Cholesterol and the Brain
By: Carol A. Rouzer, VICB Communications
Published: June 19, 2012
The interaction of cholesterol with amyloid precursor protein helps to explain why high cholesterol is a risk factor for Alzheimer’s Disease.
Alzheimer’s Disease (AD) is a devastating neurological condition that leads to dementia, primarily in elderly patients. (Figure 1). Currently one in eight older Americans suffers from AD, corresponding to 5.4 million people. AD is the sixth leading cause of death in the U.S., and the only major cause of death for which there are no effective therapies (Alzheimer’s Association). Both genetic and environmental factors contribute to the risk for developing AD. One of these risk factors is an elevated cholesterol level in neural cell membranes, although the mechanism by which cholesterol promotes the neurodegenerative changes observed in AD has not been explained. Now, Vanderbilt Institute of Chemical Biology investigator Chuck Sanders and his laboratory may have the answer [P. J. Barrett et al., (2012) Science, 336, 1168].
Figure 1. Proteolytic processing of APP. Cleavage of APP by α-secretase removes an ectodomain fragment designated APPsα, and a membrane-bound fragment, C83. γ-Secretase cleaves C83 to produce the nonamyloidogenic P3 and AICD peptides. Alternatively, cleavage of APP by β-secretase removes the APPsβ ectodomain fragment, yielding the C99 transmembrane fragment. Cleavage of C99 by γ-secretase produces the amyloidogenic Aβ1-40 and Aβ1-42 peptides and AICD. Image reproduced through the courtesy of Wikimedia Commons under the GNU Free Documentation License.
An important biochemical hallmark of AD is the accumulation of proteinaceous deposits known as amyloid plaques in brain tissue. The primary constituents of amyloid plaques are aggregated amyloid-β (Aβ) peptides, which are neurotoxic. These 40- to 42-amino acid peptides are derived from the amyloid precursor protein (APP), a single pass transmembrane protein that is widely expressed in many tissues. There are eight isoforms of APP, ranging from 365 to 770 amino acids in length. The exact function of APP is unknown, though evidence suggests that it plays a role in the formation and repair of neural synapses, where expression levels of the 695 amino acid isoform are particularly high.
APP undergoes extensive postranslational modifications, including proteolytic cleavage by a number of enzymes. There are two primary pathways for APP proteolysis. The first, nonamyloidogenic pathway is initiated by the enzyme α-secretase, which produces a transmembrane C83 fragment and a soluble APPsα fragment. Subsequent cleavage of C83 by γ-secretase produces the soluble P3 peptide, leaving behind the membrane associated AICD fragment. Alternatively, β-secretase initiates the amyloidogenic pathway by cleaving APP to yield the transmembrane C99 fragment and the soluble APPsβ. Subsequent cleavage of C99 by γ-secretase yields AICD and Aβ peptides (Figure 1). Due to its role as the direct precursor of Aβ peptides, the Sanders lab performed NMR-based structural studies of C99 embedded in lysomyristoylphosphatidylglycerol detergent micelles.
The structural data indicated that the backbone of C99 comprises a number of distinct domains, including an unstructured N-terminus followed by a micelle-associated N-helix, which is connected by an N-loop to the helical transmembrane domain. The C-terminal end of the transmembrane domain is then joined by a long unstructured C-loop to a terminal, micelle-associated C-helix (FIgure 2). The investigators used power saturation electron paramagnetic resonance measurements of spin-labeled C99 to confirm that the general topology of the protein observed in micelles was consistent with that observed in lipid vesicles, including the span of the transmembrane domain and the surface association of the N- and C-helixes.
Figure 2.Diagrammatic representation of the C99 protein in a detergent micelle (tan sphere). Regions of the protein include an unstructured N-terminus, a micelle-associated N-helix, an N-loop that joins the N-helix to the curved helical transmembrane domain, a long unstructured C-loop and a micelle-associated C-helix. From P. J. Barrett et al., (2012) Science, 336, 1168. Reprinted with permission from AAAS. Readers may view, browse, and/or download material for temporary copying purposes only, provided these uses are for noncommercial personal purposes. Except as provided by law, this material may not be further reproduced, distributed, transmitted, modified, adapted, performed, displayed, published, or sold in whole or in part, without prior written permission from the publisher.
The transmembrane domain of C99 is highly curved, due in part, to the presence of two glycine residues (G708 and G709) near its center. Mutation of these residues to leucine partially straightened the curvature and reduced the flexibility of the domain. The investigators hypothesized that the curvature may facilitate binding of C99 in the sluice-like active site of γ-secretase and help to expose the target peptide bonds for cleavage. The extracellular region of the transmembrane domain is also notable for the presence of a glycine zipper (GXXXGXXXG). These structures typically promote self-association of proteins, but there is no strong evidence that this occurs with C99, possibly because the zipper region is blocked by the N-helix and N-loop.
The organization of the N-helix, N-loop, and the extracellular region of the transmembrane domain led the Sanders lab to hypothesize that this region of the protein might form a lipid binding region. They tested their hypothesis by 1H,15N-transverse relaxation-optimized spectroscopy (TROSY) NMR using uniformly 15N-labeled protein in bicelles as model membranes. In this technique, a signal is detected for every proton that is directly bound to a labeled nitrogen atom. Since every amino acid except proline has at least one such bond, TROSY NMR provides a comprehensive picture of the protein. The data are plotted with the 1H chemical shift on the X-axis versus the 15N chemical shift on the Y-axis, with each spot representing one amino acid. The technique is a highly sensitive method for detecting ligand binding to a protein since residues involved in the ligand interaction exhibit a change in chemical shift, which is observed as a movement of the corresponding spot on the graph (Figure 3). Using this technique, the Sanders lab demonstrated binding of cholesterol to C99 with saturation occurring at a ratio of 1:1, and an affinity constant of 5.1 ± 1.2 mol%.
Figure 3. 1H,15N-Transverse relaxation optimized spectroscopy NMR spectrum of C99 in the absence of cholesterol (blue) overlaid with the spectrum of C99 in the presence of 10 mol% cholesterol (red). Figure kindly provided by the Sanders lab.
As predicted, the spectral data showed that amino acids most involved in cholesterol binding were in the N-helix, N-loop, and extracellular region of the transmembrane domain. In particular, the resonances of G700, G704, G708, and G709 showed strong shifts on cholesterol binding (Figure 3). In addition the investigators performed alanine scanning mutagenesis of all amino acids in the range of 690 to 710. The results (Figure 4) suggested that N698 and E693 are likely involved in hydrogen bonding to cholesterol. The mutation data also revealed that some residues, such as S697 are not directly involved in cholesterol binding, even though they show substantial changes in nmr chemical shift upon cholesterol addition. The Sanders group proposed that these residues are important for N-loop flexibility, which may assist binding.
Figure 4. Results of Ala-scanning mutagenesis. The color indicates the effect of mutation of the residue to alanine on cholesterol binding. In order from greatest to least effect are red>yellow>green. From P. J. Barrett et al., (2012) Science, 336, 1168. Reprinted with permission from AAAS. Readers may view, browse, and/or download material for temporary copying purposes only, provided these uses are for noncommercial personal purposes. Except as provided by law, this material may not be further reproduced, distributed, transmitted, modified, adapted, performed, displayed, published, or sold in whole or in part, without prior written permission from the publisher.
Finally, the investigators noted that the presence of G700 and G704 at the outer face of the curved transmembrane helix makes the surface of the helix flat, a feature that should facilitate interaction with the relatively flat cholesterol molecule. A model of C99 with cholesterol bound is shown in Figure 5.
Figure 5. Model of cholesterol bound to C99. (Left) Diagram of the protein showing key residues interacting with the bound cholesterol. (Right) Space filling model of cholesterol bound to C99. The transmembrane domain is shown in blue, and the N-helix and N-loop are in green. The arrow shows the movement of the N-helix upon cholesterol binding. Left figure from P. J. Barrett et al., (2012) Science, 336, 1168. Reprinted with permission from AAAS. Right figure kindly provided by the Sanders lab. Readers may view, browse, and/or download material for temporary copying purposes only, provided these uses are for noncommercial personal purposes. Except as provided by law, this material may not be further reproduced, distributed, transmitted, modified, adapted, performed, displayed, published, or sold in whole or in part, without prior written permission from the publisher.
The association of cholesterol with C99 may be of great significance to its role as the precursor for Aβ peptides. It is important to note that this interaction occurs at a cholesterol concentration typically found in mammalian cell membranes, so it is likely to be physiologically relevant. In addition, both β- and γ-secretases preferentially associate with cholesterol-rich lipid rafts, and addition of cholesterol to membranes increases cleavage of C99 by γ-secretase. Thus, cholesterol binding to C99 may play a cofactor role in substrate recognition or catalysis. Furthermore, cholesterol binding blocks the α-secretase cleavage site, potentially preventing the nonamyloidogenic processing of APP. Finally, the Sanders group notes that the cholesterol binding site of C99 is in the Aβ domain of the protein. Therefore, cholesterol binding may also contribute to Aβ’s tendency to form neurotoxic aggregates. Together, the data provide important information that will aid in our understanding of the role of cholesterol in AD pathogenesis.