Inhibition of Toxicity in the β-Amyloid Peptide Fragment β-(25–35) Using N-Methylated Derivatives

β-(25–35) is a synthetic derivative of β-amyloid, the peptide that is believed to cause Alzheimer's disease. As it is highly toxic and forms fibrillar aggregates typical of β-amyloid, it is suitable as a model for testing inhibitors of aggregation and toxicity. We demonstrate thatN-methylated derivatives of β-(25–35), which in isolation are soluble and non-toxic, can prevent the aggregation and inhibit the resulting toxicity of the wild type peptide.N-Methylation can block hydrogen bonding on the outer edge of the assembling amyloid. The peptides are assayed by Congo red and thioflavin T binding, electron microscopy, and a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) toxicity assay on PC12 cells. One peptide (Gly25 N-methylated) has properties similar to the wild type, whereas five have varying effects on prefolded fibrils and fibril assembly. In particular, β-(25–35) with Gly33 N-methylated is able to completely prevent fibril assembly and to reduce the toxicity of prefolded amyloid. With Leu34 N-methylated, the fibril morphology is altered and the toxicity reduced. We suggest that the use of N-methylated derivatives of amyloidogenic peptides and proteins could provide a general solution to the problem of amyloid deposition and toxicity.

Alzheimer's disease (AD) 1 is the most common form of senile dementia. ␤-Amyloid (A␤), a 39 -43-amino acid ␤-sheet peptide, aggregates in the brain to form the major component of characteristic deposits known as senile plaques (1)(2)(3)(4). X-ray diffraction data have shown that the conformation of A␤ is characterized by an antiparallel cross-␤-pleated sheet (5), although more recent solid state NMR evidence suggests that the peptide has a parallel ␤-sheet structure (6). Nevertheless, aggregation occurs because of hydrogen bonding between ␤-strands, and the resulting fibrils have axes perpendicular to the ␤-strand and parallel to the cross-linking hydrogen bonds (5).
Of all of the A␤ derivatives studied so far, ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), se-quence GSNKGAIIGLM, is the shortest fragment that exhibits large ␤-sheet fibrils and retains the toxicity of the full-length peptide (2,(7)(8)(9). It has been proposed that ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) represents the biologically active region of A␤. In vitro studies have shown that it does not require aging to aggregate and become toxic (8 -10), unlike the full-length peptide. As with A␤-␤, toxicity is dependent on the aggregation state of the peptide, because ␤-(25-35) that has been solubilized and unfolded in 35% acetonitrile (AcN), 0.1% trifluoroacetic acid is nontoxic (8,11). In this study, ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) has been chosen as a model for full-length A␤ because it retains both its physical and biological properties, while its short length readily allows derivatives to be synthesized and studied. A great deal of evidence, much of which comes from studying hereditary forms of the disease, supports the view that A␤ aggregation is implicated in AD (1)(2)(3)(4)(12)(13)(14)(15)(16). Controversy has raged, however, over whether these fibrils are actually a cause or a consequence of the disease. Although many mutations in the amyloid precursor protein gene have been linked to premature onset of AD, the amount of amyloid deposited in the brain does not necessarily correlate with disease severity. A resolution of this apparent paradox may be that it is a ␤-structured protofilament, which subsequently forms fibrils and then plaques, that is the pathogenic element (17,18). What is clear is that the most important factor influencing A␤ toxicity is its aggregation state, because only the non-soluble fibrillar form is neurotoxic (2,11,19). In view of the role fibril formation plays in AD and other diseases, it has been proposed that a valid therapeutic strategy would be to administer compounds that can block fibril or protofilament formation by binding to their ends (20,21). Designed peptides have already been shown to act as "␤-sheet breakers," inhibiting amyloid formation and lowering toxicity in Alzheimer's peptides and prion proteins (20,(22)(23)(24)(25)(26)(27). Our proposal is that N-methylated peptide derivatives can also act as ␤-sheet breakers.
Quantitative Congo Red-Quantitative measurement of Congo red (CR) binding was carried out essentially as described in Refs. 30 -32. Readings were taken on a Kontron Uvikon 922 spectrophotometer at 480 and 540 nm, and the amount of CR bound was calculated as follows: Cb (CR bound) [M] ϭ (A 540 /25295) Ϫ(A 480 /46306). Error bars were calculated from the range between duplicate readings.
Thioflavin T (ThT) Fluorescence-ThT fluorescence was carried out as described in Refs. 32 and 33. Measurements were carried out in triplicate on a Perkin-Elmer luminescence spectrophotometer LS50B using FLWINLAB software and an integration time of 5 s. All data were scaled using ␤-(25-35) alone as 100%. Scaled figures were plotted using the standard deviation to generate the error bars.
Electron Microscopy-Electron microscopy (EM) was carried out to determine whether the NMe derivatives were able to alter wild type peptide morphology. Peptides were solubilized at 500 M (1 mM total peptide concentration for 1:1 combinations), and 50-l drops were applied to glow discharged, carbon-coated, 400-mesh copper grids. Peptides were then negatively stained with 2% uranyl acetate before being viewed on a Philips EM 301 electron microscope at 100 kV using a magnification of 75,000.
Toxicity Assay-MTT reduction was carried out as described in Refs. 9 and 34 using rat pheochromocytoma (PC12) cells, with cells subcultured 50/50 1 day prior to assay. Prior to peptide application, cells were treated with 1 mg/ml DNase to break up any clumps, a feature of PC12 cell growth. This allows for a more accurate cell count. An overnight incubation at 37°C followed mixing of peptides and cells before the addition of MTT, and after a further 2-h incubation cells were lysed and read at 550 nm. All readings were carried out in triplicate. Positive controls consisted of addition of 20 mM MOPS, pH 7, to cells, and negative controls consisted of the addition of 0.1% Triton X-100, which lyses cells and abolishes MTT reduction. Assay values for positive controls were taken as 100%, and complete inhibition of cell function, 0%, was represented by the negative control. Raw data were then scaled as follows: (each raw data point Ϫ negative mean)/(positive mean Ϫ negative mean) ϫ 100. The scaled mean for each data set was then plotted with the error given by the standard deviation.
Reduction of ␤-(25-35) Toxicity as Measured by MTT Assay-Aggregation of A␤ and the related peptide ␤-(25-35) is known to be an important factor in toxicity because non-fibrillar soluble forms are nontoxic (11). We studied toxicity using the established MTT assay, which measures the redox potential of cells, in this case PC12 cells, and thus monitors cell condition (9). Healthy cells reduce MTT, turning the redox dye from yellow to purple/blue, whereas unhealthy cells show less of a color change. This assay is specific for A␤ toxicity, and both A␤ and ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) inhibit the cellular reduction of MTT (35).
All peptides were prepared and mixed as described above using a concentration range of 20 nM-400 M ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). Wild type peptide was assayed alone and in combination with 100 M NMe peptides to see whether the ED 50 for the ␤-(25-35) toxicity curve was altered. Interestingly, at 400 M ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) toxicity is reduced compared with slightly lower concentrations, consistent with protofibrils being the toxic form rather than highly aggregated and structured amyloid. It is possible that at this concentration aggregation is near completion, and it may be the process of aggregation that is toxic rather than the presence of the aggregates themselves (36). Apart from NMeGly 25 , none of the NMe peptides are toxic in isolation. On mixing with ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) we found that the NMe peptides had different effects depending on the mixing conditions. NMeGly 25 shows the same toxicity profile as wild type, and NMeAla 30 and NMeIle 31 have little effect using either of the mixing condi-tions, whereas NMeGly 29 and NMeLeu 34 have some inhibitory effect on toxicity when mixed with unfolded wild type. NMeGly 33 , however, appears to inhibit toxicity when premixed with either folded or unfolded ␤-(25-35) but with a greater effect when mixed with the unfolded peptide. Fig. 6 shows examples of some of these results, and Tables I and II provide a summary of the results. DISCUSSION The purpose of this study is to assess whether N-methylated derivatives of ␤-(25-35) can successfully inhibit fibril formation and the resultant toxicity of this amyloidogenic peptide. An important factor in the success of inhibition is the structural state of ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) when it is mixed with the potential inhibitor. Premixing peptides with unfolded wild type is generally more effective than premixing with aggregated wild type, because the latter requires the inhibitor to insert into and break up existing fibrils. We demonstrate that inhibition can be achieved using these NMe derivatives and that the outcome depends on which residue has been N-methylated.
NMeGly 25 is a ␤-sheet peptide that aggregates and has the same toxicity profile as the wild type peptide. Fibrils typical of wild type are seen under EM, and the same level of CR binding occurs. It does not have the same effect upon ThT fluorescence, however, and generates a very small amyloid-induced peak. The addition of an NMe group at the N terminus of ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) still allows the peptide to fold, aggregate, and thus become toxic, but it clearly affects the mechanism of ThT binding. As with CR, the binding mechanism of ThT to amyloid is unknown (20,23), but these results suggest that they must be binding different parts of the peptide because for NMeGly 25 , CR binding is typical of amyloid, whereas ThT binding is not. If the mechanism of binding of these dyes could be understood, it might give some insight into the structure of amyloid.
The remaining NMe peptides are all non-aggregating and nontoxic in isolation, demonstrating that the amide NH groups of residues 29, 30, 31, 33, and 34 are all essential to amyloid formation. These peptides show different levels of effectiveness as inhibitors of ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) aggregation and toxicity, and this does not appear to be related to secondary structure. Analysis of these peptides in isolation using circular dichroism (data not shown) shows that whereas NMeAla 30 , NMeGly 29 , and NMeGly 33 in aqueous solution are predominantly random coil, only NMeGly 33 is an effective inhibitor of both aggregation and toxicity. NMeIle 31 and NMeLeu 34 both show elements of ␤-sheet structure in isolation, but only NMeLeu 34 is effective at altering wild type fibrils and inhibiting toxicity. Clearly, the position of the NMe group is most important and determines how the derivative interacts with the wild type peptide.
NMeGly 33 is the most effective inhibitor of ␤-(25-35) aggregation and toxicity. When mixed with aggregated wild type, although CR binding and ThT fluorescence are comparable with wild type alone, EM shows that fibrils are much smaller and finer in appearance. There is also inhibition of toxicity, not seen with any of the other NMe peptides using these mixing conditions. When premixed with unfolded ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), the effects are more dramatic. CR binding is abolished, the ThT fluorescence peak falls to less than 10% of the wild type alone, and no fibrils are visible under EM. Most importantly, NMeGly 33 can inhibit ␤-(25-35) toxicity when present in equimolar amounts. The midpoint of the toxicity curve is shifted from 30 to 120 M ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). It should be noted, however, that when wild type peptide is completely unfolded in 35% AcN, at lower peptide concentrations, aggregation, and thus toxicity, does not reproducibly return during the time scale of the experiment. If it did return to the levels typical of those when wild type is solubilized in water, then NMeGly 33 would have shifted the midpoint of the curve from 3 to 120 M. This means that PC12 cells can tolerate the presence of 40-fold more ␤-(25-35) with 100 M NMeGly 33 than without it.
Mixing NMeGly 33 and NMeLeu 34 with unfolded wild type again highlights the differences in the binding mechanisms of CR and ThT. Although NMeGly 33 causes a reduction in the effect of ␤-(25-35) on both dyes, NMeLeu 34 increases the effect on CR binding but decreases the effect on ThT fluorescence. With NMeL34, the long fibril structures observed under EM must permit CR to bind but alter the binding of ThT. With NMeGly 33 , the wild type fibrils must either be reduced to a size too small to bind either dye or be observable under EM or they must disrupt fibril formation altogether.
The and aggregation levels (11). This conclusion is supported by our results, but it should be noted that not all changes to fibril morphology result in a reduction of toxicity. With NMeGly 33 , the disruption of fibrils does indeed lead to an inhibition of wild type toxicity. In the case of NMeLeu 34 , however, the morphological change is not great enough to have an inhibitory effect, illustrating that morphology must be altered in particular ways for the NMe peptide to be an effective inhibitor. Both NMeGly 33 and NMeLeu 34 alter the wild type fibrils under both conditions, but NMeGly 33 is clearly the better inhibitor of wild type toxicity and is most effective when fibrils are abolished or reduced to levels not visible under EM. It appears that these NMe derivatives are not simply binding to the end of a growing ␤-(25-35) fibril and blocking further addition of monomers but  (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35). All data are scaled with 100% MTT reduction, represented by positive controls, and 0%, represented by negative controls. Each data set was carried out in triplicate with the standard deviation generating the errors. that their structural effects are clearly more complicated. If it is the protofilament precursors of amyloid fibrils that are neurotoxic (17,18), then it is possible that the N-methylated peptides are exerting their inhibitory effects via these elements rather than fully formed fibrils. Nevertheless, the effectiveness of NMeGly 33 in particular does validate this strategy of using NMe derivatives of ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) and full-length A␤ to prevent aggregation and toxicity.
␤-(25-35) is a particularly intractable peptide because it aggregates rapidly, unlike the full-length A␤ peptide, which requires aging for up to a week before it aggregates and becomes toxic (2). ␤-(25-35) is very toxic and, according to previous work, requires only nanomolar levels to have an effect (2), although we find toxicity is apparent only above 1 M. It is therefore not surprising that although some inhibition is seen when NMeGly 33 is added to aggregated ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35), showing that the peptide must be in a dynamic state of equilibrium between the folded and unfolded state, inhibition works best when added to unfolded ␤-(25-35) before applying conditions that promote aggregation. Previous work done with ␤-sheet breaker peptides on full-length A␤ has proved effective to varying degrees (20, 22-27, 37, 38), and this provides encouragement that if NMe derivatives of ␤- (25)(26)(27)(28)(29)(30)(31)(32)(33)(34)(35) are added to ␤-(1-42) they will be able to disrupt the aggregation and toxicity of this peptide also. Future work will involve testing these derivatives and possibly NMe derivatives of A␤ itself on the full-length AD peptide.
We believe that the use of NMe derivatives could provide a general strategy for inhibiting the aggregation and toxicity of amyloid. There are more than 20 diseases that are the result of amyloid-like aggregation of particular peptides and proteins, including prion diseases, type II diabetes mellitus, Huntington's disease, and Parkinson's disease (39 -41). The use of peptides as therapeutic agents, however, is problematic in two main ways. First, they must be able to cross the blood-brain barrier, and second, they must be able to avoid degradation by proteases in the brain. Previous work has shown that not only are the D-amino acid forms of A␤ and ␤-(25-35) equally toxic to their L-amino acid counterparts but that short D-amino acid ␤-sheet breaker peptides can bind to either D or L forms of A␤ (42). D-Amino acids are not susceptible to the natural proteo-lytic processes in the brain (37,42), and therefore constructing NMe derivatives from D-amino acids could overcome one of the hurdles of creating an effective drug from these peptides. The use of small derivatives may also overcome the problem of getting these peptides to cross the blood-brain barrier (37), particularly if, like NMeGly 33 , the peptide is amphipathic. With these facts in mind, it is possible that the NMe derivatives of amyloidogenic peptides and proteins could provide a general and valid lead to therapeutic drugs for a large number of diseases that are currently untreatable.