Conceived and designed the experiments: MNW KGH. Performed the experiments: KGN JMR. Analyzed the data: KGN KE KGH. Contributed reagents/materials/analysis tools: KE MNW KGH. Wrote the paper: KGH.
Current address: Department of Applied Biochemistry, DSM Biotechnology Center, Delft, The Netherlands
The authors have declared that no competing interests exist.
Recent studies have shown that fluorescently labeled antibodies can be dissociated from their antigen by illumination with laser light. The mechanism responsible for the photounbinding effect, however, remains elusive. Here, we give important insights into the mechanism of photounbinding and show that the effect is not restricted to antibody/antigen binding.
We present studies of the photounbinding of labeled calmodulin (CaM) from a set of CaM-binding peptides with different affinities to CaM after one- and two-photon excitation. We found that the photounbinding effect becomes stronger with increasing binding affinity. Our observation that photounbinding can be influenced by using free radical scavengers, that it does not occur with either unlabeled protein or non-fluorescent quencher dyes, and that it becomes evident shortly
The experimental results exclude surface effects, or heating by laser irradiation as potential causes of photounbinding. Our data suggest that free radicals formed through photobleaching may cause a conformational change of the CaM which lowers their binding affinity with the peptide or its respective binding partner.
Fluorescent probes are commonly used in biological experiments and have provided enormous insight into cell machinery and protein dynamics. Despite their successful application over the last century, fluorescent conjugates can influence cell viability and the properties of the molecules under study
Recently it has been demonstrated that fluorescently labeled molecular complexes such as antibody-antigen
One important requirement for studying photounbinding is an assay that allows us to distinguish between the loss of a binding partner (photounbinding) from the loss of fluorescence by photobleaching. We have found that immobilizing one binding partner on a coverglass via a long chemical cross-linker
In the present photounbinding study, the emphasis was put on the dependence of the photounbinding phenomena on the initial dissociation constant of the molecular system under various experimental conditions in order to elucidate its underlying mechanism. To be able to perfom measurements using a single molecular system, we studied the binding of the signaling molecule calmodulin (CaM) to a family of peptides that mimic the CaM-binding domain of Ca2+/(CaM) dependent protein kinase II (CKII)
The introduction of a single Cys residue by conversion of Asp at amino acid 3 to Cys in a pET23d CaM expression plasmid was described previously
CaM labeling was performed as described previously
All CKII peptides
Peptide | Sequence | Kd×10−13 [ |
CKII(290–312) | CLKKFNARRKLKGAILTTMLATRN | 3 |
CKII(292–312) | CKFNARRKLKGAILTTMLATRN | 5 |
CKII(293–312) | CFNARRKLKGAILTTMLATRN | 17 |
CKII(294–312) | CNARRKLKGAILTTMLATRN | 570 |
CKII(290–312) |
LKKFNARRKLKGAILTTMLATRN | 3 |
*: high affinity peptide used for unspecific background determination as described in
Synthesis was performed with assistance of the Protein Chemistry Facility of the Research Institute of Molecular Pathology, Vienna, Austria. The peptides were purified with High Performance Liquid Chromatography (HPLC) and verified by Mass Spectrometry.
A selected CMKII peptide was covalently bound via a SM(PEG)8 crosslinker (MW 689.7) onto a coverslip by amino-silylation following the protocol recommended by Pierce (#80370, #22108), which is similar to the one described in Heinze 2009. The coverglasses were incubated with a 1 mM CKII peptide solution overnight at 4°C, rinsed thoroughly and incubated with CaM-A488, CaM-A647 (3 µM) or unlabeled CaM (60.4 µM) in buffer (25 mM MOPS, 150 mM KCL, 0.5 mM CaCl2, 0.1 mg/ml BSA) overnight at 4°C. Finally, the coated chamber was rinsed again and filled with 10 mL CaM-buffer. Proper coating was verified by fluorescence imaging.
When using peptides with lower binding affinity times, the periods between rinsing after re-incubation and imaging were kept short (less than 2 min) to minimize potential bias by spontaneous dissociation of the CaM-CKII peptide complex.
When using unlabeled CaM or the QSY 9 and Atto540 Q labeled CaM two different controls were performed to ensure the presence of the labeled nonfluorescent CaM and proper focusing onto the glass surface when inducing photounbinding. Details about the procedures and results are described in
For the Phalloidin staining, AAV-HT1080 cells (Stratagene, San Diego, CA, USA, #240109) were fixed in a 4% paraformaldehyde-PBS solution (PFA-PBS) for 15 min at RT, permeabilized with 0.1% Triton X-100 for 3 min, and blocked with 1% BSA-PBS solution for 30 min before incubation with phalloidin-Alexa488 (Ph-A488, Invitrogen, #A12379) for 60 min. After the photounbinding step cells were re-stained with phalloidin-Alexa647 (Ph-A647, Invitrogen, #A22287).
To test label-free unbinding the primary staining was done with unlabeled phalloidin and ph-A488 at a ratio of 4 (unlabeled):1 (labeled). A small amount of labeled phalloidin was necessary to visualize the actin filaments to be illuminated in the photounbinding step.
For GFP staining, PFA fixed B16 actin-GFP cells (kindly provided by the laboratory of Dr. Small, IMBA, Vienna, Austria) were permeabilized with 0.1% Triton X-100 and stained with anti-GFP-biotinylated/Streptavidin APC-Cy7 (BD Biosciences, San Jose, CA, USA, #554063). Cells were blocked in 1% BSA-PBS followed by incubation with goat anti-GFP (2.8 µg/mL) in PBS-BSA for 30 min each, washed (3×) with PBS and finally incubated with Streptavidin APC-Cy7 at the same concentration for 30 min at RT.
For establishing B-16 actin-GFP mouse melanoma and AAV-HT1080 cultures, frozen cryovials were thawed in a 37°C water bath, transferred to 10 mL of DMEM (10% FCS, 2 mM L-Glutamine, Invitrogen), collected by centrifugation at 200×
For the unbinding experiments we used a laser scanning microscope (LSM) (Zeiss LSM 510 confocal) with options for one- and two-photon excitation (1PE or 2PE). To induce photobleaching and/or photo-unbinding the laser (488, 489, 543, 561 or 633 nm for 1PE or a modelocked Titan-sapphire laser line at 800 nm, 200 fs pulses, 80 MHz for 2PE) was focused onto the CaM/CKII peptide coated glass surface or the cell sample through the objective lens (Zeiss, Plan-Apochromat 63×/1.40 Oil DIC M27). Samples were raster scanned at 3.3 sec/line ( = 61µm/s) for 2PE and 62 msec/line ( = 3.25 mm/s) for 1PE, over a square subarea (edge length = 10 or 20µm) - similar to the imaging procedures described in
The experimental procedures were equivalent to those described previously
The fluorescence intensities in the green and red detection channel were obtained from surface plots of the CaM coated surface – CaM-A488 after bleaching and CaM-A647 after reincubation. The amount of rebinding (CaM-A647 fluorescence in the previously illuminated patches) and unbinding/bleaching (loss of CaM-A488 fluorescence) were calculated based on these surface plots. Raw data was analyzed using a custom-written computer code in the R-environment (see
To investigate how the binding of CaM to a set of CKII peptides is affected by photounbinding, we illuminated immobilized CaM/CKII peptide complexes with various laser intensities in a standard LSM and tested the photo-induced unbinding effect upon 1PE and 2PE on either fluorescent or non-fluorescent probes. One iteration of laser scanning was performed to induce photounbinding, unless stated otherwise. To assay photounbinding, we re-applied CaM – but with a different label – and quantified the fluorescence intensity of the newly bound probe.
CKII peptides were attached to a glass surface via an SM(PEG)8 crosslinker followed by CaM-A488 incubation. After light illumination to induce photounbinding of the CKII peptide/CaM-488 complexes, the surface was re-incubated with CaM-A647 to visualize free binding sites in the previously illuminated regions.
Laser illumination for inducing photobleaching and photounbinding of fluorescently tagged (
A: laser power and intensity used to illuminate the corresponding patches in B: ‘bleaching’ pattern (CaM-A488 fluorescence, scale bar: 20 µm), and C: rebinding pattern (CaM-A647 fluorescence).
A: Illuminated patch of CaM-A647 and CKII(290–312) by 633 nm laser light; laser power: 190 µW (flux = 589 nJ/µm2), scale bar = 10 µm. B: rebinding pattern (CaM-A488 fluorescence). C: illumination of unlabeled CaM and CKII(290–312) peptide by 488 nm laser light within the indicated patch (yellow dashed line); laser power: 370 µW (flux = 1.15 µJ/µm2), green dots: fluorescent beads to allow proper focusing. D: no rebinding of A647 was observed after laser illumination within the corresponding patch.
The confocal images in
Furthermore, we found that photounbinding of CaM requires a fluorescent label but is not restricted to a specific label or wavelength
The calmodulin-CKII peptide system allows the study of photounbinding under different dissociation constants without changing the molecular system.
Sample preparations and reactions with different CKII-peptides were performed in parallel under identical conditions (concentrations, incubation time, illumination and imaging settings) for each series of measurements. The CaM/CKII peptide coated surface was immersed in buffer at an initital temperature of 4°C to lower off-rates by 2–4 fold
In
Remaining Fluorescence (A) and corresponding rebinding (B) at various laser powers for peptides CKII(290–312) (grey symbols), CKII(292–312) (green symbols), CKII(293–312) (blue symbols), and CKII(294–312) (red symbols). A: single exponential (solid line) and double exponential (dotted line) fits to the unbinding data. B: single rising-exponential fits to the rebinding data. C: summary of maximal photounbinding values for all tested peptides after 1PE laser illumination (λexc = 488 nm; P = 3.6 mW) and one scan iteration (solid bars) and two scan iteration (open bars) in comparison. D: photounbinding threshold decreases for the lower affinity peptides (Kd: 3–570×10−13 M) the graph shows the relative increase of rebinding when photounbinding laser power is doubled to 7.2 mW. Uncertainties for the rebinding fraction and remaining fluorescence fraction due to variablilty in CKII-CaM coatings and alignment of the coverglasses are less than 15% for each data point, whereas those associated with the laser power are negligible.
We found that photounbinding (after 1PE, λexc = 488 nm) is higher for lower dissociation constants (corresponding to initially tighter binding). In
Additionally, we found that for lower affinity complexes the intensity threshold for photounbinding is shifted to higher light doses when either doubling the scan iterations (
To understand the connection between photounbinding and photobleaching, we had a closer look at the relation between rebinding fraction
Data is shown for peptides CKII(290–312) (grey symbols), CKII(292–312) (green symbols), CKII(293–312) (blue symbols), and CKII(294–312) (red symbols); The plot shows that the rebinding fraction is not directly proportional to the loss of fluorescence, but is suppressed at lower laser powers. The solid black line is a least-square fitted power-law to the CKII(293–312) peptide data (blue symbols) and given by:
The cellular actin network and its interactions with various target proteins is one important topic in cell migration studies and is often addressed by fluorescence approaches
We found that GFP cannot be dissociated from actin (for experimental details see
Following labeling with phalloidin-A488, actin filaments were illuminated with different laser intensities (1PE: 488 nm, 20 µW–370 µW and 2PE: 800 nm, 14 mW–25 mW) and incubated with phalloidin-A647 directly after illumination.
A: Ph-488 fluorescence after illumination at 488 nm (1PE) and 20 µW (62.0 nJ/µm2) (bleached patch is indicated in yellow); B: rebinding of Ph-647 within the previously illuminated area; C: photounbinding in a human fibrosarcoma cell, three squares were bleached (2PE, 800 nm) with different laser intensities left: 14 mW (flux = 10.7 mJ/µm2); top: 20 mW (15.4 mJ/µm2); right: 24 mW (18.4 mJ/µm2).
To further investigate whether a fluorescence label is the critical driving force to induce photounbinding we performed photounbinding experiments where CaM was labeled with a quencher dye, typically used in Fluorescence Resonance Energy Transfer (FRET) experiments as an ideal acceptor. The dyes QSY 9 and Atto540 Q used in this study exhibit a large cross-section at 560 nm and 542 nm, respectively, but very low fluorescence quantum efficiency. If the photounbinding mechanism relies on absorption, we should see CaM rebinding at the previously illuminated square patches. However, we did not observe photounbinding for any of the quencher dyes for any laser intensity applied in this study (
Given the small ‘laser power window’ where photobleaching is observed without any signs of unbinding, we asked whether photounbinding and photobleaching follow independent mechanisms which occur simultaneously, or whether the two phenomena are linked.
It has been described previously that, for the case of 2PE, preventing the bleaching pathway is possible using ascorbic acid as a chemical stabilizer (scavenger)
A: Remaining CaM-A488 fluorescence (open symbols) and the corresponding rebinding (solid symbols) after two-photon excitation (Ti:Sa laser λexc: 800 nm) with the addition of 8 mM ascorbic acid (squares) and without (circles). Photobleaching (and photounbinding) is partly prevented by the stabilizer as expected. B: Control study with A488 fluorophores directly covalently bound to the SM-PEG8 crosslinker via a tripeptide (H-Gly-Gly-Cys-OH). As expected the Alexa 488 fluorescence was stabilized to a comparable extent in presence of ascorbic acid (squares), however no photounbinding was detected. The two data sets have been fitted with a (2 parameter) single exponential function. Uncertainties for the rebinding fraction and remaining fluorescence fraction due to variablilty in CKII-CaM coatings and alignment of the coverglasses are less than 15% for each data point, whereas those associated with the laser power are negligible.
Both samples (
The suggested model is mainly based on three observations:
Photounbinding increases with decreasing dissociation constant
Unbinding (and rebinding) fractions are smaller in the presence of the reducing agent ascorbic acid (
Non-radiative absorption is insufficient to induce photounbinding.
The increase in photounbinding with decreasing dissociation constant (increasing affinity), may be influenced by the unique conformational states that CaM adopts when complexed with these different peptides
Our results clearly show that the unbinding is decreased in the presence of the reducing agent ascorbic acid (
The conformational change of the CaM itself may be assumed to be caused by its interaction with the resultant radicalized molecule X*. A subsequent reaction of a radical dark-state with, for example, free radicals in the solution eventually brings the fluorophore into a stable (bleached) non-fluorescent state.
The observation that the ratio
In
The observations that
A reason why the excited singlet state bleaching may dominate for the photounbinding whereas excited triplet state bleaching does not, may be due to the higher energy provided by bleaching via the singlet excited state as compared to the triplet excited state. This speculation would also explain the larger photounbinding fractions observed for 2PE compared to 1PE, since the 2PE bleaching pathway is known to be significantly different for 2PE with no significant contribution coming from triplet state bleaching
Photounbinding has been shown to occur for various common binding systems such as antibody-antigen
Photounbinding was visualized and quantified by rebinding the same, but differently labeled binding partners in the previously illuminated (photobleached) areas.
However, our results also suggest that photounbinding does not occur for molecules attached through covalent bonds. This is based on the two observations that there was no unbinding after crosslinking the binding partners by formaldehyde fixation (
For non-covalent binding of CKII peptides and A488-CaM we found that laser intensities of <100 µW −which induce only a weak loss of fluorescence− already result in a clearly detectable CaM-A647 rebinding pattern. Obviously, pure photounbinding is hard to distinguish from photobleaching followed by photounbinding of labeled CaM in a typical imaging setup at low or moderate laser intensities. This is particularly relevant in FRAP (Fluorescence Recovery after Photobleaching) or FLIP (Fluorescence Loss in Photobleaching) where experiments photounbinding could be misinterpreted as bleaching and bias the obtained results as discussed recently
Our previous study
In this study we have found that the radical scavenger ascorbic acid prevents not only photobleaching, but also photounbinding under two-photon excitation. Our results suggest that the unbinding is either a direct consequence of photobleaching or at least follows similar pathways with similar thresholds. Whilst a reduction of photobleaching will reduce photounbinding, suggesting that photounbinding is related to a bleaching mechanism, the two are not proportional. The observed trend (increase in photounbinding fraction relative to bleaching fraction with increasing illumination energy) suggests that photounbinding may be governed solely by a sub-dominant bleaching pathway, such as that which occurs through the excited singlet state (S*).
Further experiments and theoretical work on the bleaching pathways of the chosen fluorophores would however be required to confirm this hypothesis.
A further possibility would be if unbinding were the result of a multi-photon process, where the fluorophore is excited into a higher singlet state, and photounbinding is the result of the subsequent decay. However this appears to be contradicted by the observation that doubling the time of illumination increases the photounbinding significantly more than doubling the laser intensity (
As the (CKII) peptides (dependent on their length) exhibit different dissociation constants for CaM, this system is ideally suited for learning more about photounbinding by studying its dependence on Kd. It has been demonstrated that the four different CKII peptides selected (CKII(290, 292-, 293-, and 294–312)) show different rebinding levels to CaM-A647. With
We have not yet fully understood, why the photounbinding rates increase with increasing binding affinity. It may be due to ligand-dependent CaM-oscillations
Future studies to elucidate the photounbinding mechanism would benefit from the use of single molecular fluorescence lifetime measurements in the presence of various reducing solutions to determine the dependence of the unbinding rate on the protein-peptide affinity. Molecular simulations of how these radicals interact with the CaM-peptide structure, and any conformational changes in the CaM they are able to induce, may provide us with further insights.
Figure and text S1: Gel electrophoresis to test for monomeric CaM after quencher dye labeling. Figure and text S2: CaM antibody staining. Figure and text S3: Automated image analysis to quantify photounbinding. Text S4: Calculation of the total incident laser flux (incident energy per unit area). Figure and text S5: No bias by fluorescence quenching. Text S6: control: unspecific binding of CaM. Figure and text S7: Probing photounbinding by rebinding of identically labeled A488-CaM. Figure and table S8: Mathematical correction for low affinity peptides. Figure S9: Log-log plot of the unbinding and rebinding fraction. Figure and text S10: Analysis of photounbinding in fixed cells. Figure S11: No photounbinding for quencher dye labeled CaM. Figure S12: Non-radiative energy transfer to CKII-peptide (Jablonski energy diagrams). Text S13: Fitting statistics for plot. Figure S14: Photounbinding experiments before and after PFA fixation.
(1.22 MB PDF)
We thank Dr. Wiseman and Dr. Gyözö Garab for discussions about potential photounbinding mechanisms, Mathias Madalinski for assistance in synthesis of the CKII peptides, Dr. Vic Small for providing B16 actin-GFP cells, Dr. Peter Steinlein for conceptual advice, Andreas Sommer for providing the R-script, the BioOptics facility for assistance in imaging and data analysis, and Dr. Brooke Morriswood for proof-reading the manuscript.