Robert J. Martab,
Rachel J. Erringtonc,
Catherine L. Watkinscd,
Sally C. Chappellc,
Marie Wiltshirec,
Arwyn T. Jonesd,
Paul J. Smithc and
Rudolf K. Allemann*ab
aSchool of Chemistry, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, UK. E-mail: allemannrk@cardiff.ac.uk; Fax: +44 (0)29 208 74030; Tel: +44 (0)29 208 79014
bCardiff Catalysis Institute, Cardiff University, Main Building, Park Place, Cardiff CF10 3AT, UK
cInstitute of Cancer and Genetics, School of Medicine, Cardiff University, Heath Park, Cardiff CF14 4XN, UK
dCardiff School of Pharmacy and Pharmaceutical Sciences, Cardiff University, Redwood Building, King Edward VII Avenue, Cardiff CF10 3NB, UK
First published on 13th August 2013
Dynamic physical interactions between proteins underpin all key cellular processes and are a highly attractive area for the development of research tools and medicines. Protein–protein interactions frequently involve α-helical structures, but peptides matching the sequences of these structures usually do not fold correctly in isolation. Therefore, much research has focused on the creation of small peptides that adopt stable α-helical structures even in the absence of their intended protein targets. We show that short peptides alkylated with azobenzene crosslinkers can be used to photo-stimulate mitochondrial membrane depolarization and cytochrome c release in permeabilised cells, the initial events of the intrinsic apoptosis pathway.
Programmed cell death is normally activated by different forms of cellular stress signals such as genotoxic or oxidative damage, but it is frequently dysregulated in cancerous cells, allowing aberrant cells to survive and multiply. Interactions between members of the B-cell leukaemia-2 (Bcl-2) family of proteins, which comprises both pro-apoptotic (Bak, Bim, Bid, Bad, PUMA, NOXA, Bik, Bmf) and anti-apoptotic (Bcl-2, Bcl-xL, Bcl-w, Mcl-1, A1) members, are critical for the regulation of programmed cell death. Members of the Bcl-2 family are characterised by the presence of up to four conserved Bcl-2 homology (BH) domains that contain α-helical peptide segments. These regions mediate interactions between pro- and anti-apoptotic proteins as illustrated by the solution NMR structure of the complex of Bcl-xL with a peptide derived from the BH3 region of Bak showing that the Bak peptide forms an amphipathic α-helix that binds to a hydrophobic groove on the surface of Bcl-xL.10 Normally, the concentration of proteins in a cell favour the formation of such inactive heterodimeric complexes, but stress triggers can result in increased production of pro-apoptotic BH3-containing proteins11 or the unmasking of previously buried BH3 regions. These additional signals tip the balance towards apoptosis and either liberate the key pore-forming proteins Bak and Bax from inactive complexes with anti-apoptotic proteins or otherwise activate them.12,13 Bak and Bax then homo-oligomerize in the mitochondrial outer membrane and form pore complexes that initiate loss of mitochondrial outer membrane polarisation (MOMP) and a catastrophic release of cytochrome c into the cytosol.14–16
In contrast to their well-defined structure in protein complexes, BH3 derived peptides are generally unstructured in solution and hence bind to their targets with relatively low affinity. Their α-helicity can be increased in an irreversible fashion by covalently attaching ‘staples’ in a metathesis reaction.17,18 Initial reports suggested that such stapled peptides could effectively cross cell membranes and positive annexin V assays were taken to indicate the onset of apoptosis in Jurkat19,20 and mouse embryonic fibroblast (MEF)21 cells. However, recent findings indicate that neither MEF nor Jurkat cells treated with stapled peptides become permeable to propidium iodide,23 suggesting that a combination of markers are required to assess the progress towards full apoptosis induction by BH3 peptides. The reversible photo-control of α-helicity of BH3 peptides and their affinities for target proteins has been achieved with an azobenzene-derived crosslinker (Fig. 1), thereby providing the potential to activate apoptotic processes in cellular systems in a reversible fashion.24 Stabilisation of the α-helical conformation was observed when the cross-linker was in the cis-configuration for peptides linked through cysteine residues with i, i + 7 spacing, while the trans-configuration was α-helix-stabilizing for i, i + 11 spacings. For complexes of Bcl-xL and fluorescent derivatives (N-terminal fluoresceinamide; FAM) of the Bak-derived peptides BakI81Fi,i+11-XL and BakI81Fi,i+7-XL, small changes in α-helix stabilisation have been shown to be mirrored by large differences in the stability of the peptide–protein complexes.24 The NMR solution structure of BakI81Fi,i+11-XL in complex with Bcl-xL revealed the molecular mechanism, by which the increased affinity of such biophotonic nanoswitches is achieved, and showed why the I81F mutation does not increase the affinity of the peptides for i, i + 11 spacing.25
Fig. 1 (A) Photo-isomerization of the water-soluble, cysteine-reactive azobenzene crosslinker with 360 nm light (or 720 nm two-photon irradiation)22 (B) UV-Vis spectra showing ∼75% switching to the cis-isomer followed by thermal reversion to the trans form. Idealised cartoons illustrating: (C) peptide crosslinked through i, i + 11 spaced cysteines; the trans-conformation encourages an α-helical in the dark state (FAM-Baki,i+11-XL, FAM-Bidi,i+11-XL) and a non-α-helical photoactivated state. (D) i, i + 7 Spaced cysteines (Ac-BakI81Fi,i+7-XL) reverse this with α-helical conformation is disfavoured in the dark state; whilst the light state stabilizes the α-helical conformation. |
Peptidea | Sequenceb |
---|---|
a Where Ac denotes N-terminal acetyl and FAM N-terminal fluoresceinamide.b Cysteines used for crosslinking are bold while other deviations from the wild type sequences are underlined (see ESI). | |
Ac-BakI81Fi,i+7 | Ac-72GVGRCLAIGDCINR87-NH2 |
FAM-Baki,i+11 | FAM-72GCVGRLAIGDCINR87-NH2 |
FAM-Bidi,i+11 | FAM-81DIIRCIAHLAVGDCDRSI101-NH2 |
Cancer cell lines can be divided into three general classes according to their apoptotic competence when challenged by BH3 mimics.27 Class A comprises cells with functional loss/dysfunction of sensing pathways or critical signal aggregators such as p53 resulting in a loss of the trigger signals which initiate apoptosis. Class B cells display loss of both Bak and Bax function and compromised pore formation. Class C cells show upregulation of anti-apoptotic Bcl-2 family members, which sequester elevated levels of pro-apoptotic molecules. In classes A and C, the ability of aberrant cells to commit to apoptosis remains intact but is latent. Here we have focused on class C cells, described as ‘primed for death’,28 since our peptides need only bind competitively to anti-apoptotic Bcl-2 proteins to release full-length activating pro-apoptotic to attempt to trigger the intrinsic apoptosis pathway. Human lymphoma SU-DHL-4 cells29,30 were made permeable through treatment with digitonin and the status of MOMP was monitored at the single cell level.31,32
Permeabilised cell systems may not fully reflect the behaviour of intact cells, but represent a compromise between complexity and accessibility. They provide quite well defined systems that can be analyzed without the complexities encountered in live cells.21,22 Apoptotic susceptibility can vary across a population of cells according to their position in the cell cycle and other as yet unknown factors. This heterogeneity is of crucial interest in the development of improved chemotherapeutics, making it highly desirable to see the effect of an identical apoptotic stimulus on an entire population of cells. However, azobenzene-modified peptides are typically delivered by lipidic transfection reagents22,33 or microinjection,26 producing an unavoidably heterogeneous peptide loading in target cells. Determining the differences in peptide loading and availability, e.g. whether the peptide resides in vesicles or free in the cytosol, at the same time as cell-to-cell variations in susceptibility are extremely challenging. Using permeablised cells ameliorates disparities brought about by variations in peptide delivery, compartmentalisation and potential active export of peptides from cells, whilst still representing a leap in complexity over in vitro testing with many more potential partners for BH3–protein interactions available.
Flow cytometry techniques allow correlation between the susceptibility of cells to our peptide agents and concentrations of other molecules or cellular properties via multi-wavelength fluorescence activated cell sorting, e.g. correlation between susceptibility to cytochrome c loss via immunofluorescence and cell cycle position as reported through the use of the spectrally compatible far-red DNA dye DRAQ5.34,35 Control experiments indicated that digitonin-permeablised cells in respiration buffer36 maintained their complement of anti-apoptotic Bcl-2 over at least a 30 minute period, with only a 1.5% fraction of cells exhibiting Bcl-2 levels below those found in intact cells (Fig. S7, ESI†) and hence provide a window of opportunity to tentatively assess activity in the presence of the cells complement of antiapoptotic proteins. MOMP was assayed using the mitochondria associating dye JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazol-carbocyanine iodide), which exhibits a green fluorescence in its positively-charged monomeric state, while association with healthy, membrane-polarised mitochondria neutralises the charge on the dye, promoting the formation of orange fluorescent J-aggregates.37,38 The level of cellular fluorescence at 580 nm and its distribution in the cell directly reports on mitochondrial health and morphology. SU-DHL-4 cells were pre-loaded with JC-1, then permeablised and maintained in respiration buffer before being treated with FAM-Baki,i+11-XL or FAM-Bidi,i+11-XL. In the dark state, FAM-Baki,i+11-XL and FAM-Bidi,i+11-XL adopt a mostly α-helical conformation (Fig. S3, ESI†) and hence can act as indicators of the potential efficiency for peptide-induced apoptosis. FAM-Baki,i+11-XL and FAM-Bidi,i+11-XL caused a sustained (>120 minute) decrease of membrane potential in a dose-dependent manner (Fig. 2A and B). FAM-Baki,i+11-XL was a more effective ‘initiator’ than FAM-Bidi,i+11-XL, despite both parent proteins being able to interact with the full spectrum of anti-apoptotic proteins. Confocal microscopy at 60 minutes showed a dramatic dual effect on the JC-1 aggregate signal (Fig. 2C); firstly, a severe loss of total fluorescence per cell and secondly the rearrangement of organelle morphology from a connected network to residual small punctuate structures. Clearly BH3-derived peptides alkylated with azobenzene cross-linkers have the ability to induce detectable physiological changes paralleling the early events in apoptotic cell death.
Fig. 2 Retention of orange fluorescent JC-1 signal after a 120 min incubation with (A) FAM-Baki,i+11-XL and (B) FAM-Bidi,i+11-XL (±s.d.). (C) Confocal laser scanning microscope image of SU-DHL-4 cells labeled with JC-1 and treated with 7.5 μM FAM-Bidi,i+11-XL; the orange, aggregated form of the dye initially showed complex, morphological mitochondrial patterns typical of a healthy cell. After 1 min of exposure peptide rapidly accumulated in the cell in a non-specific pattern and after 60 min the aggregate signal was greatly reduced and punctate (bar = 10 μm). |
FAM-Baki,i+11-XL proved to be a potent inducer of apoptosis, but acted less well as a photoswitch (Fig. 3A) with a maximal conversion of 35% to the less active cis-form. FAM-BakI81Fi,i+7-XL on the other hand displays a much larger change in its affinity to Bcl-xL upon irradiation than FAM-BakI81Fi,i+11-XL24 or FAM-Baki,i+11-XL.25 Ac-BakI81Fi,i+7-XL was synthesized and displayed superior photo-switching properties (65–70% cis) (Fig. 3B) and an increased half-life of the cis-form of 28 min at 37 °C. Photo-activation of Ac-BakI81Fi,i+7-XL molecules triggered a measurable decay in the mitochondrial membrane potential (Fig. 3C) despite the fraction of the ‘active’ peptide diminishing rapidly even with an improved initial switching efficiency of 70%. The ‘switch-on’ nature of the i, i + 7 spacing photoswitch is also useful for observing cytochrome c release as this irreversible change is induced with light, rather than merely delayed by a ‘switch-off’ approach.
Fig. 3 Typical absorbance spectra of cross-linked peptides in the dark (blue) and light-activated (red) state of (A) FAM-Baki,i+11-XL and (B) Ac-BakI81Fi,i+7-XL. (C) Typical cellular MOMP response after peptide activation. Geometric mean analysis of JC-1 aggregate fluorescence signal in permeabilized SU-DHL-4 cells 60 min after addition of Ac-BakI81Fi,i+7-XL in the dark and UV irradiated states and FAM-Baki,i+11-XL as a positive control. (D) Graph of the percentage of cells exhibiting cytochrome c release below a threshold in permeabilized SU-DHL-4 cell populations after 30 min exposure to 5.0 or 7.5 μM concentrations of Ac-BakI81Fi,i+7-XL with (red) and without (cyan) photoswitching. Controls include buffer only (green), FAM-Baki,i+11-XL (2.5 μM and 5.0 μM, red) and 5 μM ABT-737 (blue) (±s.d.). (E) Chart representing cytochrome c levels in permeabilized SU-DHL-4 cells exposed to ABT-737 (blue), irradiated (orange) and dark (cyan) 7.5 μM Ac-BakI81Fi,i+7-XL and no peptide (green) for 30 minutes depicting the sensitive sub-population that completes cytochrome c release before the bulk population. (F) Region analysis of cytochrome c release shows that delivery of a pulse of active peptide has a significant effect on apoptosis induction (±s.d.). |
Although MOMP necessarily precedes cytochrome c release we found, as has been observed by others,39 that the release of mitochondrial cytochrome c was more readily detected at early time points than MOMP changes. Using 30 minute incubation times to observe cytochrome c release, an early and committed step in apoptosis, allowed the matching of protein retention timescales with a single half-life of photoswitched Ac-BakI81Fi,i+7-XL. Our aim was to determine whether the test system showed evidence of a threshold for peptide-triggered cytochrome c release as a surrogate for apoptotic commitment or displayed any evidence of sub-population resistance. For the cytochrome c assay, SU-DHL-4 cells were permeablised and treated with the peptides for 30 minutes, then fixed, processed for cytochrome c detection and analysed by flow cytometry (see Materials and methods). The results suggest a finely balanced threshold for rapid commitment to apoptosis such that treatment with 5 μM of irradiated Ac-BakI81Fi,i+7-XL peptide was indistinguishable from its dark counterpart, while 7.5 μM of irradiated Ac-BakI81Fi,i+7-XL showed an increase in the fraction of cells releasing cytochrome c (Fig. 3D). Activation of components of the intrinsic apoptosis pathway resulted in the formation of a population of entirely cytochrome c depleted cells rather than an intermediate population (Fig. 3E); this result was similar to that obtained with the small molecule Bax BH3 mimic, ABT-737.40 Comparative analysis of a number of paired dark/irradiated responses showed that treatment with dark-adapted, non-α-helical Ac-BakI81Fi,i+7-XL produced only a minimal response (14.5 ± 2.0%) that was only slightly higher than that obtained with untreated cells (8.8 ± 1.3%). After irradiation, cytochrome c release increased significantly (33.2 ± 7.7%) (Fig. 3F). Nanoswitch effects were also evident by changes in the mitochondrial morphology.41 In control cells or cells exposed to unswitched Ac-BakI81Fi,i+7-XL, mitochondria formed a network of ring-like structures, whereas mitochondria exposed to photoswitched Ac-BakI81Fi,i+7-XL, FAM-Baki,i+11-XL or ABT-737 lost definition, forming dim punctuate and diffuse globular structures (Fig. S8, ESI†).
Peptide-based nanoswitches have key advantages over small molecule agents for modulating cellular pathways controlled by protein–protein interactions. Firstly, nanoswitches exploit native sequences replicating the target selectivity of the parent protein. This is critical in cases where a protein of interest binds to many different partners as is the case with Bcl-2 family members whose patterns of affinity for each-other establish a pre-set for the apoptotic trigger. Secondly, the controlled delivery of a pulse of ‘activated’ peptide both matches the temporal scale for protein–protein interactions and provides an in situ perturbation that is discrete, predictable, programmable and not subject to a ‘wash-out’ procedure. The need for generic programmable tools has been highlighted by the appreciation that cell fate, for example determined by the stress signal integrator p53, can be related to the periodicity versus amplitude of the signaling pathway output.42 The design and synthesis of photoswitches with longer switching wavelengths and either rapid43 or extended44,45 relaxation times as required, coupled with improved methods for peptide delivery to live cells,46,47 may eventually lead to photo-controlled drugs and invaluable tools to probe critical protein–protein interactions in intact cells and whole organisms.48,49
Footnote |
† Electronic supplementary information (ESI) available: Details of peptide synthesis, purification and identification, proteolytic and reductive stability, UV-Visible and circular dichroism spectra. Cell culture, and assay conditions. See DOI: 10.1039/c3mb70246d |
This journal is © The Royal Society of Chemistry 2013 |