the trans-Golgi network and elevating both Aβ40 and Aβ42 production 6- to 8-fold without altering their ratio.
• The Icelandic Protective Mutation (APP A673T, rs63750847): Discovered via whole-genome sequencing of 1,795 Icelanders (allele frequency ≈ 0.43% in Nordic populations), substituting alanine with threonine at position 2 of Aβ (Ala673Thr, immediately C-terminal to the BACE1 scissile bond) reduces BACE1 catalytic cleavage by ~40% and slows Aβ fibril nucleation. Carriers of APP A673T exhibit a 5.2-fold reduction in late-onset Alzheimer disease risk (OR = 0.19) and preserved cognitive scores past age 85, even among [APOE](/wiki/apolipoprotein-e) ε4 carriers. Conversely, a homozygous recessive substitution at the identical codon, A673V, increases BACE1 cleavage and Aβ fibrillogenesis.

Four-Subunit Cryo-EM Structure and Presenilin Autoproteolysis

Active γ-secretase (~170 kDa protein mass, ~230 kDa with glycosylation) is an obligate heterotetrameric membrane protein complex containing one copy each of four subunits in 1:1:1:1 stoichiometry:

1. Presenilin (PSEN1 or PSEN2): A 9-transmembrane (9-TM) helix subunit containing two conserved catalytic aspartate residues located on opposing transmembrane helices—Asp257 on TM6 and Asp385 on TM7 in PSEN1—positioned inside a water-accessible hydrophilic cavity within the lipid bilayer. During holocomplex assembly in the endoplasmic reticulum, the hydrophobic loop between TM6 and TM7 (Exon 9) undergoes intramolecular autoproteolysis (between Thr291 and Ala299), generating a stable heterodimer of a ~30-kDa N-terminal fragment (NTF, TM1–TM6) and a ~20-kDa C-terminal fragment (CTF, TM7–TM9).
2. Nicastrin (NCSTN): A single-pass transmembrane glycoprotein whose large bilobed extracellular ectodomain (~660 residues) projects ~70 Å above the membrane like a lid. Within the ectodomain, Glu333 binds the free N-terminal amino group of C99 (Asp1), while the bulky ectodomain sterically excludes full-length transmembrane proteins (ectodomain > 20–30 residues) that have not first undergone shedding by ADAM10 or BACE1.
3. Anterior Pharynx-Defective 1 (APH1A or APH1B): A 7-TM structural scaffold subunit that stabilizes the PSEN C-terminal helices.
4. Presenilin Enhancer 2 (PEN2, encoded by PSENEN): A hairpin-like 3-TM subunit that binds TM4 of PSEN1, triggering an allosteric conformational shift that activates PSEN1 autoproteolysis and opens the catalytic pore.

Sequential Tripeptide Trimming and PSEN1 Kinetic Destabilization

Cryo-EM structures of human γ-secretase bound to C99 (and Notch1, one of >90 physiological Type I transmembrane substrates cleaved by γ-secretase) show that the transmembrane α-helix of C99 unwinds at its C-terminal cytoplasmic boundary to form an intermolecular three-stranded hybrid β-sheet with two β-strands induced in PSEN1 (β1 in the TM6–TM7 loop and β2 preceding TM7). This β-sheet anchors the scissile bond directly between the catalytic dyad Asp257 and Asp385.

Proteolysis proceeds in two stages:

1. Initial Endopeptidase (ε-site) Cleavage: PSEN1 cleaves C99 near the cytoplasmic membrane border either between Leu49–Val50 (~75–85% of wild-type cleavage events) or between Thr48–Leu49 (~15–25%), releasing the soluble APP intracellular domain (AICD50-99 or AICD49-99) into the cytosol and leaving membrane-bound Aβ49 or Aβ48.
2. Processive Carboxypeptidase (ζ- and γ-site) Trimming: Because the α-helical pitch of the substrate is 3.6 residues per turn, γ-secretase slides the remaining peptide forward by three residues (one helical turn) at each step along two distinct production lines:

• The Aβ40 Line: Aβ49 → (+VIT) → Aβ46 → (+VIV) → Aβ43 → (+IAT) → Aβ40 (~85–90% of secreted Aβ).
• The Aβ42 Line: Aβ48 → (+ITL) → Aβ45 → (+TVI) → Aβ42 (~8–10% of secreted Aβ) → (+GGVV) → Aβ38.

Over 300 pathogenic missense mutations in PSEN1 (such as the Colombian Paisa founder mutation E280A, M146V, and L166P) and intra-Aβ APP mutations (London V717I, Florida I716V) cause autosomal dominant EOAD not by increasing total γ-secretase activity, but by thermodynamically destabilizing the γ-secretase–Aβn enzyme–substrate (E–S) complex. Quantitative enzyme kinetics (koff vs. kcat) demonstrate that mutant PSEN1 complexes release longer, hydrophobic intermediates (Aβ42 and Aβ43) from the catalytic cavity prematurely before the fourth carboxypeptidase trimming step to Aβ40 or Aβ38 can occur, elevating the Aβ42/Aβ40 ratio and accelerating amyloid fibril nucleation.

Anti-Amyloid Monoclonal Antibodies vs. Gamma-Secretase Modulators

Clinical translation of APP processing pharmacology has highlighted the mechanistic distinction between non-selective enzyme inhibition and conformation-specific clearance or allosteric modulation:

• Failure of Non-Selective γ-Secretase Inhibitors (GSIs): Pan-γ-secretase inhibitors (semagacestat, avagacestat) blocked catalytic Asp257/Asp385 cleavage across all substrates, halting Notch1 intracellular domain (NICD) signaling required for gastrointestinal goblet-cell differentiation, T-cell maturation, and epidermal homeostasis, while causing paradoxical cognitive worsening due to toxic accumulation of uncleaved C99 stubs in neuronal endosomes.
• Allosteric γ-Secretase Modulators (GSMs): Second-generation small-molecule GSMs (BPN-15606, RG6289) bind allosterically to the PSEN1 N-terminal fragment, stabilizing the E–S complex (increasing residence time τ) without blocking initial ε-site cleavage of APP or Notch1. Consequently, GSMs accelerate the final carboxypeptidase trimming steps (Aβ42 → Aβ38 and Aβ40 → Aβ37), lowering pathogenic Aβ42 while preserving Notch signaling.
• Protofibril and Pyroglutamate Anti-Aβ Monoclonal Antibodies: Passive immunotherapy with lecanemab (an IgG1 monoclonal antibody selectively binding soluble Aβ protofibrils with 10- to 15-fold higher affinity than fibrillar plaques and >1,000-fold over monomers) and donanemab (targeting N-terminally truncated, pyroglutamate-modified AβpE3–42 found exclusively in mature parenchymal plaques) clears cerebral amyloid by 60–85% on PET imaging via microglial Fcγ receptor phagocytosis, slowing clinical cognitive decline by 27–35% over 18 months in early Alzheimer disease, though requiring monitoring for amyloid-related imaging abnormalities (ARIA-E and ARIA-H) driven by vascular wall Aβ40 mobilization, particularly in APOE ε4/ε4 homozygotes.

Why do individuals with Down syndrome (Trisomy 21) develop universal early-onset Alzheimer neuropathology?

Because the APP gene resides on the long arm of chromosome 21 (Chr21q21.3), individuals with constitutional Trisomy 21 carry three wild-type copies of APP, resulting in a 1.5-fold gene-dosage elevation in APP mRNA, C99, and secreted Aβ40/Aβ42 throughout life. Virtually 100% of individuals with full Trisomy 21 develop diffuse and neuritic cortical Aβ plaques by age 40, whereas rare individuals with partial Trisomy 21 whose triplicated chromosomal segment excludes 21q21.3 (APP disomy) never develop Alzheimer pathology despite having Down syndrome phenotypes.

How do intra-Aβ mutations like the Arctic (E693G) and Dutch (E693Q) variants act differently from flanking APP mutations?

Mutations flanking the Aβ domain (Swedish KM670/671NL at the N-terminus or London V717I near the C-terminus) alter BACE1 cleavage rate or γ-secretase trimming site selection. By contrast, mutations inside the mid-domain of Aβ (codons 21–23, corresponding to APP residues 692–694: Flemish A692G, Dutch E693Q, Arctic E693G, and Iowa D694N) remove negatively charged side chains (Glu22 or Asp23) that normally repel adjacent Aβ monomers, accelerating soluble protofibril formation (Arctic, targeted by lecanemab) or vascular smooth-muscle cell toxicity and hereditary cerebral hemorrhage with amyloidosis (Dutch).

Why did BACE1 inhibitors fail to improve cognition in Phase III clinical trials?

Small-molecule BACE1 inhibitors (verubecestat, lanabecestat, atabecestat) lowered cerebrospinal fluid Aβ40 and Aβ42 by 75–90%, but produced a mild, non-progressive 2–3 point cognitive worsening within 12 weeks of initiation that reversed upon drug discontinuation. Proteomic studies demonstrated that BACE1 also cleaves essential synaptic substrates—most notably Seizure Protein 6 (SEZ6), Close Homolog of L1 (CHL1), and Neuregulin-1 (NRG1)—required for dendritic spine maintenance and axonal myelination.

1. Jonsson, T., Atwal, J. K., Steinberg, S., et al. (2012). "A mutation in APP protects against Alzheimer's disease and age-related cognitive decline." Nature, 488(7409), 96–99. https://doi.org/10.1038/nature11283
2. Zhou, R., Yang, G., Guo, X., et al. (2019). "Recognition of the amyloid precursor protein by human γ-secretase." Science, 363(6428), eaaw0930. https://doi.org/10.1126/science.aaw0930
3. Szaruga, M., Munteanu, B., Lismont, S., et al. (2017). "Alzheimer's-causing mutations shift Aβ length by destabilizing γ-secretase–Aβn interactions." Cell, 170(3), 443–456.e14. https://doi.org/10.1016/j.cell.2017.07.004
4. van Dyck, C. H., Swanson, C. J., Aisen, P., et al. (2023). "Lecanemab in early Alzheimer's disease." New England Journal of Medicine, 388(1), 9–21. https://doi.org/10.1056/NEJMoa2212948