Inducible fold-switching as a mechanism to fibrillate pro-apoptotic BCL-2 proteins

Abstract

Neurodegenerative diseases often are associated with cellular dysregulation that results in premature cell death or apoptosis. A common example is the accumulation of amyloid plaques that promotes the excessive expression of p38 mitogen-activated protein kinase. The increased abundance of this enzyme leads to mass phosphorylation and activation of a protein from the B-cell lymphoma 2 (BCL-2) family, BAX. BAX is the central regulatory protein for mitochondrial outer membrane permeabilization (MOMP), a poration process that commits cells to apoptosis by releasing death-propagating factors from the mitochondria. Recent reports identify a naturally occurring peptide, Humanin (HN), that could block amyloid-beta-associated neuronal apoptosis by interacting with BCL-2 proteins. We recently showed humanin interaction leads to the amyloid-like fibrillation of BAX and a second BCL-2 family member, BID. We proposed this as a novel anti-apoptotic mechanism that inhibits pro-apoptotic BCL-2 proteins from initiating MOMP by sequestering them into fibrils, a heretofore unprecedented phenomenon that involves refolding globular BCL-2 proteins rapidly into fibrils where they undergo significant alpha-helix to beta-sheet fold-switching. Here we seek to further characterize the fibrillation and fold-switch in conditions that are known to induce amyloid fibrillation.

Keywords: BAX; BID; amyloid; apoptosis; conformational change; electron microscopy; fibrils; fold-switching; humanin; mitochondrial outer membrane permeabilization; β-sheet.

Figure 1

BAX and BID switch folds from α-helical to β-sheet conformations upon fibrillation with HN. CD spectra show 5 μM samples of the proteins (solid grey lines) along with 50 μM samples of HN (dashed grey lines) and the resuspended pellets of fibrillation reactions after ultracentrifugation (solid black lines). Fibril spectra are plotted with a secondary y-axis on the right. A) BAX exhibits the most complete fold transition to β-sheets showing one minimum at 225 nm. B) BID does not exhibit clear β-sheet fold-switching by CD. Structural changes are indicated by broadening over the 210–220 nm region. C) The spectrum from fibrils produced using a premixture of BAX and BID shows a hybrid secondary structure. D) Fibrils interacting with ThT showing a fluorescence response.

Figure 2

BCL-2 and HN aggregation rates in various temperature conditions. Titrations were conducted at 10°C (blue circles), 24°C (green squares), 37°C (red diamonds), or at 37°C with HN titrated into reaction buffer alone (black triangles). BCL-2 protein concentrations were kept constant at A) 5 μM BAX, B) 5μM BID, or C) a mixture of 2.5 μM each of BAX and BID. BAX is generally more reactive with HN versus BID. The aggregation rate for BAX is inhibited by colder temperatures while raising it to 37°C has no effect. Conversely, for BID there is a clear enhancement by raising the temperature while lowering it to 10 °C follows the same titration curve as 24°C. The mixture shows the same behavior as BID with enhanced reactivity at all temperatures over BID alone, but the mixture does not reach the same reactivity as BAX. D) ThT reactivity of samples at 24°C (blue bars) versus 37°C (red bars). The higher temperature features reduced activity across all samples, but remains within error of the sample at 24°C. Delta values are the difference between fibril samples and control samples containing only BCL-2 proteins. This difference for HN only samples are calculated with a baseline measurement from the buffer. Error bars represent the values of duplicate samples.

Figure 3

BCL-2 and HN aggregation rates with increasing concentrations of detergent. Titrations were conducted with 0.00% OG (blue circles, reprinted from 24°C titrations), 0.25% OG (green squares), 1.00% OG (red diamonds), or with 1.00% OG and HN titrated into reaction buffer alone (black triangles). BCL-2 protein concentrations were kept constant at A) 5 μM BAX, B) 5μM BID, or C) a mixture of 2.5 μM each of BAX and BID. All titrations indicate that OG can enhance the reaction rate. BAX reactivity increased along with OG concentration while a trend is less obvious with BID. The mixture features increased reactivity at both OG concentrations, but 1.00% OG is less reactive versus titrations at 0.25% OG. D) ThT results with samples in 0% (blue bars) or 1% (red bars) OG buffers. Total fibrillation was reduced in the fibril samples, but activity was enhanced with the HN peptide alone. Delta values are the difference between fibril samples and control samples containing only BCL-2 proteins. This difference for HN only samples are calculated with a baseline measurement from the buffer. Error bars represent the values of duplicate samples.

Figure 4

BCL-2 and HN aggregation rates with increasing pH values. Titrations were conducted in various potassium phosphate buffers at pH 5.0 (blue circles), pH 6.0 (green squares), pH 7.0 (orange diamonds), pH 8.0 (red triangles), or at pH 8.0 with HN titrated into reaction buffer alone (black triangles). BCL-2 protein concentrations were kept constant at A) 5 μM BAX, B) 5μM BID, or C) a mixture of 2.5 μM each of BAX and BID. All titrations indicate that the reactivity rate is enhanced by increasing pH conditions. Titrations with mixture samples do not show a proportionate increase with pH as was observed with BAX or BID alone. This indicates some reactivity between the two proteins that effects the fibrillation mechanism. D) ThT fluorescence in pH 5 (blue bars) or pH 8 (red bars) buffer. Fibrillation is enhanced with increasing pH levels. Delta values are the difference between fibril samples and control samples containing only BCL-2 proteins. This difference for HN only samples are calculated with a baseline measurement from the buffer. Error bars represent the values of duplicate samples.

Figure 5

Negative-stain electron microscopy of fibril samples made with 5 μM of BAX, BID, or Mix and 50 μM HN in 1% (w/v) OG (left) or pH 8 (right) buffers with histograms featuring their width (blue) and length (red) distributions. BAX (A & B), BID (C & D), and Mix (E & F) fibrils all have narrow width distributions under these conditions. The most common fibril length is about 45 nm. It was noted that there is a distinct lack of long, networking BAX fibrils under these conditions. EM images are unaltered from their original capture.