**5. Interaction partners of the AICD**

phorylation regulates both AICD peptide formation and AICD-dependent cellular responses (Figure 2). These data point to a sorting function regulated by Tyr682 phosphorylation, with non-phosphorylated APP kept at the plasma membrane and therefore processed by αsecretase, and a phosphorylation-dependent re-localization resulting in β-cleavage. Sorting implies docking to respective intracellular trafficking machineries and their adaptors, includ‐ ing PTB domain containing proteins. Consistently, an APPYG/YG mutation introduced into the endogenous APP locus by knock-in led to a marked shift toward the non-amyloidogenic pathway in brain with increased levels of full length APP, sAPPα, αCTF, unaltered βCTF and

Sorting due to differentially phosphorylated residues is one side of the medal, signaling is the other [40]. Two signaling proteins are well known to require Tyr682 phosphoryla‐ tion for binding to APP-CTFs, namely ShcA and Grb2. ShcA is a member of a family of cytoplasmic adaptor proteins (ShcA, ShcB, ShcC) that interacts with its PTB and Src ho‐ mology2 (SH2) domains with receptor tyrosine kinases (RTKs) and activated growth fac‐ tor receptors, which is the case also for SH2/SH3 domains containing Grb2 [42]. The initiated cascades are involved both in cell proliferation and gene transcription events, like i.e. the MAP kinase pathway. Again, binding occurs only to pTyr682 of βCTFs but not of αCTFs [41] (Figure 2). Whereas the reasons for the different binding preferences remain elusive, the underlying structural transitions within the AICD itself modulating

First structural insights on the AICD peptide came from NMR experiments, revealing most of the AICD to be unstructured. The transient structure (also termed intrinsic disorder: ID) of cytoplasmic domains of membrane proteins is well suited for the molecular recognition in intracellular signaling events for a number of reasons [43]: (i) modulation of the structural propensity provides ID proteins with the capability to combine high specificity with low affinity; (ii) binding diversity in which one region specifically recognizes differently shaped partners by structural accommodation at the binding interface, a phenomenon known as oneto-many signaling; (iii) binding commonality in which distinct sequences recognize a common binding site (with eventually different folds); (iv) the formation of large interaction surfaces as the ID region wraps up or surrounds its binding partner, making it possible to overcome steric restrictions; (v) faster rates of association by reducing dependence on orientation factors and by enlarging target sizes; (vi) faster rates of dissociation by unzipping mechanisms; (vii) the precise control and simple regulation of the binding thermodynamics; and (viii) the reduced life-time of ID proteins in the cell, possibly representing a mechanism of rapid turnover of important regulatory molecules. A prominent example of intrinsically disordered proteins is α-synuclein, a protein critically involved in Parkinson's disease, which binds to a multitude of partners differentially by alternative folding [44], a feature that equally applies

reduced sAPPβ and Aβ40 levels [31].

8 Understanding Alzheimer's Disease

sorting and signaling have been studied in some detail.

**4. Structural transitions within the AICD**

to the intracellular domain of APP.

More than 20 proteins have been reported to interact with the AICD [47] (Table 1). However, little is known whether these complexes occur also *in vivo* and what relevance they may have for cell physiology or AD pathogenesis. Basically, they can be classified in modifying, sorting, or signaling interactions. The modifying enzymes have been already mentioned and account for phosphorylation and prolyl *cis/trans* isomerization events. Basolateral sorting is guided by the protein PAT1, which is the only protein that has been shown to directly interact with the 653YTSI motif and is associated with microtubules [48].

Knowledge about the interaction partners for the 667VTPEER motif is similar scarce. Major binder for the motif, and as well for the complete AICD, are the multi-domain adaptor/ scaffolding proteins of the Fe65 family (Fe65, Fe65L1, and Fe65L2) [49]. The only additional binding partner to the 667VTPEER motif is the dimeric adaptor protein 14-3-3γ, which seems to stabilize the AICD/Fe65 interaction [50]. Fe65 is enriched in brain, whereas Fe65L1 and Fe65L2 are more widely expressed. All three members contain a WW domain and two PTB domains (PTB1 and PTB2). Through the PTB2 domain, they interact with the AICD and can alter APP processing. After proteolytic processing of APP and release of the AICD to the cytoplasm, Fe65 can translocate to the nucleus to participate in gene transcription events (Figure 2), which is modulated by 14-3-3γ. This role is further mediated by interactions of Fe65- PTB1 with the transcription factors CP2/LSF/LBP1 [51] and Tip60 [52] and the WW domain with the nucleosome assembly factor SET [53]. Possible target genes identified by reporter assays include GSK3β, Neprilysin, KAI1, and the low-density lipoprotein receptor-related protein 1 (LRP1), but the physiological relevance for endogenous transcriptional regulation has been discussed controversially [54]. Fe65-PTB1 also interacts with two cell surface lipoproteins receptors, namely LRP1 [55] and ApoEr2 [56], forming trimeric complexes with APP. The Fe55 WW domain further binds to mammalian Ena (mEna) [57], through which it functions in regulation of the actin cytoskeleton, cell motility, and neuronal growth cone formation [49]. The interaction has been implicated in a role for AICD signaling, in synaptic plasticity and memory [58]. Moreover, Fe65 family proteins have attracted attention, as Fe65 or Fe65L1 double knockout mice revealed defects in cortical development with neuronal mispositioning and ectopia, resembling human lissencephaly type 2 [59]. Interestingly, very similar cortical defects were also found in APP-/-APLP1-/-APLP2-/- triple knockout mice lacking all APP family members, suggesting a lack of APP/Fe65 dependent signaling as the underlying cause of defects in both mouse mutants [60].

constructs lacking the AICD are still transported to the nerve terminal by the fast axonal

Basolateral sorting Endocytosis, signaling and transcription activation, ... AICD/Fe65 stabilization

synapse formation, ...

Transport, signaling

Exocytosis,

Transport

Signaling Notch crosstalk

**Table 1. Selected interaction partners of the AICD.** \*Data depend on cell line studied and are sometimes conflicting. \*\*Due to basolateral sorting and independent of PAT1 binding. Pat1 binding as such increases Aβ levels [48]. \*\*\*Numb isoform dependent. ↓ denotes changes of non-amyloidogenic (α) or amyloidogenic (β) APP processing.

The Dab family member Dab1 regulates neuronal migration in mammals as an essential component of the Reelin signaling pathway. Dab1 binds not only to APP family proteins [64] but is well known to also bind to ApoE receptors (ApoEr2, VLDLR, and LRP) [69]. Dab1 increases cell surface expression of APP and ApoEr2, increases α-cleavage of APP and ApoEr2, and decreases APP βCTF formation and Aβ production in transfected cells and in primary neurons. The Dab family represents a prototype of PTB domains that bind their ligands in a pY-independent manner [36]. In addition Dab proteins bind specifically to the phosphoinositide (PI) PI-4,5-P2, which is predominantly located at the cellular membrane [70]. Binding of PTB domains to PIs is a common principle to locate and orientate the adaptors at the target membrane and to facilitate downstream events that accompany NPXY peptide binding. Since PTB domains structurally belong to the pleckstrin homology (PH) superfold family and PH domains are the prototypical PI binding domains, this function seems to be evolutionarily conserved within PTB domains [36]. The crystal structures of ternary complexes of Dabs bound to ApoEr2 [71] or APP [72] peptides and lipid revealed the lipid head group (IP3) to be recognized by a large basic patch opposite the peptide-binding groove (Figure 4A). This patch, also termed as "phospholipid binding-crown", is conserved in many PTB domains [36].

Finally, binding of the AICD to the Numb PTB domain has been found to inhibit Notch signaling [65], thereby establishing a crosstalk between the APP family and Notch in the development of the peripheral nervous system (PNS) [73]. Like APP, the Notch receptor undergoes a series of proteolytic cleavages that release the Notch intracellular domain (NICD) that functions in transcriptional activation and subsequent signal transduction events, including proliferation, differentiation, or apoptotic cues [74]. Similar to the NICD, the AICD has been also found to regulate PI-mediated calcium signaling through a γ-secretase depend‐

**Function Processing\* Selected**

Structure and Function of the APP Intracellular Domain in Health and Disease

α↑, β↓\*\* β↓ n.a. β↓

http://dx.doi.org/10.5772/54543

β↓ α↑, β↓ - \*\*\*

**citations**

11

[48] [49] [50] [61] [62]

[63] [64] [42] [65]

**Interacting region within AICD**

YTSI AICD-C31: VTPEER + GYENPTY

VTPEER GYENPTY

GYENPTY GYENPTY G(pY)ENPTY GYENPTY

transport mechanism [63].

**domain**

n.a. PTB2 n.a. PTB

PTB PTB PTB/SH2 PTB

**Protein Interacting**

PAT1 Fe65, Fe65L1, -L2 14-3-3-γ X11/Mint

JIP1 Dab1 ShcA/Grb2 Numb

Fe65 binding to the AICD is unique, as its extended binding interface ranges from the 667VTPEER up to the 681GYENPTY motif and thus includes almost the entire AICD-C31 fragment (Figures 2 and 4A). Most other AICD interacting proteins recognize the 671GYENPTY motif and neighbouring residues, with the interaction site spanning only about 10 residues. As 681GYENPTY is essential for APP trafficking, the respective complexes can also alter APP processing. Like Fe65, the binders for this motif are PTB-containing proteins including members of the X11/Mint, JIP, Dab, and Shc families, as well as the Numb protein.

Mints consist of a divergent N-terminal region and conserved C-terminal sequences composed of one PTB domain and two tandem PDZ domains. Although their regulatory role for APP metabolism and transport is unresolved, it seems that they slow cellular APP processing and reduce Aβ40 and Aβ42 secretion [61] by suppressing translocation of APP into BACE- and γsecretase-rich detergent-resistant membrane (DRM) domains, the so-called rafts [62, 66]. In addition, there is evidence for a functional role of the AICD interaction with X11/Mints for synapse formation [62, 67] and synaptic neurotransmitter release [68]. c-Jun N-terminal kinase (JNK) interacting protein-1 (JIP1), a scaffolding protein for the JNK kinase cascade, has been suggested to mediate anterograde transport of APP by the molecular motor kinesin-1. However, this initial view has been challenged recently, as in contrast to this model, APP


constructs lacking the AICD are still transported to the nerve terminal by the fast axonal transport mechanism [63].

Knowledge about the interaction partners for the 667VTPEER motif is similar scarce. Major binder for the motif, and as well for the complete AICD, are the multi-domain adaptor/ scaffolding proteins of the Fe65 family (Fe65, Fe65L1, and Fe65L2) [49]. The only additional binding partner to the 667VTPEER motif is the dimeric adaptor protein 14-3-3γ, which seems to stabilize the AICD/Fe65 interaction [50]. Fe65 is enriched in brain, whereas Fe65L1 and Fe65L2 are more widely expressed. All three members contain a WW domain and two PTB domains (PTB1 and PTB2). Through the PTB2 domain, they interact with the AICD and can alter APP processing. After proteolytic processing of APP and release of the AICD to the cytoplasm, Fe65 can translocate to the nucleus to participate in gene transcription events (Figure 2), which is modulated by 14-3-3γ. This role is further mediated by interactions of Fe65- PTB1 with the transcription factors CP2/LSF/LBP1 [51] and Tip60 [52] and the WW domain with the nucleosome assembly factor SET [53]. Possible target genes identified by reporter assays include GSK3β, Neprilysin, KAI1, and the low-density lipoprotein receptor-related protein 1 (LRP1), but the physiological relevance for endogenous transcriptional regulation has been discussed controversially [54]. Fe65-PTB1 also interacts with two cell surface lipoproteins receptors, namely LRP1 [55] and ApoEr2 [56], forming trimeric complexes with APP. The Fe55 WW domain further binds to mammalian Ena (mEna) [57], through which it functions in regulation of the actin cytoskeleton, cell motility, and neuronal growth cone formation [49]. The interaction has been implicated in a role for AICD signaling, in synaptic plasticity and memory [58]. Moreover, Fe65 family proteins have attracted attention, as Fe65 or Fe65L1 double knockout mice revealed defects in cortical development with neuronal mispositioning and ectopia, resembling human lissencephaly type 2 [59]. Interestingly, very similar cortical defects were also found in APP-/-APLP1-/-APLP2-/- triple knockout mice lacking all APP family members, suggesting a lack of APP/Fe65 dependent signaling as the

Fe65 binding to the AICD is unique, as its extended binding interface ranges from the 667VTPEER up to the 681GYENPTY motif and thus includes almost the entire AICD-C31 fragment (Figures 2 and 4A). Most other AICD interacting proteins recognize the 671GYENPTY motif and neighbouring residues, with the interaction site spanning only about 10 residues. As 681GYENPTY is essential for APP trafficking, the respective complexes can also alter APP processing. Like Fe65, the binders for this motif are PTB-containing proteins including

Mints consist of a divergent N-terminal region and conserved C-terminal sequences composed of one PTB domain and two tandem PDZ domains. Although their regulatory role for APP metabolism and transport is unresolved, it seems that they slow cellular APP processing and reduce Aβ40 and Aβ42 secretion [61] by suppressing translocation of APP into BACE- and γsecretase-rich detergent-resistant membrane (DRM) domains, the so-called rafts [62, 66]. In addition, there is evidence for a functional role of the AICD interaction with X11/Mints for synapse formation [62, 67] and synaptic neurotransmitter release [68]. c-Jun N-terminal kinase (JNK) interacting protein-1 (JIP1), a scaffolding protein for the JNK kinase cascade, has been suggested to mediate anterograde transport of APP by the molecular motor kinesin-1. However, this initial view has been challenged recently, as in contrast to this model, APP

members of the X11/Mint, JIP, Dab, and Shc families, as well as the Numb protein.

underlying cause of defects in both mouse mutants [60].

10 Understanding Alzheimer's Disease

**Table 1. Selected interaction partners of the AICD.** \*Data depend on cell line studied and are sometimes

conflicting. \*\*Due to basolateral sorting and independent of PAT1 binding. Pat1 binding as such increases Aβ levels

[48]. \*\*\*Numb isoform dependent. ↓ denotes changes of non-amyloidogenic (α) or amyloidogenic (β) APP processing.

The Dab family member Dab1 regulates neuronal migration in mammals as an essential component of the Reelin signaling pathway. Dab1 binds not only to APP family proteins [64] but is well known to also bind to ApoE receptors (ApoEr2, VLDLR, and LRP) [69]. Dab1 increases cell surface expression of APP and ApoEr2, increases α-cleavage of APP and ApoEr2, and decreases APP βCTF formation and Aβ production in transfected cells and in primary neurons. The Dab family represents a prototype of PTB domains that bind their ligands in a pY-independent manner [36]. In addition Dab proteins bind specifically to the phosphoinositide (PI) PI-4,5-P2, which is predominantly located at the cellular membrane [70]. Binding of PTB domains to PIs is a common principle to locate and orientate the adaptors at the target membrane and to facilitate downstream events that accompany NPXY peptide binding. Since PTB domains structurally belong to the pleckstrin homology (PH) superfold family and PH domains are the prototypical PI binding domains, this function seems to be evolutionarily conserved within PTB domains [36]. The crystal structures of ternary complexes of Dabs bound to ApoEr2 [71] or APP [72] peptides and lipid revealed the lipid head group (IP3) to be recognized by a large basic patch opposite the peptide-binding groove (Figure 4A). This patch, also termed as "phospholipid binding-crown", is conserved in many PTB domains [36].

Finally, binding of the AICD to the Numb PTB domain has been found to inhibit Notch signaling [65], thereby establishing a crosstalk between the APP family and Notch in the development of the peripheral nervous system (PNS) [73]. Like APP, the Notch receptor undergoes a series of proteolytic cleavages that release the Notch intracellular domain (NICD) that functions in transcriptional activation and subsequent signal transduction events, including proliferation, differentiation, or apoptotic cues [74]. Similar to the NICD, the AICD has been also found to regulate PI-mediated calcium signaling through a γ-secretase depend‐ ent pathway [75, 76]. Cells lacking APP were shown to exhibit deficits in calcium storage that could be reversed by transfection with APP constructs containing an intact AICD. Constructs lacking the AICD were not able to rescue the phenotype, strongly indicating that this domain is critically involved in endoplasmic reticulum (ER) calcium filling [76]. The multitude of interactions with the AICD raises the question of the spatial and temporal regulation of all these complexes, which needs a detailed structural analysis and a thorough biochemical characterization.

<sup>667</sup>VTPEER<sup>672</sup>

V<sup>667</sup>

L<sup>674</sup>

E<sup>671</sup>

E<sup>670</sup>

P<sup>669</sup>

T<sup>668</sup>

R<sup>672</sup>

M<sup>675</sup>

<sup>681</sup>GYENPTY<sup>687</sup>

E<sup>683</sup>

F<sup>689</sup>

IP3

Structure and Function of the APP Intracellular Domain in Health and Disease

F<sup>690</sup>

T<sup>686</sup>

N<sup>684</sup> P<sup>685</sup>

Y<sup>687</sup>

Y<sup>682</sup>

http://dx.doi.org/10.5772/54543

13

β5

N<sup>680</sup>

G<sup>681</sup>

**Figure 4. Structure of the AICD in PTB domain complexes.** A. Crystal structures of AICD peptides in complex with PTB domains: Fe65-PTB2/AICD (PDB code 3DXC), X11α-PTB/AICD (1X11), and Dab1-PTB/AICD (1OQN). AICD peptides are colour coded from blue (N-terminus) to red (C-terminus) and PTB domains are given in gray. In Fe65-PTB2/AICD, the visible AICD structure corresponds to AICD-C31 and includes both the 667VTPEER and the 681GYENPTY sequences. Dab1 is also bound to the polar head group of the lipid PI-4,5-P2 (IP3: inositol-1,4,5-triphosphate). B. Superposition of the three AICD fragments as shown in Figure 3A (complex with Fe65-PTB2: red; X11α: blue; Dab1: green). The alterna‐ tive side chain conformations of Tyr682 are highlighted. C. Close-up view on the AICD helix αN in complex with Fe65- PTB2. The 667VTPEER motif is highlighted in blue and hydrogen bonds within the capping box are given as dashed lines. D. Interaction of the 680NGYE motif with Fe65-PTB2. The AICD stretch forms a β sheet *in trans* with strand β5 from the PTB domain. The side chain of Tyr682 is accommodated in a hydrophobic pocket created by the C-terminal helix of the PTB domain. E. Interaction of the 684NPTY motif and helix αC of the AICD with Fe65-PTB2. Tyr687 is rather solvent exposed and helix αC is fixed to the PTB domain by hydrophobic interactions of two subsequent phenylalanines.

As already described, the 684NPTY sequence is forming a type I β-turn structure, which is retained within the complexes and forms the N-terminal cap of an induced α-helix at the very C-terminus of AICD (helix αC) (Figure 4E). Asn684 has a conserved structural role, with the carboxamide of the side chain hydrogen bonding to the main chain of Thr686. As the carbox‐ amide is also tightly bonded to the PTB domains, the preformed NPTY conformation is a major determinant and probably also a starting point for AICD folding and complex formation. The conserved proline initiates and stabilizes the subsequent helix as found in many α helices. The most prominent residue, however, is Tyr687, as the tyrosine at this position is the discriminator for the classification in pY-dependent and pY-independent PTB domains [36]. In all structur‐ ally solved AICD/PTB domain complexes the peptide is non-phosphorylated, which reflects the *in vivo* situation within neurons. The pY-independence is readily explained, as the binding pocket is rather solvent exposed, and besides some van-der-Waals interactions of the benzene ring the tyrosine is not coordinated further. The binding mode is quite different in pY-

Y<sup>682</sup>

<sup>α</sup><sup>C</sup> <sup>α</sup><sup>N</sup>

<sup>A</sup> <sup>B</sup>

Fe65-PTB2/AICD X11α-PTB/AICD Dab1-PTB/AICD

<sup>C</sup> <sup>D</sup> <sup>E</sup>

#### **6. Structure-function relationship of AICD complexes**

The structure-function relationship of AICD complexes is governed by the one-to-many principle with the intrinsically disordered AICD folding onto its manifold adaptor proteins, in particular the PTB domain containing proteins. The recurrent interaction pattern includes the recognition of the 681GYENPTY sequence, which shall be described in the following. High resolution structures for this interaction are known for Dab1 and 2 [72], X11α [77], and the Fe65-PTB2 domains [30] (Figure 4A). All PTB domains comprise a pleckstrin homology (PH) fold consisting of a central β sandwich structure and a C-terminal α helix. Overall, complex formation can be described as an induced-fit docking of the AICD to a rigid PTB domain scaffold. Common to all the complexes is the binding of the 681GYEN sequence to the β5 strand of the respective PTB domain by a mechanism called β completion, where a (antiparallel) β sheet is created between two polypeptide chains (*in trans*) (Figure 4D). This interaction occurs between the protein backbones and therefore strong sequence conservation is not present on the PTB domain side. The conservation of AICD Gly681 is explained as longer side chains would cause steric clashes with the PTB domains, as shown for the Fe65-PTB2/AICD interac‐ tion, where a G681A mutation abolishes the binding and Gal4-Tip60-dependent transactiva‐ tion [78]. The importance of the flexible glycine becomes evident when comparing the solved PTB/AICD complexes (Figure 4B), revealing that Gly681 forms a hinge that allows for different AICD conformations in the N-terminal direction. The hinge function correlates with a peptideflip of the glycine [30].

The side chain of Tyr682 is accommodated in the center of the interface and faces the C-termi‐ nal helix of PTB domains (Figures 3A and 3D). In all complexes it lays in a hydrophobic pocket, however, the conformations between the Fe65-PTB2 and Dab1 in respect to X11α and Dab2 complexes are different. The hydrophobic nature of the pocket explains the general conserva‐ tion of a tyrosine or phenylalanine in this position in the context of NPXY sequences. All crys‐ tallized complexes are specific for non-phosphorylated Tyr682, which can be readily explained, as there is no space available to accommodate the extra phosphate moiety. This is in contrast to ShcA, where the binding site is more open [79], which apparently allows for binding of a phosphorylated Tyr682 (although no structure of this complex is available). The readout of the conserved glutamate is again different in the PTB complexes, although its function as selec‐ tivity filter seems to be minor. Whereas it forms i.e. a salt bridge with an arginine of X11α, in the Fe65-PTB2 complex it is fixed *in cis* to Lys688 following the NPTY motif.

ent pathway [75, 76]. Cells lacking APP were shown to exhibit deficits in calcium storage that could be reversed by transfection with APP constructs containing an intact AICD. Constructs lacking the AICD were not able to rescue the phenotype, strongly indicating that this domain is critically involved in endoplasmic reticulum (ER) calcium filling [76]. The multitude of interactions with the AICD raises the question of the spatial and temporal regulation of all these complexes, which needs a detailed structural analysis and a thorough biochemical

The structure-function relationship of AICD complexes is governed by the one-to-many principle with the intrinsically disordered AICD folding onto its manifold adaptor proteins, in particular the PTB domain containing proteins. The recurrent interaction pattern includes the recognition of the 681GYENPTY sequence, which shall be described in the following. High resolution structures for this interaction are known for Dab1 and 2 [72], X11α [77], and the Fe65-PTB2 domains [30] (Figure 4A). All PTB domains comprise a pleckstrin homology (PH) fold consisting of a central β sandwich structure and a C-terminal α helix. Overall, complex formation can be described as an induced-fit docking of the AICD to a rigid PTB domain scaffold. Common to all the complexes is the binding of the 681GYEN sequence to the β5 strand of the respective PTB domain by a mechanism called β completion, where a (antiparallel) β sheet is created between two polypeptide chains (*in trans*) (Figure 4D). This interaction occurs between the protein backbones and therefore strong sequence conservation is not present on the PTB domain side. The conservation of AICD Gly681 is explained as longer side chains would cause steric clashes with the PTB domains, as shown for the Fe65-PTB2/AICD interac‐ tion, where a G681A mutation abolishes the binding and Gal4-Tip60-dependent transactiva‐ tion [78]. The importance of the flexible glycine becomes evident when comparing the solved PTB/AICD complexes (Figure 4B), revealing that Gly681 forms a hinge that allows for different AICD conformations in the N-terminal direction. The hinge function correlates with a peptide-

The side chain of Tyr682 is accommodated in the center of the interface and faces the C-termi‐ nal helix of PTB domains (Figures 3A and 3D). In all complexes it lays in a hydrophobic pocket, however, the conformations between the Fe65-PTB2 and Dab1 in respect to X11α and Dab2 complexes are different. The hydrophobic nature of the pocket explains the general conserva‐ tion of a tyrosine or phenylalanine in this position in the context of NPXY sequences. All crys‐ tallized complexes are specific for non-phosphorylated Tyr682, which can be readily explained, as there is no space available to accommodate the extra phosphate moiety. This is in contrast to ShcA, where the binding site is more open [79], which apparently allows for binding of a phosphorylated Tyr682 (although no structure of this complex is available). The readout of the conserved glutamate is again different in the PTB complexes, although its function as selec‐ tivity filter seems to be minor. Whereas it forms i.e. a salt bridge with an arginine of X11α, in the

Fe65-PTB2 complex it is fixed *in cis* to Lys688 following the NPTY motif.

**6. Structure-function relationship of AICD complexes**

characterization.

12 Understanding Alzheimer's Disease

flip of the glycine [30].

**Figure 4. Structure of the AICD in PTB domain complexes.** A. Crystal structures of AICD peptides in complex with PTB domains: Fe65-PTB2/AICD (PDB code 3DXC), X11α-PTB/AICD (1X11), and Dab1-PTB/AICD (1OQN). AICD peptides are colour coded from blue (N-terminus) to red (C-terminus) and PTB domains are given in gray. In Fe65-PTB2/AICD, the visible AICD structure corresponds to AICD-C31 and includes both the 667VTPEER and the 681GYENPTY sequences. Dab1 is also bound to the polar head group of the lipid PI-4,5-P2 (IP3: inositol-1,4,5-triphosphate). B. Superposition of the three AICD fragments as shown in Figure 3A (complex with Fe65-PTB2: red; X11α: blue; Dab1: green). The alterna‐ tive side chain conformations of Tyr682 are highlighted. C. Close-up view on the AICD helix αN in complex with Fe65- PTB2. The 667VTPEER motif is highlighted in blue and hydrogen bonds within the capping box are given as dashed lines. D. Interaction of the 680NGYE motif with Fe65-PTB2. The AICD stretch forms a β sheet *in trans* with strand β5 from the PTB domain. The side chain of Tyr682 is accommodated in a hydrophobic pocket created by the C-terminal helix of the PTB domain. E. Interaction of the 684NPTY motif and helix αC of the AICD with Fe65-PTB2. Tyr687 is rather solvent exposed and helix αC is fixed to the PTB domain by hydrophobic interactions of two subsequent phenylalanines.

As already described, the 684NPTY sequence is forming a type I β-turn structure, which is retained within the complexes and forms the N-terminal cap of an induced α-helix at the very C-terminus of AICD (helix αC) (Figure 4E). Asn684 has a conserved structural role, with the carboxamide of the side chain hydrogen bonding to the main chain of Thr686. As the carbox‐ amide is also tightly bonded to the PTB domains, the preformed NPTY conformation is a major determinant and probably also a starting point for AICD folding and complex formation. The conserved proline initiates and stabilizes the subsequent helix as found in many α helices. The most prominent residue, however, is Tyr687, as the tyrosine at this position is the discriminator for the classification in pY-dependent and pY-independent PTB domains [36]. In all structur‐ ally solved AICD/PTB domain complexes the peptide is non-phosphorylated, which reflects the *in vivo* situation within neurons. The pY-independence is readily explained, as the binding pocket is rather solvent exposed, and besides some van-der-Waals interactions of the benzene ring the tyrosine is not coordinated further. The binding mode is quite different in pY- dependent Shc or IRS1 peptide complexes, where the phosphate moiety is read out by a set of conserved arginine residues and the binding pocket is much more pronounced [36].

the majority of APP locates to the Golgi apparatus and trans-Golgi network [10]. APP not shed at the surface is internalized within minutes [82], delivered to endosomes, and if not degraded in lysosomes recycled to the cell surface [83]. AICD is even more difficult to study, as due to its small size it is rapidly degraded once it is released from the membrane by the insulin degrading enzyme (IDE) [84], that also degrades the Aβ peptide, by the proteasome [85], or by the endosomal/lysosomal system [86]. However, AICD found in the nucleus appears to be more stable, suggesting that AICD involved in signal transduction escapes rapid degradation [87]. Nuclear AICD is stabilized via interaction with Fe65 [88, 89], which accordingly has a dominant function in AICD mediated physiological and pathophysiological processes.

Structure and Function of the APP Intracellular Domain in Health and Disease

http://dx.doi.org/10.5772/54543

15

From a structural viewpoint it is evident that the enlarged and unique protein-protein interface coupled with high affinity binding prevents the AICD from degradation. Interestingly, AICD-C31 (starting at Ala665), which is believed to induce apoptosis and is enriched in AD brains [34], fits exactly in length with the AICD part interacting with Fe65-PTB2. Hence, two scenarios comprising a modulating role for Fe65 in AICD-C31 mediated neurotoxicity might be envis‐ aged: (i), under physiological conditions Fe65 protects the AICD from caspase cleavage occurring at Asp664 and might therefore inhibit apoptosis as shown previously [90] and (ii), increased levels of AICD-C31 compete with AICD binding as part of full-length APP and therefore interfere with physiological Fe65 functions including nuclear signaling and traffick‐ ing of APP. In any case, modifying the protein-interacting network around the AICD seems

Despite enormous efforts to develop an efficient treatment for AD, only symptomatic treat‐ ments with modest impact on the progress of the disease are available [6]. Drugs currently approved for the treatment of AD are either acetylcholine esterase inhibitors to increase the level of the neurotransmitter, which is depleted in AD brains, or antagonize the NMDA receptor to prevent abnormal neuronal stimulation [91]. None of them directly targets the amyloid cascade and would thereby allow for a disease-modifying treatment. Many current therapeutic approaches for AD focus on the reduction of the Aβ load either by inhibiting the involved secretases BACE and γ-secretase, or by augmenting the elimination of amyloid peptides, e.g. by active or passive immunotherapy [6]. Finally, a smaller number of trials have targeted ApoE4 levels or either tau phosphorylation or tau aggregation. None of the ap‐ proaches was successful so far, which means that either there were not enough clinical trials or the ideas were too simplistic to be potent for a complex disease. Like for other complex diseases (i.e. hypertension or AIDS), a combination of drugs that have different modes of action

In this sense, the AICD might be re-evaluated as a potential drug target. In contrast to Aβ, the AICD is a physiological highly relevant protein domain modulating a diverse set of important APP functions including trafficking and signal transduction. As both proc‐ esses are also directly affecting Aβ production, upstream targeting of AICD might be

to be a valid target for decreasing neurotoxicity and the treatment of AD.

**8. Conclusion**

could be the key to success.

The NPTY sequence is followed by the 688KFFEQMQN695 sequence, which forms the C-termi‐ nus of the AICD (Figures 4A and 4E). The conformation of this region is slightly different and not always present in the structures, as the complexes have mostly been formed with truncated synthetic peptides. In the Fe65-PTB2 (which contains the entire C-terminus) and X11α com‐ plexes, the region is part of the C-terminal helix αC. The helix is fixed to the PTB domains by hy‐ drophobic interactions of the two phenylalanines (Phe689 and Phe690) with the C-terminal helices of the respective PTB domains. These helices are three turns longer than those of Shc [79] and IRS1 [80] PTBs, and therefore the total interaction surfaces are significantly larger.

In most PTB domain complexes bound to an NPXY motif the described surfaces comprise the entire interaction, however, there is a single exception to the rule: the Fe65-PTB2/AICD complex, where the interface is about three times as large and includes an additional α helix (helix αN, 669PEERHLSKMQQ679) N-terminal to the 681GYENPTY sequence (Figure 4C) [30]. This helix is N-terminally capped by the 667VTPEER motif comprising the phosphorylatable Thr668 as already described. Like helix αC, helix αN is of amphipathic character and binds on a hydrophobic patch on the Fe65-PTB2 surface located in between strand β5 and the Nterminus of the C-terminal helix, which is almost perpendicularly crossed by helix αN. Whereas Leu674 and Met677 cover the hydrophobic patch, Glu670, His673, and Gln678 are involved in polar interactions with the PTB domain. With the exception of Glu670, the 667VTPEER capping box is not touching the PTB domain, which is somewhat astonishing, as it was afore known that phosphorylation of Thr668 is detrimental to complex formation [20]. As described for free AICD, the side chain of Thr668 is hydrogen-bonded to the main chain of Glu671, and Pro668 is *in trans* configuration. Furthermore, the side chain of Glu671 is tied back to the main chain nitrogen of Thr668, and thus completing the rigid helix cap.

The most important question arising from structural data is how phosphorylation is able to regulate Fe65-PTB2/AICD complex formation in a process that is critically involved in Aβ gen‐ eration and AD pathogenesis? Phosphorylation induces a *cis* configuration of Pro669 [46], which is incompatible with the formation of helix αN. As found by mutational studies [30], the destruction of the helix cap increases the entropy of the system and reduces the binding affini‐ ty, and once the helix is dissolved, the remaining interfaces are not sufficient for maintaining the complex. This molecular switch model is only valid for the Fe65-PTB2/AICD interaction, as all other PTB domains do not contact Thr668 and phosphorylation does therefore not alter their binding affinity [20]. Intriguingly, the Fe65-PTB2/AICD interface spans almost the entire AICD-C31 fragment, which has been implicated in apoptotic events. This raises the next ques‐ tion: what determines stability, lifetime, and eventually toxicity of the AICD?
