Cell-fate decisions in metazoans are frequently guided by the Notch signalling pathway. Notch signalling is orchestrated by a type-1 transmembrane protein, which, upon interacting with extracellular ligands, is proteolytically cleaved to liberate a large intracellular domain [NICD (Notch intracellular domain)]. NICD enters the nucleus where it binds the transcription factor CSL (CBF1/suppressor of Hairless/Lag-1) and activates transcription of Notch-responsive genes. In the present paper, the interaction between the Drosophila NICD and CSL will be examined. This interaction involves two separate binding regions on NICD: the N-terminal tip of NICD {the RAM [RBP-Jκ (recombination signal-binding protein 1 for Jκ)-associated molecule] region} and an ankyrin domain ∼100 residues away. CD studies show that the RAM region of NICD lacks α-helical and β-sheet secondary structure, and also lacks rigid tertiary structure. Fluorescence studies show that the tryptophan residues in RAM are highly solvated and are quenched by solvent. To assess the impact of this apparent disorder on the bivalent binding of NICD to CSL, we modelled the region between the RAM and ANK (ankyrin repeat)-binding regions using polymer statistics. A WLC (wormlike chain) model shows that the most probable sequence separation between the two binding regions is ∼50 Å (1 Å=0.1 nm), matching the separation between these two sites in the complex. The WLC model predicts a substantial enhancement of ANK occupancy via effective concentration, and suggests that the linker length between the two binding regions is optimal for bivalent interaction.

Introduction

The Notch pathway is a transmembrane signal transduction pathway used in cell-fate determination in metazoan development, homoeostasis and stem-cell maintenance and differentiation. Through Notch signalling, neighbouring cells determine each other's fates, producing complex differentiated tissues. Components of the pathway were first identified in Drosophila melanogaster through genetic studies [1]. Further classical and molecular genetic and cell biological studies in Drosophila, Caenorhabditis elegans, Xenopus laevis and mammals have defined an elaborate (and growing) network of interactions among Notch pathway components, and have helped define the biochemical events that control signalling [25].

Determining the molecular mechanisms of Notch signalling is important for understanding a number of human diseases. Some Notch-related diseases affect early development, such as spondylocostal dysostosis and Alagille's syndrome (resulting from mutations in Delta-3 [6] and Jagged-1 ligands [7]), and Noonan's syndrome (for which mutations map to the Deltex-1 locus [8]). Other Notch-related diseases appear to impair neurological function in adults [such as CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy), resulting from mutations in the Notch3 receptor [9]. In addition, mutations in Notch pathway genes are associated with a variety of cancers, including T-cell acute lymphoblastic leukaemia [10], breast cancer [11] and B-cell lymphoma [12]. Finally, the Notch pathway is used in the infectivity cycles of a number of opportunistic viruses, including Epstein–Barr virus, adenovirus and Kaposi's sarcoma-related herpesvirus [13].

The Notch receptor

At the heart of Notch signalling is a family of single-pass transmembrane receptors encoded by the Notch genes (Figure 1) [14,15]. Whereas Drosophila only has one copy of the Notch gene, C. elegans encodes two paralogous genes, and mammals have four (termed Notch1Notch4). Notch receptors have a common domain/sequence motif organization. The Drosophila Notch gene (Figure 1) encodes a 2703-residue protein, of which 1745 residues are predicted to be extracellular [15]. The N-terminal extracellular portion contains 36 EGF repeats, and three additional LN (Leu-Asn) repeats. The intracellular 937 residues contain a highly conserved ANK (ankyrin repeat) domain (240 residues), beginning approx. 140 residues C-terminal to the putative transmembrane segment. The ∼140 residues between the transmembrane segment and the ANK domain, referred to as the RAM [RBP-Jκ (recombination signal-binding protein 1 for Jκ)-associated molecule] segment, shows only modest conservation among Notch proteins (Figure 2). The region C-terminal to the ANK domain includes an NLS (nuclear localization sequence), a transactivation domain and a PEST (Pro-Glu-Ser-Thr) sequence (involved in proteolytic degradation).

The Notch signalling pathway

Figure 1
The Notch signalling pathway

Following binding of extracellular ligand to the Notch receptor (shown here as a single chain for simplicity), two proteolytic cleavages release NICD in the receiving cell, providing access to the nucleus and complex formation with CSL (see the text for details). ADAM, a disintegrin and metalloproteinase; E(spl), Enhancer of split.

Figure 1
The Notch signalling pathway

Following binding of extracellular ligand to the Notch receptor (shown here as a single chain for simplicity), two proteolytic cleavages release NICD in the receiving cell, providing access to the nucleus and complex formation with CSL (see the text for details). ADAM, a disintegrin and metalloproteinase; E(spl), Enhancer of split.

The ANK and RAM regions of the Notch receptor

Figure 2
The ANK and RAM regions of the Notch receptor

(A) The ANK domain, coloured by repeats (PDB code 1OT8 [28]), and some interactions with effectors in the receiving cell. Left, the two WWE (Trp-Trp-Glu) modules of the ANK-binding domain of Deltex (orange and purple [53]). Right, a RAM–ANK homologue from C. elegans complexed with CSL [32]. The WXP-containing peptide is shown in black bound to the BTD (light pink), whereas ANK binds NTD and CTD (light blue) with MAM (grey helix). FIH, factor inhibiting HIF (hypoxia-inducible factor); Nrarp, Notch-regulated ankyrin repeat protein. (B) Sequence alignment (ClustalW) of the RAM regions of divergent Notch homologues shows low sequence similarity, especially in the central linker region. (C) Distribution of linker lengths between WXP and ANK (red) for the Notch homologues in (B). The grey distribution shows a much more heterogeneous distribution of lengths from ANK to the C-terminus.

Figure 2
The ANK and RAM regions of the Notch receptor

(A) The ANK domain, coloured by repeats (PDB code 1OT8 [28]), and some interactions with effectors in the receiving cell. Left, the two WWE (Trp-Trp-Glu) modules of the ANK-binding domain of Deltex (orange and purple [53]). Right, a RAM–ANK homologue from C. elegans complexed with CSL [32]. The WXP-containing peptide is shown in black bound to the BTD (light pink), whereas ANK binds NTD and CTD (light blue) with MAM (grey helix). FIH, factor inhibiting HIF (hypoxia-inducible factor); Nrarp, Notch-regulated ankyrin repeat protein. (B) Sequence alignment (ClustalW) of the RAM regions of divergent Notch homologues shows low sequence similarity, especially in the central linker region. (C) Distribution of linker lengths between WXP and ANK (red) for the Notch homologues in (B). The grey distribution shows a much more heterogeneous distribution of lengths from ANK to the C-terminus.

Molecular mechanisms of Notch signalling

The best-understood pathway by which Notch signalling occurs is often referred to as ‘canonical Notch signalling’ (Figure 1). Notch receptors in signal-receiving cells bind extracellular transmembrane protein ligands of the DSL (Delta, Serrate, Lag) class attached to transmitting cells. Ligand binding activates two proteolytic cleavages, one just outside the cell membrane mediated by an ADAM (a disintegrin and metalloproteinase) protease [1618], and a second just inside the cell membrane mediated by the γ-secretase–presenillin complex [19,20]. The second cleavage frees the intracellular portion of the Notch receptor [referred to as NICD (Notch intracellular domain)] from the membrane, allowing it to enter the nucleus via its NLS. In the nucleus, NICD binds to CSL (CBF1/suppressor of Hairless/Lag-1), a transcription factor bound to specific sequences upstream of genes activated by canonical Notch signalling [21]. Binding of NICD to CSL triggers a transcriptional switch, displacing transcriptional repressors, recruiting co-activator proteins, including Mastermind (MAM), and initiating transcription in the receiving cell [22].

Structure and interactions of the Notch ANK domain

In addition to the CSL transcription factor, MAM and other nuclear co-activators and co-repressors, a large number of modifiers of Notch signalling have been identified through genetic and biochemical approaches. Many of these are cytosolic proteins that appear, by yeast two-hybrid, co-localization or co-precipitation studies, to interact with NICD. Within NICD, the ANK domain appears to be a ‘hotspot’ for these interactions. Consistent with its role as a hub for protein–protein interactions, the ANK domain has been shown to be critical for signalling in molecular genetic studies. For example, in C. elegans, the ANK domain is sufficient to activate Notch [23,24].

ANK domains are 33-residue-repeat sequences that contain a pair of antiparallel α-helices capped by a short β-turn. ANK domains are found in tandem arrays, usually with a minimum of four, and as many as 30, repeats [25]. Initially, sequence analysis identified six ANK domains in NICD. However, a combination of thermodynamic and hydrodynamic analysis indicated that the Notch receptor has an additional seventh ANK domain [26,27]. X-ray crystallography studies from our laboratory confirmed that, indeed, the Notch ANK domain is structured to its C-terminal seventh repeat (Figure 2A). However, when not complexed with binding partners, the first (N-terminal) sequence repeat is largely disordered [28,29], a finding that was initially predicted from thermodynamic analysis [30].

Although the modular architecture of ANK might predict a conformational heterogeneity in which repeats fold independently of one another, thermodynamic studies from our laboratory have clearly shown the Notch ANK domain to undergo folding–unfolding transitions in a single all-or-none reaction, despite its modular architecture [26,30]. Surprisingly, even the partly disordered N-terminal repeat contributes to stability, as deletion of the first repeat of the Drosophila ANK domain decreases stability (represented as the free energy of unfolding, ΔG°) by more than 1 kcal/mol (1 kcal=4.184 kJ) [31]. Recently, two beautiful crystal structures have demonstrated that, when complexed with CSL, the first repeat of the Notch ANK domain becomes structured, adopting a canonical ANK fold and stacking on to the ANK domain array [32,33]. These results demonstrate the importance of both structured and disordered regions of the Notch receptor in interacting with effector proteins, as well as binding-mediated conformational transitions between disorder and order.

Interactions involving the RAM region of the Notch receptor

Immediately N-terminal to the ANK domain lies another important interaction region of the Notch receptor, loosely referred to as the RAM region. This region, which spans approx. 140 residues from the transmembrane segment (and S3 cleavage site) to the first ANK domain, has been implicated in interaction with CSL [3436]. In addition, reports have implicated interaction of the RAM region with the cytosolic/membrane-associated proteins Numb and Disabled [37,38]. Unlike the ANK domain, Notch homologues show low sequence similarity through the RAM region (Figure 2B), with the exception of two short stretches of sequence near the transmembrane segment the ANK domain. The sequence-distal (relative to ANK) region, which includes a conserved WXP (Trp-Xaa-Pro) tripeptide, has been demonstrated to be important for interaction with CSL [34,36], and has been shown to bind directly to the BTD (β-trefoil domain) of CSL [32,36] (Figure 2A).

We have recently probed the energetics of interaction of the WXP region of RAM with mammalian CSL using ITC (isothermal titration calorimetry) [39]. Most of the affinity (Kd in the range 0.2–1 μM) of interaction of RAM (and RAM–ANK constructs) with BTD is maintained in short (∼16 residue) peptide segments centred on the WXP sequence. Substitution of the conserved tryptophan and proline residues substantially decreases affinity; however, studies of a variant peptide derived from Notch2 show that a WXP sequence is not sufficient for binding, suggesting additional sequence determinants proximal to the WXP sequence [39]. This observation is consistent with the substantial sequence conservation surrounding the WXP motif; however, in this context, the similar binding affinities of WXP-containing peptides derived from four different mammalian (mouse) Notch receptors, which show some sequence variation, is somewhat unexpected [39].

Although the WXP motif of the RAM region engages CSL with reasonably high affinity, it is not the only region of NICD to interact. In the recent structures of CSL complexed with MAM and either ANK [33] or RAM–ANK [32], there is substantial interaction between the ANK domain and the N- and C-terminal domains of CSL (NTD and CTD respectively) (Figure 2A). This interaction, which is buttressed structurally by the extended α-helical MAM polypeptide, is consistent with earlier biochemical studies showing that, although RAM can bind independently to CSL, the binding of ANK requires MAM [35], and helps to explain earlier studies that suggest involvement of both regions of NICD in interaction with CSL [23,34,40,41]. This bivalent mode of interaction between NICD and CSL involves considerable separation, both in terms of primary sequence for NICD [∼100 residues separate WFP (Trp-Phe-Pro) from ANK, see below] and in structural terms, with approx. 50 Å (1 Å=0.1 nm) separation between the two binding sites on CSL [32].

Solution structure of the RAM region

Given the bivalent mode by which NICD binds CSL (through both ANK and the WXP motif of RAM), what is the status of the intervening RAM region? As mentioned above, the sequence conservation between the WXP and ANK motifs is rather low, and electron density is only visible for segments of RAM near to these two anchor-points for binding [32]. To get a better understanding of the structural and energetic consequences of this bivalent arrangement, we have been using a variety of spectroscopic methods to explore the conformational preferences of the RAM region in solution.

CD spectroscopy provides one means to assess the secondary and tertiary structure of polypeptides in dilute solutions. The far-UV spectrum of the Drosophila RAM region shows little α-helical or extended β-sheet structure, as is evident by a lack of ellipticity at 222 or 215–217 nm respectively (Figure 3A, blue). This low level of hydrogen-bonded secondary structure is independent of the ANK domain, as the CD spectrum of RAM–ANK (Figure 3A, black) is exactly the sum of the spectra of the separate RAM and ANK domains (when corrected for the different chain lengths of the two constructs). Similar findings have been reported for the human Notch1 RAM region [35].

CD spectroscopy of the Drosophila Notch RAM and ANK polypeptides

Figure 3
CD spectroscopy of the Drosophila Notch RAM and ANK polypeptides

(A) Far-UV and (B) near-UV spectra of RAM–ANK (black), ANK (red), RAM (blue), the weighted sum of RAM and ANK spectra (purple) and RAM–ANK W4F (green). (C) Far-UV spectra of RAM–ANK in the absence (––) and presence (—) of LUVs containing PC:PS at a 1:1 molar ratio. LUVs were prepared by extrusion of an aqueous suspension of phospholipids (Avanti Polar Lipids) through a mini-extruder, and were added to a final concentration of 2.5 mg/ml. Instrumental details can be found in [26,27]. Conditions: 25 mM Tris/HCl and 150 mM NaCl, at pH 8 (RAM at pH 7) and 20°C; 10 μM protein (A and C), 20–50 μM protein (B).

Figure 3
CD spectroscopy of the Drosophila Notch RAM and ANK polypeptides

(A) Far-UV and (B) near-UV spectra of RAM–ANK (black), ANK (red), RAM (blue), the weighted sum of RAM and ANK spectra (purple) and RAM–ANK W4F (green). (C) Far-UV spectra of RAM–ANK in the absence (––) and presence (—) of LUVs containing PC:PS at a 1:1 molar ratio. LUVs were prepared by extrusion of an aqueous suspension of phospholipids (Avanti Polar Lipids) through a mini-extruder, and were added to a final concentration of 2.5 mg/ml. Instrumental details can be found in [26,27]. Conditions: 25 mM Tris/HCl and 150 mM NaCl, at pH 8 (RAM at pH 7) and 20°C; 10 μM protein (A and C), 20–50 μM protein (B).

Near-UV CD spectroscopy probes the extent of rigid tertiary structure around aromatic side chains (tyrosine, tryptophan and phenylalanine). Strong near-UV signals require aromatic groups to be maintained in close proximity and to adopt specific orientations between their electronic transition dipoles [42]. Unlike far-UV CD studies on the peptide backbone, it is not possible to determine what kind of structure a protein with near-UV optical activity possesses, but rather that a polypeptide either possesses or lacks tertiary structure. The Notch ANK domain, with rigid tertiary structure, shows substantial CD in the near-UV region, consistent with rigid tertiary structure around its single tryptophan and tyrosine (and four phenylalanine) residues (Figure 3B, red). Much of this signal involves coupling to the single tryptophan residue, as replacement with phenylalanine eliminates most of the near-UV CD (not shown). In contrast, the RAM region shows no detectable near-UV CD in solution (Figure 3B, blue) despite having three tryptophan, two tyrosine and four phenylalanine residues with roughly uniform sequence distribution. As with the far-UV spectrum, this lack of optical activity in the near-UV region is independent of ANK, since the near-UV spectrum of RAM–ANK (Figure 3B, black) is identical with that of ANK (and thus to the sum of RAM and ANK); likewise, a RAM–ANK construct with the single ANK tryptophan residue replaced with phenylalanine lacks near-UV CD, as does RAM alone (Figure 3B, green).

In addition to serving as probes for tertiary structure, aromatic residues in proteins can provide information on local solvation and dynamics [43]. The tryptophan residues in RAM and ANK are particularly well-distributed to probe the local structure in NICD, as three tryptophan residues are roughly equally spaced along RAM, and the fourth tryptophan residue is located in the fifth repeat of the ANK domain (Figure 4A). The fluorescence emission spectrum of ANK, when excited at 295 nm (a wavelength that stimulates tryptophan fluorescence exclusively), has a maximum at 335 nm (Figure 4B, red curve), indicating that the tryptophan residue is substantially shielded from solvent, and is in an environment of decreased polarity. This interpretation is consistent with the crystal structure of the Drosophila Notch ANK domain [28].

Tryptophan fluorescence spectroscopy of Drosophila Notch RAM and ANK polypeptides

Figure 4
Tryptophan fluorescence spectroscopy of Drosophila Notch RAM and ANK polypeptides

(A) Schematic diagrams of the locations of the four tryptophan residues in RAM and ANK, and constructs presented here, including tryptophan-to-phenylalanine substitutions. (B) Emission spectra resulting from excitation at 295 nm. RAM–ANK (black), ANK (red), W4F (blue) and an exposed tryptophan model (ANK with its single tryptophan moved from repeat 5 to an unstructured site on the N-terminus; green). Fluorescence spectra were collected with an SLM 48000 fluorimeter, with excitation at 295 nm. Spectra were normalized to the same concentration, were collected using identical instrument settings, and were corrected for inner-filter effects so that absolute intensities could be compared. Conditions are as in Figure 3.

Figure 4
Tryptophan fluorescence spectroscopy of Drosophila Notch RAM and ANK polypeptides

(A) Schematic diagrams of the locations of the four tryptophan residues in RAM and ANK, and constructs presented here, including tryptophan-to-phenylalanine substitutions. (B) Emission spectra resulting from excitation at 295 nm. RAM–ANK (black), ANK (red), W4F (blue) and an exposed tryptophan model (ANK with its single tryptophan moved from repeat 5 to an unstructured site on the N-terminus; green). Fluorescence spectra were collected with an SLM 48000 fluorimeter, with excitation at 295 nm. Spectra were normalized to the same concentration, were collected using identical instrument settings, and were corrected for inner-filter effects so that absolute intensities could be compared. Conditions are as in Figure 3.

In contrast, the emission spectrum of RAM–ANK is shifted to a longer wavelength (340 nm) (Figure 4B, black curve), suggesting an increase in overall polarity surrounding the four tryptophan residues. Although this increase in emission wavelength is rather modest compared with the value expected for fully hydrated tryptophan side chains (∼350–360 nm) (Figure 4B, green curve), the three tryptophan residues in RAM may be substantially quenched if they have high solvent accessibility, which would decrease their overall contribution to the emission spectrum. Consistent with this, the integrated intensity of the emission spectrum of RAM–ANK is only 1.6 times that of the ANK spectrum, despite having four times as many tryptophan chromophores. Interpretation of the emission spectrum of RAM–ANK, as resulting from an unperturbed (from ANK) tryptophan residue with a 335 nm maximum and three tryptophan residues (from RAM) with substantially red-shifted but quenched fluorescence, is consistent with the overall shape of the RAM–ANK spectrum, which shows a peak near that of the ANK spectrum and a shoulder at around 350 nm. To confirm this interpretation, the tryptophan residue from ANK was replaced with a phenylalanine residue (Figure 4A, W4F). Indeed, the fluorescence emission spectrum of RAM–ANK W4F (blue curve, Figure 4B) has an emission maximum of 350 nm, with an intensity roughly equal to that for the single tryptophan residue of ANK, despite having three tryptophan residues in the RAM region. Together, these findings suggest that, unlike the ANK domain, the RAM region lacks a rigid tertiary structure and is instead substantially solvated along its length.

As described above, binding of RAM to CSL causes a limited amount of rigid structure to be formed in the region of WFP and also near to the ANK domain [32]. In the unbound state, the RAM region is not entirely free in solution, but spends much of its time in close proximity to the cell membrane, being adjacent to the transmembrane region of the receptor. To examine whether the presence of lipid bilayers induces structure in RAM, we compared the far-UV CD spectra of RAM–ANK in the presence of LUVs (large unilamellar vesicles) made of PC (phosphatidylcholine) and PS (phosphatidylserine) (Figure 3C). On the basis of the shape and magnitude of the far-UV CD spectra, there appears to be no change in secondary structure upon addition of LUVs.

Thermodynamic independence of the RAM and ANK regions of NICD

In addition to showing that RAM possesses very little rigid tertiary structure in the unbound state, the solution spectroscopy studies above indicate that RAM and ANK domains do not influence each other's structure. A more accurate way to examine the degree to which RAM and ANK influence each other is thermodynamic analysis of unfolding. When unfolding transitions are well resolved and can be approximated as a conversion between two thermodynamic states, unfolding midpoint and stabilities can be determined with very high precision. Here, we compare urea-induced unfolding of RAM and RAM–ANK constructs, and examine the sensitivity of unfolding to point substitutions in RAM and ANK.

Like ANK, the urea-induced unfolding transition of RAM–ANK, as monitored by CD spectroscopy in the α-helical region, is very sharp, with a precisely defined midpoint of 2.43±0.03 M (Figure 5A), which is very similar to that determined for ANK (2.50±0.03 M). When converted into an unfolding free energy via a two-state model, unfolding free energies for RAM–ANK and ANK are nearly identical (6.95 and 6.85 kcal·mol−1 respectively). Similar results were obtained by tryptophan fluorescence (not shown). Given that the change in CD signal results from structural changes in the α-helices of ANK, the similarity of these unfolding energies indicates that RAM and ANK are conformationally independent of one another. Considering the high sensitivity of the unfolding transition of the ANK domain to the presence of the first ANK domain, which is partly disordered in the unbound state, such independence is remarkable. Similar independence has been observed previously in thermal denaturation of human Notch1 [35]. In contrast, the RAM region alone shows absolutely no sign of any structural transition from 0 to 6 M urea (Figure 5A).

Urea-induced denaturation of Drosophila Notch RAM and ANK polypeptides monitored by CD spectroscopy

Figure 5
Urea-induced denaturation of Drosophila Notch RAM and ANK polypeptides monitored by CD spectroscopy

(A) Unfolding transitions monitored by CD at 222 nm for RAM, RAM–ANK and ANK. (B) Normalized unfolding transitions of ANK-containing constructs, including tryptophan-to-phenylalanine mutations (see Figure 4A). Curves resulted from fitting a two-state unfolding transition to the data. Conditions: 2 μM protein, 25 mM Tris/HCl and 150 mM NaCl, at pH 8 and 20°C.

Figure 5
Urea-induced denaturation of Drosophila Notch RAM and ANK polypeptides monitored by CD spectroscopy

(A) Unfolding transitions monitored by CD at 222 nm for RAM, RAM–ANK and ANK. (B) Normalized unfolding transitions of ANK-containing constructs, including tryptophan-to-phenylalanine mutations (see Figure 4A). Curves resulted from fitting a two-state unfolding transition to the data. Conditions: 2 μM protein, 25 mM Tris/HCl and 150 mM NaCl, at pH 8 and 20°C.

As another example of the conformational independence of RAM and ANK, we find the effect of point substitutions that we have made to isolate the contributions of individual tryptophan residues to fluorescence (Figure 4) (X. Wen, D. Toptygin and L. Brand, unpublished work) on unfolding energy to be autonomous to the RAM and ANK region (Figure 5B). Replacement of the tryptophan residue in ANK with phenylalanine substantially decreases the stability of the ANK domain, and the extent of destabilization is independent of whether RAM is present or not (ΔΔG° of 1.95 and 2.00 kcal·mol−1 respectively). In contrast, replacement of the tryptophan residues in the RAM region of RAM–ANK with phenylalanine have no effect on the conformational transition (Figure 5B), consistent with the idea that the RAM region is not involved (either directly or indirectly) in the conformational transitions of ANK.

Implications of NICD disorder on binding to CSL

Binding reactions involving conformational disorder in one or both partners have been described for a number of proteins, and can have profound implications for binding and allosteric control [4446]. In most well-studied examples, disorder is lost upon binding. For the RAM region, the spectroscopic and thermodynamic studies above indicate that, in the unbound state, the RAM region of NICD lacks rigid tertiary structure, and its non-polar tryptophan residues are substantially solvated; moreover, crystallographic data are consistent with a considerable amount of disorder in the bound state [32]. Because the disordered region of RAM separates two distinct binding sites for CSL (WXP and ANK), the properties of the conformational ensemble are critical for determining the overall affinity. For example, the distribution of distances between the WXP motif and ANK domain determines the joint probability that these two regions will be able to engage their respective sites on CSL, as well as the dynamics of interaction. To understand the contribution of this region of RAM to NICD–CSL, a statistical polymer model must be used instead of a rigid well-defined structure.

One relatively simple model that can be used to model a disordered polypeptide chain is the WLC (wormlike chain) [42,47]. In this model, the polypeptide is treated as a continuous cylinder with a fixed (but randomly directed) radius of curvature. The only two variables in the WLC model are the contour length [lc, which is the distance per residue (3.8 Å) times the number of residues] and the persistence length (lp, the amount of chain length it takes for the direction to become uncorrelated, taken here as 4 Å, based on a combination of experiment and modelling [47]). Approximating the WLC with spherical harmonics [48], the end-to-end distribution can be written as:

 
formula
(1)

where the last term is defined as:

 
formula

This distribution gives the probability that two segments of chain (WXP and ANK, for example) are separated by a radial distance r. With a sequence separation of 103 residues (as between human Notch1 WXP and ANK), p(r) shows a maximum at slightly less than ∼50 Å (Figure 6A), the distance between the corresponding binding sites on CSL, indicating that these two sites are appropriately spaced for binding. We note that excluded volume effects [49,50], which are not considered in the WLC model, would increase this maximum to slightly larger distances.

WLC modelling of the RAM–ANK linker

Figure 6
WLC modelling of the RAM–ANK linker

(A) Radial and Cartesian probability distributions [p(r) and p(x,y,z), solid and broken lines respectively], with WLC parameters of lp=4 Å, and lc=103×3.8=391.4 Å. p(r) is given by eqn (1); p(x,y,z)=p(r)/4πr2. (B) Effective concentration of WXP relative to the N-terminus of ANK from eqn (2), assuming lp=4 Å and lc=3.8×nres (nres is number of residues). (C) Comparison of ANK binding to CSL in cis (solid lines) and in trans (broken lines) to WXP. An NICD (or ANK when in trans) concentration of 2 μM was assumed in both cases. Saturation in trans is calculated as fbivalent=KANK[NICD]/(1+KANK[NICD]) and in cis as fbivalent=KRAMKANKCeff [NICD]×{1+(KRAM+KANK+KRAMKANKCeff)[NICD]}−1.

Figure 6
WLC modelling of the RAM–ANK linker

(A) Radial and Cartesian probability distributions [p(r) and p(x,y,z), solid and broken lines respectively], with WLC parameters of lp=4 Å, and lc=103×3.8=391.4 Å. p(r) is given by eqn (1); p(x,y,z)=p(r)/4πr2. (B) Effective concentration of WXP relative to the N-terminus of ANK from eqn (2), assuming lp=4 Å and lc=3.8×nres (nres is number of residues). (C) Comparison of ANK binding to CSL in cis (solid lines) and in trans (broken lines) to WXP. An NICD (or ANK when in trans) concentration of 2 μM was assumed in both cases. Saturation in trans is calculated as fbivalent=KANK[NICD]/(1+KANK[NICD]) and in cis as fbivalent=KRAMKANKCeff [NICD]×{1+(KRAM+KANK+KRAMKANKCeff)[NICD]}−1.

Another useful expression that can be obtained from the WLC model is the effective concentration, Ceff, which is given by:

 
formula
(2)

where L0 is Avogadro's number. The two equations above provide a means to calculate Ceff as a function of the sequence separation between WXP and ANK (Figure 6B). Within the WLC approximation, Ceff is maximal at approx. 80 residues, falling off abruptly below approx. 50 residues and more gradually at higher separations. Interestingly, although the sequence conservation in the linker separating WXP and ANK is quite low, the number of residues in this linker region is highly conserved (Figure 2C) (for comparison, the region of NICD C-terminal to the ANK domain, which also shows low sequence conservation, is highly variable in length). These observations suggest that linker length between WXP and ANK may be evolutionarily conserved to maintain a constant effective concentration.

For bivalent interactions such as that between NICD and CSL, the binding affinity is directly proportional to Ceff. This can be seen by considering the binding reaction as a two-step process, and comparing it with formation of a ternary complex (Figure 7). Binding of the WXP motif in RAM to CSL is reasonably tight (Kd ∼0.2–0.5 μM [39]), thus the first step in the two schemes of Figure 7 should be saturable. However, binding of the ANK domain to CSL, which is likely to be required to activate transcription of Notch-responsive genes, is undetectably weak in the absence of RAM [35]. In other words, the second step in the lower scheme of Figure 7 does not saturate without very high concentrations of ANK (unless a large amount of positive co-operativity between WXP and ANK were to be invoked). However, in the upper scheme, where WXP and ANK are in cis on the same polypeptide, the second step is unimolecular; thus the concentration of ANK is replaced by Ceff. For a 103-residue linker length, Ceff reaches approx. 0.5 mM (Figure 6B). The effect of this enhancement over the binding of ANK to CSL in trans is shown in Figure 6(C), where the fraction of CSL with fully bound RAM–ANK (i.e. with ANK bound to its site on CSL) (solid line) or ANK (broken line) is plotted as a function of the association constant for ANK to CSL. In the range 103–106 M−1 (a weak binding regime consistent with the data of [35]), the WLC model predicts a significant enhancement in formation of active transcription complex when RAM and ANK are tethered.

Binding of the ANK domain in cis (upper) and in trans (lower) to RAM

Figure 7
Binding of the ANK domain in cis (upper) and in trans (lower) to RAM

The three domains of CSL (N, NTD; B, BTD; C, CTD) are shown in red, and the RAM and ANK regions of NICD are shown in light blue. RAM and ANK binding to CSL in trans (represented by the association constants KRAM and KANK respectively, in reciprocal molar units) are assumed to be independent of each other (i.e. non-cooperative).

Figure 7
Binding of the ANK domain in cis (upper) and in trans (lower) to RAM

The three domains of CSL (N, NTD; B, BTD; C, CTD) are shown in red, and the RAM and ANK regions of NICD are shown in light blue. RAM and ANK binding to CSL in trans (represented by the association constants KRAM and KANK respectively, in reciprocal molar units) are assumed to be independent of each other (i.e. non-cooperative).

Other roles for conformational disorder in Notch signalling?

As described above (e.g. Figure 2A), Notch signalling via NICD is influenced by a number of proteins besides CSL. Several of these bind directly to RAM–ANK. It is tempting to wonder whether, like CSL, the disorder and folding transitions of RAM–ANK play a role in the binding of these proteins. There is recent evidence that FIH [factor inhibiting HIF (hypoxia-inducible factor)] binds to repeat 2 of ANK in an unfolded state [51]; this is the reverse of the folding of repeat 1 induced by the binding of CSL. Ultracentrifugation studies from our laboratory (A. Allgood and D. Barrick, unpublished work) indicate that Deltex binds to ANK via the N-terminal repeats, and the first repeat may be partially, but not completely, responsible. Whether this involves induced folding of the first repeat {as with CSL, and as with the interaction of IκBα [inhibitor of NF-κB (nuclear factor κB) α] with NF-κB [52]}, or unfolding of subsequent repeats, remains to be seen.

Note added in proof (received 31 January 2008)

A recent FRET study of the binding of NICD to CSL has demonstrated a modest, but significant, affinity enhancement in the binding of RAM to NICD when in cis to ANK. Combined with the weak binding of ANK to CSL in the absence of RAM, this observation is consistent with a substartial increase in AMK occupancy in cis via effective concentration.

Structure and Function in Cell Adhesion: Biochemical Society Annual Symposium No. 75 held at The Palace Hotel, Manchester, U.K., 3–5 December 2007. Organized and Edited by David Garrod (Manchester, U.K.).

Abbreviations

     
  • ANK

    ankyrin repeat

  •  
  • BTD

    β-trefoil domain

  •  
  • CSL

    CBF1/suppressor of Hairless/Lag-1

  •  
  • CTD

    C-terminal domain

  •  
  • LUV

    large unilamellar vesicle

  •  
  • MAM

    Mastermind

  •  
  • NF-κB

    nuclear factor κB

  •  
  • NICD

    Notch intracellular domain

  •  
  • NLS

    nuclear localization sequence

  •  
  • NTD

    N-terminal domain

  •  
  • PC

    phosphatidylcholine

  •  
  • PS

    phosphatidylserine

  •  
  • RAM

    RBP-Jκ (recombination signal-binding protein 1 for Jκ)-associated molecule

  •  
  • WLC

    wormlike chain

We thank Raphael Kopan, Rohit Pappu, Olga Lubman, Rhett Kovall and Steven Blacklow for insightful comments and suggestions. This work was supported by NIH (National Institutes of Health) grant GM60001 to D.B.

References

References
1
Welshons
W.J.
The analysis of a pseudoallelic recessive lethal system at the Notch locus of Drosophila melanogaster
Cold Spring Harbor Symp. Quant. Biol.
1958
, vol. 
23
 (pg. 
171
-
176
)
2
Artavanis-Tsakonas
S.
Rand
M.D.
Lake
R.J.
Notch signaling: cell fate control and signal integration in development
Science
1999
, vol. 
284
 (pg. 
770
-
776
)
3
Kimble
J.
Simpson
P.
The LIN-12/Notch signaling pathway and its regulation
Annu. Rev. Cell Dev. Biol.
1997
, vol. 
13
 (pg. 
333
-
361
)
4
Kopan
R.
Cagan
R.
Notch on the cutting edge
Trends Genet.
1997
, vol. 
13
 (pg. 
465
-
467
)
5
Le Borgne
R.
Bardin
A.
Schweisguth
F.
The roles of receptor and ligand endocytosis in regulating Notch signaling
Development
2005
, vol. 
132
 (pg. 
1751
-
1762
)
6
Bulman
M.P.
Kusumi
K.
Frayling
T.M.
McKeown
C.
Garrett
C.
Lander
E.S.
Krumlauf
R.
Hattersley
A.T.
Ellard
S.
Turnpenny
P.D.
Mutations in the human Delta homologue, Dll3, cause axial skeletal defects in spondylocostal dysostosis
Nat. Genet.
2000
, vol. 
24
 (pg. 
438
-
441
)
7
Oda
T.
Elkahloun
A.G.
Pike
B.L.
Okajima
K.
Krantz
I.D.
Genin
A.
Piccoli
D.A.
Meltzer
P.S.
Spinner
N.B.
Collins
F.S.
Chandrasekharappa
S.C.
Mutations in the human Jagged1 gene are responsible for Alagille syndrome
Nat. Genet.
1997
, vol. 
16
 (pg. 
235
-
242
)
8
Tramboo
N.A.
Iqbal
K.
Dar
M.A.
Malik
R.A.
Naikoo
B.A.
Andrabi
M.A.
Unusual dysmorphic features in five patients with Noonan's syndrome: a brief review
J. Paediatr. Child Health
2002
, vol. 
38
 (pg. 
521
-
525
)
9
Joutel
A.
Corpechot
C.
Ducros
A.
Vahedi
K.
Chabriat
H.
Mouton
P.
Alamowitch
S.
Domenga
V.
Cecillion
M.
Marechal
E.
, et al. 
Notch3 mutations in cadasil, a hereditary adult-onset condition causing stroke and dementia
Nature
1996
, vol. 
383
 (pg. 
707
-
710
)
10
Weng
A.P.
Ferrando
A.A.
Lee
W.
Morris
J.P.
4th
Silverman
L.B.
Sanchez-Irizarry
C.
Blacklow
S.C.
Look
A.T.
Aster
J.C.
Activating mutations of NOTCH1 in human T cell acute lymphoblastic leukemia
Science
2004
, vol. 
306
 (pg. 
269
-
271
)
11
Callahan
R.
Egan
S.E.
Notch signaling in mammary development and oncogenesis
J. Mammary Gland Biol. Neoplasia
2004
, vol. 
9
 (pg. 
145
-
163
)
12
Suzuki
T.
Shen
H.
Akagi
K.
Morse
H.C.
Malley
J.D.
Naiman
D.Q.
Jenkins
N.A.
Copeland
N.G.
New genes involved in cancer identified by retroviral tagging
Nat. Genet.
2002
, vol. 
36
 (pg. 
166
-
174
)
13
Hayward
S.D.
Viral interactions with the Notch pathway
Semin. Cancer Biol.
2004
, vol. 
14
 (pg. 
387
-
396
)
14
Kidd
S.
Kelley
M.R.
Young
M.W.
Sequence of the Notch locus of Drosophila melanogaster: relationship of the encoded protein to mammalian clotting and growth factors
Mol. Cell Biol.
1986
, vol. 
6
 (pg. 
3094
-
3108.
)
15
Wharton
K.A.
Johansen
K.M.
Xu
T.
Artavanis-Tsakonas
S.
Nucleotide sequence from the neurogenic locus Notch implies a gene product that shares homology with proteins containing EGF-like repeats
Cell
1985
, vol. 
43
 (pg. 
567
-
581
)
16
Brou
C.
Logeat
F.
Gupta
N.
Bessia
C.
LeBail
O.
Doedens
J.R.
Cumano
A.
Roux
P.
Black
R.A.
Israel
A.
A novel proteolytic cleavage involved in Notch signaling: the role of the disintegrin–metalloprotease TACE
Mol. Cell
2000
, vol. 
5
 (pg. 
207
-
216
)
17
Mumm
J.S.
Schroeter
E.H.
Saxena
M.T.
Griesemer
A.
Tian
X.
Pan
D.J.
Ray
W.J.
Kopan
R.
A ligand-induced extracellular cleavage regulates γ-secretase-like proteolytic activation of Notch1
Mol. Cell
2000
, vol. 
5
 (pg. 
197
-
206
)
18
Gordon
W.R.
Vardar-Ulu
D.
Histen
G.
Sanchez-Irizarry
C.
Aster
J.C.
Blacklow
S.C.
Structural basis for autoinhibition of notch
Nat. Struct. Mol. Biol.
2007
, vol. 
14
 (pg. 
295
-
300
)
19
De Strooper
B.
Annaert
W.
Cupers
P.
Saftig
P.
Craessaerts
K.
Mumm
J.S.
Schroeter
E.H.
Schrijvers
V.
Wolfe
M.S.
Ray
W.J.
, et al. 
A presenilin-1-dependent γ-secretase-like protease mediates release of Notch intracellular domain
Nature
1999
, vol. 
398
 (pg. 
518
-
522
)
20
Schroeter
E.H.
Kisslinger
J.A.
Kopan
R.
Notch-1 signalling requires ligand-induced proteolytic release of intracellular domain
Nature
1998
, vol. 
393
 (pg. 
382
-
386
)
21
Jarriault
S.
Brou
C.
Logeat
F.
Schroeter
E.H.
Kopan
R.
Israel
A.
Signalling downstream of activated mammalian Notch
Nature
1995
, vol. 
377
 (pg. 
355
-
358
)
22
Lubman
O.Y.
Korolev
S.V.
Kopan
R.
Anchoring Notch genetics and biochemistry: structural analysis of the ankyrin domain sheds light on existing data
Mol. Cell
2004
, vol. 
13
 (pg. 
619
-
626
)
23
Roehl
H.
Bosenberg
M.
Blelloch
R.
Kimble
J.
Roles of the RAM and ANK domains in signaling by the C. elegans GLP-1 receptor
EMBO J.
1996
, vol. 
15
 (pg. 
7002
-
7012
)
24
Roehl
H.
Kimble
J.
Control of cell fate in C. elegans by a GLP-1 peptide consisting primarily of ankyrin repeats
Nature
1993
, vol. 
364
 (pg. 
632
-
635
)
25
Mosavi
L.K.
Cammett
T.J.
Desrosiers
D.C.
Peng
Z.Y.
The ankyrin repeat as molecular architecture for protein recognition
Protein Sci.
2004
, vol. 
13
 (pg. 
1435
-
1448
)
26
Zweifel
M.E.
Barrick
D.
Studies of the ankyrin repeats of the Drosophila melanogaster Notch receptor. 2. Solution stability and cooperativity of unfolding
Biochemistry
2001
, vol. 
40
 (pg. 
14357
-
14367
)
27
Zweifel
M.E.
Barrick
D.
Studies of the ankyrin repeats of the Drosophila melanogaster Notch receptor. 1. Solution conformational and hydrodynamic properties
Biochemistry
2001
, vol. 
40
 (pg. 
14344
-
14356
)
28
Zweifel
M.E.
Leahy
D.J.
Hughson
F.M.
Barrick
D.
Structure and stability of the ankyrin domain of the Drosophila Notch receptor
Protein Sci.
2003
, vol. 
12
 (pg. 
2622
-
2632
)
29
Ehebauer
M.T.
Chirgadze
D.Y.
Hayward
P.
Martinez Arias
A.
Blundell
T.L.
High-resolution crystal structure of the human Notch 1 ankyrin domain
Biochem. J.
2005
, vol. 
392
 (pg. 
13
-
20
)
30
Bradley
C.M.
Barrick
D.
Limits of cooperativity in a structurally modular protein: response of the Notch ankyrin domain to analogous alanine substitutions in each repeat
J. Mol. Biol.
2002
, vol. 
324
 (pg. 
373
-
386
)
31
Mello
C.C.
Barrick
D.
An experimentally determined protein folding energy landscape
Proc. Natl. Acad. Sci. U.S.A.
2004
, vol. 
101
 (pg. 
14102
-
14107
)
32
Wilson
J.J.
Kovall
R.A.
Crystal structure of the CSL–Notch–Mastermind ternary complex bound to DNA
Cell
2006
, vol. 
124
 (pg. 
985
-
996
)
33
Nam
Y.
Sliz
P.
Song
L.
Aster
J.C.
Blacklow
S.C.
Structural basis for cooperativity in recruitment of Maml coactivators to Notch transcription complexes
Cell
2006
, vol. 
124
 (pg. 
973
-
983
)
34
Tamura
K.
Taniguchi
Y.
Minoguchi
S.
Sakai
T.
Tun
T.
Furukawa
T.
Honjo
T.
Physical interaction between a novel domain of the receptor notch and the transcription factor RBP-Jκ/Su(H)
Curr. Biol.
1995
, vol. 
5
 (pg. 
1416
-
1423
)
35
Nam
Y.
Weng
A.P.
Aster
J.C.
Blacklow
S.C.
Structural requirements for assembly of the CSL–intracellular Notch1–Mastermind-like 1 transcriptional activation complex
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
21232
-
21239
)
36
Kovall
R.A.
Hendrickson
W.A.
Crystal structure of the nuclear effector of Notch signaling, CSL, bound to DNA
EMBO. J
2004
, vol. 
23
 (pg. 
3441
-
3451
)
37
Giniger
E.
A role for Abl in Notch signaling
Neuron
1998
, vol. 
20
 (pg. 
667
-
681
)
38
Guo
M.
Jan
L.Y.
Jan
Y.N.
Control of daughter cell fates during asymmetric division: interaction of Numb and Notch
Neuron
1996
, vol. 
17
 (pg. 
27
-
41
)
39
Lubman
O.Y.
Ilagan
M.X.
Kopan
R.
Barrick
D.
Quantitative dissection of the Notch:CSL interaction: insights into the Notch-mediated transcriptional switch
J. Mol. Biol.
2007
, vol. 
365
 (pg. 
577
-
589
)
40
Tani
S.
Kurooka
H.
Aoki
T.
Hashimoto
N.
Honjo
T.
The N- and C-terminal regions of RPB-J interact with the ankyrin repeats of Notch1 RAMIC to activate transcription
Nucleic Acids Res.
2001
, vol. 
29
 (pg. 
1373
-
1380
)
41
Kurooka
H.
Kuroda
K.
Honjo
T.
Roles of the ankyrin repeats and C-terminal region of the mouse Notch1 intracellular region
Nucleic Acids Res.
1998
, vol. 
26
 (pg. 
5448
-
5455
)
42
Cantor
C.R.
Schimmel
P.R.
Biophysical Chemistry Part II: Techniques for the Study of Biological Structure and Function, W.H
1980
New York
Freeman and Company
43
Lakowicz
J.R.
Principles of Fluorescence Spectroscopy
1999
New York
Kluwer Academic/Plenum Publishers
44
Levy
Y.
Cho
S.S.
Onuchic
J.N.
Wolynes
P.G.
A survey of flexible protein binding mechanisms and their transition states using native topology based energy landscapes
J. Mol. Biol.
2005
, vol. 
346
 (pg. 
1121
-
1145
)
45
Hilser
V.J.
Thompson
E.B.
Intrinsic disorder as a mechanism to optimize allosteric coupling in proteins
Proc. Natl. Acad. Sci. U.S.A.
2007
, vol. 
104
 (pg. 
8311
-
8315
)
46
Wright
P.E.
Dyson
H.J.
Intrinsically unstructured proteins: re-assessing the protein structure–function paradigm
J. Mol. Biol.
1999
, vol. 
293
 (pg. 
321
-
331
)
47
Zhou
H.-X.
Polymer models of protein stability, folding, and interactions
Biochemistry
2004
, vol. 
43
 (pg. 
2141
-
2154
)
48
Gobush
W.
Yamakawa
H.
Stockmayer
W.H.
Magee
W.S.
Statistical mechanics of wormlike chains. I. Asymptotic behavior
J. Chem. Phys.
1972
, vol. 
57
 (pg. 
2839
-
2843
)
49
Tran
H.T.
Pappu
R.V.
Toward an accurate theoretical framework for describing ensembles for proteins under strongly denaturing conditions
Biophys. J.
2006
, vol. 
91
 (pg. 
1868
-
1886
)
50
Tran
H.T.
Wang
X.
Pappu
R.V.
Reconciling observations of sequence-specific conformational propensities with the generic polymeric behavior of denatured proteins
Biochemistry
2005
, vol. 
44
 (pg. 
11369
-
11380
)
51
Coleman
M.L.
McDonough
M.A.
Hewitson
K.S.
Coles
C.
Mecinovic
J.
Edelmann
M.
Cook
K.M.
Cockman
M.E.
Lancaster
D.E.
Kessler
B.M.
, et al. 
Asparaginyl hydroxylation of the Notch ankyrin repeat domain by factor inhibiting hypoxia-inducible factor
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
24027
-
24038
)
52
Truhlar
S.M.
Torpey
J.W.
Komives
E.A.
Regions of IκBα that are critical for its inhibition of NF-κB·DNA interaction fold upon binding to NF-κB
Proc. Natl. Acad. Sci. U.S.A.
2006
, vol. 
103
 (pg. 
18951
-
18956
)
53
Zweifel
M.E.
Leahy
D.J.
Barrick
D.
Structure and Notch receptor binding of the tandem WWE domain of Deltex
Structure
2005
, vol. 
13
 (pg. 
1599
-
1611
)
54
Del Bianco
C.
Aster
J.C.
Blacklow
S.C.
Mutational and energetic studies of Notch1 transcriptional complexes
J. Mol. Biol.
2008
, vol. 
376
 (pg. 
131
-
140
)

Author notes

1

Present address: US Patent and Trademark Office, 600 Dulaney Street, Alexandria, VA 22314, U.S.A.