Lührs et al. 10.1073/pnas.0506723102. |
Fig. 6. Conceivable intermolecular interaction patterns of strand b2 with strand b1 of neighboring Ab(1–42) molecules. In the "odd" scenario (A), strand b2 interacts with strand b1 of the odd neighbor, whereas in the "even" scenario (B), it interacts with strand b1 of the even neighbor. The inhibitor Ab16–22m (14) contains N-methylated amino acids and can therefore only bind to the odd end of the fibril. Wild-type Ab molecules are indicated by gray diagonal bars that correspond to a cross section along the protofilament axis through the Ca atom positions of the residues F19 and G38. Correspondingly, the inhibitor Ab16–22m is indicated by dark-gray diagonal half-bars. The backbone amide and the carbonyl groups of the odd-numbered residues are sketched as green horizontal bars; the corresponding groups of the even residues are sketched as orange horizontal bars. The N methylation on the odd-numbered residues of the inhibitor Ab16–22m is indicated by a yellow circle. The fibrillar axis is indicated by a two-sided arrow.
Fig. 7. Results of the structure calculation of Ab(1–42) protofilament. (A) Bundle of 10 calculated CYANA conformers that resulted from the superposition of the heavy backbone atoms of residues 18–41 obtained from the refined calculation (Table 1). The peptide backbone is shown in cyan, and the side chains are green. Only the middle monomer of the calculated pentameric protofilament fragment is shown. (B) Bundle of the 10 best calculated CYANA conformers obtained without polymer constraints (Table 1, column 2). The coloring and superposition correspond to A.
Fig. 8. The 3D structure of the 35MoxAb(1–42) fibril. Ribbon diagram of the protofilament core structure comprising residues 17–41. The b-sheets are indicated by cyan arrows, and nonregular secondary structure is indicated by gray spline curves through the Ca atom coordinates of the corresponding residues. The bonds connecting the heavy side-chain atoms of hydrophobic, polar, negatively charged, and positively charged residues are shown in yellow, green, red, and blue, respectively. The intermolecular salt bridge connecting D23 and K28 is indicated by dotted lines. The direction of the fibril axis is indicated by arrows pointing from "even" to "odd."
Fig. 9. Comparison of the Ab(1–42) fibril structure to the model that was proposed for Ab(1–40) fibrils. (A) Ball-and-stick drawing of the Ab(1–42) structure. Only the heavy atoms of the peptide segment 17–40 are shown. Hydrophobic, negatively charged, positively charged, and polar residues are green, red, blue, and purple, respectively. (B) Ball-and-stick drawing of the Ab(1–40) model. The peptide segment shown and the coloring correspond to A. The dotted white line aligns the Ab(1–42) structure and the Ab(1–40) model through the Ca position of residues F19. Some side-chain identities are indicated using the single-letter amino acid code and the residue number.
Fig. 10. van der Waals contact surface polarity and ribbon diagram at the even end of the 35MoxAb(1–42) protofilament core structure comprising residues 17–42. The b-sheets are indicated by cyan arrows, and nonregular secondary structure is indicated by gray spline curves through the Ca atom coordinates of the corresponding residues. The bonds connecting the heavy side-chain atoms are colored as in Fig. 8. Positively and negatively charged surface patches are shown in blue and red, respectively, and all others are shown in white. The protofilament section has been rotated by 180° around the y axis relative to Fig. 4C. The direction of the fibril axis is indicated by arrows pointing from "even" to "odd."
Table 1. Parameters characterizing the structure calculation of the Ab(1–42) protofilament
Stage of structure calculation | Refined | Unrefined | |||
No. of Ab(1–42) monomers | 5 | 5 | 4 | 3 | 2 |
No. of distance constraints | |||||
H bonds* | 170 | 170 | 126 | 82 | 38 |
Long-range† | 12 | 12 | 9 | 6 | 2 |
Corresponding heavy atoms‡ | 520 | 0 | 0 | 0 | |
No. of dihedral angle constraints § | 210 | 210 | 168 | 126 | 84 |
CYANA target function | 0.20 ± 0.07 | 5.19 ± 3.44 | 3.55 ± 2.31 | 1.26 ± 0.99 | 0.00 ± 0.00 |
Coordinate precision** | |||||
Backbone N, C, C' rms deviation, Å | 1.18 ± 0.33 | 1.80 ± 0.35 | 1.97 ± 0.30 | 3.76 ± 1.40 | 3.84 ± 1.37 |
All heavy atoms rms deviation, Å | 1.70 ± 0.34 | 2.69 ± 0.40 | 2.89 ± 0.28 | 4.64 ± 1.43 | 4.82 ± 1.41 |
*
Constraints of dmax(HN,O) = 2.0 Å and dmax(N,O) = 2.9 Å were set to be consistent with parallel, in-register b-sheets formed by residues 18–26 and 31–42, respectively.†
To enforce the salt bridge between residues D23 and K28 and correct intersheet stacking, dmax(D23Cg,K28Nz) = 2.8 Å and dmax(F19Ca,G38Ca) = dmax(A21Ca,V36Ca) = 8.0 Å were used, respectively.‡
The distance between corresponding heavy atoms of neighboring Ab(1–42) monomers was set to dmax(heavy,heavy) = 5.0 Å.§
To enhance the convergence of the structure calculation, the following backbone torsion angle constraints were used for the peptide segments 18–26 and 31–42: -164 £ F £ -74; 68 £ Y £ 158. The CYANA target function is the sum of the square of the distance between calculated and experimental constraints.**Relative to the mean coordinates, which were obtained by fitting the N, Ca, C’ positions of the peptide segments 18–41 of the 10 best CYANA conformers after the final calculation.
Table 2. Summary of Ab peptide properties
Peptide | Toxicity* | Fibril morphology† |
35L Ab(1–42) | 390 | WT (Fig. 2J) |
F19A, A21F | 1,400 | Premature fibrils (Fig. 2G) |
F19A, A21F, V36G, G38V | 540 | WT-like (Fig. 2H) |
K28D | 270 | Amorphous (Fig. 2A) |
D23K | 480 | Rod-like (Fig. 2B) |
D23K, K28D | 380 | WT-like, some premature (Fig. 2C) |
F19G | 760 | Rod-like (Fig. 2E) |
G38F | 890 | Seed-like (Fig. 2D) |
F19G, G38F | 2,100 | WT-like, mostly (Fig. 2F) premature |
35L Ab(1–40) | 530 | More regular than WT (data not shown) |
35M Ab(25–35) | 350 | n.d. |
n.d., not determined.
*Relative neurotoxicity as determined by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT)-formazan cell viability assay (see Supporting Text). Toxicities are given as EC50/nmol·liter-1. The EC50 is the peptide concentration, which shows the half-maximal effectiveness in the MTT-formazan assay.
†
Morphology as determined by negative staining EM (see Supporting Text).Supporting Text
Structure Comparison to the "Tycko Model" of Ab(1–40).
Previously, a model for fibrils of Ab(1–40) has been proposed on the basis of solid-state NMR studies and electron microscopy (1). Evidently, no experimental long-range constraints were used to generate the model Ab(1–40) fibrils, which is in stark contrast to the presented Ab(1–42) structure. Fig. 9 illustrates the significant deviations of this model from the presented structure of Ab(1–42): (i) In Ab(1–40), the side chain interactions are intramolecular, which is in contrast to the intermolecular nature of the side chain interactions in Ab(1–42) (Fig. 4A). (ii) In Ab(1–40), the inter-b-sheet side chain contacts are between odd residues of b1 and odd residues on b2, which is in contrast to Ab(1–42), where the intermolecular inter-b-sheet side chain contacts are between odd residues on b1 and even residues on b2. (iii) Taking the Ca positions of F19 and M35 as a reference, it is immediately apparent that the register of the inter-b-sheet side chain interactions differs by about 10 Å (Fig. 9). (iv) In Ab(1–40), the protofilament supposedly consists of two to four Ab(1–40) molecules per unit length, which is in contrast to Ab(1–42), which comprises only one Ab(1–42) molecule per protofilament unit length (Fig. 4E). (v) In Ab(1–40), the b1 strand is extended up to residue 12 (not shown), which is in contrast to Ab(1–42), where structural heterogeneity was detected for residue 1–17 (Fig. 1 E and F). A very recent publication also suggest that different types of Ab(1–40) fibrils can be obtained depending on the conditions of fibril formation (2). Amyloidogenesis of Ab(1–40) and Ab(1–42) might therefore follow different mechanisms and/or lead to different 3D structures. However, final conclusions on this will only be possible when an experimental structure of Ab(1–40) fibrils becomes available.H/D Exchange.
Homogenously 15N-labeled Ab(1–42) peptide that was used for the H/D-exchange studies was prepared as described in ref. 3. Using this procedure, the methionine M35 is chemically oxidized. Aged fibrils were collected by centrifugation at 100,000 ´ g for 10 min. Except for residue Q15, a complete chemical shift assignment of backbone amide resonances to individual amide groups in the peptide backbone of 35MoxAb(1–42) was obtained in perdeuterated DMSO containing 0.01% deuterated trifluoroacetic acid (TFA) using conventional strategies (4). To start the H/D exchange, the aged fibrils were resuspended in 10 mM Tris DCl (pH* = 7.7) containing 150 mM sodium chloride and D2O as the solvent. At suitable intervals, aliquots of the fibrillar suspension were sedimented at 100,000 ´ g for 10 min. The pellet was resuspended in pure D2O and frozen in liquid nitrogen to quench H/D exchange. The sample was then lyophilized in high vacuum (pressure < 40 mmHg). For the NMR analysis, Ab(1–42) fibrils were dissociated in DMSO containing 0.01% TFA.Within 30 s, the sample was diluted 10-fold into DMSO, yielding a final concentration of 0.01% TFA. The residual water concentration was about 100 mM as estimated from NMR measurements yielding slow intrinsic exchange rates in DMSO (Fig. 1C). Immediately after dissolving the fibrils, a series of 80 2D [15N,1H]-correlation spectra were measured for 4 h (3 min per spectrum). Residues that display slow exchange in the fibrils and residues that have high intrinsic exchange rates in DMSO would both yield intense peaks in the [15N,1H] correlation spectrum. To distinguish these two species, 0.5 M H2O and 0.5 M D2O were added to the sample, and a second series of 80 2D spectra were measured. Fig. 1C shows the peak volumes for the residues D7, V24, and G29. While G29 displayed an exchange rate of about 10-4 s-1, the intensity of the peaks corresponding to D7 and V24 remained virtually constant over the period of observation. However, after the addition of H2O/D2O (Fig. 1C, arrow), the volume of peak V24 remained unchanged while the peak D7 dropped to 50% of the initial intensity before the first spectrum could be acquired. This observation shows that the intrinsic exchange of D7 in DMSO is too fast to allow conclusions about the protonation state of this residue in the fibrils. Using this control measurement, the residues D1, A2, D7, H14, and A30 were excluded form the H/D exchange data analysis. The spectra were analyzed using the program PROSA (5) and a specially written VISUAL BASIC program in combination with EXCEL (Microsoft). Overall, our procedure yielded precise exchange rates for 88% of the amino acid residues in the sequence of Ab(1–42) and 96% of the amino acid in the core of the protofilament (see Figs. 1E and 4A).
Ab(1–42) Expression and Purification.
The strategy used to obtain monomeric Ab(1–42) is based on previous work (3, 6) but further improved and adapted to get a large amount of each of the mutated peptides. We cloned the DNA construct for the expression of the protein sequence (NANP)19-RSM- DAEFRHDSGYEVHHQKLVFFAEDVGSNKG-AIIGLLVGGVVIA into the pRSETA-Vector (Invitrogen) using the BamHI and EcoRI restriction sites. The substitution of MET35 by LEU was introduced to allow the separation of the 35LAb(1–42) peptide from the N-terminal fusion peptide by cyanogen bromide cleavage. Site-directed mutants of 35LAb(1–42) were generated by using the QuikChange Strategy. For peptide expression, freshly transformed Escherichia coli BL21(DE3) were grown to an OD600 of 1.0 in 1–2 liters of TB medium (7). The cells were harvested by centrifugation, resuspended in 100 ml of buffer G (100 mM sodium-phosphate/10 mM Tris·HCl, pH 8.0/6.0 mol·liter-1 guanidinium hydrochloride) and lysed by sonication (300 W for 6 min) on ice. Cell debris was removed by centrifugation at 18,000 ´ g for 30 min, and the supernatant was incubated with 12.5 ml of Ni-Sepharose (Pharmacia) for 1 h before packing it into a disposable Bio-Rad column. The column was washed with 10 volumes (vol) of buffer G, 10 vol of buffer U (10 mM Tris·HCl, pH 8.0/8.0 mol·liter-1 urea), and 4 vol of buffer A (40% (vol/vol) acetonitril). The fusion peptide was then eluted with buffer A containing 1.0 mol·liter-1 formic acid. Peptide containing fractions were combined, and an equal volume of 99% formic acid was added. Quantitative cleavage of the peptide was obtained by incubation at room temperature for 3 h in the presence of 50 mg/ml cyanogen bromide. The cleavage products were stored frozen at -80°C until usage. Monomeric Ab(1–42) was obtained by applying 20 mg of raw peptide in 0.7% (vol/vol) formic acid to a cation-exchange column (Toyopearl SP-650M, Tosoh Bioscience). The column was washed with 5 vol of buffer C1 (1.0% (vol/vol) acetic acid/1.0 mol·liter-1 sodium chloride in 20% (vol/vol) acetonitril), 5 vol of buffer C2 (1.0% (vol/vol) acetic acid in 20% (vol/vol) acetonitril), and finally 2 vol of buffer C3 (1.0% (vol/vol) acetic acid in water). Monomeric Ab-peptide was then eluted with 50 mM ammonium hydroxide (pH » 10.5). Amorphous aggregates were removed by centrifugation at 200,000 ´ g for 30 min. For each purification of Ab, new materials were used, including columns, Ni-Sepharose, cation exchanger, and polycarbonate ultracentrifugation tubes. The identity of each batch of Ab peptide was verified by MALDI-TOF mass spectrometry, and the peptide was stored on ice for no more than 4 days before usage. We obtained highly reproducible behaviors between different batches of peptide. This procedure yields seed-free Ab(1–42) stock solutions as indicated by the observation of a substantial lag phase of fibril formation under our experimental conditions (data not shown; see below).Electron Microscopy.
For cryoelectron microscopy and cryonegative staining, the grid with fibrillar 35MoxAb(1–42) solution at 50 mM was placed upside down on a drop of saturated ammonium molybdate for 10 s before blotting and vitrification. The method of cryonegative staining was applied previously for imaging Ab fibrils and intermediates with improved contrast (8). The frozen grids were mounted under liquid nitrogen in a cryospecimen holder and examined at approximately -172°C in a Philips CM12 electron microscope operated at 80 kV. Images were recorded on Kodak 4489 (Eastman Kodak) electron image film at a nominal magnification of ´45,000. Magnification calibration was performed by using catalase crystals.For negative staining, EM grids (EMS, formvar carbon-coated, 200 MESH) were layered on top of one drop of sample. After incubation, excess liquid was carefully removed from the grid by using a filter paper before washing the grid with three drops of distilled water. The EM grid was stained for 1 min with one drop of 2% (w/v) uranylacetate and then analyzed on a JEOL JEM-100CXII electron microscope at 80 kV with nominal magnifications between ´10,000 and ´72,000. Images were recorded digitally by using the SIS Megaview III imaging system.
Fibril Formation and ThioT Assay.
We used stock solution of 250 mM freshly prepared Ab(1–42) peptides (see above) in 50 mM ammonium hydroxide (Ab stock). To induce fibril formation, the 4 volumes of Ab stock were mixed rapidly with 1 volume of 5-fold concentrated aggregation buffer (5´ buffer, 1.0 mol·liter-1 Tris·HCl, pH 7.7), which yielded a final pH of 8.0. All samples were incubated at room temperature. ThioT fluorescence was measured by mixing 8–10 ml of fibril suspension with 1 ml of 50 mM glycine·NaOH buffer that contained 10 mM ThioT. This yielded an Ab(1–42):ThioT ratio of about 1:5. The ThioT fluorescence was then detected using an excitation wavelength of 442 nm and an emission wavelength of 485 nm. The ThioT signal was normalized to the fluorescence signal of the single tyrosine in Ab(1– 42) (excitation 278 nm, emission 307 nm). Each experiment was repeated at least twice.Structure Calculation.
The calculation of protein structures from NMR-derived upper limit restraints (UPLs) is a well established technique (4). Toward the calculation of the Ab(1–42) protofilament structure, we have obtained an number of UPLs. Pairwise mutagenesis showed that the side chains of F19/G38 and A21/V36 form intermolecular inter-b-sheet interactions. This suggested that the Ca positions of residue pairs should be no further apart than about the sum of their side chains’ length plus one van der Waals diameter of a carbon atom. Hence, two long-range UPLs per monomer of dmax(F19Ca,G38Ca) = 8.0 Å and dmax(A21Ca,V36Ca) = 8.0 Å were obtained. Note that these distances are close to the broad x-ray fiber diffraction band of ~10 Å, which has been observed in fibril diffraction studies and attributed to intersheet distances (9). The intermolecular salt bridge between residues D23 and K28, which has also been identified by pairwise mutagenesis, was implemented by the intermolecular UPL of dmax(D23Cg,K28Nz) = 2.8 Å. Next, H-bond constraints, as identified from H/D exchange, were set to be consistent with intermolecular, parallel, in-register b-sheets (10) formed by residues 18–26 (b1) and residues 31–42 (b2). H bonds were implemented in a standard manner (11) by two UPLs per H bond of dmax(HN,O) = 2.0 Å and dmax(N,C’) = 2.9 Å between the corresponding intermolecular H-bonding partners. These H-bond constraints also provided backbone torsion angle constraints that restrict the peptide backbone conformation of residues 18–26 (b1) and residues 31–42 (b2) to the region of the Ramachandran plot of extended conformation (-164 £ F £ -74 and 68 £ Y £ 158). This is analogous to nuclear Overhauser effect data-derived torsion angle constraints, which are used to enhance the performance of the structure calculation (12).Initial structure calculations were performed for an Ab(1–42) pentamer using the software package CYANA (13) starting from 100 randomized conformers, and the 10 conformers with the lowest CYANA target functions were used for further analysis (Table 1, second column). A relatively well defined structure was obtained for the middle molecule of the Ab(1–42) pentamer illustrated by the backbone rms deviation (rmsd) of 1.8 Å (Fig. 7B). This structure is very similar to the final result (Fig. 7A). However, the CYANA target function was relatively high, indicating a significant number of UPL distance violations. This high target function appeared likely to be caused by convergence problems of the calculation.
By taking advantage of the polymeric, repetitive nature of the Ab(1–42) fibril, we refined the structure calculation by implementing polymer constraints of UPL (heavy, heavy) = 5.0 Å between all corresponding heavy atoms in the fibrillar core of Ab(1–42). Note that the size of the polymer constrains are close to the sharp diffraction band of 4.8 Å, which is observed in Ab fibril diffraction studies, and which have been interpreted as interstrand distances (9). Further evidence of a repetitive structure of Ab fibrils is the observation of good quality solid-state NMR spectra, which are indicative of a very similar chemical environment for most of the spins in the fibrils (10). The corresponding UPLs can therefore also be regarded as experimental constraints.
For the refined calculation of the Ab(1–42) protofilament structure, all H-bond constraints, backbone torsion angle constraints, salt bridge constraints, inter-b-sheet constraints, and polymer constraints were used, yielding a total of 702 UPLs and 210 angle constraints for an Ab(1–42) pentamer (Table 1, first column). After the CYANA calculation, the 10 conformers with the lowest CYANA target functions had a backbone rmsd of 1.18 Å for residues 18–41 (Table 1) and a small target function of 0.2, which indicates minimal constraint violations. In these final conformers, the deviation from ideal geometry was minimal. The bundle of the 10 best conformers (Fig. 7A) shows that both the peptide backbone and the side chains are well defined. We therefore concluded that the experimental input data represent a self-consistent set and that the constraints are well satisfied in the calculated 10 best conformers.
To inquire about the influence of the polymeric nature of the Ab(1–42) fibril in the structure calculation, a series of calculations with Ab(1–42) dimers to Ab(1–42) pentamers were performed (Table 1). Inspection of Table 1 shows that the backbone rmsd values relative to the refined structure are strongly correlated with the number of molecules included in the calculation. The highest rmsd value of 3.8 Å was observed for the dimer calculation, indicating that the structure is not defined. Thus, the repetitive nature of the fibril restricts the conformational space of the Ab(1–42) molecules dramatically. We also performed structure calculations varying the polymer UPLs between 5 and 8 Å and the interstrand distances between 8 and 10 Å (data not shown). These modifications did not significantly influence the final outcome.
Cell Viability Test.
The 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assays were done in 96-well microtiter dishes containing 100 ml of medium per well. The B12 cells were plated in medium containing 5% dialyzed sera, and the day after plating, the test reagents were added. After 20 h, 10 ml of a 5 mg/ml MTT stock in PBS was added per well, and the incubation continued for 2 or 4 h. Finally, 100 ml of a solution containing 50% dimethylformamide and 20% SDS (pH 4.8) was added. The next day, absorption values at 570 nm were determined with an automatic microtiter plate reader.1. Petkova, A. T., Ishii, Y., Balbach, J. J., Antzutkin, O. N., Leapman, R. D., Delaglio, F. & Tycko, R. (2002) Proc. Natl. Acad. Sci. USA 99, 16742-16747.
2. Petkova, A. T., Leapman, R. D., Guo, Z., Yau, W. M., Mattson, M. P. & Tycko, R. (2005) Science 307, 262-265.
3. Dobeli, H., Draeger, N., Huber, G., Jakob, P., Schmidt, D., Seilheimer, B., Stuber, D., Wipf, B. & Zulauf, M. (1995) Bio/Technology 13, 988-993.
4. Wuthrich, K. (1986) NMR of Proteins and Nucleic Acids (Wiley, New York).
5. Guntert, P., Dotsch, V., Wider, G. & Wuthrich, K. (1992) J. Biomol. NMR 2, 619-629.
6. Fezoui, Y., Hartley, D. M., Harper, J. D., Khurana, R., Walsh, D. M., Condron, M. M., Selkoe, D. J., Lansbury, P. T., Jr., Fink, A. L. & Teplow, D. B. (2000) Amyloid 7, 166-78.
7. Sambrook, J. & Russell, D. W. (2001) Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press, Cold Spring Harbor, NY), Vol. 3.
8. Bohrmann, B., Adrian, M., Dubochet, J., Kuner, P., Muller, F., Huber, W., Nordstedt, C. & Dobeli, H. (2000) J. Struct. Biol. 130, 232-246.
9. Kirschner, D. A., Abraham, C. & Selkoe, D. J. (1986) Proc. Natl. Acad. Sci. USA 83, 503-507.
10. Balbach, J. J., Petkova, A. T., Oyler, N. A., Antzutkin, O. N., Gordon, D. J., Meredith, S. C. & Tycko, R. (2002) Biophys. J. 83, 1205-1216.
11. Guntert, P., Mumenthaler, C. & Wuthrich, K. (1997) J. Mol. Biol. 273, 283-298.
12. Guntert, P., Billeter, M., Ohlenschlager, O., Brown, L. R. & Wuthrich, K. (1998) J. Biomol. NMR 12, 543-548.
13. Herrmann, T., Guntert, P. & Wuthrich, K. (2002) J. Mol. Biol. 319, 209-227.
14. Gordon, D. J., Sciarretta, K. L. & Meredith, S. C. (2001) Biochemistry 40, 8237-8245.