Rasmus K.
Jakobsen‡
,
Stine G.
Stenspil‡
,
Junsheng
Chen
and
Bo W.
Laursen
*
Nano-Science Center & Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark. E-mail: bwl@chem.ku.dk
First published on 18th March 2025
Fluorescent off–on probes based on a modular design where an analyte sensitive PET moiety is attached to a fluorophore are extremely successful. Here we report a new modular fluorescence probe design switched by dynamic quenching due to proton coupled electron transfer (PCET) mediated by collision with weak bases in solution. The fluorescence lifetime of this probe directly reports on the rate of deprotonation by the weak bases in the solution. We investigate the probe design, mechanism of response, and sensitivity to various abundant weak bases/metabolites including acetate, glutamate, phosphate, valine, and amines. We find that this modular PCET based probe design, in contrast to traditional PET probes, can work efficiently with a fluorescence lifetime readout providing a calibration free probe for weak bases. Upon further development we envision such dynamic PCET probes as sensitive tools for studies of cellular buffer systems and metabolite pools.
![]() | ||
Fig. 1 Photoinduced electron transfer probes and mechanism. (A) 1 (LysoSensor Blue), a pH sensitive probe in which protonation of the two morpholine side groups inhibits PET quenching of the anthracene unit. Adapted from de Silva et al.52 (Rhod-2), an example of a calcium sensitive probe. Adapted from Minta et al.15 (B) Schematic representation of a molecular PET probe displaying fluorescence when bound to an analyte or undergoing PET quenching when no analyte is bound. |
The readout for PET probes is generally based on the fluorescence emitted from the fraction of probes that are bound to analytes. These probes will display similar photophysical properties to the chromophore scaffold without a quenching moiety (Fig. 1B). Probes not bound to an analyte do not contribute to the fluorescence signal, since PET quenching is much faster than the fluorescence process. Consequently, the readout of the probe only varies in fluorescence intensity and not in fluorescence lifetime, thus corresponding to static quenching.20 The fluorescence intensity will be proportional to the ground-state equilibrium between probes bound and not bound to an analyte. While such turn-on of fluorescence upon analyte binding is excellent for qualitative or relative measurements, this simple one-channel intensity readout is also a drawback and makes quantitative measurements dependent on the use of internal standards and/or comprehensive calibration.20,21 PET quenching that instead takes place upon collision of the freely diffusing analyte and the excited fluorophore results in a shortening of the observed fluorescence lifetime and a proportional loss of fluorescence intensity (dynamic quenching).20 Since the fluorescence lifetime readout fundamentally is independent of the concentration of the probe it provides a simple and absolute measure of analyte concentration.20,22 The advantageous effect of dynamic quenching and fluorescence lifetime readout is successfully demonstrated by the quantitative imaging of Cl− ions in biological systems.23–26 However, this is a special case, only feasible due to the high abundance and relatively low oxidation potential of Cl−, allowing this analyte to act directly as the PET quencher without binding to a PET active receptor unit.
Here we demonstrate how a modular PET probe design with a fluorophore and a covalently linked PET quencher can be utilized in the dynamic quenching regime, and via the fluorescence lifetime readout, report on biologically abundant analytes such as glutamate and phosphates. The sensing mechanism relies on the analytes' ability to be proton acceptors in proton coupled electron transfer (PCET) quenching of the excited fluorophore.27,28
For the PCET probe quenching occurs by a diffusion controlled deprotonation of the phenol group by a basic analyte (B:) and transfer of an electron from the phenol/phenolate to the excited DAOTA leading to the formation of the dark PET state. This PCET quenching can either take place stepwise by proton transfer (PT) followed by electron transfer (ET) or by a concerted proton electron transfer (CPET) process (highlighted as the blue colored processes in Fig. 2). Regardless of the mechanism, the quenching rate (kq) will depend on the concentration of the base assisting the PCET process; kq = kPCET[B:]. This dynamic quenching process leads to a reduction of the observed fluorescence lifetime of the probe, proportional to the quenching rate constant (kq) and thus the concentration of the analyte/base. This opens up the possibility to determine analyte abundance from one simple fluorescence lifetime measurement independent of probe concentration.
This follows the classical PET probe model shown in Fig. 1 and the PCET model in Fig. 2 when [B:] = 0 (and hence kq = 0). In buffered solutions the fluorescence intensity however displays an additional decrease at pH ≈ pKa of the buffer. For a 100 mM acetate buffer this results in a more complex fluorescence intensity response shown as the red points in Fig. 3B, significantly deviating from the response expected from the classical PET model and that observed in pure water (blue points). Investigation of the fluorescence lifetime (τ) as a function of pH shows similar buffer dependence, as shown in Fig. 3C.
In the acetate buffer solution τ = 14.0 ns at low pH, but decreases to τ = 12.0 ns at higher pH, the decays measured at all pH values were monoexponential (ESI §12†). According to the proposed PCET probe model (Fig. 2) this decrease in τ arises from collision of the excited probe with a base (B:), which is able to assist in the deprotonation of the phenol and thus the overall PCET process, leading to fluorescence quenching due to the formation of the PET state of the phenolate. The change in τ is proportional to the increased concentration of acetate ions (pKa = 4.76), which in this case is the base assisting the PCET process. According to the PCET model, the observed 14.5% drop in τ should be accompanied by a similar drop in fluorescence intensity. This is indeed confirmed when comparing the measured fluorescence intensity (red points in Fig. 3B) with a theoretical Henderson–Hasselbalch curve for the system based on a 14.0% quenching from a base with pKa = 4.76 and a full quenching from the ground-state deprotonation of the probe with pKa = 8.90, as shown by the red line (see ESI §4† for details). In agreement with the PCET model τ is unchanged in pure water (Fig. 3C, blue points), except for a small decrease in the most basic solutions corresponding to dynamic quenching from hydroxide ions (Fig. S2†). To exclude the possibility that any of the changes in fluorescence properties are due to direct interactions with the DAOTA fluorophore we also studied anisole-DAOTA, where the phenol group has been methylated and thus cannot take part in proton transfer processes. As expected, anisole-DAOTA showed no changes in fluorescence intensity or lifetime (Fig. 3C, S3 and Table S2†) as a function of buffer or pH. Importantly, anisole-DAOTA displays the same fluorescence lifetime as phenol-DAOTA in pure water (τ0), proving that water does not act as a base (proton acceptor) in the PCET process. According to our proposed model (Fig. 2) the observed τ, in the presence of bases, can thus easily be converted to a quenching rate constant (kq), which directly reflects the rate of base assisted PCET experienced by the excited probe (kPCET[B:]), as described by eqn (1).
![]() | (1) |
An important feature of both the dynamic PCET and the traditional PET sensing modalities is the very fast formation of the non-emissive PET state from the excited phenolate form, since this makes the base assisted excited state proton transfer irreversible and explains why no emission is observed from the excited phenolate. The formation of the PET state from the deprotonated probe was therefore investigated by femtosecond transient absorption spectroscopy (details in ESI §3†). Femtosecond transient absorption spectra of the phenolate-DAOTA showed two characteristic signals.
The first is an excited state absorption (ESA) at 400 nm (black data points in Fig. 4). The ESA signal has a rise time corresponding to the instrument response function (IRF) (≈100 fs) and an additional component of 0.4 ps (highlighted in pink) and decay with a 1 ps time constant. The second feature observed is a ground state bleach (GSB) of the first DAOTA absorption band at 580 nm with a rise time identical to the instrument response function (IRF) (≈100 fs), and a decay of 1.2 ps (red data). From this we deduce that the excited state of the phenolate form only has a lifetime of 0.4 ps before converting to the PET state, which itself has a short lifetime of 1 ps before returning non-radiatively (by back electron transfer) to the ground state. Transient absorption spectra of the protonated form support this interpretation with ESA and GSB signals rising with the 100 fs response time of the instrument and not decaying on the ps time scales (in agreement with the 14 ns lifetime measured from the fluorescence decay). Femtosecond transient absorption spectra of phenolate- and phenol-DAOTA are shown in Fig. S4 and S5.† Very fast formation of the PET state from the excited phenolate-DAOTA is observed in the femtosecond transient absorption measurements and confirms that base assisted proton transfer from phenol-DAOTA in the excited state will be irreversible, thus supporting the PCET model (Fig. 2).
As the operation of the PCET probe relies on a diffusion controlled interaction with basic molecules, the most obvious analytes in biological systems are the abundant metabolites making up cellular buffer systems. Cellular systems contain complex mixtures of hundreds of small molecule metabolites, reaching a total concentration of approximately 300 mM.30
This metabolite pool is dominated by weak bases such as glutamate and organic phosphates reaching concentrations of ∼100 mM and ∼50 mM respectively. Also, inorganic phosphate is abundant inside cells reaching concentrations of ∼50 mM. With this in mind, we investigated the response of the phenol-DAOTA probe in a series of bases covering a range of pKa values. The probe response was investigated in solutions of acetate, formate, valine, glutamate, phosphate, glucose-6-phosphate, and ethanolamine where the concentrations of the individual bases were varied by adjusting the pH.
For a 100 mM glutamate solution the fluorescence lifetime response as a function of pH is shown in Fig. 5. At low pH τ = 14 ns, identical to the lifetime for pure water solutions, showing that the fully protonated glutamic acid (Glu+) is not facilitating the PCET quenching. At pH = 3 where the solution contains ≈100 mM Glu0, the measured fluorescence lifetime has only dropped ≈0.5% making it clear that Glu0 has very little impact on the probe and that this is on the limit of detection. However, clear drops in fluorescence lifetime are observed at pH values equal to the second and third pKa values proving that both Glu− and Glu2− indeed promote the PCET quenching, with Glu2− being more efficient than Glu−. As seen for acetate the fractional drop in τ for glutamate is in all cases followed by a corresponding drop in fluorescence intensity in full agreement with the PCET model (Fig. S8†). For all the investigated bases, we found perfect correspondence between the changes in the fluorescence lifetime of the probe and their reported pKa values. Quantitative correlation between the change in fluorescence intensity and the observed fractional drop in fluorescence lifetime was also observed (Fig. S6 to S14†). Stronger bases most efficiently promote the PCET quenching. This is clearly seen when the fluorescence lifetime as a function of concentration of each base is displayed in a Stern–Volmer plot, as shown for selected bases in Fig. 6 (plots for all bases are shown in Fig. S15–S23†). The concentrations of each base are calculated based on the pKa of their corresponding acid, pH in the solution, and the total buffer concentration. The perfect straight lines of these plots confirm that each base acts as a dynamic quencher according to the classical Stern–Volmer equation (reorganized from eqn (1)):
![]() | (2) |
For hydrogenphosphate the Stern–Volmer plot shows a significant upward curvature at concentrations above 40 mM (shown in Fig. S20†), indicating a departure from an ideal solution caused by electrostatic attraction between the cationic probe and dianionic base (HPO42−), leading to a local increase of the base concentration around the probe. This effect arising for oppositely charged fluorophores and quenchers is predicted by the Debye–Hückel theory and has been reported previously.31,32 In agreement with this explanation, increased ionic strength suppresses this effect and reinstates linear Stern–Volmer plots. We demonstrate this by addition of NaCl (150 mM) as shown in Fig. S24.†
To demonstrate the potential of the phenol-DAOTA probe in more complex buffer/metabolite systems, lifetime measurements in a mixed system of acetate and phosphate showed very good correlation with expected lifetimes based on the kPCET rates established for the individual bases (see ESI §10†).
Measurements of the PCET quenching of phenol-DAOTA in a 100 mM glutamate dissolved in D2O display PCET rates 2–3 times lower than in H2O (ESI §9†). This kinetic isotope effect confirms that proton transfer is indeed involved in the rate-determining step of the quenching as expected from our model for PCET in this system (Fig. 2).27
As expected, the rate of PCET (kPCET) derived from the slopes of the Stern–Volmer plots increases with the base strength of the proton acceptors as shown in Fig. 7.27,33,34 Here log(kPCET) is plotted as a function of ΔpKa, which represents the difference between the pKa of the conjugated acid of the proton acceptor (BH) and the proton donor (phenol-DAOTA). The plot reveals several interesting features of the base promoted quenching process. Firstly, the rate of PCET reaches a plateau at kPCET ≈ 108.5 M−1 s−1 for the stronger bases (ΔpKa > −2) with the exception of hydroxide, which reaches the expected diffusion limit around 1010 M−1 s−1. This behavior is quite similar to ground state deprotonation of phenol itself (indicated by the gray line), when taking the larger size and thus slower diffusion of phenol-DAOTA into account, as shown in the diffusion corrected plot in Fig. S25 (ESI §6†).35,36
![]() | ||
Fig. 7 Rate of PCET as a function of the base strength of the quencher. Log(kPCET) as a function of ΔpKa from the base strength of the quencher (given as pKa of the corresponding acid, BH) relative to the pKa of the proton donor. Data points for phenol-DAOTA quenching in blue, the black line is a guide to the eye. The gray line shows literature values measured of the ground state deprotonation rate of phenol (see ESI §6† for details). |
Secondly, phenol-DAOTA shows a higher sensitivity for weaker bases than phenol itself by ∼3 ΔpKa units, as seen by the horizontal shift of the full black vs. gray line. For the diffusion corrected values, this offset is close to 4 pKa units (Fig. S25†). This suggests that phenol-DAOTA is a stronger acid in the excited state ( ≈ 5) than in its ground state (pKa = 8.9). The minute shift in the absorption spectra of phenol-DAOTA (Fig. 3A) upon deprotonation shows that the excited state acidity does not originate from stabilization of the phenolate form itself37 but must be a result of a concerted proton electron transfer (CPET) directly to the low energy PET state that is more favorable than the sequential PCET transfer.27,38 Thus, the additional driving force for proton transfer comes from formation of the low energy PET state (Fig. 2). From a sensor point of view, this enhanced sensitivity to weak bases is highly advantageous since the ground state pKa of 8.9 allows the probe to be fully protonated (and thus emissive) at all biologically relevant pH values while still being sensitive to (quenched by) relevant weak bases/metabolites. The sensitivity of the phenol-DAOTA probe for determining concentrations of bases in the various buffer solutions can be estimated from the Stern–Volmer equation (eqn (2)) and the accuracy of the fluorescence lifetime measurements. The minimum analyte concentration that is measurable by the phenol-DAOTA probe is 38 mM for formic acid, while acetate and phosphate can be detected at concentrations as low as 6 mM and 2 mM respectively (see ESI §7 and §8† for details on experimental reproducibility and detection limits). These detection limits may not seem impressive compared to PET probes binding to analyte with high affinity receptors in their ground state (like the calcium probe 2 in Fig. 1), however they are actually in the relevant range for the abundant and important metabolites we have tested here.30 In the tested buffer systems the fluorescence lifetimes (and intensities) are quenched by ∼4–50% (Table S4†). This range is close to ideal as this leaves sufficient signal to obtain reliable lifetime measurements, in contrast to 80–100% quenching where little to no signal is measurable resulting in large errors in lifetime values. The long fluorescence lifetime of the DAOTA fluorophore furthermore allows for time-gated measurements, eliminating interference from other fluorescent dyes and autofluorescence in biological samples.39,40 We demonstrate this by reproducing the probe lifetimes in a phosphate buffer in the presence of a cyanine fluorophore providing an overlapping fluorescence signal accounting for up to ≈25% of the total intensity (see ESI §11†). We thus envision that dynamic PCET probes, like phenol-DAOTA, can be developed into general tools for investigating complex biological buffer systems and changes in metabolite pools, even in the presence of overlapping scattering and autofluorescence.
Dynamic PCET probes cannot be expected to be selective for any specific analyte in a complex system, but will provide a response that is the sum of contributions from each base, weighted by its concentration and kPCET, as we showed for the acetate/phosphate mixture (ESI §10†). In cases where one analyte is all dominating it may be possible to extract its concentration, as in the case for Cl− sensing in cellular systems by PET probes.25 While the PCET probes in most cases will not provide a measure on one specific analyte/metabolite, we envision that the cumulative response of dynamic PCET probes will provide a new general descriptor of a buffer system. In conjunction with the pH value, this may provide new insights on the state of, or changes to, cellular buffer systems. For instance, it has been speculated that cells may have differential local buffers to support various biochemical processes.46 PCET probes could map out spatiotemporal variations in metabolite/buffer composition and/or concentration to elucidate such questions. Likewise, PCET probes may help us understand how different buffers and anions influence other probes and biomolecules.47,48
For the development of PCET probes, we can consider the optimal properties of such probes. The probes should fulfil the following design criteria: (1) have only one emitting state, for which the fluorescence lifetime is proportional to the rate of PCET quenching induced by collision with the analytes (eqn (1) and Fig. 2). (2) The probe should not be quenched by water, as is the case for many photoacids.49 (3) The probe should be fully ON at physiologically relevant pH values and still be effectively quenched by most relevant metabolites/bases. (4) Long excited state lifetimes (τ0) will increase sensitivity (eqn (2)). For example, the 14 ns lifetime of the DAOTA fluorophore50,51 in water enhances its sensitivity to diffusion limited quenching by a factor of 3 or more compared to most fluorophores with lifetimes below 4 ns. The long fluorescence lifetime also allows the use of time-gated detection to eliminate the interference from autofluorescence.40,52–54 (5) The modularity of the probe will allow for tuning properties of the dye component somewhat independent of the quencher group. For example, the pKa of the quencher can easily be tuned by substitution,17,18,27 while decreasing the reduction potential of the dye will increase the PCET driving force, even to the point where water will facilitate quenching.55 (6) Finally, fluorescence lifetime (τ0) and spectra of the fluorophore should be insensitive to salt and buffer variations, as is the case for phenol-DAOTA (Tables S1 and S4†).
All of the above listed properties are fulfilled by the phenol-DAOTA. However, for use in cells and other complex biological systems we will need further improvements of phenol-DAOTA. This includes introducing reactive functionalities so the probes can be conjugated to dextrans or other carrier biomolecules in order to enhance their water solubility,56 avoid undesired localization and control the probe distribution in complex biological systems.57–59
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5sc00326a |
‡ Contributed equally. |
This journal is © The Royal Society of Chemistry 2025 |