Conformational variability of the intracellular domain of Drosophila Notch and its interaction with Suppressor of Hairless
- *Department of Cell Biology, Harvard Medical School, 240 Longwood Avenue, Boston, MA 02115;
- †Massachusetts General Hospital Cancer Center, 149 13th Street, Charlestown, MA 02129; and
- ‡Collège de France, 75231 Paris, France
-
Communicated by Vincent T. Marchesi, Yale University School of Medicine, New Haven, CT, April 9, 2007 (received for review December 28, 2006)
Abstract
The Notch receptor is the central element in an evolutionarily conserved signal transduction pathway that controls cell fates in metazoans. Receptor–ligand interactions trigger a cascade of proteolytic events that release the entire Notch intracellular domain (NICD) from the membrane, permitting its translocation into the nucleus and participation in a transcriptionally active complex. Using electron microscopy, we examined the structure of NICD and its interaction with the DNA-binding effector of Notch signaling, Suppressor of Hairless [Su(H)]. In conjunction with biochemical analyses, we found that Drosophila NICD is monomeric and exists in two primary conformational states, only one of which can bind Su(H). Furthermore, we show that changes in divalent cation concentrations lead to NICD self-association, which seems to be mediated by the polyglutamine-containing, opa-repeat region of NICD. Our study suggests that conformational modulation of NICD may define a mechanism of Notch pathway control.
Notch signaling couples the cell fate acquisition of an individual cell with the cell fate choices made by its next-door neighbors. This cell signaling mechanism relies on interactions between the Notch receptor expressed on the surface of one cell with membrane-bound ligands expressed on the surface of adjacent cells. Receptor–ligand interactions trigger proteolytic events that eventually release the intracellular domain of Notch (NICD). The NICD then translocates to the nucleus, where it can assemble with the DNA-binding effector Suppressor of Hairless [Su(H)] and other nuclear proteins into a transcriptionally active complex that drives Notch-dependent gene expression (1–3). The accumulated molecular evidence indicates that the signal transduction mechanism of Notch does not involve an enzymatic amplification step but relies on a cascade of stoichiometric molecular interactions that lead to the transmission of the Notch signal into the nucleus. The signal itself is, therefore, the ≈1,000-aa long NICD, which participates directly in the initiation of downstream gene activity.
The molecular steps leading to the translocation of NICD into the nucleus and regulating the formation of the active complex remain unknown. However, it is clear that NICD physically interacts with many proteins both inside and outside the nucleus, and, thus, competitive protein binding could potentially influence the signaling capacity of NICD. For instance, under certain experimental conditions, Su(H) has been shown to associate with NICD in the cytoplasm (4). This interaction appears to involve Notch binding sites that overlap with those mediating the interaction between NICD and the cytoplasmic Notch signal modulator, Deltex (5, 6).
Despite the wealth of information that has accumulated over the past two decades regarding the mechanism by which signals are transmitted through the Notch cell surface receptor, very little is known about the stoichiometry of pathway–component interactions and the structural changes associated with these interactions. Additionally, because of the large size and domain complexity of the Notch receptor, structural studies have thus far been confined to only fragments of the receptor. The crystal structures of NICD fragments relevant to the recruitment of transcriptional coactivators on DNA, as well as the ankyrin-repeat region of NICD, have been recently solved (7–10). However, information about the conformation of the entire NICD molecule or, indeed, its oligomeric state, is lacking. Given the nature of Notch signaling, such information is important if we are to understand the molecular mechanisms that govern the transmission of Notch signals from the cell surface to the nucleus. Therefore, using EM, we sought to examine the structure of the entire NICD and its interaction with the major Notch signal effector Su(H).
Here we report the results of our studies involving single-particle EM on purified NICD and Su(H), as well as the complex formed between these two molecules. In combination with analytical ultracentrifugation studies, we show that under physiological conditions, full-length NICD is a monomer and that it interacts with the transcription factor Su(H) in a 1:1 stoichiometry. We also show that both monovalant and divalent cations can affect the conformation of NICD and its ability to form a complex with Su(H). Finally, we provide evidence indicating that divalent cations can induce the formation of filamentous structures via the polyglutamine-rich, opa-repeat domain of NICD (11), demonstrating the potential of NICD to self-associate.
Results
Drosophila NICD Is an Asymmetric Monomer.
A polypeptide fragment corresponding to the Drosophila NICD (amino acids 1765–2703) (Fig. 1 a) and the full-length Su(H) protein were independently expressed in Sf9 insect cells (see Methods). Both proteins were engineered with C-terminal Flag-tags and sequentially purified by anti-Flag M2 affinity chromatography and gel filtration chromatography. Elution profiles from the gel-filtration column revealed that each preparation contained a single component (Fig. 1). NICD eluted from the gel filtration column with an M r of 250 kDa (Fig. 1 b), whereas Su(H) eluted with an M r of 75 kDa (Fig. 1 c). To assess further the purity of the preparations, eluate fractions were resolved by SDS/PAGE. Coomassie staining confirmed the presence of a single predominant protein in each preparation, with NICD migrating at an M r of ≈110 kDa and Su(H) migrating at an M r of ≈70 kDa (Fig. 1 d), consistent with the respective values of 98 and 67 kDa expected from their primary sequences.
Purification of NICD and Su(H). (a) Schematic representation of NICD with domain positions demarcated (TM, transmembrane domain). FPLC chromatographs of purified NICD (b) and purified Su(H) (c) are shown. NICD fraction 10 (11.07 ml) and Su(H) fraction 13 (13.59 ml) were used for further characterization and EM imaging. au, absorbance units. (d) Coomassie-stained gel of NICD and Su(H). Fractions shown are from the final gel filtration column used to purify each protein. (e) Velocity sedimentation data obtained from analytical ultracentrifugation of NICD. Concentrations used were 1.0 mg/ml in 50 mM Hepes (pH 7.5), 150 mM KCl (thick line), 0.7 mg/ml in 50 mM Hepes (pH 7.5), 150 mM KCl (thin line), 0.7 mg/ml in 50 mM PBS Na2PO4 (pH 7.5), 150 mM KCl (dotted line), and 0.3 mg/ml in 50 mM Hepes, 150 mM KCl (dashed line). g(s*), apparent sedimentation coefficient distribution; S*, apparent sedimentation coefficient. (f) Molecular mass distribution of the NICD at the same concentrations. c(M), molar mass distribution.
Although Su(H) behaved as predicted when analyzed by both gel filtration chromatography and SDS/PAGE, suggesting an overall spherical structure, the discrepancy between the elution profile of NICD revealed by gel filtration chromatography and its behavior observed by SDS/PAGE suggested that this protein is either asymmetric or forms oligomers. To examine further the molecular organization of NICD, we performed analytical ultracentrifugation, which, unlike gel filtration chromatography, allows molecular mass determinations based on first-order principles. Using velocity sedimentation analysis at three different protein concentrations, we found that NICD had a mean sedimentation coefficient (S 20,w) of 4.33, a frictional ratio (f/f 0) of 1.6 (Fig. 1 e), and a molecular mass of ≈110 kDa (Fig. 1 f). Moreover, as shown in Fig. 1 e and f, these values were observed when NICD was analyzed in either a Hepes buffer (pH 7.5, 150 mM KCl) or a phosphate buffer (PBS). Thus, we conclude that in physiological concentrations of monovalent ions, Drosophila NICD is an asymmetric protein that exists as a monomer.
Conformational Flexibility of Drosophila NICD.
Although specific regions of the Notch receptor from different species have been examined by x-ray diffraction, these studies focused on small portions of NICD that only included the RAM and ankyrin-repeat domains (7–10). Thus, information about the structural and oligomeric state of the entire NICD that is critical for signaling is still lacking. To further examine the oligomeric and conformational states of the entire NICD, the purified protein was subjected to single-particle EM analysis. In agreement with our previous observations, electron micrographs of negatively stained protein preparations not only confirmed the notion that NICD is a monomer but also revealed that this molecule consists of four globular domains that are linearly arranged with an overall length of ≈134 Å. The length of the extended molecule, however, cannot be determined precisely because of its inherent flexibility, thus limiting resolution of the molecules' ends during averaging (Fig. 2 a and b, averages 7–9). Typically, the four domains are folded to form a U-shaped particle, with each “leg” of the U having a length of ≈72 Å (Fig. 2 a and b, averages 1–6). The connectivity of the domains could be seen clearly in contour plots displaying only pixels above a significance level of 3σ (Fig. 2 c), which we calculated with the SPIDER software package (12). Overall, these data indicate that Drosophila NICD is flexible and can adopt compact and extended conformations.
Electron microscopy of NICD. (a) Raw image of negatively stained NICD particles in 50 mM Hepes (pH 7.5), 150 mM KCl, and 1 mM EGTA. (b) Representative class averages of NICD particles showing the compact state (averages 1–6) as well as extended conformations (averages 7–9). (c) Contour plots of the averages in b showing pixels at a significance level of 3σ and above. In the first panel, the legs of the U-shaped particle are labeled, and their connection is marked by a black arrow. Raw images of negatively stained NICD particles in the presence of 1 M KCl (d) and 30 mM KCl (e), incubated for 24 h at 4°C, are shown. (f) Velocity sedimentation data. g(s*), apparent sedimentation coefficient distribution; S*, apparent sedimentation coefficient. (g) Molecular mass distribution of NICD incubated in 30 mM KCl. (Scale bars: 100 Å.)
Interestingly, the distance between the two legs is more variable and ranges from 22 to 29 Å (Fig. 2 b and c), suggesting weak interactions between the two legs of U-shaped NICD. Examining the charge distribution across the NICD molecule, we predict the isoelectric point (pI) of the N-terminal region (W1821–P1900) to be 4.23, whereas the C-terminal region (A2521–G2571) has an isoelectric point of 8.80. This observation raises the possibility that the N- and C-terminal regions form contacts through electrostatic interactions. Consistent with this possibility, the shape of NICD was influenced by different concentrations of monovalent cations. NICD was incubated for 2 h in the presence of either 1 M KCl or 30 mM KCl and subsequently analyzed by negative-stain EM. Both U-shaped particles and extended conformations were observed when NICD was incubated with 1 M KCl (Fig. 2 d). On the other hand, in the presence of 30 mM KCl, the majority (≈90%) of NICD had a compact U-shaped appearance (Fig. 2 e). Velocity sedimentation analysis in 30 mM KCl revealed that NICD had a mean S 20,w of 4.5, an f/f 0 of 1.46 (Fig. 2 f), and a molecular mass of 110 kDa (Fig. 2 g). The increase in the sedimentation coefficient and decrease in the frictional ratio we observed when NICD was examined in 30 mM KCl as compared with 150 mM KCl (see above) indicates that the conformation of the entire NICD at low monovalent ion concentrations is more compact.
Taken together, these data suggest that Drosophila NICD exists in both extended and compact states, with the compact conformation possibly being stabilized through N- and C-terminal electrostatic interactions. This notion is supported by the finding that the equilibrium between these states can be influenced by the ionic strength of the environment.
Formation of the NICD–Su(H) Complex Is Conformation-Dependent.
An essential step in the transmission of Notch signals is the association of NICD, generated from the ligand-triggered proteolytic cleavage of the receptor, with Su(H), the major DNA-binding effector of Notch signaling. Recent x-ray diffraction studies showed that Su(H) does not self-associate in the presence of NICD and that it forms a 1:1 molar complex with NICD through the RAM-ankryin repeat region of Notch (9, 10). When NICD and Su(H) were mixed in a 1:1 molar ratio and analyzed by gel filtration chromatography in the presence of 150 mM KCl, the majority of NICD and Su(H) migrated through the column as a single component (Fig. 3 a). Similarly, when an equimolar mixture of NICD and Su(H) was examined by velocity sedimentation analysis in the presence of 150 mM KCl, a single component was identified that had a mean S 20,w of 5.42 and an f/f 0 of 1.8 (Fig. 3 b). The sedimentation profile of the complex was, however, slightly asymmetric, indicating minor heterogeneity in the sample. The molecular mass of the complex was calculated to be ≈170 kDa (Fig. 3 c), close to the predicted value of 165 kDa for a heterodimeric NICD–Su(H) complex. Therefore, at physiological concentrations of monovalent ions (150 mM KCl), NICD and Su(H) form a stable, equimolar complex.
NICD–Su(H) complex formation. (a) Gel filtration chromatograph and Coomassie-stained gel of the peak fractions showing the NICD–Su(H) complex in the absence of Ca2+. au, absorbance units. (b) Velocity sedimentation data of the NICD–Su(H) complex in the presence (solid line) and absence (dashed line) of 1 mM CaCl2. g(s*), apparent sedimentation coefficient distribution; S*, apparent sedimentation coeffiecient. (c) Molecular mass distribution of the NICD–Su(H) complex calculated from the velocity sedimentation data. c(M), molar mass distribution. (d) EM image of NICD–Su(H) complex in the presence of 1 mM CaCl2. (e) Representative class averages of the NICD–Su(H) complex. For comparison, NICD averages 1, 4, and 7 from Fig. 2 b are shown in the first column. Velocity sedimentation data (f) and molecular mass distribution (g) of the NICD and Su(H) mixture in 30 mM NaCl are shown. (h) Raw image of the NICD and Su(H) mixture under low-salt conditions. Representative NICD particles are circled in black and Su(H) particles are circled in white. (Scale bars: 100 Å.)
Examination of the NICD–Su(H) complex by negative-stain EM revealed that the majority of particles were larger than NICD (Fig. 3 d). Representative class averages of the complexes are shown in Fig. 3 e. For comparison, we also show NICD averages 1, 4, and 7 from Fig. 2 b. The length of the complex ranges from 122 to 134 Å, and the width varies between 53 and 67 Å. Based on these dimensions and its appearance in class averages, NICD in the NICD–Su(H) complex appears to be more similar to the extended molecule but bent around additional density representing the Su(H) molecule (Fig. 3 e).
To determine whether the compact form of NICD is also capable of interacting with Su(H), we examined the capacity of NICD to bind to Su(H) in a low-ionic-strength environment, in which the EM analysis had shown that NICD adopts a compact conformation. We, therefore, performed velocity sedimentation analysis in 30 mM KCl using a mixture of Su(H) and excess NICD to ensure complete binding. This analysis revealed two separate components (Fig. 3 f), corresponding to the molecular masses of Su(H) (60 kDa) and NICD (110 kDa) (Fig. 3 g). Examination of a parallel sample, incubated for the length of the analytical ultracentrifugation time course, revealed particles consistent with individual NICD and Su(H) molecules but not with the complex (Fig. 3 h). These results indicate that the compact form of NICD, which predominates in a low-ionic-strength environment, cannot form a stable complex with Su(H).
We finally note that attempts to bind NICD to Su(H) at high ionic strengths (1 M KCl) were not successful, consistent with the notion that electrostatic interactions play an important role in the formation of the NICD–Su(H) complex. Additionally, x-ray diffraction studies of complexes involving Su(H) and NICD peptide fragments, deriving from either human or Caenorhabditis elegans proteins, showed many electrostatic interactions between the interacting protein domains (9, 10).
Calcium Stabilizes the Extended Conformation of Drosophila NICD.
The susceptibility of NICD conformation, and indeed NICD–Su(H) complex formation, to monovalent ion concentrations prompted us to examine the effects of divalent cations on the structure of NICD. We found that the overall shape of NICD changed dramatically when the protein was exposed to CaCl2. After 2 h of incubation in 1 mM CaCl2, the majority of NICD molecules adopted an extended conformation (Fig. 4 a). The conformation of the particles could be reversed to a mixture of compact and extended forms by chelating the calcium with EGTA (data not shown). Upon a further increase of the CaCl2 concentration to 10 mM, the NICD not only extended but also assembled into larger, filament-like structures. After 1–2 h of incubation in 10 mM CaCl2, NICD started to form thin filaments, and after 4 h of incubation, thin filaments became the dominant species (thin arrowheads in Fig. 4 b) while thicker filaments started to appear (thick arrowhead in Fig. 4 b). This effect was also observed, albeit to a much lesser degree, when NICD was incubated with MgCl2 at concentrations ≥200 mM, which produced a few thin filaments that were much smaller than those observed with 10 mM CaCl2 as well as a mixture of extended and compact particles (Fig. 4 c). Thick filaments, on the other hand, failed to form even after 24 h of incubation.
Influence of divalent cations on NICD structure. (a) Raw image of NICD incubated with 1 mM CaCl2 at room temperature for 2 h, showing the marked increase in molecules adopting an extended conformation. (b) Raw image of NICD filaments formed in 10 mM CaCl2 after 4 h. Thin NICD filaments are indicated with thin arrowheads, and a thick NICD filament is indicated with a thick arrowhead. (c) Raw image of negatively stained NICD particles in the presence of 200 mM MgCl2 incubated for 24 h at 4°C. Both compact (white circles) and extended (white arrows) monomers are present. Thin filaments are indicated with black arrowheads. (Scale bars: 100 Å.)
Attempts to bind Su(H) to filamentous NICD structures formed in the presence of 10 mM CaCl2 were unsuccessful as assessed by negative-stain EM (data not shown), suggesting that filament formation is not compatible with Su(H) binding. Thus, these observations not only demonstrate the profound impact calcium ions can exert on the structure of NICD but also indicate that NICD has the potential to self-associate and oligomerize.
On the other hand, the presence of 1 mM CaCl2, a concentration that cannot induce the formation of filamentous NICD structures, does not affect the complex formation of NICD with Su(H). The NICD–Su(H) complex formed in the presence of this low CaCl2 concentration eluted from the gel filtration column similar to the complex formed in the absence of CaCl2 (data not shown). Additionally, velocity sedimentation analysis revealed that the NICD–Su(H) complex behaved identically in the presence (1 mM) and absence of CaCl2 (Fig. 3 b and c).
The Polyglutamine Region of Drosophila NICD Is Involved in Filament Formation.
It is known that polyglutamine domains are associated with the formation of protein filaments as well as aggregates (13, 14), and Drosophila NICD contains a stretch of 31 glutamine residues near its C terminus (11). Such polyglutamine regions are also present, but to a lesser extent, in mammalian Notch 1 proteins. To probe the possible involvement of the polyglutamine region of NICD in filament formation, we took advantage of the polyglutamine aggregation inhibitor C2-8, which has been shown to prevent polyglutamine aggregation in Huntington's disease neurons (15). NICD was incubated with 10 mM CaCl2, and filament formation was monitored over the course of 72 h in the presence and absence of the C2-8 inhibitor (50 mM) (Fig. 5). Immediately upon mixing (0 h), the two samples appeared identical, containing both compact and extended NICD molecules (Fig. 5 a and d). After 4 h, the NICD sample containing C2-8 was similar to the 0-h mixture (Fig. 5 b), whereas the sample without C2-8 showed the presence of thin filaments (Fig. 5 e). After 24 h, the NICD sample lacking C2-8 formed thicker filaments and filamentous aggregates (Fig. 5 f), and after 72 h, the sample contained excessive plaques consisting of tangled thick filaments (data not shown). By contrast, in the sample containing C2-8, no NICD filaments could be observed after 24 h (Fig. 5 c) or 72 h of incubation in 10 mM CaCl2. These results strongly indicate that the polyglutamine region is responsible for filament formation, a region that may be naturally occluded by the N terminus in the compact NICD conformation.
NICD in the presence of CaCl2 and C2-8 polyglutamine inhibitor. Shown are raw images of NICD incubated with 10 mM CaCl2 and the C2-8 polyQ aggregation inhibitor (50 μM) for 0 h (initial addition) (a), 4 h (b), and 24 h (c). Also shown are raw images of NICD incubated with only 10 mM CaCl2 for 0 h (d), 4 h (e), and 24 h (f). After 4 h of incubation in only 10 mM CaCl2 (e), equal aliquots of sample were removed and further incubated in either 50 μM C2-8 (g) or buffer (control) (i) for up to 24 h and reexamined on continuous carbon grids using EM. In g, both compact (white circles) and extended (white arrows) monomers were present. After an additional 24-h incubation period (h) the structural distribution of the population did not change. (j) Filamentous networks formed in only calcium (i) were further incubated with 50 μM C2-8 for 24 h. (Scale bars: 100 Å.)
To determine whether the C2-8 inhibitor is able to revert preformed filaments back to its constituent NICD particles, two separate aliquots from the NICD sample containing thin filaments (Fig. 5 e) were removed. To one aliquot, 50 mM C2-8 was added, whereas the other aliquot served as a control for filament formation. After 24 h (Fig. 5 g) or 48 h (Fig. 5 h) of incubation in the presence of 50 mM C2-8 (Fig. 5 g), only compact and extended NICD particles were observed. In contrast, the sample that did not receive C2-8 contained thick, tangled filaments (Fig. 5 i), and these thick filaments did not disassemble upon incubation with 50 mM C2-8 (Fig. 5 j). These experiments demonstrate that the C2-8 polyglutamine inhibitor prevents the progression of filament formation and that the inhibitor can disassemble preformed thin filaments but not tangled thick filaments, suggesting that the polyglutamine repeats remain accessible only within the thin filaments.
Discussion
The results we report here suggest that the interaction of NICD with other proteins may be regulated by factors that influence NICD conformation. The data we obtained from analytical ultracentrifugation and EM analysis all point to NICD being a highly flexible monomer that can exist as an extended, linear molecule or in a U-shaped, more compacted state. Importantly, our data demonstrate that the ability of NICD to interact with Su(H) is conformation-dependent. The single particle averages indicate that binding is compatible only with a molecule that is at least partially extended (see Results), and this conclusion is supported by the observation that Su(H) does not interact with NICD in a low-ionic-strength environment, in which NICD adopts the compact, U-shaped state, but readily interacts with NICD in 1 mM CaCl2, a condition favoring the extended form of NICD. Thus, any factor that regulates NICD conformation could potentially regulate its association with Su(H) and, consequently, its activity. Importantly, these results imply that protein interaction assays between NICD and other protein species must be carefully controlled.
The role monovalent and divalent cations play in regulating NICD conformation and signaling in vivo remains to be determined. Our attempts to address this issue have not been conclusive, as we have found that manipulating intracellular calcium levels in cultured insect cells appears to induce apoptosis. Nonetheless, it must be emphasized that the cellular response to Notch signaling seems to be uniquely sensitive to signal dosage (3), and therefore changes in ion concentrations, which could modulate the availability of NICD to interacting proteins, may play a significant role in influencing the biological output of Notch activity. In this regard, it is important to note that changes in both extracellular and intracellular calcium concentrations appear to be important in localized Notch signaling within the developing organism (16, 17).
Our structural studies also reveal a potential of NICD for self-association. In the presence of high calcium concentrations, NICD forms filament-like structures, and polyglutamine aggregation inhibitors can reverse this physical state of NICD. Polyglutamine stretches have been shown to play important roles in protein function and have been associated with a number of neurodegenerative diseases. Moreover, it is thought that polyglutamine-mediated protein interactions can influence the oligomerization state of proteins, affecting their activity (18, 19). One such example in Drosophila occurs in the case of Groucho, which is a member of a family of metazoan corepressors that is not only linked to Notch-dependent signaling but has widespread roles in development (20). Although a cellular role of NICD filaments we observe in the presence of calcium has yet to be determined, it is attractive to speculate that formation of such structures could be used as a means to sequester NICD in a nonactive state by virtue of our finding that they cannot bind Su(H).
This study emphasizes the utility of EM analysis in addressing aspects of Notch signaling that have hitherto been unexplored and raises the hypothesis that conformational modulation of NICD may, in fact, define an important and fundamental mechanism in Notch pathway control.
Methods
Protein Expression and Purification.
The entire intracellular domain of Drosophila Notch (amino acids 1765–2703) was cloned into pFastBac1 with sequences encoding a Flag tag at its C terminus and a 6xHis tag at its N terminus. Similarly, the entire Su(H) protein was cloned into pFastBac1 with sequences encoding myc and Flag tags at its C terminus. NICD and Su(H) were expressed in Sf9 cells using a baculovirus overexpression system and purified using M2-affinity (Sigma, St. Louis, MO) chromatography. Flag-tagged proteins were eluted from the M2-column in 0.8 mg/ml Flag peptide in BC300 (20 mM Hepes, pH 7.9/300 mM KCl/10% glycerol/0.2 mM EDTA). Affinity-purified Flag-tagged proteins were further purified over a Superdex 200 column by using an ÄKTA FPLC.
Analytical Ultracentrifugation.
Velocity sedimentation experiments were performed by using a Beckman (Fullerton, CA) Optima XL-I analytical ultracentrifuge with an An-60 Ti rotor operating at a speed of 50,000 rpm at 20°C. For the NICD and the NICD–Su(H) complex, 400 ml of sample was loaded into a double-sector quartz cell and blanked against the sample buffer containing either (i) 50 mM Hepes (pH 7.5) and 150 mM KCl, (ii) 50 mM Na2PO4 (pH 7.5) and 150 mM KCl, or (iii) 50 mM Na2PO4 (pH 7.5) and 30 mM KCl. Because Su(H) at concentrations of ≈1 mg/ml tends to form aggregates at KCl concentrations <100 mM, for the experiments performed in 30 mM KCl, we first centrifuged the Su(H) sample (0.3 mg/ml) at 100,000 × g for 30 min and examined the protein by negative-stained EM. The Su(H) sample appeared to be free of aggregates and remained so even after incubation in 30 mM KCl for up to 24 h. Absorbance data were collected in triplicate by using a step size of 0.005 cm in continuous mode at either 230 nm or 280 nm depending on the protein concentration. Absorbance data were analyzed by using the SEDFIT program (21).
Electron Microscopy and Image Processing.
Three milliliters of sample solution were adsorbed to glow-discharged, carbon-coated, 200-mesh copper grids. Grids were washed with five drops of Milli-Q water and stained with two drops of 0.75% uranyl formate as described (22). Negatively stained specimens were examined by using an FEI (Hillsboro, OR) Tecnai 12 electron microscope operating at an accelerating voltage of 120 kV. Images were recorded on imaging plates under low-dose conditions at a magnification of ×67,000 and a defocus value of approximately −1.5 μm. Imaging plates were scanned by using a Ditabis scanner (Pforzheim, Germany) at a step size of 15 μm and binned over 2 × 2 pixels for a final sampling of 4.48 Å per pixel on the specimen level.
Particles of each sample were selected from the images by using the WEB display program associated with the SPIDER software package (12). For NICD in EGTA, 7,301 particles were selected from 36 images, and for the NICD–Su(H) complex, 7,708 particles were selected from 55 images. For each data set, the selected particles were windowed into individual images of 70 × 70 pixels, and the particle images were subjected to 10 rounds of multireference alignment and K-means classification, specifying 30 output classes.
PolyQ Inhibitor Experiments.
For the experiments using the C2-8 polyQ aggregation inhibitor (Calbiochem, San Diego, CA), a 100-ml sample of purified NICD (0.3 mg/ml in 50 mM Hepes, pH 7.5/150 mM KCl) was mixed with 10 mM CaCl2 (final concentration) in the presence and absence of C2-8 (50 mM) and incubated for 72 h at 4°C. At time points of 0, 4, 24, and 72 h, negatively stained specimens were prepared of both mixtures and examined by EM. Additionally, at the 4-h time point, equal volumes of sample (25 ml each) were removed from the NICD mixture containing 10 mM CaCl2 but not C2-8. To one of the aliquots, 25 ml of the above calcium buffer was added, and to the other aliquot, 25 ml of buffer containing C2-8 (50 mM final concentration) was added. Negatively stained specimens were prepared of each mixture after 24 h of incubation. After this step, we also incubated the sample that received the additional calcium buffer with 50 mM C2-8 for 24 h. Negatively stained specimens were again prepared after this final incubation step.
Acknowledgments
This work was supported by National Institutes of Health Grants NS26084, GM62931, and CA098402 (to S.A.-T.). The molecular EM facility at Harvard Medical School was established by a generous donation from the Giovanni Armenise Harvard Center for Structural Biology and is supported by National Institutes of Health Grant GM62580 (to T.W.).
Footnotes
- §To whom correspondence should be addressed. E-mail: tsakonas{at}helix.mgh.harvard.edu
-
Author contributions: D.F.K. and R.J.L. contributed equally to this work; D.F.K., R.J.L., T.W., and S.A.-T. designed research; D.F.K. and R.J.L. performed research; R.J.L. and S.A.-T. contributed new reagents/analytic tools; D.F.K., R.J.L., T.W., and S.A.-T. analyzed data; and D.F.K., R.J.L., T.W., and S.A.-T. wrote the paper.
-
The authors declare no conflict of interest.
- Abbreviations:
- NICD,
- Notch intracellular domain;
- Su(H),
- Suppressor of Hairless.
- © 2007 by The National Academy of Sciences of the USA
