FRET (Fluorescence Resonance Energy Transfer)
- FRET is a spectroscopic technique that enables us to determine distances at the nanoscale using chemical compounds that absorb and emit light
- Sealed closed boxes exhibit FRET - as such, we know we can use FRET as an analysis technique
- Unsealed boxes do not exhibit FRET (regardless of hinge torque or opening bracing angle)
- The boxes were redesigned to include latches. Latched boxes give a FRET signal and can therefore close the box
- Boxes with a Ni-NTA linker exhibit FRET with his-GFP in a non-sealed box, but not in sealed boxes. This indicates that a protein can be localised in the box
- Our braces work. This was elucidated through a difference in the intensity of the the emission spectrum at the acceptor’s peak emission wavelength between latched boxes with no brace and latched boxes with braces
- Our EcoRI-Spytcatcher trigger was shown to be functional (i.e. its ability to bind to the box and cut its braces) through a difference in the intensity of the the emission spectrum at the acceptor’s peak emission wavelength
Don’t FRET, we’ll fill you up on all the raw data and analysis here
Imagine a strand of human hair and think about how thin it is. Now try to picture something that’s a thousandth of that size. A hard ask, right? That’s how tiny our boxes are – they’re smaller than the wavelength of visible light and that’s why we can’t see them.
While we can’t use visible light to see our tiny traps, we can use it to observe their behavior using a phenomenon known as fluorescence/Förster resonance energy transfer, or FRET. When excited by light within their excitation spectrum, chemical compounds called fluorophores remit light at wavelengths within their emission spectrum. If they are tagged onto structures or other molecules (e.g. DNA), fluorophores can be used as dyes for microscopy coupled with spectroscopic techniques – they can indicate the presence of a molecule or particle and trace their path.
However, if two fluorophores have emission and excitation spectra that overlap and are in close proximity, excitation of a ‘donor’ fluorophore can result in (radiationless) energy transfer to an ‘acceptor’ fluorophore which remits within its emission spectrum. The distance over which FRET occurs (1-10nm), its incredible sensitivity to changes in the radius of interaction and the simplicity of the process make it an attractive technique for verifying our DNA design.
The Overall Aim of FRET
By attaching a donor dye to one lid and an acceptor dye to another, FRET can help us confirm whether our brace and hinge design are working. Furthermore, if our boxes behave as per our design, FRET can be used to determine if our trigger can be localized within our box and cut the braces.
In order to conduct our FRET experiments, we opted to attach one fluorophore to one lid and its FRET pair to the other. This would require modifying an existing staple oligonucleotide in each lid with a fluorescent dye.
When a sample is exposed to a wavelength of light that results in donor-only excitation and the two fluorophores are far apart, the intensity of detected light within the acceptor’s emission spectrum should be relatively low, perhaps even negligible. However, if the experiment is repeated and the dyes are within close proximity of each other, FRET will occur. This should be apparent from a relative increase in acceptor emission intensity and, if the concentrations of the samples are standardized, a relative decrease in donor emission intensity. If a fluorophore is attached to each of the lids, FRET will indicate the extent to which the box is open or closed (Figure 1).
Figure 1: (Left) A braced (open) box and the emission spectrum when no FRET is occurring and (Right) a closed or sealed box and its emission spectrum when FRET occurs
There were several key considerations that had to be taken into account when choosing a FRET pair:
Good spectral overlap (integral of shared area between emission spectrum of the donor and excitation spectrum of the acceptor on an absorbance-wavelength chart) whilst little to no overlap of each of the fluorophore’s respective excitation spectra (Figure 2)
The FRET pair must have an appropriate Förster radius (distance at which FRET efficiency is 50% - indicative of the radius of interaction)
Quantum yield (ratio of emitted photons to absorbed photons) of the fluorophores must be sufficient to see a signal
Must have a good lifetime (time taken to return to ground state from excitation) and be photostable so that the same fluorophores can be used in fluorescence lifetime experiment (FLIM-FRET) or single molecule FRET
Figure 2: Excitation (dashed) and emission (solid) spectra for an ideal FRET pair
ATTO550 and ATTO647n (NHS-ester modified and attached to a DNA staple oligonucleotide) were chosen as they meet these criteria and have been used with DNA origami and buffer similar to ours (. Their spectra are shown in Figures 3 and 4.
(Left) Figure 3: ATTO550 Spectra, (Right) Figure 4: ATTO647n Spectra
The donor (ATTO550) is located on the lid lip helices of the bottom lid (helices 46-48, Figure 5). The staple tagged with the donor fluorophore is 5’ modified.
Figure 4: Staple with ATTO550 highlighted (caDNAno view)
The acceptor (ATTO647n) is located on the lid lip helices of the top lid (helices 24-25, Figures 6 and 7). The staple tagged with the donor fluorophore is 5’ modified. As the ideal staple location was occupied by a sealing (or non-sealing) staple, two 5’modified oligonucleotides were required.
Figure 5: Non-sealing box staple with ATTO647n highlighted (caDNAno view)
Figure 6: Sealing box staple with ATTO647n highlighted (caDNAno view)
Experiment 1: Origami Formation and Dye Colocalisation (Gel Electrophoresis)
We needed to confirm that the dyes (ATTO550 and ATTO647n) were bound to a fully formed box after a slow-ramp PCR synthesis protocol and PEG purification (see ‘Origami Synthesis Protocol’ under ‘DNA Design’). Colocalsation is crucial to conducting FRET experiments as the absence of one or both fluorophores from the box will mean that there is no clear indicator of a closed (or semi-closed) configuration.
The first agarose gel (2% w/w agarose, 0.5x TBE Buffer, addition of 4 microL of RedSafe DNA Stain) was run (2 hours, voltage of 70V) in order to confirm fluorescent box formation against controls which were boxes made using ‘normal’ non-fluorescent staple mixes (Figure 8). The gel was then imaged using a Cambridge Uvitec Machine with UV excitation and no filters applied.
A second gel (1.5% w/w agarose, 0.5x TBE buffer, no RedSafe DNA Stain) was run (2 hours, voltage of 70V) to confirm that the fluorophores were colocalised, i.e. both fluorophores were attached to the box. RedSafe was omitted as its excitation and emission spectra coincide with ATTO550, thereby making it difficult to isolate the donor fluorophore with an EPI GREEN excitation wavelength (520nm). The gel was imaged using a Cambridge Uvitec Machine with an EPI RED excitation wavelength of 650nm and a 695nm filter (Figure 9) and an EPI GREEN excitation wavelength of 520nm and a 595nm filter (Figure 10).
Results and Discussion
From gel 1 (Figure 8), it is clear that the fluorescent origami structures appear to have the same or similar bands to the controls (non-fluorescent boxes). Since these band structures match those observed in previous gels, we can conclude that the fluorescent boxes have formed as expected (a defined origami band).
Figure 9 is an image of gel 2 exposed to an excitation wavelength of 650nm with a 695nm cut-off filter, whilst Figure 10 is an image of the same gel exposed to an excitation wavelength of 520nm with a cut-off filter of 595nm. Despite the low emission intensity of Figure 10 (possibly due to ATTO550’s low absorbance of light at 520nm and the cut-off filter), it is apparent that both fluorophores are present in the origami bands. As such, it can be concluded that the fluorophores have been colocalised on the boxes. We can now conduct FRET experiments in order to functionally characterize our boxes.
Figure 7: Gel 1 - Impure and purified boxes run against a 1kb ladder and controls (Modified using Microsoft Word - Brightness 0+ and Contrast 40%+)
Figure 8: Gel 2 - Impure and purified boxes run against a 1kb ladder and controls. Imaged using a Cambridge Uvitec machine with an excitation wavelength of 650nm and a filter cutoff of 695nm (Modified using Microsoft Word - Brightness 0+ and Contrast 40%+)
Figure 9: Gel 2 - Impure and purified boxes run against a 1kb ladder and controls. Imaged using a Cambridge Uvitec machine with an excitation wavelength of 650nm and a filter cutoff of 695nm (Modified using Microsoft Word - Brightness 0+ and Contrast 40%+)
We have confirmed colocalisation of the fluorophores on our PEG purified origami boxes. As such, we can now carry out bulk FRET experiments on our tiny traps.
FRET Experiments: Materials and Method
An initial attempt at fluorescence lifetime analysis (FLIM-FRET) produced a donor-only lifetime estimate of 0.8ps. As this lifetime approached the limit of the resolution of the instrumentation (HORIBA DeltaFlex), this method was not pursued (any change in fluorophore lifetime due to FRET would be difficult to detect). Therefore, all FRET experiments were conducted as bulk FRET experiments in which a sample is excited and an emission spectrum is observed.
100 microlitres of fluorescently-labelled PEG purified origami structures (see DNA design for synthesis and purification protocols) was pipetted into a cuvette and placed into a Horiba FluoroMax 4. The nominal and actual origami concentrations (measured using a NanoDrop 1000) were recorded for each sample – no normalization was carried out as the different box designs were first tested in order to establish the limits or presence of a binary (FRET or not FRET) outcome.
The excitation wavelength was chosen such that most, if not all, of the donor and/or acceptor emission spectrum was observable. The emission spectrum scan was set to commence ~25-30nm away (in the direction of red shift) to avoid including Raman scattering peaks. Two emission spectra (with emission wavelength plotted against counts per second) was obtained for each sample – the first with donor excitation (and FRET/no FRET) and the second with acceptor excitation (to confirm the presence of the acceptor fluorophore in the sample).
Experiment 2: Protocol Optimisation and Brace Pos. 1 vs Brace Pos. 2
Optimise FRET protocol
Determine if FRET is observable in the sealed box and any braced (but untorqued) boxes
Test the performance of Brace Position 1 (braces further from hinge) vs Brace Position 2 (closer to hinge)
A sealed box, a control box (no closing hinge torque, no opening braces), and boxes with no closing hinge torque but all opening brace angles in position 1 and 2 were synthesized for this FRET experiment.
Results and Discussion
We first compared the FRET signal from boxes with and without sealing staples, where neither box had any closing hinge torque nor braces. The emission spectrum of the box with no sealing staples, no closing hinge torque and no braces (Figure 12) resembled that of a staple mix that only contained the donor fluorophore (Figure 10), whereas a clear FRET signal was seen in samples with sealing staples (Figure 11). The emission spectra of the boxes with no sealing staples, no closing hinge torque but various opening bracing angles (in brace positions 1 and 2) also resembled that of the donor fluorophore (i.e. no FRET signal), This data indicates that in the absence of a closing hinge torque, boxes tend to exist in a sufficiently open conformation such that the two halves are separated by a distance greater than the Förster radius of ATTO550 and ATTO647n (6.5nm) at the side opposite the hinge.
Figure 10: Emission spectrum of donor only staple mix
Figure 11: Emission spectrum of sealed box with FRET visible
FRET was observed for the sealed box, which suggests that a FRET binary test criterion can be established. However, as none of the braced boxes demonstrated any FRET (despite the acceptor fluorophore being present in the sample), no conclusions can be made about the efficacy of the brace or the difference in camming action between brace position 1 and brace position 2. In order to show that the braces work as per our design, we must be able to observe some level of FRET in an unbraced box that is not sealed (i.e. a box with closing hinge torque).
Figure 12: Emission spectrum of control box (L) and verification of the acceptor fluorophore (R)
Figure 13: Emission spectrum of box with opening bracing angle of 90 degrees (L) and verification of the acceptor fluorophore (R)
Whilst FRET can be achieved using our fluorophore/experimental design, we cannot see any FRET in the unsealed boxes. As no FRET was observed in the control box (no opening brace, no closing hinge torque), an absence of FRET in the braced boxes could suggest that the braces work, or that the braces do not work as per our design and that the lids move freely beyond the range of FRET between our two fluorophores (as in the control box). As such, no conclusion can be made about the efficacy of the braces in either brace position 1 or brace position 2.
Experiment 3: First Attempt at Functional Characterisation
Determine if FRET can be achieved by boxes with closing hinge torque (i.e. replicate the sealed box result - indicate boxes are either shut or close to shut)
Determine if a FRET binary (or different degrees of FRET) is observable for different closing hinge torques for a constant opening brace angle (characterization)
Test the performance of our SpyCatcher-EcoRI trigger. Can it cut the brace? Seeing this result will depend on the success of aims 1 and 2
All possible permutations (varying opening brace angle, varying closing torque hinge) were synthesized for this experiment. A series of opening braces angles of 90 degrees in brace position 2 (with varying closing hinge torque) were incubated with EcoRI (37 degrees Celsius incubation for one hour).
Results and Discussion
The sealed box (Figure 14) showed no signs of FRET. This could be due to the wrong staple mix being used in synthesis – the magnitude of intensity in the acceptor verification emission spectrum is very low. From previous experiments, it is clear that the sealed box should FRET. The acceptor is unlikely to have bleached as all other samples have a much larger intensity for acceptor.
Figure 14: Emission spectrum of sealed box (L) and verification of the acceptor fluorophore (R)
All boxes with no opening brace but various closing hinge torques (much like Figure 15, a box with a closing hinge torque of 105 degrees and no brace) showed no signs of a FRET signal. A similar no- FRET result was obtained for all other combinations of closing hinge torque and opening bracing angle. It could be argued that some FRET can be observed in the no brace, closing hinge torque of 70 degrees sample (Figure 16) - this, however, is likely a result of low sample concentration and background fluorescence becoming more significant at low signal magnitudes rather than FRET. Furthermore, 105 degree closing hinge torque shows no sign of FRET, so it is unlikely that a box with 70 degrees closing hinge torque would do so.
Figure 15: Emission spectrum of a box with a closing hinge torque of 135 degrees (L) and verification of the acceptor fluorophore (R)
Figure 16: Emission spectrum of a box with a closing hinge torque of 70 degrees (L) and verification of the acceptor fluorophore (R)
From this experiment, is clear that the current experimental design is inadequate to test the functional capabilities of our boxes, which are deviating from their expected behaviour. As no unbraced boxes demonstrate any FRET, we cannot conclude whether the absence of FRET in the braced boxes is a result of the braces meeting their design specifications (and thereby separating the two halves/fluorophores) or if the braces do not hybridise and the lids are free moving (well beyond the Förster radius of the FRET pair). Furthermore, as boxes with significant closing hinge torque and no brace do not produce a FRET emission spectrum similar to the sealed box, it is apparent that boxes with hinge torque are not behaving as expected. While this does not suggest that the hinges do not have any effect in shifting the mean equilibrium position of the two halves, the inability to establish a FRET binary with sealed and unsealed boxes does affect our capacity to functionally characterise our boxes in solution.
It is possible that the Förster radius of our FRET pair, and that of any FRET pair (2-10nm), may be inadequate to test our design and characterise the boxes given that closing hinge torque does not allow us force FRET. Using the brace design Python script (based on triangular geometry), positioning the braces at the lip of the box lids gives a distance of 48 base pairs (16.32nm) for a 45 degree brace and 89 base pairs (30nm). Since these distances are greater than some of the largest Förster radius’, a new experimental design will be required to test the functionality of the braces and their response to the introduction of EcoRI.
The experimental design has been shown to be inadequate and, as such, no solid conclusions could be drawn from Experiment 3. An aspect of the box or the fluorophore design should be reconsidered before considering any further bulk FRET experiments.
Box Redesign – Latches and Conjugation
The results from Experiment 3 suggested that the closing hinge torque in our designs isn’t sufficient to bring the two halves of the box together, as compared with sealing the box with staple strands extending from one half to another. However, it begs to be asked whether there is some way we can test the action of the braces without radically changing key elements of our original design.
Two modifications were proposed and designed. The first was a set of latches, which consist of eight base pair staple extensions in in the top half that are complementary to similar extensions in the bottom half of the box (Figure 17). Figure 17: The latch extensions and a box in a latched configuration
These latches should seal the box in the absence of a brace, or, if they are added after the braces in the synthesis phase, should seal the box if a set of braces are non-functional or cut by our trigger. However, if the braces assemble as we expect them to, the relatively short length of the extensions (8bp) should mean that the two halves of the box will not be forced together.
The second modification was a set of hinge extensions that were designed to extend into or close to the cavity of the box – one complementary to a nickel-nitrilotriacetic acid (Ni-NTA) modified strand and another complementary to a red acceptor fluorophore (ATTO565N) modified strand. Any protein with a histidine, or histag (including his-Green Fluorescent Protein or our EcoRI-Spycatcher trigger) can be bound to the Ni-NTA (Figure 18). Figure 18: A latched box with both of the hinge modifications
GFP and ATTO565n are a potential FRET pair, so these hinge modifications can be used for colocalisation experiments (Figure 19). Figure 19: Spectra for GFP and Alexa Fluor 568, which has similar spectra to ATTO565n (unavailable on ThermoFisher SpectraViewer)
For a more detailed explanation of these modifications, refer to DNA Trap Design. Two experiments were designed that utilized these new modifications. Experiment 4 involved confirming that a protein (or enzyme such as our SpyCatcher-EcoRI trigger) could be localized inside the box. The purpose of Experiment 5 was to attempt to test the functionality of our EcoRI-Spycatcher trigger (i.e. whether it can cut the braces and whether it can do so whilst localized within the box).
Experiment 4: Protein Localisation
- Determine if a his-tagged protein, such as his-GFP, can be conjugated to the box via a Ni-NTA bridge. If this is possible, we should observe FRET between GFP and ATTO565n (hinge modification) in a braced open box sample and no FRET in a sealed box sample.
Both samples had no closing hinge torque and both of the hinge modifications (Ni-NTA bridge and ATTO565n modified staple). However, one sample (2A) was braced open (opening bracing angle of 90 degrees) whilst the other (2B) was sealed. The boxes were synthesized as per the regular protocol (without the fluorescent staples on each half). ATTO 565n dye was added well in excess and the sample was allowed to incubate at room temperature for a day. Following this, an excess of NTA solution was added and the sample was incubated for an hour. The same procedure was carried out for Nickel Sulfate. Finally, once the boxes had been PEG purified, a tenfold excess of GFP was added and allowed to incubate for 3 hours.
Results and Discussion
The braced open sample with the hinge modifications (2A) clearly demonstrated FRET, indicating that GFP and the red acceptor (ATTO565n) were colocalised within the boxes (Figure 20). Figure 20: Sample 2A - 90 degree opening bracing angle with hinge modifications
In contrast, excitation of the sealed sample with hinge modifications (2B, Figure 21) did not produce as large a FRET signal as observed in sample 2A. Figure 21: Sample 2B - 90 degree opening bracing angle with hinge modifications
A distinct difference in the peak emission wavelength of the acceptor fluorophore between samples 2A and 2B indicates that the GFP is quite close to the red acceptor in the braced open sample and not so in the sealed sample. The most likely explanation for this phenomenon is that the his-GFP has been conjugated with the box and is sufficiently close to the red acceptor for FRET to occur.
his-GFP was successfully localized within the box, as seen in the emission spectra of the closed and braced open samples. This allows us to perform experiments with our box and our his-tagged EcoRI-Spycatcher trigger (Experiment 5).
Experiment 5: Second Attempt at Functional Characterisation with Latches
- To test the functionality of the EcoRI-Spycatcher trigger and of the braces. In order to do this, we must show that the trigger is more likely to cut a functional set of braces when it can be bound to the box – this corresponds to a larger FRET signal when EcoRI is incubated with boxes with latches and a NiNTA link as opposed to boxes with latches and no NiNTA link.
The following samples were synthesized using a fluorescent staple mix (ATTO550 and ATTO647n):
- Opening bracing angle of 90 degrees, latched box (0A)
- No brace, latched box (0B)
- Opening bracing angle of 90 degrees, latched box (1A) box incubated with EcoRI (37 degrees Celsius, 40 minutes)
- Opening bracing angle of 90 degrees, latched box (1B)
- Opening bracing angle of 90 degrees, NiNTA, latched box incubated with EcoRI (37 degrees Celsius, 40 minutes) (3A)
- Opening bracing angle of 90 degrees, no NiNTA, latched box box incubated with EcoRI (37 degrees Celsius, 40 minutes) (3B)
The synthesis method and incubation period for the NiNTA link is as described in previous experiments.
Results and Discussion
The samples can be simplified as such:
- 0A: box + brace + latches
- 0B: box - brace + latches
- 1A: box + brace + latches + trigger
- 1B: box + brace + latches
- 3A: box + brace + latches + NiNTA linker + trigger
- 3B: box + brace + latches + trigger
A FRET signal was observed in all samples, however, there was a subtle but noticeable shift in the magnitude of the acceptor peak between the samples. In order to compare the curves of different samples at different concentrations (and hence different fluorescence intensities), each curve was scaled to an arbitrary reference curve by varying the scaling factor to minimize total RMSD (using Excel Solver). The result of this scaling can be seen in Figure 22.
Figure 22: Scaled sample emission spectra
From the ‘low resolution’ scaled curves, it is evident that there is some deviation in the magnitude of the acceptor peak between the various sample emission spectra. By changing the scale of x-axis and y-axis, we can ‘zoom in’ on this region (Figure 23). We can also look at the magnitude of the curve at the acceptor emission peak for each sample in order to simplify the trend a little further (Figure 24). Figure 23: Scaled sample emission spectra, zoomed into the range near the acceptor emission peak wavelength
Figure 24: Intensity of ATTO647N acceptor at peak emission
The progression from high to low acceptor peak magnitude is logical if we conclude that the trigger is functional as per its design. A larger FRET signal corresponds to a greater proportion of latched boxes in solution, whereas a lower FRET signal indicates that a greater proportion of boxes with braces are present in solution. It can be seen that the FRET response is bounded by a maximum (0B – a box with no brace and latches) and a minimum (Sample D from Experiment 3, box with no brace and a closing hinge torque of 70 degrees) acceptor peak magnitude between ~640-700nm. These correspond to a configuration where the box is latched with no resistance (highest proportion latched in solution) and another where there is an absence of an acceptor peak and no observable FRET, respectively.
The interesting result is that sample 3A (latched boxes incubated with NiNTA and with EcoRI) had a higher magnitude acceptor emission peak than 3B (latched boxes without NiNTA and with EcoRI). This suggests that the triggers were able to cut functional braces in a larger proportion of boxes (higher FRET signal) as they were able to be localised via the NiNTA link.
One would expect that similar logic could be applied to sample 1B, which has boxes with latches and braces but which were not incubated with the trigger. As this sample should have a greater proportion of boxes that are braced open, we should see a decrease in FRET signal. However, the opposite is true. This is likely a result due to error in the scaling method and the noise of the data in combination with a small intensity difference. As all the other samples seem to fit the trend, it can be concluded that the trigger is functional and that the braces are also functional. This experiment should also be repeated (with an opening bracing angle of 90 degrees and with other opening bracing angles) in order to confirm if this is a systematic trend.
It was successfully shown that the trigger and braces are functional. This means that the trigger is able to be bound to the box and cut braces that hold the two halves of the box apart. This conclusion can be made thanks to the results of Experiment 4 and the increase in FRET signal observed in the sample with boxes with a NiNTA link (compared to a sample of boxes without the NiNTA link), indicating that a greater proportion of boxes with braces had been cut and then latched.
ATTO-TEC, (2016). Product Information: ATTO 550. 1st ed. Siegen: ATTO-TEC.
ATTO-TEC, (2016). Product Information: ATTO 647N. 1st ed. Siegen: ATTO-TEC.
Behlke, M., Huang, L., Bogh, L., Rose, S. and Devor, E. (2011). Fluorescence and Fluorescence Applications. 3rd ed. [ebook] Integrated DNA Technologies. Available at: https://www.idtdna.com/pages/docs/technical-reports/fluorescence-and-fluorescence-applications.pdf?sfvrsn=5 [Accessed 21 Oct. 2016].
Fábián, Á., Rente, T., Szöllősi, J., Mátyus, L. and Jenei, A. (2010). Strength in Numbers: Effects of Acceptor Abundance on FRET Efficiency. ChemPhysChem, 11(17), pp.3713-3721.
Loura, L. (2012). Simple Estimation of Förster Resonance Energy Transfer (FRET) Orientation Factor Distribution in Membranes. IJMS, 13(12), pp.15252-15270.
Wahl, M. (2014). Time-Correlated Single Photon Counting. 1st ed. [ebook] Berlin, Germany: PicoQuant. Available at: https://www.picoquant.com/images/uploads/page/files/7253/technote_tcspc.pdf [Accessed 21 Oct. 2016].