Makay T.
Murray
a,
Austin
Pounder
a,
Ryan E.
Johnson
b,
Keenan T.
Regan
b,
Richard A.
Manderville
b and
Stacey D.
Wetmore
*a
aDepartment of Chemistry and Biochemistry, University of Lethbridge, 4401 University Drive West, Lethbridge, AB T1K 3M4, Canada. E-mail: stacey.wetmore@uleth.ca
bDepartment of Chemistry & Toxicology, University of Guelph, 50 Stone Road East, Guelph, ON N1G 2W1, Canada
First published on 1st August 2025
Fluorescent probes are powerful tools for the detection of proteins in biomedical applications. However, the design of selectively active fluorescent probes is challenging due in part to difficulties predicting the functions of novel modifications, especially in different cellular environments. In the present study, a family of cyclic N-glycol-linked 4-formyl-aniline probes (denoted AnMeInd, AnMeBtz, and AnBtz), which have distinct thrombin binding affinities and fluorescent responses, were investigated at the T3 position of the thrombin binding aptamer (TBA) using TD-DFT calculations and classical and driven-adaptive bias molecular dynamics (MD) simulations. Classical MD and D-ABMD simulations corroborate the experimentally observed differences in thrombin binding affinities relative to canonical TBA, highlighting that neutral, shorter probes that lack an abundance of exocyclic substituents are better accommodated in a thrombin binding pocket that is rich in aromatic and charged residues. TD-DFT suggests that cationic AnMeInd and AnMeBtz exhibit the same inherent tendency to undergo twisted intramolecular charge transfer (TICT). However, the methyl substituents of AnMeInd reduce probe contacts with the aptamer scaffold to enhance TICT in the unbound state, and adjust probe binding location in the thrombin hydrophobic pocket, which increases solvent shielding, probe rigidity, and TICT suppression upon target binding compared to AnMeBtz. Unlike the cationic analogues, combined computational and experimental data suggest the best AnBtz probe functions as a photobase. Although AnBtz undergoes protonation in the excited state in solvent exposed environments (i.e., unbound TBA), which facilitates nonradiative decay, this pathway is suppressed upon thrombin binding due to reduced solvent accessibility, thereby enhancing fluorescence. These findings underscore the importance of tuning probe charge, linker length, exocyclic substituents, flexibility, and microenvironment in both the unbound and bound states to optimize fluorescence response. Our results provide a strategic foundation for designing high-performance fluorescent probes for biosensing and nucleic acid-based diagnostics.
Protein induced fluorescent enhancement (PIFE) is an extremely useful ‘turn-on’ mechanism for fluorogenic probe design. Fluorescent molecular rotors (FMRs) that contain flexible, bridged aryl species with electron donor and acceptor moieties, such as cyanines, hemicyanines, and coumarins (among others),12 can exhibit PIFE due to changes in probe rigidity and/or solvent quenching upon target binding.13 A key underlying photophysical phenomenon that can give rise to PIFE for FMRs is twisted intramolecular charge transfer (TICT, Fig. 1A).12 Specifically, upon excitation of an electron to the S1 state, electron density shifts toward the acceptor moiety, followed by structural relaxation. The new geometrically-relaxed structure in the S1 state is accessible through free rotation about the bridge, resulting in orthogonal donor and acceptor ring planes. Electronic relaxation (S1 → S0) in this orthogonal conformation goes through a nonradiative decay pathway, quenching fluorescence compared to that emitted from a planar conformation (Fig. 1B). As a result, FMRs demonstrate viscosity-sensitive fluorescence, exhibiting minimal emission in low-viscosity environments (water) and enhanced fluorescence in high-viscosity media (polyethylene glycol, glycerol). In parallel, protein binding can inhibit inter-aryl rotation and promote the (planar) emissive intramolecular charge transfer (ICT) state. The optimum probe design to maximize PIFE is not well understood, necessitating studies on candidate probes with a range of target binding affinities and fluorescence readouts.
The thrombin binding aptamer (TBA) provides an excellent platform for testing probe performance.8,10,14–17 TBA binds to the fibrinogen recognition site (exosite I) of human α-thrombin with an apparent dissociation constant (Kd) ranging from low- to mid-nanomolar.18 TBA (5′-GGTTGG-TGT-GGTTGG-3′) folds into a chair-like structure, with two guanine tetrads forming an antiparallel guanine-quadraplex (GQ) topology stabilized by a central potassium ion. Folding results in two parallel TT loops on one TBA face and a TGT loop on the opposing face.19 TBA binds to the thrombin target via the TT loops (T3/T4 and T12/T13), while the TGT loop is solvent exposed. Although the direct incorporation of hemicyanine probes into TBA has already demonstrated considerable success for thrombin detection,8–10,20 there is room for improvement in probe design. However, de novo prediction of the structure–function relationship of aptamers containing a PIFE-based probe, particularly in terms of enhanced target binding affinity and response, is challenging. Although structural information would facilitate our understanding of the intricacies of probe function, only a few X-ray crystal structures of modified TBA bound to thrombin are available to our knowledge,18,21–26 none of which contain a PIFE-based probe. Furthermore, computational investigations of the structure and function of fluorescent probes incorporated into TBA remain scarce.8,10,27–29
In the present work, we investigate modified TBA and TBA–thrombin complexes containing three fluorogenic probes that differ in their charge (neutral and cationic), acceptor ring systems (benzothiazole and indoline), and inter-ring linker (olefin and single bond). Specifically, we investigate AnBtz (2-(4-dimethylaniline)benzothiazole), AnMeInd (2-[4-(dimethylaniline)ethenyl]-1,3,3-trimethyl-3H-indolium), and AnMeBtz (2-[4-(dimethylaniline)ethenyl]-3-methylbenzothiazolium; Fig. 1C). These modifications were selected due to the availability of accurate experimental data for the probes incorporated at the T3 position of TBA,8 which highlight the effects of the probe chemical composition on thrombin binding affinities (Kd), fluorescent absorption/emission (λex/λem) spectra, relative fluorescence intensity upon target binding (Irel), and quantum yield (Φf). We have employed molecular dynamics (MD) simulations to unravel the structural dynamics of the probe-modified TBA aptamers in isolation and when bound to thrombin to identify factors responsible for unbound probe quenching and bound probe fluorescent enhancement. Driven-adaptive biased MD (D-ABMD) was used to further explore the relative thrombin binding affinity of the decorated TBAs. Time-dependent density functional theory (TD-DFT) along with experimentally-determined fluorescence intensities of the free probes across a range of solvents were utilized to investigate the inherent FMR characteristics of the probes and thereby rationalize differences in the photophysical responses. Together, these findings provide critical insights into the behavior of probe-modified TBAs and establish key design principles for the future development of improved fluorescent biosensors. By highlighting the roles of charge, linker length, conformational flexibility, and photobasicity in the interplay between TICT and FMR behavior, this study lays the groundwork for the rational development of next-generation fluorogenic aptamer sensors for biomolecular detection.
Each TBA/TBA–thrombin model was prepared for simulation using the tLEaP module of AMBER 22.44 Specifically, each system was solvated in a truncated-octahedral box of TIP4P-EW water,33 with the nearest box face being a minimum distance of 10 Å from the solute and the probe identity dictating total box size. This added ∼2100–4000 water molecules for each unbound TBA and ∼11000 water molecules for each TBA–thrombin complex (Table S3). Next, each system was neutralized with Na+ and then brought to an intracellular physiological NaCl concentration of 150 mM, with ion count determined using the SLTCAP45 calculator.
Bound TBA–thrombin complexes were initially simulated for 1 μs in triplicate for the three distinct probe poses across both crystal structures (Fig. S2). Analysis of the resulting trajectories indicated that the 4DII thrombin–TBA binding orientation is not viable for the probe-modified TBAs and therefore these models were not further considered (see results). Due to the observed high dynamics of all probes in the 1HAO orientation, the corresponding simulations were treated as an extended equilibration step. The trajectories were clustered with respect to the root-mean-square deviation (RMSD) of residues at the aptamer–thrombin interface (defined as G2–G5 and G11–G14 (TBA), and Phe34, Lys36, Gln38, Leu64–Arg67, Tyr76, Arg75–Asn79, and Glu81–Leu86 (thrombin)). The representative structure from the dominant cluster was re-solvated and re-ionized (Table S3) prior to initiating 9 × 1 μs final MD production simulations. For unbound TBA models, 1 μs MD production simulations were performed in triplicate from each of the three initial probe orientations (Fig. S3). All simulations were run as an NPT ensemble at 310 K (Langevin thermostat)47 and 1 bar (Berendsen barostat).48 The water density was ∼1 g mL−1 and the periodic boundary condition was enabled. Frames from these final production simulations were saved to disk every 20 ps for analysis (a total of 20000 frames per simulation replica).
The synthesis of AnBtz, (2-(4-dimethylaniline)benzothiazole), was carried out following known procedures.58 4-(Dimethylamino)benzaldehyde (202.6 mg, 1.34 mmol) and 2-aminothiophenol (187.0 mg, 1.47 mmol) were combined in 15 mL of EtOH and heated to reflux for 18 h. The reaction mixture was subsequently cooled to 0 °C and filtered to obtain the product as yellow crystals (286.7 mg, 1.12 mmol, yield = 83%). The product was characterized by 1H NMR and 13C NMR and was in agreement with the literature.58
A second canonical TBA–thrombin binding orientation has been identified using cytolysin A (ClyA) biological nanopores,61 which was attributed to the configuration observed in the PDB ID: 1HAO62 crystal structure that differs from 4DII by a 180° rotation about the TBA central axis prior to thrombin binding (Fig. S2). Furthermore, the same alternate binding configuration occurs when TBA is bound at thrombin exosite I and the larger HD22 aptamer is simultaneously bound at exosite II (PDB IDs: 5EW1 and 5EW2).26 There is also evidence that the preferred TBA–thrombin binding configuration is impacted by modification of the TT loops.14,23,61 For example, an abasic site (tetrahydrofuran) at T3 or T12 enhances the Kd of a particular pose,61 with an abasic site at T3 inducing binding as per 1HAO (PDB ID: 4LZ4) and an abasic site at T12 resulting in binding as per 4DII (PDB ID: 4LZ1).22 Furthermore, X-ray crystal structures of TBA containing lactose, glucose, or leucine bound via a triazole moiety to N3 of T3 (PDB IDs: 6Z8V, 6Z8W, and 6Z8X)23 show the modified nucleotides bound in a thrombin hydrophobic pocket similar to that encapsulating T3 in PDB ID: 1HAO.62 Previous computational studies have also revealed that bulky modifications at T3 prefer the 1HAO binding orientation.14 Therefore, the 1HAO TBA–thrombin binding configuration was also considered for TBA containing the hemicyanine probes, with the complexes remaining well bound and stable throughout the simulations (Fig. S8–S10).
When canonical TBA is bound to thrombin according to the 1HAO binding pose (Fig. 2), T3 and T12 form many noncovalent interactions with thrombin over the course of the classical MD simulations (Fig. S11). Specifically, T3 π–π stacks with Tyr76 (47 ± 25% occupancy) and T12 is positioned in a hydrophobic pocket comprised of Ile24, Tyr117, His71, and Ile79. On the other hand, Tyr76 hydrogen bonds with O4′ in the T4 sugar moiety (67 ± 27%), and the T4 and T13 nucleobases stack against the GQ of TBA, which permits hydrogen bonding to each other as well as Arg75 and Arg77A. At the same time Arg75 stacks with G5 (83 ± 13%), while Arg77A stacks with G14 (38 ± 17%). Visual inspection of the D-ABMD simulation trajectories reveal that TBA disengagement from thrombin initially occurs in the region near T12, which is followed by disruption of Arg75 hydrogen bonds with T4, T13, and G5. To preserve the remaining strong contacts with thrombin, the TBA TT loops distort to contract around Tyr76 and Arg77A, while exosite I distorts to maintain hydrogen bonding between Tyr76 and T4. The hydrogen bonds between T4 and T13 and Arg77A are the last TBA–thrombin contacts to be severed as the complex dissociates, highlighting the importance of these interactions for the high affinity of TBA to thrombin. Overall, unbinding of canonical TBA from thrombin requires maximum work of −3.9 ± 0.3 kcal mol−1 (Fig. 3A).
![]() | ||
Fig. 2 (A) Canonical TBA–thrombin complex bound according to PDB ID: 1HAO. TBA–thrombin binding interface around (B) T3 and T4, and (C) T12, T13, and G14. Stacking interactions shown with red dotted lines and hydrogen-bonding interactions with yellow dotted lines. Select average occupancies and standard deviations across replicates are provided. |
The AnMeInd probe is highly dynamic at the T3 position of the TBA when bound to thrombin (Fig. S8–S10), with overlays of classical MD snapshots highlighting that the probe primarily rests against the thrombin surface (Fig. 4A-i). Indeed, the steric repulsion between the exocylic methyl groups on the cationic AnMeInd acceptor ring and Arg73 coupled with electrostatic repulsion between the positively charged probe and Arg67, Lys70, and Arg75 likely prevent probe insertion deep into the thrombin pocket (Fig. 4B-i). At the thrombin surface, the representative structure from the dominant conformation reveals the edge of the indolinium ring forming a T-shaped interaction with the face of Phe34 (32 ± 13% occupancy), the backbone of Gln38 directed toward the charged ring (7 ± 4%), and the acceptor moiety π–π stacking with Arg67 (45 ± 24%). The donor ring of the probe stacks with Tyr76 (62 ± 18%). Tyr76 and Arg75 hydrogen bond with the neighboring T4 of TBA (Fig. S12), albeit with reduced occupancy relative to canonical TBA (Fig. S11). In the representative structures from other (minor) clusters, slight distortions from this dominant conformation occur that primarily shift the G14–G15 edge of TBA closer to the thrombin center of mass, movement that correlates with the probe detachment from the protein periodically seen as part of the system dynamics (Fig. 4A-i). Indeed, D-ABMD simulations suggest that minimal work (−3.4 ± 0.2 kcal mol−1, Fig. 3B) is required to break the Tyr76 and Arg75 hydrogen bonds with T4 in conjunction with detachment of AnMeInd from the thrombin surface, while maintaining the global fold of canonical TBA, which correlates with the measured decreased thrombin binding affinity upon probe incorporation into TBA.8
![]() | ||
Fig. 4 (A) 100 frame overlays of the T3 probe across all MD replicates initiated from PDB ID: 1HAO. Representative structures of (B) thrombin bound and (C) unbound TBA containing (i) AnMeInd, (ii) AnMeBtz, and (iii) AnBtz at T3. π–π interactions shown as red dotted lines. |
As discussed for AnMeInd, the positively-charged AnMeBtz likely experiences electrostatic repulsion with Arg67, Lys70, and Arg75, which prevents insertion of the probe deep into the thrombin pocket. As a result, AnMeBtz similarly rests against the edge of thrombin, although exhibits increased dynamics (Fig. 4B-ii and Fig. S8–S10, S13). In the dominant conformation, the acceptor ring forms π–π stacking interactions with Phe34 (6 ± 21%) and Arg67 (18 ± 33%) as well as edge-on interactions with Met32 (12 ± 24%) and Gln38 (10 ± 11%, Fig. 4B-ii). The donor moiety does not form long-lasting interactions with thrombin. Nevertheless, during D-ABMD simulations, the TT loops visibly resist dissociation, with the loops stretching to maintain interactions with thrombin as TBA disengages. Specifically, a T12 (N3⋯O) Ile61 hydrogen bond is maintained, which outlasts the interactions between Arg77A or Arg75 and TBA. Unlike the rapid unbinding of AnMeInd from the thrombin surface during D-AMBD, AnMeBtz slides along the face of thrombin, forming interactions with Phe34, Met32, Lys80, Gln39, Tyr76, and Arg75 before unbinding from the protein. This drawn-out disengagement process results in a greater mean of the maximum work for AnMeBtz (−3.7 ± 0.2 kcal mol−1, Fig. 3C) compared to AnMeInd (−3.4 ± 0.2 kcal mol−1, Fig. 3B), but less than canonical TBA (−3.9 ± 0.3 kcal mol−1, Fig. 3A), rationalizing the relative experimental binding affinities.8
Unlike the probes with the olefin linker, classical MD simulations demonstrate that AnBtz can deeply intercalate into a thrombin pocket located above Tyr76, which is formed by Arg67, Phe34, Leu65, Gln38, and Met32 (Fig. 4A-iii and B-iii). This binding configuration is well maintained across all replica simulations, with slightly different relative arrangements of thrombin and TBA throughout the simulations highlighting the dynamic nature of the complex (Fig. S8–S10 and S14). Nevertheless, AnBtz is much less dynamic in this pocket than the other probes, which bind to the surface of thrombin. Indeed, the acceptor ring of AnBtz forms persistent π–π interactions with both Phe34 (69 ± 39%) and Arg67 (64 ± 12%), while also exhibiting a consistent edge-on interaction with Met32 (85 ± 37%). Although the donor ring of AnBtz does not π–π stack with Tyr76 (occupancy < 1%), and the hydrogen bond between Tyr76 and T4 is significantly weakened (36 ± 30%) compared to canonical TBA (67 ± 37%), no other canonical TBA–thrombin interactions are significantly altered by the presence of AnBtz at T3 (Fig. S11 and S14). The persistent intercalation of the probe into the thrombin pocket contributes to the enhanced thrombin binding affinity of AnBtz–TBA compared to canonical TBA, with D-ABMD revealing −11.7 ± 0.5 kcal mol−1 is required to disengage the modified TBA from the target (Fig. 3D). Along the disengagement pathway, AnBtz π–π stacking interactions with Phe34 and Arg67 are the last to breakdown, being persistent even after the hydrogen bonds with Arg75, Tyr76, and Arg77A are disrupted. This correlates with the reported enhanced thrombin binding affinity upon incorporation of AnBtz into TBA.8
Overall, classical MD combined with D-ABMD simulations provide structural rationalization for the observed differential binding affinities upon probe incorporation at T3 of TBA. Indeed, the qualitative trend of decreasing binding affinity seen experimentally, namely AnBtz > TBA > AnInd > AnMeBtz, is reproduced by the D-ABMD simulations. Furthermore, classical MD simulations uncover key differences in probe binding that explain this trend. Specifically, our simulations indicate that a deep hydrophobic pocket that can bind the probes exists above Tyr76, which is formed by Arg67, Lys70, and Lys14A. As this pocket is rich in both aromatic and charged residues, non-decorated, planar, and neutral probes (e.g., AnBtz) are better accommodated. In contrast, electrostatic repulsion from Arg67 and Lys70, as well as Arg75, prevents cationic probes (e.g., AnMeBtz and AnMeInd) from binding. Out-of-plane exocyclic substituents (e.g., methyl groups in AnMeInd) further generate steric interactions with Arg67 that prevent efficient intercalation. Furthermore, the olefin bridge of both AnMeInd and AnMeBtz is longer than the C–C linker of AnBtz (by ∼2.2 Å), which further hinders effective intercalation into the thrombin pocket. Indeed, overlays of each probe onto the T3 position according to key (O3′ and O5′) backbone atoms illustrates that the probes containing the olefin bridge do not fit in this pocket (Fig. S7B). Thus, our MD simulations have identified key chemical features for optimal probe design in terms of the charge of the acceptor moiety and the linker length.
TD-DFT calculations show that both AnMeInd and AnMeBtz undergo a HOMO → LUMO transition with a strong oscillator strength, indicating efficient electronic excitation (Fig. 5A). Although the frontier molecular orbitals are uniformly distributed in both probes, the S0 HOMO is slightly more localized on the donor moiety, while the S0 LUMO exhibits a stronger localization on the acceptor moiety. Examination of the S1 EEP suggests that rotation about the amine donor does not substantially contribute to TICT as EDE is endergonic for both probes. Instead, consideration of rotation about each bond in the linker reveals the two cationic probes likely undergo TICT formation upon 90° rotation about the olefin bond, which results in a lower barrier and exergonic process (Fig. 5C-i and ii). Indeed, rotation with respect to other bonds along the flexible π-bridge likely do not significantly contribute to TICT due to a higher ERB and less favorable EDE (Fig. S16). In fact, the low oscillator strength of the twisted conformation following 90° olefin rotation supports the nonemissive nature of this state (Fig. 5B). Nevertheless, the ERB for twisting about the olefin bond is comparable for AnMeInd and AnMeBtz (4.9 and 4.5 kcal mol−1, respectively). Furthermore, both probes possess a sufficient EDE for TICT state formation (−7.5 and −10.3 kcal mol−1, respectively; Fig. 5C-i and ii). This indicates that the inherent propensity for AnMeInd and AnMeBtz to undergo TICT is unlikely to be the primary factor behind the observed differences in the fluorescence responses upon probe-decorated TBA binding to thrombin.
To evaluate the impact of changes in the local microenvironment of the probe upon binding to thrombin on the observed fluorescence enhancement, classical MD simulations were considered for both the unbound probe–TBA complexes and the corresponding thrombin-bound forms. The unbound aptamer containing AnMeBtz or AnMeInd remained highly stable over the simulations and exhibited similar dynamics in each nucleotide (Fig. S17 and S18). Both probes are consistently aligned with respect to T4 and T13 in unbound TBA (representative structure >90% occupancy, Fig. 4C). Both probes exhibit comparable levels of solvent exposure in this state, ranging from 48–50% (Fig. 6A). However, subtle differences in stacking interactions were observed, with AnMeInd exhibiting stacking occupancies of 75 ± 28% with T4 and 80 ± 30% with T13, whereas AnMeBtz showed stronger interactions (89 ± 6% and 95 ± 2% occupancy, respectively; Fig. 4C-i, ii and Fig. S17 and S18). Indeed, AnMeInd is more conformationally flexible in the unbound TBA state.
Upon thrombin binding, AnMeInd undergoes a greater increase in solvent shielding compared to AnMeBtz (10.8% vs. 9.6%, Fig. 6A). Indeed, as previously discussed, AnMeBtz exhibits greater fluctuations than AnMeInd in the TBA–thrombin complex over the course of the classical MD simulations (Fig. 4A and Fig. S8–S10), which permits AnMeBtz to explore more diverse and solvent exposed conformations when bound to thrombin. As a result, the nonemissive TICT state, which is typically favored in polar environments, is more prominent for AnMeBtz, at least in part leading to the observed reduced fluorescent response (Irel = 2.0 vs. 4.0). In contrast, the nonemissive TICT state is suppressed for AnMeInd, which results in fluorescence recovery and prolonged excited-state lifetimes, rationalizing the observed fluorescent enhancement.69 Notably, both AnMeInd and AnMeBtz exhibit similar conformational rigidity upon thrombin binding (173.5 ± 33.4° vs. 189.5 ± 33.5°, Fig. 6C), with minimal change between the unbound and bound states (Fig. 6B). This suggests that both probes experience comparable structural constraints in the thrombin-bound complex.
Given that both cationic probes share a comparable intrinsic propensity for TICT and exhibit similar structural dynamics in both unbound and thrombin-bound states, these factors are unlikely to be the primary drivers of the observed differences in fluorescence response. Rather, the reduced fluorescence response of AnMeBtz likely stems from its increased solvent exposure in the thrombin-bound state, which facilitates TICT. Moreover, the exocyclic methyl groups of AnMeInd likely prevent optimal contact with T4/T13 in the unbound aptamer, promoting partial probe disengagement, which increases rotational freedom upon excitation and contributes to lower background fluorescence, thereby amplifying the observed fluorescence enhancement upon thrombin binding. Thus, although careful design is required to preserve target binding affinity, intentionally preventing stable, noncovalent interactions in the unbound aptamer state using exocyclic substituents offers a viable strategy to reduce background fluorescence by enabling TICT without the need to first overcome stabilizing contacts with the aptamer scaffold.
Classical MD simulations and TD-DFT results suggest that another microenvironmental factor must be responsible for the fluorescence response of AnBtz. Indeed, the simulations highlight that the neutral probe experiences the greatest increase in solvent shielding upon thrombin binding (21.2%, Fig. 6A), indicating a distinct environmental shift compared to the cationic probes. One plausible explanation for the differential fluorescence modulation is that AnBtz functions as a photobase – a molecule that exhibits higher proton affinity in the excited state than in the ground state. A photobase can result from a change in electronic structure upon photoexcitation, which allows the excited molecule to abstract a proton from the environment. Several experimental and computational investigations by the Dawlaty and Petit laboratories have showcased the photobasicity of nitrogen-containing heteroaromatic ring systems, with KB values increasing by up to 10 orders of magnitude upon photoexcitation.70–75
TD-DFT calculations suggest that the nitrogen of the benzothiozole moiety becomes more electron rich upon photoexcitation of AnBtz (Fig. 7B). Compared to the neutral species (Fig. 5C-iii), AnBtz-H likely exhibits TICT state formation upon rotation of both the C–N of the donor (ERB = 5.3, EDE = −7.9 kcal mol−1) and the C–C biaryl bridge (ERB = 2.0, EDE = −5.9 kcal mol−1, Fig. 7C). Protonation to yield AnBtz-H creates a stronger acceptor moiety, which promotes TICT. To experimentally validate the feasibility of proton abstraction by AnBtz, we measured the fluorescence intensity of AnBtz in various nonpolar, polar-protic, and polar-aprotic solvents. We found the fluorescence intensity was similar irrespective of the dielectric constant; however, there was a dramatic drop in intensity in trifluoroethanol (TFE to EtOH, Irel = 4.00, Fig. 7D and Table S6). To establish a relationship between the pKa of the solvent and viscosity, we measured the fluorescence intensity change from TFE to a 25% TFE:glycerol mixture. While MeOH to MeOH:glycerol led to a mild decrease in intensity upon increased viscosity (Irel = 0.97), TFE to TFE:glycerol led to substantial increase in intensity (Irel = 3.21), supporting FMR characteristics (Fig. 7A). Therefore, the AnBtz probe appears to act as a photobase, with excited-state protonation likely driving the observed turn-on response.
Overall, we proposed that upon photoexcitation in an acidic enough environment, such as the 10 mM Tris-HCl buffer at pH 7.3 used experimentally,8 a protonated AnBtz-H species is generated, which functions as an FMR and will efficiently undergo TICT. Thus, AnBtz in unbound TBA is likely protonated through solvent exposure and does not exhibit significant fluorescence, consistent with the low background signal observed experimentally for the free AnBtz probe (this work) and when positioned at T3 of TBA.8 However, when intercalated within the hydrophobic pocket of thrombin, the probe is significantly shielded from solvent (Fig. 6A), which prevents protonation and permits strong fluorescence (Fig. 7E). This interplay between the local environment photoinduced protonation and TICT offers a mechanistic explanation for the fluorescence enhancement observed for AnBtz (Irel = 4.0),8 and points to a new important design principle that can direct development of improved probes.
While TD-DFT calculations suggest similar propensities for TICT formation among the cationic probes, MD simulations reveal differences in probe microenvironments upon thrombin binding. Indeed, the methyl substituents of AnMeInd reduce probe contacts with the aptamer scaffold, which enhances TICT in the unbound state compared to AnMeBtz. In direct complement, the additional substituents adjust probe binding location in the thrombin hydrophobic pocket, which increases solvent shielding, probe rigidity, and TICT suppression upon target binding, rationalizing the observed enhanced fluorescence response for AnMeInd compared to AnMeBtz.
Distinct from the cationic probes, combined classical MD, TD-DFT, and experimental fluorescence studies suggest that the neutral AnBtz probe with the most promise operates via a different mechanism involving excited-state proton transfer. Specifically, AnBtz behaves as a photobase, which undergoes protonation upon excitation in aqueous solution. In the unbound aptamer, the solvent exposed AnBtz is susceptible to excited state protonation and subsequent TICT, which induces rapid decay through nonradiative pathways that results in fluorescence quenching. However, when the probe is intercalated into thrombin, the hydrophobic, solvent-shielded environment inhibits protonation, stabilizing an emissive neutral state and enabling fluorescence recovery.
Our findings highlight the importance of considering conformational rigidity, noncovalent interactions with the aptamer scaffold, and local environmental effects when designing fluorescent probes. To optimize fluorescence enhancement via PIFE, future probe designs should reduce background fluorescence by prioritizing TICT suppression through enhanced solvent shielding and increased rigidity upon target binding. Incorporating photobasicity offers another promising approach, allowing for a more pronounced TICT-based fluorescence turn-on response without the need for an intrinsically charged probe. These design principles establish a framework for the rational development of high-performance fluorescent probes, expanding their applicability in biosensing, nucleic acid biotechnology, and other fluorescence-based detection platforms.
NMR characterization, photophysical properties, synthetic details, computational details and SI Figures/Tables. See DOI: https://doi.org/10.1039/d5cp01851j
This journal is © the Owner Societies 2025 |