Ultrafast Excited-State Dynamics and Two-Photon Near-Infrared Induced Photodynamic Therapy Performance of 5‑Phenylethynyl-4-thiouridine

Abstract

5-Phenylethynyl-4-thiouridine (5ph-4TU), featuring both π-conjugated phenylethynyl and carbonyl-sulfur substitutions, overcomes the UVA absorption limitations of conventional thio-nucleosides by extending absorption into visible and potentially longer-wavelength regions under single or two photon excitationan essential advancement for enhancing nucleosides’ photoactivity toward deep-tissue cancer photodynamic therapy (PDT). However, the ultrafast excited-state dynamics and synergistic effects of these dual modifications remain unexplored, despite their critical role in generating unique excited-state characters inaccessible for canonical nucleosides. Here, we employ ultrafast spectroscopy combined with QM­(MS-CASPT2//CASSCF)/MM calculations to map out the unusual excited-state pathways of 5ph-4TU. We reveal that the C5-phenylethynyl group can effectively suppress C5C6 bond twisting to incur a high barrier of 0.8 eV from the S₂ (¹ππ*)_min to access S₂ (¹ππ*)/S₁ (¹nπ*) conical intersection. This hinders internal conversion from S₂ (¹ππ*) to S₁ (¹nπ*), resulting in a long-lived S₂ (¹ππ*)_min state (∼600 ps in CH₃CN) from which enhanced fluorescence occurs (Φ_FL = 0.11 ± 0.03). Interestingly, once reaching S₁ (¹nπ*)_min, the triplet manifold can be effectively populated through the S₁ (¹nπ*)/T₂ (³ππ*)/T₁ (³nπ*) triple quasi-degeneracy region with significant spin–orbit coupling (SOC = 123 cm^–1). Furthermore, we find that the extended π-conjugation enhances two-photon absorption into the near-infrared range while preserving the same excited-state dynamics as single-photon excitation. In vitro experiments confirm excellent biocompatibility and efficient singlet oxygen (¹O₂) generation, inducing cancer cell death under 760–770 nm two-photon irradiation, underscoring its potential for deep-tissue PDT. These findings provide a new paradigm for designing next-generation biophotosensitizers with combined imaging and therapeutic functionalities.

Introduction

Nucleobases are the primary chromophores responsible for the photophysical and photochemical behavior of DNA and RNA. , Canonical nucleobases predominantly absorb ultraviolet C (UVC) light (200–280 nm) and undergo ultrafast internal conversion (IC) to the electronic ground state within the subpicosecond range, facilitated by a ¹ππ*/S₀ conical intersection. − This highly efficient nonradiative decay mechanism minimizes the risk of DNA photodamage, playing a crucial role in the exceptional photostability of genetic material. However, in certain biomedical applicationssuch as photochemotherapy or photodynamic therapy (PDT)enhancing the photoactivity of nucleobases is imperative, particularly for selectively disrupting the DNA of pathological cells like cancer cells.

Thio-nucleobases, a class of nucleobase analogues derived from the substitution of the carbonyl oxygen with a sulfur atom in purine and pyrimidine (Scheme ), have garnered increasing attention due to their unique photophysical and photochemical properties. , These modified nucleobases can be metabolically incorporated into cellular DNA or RNA (existent in the form of thio-nucleosides), acting as embedded photosensitizers that preferentially accumulate in cancer cells. Upon photoexcitation, they generate cytotoxic singlet oxygen (¹O₂) with high yields site-specifically, distinguishing them from conventional photosensitizers by their biocompatibility and selective tumor targeting ability. As a result, there has been a surge of interest in both experimental and theoretical investigations into the distinct excited-state decay mechanisms of this important class of biophotosensitizer. −

1. Molecular Structures of Uridine (U), 4-Thiouridine (4TU) and 5-Phenylethynyl-4-thiouridine (5ph-4TU).

For thio-nucleosides, such as 4-thiouridine (4TU), 4-thiothymidine (4TT), , and 6-thioguanine (6TG), ,,, prevalent studies show that the UV excitation populate the ¹ππ* state first, which then relaxes to ¹nπ* in the subpicosecond range through a ¹ππ*/¹nπ* conical intersection (CI). The ¹nπ* state serves as a gateway to the triplet manifold due to the heavy-atom effect of sulfur, leading to near-unity triplet quantum yields. Unlike canonical nucleosides, which undergo ultrafast IC to the ground state, thio-nucleosides efficiently populate long-lived, reactive triplet states that facilitate energy transfer to molecular oxygen, generating reactive oxygen species (ROS) such as ¹O₂an essential feature for PDT. Moreover, sulfur substitution red-shifts the absorption spectrum into the UVA region, enabling selective excitation at longer wavelengths and reducing damage to canonical nucleobases in surrounding healthy tissues. Despite these advantages, the absorption spectrum of thio-nucleosides remains limited to the UVA region, with peak absorption wavelengths of 330 nm (4TU), 337 nm (4TT), and 341 nm (6TG). This spectral limitation restricts their use in phototherapeutic applications, as UVA light has limited penetration depth in biological tissues.

A promising approach that has emerged to overcome the spectral limitations of thio-nucleosides is the introduction of π-conjugated functional groups, which can further red-shift absorption into the visible range. Recently, a novel thio-nucleoside, 5-phenylethynyl-4-thiouridine (5ph-4TU) as shown in Scheme , was synthesized by covalently attaching a conjugated phenylethynyl group at the C5 position of 4TU. This modification significantly red-shifts the absorption spectrum, with the first absorption peak at 375 nm and a long tail reaching 420 nmone of the longest absorption wavelength observed among thiopyrimidine derivatives. Intriguingly, 5ph-4TU emits bright fluorescence at 450 nm with a quantum yield (Φ_FL) of ∼0.09, a rare property for nucleosides. , A further study of laser flash photolysis revealed that 5ph-4TU generates a long-lived triplet state with a quantum yield (Φ_T) of 0.55 at 355 nm excitation, a little inferior to its predecessor 4TU (0.9 ± 0.1). Nontheless, TDDFT calculations predicted that 5ph-4TU possesses a strong two-photon absorption (TPA) band at 476 nm with a cross-section of ∼400 GM, further red-shifted compared to single-photon peak wavelength, suggesting that it could even enable deeper tissue penetration by TPAan advantage not found in conventional thio-nucleosides.

Although the introduction of a phenylethynyl group at the C5-position of thiopyrimidine dramatically enhances the longer-wavelength photon absorption, this modification will inevitably regulate the excited-state process as well. How would C5-phenylethynyl substitution of thio-nucleoside manipulate the earliest excited-state events upon single-photon or two-photon excitation and bring about unique photophysical and photochemical characters inaccessible for conventional nucleosides? Besides their practical applications, 5ph-4TU provides an excellent model system of dual-functionality substitution of DNA/RNA nucleobases, greatly extending the scope of the current research on the single functionality substitution (thio-nucleosides etc.) in the literature.

Herein, by means of ultrafast time-resolved spectroscopy and the QM­(MS-CASPT2//CASSCF)/MM calculations, we decipher the critical tuning effects of C5-phenylethynyl (as in 5ph-4TU) on shaping the excited-state decay pathways and two-photon photoactive properties of thio-nucleoside. Our findings reveal that the C5-phenylethynyl group effectively suppresses the C5C6 bond twistinga critical vibrational motion repsonsible for singlet-state conical intersectionsthereby elevating the energy barrier (ca. 0.8 eV) from the S₂ (¹ππ*)_min to access the S₂ (¹ππ*)/S₁ (¹nπ*) conical intersection. Consequently the IC channel from S₂ (¹ππ*) to S₁ (¹nπ*) is hindered, exhibiting a prolonged S₂ (¹ππ*) lifetime (∼600 ps in CH₃CN) and significantly enhanced S₂ (¹ππ*) fluorescence emitting quantum yield (Φ_FL = 0.11 ± 0.03). Aside from emitting fluorescence, depopulation to S₁ (¹nπ*) minimum still occurs, which is followed by instantaneous intersystem crossing to a triplet manifold since there exists a triple S₁/T₂/T₁ quasi-degeneracy region. Moreover, we demonstrate that 5ph-4TU exhibits significant two-photon absorption even to near-infrared (NIR) region and retains the same excited-state behavior under two-photon excitation, making it a promising candidate for PDT applications under NIR. In conjunction with cellular imaging experiments and MTT assays, we confirm that 5ph-4TU possesses excellent biocompatibility and effectively generates ¹O₂ upon 760–770 nm two-photon excitation, leading to cancer cell death. From ultrafast and mechanistic perspective, these findings for 5ph-4TU disclose for the first time the C5-π-conjugated functionality in synergy with thionation tuning dramatically the excited state electronic structure and dynamics, paving the way for the development of next-generation biophotosensitizers that integrate bright fluorescence, efficient triplet-state formation, and strong near-infrared two-photon photoactivitycharacteristics challenging to achieve in one-molecule, but essential to potentially endow 5ph-4TU a promising candidate as a dual-function biophotosensitizer for imaging-guided cancer therapy.

Materials and Methods

Materials and Chemical Synthesis

3′,5′-Di-O-acetyl-5-phenylethynyl-4-thio-2′-deoxyuridine (5ph-4TU, >97%) was synthesized according to the previous literature. Due to the high reactivity of the hydrogen on the hydroxyl group of sugar moiety, it was protected with an acetyl group during the synthesis process. The ¹H NMR spectra for the synthesized compounds were shown in Figure S1. The other compound for comparison (uridine, 4-thiouridine) were purchased from Sigma-Aldrich. The DNA-specific dye DAPI and PBS were both purchased from Beijing Solarbio Science & Technology Co., Ltd. Commercially available reagent acetonitrile (ACN, Sigma-Aldrich, 99.9%) and dioxane (Sigma-Aldrich, 99.9%) were used without further purification. Notably, the mixed solvents ACN-H₂O (v/v, 1:1) and ACN-H₂O (v/v, 1:2) have been applied to investigate the photoproperties of 5ph-4TU in more polar solvents. This is due to the limited water-solubility of 5ph-4TU, which has an acetylated hydroxyl group on the sugar moiety.

Femtosecond Transient Absorption Spectra (fs-TA)

Femtosecond transient absorption spectra (fs-TA) were measured using a commercial femtosecond pump–probe spectrometer (Harpia-TA, light conversion). The fundamental pulses (800 nm, 40 fs, 1 mJ, 1 kHz) were generated with a Ti:Sapphire laser system (Coherent Astrella) and split by a beam splitter to generate the pump and probe beams. A small portion of the fundamental pulses was sent to a computer-controlled optical delay line and then focused onto a CaF₂ plate to generate the white light continuum probe beam. The rest of the fundamental pulses were sent to the optical parametric amplifier (TOPAS-C, Coherent Inc.) to generate the 350 nm/400 nm/800 nm pump beam, which was chopped at 500 Hz. To avoid anisotropic signals, the polarizations of pump and probe beams were set to the magic angle (54.7°). The focused probe and pump beams were overlapped into a fused silica cuvette with a 1 mm beam path length. The frequent movement of the electrically driven sample holder ensures that the sample in the probed volume was continuously renewed to avoid re-excitation of the excited volume by successive laser pulses, similar to the function of a magnetic stirrer. The absorption spectra and fluorescence spectra of the solutions before and after the fs-laser experiments were consistent with each other, indicating no occurrence of photobleaching. The OD of 5ph-4TU solution (500 μM) at 350 nm was ∼0.5. All experiments were carried out at room temperature. The differential absorbance ΔA (t, λ) obtained by femtosecond transient absorption spectroscopy (fs-TA) as a function of wavelength and time delay was corrected for the frequency-chirp and then analyzed using the Glotaran program.

Time-Resolved Fluorescence Spectroscopy (TRF)

Time-resolved fluorescence spectroscopy (TRF) was carried out using a high-resolution streak camera system: Universal Streak Camera (C10910-05, Hamamatsu) combined with a CMOS camera (C13440-20CU) and a spectrometer (HRS-300-S). The camera had two time sweep regimes: fast (minimum duration, 70 ps) and slow (minimum duration, 1 ns). In our experiments, the time-resolved fluorescence spectra and dynamics on the 5 ns time scales were collected through the slow sweep regimes. The fs 350 nm (∼500 nJ) and 800 nm (∼1 μJ) pump beam were generated from the optical parametric amplifier (TOPAS-C, Coherent Inc.) for single-photon excitation and two-photon excitation, respectively. The pump beam (0.6 cm beam diameter) was gently focused into the sample using a lens with a focal length of 21 cm. The sample solutions were in a 1 mm fused silica cuvette, and the sample cell was continuously moved.

Time-Resolved Laser Flash Photolysis

Nanosecond transient absorption spectra were measured using a flash photolysis setup, an Edinburgh LP980 spectrometer (Edinburgh Instruments Ltd.) combined with a Nd:YAG laser (Spectra-Physics Lab 170, Newport Corp.). Each measurement was performed in a 1 cm path length quartz cuvette at room temperature. The sample was excited by a 355 nm laser pulse (1 Hz, 10 mJ/pulse/cm², fwhm ≈ 7 ns). The analyzing light came from a 150 W pulsed xenon lamp. A monochromator equipped with a photomultiplier to collect the spectral range from 300 to 800 nm was used to analyze the transient absorption spectra. The signals from the photomultiplier were displayed and recorded as a function of time on a 100 MHz (1.25 Gs/s sampling rate) oscilloscope (Tektronix, TDS 3012C), and the data were transferred to a personal computer. The data were analyzed using the online software of the LP980 spectrophotometer. The fitting quality was judged by weighted residuals and the reduced χ² value.

Steady-State UV Absorption Spectra

Steady-state UV absorption spectra were measured using a double-beam spectrometer (Model U-3010, Hitachi). Each measurement was collected from 600 to 200 nm at a scan rate of 600 nm/min. The spectrum of a blank sample containing only ultrapure water or acetonitrile was used as the background and was subtracted from the averaged spectra. All measurements were performed in 1 cm quartz cuvettes at an ambient temperature.

Steady-State Emission Spectra

Steady-state emission spectra were recorded using a spectrofluorometer (F4500, HITACHI) in 1 cm path length quartz cuvettes at ambient temperature.

Singlet Oxygen Phosphorescence Detection

The characteristic 1275 nm phosphorescence emission of ¹O₂ was measured by using an Edinburgh Instruments FLS980 steady-state and transient fluorescence spectrometer equipped with a 450 W xenon lamp. All experiments were conducted under ambient temperature and pressure conditions, employing a liquid-nitrogen-cooled near-infrared photomultiplier tube (NIR PMT) as the detector. Measurements were performed using standard 1 cm path length quartz cuvettes in a direct excitation–emission geometry configuration. Spectral scans were acquired from 1200 to 1400 nm with 1 nm increments and a 0.1 s dwell time.

Cell Culture

A human cervical cancer cell line (HeLa) was cultured in a humidified incubator of 5% CO₂ at 37 °C in a standard culturing/growth medium (McCoy’s 5A medium containing 10% fetal bovine serum and 1% antibiotics and antimycotics). After the cell growth density reached more than 80%, the cells were spread on a 24-well plate. A 1 mM stock solution of 5ph-4TU was prepared in DMSO and was diluted in PBS as needed in the cellular experiment. HeLa cells were treated for 24 h in the dark with 50 μM 5ph-4TU to ensure the cell uptake. As a nucleobase analog, 5ph-4TU is also expected to be metabolically incorporated into DNA in a manner similar to 6TG or 4TU. , Subsequent to the drug treatment, the cells were washed three times with PBS (pH 7.4) to remove the free 5ph-4TU and medium solution. In the imaging studies, the cells were incubated with Singlet Oxygen Sensor Green (SOSG) solution (10 μM) in the dark for 2 h prior to the experiment. Currently SOSG has been successfully used as a specific probe for detecting intracellular ¹O₂. In this procedure, the cells were first washed three times with standard maintenance medium (SMM) and under the irradiation of an unfocused external blue light LED (λ_max = 390–400 nm, 150 mW/cm²) or red light LED (λ_max = 760–770 nm, 150 mW/cm²) light source for 20 min, respectively. After irradiation, SMM medium was removed. We also set up three control groups: cells treated (i) without 5ph-4TU and without irradiation, (ii) without 5ph-4TU, and (iii) without irradiation. These procedures ensured sufficient cellular uptake of SOSG. Then, these steps were followed: add methanol and fix cells at −20 °C for 10 min; rinse with PBS buffer three times; add ×2000 DAPI dye; rinse with PBS buffer twice; sealing piece. The blue fluorescence of DAPI (λ_max = 488 nm, under 405 nm excitation) and green fluorescence (λ_max = 525 nm, under 488 nm excitation) of the reaction product between SOSG and singlet oxygen ¹O₂ were then monitored separately using a Laser Scanning Confocal Microscope (Nikon A1-DUT, Japan). ImageJ 1.8.0 software was used to measure the mean luminescence intensity.

Cell-Viability Assay

The study of the effect of 5ph-4TU on the viability of cells was carried out using a methylthiazolyldiphenyl-tetrazolium bromide (MTT) assay. HeLa cells growing in the log phase were seeded into 96-well plates (ca. 1 × 10⁴ cells per well) and allowed to adhere for 24 h. 5ph-4TU (200 μL per well) at a concentration from 10 μM to 500 μM was added into the wells of the treatment group and 200 μL per well of DMSO diluted in DMEM at a final concentration of 0.2% was added to the negative control group. The cells were incubated with 5ph-4TU for 10 h at 37 °C under 5% CO₂, which were irradiated for 20 min with an unfocused external blue light LED (λ_max = 390–400 nm, 150 mW/cm²) or red light LED (λ_max = 760–770 nm, 150 mW/cm²), then MTT (5 mg/mL in phosphate buffer solution) was added into each well. After 4 h of incubation at 37 °C, 200 μL DMSO was added to dissolve the purple crystals. After 20 min of incubation, the optical density readings at 492 nm were taken using a plate reader. The cytotoxicity experiment was repeated three times.

Theoretical Calculations

The 5ph-4TU molecule was constructed manually. It was then solvated in a spherical acetonitrile box with a radius of 30 Å (1853 acetonitrile molecules). This solvated system was relaxed by sequentially undergoing constrained molecular mechanics (MM) minimizations (2000 steps) and molecular dynamics runs (100 ps, MD). In MM minimizations and MD simulations, 5ph-4TU and acetonitrile molecules were described using the generalized Amber force field (GAFF). All these MM calculations were performed using the sander module in AMBER22. The final MD snapshot including the 5ph-4TU molecule and the acetonitrile molecules was taken as the starting structure for the following QM­(MS-CASPT2//CASSCF)/MM computation. In all QM/MM calculation, the MM region including all the acetonitrile molecules was described using the force field used in the MD simulation mentioned above. The QM region of the investigated system was only the 5ph-4TU molecule. The geometries of 5ph-4TU were optimized in the five lowest electronic states (S₀, S₁, S₂, T₁ and T₂, shown in Figure and Table S2), by use of the complete active space self-consistent field (CASSCF) method, with the active spaces of 12 electrons in 10 orbitals (sulfur lone pair, five π and four π* orbitals as shown in Figure S12). A larger active space yields the same relative order of bright and dark states as the current result, and the energy is also similar, as shown in Tables S3 and S4. All the excited minimum structures were optimized starting from FC regions. An imaginary level shift of 0.2 au was used to avoid the intruder-state problem. The single-point energies of all CASSCF optimized structures were obtained at the multistate second-order perturbation theory (MS-CASPT2) level, and the relative energies of involved states for all stationary points are shown in Table . The basis sets of ANO-L-VDZP are applied in the CASSCF and MS-CASPT2 calculations. All of the optimization calculations at the CASSCF level and the single-point energy calculations at the MS-CASPT2 level were carried out by using the OpenMolcas22.02 package interfaced with the TINKER6.3.2 package. The definition of the key dihedral angles of 5ph-4TU is shown in Figure .

4.

CASSCF­(12, 10)/ANO-L-VDZP optimized minimum-energy as well as CI structures of 5ph-4TU. Also shown are the selected bond lengths (in Å) and the dihedral angle between pyrimidine and benzene ring.

1. Experimental and Calculated UV-Vis Absorption as well as Emission Properties of 5ph-4TU in ACN .

States	Calcd (eV)	Exp (in ACN)	f (SOC)
S1 (1nπ*)FC	2.87		0.0003
S2 (1ππ*)FC	3.52	3.31 (λabs = 375 nm)	0.5628
S1 (1nπ*)min	2.55		0.0002
S2 (1ππ*)min	3.11
	3.24 (in H2O)
S2 (1ππ*)FL	2.88	2.76 (λem = 450 nm)	0.3122
T1 (3nπ*)min	2.52
T2 (3ππ*)min	2.52
S2/S1 CI	3.90/3.81
	5.90/5.84 (in H2O)
S1/T2/T1	2.53/2.53/2.51		S1/T2 (123 cm–1)
S1/S0	3.46/3.43		S1/T1 (60 cm–1)

^aAlso shown are adiabatic excitation energies of optimized excited state minima and intersection structures, as well as calculated oscillator strengths (f) or spin-orbit coupling (SOC) values.

^bS₂ (¹ππ*)_min is referred to adiabatic energy at the minimum point of the S₂ state, which represents the energy difference between S_2,min and S_0,min.

^cS₂ (¹ππ*)_FL is referred to the fluorescence emission energy, which represents the vertical energy difference between S_2,min and a specific vibrational level of S₀.

In order to simulate the excited-state absorption spectra of the S_2FC, S_2 min, and T₁ states of 5ph-4TU, we employed the TD-B3LYP/PCM/6-311G** method based on the CASSCF-optimized geometries. This hybrid approach leverages the structural accuracy of multiconfigurational methods while utilizing the computational efficiency of TD-DFT for calculating higher-lying transitions, providing a qualitative yet reliable interpretation of the experimental spectral features.

Results and Discussion

Steady-State Spectral Properties and Excited State Electronic Structure Calculations

Figure a shows that the phenylethynyl substitution has striking effect on the steady-state spectra of 5ph-4TU, in comparison to those for 4TU and uracil (U). Contrary to the canonical uracil which only absorbs strongly in the UVC (100–280 nm) region of the electromagnetic spectrum, sulfur substitution of the C4 carbonyl in uracil results in a significant red-shift in the absorption maximum. The lowest-energy absorption of 4TU has a maximum at 330 nm with an extinction coefficient of (1.7 × 10⁴) M^–1 cm^–1 in acetonitrile (ACN), whereas 5ph-4TU exhibits an absorption maximum at 375 nm with an extinction coefficient of (0.7 × 10⁴) M^–1 cm^–1. It reveals that the addition of the phenylethynyl substituent at the 5-position in 5ph-4TU further red-shifts the ground-state absorption spectrum, which could be ascribed to the extended π-conjugation of the molecule. Specifically, the absorption band of 5ph-4TU extends into the visible region, with a ε value of ∼4000 M^–1 cm^–1 at 400 nm on the red side (Figure a), which allows for selective excitation at longer wavelength. The UV–vis absorption spectrum of 5ph-4TU in different solvents with increasing polarity (Dioxane < ACN < ACN-H₂O = 1:1 < ACN-H₂O = 1:2) are also presented in Figure S2, exhibiting no noticeable absorbance changes at 375 nm.

1.

(a) UV–vis absorption spectra of 5-phenylethynyl-4-thiouridine (5ph-4TU, red), 4-thiouracil (4TU, blue) and uracil (U, black) in acetonitrile (ACN) at room temperature (50 μM, path length: 1 cm). (b) Fluorescence spectra of 5ph-4TU in different solvents (Dioxane, ACN, ACN-H₂O = 1:1, ACN-H₂O = 1:2) and 4TU in ACN. The excitation wavelength was set at 350 nm for 5ph-4TU and 330 nm for 4TU. (c) Molecular orbitals involved in transitions to the first and second excited states of 5ph-4TU by using CASSCF (12, 10)/ANO-L-VDZP. (d) Phosphorescence of ¹O₂ peaked at 1275 nm observed in the emission spectra of PN and 5ph-4TU solution under air-saturated conditions after 355 nm excitation (A₃₅₅ = 0.4). To ensure time-resolved detection of ¹O₂, the solvent D₂O is used since the lifetime of ¹O₂ in D₂O is much longer than that in H₂O.

In the emission spectra (Figure b), 4TU produces no emission signal upon exciting its maximum absorption at 330 nm, revealing that 4TU has strikingly low quantum yields of fluorescence (Φ_FL), in accordance with previous results on 4TU of near-unity triplet yields. In contrast, 5ph-4TU emits bright fluorescence emission at 450 nm in ACN upon 350 nm excitation, and the Φ_FL is further determined to be 0.11 ± 0.03 using quinine bisulfate as a reference standard. This value is consistent with the reported Φ_FL of 0.09 ± 0.01. The emission is further confirmed to be fluorescence instead of phosphorescence by a deoxygenation experiment (Figure S3), in agreement with a previous study that no phosphorescence of 5ph-4TU was observed even at 77 K. Notably, the Φ_FL of 5ph-4TU was found to be substantially higher than those of other thio-nucleosides, which is a rare photoproperty for nucleosides. Additionally, it can be observed that the emission intensity is clearly enhanced by changing the solvent from nonpolar dioxane to polar solvent mixture ACN-H₂O = 1:2, accompanied by typical red-shift of about 10 nm. Quantitatively, the Φ_FL in dioxane, ACN-H₂O = 1:1 and ACN-H₂O = 1:2 are 0.09 ± 0.02, 0.13 ± 0.02 and 0.14 ± 0.03, respectively, exhibiting an enhancement with increasing solvent polarity. The above spectral difference between 4TU and 5ph-4TU suggests that the photophysical pathways in the excited state have been altered dramatically by C5-phenylethynyl substitution.

To aid the interpretation of the above steady-state and the following ultrafast spectra, high-level electronic structure calculations were performed for 5ph-4TU. The ground state (S₀) and five low-lying excited state minima are optimized using the CASSCF (12, 10)/ANO-L-VDZP, and all optimized structures are shown in Figure . Further, the MS-SA5-CASPT2 (12, 10)/ANO-L-VDZP calculations were employed to obtain energies of the vertical absorption and excited state minima by state-average of five roots with equal weights. The vertical absorption energies and oscillator strength (f) values as well as minima energies for the five lowest-lying singlet and triplet electronic states of 5ph-4TU are collected in Table , as well as corresponding transition orbitals in Figure c. As we can notice, the S₁ (¹nπ*) state is 2.87 eV vertically above the ground state with negligible oscillator strength (f = 0.0003), being derived from the n_s lone pair orbital (HOMO) to the π* orbital (LUMO). The S₂ state has ππ* character with f computed to be 0.56, corresponding to HOMO–1 → LUMO transition, which is directly populated in the Franck–Condon region (¹ππ*)_FC. In fact, 5ph-4TU exhibits an absorption band with a maximum at 375 nm (3.31 eV) in the UV–vis absorption spectrum (Figure a), which is in agreement with the calculated S₂ (¹ππ*)_FC energy (3.52 eV). This energy is lower than the S₂ (¹ππ*)_FC of 4TU (3.95 eV), explaining the red-shift of absorption maxima of 5ph-4TU relative to 4TU. Moreover, the S₁ (¹nπ*) state is a dark state with calculated emission oscillator strength value f = 0.0002, while the calculated emission f is 0.3122 for S₂ (¹ππ*)_min state. In fact, the emission wavelength of S₂ (¹ππ*)_min is calculated to be 431 nm (2.88 eV), which is in good agreement with the experimental fluorescence peak wavelength of 450 nm (2.76 eV), indicating the fluorescence emission should originate from the S₂ (¹ππ*)_min state.

Noticeably, the lowest S₁ and S₂ excited states involve orbital transitions predominantly in the 4TU moiety, with only a small portion associated with the phenylethynyl group (Table S2). However, the phenylethynyl group may participate in the orbital transitions related to the higher excited state. As shown in Figure a, comparison of the steady-state absorption spectra reveals an important difference between 4TU and 5ph-4TU. 4TU exhibits main absorption peaks at 330 and 245 nm, and these two bands are red-shifted for 5ph-4TU (375 and 275 nm). In addition, an additional absorption band emerges at 310 nm in 5ph-4TU. This new absorption feature should correspond to a higher excited state involving the phenylethynyl group, given that the S₂ state due to the 4TU moiety excitation accounts for the first absorption peak at 375 nm.

Time-Resolved Fluorescence Spectroscopy Characterizing Fluorescence Decay Dynamics

To examine the evolving photophysical dynamics of the emissive state, time-resolved fluorescence (TRF) spectra of 5ph-4TU were measured in various solvents after excitation at 350 nm. Figure a,d–f show the TRF spectrum of 5ph-4TU in different solvents with increasing polarity (Dioxane < ACN < ACN-H₂O = 1:1 < ACN-H₂O = 1:2), obtained after 350 nm excitation. In ACN (Figure a), a fluorescence emission peak at 450 nm was clearly observed, which is consistent with the steady-state fluorescence spectra. The fluorescence decay profile in ACN is shown in Figure b, which can be fitted by a monoexponential decay kinetics, indicating that there is only one luminescent state, which should originate from the bright state S₂ (¹ππ*)_min. Fluorescence decay profiles in different solvents are compared in Figure c. Notably, the fitted fluorescence lifetimes show obvious solvent-dependence, exhibiting a longer lifetime with the increase of solvent polarity (544 ± 12 ps in Dioxane, 599 ± 10 ps in ACN, 910 ± 18 ps in ACN-H₂O = 1:1, 950 ± 16 ps in ACN-H₂O = 1:2) and a red-shift of about 10 nm. The increase of fluorescence lifetime with solvent polarity rationalizes well the steady-state emission intensity and Φ_FL measured above.

2.

Three-dimensional time-resolved fluorescence spectra of 5ph-4TU (500 μM, path length: 1 mm) upon 350 nm fs laser excitation, in different solvents: (a) Dioxane, (d) ACN, (e) ACN-H₂O = 1:1 and (f) ACN-H₂O = 1:2. (b) Fluorescence decay profiles (circles) and fits (lines) in ACN. (c). Comparative fluorescence decay profiles in different solvents.

To understand why phenylethynyl substitution results in much longer lifetimes and strong fluorescence for 5ph-4TU, some pertinent observations and facts are taken into account. (i) It has been shown that the barrierless nature between ¹ππ* state and the ¹ππ*/S₀ conical intersections (CIs) are key to ultrafast excited-state decay (subpicosecond time scales) in nucleic acid pyrimidine bases, wherein the pyrimidine ring deformation induced by C5C6 bond twisting can promote internal conversion (IC) of the ¹ππ* state. ,− Notably, C5 modifications can hinder ring distortion, increasing the energetic barrier along the ¹ππ*-to-S₀ IC pathway and thereby leading to longer ¹ππ* lifetimes. For example, when the hydrogen atom at the C5 position of cytosine is replaced by a methyl group or a fluorine groupforming 5-methylcytosine (5mC) and 5-fluorocytosine (5FC), respectivelyit is found that these substitutions at the C5 position can attenuate the torsional motion of the C5C6 bond and relevant out-of-plane vibration due to the increased mass relative to the H atom. This effect imposes an energy barrier for the ¹ππ*_min→¹ππ*/S₀ CI transition in 5mC (0.08 eV) and in 5FC (0.15 eV), ultimately prolonging the excited-state lifetimes of ¹ππ* state (7.6 ps for 5mC; 40 ps for 5FC) compared to that of cytosine (1.5 ps) in aqueous solution. − (ii) Analogously, the ab initio quantum-chemical study on the deactivation processes undertaken by uracil (U), indicated deformation of the C5C6 bond as well as out-of-plane vibration of the H5 atom is the dominant mechanism leading to ¹ππ* to CIs transition. Whereas, methyl (5-methyluracil, 5mU) and fluorine (5-fluorouracil, 5-FU) substitution at C5 position both can reduce the C5C6 bond twisting, suggesting that accessing the ¹ππ*/S₀ CI from ¹ππ*_min requires to overcome a larger energetic barrier and ultimately results in elongated excited-state lifetime in contrast to U. , (iii) For thiobases such as 4-thiouracil (4TU), the C4-thio substitution enforces a planar pyrimidine ring geometry in both the S₂ (¹ππ*)_min state and the S₂ (¹ππ*)/S₁ (¹nπ*) CI. This structural similarity results in an exceptionally low energy barrier for the S₂ (¹ππ*)_min → S₂ (¹ππ*)/S₁ (¹nπ*) CI transition, ultimately yielding an ultrafast S₂ (¹ππ*) lifetime of ∼76 fs. Interestingly, methylation at the C5 position of 4-thiouracil (forming 4-thiothymidine, 4-TT) displays distinct geometric configurations: both planar and twisted conformations were identified for the S₂ (¹ππ*)_min state. The planar configuration S_2P is energetically favored, whereas the S₂ (¹ππ*)/S₁ (¹nπ*) CI adopts a twisted geometry involving C5C6 bond torsion. Consequently, an activation energy is required to reach the S₂/S₁ CI from the S_2P, leading to the relatively longer-lived S₂ (¹ππ*) state (3.6 ps) and explaining a weak fluorescence from S₂ state.

As discussed above, ring deformations initiated by C5C6 bond twisting as well as out-of-plane vibration is responsible for the fast access to CIs. However, the substitution at C5 position can restrict torsion about C5C6 bond and weaken out-of-plane vibrations, which hinders the internal conversion (IC) between the ¹ππ* and ¹nπ* or S₀ states and will ultimately prolong the lifetime of ¹ππ* state as well as increases radiation probability from it. These past studies give us a hint that for 5ph-4TU, the π-extended functionality of phenylethynyl substituted at C5 position, being more massive and rigid than the methyl group in 4TT, may impose stronger steric constraints that suppress C5C6 bond twisting of the pyrimidine ring. This enhanced restriction is expected to hinder effectively the IC channel from S₂ (¹ππ*) to S₁ (¹nπ*), thereby significantly prolonging the ¹ππ* excited-state lifetime and causing efficient emitting behavior for 5ph-4TU. Moreover, the phenylethynyl substituent at C5 atom of the pyrimidine ring can increase the rigidity of the structure, which is another reason for the enhanced fluorescence. The restriction of intramolecular motions is commonly adopted to promote fluorescence emitting, since flexible molecular motions usually favor nonradiative decay along with the transformation of excited-state energy to other forms such as thermal energy. These analyses help us conclude on the prevailing role played by the phenylethynyl group, yielding an emissive long-lived S₂ (¹ππ*) state along with increased Φ_FL.

Femtosecond Transient Absorption Spectroscopy Monitoring The Excited State Relaxation Dynamics

The above time-resolved fluorescence spectroscopy provides neat information related to the emissive state. To further obtain the holistic excited-state relaxation picture for 5ph-4TU, the femtosecond transient absorption (fs-TA) spectroscopy measurements were carried out. Figure and Figure S4 show the fs-TA contour plots and transient absorption spectra at typical time delays of 5ph-4TU obtained after 350 nm irradiation in different solvents with increasing polarity (Dioxane < ACN < ACN-H₂O = 1:1 < ACN-H₂O = 1:2). For 5ph-4TU in ACN (Figures a,d–f), immediately after 350 nm excitation, a negative band at 375 nm is observed, which should be ascribed to the depletion of ground state molecules (GSB), consistent with the steady-state absorption spectrum (Figure a). Simultaneously, an intense positive band centered at 460 nm is developed along with a featureless broad band in the 500–650 nm range, which should correspond to the excited state absorption (ESA) of the initially populated S₂ (¹ππ*)_FC. In the first 12 ps, the ESA band of S₂ (¹ππ*)_FC at 460 nm decays gradually and shows an ∼15 nm blue shift to 445 nm, with the simultaneous buildup of two new ESA absorption bands centered at 550 and 600 nm. This spectral evolution suggests the decay of S₂ (¹ππ*)_FC features, corresponding to the appearance of new excited species, as an intermediate state prior to forming T₁. Afterward, from 12 to 1800 ps, while the 445, 550 and 600 nm bands progressively decrease in intensity, two new ESA peaks at 390 and 470 nm appear accordingly, with a long-tail featureless absorption extending to 700 nm. Two isosbestic points are clearly observed at 410 and 520 nm, indicating a state to state transition as well. These eventually formed excited-state features then do not show any further evolution from 1800 ps until ∼7 ns, which are ascribed to the triplet–triplet state absorption bands of the long-lived T₁ state of 5ph-4TU, as can be confirmed by nanosecond laser flash photolysis (Supporting Information (SI), Figure S5). The observed absorption features for the T₁ state also agree well with the literature. Additionally, the triplet quantum yields (Φ_T) are determined by comparing the initial transient absorption intensities of 5ph-4TU triplet states at 470 nm across different solvents under identical conditions (Figure S6a), using the reported Φ_T of 0.55 in ACN as a reference standard. Our TD-DFT calculations show that the first T₁ state absorption band exhibits comparable oscillator strengths in ACN and H₂O (Table S5), suggesting that the variations of the triplet absorption coefficient (ε_T) in different solvents are likely negligible. Therefore, under the reasonable approximation that the absorption coefficient of ε_T in different solvents remains nearly unchanged, the Φ_T values are obtained to be 0.59 ± 0.01, 0.52 ± 0.03 and 0.49 ± 0.03 in dioxane, ACN-H₂O = 1:1 and ACN-H₂O = 1:2, respectively.

3.

Three-dimensional and femtosecond transient absorption spectra of 5ph-4TU (500 μM path length: 1 mm) upon 350 nm fs laser excitation in ACN (a, d–f). (b) Kinetic traces at typical wavelength (circles) and fits (lines) from global analysis. (c) Evolution-associated difference spectra (EADS) of the corresponding components obtained from global analysis of the fs-TA spectra for 5ph-4TU in ACN.

Following the efficient triplet formation of 5ph-4TU, triplet–triplet energy transfer with oxygen in solution can lead to ¹O₂. Therefore, using phenalenone (PN) as a reference standard (Φ_Δ= 0.98 in air-saturated ACN), singlet oxygen quantum yields (Φ_Δ) of 5ph-4TU under 355 nm excitation in different solvents (Dioxane, ACN, ACN-H₂O = 1:1 and ACN-H₂O = 1:2) were determined by detecting 1275 nm-centered ¹O₂ phosphorescence emission (Figure d), yielding values of 0.46 ± 0.03, 0.41 ± 0.02, 0.37 ± 0.02 and 0.35 ± 0.02, respectively. A slight decrease in Φ_Δ with increasing solvent polarity (from Dioxane to ACN-H₂O mixtures) aligns with the trend observed in our Φ_T measurements.

Kinetics traces corresponding to the fs-TA in Figure were analyzed globally with three time constants, and satisfactory fitting to experimental data is obtained (Figure b). The evolution-associated difference spectra (EADS) extracted from global analysis (Figure c) match nicely with the experimental spectral feature, indicating high fitting quality too. Table lists the fitted time constants in different solvents. As in ACN, the first time constant τ₁ (3.6 ± 0.4 ps) corresponds to the decay of S₂ (¹ππ*)_FC to the intermediate state, most likely a process of vibrational cooling to the S₂ (¹ππ*)_min as manifested by the blue-shift and narrowing of the ESA band during this period. This assignment can be further confirmed by the comparison with fs-TA upon 400 nm excitation, where τ₁ becomes faster (Figure S7 and Table S1) due to the lesser excess vibrational energy to be dissipated under lower-energy photon excitation. The third time constant τ₃ (>7 ns) corresponds to the decay of the long-lived T₁ state. The exact lifetime of the T₁ state can be further determined to be ∼12 μs by nanosecond laser flash photolysis in deoxygenated conditions (SI, Figure S5), in accordance with a report.

2. Fitted Time Constants from Time-Resolved Fluorescence Spectroscopy (Upper) and Femtosecond Transient Absorption Spectroscopy (Lower) for 5ph-4TU upon 350 nm Excitation in Solvents of Different Polarity.

	Dioxane	ACN	ACN-H2O = 1:1	ACN-H2O = 1:2
τ/ps	544 ± 12	599 ± 10	910 ± 18	950 ± 16
τ1/ps	4.0 ± 0.2	3.6 ± 0.4	3.4 ± 0.2	3.2 ± 0.1
τ2/ps	553 ± 10	600 ± 15	915 ± 14	964 ± 19
τ3/ns	>7	>7	>7	>7

The most intriguing temporal component in 5ph-4TU is the second time constant τ₂ (600 ± 15 ps), which agrees well with the lifetime of the S₂ (¹ππ*)_min state (599 ± 10 ps) observed in the above TRF. Besides, τ₂ also exhibits an increase of lifetime with solvent polarity, and the obtained time constants in different solvents are consistent with the results of TRF. Therefore, τ₂ represents the decay lifetime of S₂ (¹ππ*)_min. As can be clearly demonstrated by the representative kinetic traces, the decay of S₂ (¹ππ*)_min at 450 and 600 nm is concomitant with the buildup of T₁ at 390 nm, suggesting that the S₂ (¹ππ*)_min state decay corresponds to the eventual formation of the long-lived T₁ triplet state (Figure b).

To further support the assignments of the excited-state transient absorption species and relaxation mechanism, the absorption spectra of the S_2FC, S_2,min, and T₁ of 5ph-4TU were calculated using the TD-B3LYP/PCM/6-311G** method. As shown in Figure S14, the S_2FC state exhibits a prominent absorption band centered at 470 nm with relatively flat absorption extending from 550 to 700 nm. For the S_2,min state, the 470 nm absorption band undergoes a blue shift while simultaneously exhibiting an enhanced intensity at 600 nm. These features reproduce well the EADS spectra derived from the global fit of fs-TA experiments (Figure c). Moreover, the calculated spectra for the T₁ state agree with experiments, exhibiting strong absorption maxima at 390 nm and a flat absorption band with two maxima at 480 and 560 nm.

Characterizing the Excited State Potential Energy Surfaces

To further interpret the above experimental results, we mapped out the excited-state PESs (Scheme ) by high-level ab initio calculations. The minimum energy path can be derived on the basis of the minimum-energy structures in the lowest four electronically excited states, S₁ (¹nπ*), S₂ (¹ππ*), T₁ (³nπ*), and T₂ (¹ππ*), as well as relvant CIs (Figure ), obtained by MS-SA5-CASPT2 (12, 10)//CASSCF (12, 10)/ANO-L-VDZP calculations. Upon excitation at 350 nm, the S₂ (¹ππ*) state is directly populated at the Franck–Condon point, denoted as S₂ (¹ππ*)_FC, which is followed by vibrational relaxation to S₂ (¹ππ*)_min. Compared to S_0 min, the S₂ (¹ππ*)_min also features a mostly planar pyrimidine ring, and the major structural difference is the elongated C5C6 and C4S bonds. Interestingly, the calculation reveals that the S₂ (¹ππ*)/S₁ (¹nπ*) CI exhibits a distorted C5C6 bond, with a dihedral angle (C4C5C6N1) of 10.8°. Besides, the dihedral angle between pyrimidine and benzene ring exhibits a marked difference (∼5°) from all other optimized structures (∼30°). Therefore, a considerable degree of structural change is needed to access this CI from S₂ (¹ππ*)_min, suggesting that a large activation energy is required. Indeed, computational results reveal this barrier to be approximately 0.8 eV which rationalize well the observed prolongation of the S₂ (¹ππ*) state lifetime.

2. Schematic Description of the Main Excited State Local Minima and CIs of 5ph-4TU as Computed by MS-SA5-CASPT2 (12, 10)/ANO-L-VDZP .

^a Energy given in eV alongside.

It is worth noting that there is no amenable S₂/S₀ intersections found for the S₂ (¹ππ*) of 5ph-4TU. Combined with its prolonged lifetime of S₂ (¹ππ*), the chances of fluorescence emission are greatly increased for 5ph-4TU, explaining the observation of bright fluorescence for 5ph-4TU. The observed prolongation of the fluorescence lifetime from 553 ps in nonpolar dioxane to 964 ps in the more polar ACN-H₂O = 1:2 can be explained by the polarity-dependent elevation of the S₂/S₁ CI. As shown in Table , computational results reveal that while the energy of S₂(¹ππ*)_min remains largely unaffected by solvent polarity (3.24 eV in H₂O vs 3.11 eV in CH₃CN), the energy of S₂/S₁ CI is markedly elevated in H₂O (5.90/5.84 eV) compared to CH₃CN (3.90/3.81 eV). Consequently, the energy barrier for accessing the S₂/S₁ CI from S₂(¹ππ*)_min is increased to approximately 2.6 eV in H₂O, thereby further extending the S₂ lifetime.

As shown in Scheme , below the S₁ state, two energetically comparable triplet states (T₁ and T₂) are located, both lying geometrically and energetically very close to S₁ (¹nπ*)_min. This region of triple quasi-degeneracy can act as a gateway to the ISC populating the triplet manifold. Considering the ³nπ* character of T₁-min, ³ππ* character of T₂-min, and the El-Sayed propensity rules, direct ISC from the S₁ (¹nπ*) state to the T₁ (³nπ*) state is less likely which is a forbidden transition. Furthermore, the computed spin orbit coupling term (SOC) for S_1 min to T₁ (59.86 cm^–1) is small, whereas S_1,min to T₂ SOC (122.68 cm^–1) is rather large. Additionally, the ³ππ* nature of the T₂ state renders the S₂(¹ππ*) → T₂(³ππ*) ISC symmetry-forbidden, resulting in a considerably weaker SOC (4.0 cm^–1) compared to the S₁/T₂ SOC. These results suggest that ISC occurs preferentially from S₁ (¹nπ*_min) to T₂ (³ππ*_min), which is followed by depopulation to the energetically degenerate triplet T₁ (³nπ*_min). Given the energetic degeneracy of the T₁ (³nπ*) and T₂ (³ππ*) states, the finally populated triplet state likely possesses a mixed electronic character of both nπ* and ππ* configurations, rather than a pure nπ* state. Consequently, 5ph-4TU exhibits a relatively high singlet oxygen quantum yield (Φ_Δ = 0.38–0.47 in various solvents). Notably, because of the triple S₁/T₂/T₁ quasi-degeneracy region, extremely rapid relaxation from S₁ → T₂/T₁ state is expected; therefore the intermediate S₁ state has a lack of concentration accumulation and does not display any prominent absorption features in the fs-TA spectra. Overall, the excited state decay pathway is ascertained: S₂ (¹ππ*)_FC → S₂ (¹ππ*)_min → S₁ (¹nπ*)_min → T₂ (³ππ*)_min/T₁ (³nπ*)_min. This pathway is also verified by high-level MS-CASPT2 computations of the energy paths along the constructed linearly interpolated internal coordinate (LIIC), providing qualitative evidence for the energetic and dynamic connections between the various excited state minima and CIs involved in the deactivation processes (Figure S13). Moreover, our CASSCF//MS-CASPT2 calculations locate the S₁/S₀ CI lying at 3.46/3.43 eV (Table ), which is accessible from the S₁ (¹nπ*)_min state (2.55 eV) with a barrier of approximately 0.88 eV. Therefore, there should also exist a nonradiative decay pathway to the ground state, in addition to the fluorescence emission and ISC pathways. Considering the substantial energy barrier of 0.88 eV for IC to the ground state and the large S₁ → S₀ energy gap of 2.55 eV, nonradiative decay to the ground state is expected to be inefficient. In addition to fluorescence emission, ISC and nonradiative decay, there may also exist a photochemical pathways of 5ph-4TU (formation of photooxidation products was postulated). According to the transient absorption spectra of 5ph-4TU in air-saturated acetonitrile upon 355 nm excitation (Figure S5c), the ground state bleach is not completely recovered with time. Dividing the amplitude of the residual bleach by that of the maximum bleach signal at 310 nm, the yield of the photoproduct is roughly estimated to be ∼0.06, indicating a quite minor contribution of the photooxidation pathway.

Two-Photon Absorption Properties Characterization with Time-Resolved Fluorescence Spectroscopy

Previous TDDFT calculation predicted the two-photon absorption property of 5ph-4TU, with a two-photon absorption cross-section calculated to be ∼400 GM at 476 nm. Besides this peak wavelength of 476 nm, to explore whether there is still two-photon absorption in the near-infrared region, we measured the time-resolved fluorescence (TRF) intensity of 5ph-4TU under 800 nm excitation as a function of the incident laser power (Figure S8). The nonlinear relationship between the TRF intensity and the incident laser power indicates the occurrence of multiphoton absorption. For a simplified n-photon absorption process, the TRF signal intensity is proportional to the n-th power of the incident laser power. Therefore, from the log–log plot in Figure S8, the slope was determined to be 1.99, which reveals that the 800 nm excitation is attributed to the two-photon absorption process.

Furthermore, we measured the two-photon absorption cross-section of 5ph-4TU in CH₃CN at 800 nm using Rhodamine B (RB) as a reference. The value of two-photon absorption cross-section σ⁽²⁾ can be determined according to the equation (reference RB, σ_r ⁽²⁾ = 152 GM at 800 nm)­

$${\sigma }_s={\sigma }_r\times \frac{{\mathrm{\Phi }}_r}{{\mathrm{\Phi }}_s}\times \frac{C_r}{C_s}\times \frac{F_s}{F_r}$$ 1

where the subscripts of s and r represent the paraments of the sample (5ph-4TU) and the reference (RB), respectively. Φ is the fluorescence quantum yield; C is the concentration of the solution; F is the fluorescence intensity upon two-photon excitation measured by TRF here. The σ ⁽²⁾ value for 5ph-4TU in CH₃CN is obtained to be 44 GM at 800 nm, which is a substantial two-photon absorption cross-section for near-infrared light. We will then use 800 nm for the two-photon excitation experiments below. To our knowledge, this is the first experimental characterization of the two-photon photoproperties for 5ph-4TU.

Time-resolved fluorescence spectra of 5ph-4TU in ACN or ACN-H₂O = 1:1 under two-photon excitation were measured (Figure ). Basically, the fluorescence spectra and the fluorescence lifetimes by two-photon excitation are similar to those obtained by one-photon excitation, showing the emission at ∼450 nm with decay lifetimes of 590 and 900 ps in ACN and ACN-H₂O = 1:1, respectively. This observation suggests that the near-infrared 800 nm two-photon absorption (with an energy equal to a single 400 nm photon) also populates 5ph-4TU to S₂ (¹ππ*)_FC, which then decays to S₂ (¹ππ*)_min, emitting fluorescence. Meanwhile, partial population of S₂ (¹ππ*)_min undergoes IC to S₁ (¹nπ*)_min, and ISC occurs through this doorway state, forming the lowest triplet state as has been observed in the above fs-TA spectra under single-photon excitation. Due to the much lower detection sensitivity of fs-TA compared to time-resolved fluorescence spectroscopy, it is not applicable to use fs-TA to trace the weak excited state absorption signals of triplet formation upon two-photon excitation. However, the triplet formation under 800 nm two-photon excitation can be proved indirectly by detecting the triplet induced ¹O₂ formation in the steady-state experiments below.

5.

Three-dimensional time-resolved fluorescence spectra of 5ph-4TU (500 μM, path length: 1 mm) upon two-photon excitation at 800 nm, in different solvents: (a) ACN and (b) ACN-H₂O = 1:1. (c) Fluorescence decay profiles (circles) and fits (lines) in two different solvents.

In Cell Singlet Oxygen Detection and 5ph-4TU Imaging with Confocal Laser Scanning Microscopy

The two-photon absorption performance of 5ph-4TU makes it an attractive candidate as two-photon PDT agents. To investigate its potential application in photodynamic therapy, we further performed cell imaging experiments in combination with MTT (methylthiazolyldiphenyl-tetrazolium bromide) assays. First, confocal microscopy experiments were performed for the HeLa cells incubated with 150 μM 5ph-4TU alone. As shown in Figure S9, the intrinsic blue fluorescence of 5ph-4TU under 405 nm excitation is clearly seen. Imaging was performed using a Nikon confocal microscope (Model:A1-DUT) equipped with differential interference contrast (DIC). By converting refractive index gradients into contrast, DIC significantly enhances the edges of highly transparent cells and organelles of the bright-field image, allowing for clear visualization of the circular nuclear structures without staining. Therefore, merge of the bright-field image with the fluorescence image (Figure S9) reveal that the blue fluorescence of 5ph-4TU is predominantly localized in the nucleus, with a minor distribution in the cytoplasm. These results provide direct evidence of the cell imaging capability of 5ph-4TU and its subcellular distribution.

Second, the ¹O₂ formation and cytotoxicity induced by 5ph-4TU under respective one-photon 390–400 nm and 760–770 nm two-photon irradiation were demonstrated in HeLa cells using confocal microscopy, as shown in Figure . The Singlet Oxygen Sensor Green (SOSG), which can selectively react with ¹O₂ to form a highly bright green fluorescent product SOSG-EP emitting at 525 nm, was used to testify whether the phototoxic species ¹O₂ could be generated following the efficient triplet formation of 5ph-4TU. Moreover, we performed colocalization experiments using 4′,6′-diamidino-2-phenylindole (DAPI), a commonly used nuclear counterstain in fluorescence microscopy, which emits blue fluorescence at 488 nm representing the position of cell nucleus. The HeLa cells were incubated in cell culture medium with 50 μM 5ph-4TU for 20 min before the imaging experiments. As shown in Figure d, intensive green fluorescence signals are observed inside the cytoplasm and nucleus for cells treated with 5ph-4TU after single-photon irradiation at 390–400 nm, while no fluorescence is seen in extracellular spaces. This result also implies that 5ph-4TU exhibits suitable plasma membrane permeability and could enter into the cells simply by incubation. By contrast, no fluorescence was observed in the control groups (no 5ph-4TU treated and no irradiation in Figure a; only 50 μM 5ph-4TU-treated in Figure b; only light-irradiated for 20 min in Figure c), confirming that 5ph-4TU can also maintain its good optical properties to efficiently generate a triplet state which forms ¹O₂ in a cellular environment upon single-photon irradiation at 390–400 nm. Notably, the strength of ¹O₂ signals for 5ph-TU treated Hela cells under 760–770 nm two-photon irradiation (Figure f) is inferior than that produced via 390–400 nm single-photon excitation, consistent with the relatively small two-photon-absorption cross-section in near-infrared region as addressed above. Nonetheless, the fluorescence microscopy images indeed demonstrate the prominent formation of ¹O₂ due to the 760–770 nm two-photon irradiation of cellular 5ph-4TU, which provides key basis for the potential usage of 5ph-4TU as a near-infrared two-photon PDT agent. A separate power dependence experiment was performed, measuring the ¹O₂ formation in 50 μM 5ph-4TU solution (mixed with 10 μM SOSG) under the same 760–770 nm LED irradiation conditions of cellular assays in Figure f. The 525 nm fluorescence intensity of SOSG-EP (¹O₂ oxidation product) was recorded, reflecting the ¹O₂ yield. As shown in Figure S10, the ¹O₂ yields resulting from 760 to 770 nm LED irradiation of 5ph-4TU show a nonlinear increase with the increased LED power and the log–log plot has a slope of 1.93. These results provide evidence supporting the two-photon absorption mechanism of 5ph-4TU to generate cytotoxic ¹O₂ under our 760–770 nm LED illumination conditions. Although CW LED light-induced two-photon excitation is inefficient, it remains experimentally observable, , possibly due to the superior near-infrared penetration depth and reduced scattering to facilitate photon delivery and utilization within solution and biological media. Especially for 5ph-4TU, the near-infrared two-photon excitation also triggers fluorescence emission at ∼450 nm, which can be reabsorbed by 5ph-4TU under two-photon excitation with a significantly increased TPA cross section (a strong TPA band predicted at 476 nm with σ ⁽²⁾ ∼ 400 GM). As such, the TPA effect should be greatly amplified, resulting in the robust TPA-induced PDT performance of 5ph-4TU.

6.

Confocal fluorescence image of 5ph-4TU-treated (50 μM) or nontreated HeLa cells stained with SOSG (10 μM) or DAPI (5 μg/mL) upon single-photon of 390–400 nm (d) or two-photon of 760–770 nm (f) irradiation for 20 min. The control groups: (a) no 5ph-4TU treated and no irradiation; (b) only 50 μM 5ph-4TU-treated; (c) only 390–400 nm-irradiated for 20 min and (e) only 760–770 nm irradiated for 20 min. An MTT assay of HeLa cells treated with different concentrations of 5ph-4TU under single-photon at 390–400 nm (g) or 760–770 nm two-photon irradiation (h) for 20 min, as well as control groups (i): untreated cells, cells irradiated at 390–400 nm or 760–770 nm alone, and cells treated with 100 μM 5ph-4TU in the dark.

Furthermore, the photocytotoxicity of 5ph-4TU was evaluated in HeLa cells using MTT assays. Figures g–i display the cell viability assay data from HeLa cells treated with different concentrations of 5ph-4TU under 390–400 nm or 760–770 nm irradiation for 20 min. Following exposure to light irradiation in the presence of 5ph-4TU with different concentrations, obvious decrease of cell viability is observed, being stabilized at ∼22% at a 5ph-4TU concentration of 100 μM upon 390–400 nm irradiation and ∼31% at a 5ph-4TU concentration of 500 μM upon 760–770 nm irradiation. In contrast, almost no decrease of cell viability is observed in the control groups (either no-5ph-4TU treated or no-irradiation in Figure c). These results demonstrate further that the effectively generated ¹O₂ can lead to significant cell death even under two-photon irradiation, albeit requiring much higher 5ph-4TU concentrations (IC₅₀ ≈ 109 μM, derived from Figure h) to reach comparable biological outcomes of single-photon irradiation (IC₅₀ ≈ 27 μM, derived from Figure g). Consequently, 5ph-4TU is anticipated to hold promising PDT applications as a near-infrared two-photon photosensitizer, with great biocompatibility and deep-tissue penetrating ability that is highly desirable for solid tumor treatment. Critically, control experiments confirmed that 5ph-4TU remains photostable under these irradiation conditions, with virtually no spectral changes being observed after prolonged exposure to the LEDs used in our cellular assays (Figures S11c–f). This confirms that the observed ¹O₂ generation and subsequent cytotoxicity mainly originate from the intrinsic activity of 5ph-4TU, rather than from potential photooxidation products. Moreover, it should be noted that the rate constant for oxygen quenching of the triplet state (2.7 × 10⁹ M^–1 s^–1) is much faster than the triplet self-quenching (k _q = 7.3 × 10⁸ M^–1 s^–1). Under the experimental conditions, the oxygen concentration (hundreds of μM to mM) also exceeds the 5ph-4TU concentration (tens to hundreds of μM), favoring the energy transfer to oxygen over self-quenching. Therefore, triplet self-quenching is unlikely to compete with ¹O₂ generation and should not affect the PDT efficacy of 5ph-4TU.

Mechanistic Implications and Significances

The experimental and theoretical results lead to insightful mechanistic implications, as follows:

Triplet Population

Unlike 4-TU where excitation to the S₂ (¹ππ*) state accesses a nearly barrierless (∼0.2 eV) pathway to the S₂/S₁ CI and consequently exhibits ultrafast S₂ (¹ππ*) decay (∼76 fs), 5ph-4TU exhibits markedly different dynamics. As discussed above, the phenylethynyl substitution on the C5-position in 5ph-4TU can sufficiently restrict the C5C6 bond twisting in the S₂ (¹ππ*) state, maintaining a nearly planar geometry, while the S₂ (¹ππ*)/S₁ (¹nπ*) CI exhibits a “bent” pyrimidine ring configuration. Such an important conformational deformation elevates the energy barrier (∼0.8 eV) for the S₂ (¹ππ*)_min → S₂ (¹ππ*)/S₁ (¹nπ*) CI, thereby hindering IC channel from S₂ (¹ππ*) to S₁ (¹nπ*) and slowing down the decay of the S₂ (¹ππ*) state with the time constant of ∼600 ps (in CH₃CN). The subsequent ISC mediated by the S₁/T₂/T₁ quasi-degenerate region is extremely fast. Therefore, the overall T₁ formation (S₂ → S₁ → T₂ /T₁) is kinetically controlled by the rate-limiting step of S₂ (¹ππ*) → S₁ (¹nπ*) internal conversion (∼600 ps), accounting for the decreased triplet quantum yield (Φ_T = 0.55) relative to 4-TU (Φ_T = 0.9). On the other hand, once reaching the S₁ (¹nπ*)_min, the S₁/T₂/T₁ three-state intersection (a typical characteristic for thiobases) is favorable for the occurrence of ISC for 5ph-4TU. Therefore, it still yields triplet with a substantial quantum yield (Φ_T = 0.55), which is essential for generating ¹O₂ to damage cancer cells. The possibility of the ISC induced by the intramolecular charge transfer (ICT) state as that for 5fdCyd has been ruled out since our high-level ab initio calculations did not locate any structure with an ICT character.

Enhancement of Fluorescence

Another important effect of the C5-phenylethynyl substitution is that it significantly prolongs the lifetime of S₂ (¹ππ*), which rationalizes the enhanced fluorescence behavior of 5ph-4TU. Based on the experimental S₂ lifetime (τ ∼ 600 ps) and Φ_F (∼0.11) in CH₃CN, the radiative lifetime (τ_r) is calculated to be ∼5.5 ns (τ_r = τ/Φ_F). This value is characteristic of an allowed radiative transition (typically in the ns range). The nonradiative decay lifetime τ_nr is further calculated to be ∼674 ps [τ_nr = τ/(1 – Φ_F)], much slower than the usual internal conversion rate. This suppression of the nonradiative channel is consistent with the substantial energy barrier (∼0.8 eV) for the S₂ to S₁ internal conversion imposed by the S₂/S₁ conical intersection, on one hand. On the other hand, there is a large energy gap ΔE_(S2–S1) of 0.56 eV for 5ph-4TU (Table ), pointing to an “anti-Kasha emission” that usually occurs when an energy gap (ΔE) between S_n and S_n–1 is greater than 0.35 eV to sufficiently slow down IC and permit emission from the higher state. Also importantly, the oscillator strength of S₂ fluorescence (calculated f = 0.3122) is a magnitude larger than that of the S₁ (¹nπ*) (f = 0.0002), supporting the S₂ (¹ππ*) state as the lowest-lying emissive state.

Moreover, the S₂/S₁ conical intersection can be elevated energetically with increasing solvent polarity, thereby hindering the relaxation pathway from the S₂ (¹ππ*)_min to the S₁ (¹nπ*) state. Varying the solvent from nonpolar dioxane to more polar ACN-H₂O = 1:2 retards the decay of the ¹ππ* from ∼553 ps to ∼964 ps. It means that further increasing of solvent polarity to a physiological aqueous environment can prolong the emissive lifetime of S₂ (¹ππ*) to nanoseconds above. It is thus anticipated that the fluorescence quantum yield of 5ph-4TU can be further enhanced in aqueous environment, which is beneficial for its application as a fluorescent imaging marker. Because of the insoluble character of 5ph-4TU in H₂O, spectroscopy experiments cannot be performed under real physiological conditions and the current spectroscopy results in CH₃CN or CH₃CN/H₂O mixed solvents provide a reference trend.

The bright fluorescence of 5ph-4TU is a rare character that should be noted. As the natural nucleosides are essentially nonfluorescent, significant modifications to nucleoside structure are required before they can be broadly used as emissive tools. Given that the increased π-conjugation and rigidity contributes greatly to improving optical emission properties, an ethynyl linker is used to allow a diversity of chromophores, such as a phenyl ring, to be attached to the pyrimidine ring as a molecular design strategy for enhancing fluorescence. This modification makes 5-phenylethynyl uridine emit fluorescence at 400 nm under UVA excitation (absorption at 320 nm). , However, for applications of fluorescent nucleosides in vivo, it is often desirable to red-shift their absorption and emission maxima as much as possible, in order to help avoiding background emission by native compounds in the cell, as well as cytotoxicity arising from ultraviolet irradiation. For this purpose, we show that 5ph-4TU displays superior performance. We find that phenylethynylation in synergy with thionation greatly lowers the excited-state energy, thus further red-shifting the absorption peak wavelength to 375 nm and emission peak wavelength to 450 nm with increased quantum yields (Φ_FL). Moreover, as demonstrated by our cellular experiment, 5ph-4TU can be readily taken up by cells and shuttled to the nucleus emitting blue fluorescence under 405 nm excitation, indicating its good biocompatibility and imaging property. Actually, the C5-phenylethynyl substitution makes 5ph-4TU resemble the structure of thymidine in the way that they are both C5-substituted and thus could be possibly incorporated into DNA by metabolism in a manner similar to 6TG or 4TU. , It shows here that, as in 5ph-4TU, the C5-phenylethynylation in synergy with thionation will empower superior fluorescence imaging properties for nucleosides, and can be employed as a new molecular design strategy to develop fluorescent bioprobes visualizing cellular nucleus etc.

Two-Photon Photoactive Properties

PDT as a noninvasive clinical treatment has been extensively utilized in superficial cancer therapy due to its unique selectivity and tumor ablation properties. Intriguingly, Crespo and co-workers recently reported on the synthesis of another pyrimidine nucleoside analogue, 5-(5-phenylthiophen-2-yl)-6-azauracil (PTAU), which, akin to 5ph-4TU, exhibits both fluorescence and a triplet-state population under UVA irradiation, rendering it suitable for applications in bioimaging and PDT. Unfortunately, the efficacy of PDT is often hindered by the poor penetration depth of ultraviolet and visible light. Near-infrared two-photon-excited PDT can increase the penetration depth of light, and thus has drawn ever-increasing research efforts. Here, we show that 5ph-4TU possesses substantial two-photon-absorption cross sections σ⁽²⁾ at 800 nm, owing to the extended π-conjugation structures. Although near-infrared is not at the peak wavelength of the two-photon absorption, irradiation of 5ph-4TU around 800 nm still results in strong fluorescence and robust generation of ¹O₂ and phototoxicity killing cancer cells. Accordingly, 5ph-4TU is demonstrated to be a promising near-infrared two-photon photosensitizer for PDT treatment targeting deep-seated tumors. Its great biocompatibility of incorporation into the nucleus is also particularly appealing.

Conclusions

In summary, through a combination of femtosecond transient absorption (fs-TA), time-resolved fluorescence spectroscopy (TRF), confocal microscopy, and the QM­(MS-CASPT2//CASSCF)/MM calculations, we have unraveled the ultrafast excited state decay pathways of 5ph-4TU upon single-photon or two-photon excitation. Following excitation to the S₂ (¹ππ*)_FC state, the 5ph-4TU quickly undergoes vibrational relaxation to the local state minima of S₂ (¹ππ*)_min within several picoseconds. Although there is a dark S₁ (¹nπ*) underneath, the internal conversion channel from S₂ (¹ππ*) to S₁ (¹nπ*) state, typically involving C5C6 bond twisting, is significantly hindered by the C5-phenylethynyl substitution as manifested by the high barrier of 0.8 eV from S₂ (¹ππ*)_min to access the S₂ (¹ππ*)/S₁ (¹nπ*) conical intersection. Consequently, the decay of the S₂ (¹ππ*) state was slowed down with the time constant of ∼600 ps (in CH₃CN). Considering that the S₁/T₂/T₁ triple quasi-degeneracy can act as an efficient gateway to the population of the triplet manifold, this sluggish S₂ (¹ππ*) → S₁ (¹nπ*) process creates a bottleneck in the overall triplet formation pathway, causing a bit reduced triplet quantum yield of 0.55, but still sufficient for subsequent ¹O₂ generation.

Moreover, the restriction of C5C6 bond twisting by the phenylethynyl group also impedes direct internal conversion from S₂ (¹ππ*) to the ground state S₀. The significantly prolonged S₂ (¹ππ*) state lifetime therefore increases the probability of fluorescence emission from S₂ (¹ππ*_min). Indeed, prominent fluorescence is observed in 5ph-4TU (Φ_FL = 0.11 ± 0.03) upon single-photon or even two-photon NIR excitation. The extending of π-conjugation by phenylethynyl together with thionation greatly lowers the excited-state energy of S₂ (¹ππ*), thus red-shifting the absorption peak wavelength to 375 nm and emission peak wavelength to 450 nm. Besides, the fluorescence lifetimes exhibit a consistent increase with solvent polarity (∼544 ± 12 ps in dioxane, 599 ± 10 ps in ACN, 910 ± 18 ps in ACN-H₂O = 1:1, 950 ± 16 ps in ACN-H₂O = 1:2). This suggests that the fluorescence lifetime could be further extended, and thus fluorescence enhancement may be more pronounced under aqueous physiological conditions. The intrinsic blue fluorescence of 5ph-4TU in the HeLa cells under 405 nm excitation is indeed vividly seen by confocal microscopy, making 5ph-4TU a bright fluorescent tool that is rare for common nucleosides.

Interestingly, significant two-photon absorption and photoactivities are observed in 5ph-4TU, owing to the extended π-conjugation structures. The two-photon absorption cross-section, σ⁽²⁾, is determined to be 44 GM at 800 nm, representing a large σ⁽²⁾ among related nucleosides in the near-infrared region. Upon two-photon excitation at 800 nm, the excited state dynamics closely resemble those under single-photon excitation. Aside from emitting fluorescence, the competing ISC pathway populates the long-lived triplet states, which further generates the toxic singlet oxygen ¹O₂ in HeLa cells as vividly visualized by confocal microscopy. The severe cytotoxicity killing cancer cells is observed even under 760–770 nm irradiation, demonstrating that 5ph-4TU holds promising PDT applications as a two-photon photosensitizer, with great biocompatibility and light-penetrating ability that is highly desirable for deep-seated tumor treatment.

From an ultrafast dynamics perspective, our findings provide in-depth mechanistic understandings of the synergistic effects of C5-π-conjugation and thionation in tuning the excited state decay pathways and creating superior photophysical and photochemical properties of 5ph-4TU. These results for 5ph-4TU will inspire the rational design of novel biophotosensitizers coupling the effective triplet formation, bright fluorescence, and near-infrared two-photon photoactivity together in one molecule, highly desirable for biomedical applications. Dual functionality substitution as in 5ph-4TU thus can be uitilized to overcome traditional limitations in combined bioimaging and cancer therapeutic applications, where one- or two-photon excitation of a photosensitizer can concurrently activate both cytotoxic singlet oxygen production for PDT and intrinsic fluorescence emission for real-time treatment monitoring of therapeutic progression.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01496.

• Additional experimental results and Cartesian coordinates of critical structures (PDF)

S.W., Q.Z., G.-N.P., and D.W. contributed equally to this work.

The authors declare no competing financial interest.