Activatable near-infrared fluorescence and chemical exchange saturation transfer MRI multimodal imaging probe for tumor detection in vitro and in vivo

Abstract

Multimodal imaging has emerged as a powerful tool in biomedical research and clinical diagnostics. Ideally, it combines multiple imaging techniques to provide complementary anatomical and molecular information in living subjects. Particularly desirable are multimodal imaging probes capable of providing differential diagnostic signals upon interaction with specific molecular targets. Hydrogen peroxide (H₂O₂) is a key target in this regard, since it is typically overexpressed in cancer cells. In this study, we present a small-molecule probe that not only selectively detects endogenous H₂O₂ through multimodal imaging, with a significant H₂O₂-triggered 15-fold fluorescence enhancement but also turns “on” a chemical exchange saturation transfer (CEST) magnetic resonance (MR) response with 60-fold signal enhancement at pH 7.4. Excellent selectivity against various other biologically relevant species is seen. Using this probe, we observed 3.4–4.5-fold and 2.8–5.8-fold higher H₂O₂ levels in cancerous cell lines and tumor tissues compared to normal cell lines and tissues, respectively. Time-dependent in vivo fluorescence and CEST imaging in a HeLa (Henrietta Lacks) tumor xenograft mouse model revealed probe-dependent tumor detection by fluorescence and CEST MRI contrast in the tumor area. These observations are attributed to the relatively high endogenous H₂O₂ levels produced during mitosis. This newly developed probe holds promise for advancing our understanding of H₂O₂-related biology and for cancer detection both in vitro and in vivo.

1. Introduction

Molecular imaging has emerged as a versatile and indispensable tool for modern biological research and disease diagnosis. By providing insights into biology fundamentals and providing enhanced disease diagnostic capabilities, molecular imaging has transformed our understanding of basic biology and improved healthcare outcomes [1,2]. Various techniques, including fluorescence imaging (FI), magnetic resonance imaging (MRI), single-photon emission computed tomography (SPECT), positron emission tomography (PET), ultrasound imaging (USI), computed tomography (CT), and photoacoustic imaging (PAI), have been developed in an effort to permit and improve in vivo molecular imaging [3,4].

Each imaging method has unique advantages and inherent limitations [5,6]. For example, FI has high sensitivity, ease of use, affordability, and rapid image acquisition [7,8]. However, shallow tissue penetration and susceptibility to interference from biological tissues limits its effectiveness. In contrast, PET offers extraordinary sensitivity and deep tissue penetration but is hindered by its poor spatial resolution [9–11]. MRI, allows remarkable spatial resolution and good tissue penetration; it excels in providing high-quality three-dimensional (3D) images but falls short in terms of sensitivity [12]. Photoacoustic imaging (PAI), a recently introduced technique that produces three-dimensional (3D) images with exceptional sensitivity, spatial resolution, and deeper tissue penetration [13,14], suffers from limitations when visualizing biomolecules deeply seated within the human body [15,16]. So-called multimodal molecular imaging offers a potential solution to the limitations of individual imaging techniques. By combining two or more imaging modalities, multimodal imaging harnesses the inherent characteristics of each modality and thus has the potential to yield detailed anatomical and cellular information. Multimodal molecular imaging thus holds the promise of enhancing our understanding of biological processes and improving our ability to monitor disease pathogenesis [17,18]. However, for the promise of multimodal molecular imaging to be realized new probes that enhance the sensitivity of complementary imaging methods need to be developed. Here, we report the preparation and study of a multimodal imaging probe, BODIPY-perox that acts as a combined cancer specific long near-infrared (NIR) luminescent chemical exchange saturation transfer (CEST) magnetic resonance imaging (MRI) agent both in vitro and in vivo (Scheme 1). As shown in Scheme 2, BODIPY-perox is comprised of two distinct moieties, namely the NIR fluorescent emitter BODIPY and a hydrogen peroxide (H₂O₂)-responsive linker that upon reaction releases a free phenolate hydroxyl (-OH) group to enhance CEST MRI. A control system, DOTA-perox, was also prepared. It contains an Eu-chelating DOTA core and a H₂O₂ sensing subunit. However, in contrast to BODIPY-perox it lacks a fluorophore.

Scheme 1.

Synthesis of DOTA-perox and BODIPY-perox. Reagents and reaction conditions: (a) Bromine in chloroform (CHCl_{3) at room temperature (12} h); (b) t-butyl bromoacetate, potassium carbonate (K_{2CO3), dimethylformamide (DMF) at 75°C (18} h); (c) compound 2, K_{2CO3, DMF at ~90°C (16} h); (d) trifluoroacetic acid (TFA) at room temperature (24 h); (e) europium(III) chloride hexahydrate (EuCl_{3⋅6 H2O); (f) triflic anhydride (Tf2O), triethyl amine (Et3N), dichloromethane (DCM) at room temperature (12} h); (g) bis (pinacolato)diboron, [1,1´-bis(diphenylphosphine)ferrocene]dichloropalladium(II) (PdCl_{2(dppf), potassium acetate (KOAc), dimethylsulfoxide (DMSO), 80°C}.

Scheme 2.

Design strategies for H₂O₂-activatable multimodal probes (DOTA-perox and BODIPY-perox). Expected reductive reaction activation of DOTA-perox and BODIPY-perox by hydrogen peroxide (H₂O₂).

H₂O₂ is a quintessential reactive oxygen species (ROS) generated primarily by active triphosphopyridine nucleotide (NADPH) oxidase in cells [19,20]. H₂O₂ plays a crucial role as a messenger in cellular signal transduction for various physiological functions and serves as a marker for oxidative stress [21,22]. Its synthesis in cellular organelles, particularly within mitochondria, is essential for cell survival, proliferation, differentiation, and maintenance under normal conditions [23,24]. However, abnormal H₂O₂ production or accumulation within cellular mitochondria due to oxidative stress and/or genetic abnormalities is linked to numerous diseases, including cancer, diabetes, obesity, stroke, and neurological disorders [25–28]. This dichotomy has motivated the development of innovative tools for localizing and monitoring endogenous H₂O₂ production within cells.

MRI offers an appealing alternative to optical imaging due to its superior spatial resolution, substantial penetration depth, and lack of harmful radiation exposure. Chemical exchange saturation transfer (CEST) agents, with their distinctive magnetic resonance frequencies, facilitate precise detection of these agents and their associated biomarkers in comparison to traditional MRI techniques [29,30]. CEST agents operate by selectively reducing the magnetization of water signals while minimally affecting their longitudinal relaxation rate. Realizing these potential benefits requires the use of molecules with exchangeable protons in conjunction with the magnetization transfer nuclear magnetic resonance (MT-NMR) technique pioneered by Forsen and Hoffman in 1963 [31].

To date, various methods have been explored for H₂O₂ detection, with fluorescent probes standing out for their sensitivity in measuring endogenous H₂O₂ [32–34]. MRI-based H₂O₂ detection methods are also appealing; they offer excellent resolution but typically lack specificity and sensitivity for endogenous H₂O₂ (Fig. 1 and Table S1) [35–37]. Recent developments in sensing platforms and activatable luminescent chemical exchange saturation transfer (CEST) MRI probes have shown promise [38]. However, challenges persist, such as the limited reach of intravenously injected activatable luminescent MRI agents at tumor sites and a lack of specificity for H₂O₂. We posited that a probe that allowed the concurrent use of both techniques (fluorescence detection and CEST MRI) might provide for improved H₂O₂ detection and, accordingly, improved diagnoses of hydrogen peroxide-related pathologies. Using BODIPY-perox we show that this approach is effective in the context of cancer detection, both in vitro and in vivo. This success is ascribed to the presence of elevated H₂O₂ concentrations in the tumor micro-environment (TME).

Fig. 1.

Timeline showing the development of small molecule H₂O₂-activatable probes and associated imaging modalities.

2. Materials and methods

2.1. Materials and general experimental methods

Details are provided in the Supporting Information.

2.2. Synthesis of compound DOTA-perox

Compound 4 (759 mg, 1 mmol) was dissolved in TFA (2 mL) and stirred for 24 h. The mixture was evaporated to dryness. The resulting product was purified by silica gel column chromatography using dichloromethane/methanol (DCM/MeOH = 95:5) as the eluent, yielding a free ligand (410 mg). The free ligand (59 mg, 0.1 mmol) and K₂CO₃ (41 mg, 0.03 mmol) were dissolved in water (10 mL), and the pH was adjusted to 7 with NaOH (0.1 M). Excess EuCl₃•6 H₂O was added, and the pH was adjusted to 6.5. After stirring at room temperature for 12 h, the excess Eu³⁺ was precipitated as Eu(OH)₃ by increasing the pH above 8 using 1 M aqueous NaOH. The solution was filtered, and the pH was adjusted to 7 using 1 M HCl. The solution was freeze-dried to obtain DOTA-perox (68 mg, 91.9%). A xylenol orange indicator test was used to examine an aqueous solution of the Eu³⁺-(4) complex for the absence of free Eu³⁺ ions. The purity of DOTA-perox was confirmed by a high performance liquid chromatography (HPLC) with a retention time (R_f) of 3.4 min being seen on a Sunfire C18 column (4.6 mm × 150 mm, 5 μm; Waters, MA, USA) and with a flow rate of 1 mL/min, resulting in 96.1%. ¹H NMR (400 MHz, DMSO-d₆): δ 7.86 (d, J = 6.64, 2 H); 7.797 (s, 2 H); 3.9 (s, 2 H); 3.515 (s, 6H); 3.377 (d, J = 17.12, 6 H); 3.019 (m, 8 H); 2.799 (s, 2 H); 1.108 (s, 12 H). ¹³C NMR (100 MHz, DMSO-d₆): 174.342, 133.869, 127.150, 117.80, 114.896, 75.605, 53.013, 51.703, 48.999, 47.653, 42.189, 23.677 ppm. ESI HRMS m/z[M+H]⁺: calculated 764.199, found 764.197.

2.3. Synthesis of compound BODIPY-perox

To a solution of 6 (408 mg, 0.526 mmol) in DMF (2 mL), bis(pinacol) diboron (174 mg, 0.684 mmol) and potassium acetate (155 mg, 1.58 mmol) were added. The flask was purged with nitrogen gas for 15–20 min. Pd(dppf)Cl₂ (129 mg, 0.158 mmol) was added to the solution, and the reaction mixture was purged with nitrogen gas for another 10 min. The reaction mixture was stirred at 110 °C for 2 h. The volatiles were evaporated off, and the residual was extracted with ethyl acetate. The organic layer was dried over anhydrous sodium sulfate, and the crude product was purified by silica gel column chromatography using dichloromethane/methanol (DCM/MeOH = 95:5) as the eluent, yielding a dark green solid (298 mg, 76%). ¹H NMR (400 MHz, DMSO-d₆): δ 8.149 (m, 8 H), 8.06 (d, J = 10.68, 1 H), 7.461 (m, 7 H), 7.096 (d, J = 10.12, 1 H), 7.069 (m, 3 H), 4.614 (s, 2 H), 1.535 (s, 12 H), 1.396 (s, 9 H). ¹³C NMR (100 MHz, DMSO-d₆). 167.47, 160.63, 146.245, 144.931, 142.786, 135.532, 134.791, 134.433, 132.593, 132.12, 131.451, 131.364, 129.61, 129.453, 129.408, 129.284, 129.186, 128.616, 128.584, 128.242, 126.221, 124.434, 119.398, 118.528, 114.923, 84.003, 82.792, 65.646, 28.067,24.903. ESI-HRMS m/z [M+H]⁺: calcd. 754.336, found 754.3302.

3. Results and discussion

3.1. Design and synthesis of the DOTA-perox probe

The DOTA-perox control probe was synthesized as follows. A first key precursor, (4-(2-bromoacetyl)phenyl)boronic pinacol ester (compound 2), was obtained by treating 4-acetylbenzeneboronic pinacol ester (compound 1) with bromine in chloroform (CHCl₃) at room temperature (Scheme S1). A second precursor, compound 3, was synthesized by selectively protecting 1,4,7,10-tetraazacyclododecane (CYCLEN) through triple N-alkylation with tert-butyl bromoacetate [39]. Compound 4 was then prepared by reacting 3 with 2. Deprotecting the tert-butyl groups using trifluoroacetic acid yielded DOTA-perox. Treating this ligand with an excess of EuCl₃⋅6 H₂O gave the corresponding Eu³⁺-chelated complex. The key precursor to BODIPY-perox, compound 6, was prepared by treating the known BODIPY derivative 5 [40] with triflic anhydride. Palladium(II)-mediated coupling with bis (pinacolato)diboron then yielded BODIPY-perox. Both probes and intermediates 1–5 were characterized by ¹H and ¹³C NMR spectroscopy and HR-MS (Figures S1–S15).

Fig. 2 presents the DFT-optimized structures and frontier molecular orbitals for DOTA-perox (A) and BODIPY-perox (B). In both species, a presumed hydrogen-bonding interaction between the -BF₂ group and the phenyl ring is inferred based on their geometric proximity. The computed energy gaps between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of BODIPY-perox and DOTA-perox were found to be 0.113 and 0.144 eV, respectively. These values lead us to suggest that BODIPY-perox is more reactive and sensitive than DOTA-perox. BODIPY-perox, as shown in Fig. 2C, exhibits sp² hybridization at the boron atoms [41]. When the nucleophile anionic form of hydrogen peroxide, O₂H^–, attacks the boron atom in BODIPY-perox, a boronate complex is generated with concurrent rehybridization to form a sp³ hybridized boron center (Int-1). As a consequence of the high electron density on the boron atom, the C–B bond dissociates, followed by aryl migration to the adjacent oxygen acceptor atom to generate intermediate Int-2. The calculated activation energy (ΔG^≠) for this step (Int-1 → Int-2) is 24.54 kcal/mol for a pathway proceeding via inferred transition state TS-1. Hydrolysis of Int-2 results in the formation of a phenolic –OH group and full cleavage of BODIPY-perox. The calculated ΔG^≠ for the rate-limiting step in the anionic O₂H^– oxidative pathway (Int-1 → Int-2) is much lower than that computed for the neutral pathway, leading us to propose that the anionic pathway is energetically feasible and likely dominant [42]. We also optimized the transition state (TS) of oxidative conversion of the boronic ester to the electron-donating phenol of BODIPY-perox with a Gaussian 09 program. Geometry optimizations were performed using B3LYP-D3/6–31 G** (water, ε = 78.3553) B3LYP-D3/6–311+G** level of theory.

Fig. 2.

DFT-optimized structures and computed HOMO/LUMO gaps for DOTA-perox (A) and BODIPY-perox (B). The calculated energy gap for DOTA-perox is ΔE_DOTA-perox = (HOMO–LUMO) = −0.143 eV, while that for the BODIPY-perox is − 0.113 eV. (C) DFT-computed schematic energy diagrams and corresponding optimized geometries for intermediates proposed to be involved in probe cleavage. P = probe.

3.2. Optical properties of BODIPY-perox and DOTA-perox

To determine whether H₂O₂ could reduce the boronic ester or boronic acid moiety in the probes (BODIPY-perox and DOTA-perox) to their corresponding phenol (hydroxyl) forms, we monitored the changes in the UV–vis absorption spectra and fluorescence emission of the compounds in the presence of various concentrations of H₂O₂ under simulated physiological conditions. Initially, we recorded the UV absorbance spectrum of BODIPY-perox dissolved in phosphate buffered saline (PBS) at different H₂O₂ concentrations (Fig. 3A). The UV–vis absorption maximum of probe BODIPY-perox (5 μM) at 700 nm increased in intensity by approximately 25-fold upon adding 200 μМ H₂O₂, as illustrated in Fig. 3B. Similarly, the emission intensity at 730 nm increased 10-fold when exposed to 200 μМ of H₂O₂ (Fig. 3C) with the changes proving concentration dependent (Fig. 3D). Specifically, the relative mean fluorescence intensity of BODIPY-perox (5 μM) was measured at different concentrations of H₂O₂ (0, 50, 100, 150, and 200 μM) using an optical imaging system (IVIS spectrum, Perkin Elmer) (Figs. 3G and 3H) with an increase in fluorescence intensity as a function of increasing H₂O₂ concentration being observed. The time required to reach fluorescence intensity saturation decreased with increasing H₂O₂ concentration with the maximum intensity being attained in most cases within approximately 12 min. The detection limit was determined to be 1.41 ng/mL. A HPLC analysis of probe BODIPY-perox (5 μM) in the presence of H₂O₂ revealed a new band, consistent with the H₂O₂-promoted reduction of the boronic ester group to a phenol (Fig. 3I). Further support for the proposed reaction between H₂O₂ and BODIPY-perox came from an HR-MS analysis carried out after treatment with 200 μM H₂O₂ at 37 °C for 30 min. A major [M + H]⁺ peak was seen at 644.248 reflecting the generation of an activated probe³⁷ (Figure S16).

Fig. 3. Photophysical properties of BODIPY-perox before and after H₂O₂ treatment.

(A) UV absorption spectra of BODIPY-perox (5 μM) recorded in the presence of various H₂O₂ concentrations (0–200 μM). Fluorescence absorption (B) and emission (C) spectra (λ_ex = 685 nm) of BODIPY-perox recorded in the presence of different H₂O₂ concentrations (0, 50, 100 and 200 μM). (D) Emission intensity of BODIPY-perox (5 μM) recorded at various H₂O₂ concentrations (pre-incubated for 15 min at 37 °C) (λ_ex/λ_em = 685/730 nm for all experiments, unless noted otherwise). (E) Fluorescence response of the BODIPY-perox probe (5 μM) to different potential interferants in blood serum: (a) blank (probe in PBS), and serum containing: (b) 150 mM KCl, (c) 2.5 mM MgCl₂, (d) 2.5 mM CaCl₂, (e) 200 μM H₂O₂, (f) 1 mM vitamin C, (g) 1 mM vitamin B₆, (h) 100 μM HSA, (i) 10 mM glucose, (j) 200 U/L lipase, (k) 100 μg/L pepsin, (l) 200 ng/mL trypsin (m) 200 U/L phosphatase, (n) 1 mM arginine, (o) 1 mM serine, (p) 5 mM glutathione, (q) 1 mM cysteine, (r) 1 mM lysine, (s) 1 mM glutamic acid, (t)) 1 mM tyrosine, (u)) 1 mM histidine. (F) Fluorescence response of BODIPY-perox (5 μM; at λ_em = 730 nm) to different species present in blood serum (graphically). (G) Change in the fluorescence intensity of BODIPY-perox seen in response to treatment with various H₂O₂ concentrations (0, 50, 100, 150, and 200 μM) as measured using an optical imaging system (IVIS Spectrum, PerkinElmer) and (H) corresponding quantitative comparison of the fluorescence intensity with H₂O₂ concentration. (I) HPLC profiles of BODIPY-perox (5 μM) in the presence of H₂O₂ (200 μM) as determined a different indicated time. The data are the average of four independent experiments; the error bars represent standard deviations.

Before applying probe BODIPY-perox to an in vitro cellular test system, we examined whether other biologically relevant analytes, including metal ions (K⁺, Mg²⁺, Ca²⁺), redox active species (ascorbic acid, vitamin B₆, human serum albumin (HSA), H₂O₂, and ${\text{S}}_2{{\text{O}}_3^2}^{-}$), enzymes (lipase, pepsin, trypsin, and phosphatase), and amino acids (cystine, alanine, serine, lysine, histidine, arginine, glutamic acid, and tyrosine), gave rise to interference. No significant spectroscopic changes were observed in the presence of these potentially interfering analytes (Fig. 3E and F).

We also monitored changes in the fluorescence intensity of the probe at 730 nm in the presence of H₂O₂. The fluorescence intensity reached a maximum at an H₂O₂ concentration of 200 μM. These spectral changes are attributed to converting the boronic ester group into the corresponding phenol (–OH) group, which is an excellent electron donor. The resonance process results in the “turn on” of the fluorescence signal. Specifically, we suggest that intramolecular charge transfer between the OH oxygen atom lone pair electrons and the electron-deficient europium ion is switched on, activating the emissive features of DOTA-perox (Scheme 2).

The optical properties of DOTA-perox in the presence of H₂O₂ were also monitored using UV absorbance and luminescence emission spectroscopies (Figures S17A and S17B). The emission maximum at 620 nm steadily increased as a function of the H₂O₂ concentration (Figure S17C). The time required to reach saturation in terms of the luminescence intensity gradually decreased with increasing H₂O₂ concentration. This is ascribed to increasing H₂O₂ concentrations serving to accelerate the reduction of DOTA-perox to give activated DOTA-perox. In most cases, saturation was reached within approximately 20 min. The detection limit was calculated to be 58.3 ng/mL H₂O₂ (Figure S17D). The probe was exposed to 200 μM H₂O₂ at 37 °C for 30 min to verify that it was H₂O₂ that was mediating the activation of DOTA-perox. Only one significant peak, corresponding to the activated probe, was observed by HRMS analysis ([M + H] = 630.110) (Figure S18).

Mirroring what was done for BODIPY-perox, before testing the ability of probe DOTA-perox to signal the presence of H₂O₂ in a cellular model system, we examined whether interference was seen in the presence of other biologically relevant analytes, including thiol amino acids (cystine, homocysteine and glutathione) biologically relevant metal ions (K⁺, Mg²⁺, Ca²⁺, Zn²⁺), redox-active species (ascorbic acid, vitamin B₆, HSA, H₂O₂, and glucose), enzymes (lipase, pepsin, trypsin, and phosphatase), thiol amino acids (cysteine and glutathione) and non-thiol amino acids (alanine, serine, lysine, arginine, and glutamic acid). We observed no spectroscopic changes in the presence of these potentially interfering analytes (Figure S19A). The fluorescence intensity of DOTA-perox in the presence and absence of H₂O₂ was then monitored as a function of pH and time at λ_em =615 nm (Figures S19B-D). The time required to reach fluorescence saturation gradually decreased as a function of increasing H₂O₂. In the case of the enzyme studies, an increase in the enzyme concentration also accelerated the reduction. BODIPY-perox was also subjected to various important oxidizing agents, including peroxyacetic acid (CH₃CO₃H), hypochlorite (ClO^–), singlet oxygen (¹O₂), superoxide (O₂^–), and peroxynitrile (ONOO^–), at a concentration of 200 μM for each of these potential interferants. The fluorescence intensity did not notably increase in response to these additional substances (Figures S20). Notably, BODIPY-perox exhibited a weaker response to ONOO^– compared to H₂O₂. We also carried out tests with lipid droplet analogs, such as phosphatidylserine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylserine (PS), and phosphatidylinositol (PI) (Figure S21). The results revealed that the presence of other lipid droplet analog substances did not significantly increase the fluorescence intensity. Both probes, DOTA-perox and BODIPY-perox, demonstrated favorable plasma stability over a 72-hour period (Figure S22). Photostability was also observed throughout the observation period (Figure S23). Taken in concert, these findings provide support for the suggestion that BODIPY-perox will exhibit selectivity for H₂O₂ in complex biological environments.

3.3. CEST MRI analysis of the activated probes in solution

As a complement to the above studies, we investigated whether probe activation could be monitored using CEST MRI. The reduced probes possess a phenolic OH group capable of proton exchange with the surrounding water. We thus expected that saturation of these protons with a radio frequency (RF) pulse would increase the CEST MRI signal owing to a decrease in the water signal (cf. Scheme 3). Because the CEST signal was expected to be pH-dependent, we initially examined the CEST MTR_asym (%) (Fig. 4A), Z-spectra (Fig. 4B), and MTR_asym (%) color map (Fig. 4C) of 5 mM aqueous BODIPY-perox solutions at pH values ranging from 5 to 8 in the presence of H₂O₂ (1 mM).

Scheme 3.

Schematic representation of the H₂O₂ activation process, CEST MRI detection of multimodal probes DOTA-perox (A) and BODIPY-perox (B). The probes are reduced by hydrogen peroxide (H₂O₂) containing exchangeable phenolic OH protons, leading to the saturation of surrounding water protons and the activation of the CEST MRI signal. See the text for further discussion.

Fig. 4.

CEST signals produced by activated BODIPY-perox. (A) Calculated MTR_asym (%) values of aqueous 10 mM BODIPY-perox solutions at different pH values, (B) Z-spectra, and (C) representative MTR_asym (%) CEST color map at 6 ppm. (D) Saturation power (B₁) dependence of the MTR_asym (%) spectra of BODIPY-perox observed at pH = 6.5. (E) Calibration curve relating the MTR_asym (%) value to the saturation power based on the spectra. (F) MTR_asym (%) CEST contrast color map at 6 ppm. (G) Z-spectra values for different BODIPY-perox concentrations at pH = 6.5. (H) MTR_asym (%) and (I) corresponding MTR_asym (%) CEST image at 6 ppm.

BODIPY-perox exhibited a broad CEST spectrum of 0.5–12 ppm, with a peak of approximately 6 ppm. Subsequent CEST imaging confirmed that the probe BODIPY-perox could be used to monitor H₂O₂ activity over a wide range of pH values using MRI. As seen in a previous study, the CEST signal produced by activated BODIPY-perox and ascribed to the OH protons proved pH-sensitive [43]. MTR_asym (%) spectra for different saturation field strengths (B₁) from 1 to 6 μT for solutions at pH = 6.5 are presented in Fig. 4D. MTR_asym (%) values (Fig. 4E) and MTR_asym (%) CEST color map (Fig. 4F) at 6 ppm increased with saturation power from 1 to 5 μT, eventually saturating at 6 μT. We then selected B₁ =4.8 μT for all subsequent experiments, as it provided a comparable CEST signal at ~6 ppm to that at the higher B₁ value but with a narrower spectrum. We assessed the concentration dependence of the CEST spectra under slightly acidic conditions (pH = 6.5), considering that malignant tumors are often acidic [44]. The 6 ppm peak from the phenolic OH group could be observed in the CEST Z-spectra (Fig. 4G) and is reflected in the MTR_asym (%) values (Fig. 4H). Even at the lowest concentration of BODIPY-perox (0.5 mM), the MTR_asym (%) peaks were greater than 10%, making them easily detectable. The corresponding CEST images at 6 ppm demonstrated readily discernible CEST signal changes as a function of BODIPY-perox concentration. Additionally, we performed CEST experiments with DOTA-perox and observed behavior similar to that seen for BODIPY-perox as reflected in the corresponding Z-spectra (Figure S24A), MTR_asym (%) spectra (Figure S24B), and MTR_asym (%) color maps (Figure S24C).

3.4. In vitro activation of probes and imaging of HeLa cells

Before conducting fluorescence imaging of HeLa cells, we assessed the biocompatibility of both DOTA-perox and BODIPY-perox against normal cells (NIH3T3, WI38, HEK293, and BMMC), as well as cancer cells (HepG2, U87MG, A549, HeLa, and MDA-MB-231) at concentrations ranging from 0 to 20 mM for 24 h. No appreciable cytotoxicity was observed (Figure S25). On this basis we considered it likely that these probes would possess sufficient biocompatibility to allow for their use in H₂O₂ detection. Flow cytometric analysis of BODIPY-perox (5 μM) revealed 3.2–4.8 times higher H₂O₂ levels in cancer cell lines than normal cells (Figure S26A). In contrast, only a 1.3–2.0-fold difference in H₂O₂ levels in cancer cells compared to normal cells was seen for DOTA-perox (5 μM) (Figure S26B). Given its lower H₂O₂ sensitivity, DOTA-perox was not used for further studies. In contrast, an effort was made to analyze the features of BODIPY-perox in detail. Further evidence supporting selective cancer cell imaging with HeLa cells as compared to normal cells ((NIH3T3, WI38, and HEK293), was obtained through confocal scanning laser microscopy (CLSM) imaging studies. These investigations revealed that BODIPY-perox produces a statistically significant level of fluorescence increase under simulated oxidative conditions (Figure S27).

We investigated the cellular uptake of BODIPY-perox and its intracellular response to endogenous H₂O₂ by CLSM. After incubation with 4 μM BODIPY-perox for 15 min, a faint red fluorescence inside HeLa cells was observed. The fluorescence intensity increased when the HeLa cells were pretreated with 50 μM H₂O₂ for 30 min, followed by BODIPY-perox treatment for 15 min (Fig. 5A). The relative mean fluorescence intensity (MFI) was enhanced approximately two-fold when the cells were pretreated with H₂O₂ (Fig. 5B). The observed increase in MFI is ascribed to the elevated endogenous H₂O₂ levels. The H₂O₂ serves to reduce catalytically BODIPY-perox, resulting in an increase in the fluorescence intensity. Further evidence of selective H₂O₂-reduced cell imaging was provided by fluorescence-activated cell sorting (FACS) analysis of HeLa cells (Fig. 5C). HeLa cells pretreated with H₂O₂ (50 μM) exhibited approximately 1.5 times brighter fluorescence than the probe-treated cells alone.

Fig. 5.

In vitro imaging of BODIPY-perox in HeLa cells. (A) Confocal laser scanning microscopy (CLSM) images of HeLa cells pre-treated with H₂O₂ or vehicle for 30 min, washed and incubated with BODIPY-perox (4 μM) for 15 min. Scale bars, 20 μm. (B) Corresponding quantitative comparison of fluorescence intensities measured from 15 individual cells using ImageJ (nonpaired Student’s t-test). (C) Representative fluorescence-activated cell sorting (FACS) of HeLa cells pre-treated H₂O₂ (50 μM) for 30 min, washed with PBS three times, and incubated with BODIPY-perox for 15 min. (D) Fluorescence imaging of HeLa cell spheroids with BODIPY-perox (10 and 20 μM) for 30 min and (E) quantitative values of mean fluorescence intensity (MFI) of HeLa spheroids incubated with BODIPY-perox. Scale bars, 100 μm. (F) CEST spectrum of HeLa cells incubated in different concentrations of BODIPY-perox (0, 0.5, and 1 mM). (G) CEST contrast color map at 2.5 ppm and (H) corresponding statistical comparison of signal intensities determined from the color map. The data are averages of three experiments; error bars represent the standard deviations. p values were determined using a nonpaired Student’s t-test: n.s., nonsignificant; **, p<0.01; and ***, p<0.001.

To evaluate the efficacy of BODIPY-perox to map H₂O₂ levels in a tumor microenvironment, we performed fluorescence microscopy imaging of HeLa cancer cells cultured in 3D spheroids (Fig. 5D and E). After incubation with BODIPY-perox at concentrations of 10 or 20 μM for 30 min, the MFI increased at both concentrations under tumor microenvironment conditions as determined relative to the non-treated group. We then recorded the MTR_asym (%) curves for HeLa cells in the presence or absence of the probe BODIPY-perox (0, 0.5, and 1 mM). The CEST spectra (Fig. 5F) contained a main peak at a maximum of ~2.5 ppm, which increased with BODIPY-perox concentration. The CEST color map generated at 2.5 ppm showed distinctly enhanced CEST contrast with increasing BODIPY-perox concentration. Based on CEST experiments involving independent HeLa cells at different concentrations of BODIPY-perox, the signal intensity in HeLa cells containing the probe (0, 0.5, and 1 mM; n = 3 each) was statistically distinct from that in the non-treated cells under identical conditions (Fig. 5H). These results support the core contention of this study, namely, that BODIPY-perox can be used to map H₂O₂ levels in cancer cells and thus serve as a probe that may have a role to play in assessing malignancy.

We also tested the ability of BODIPY-perox to detect endogenous H₂O₂ in HeLa cells generated by lipopolysaccharide (LPS) treatment, which is known to generate reactive oxygen species as the result of H₂O₂ production [34]. In this study HeLa cells were incubated with 100 nM/mL LPS for 24, 48, or 72 h and then treated with probe BODIPY-perox (0.5 μM) for 15 min and washed and imaged (Fig. 6A). FACS analysis showed a clear increase in fluorescence intensity in the LPS-stimulated HeLa cells compared to the control cells (Fig. 6B). Additionally, the fluorescence signal increased proportionally with the LPS treatment time reflecting the expectation that longer incubation with LPS would produce higher levels H₂O₂ in the HeLa cells. The production of H₂O₂ by LPS treatment was monitored by CLSM. As shown in Fig. 6C, the fluorescence intensity increased with LPS treatment time (24, 48, and 72 h) after incubation with BODIPY-perox (0.5 μM). In contrast, we observed a relatively weak fluorescence in the LPS non-treated group. The MFI increased by approximately 1.5-fold after treating with LPS for 24 h. An approximately 2.1-fold increase was seen after 48 h, and an approximately 3-fold increase was seen after 72 h as compared to the untreated group (Fig. 6D).

Fig. 6.

In vitro imaging of BODIPY-perox in HeLa cells treated with lipopolysaccharide (LPS). (A) Schematic showing the protocol used for LPS incubation with subsequent probe treatment. (B) CLSM images of HeLa cells pre-treated with LPS (100 ng/mL) or vehicle for 24, 48, and 72 h, prior to being incubated with BODIPY-perox (0.5 μM) for 15 min and washed and imaged. Scale bars, 20 μm. (C) Corresponding quantitative comparison of fluorescence intensities from 15 individual cells using the ImageJ software (nonpaired Student’s t test). (D) FACS analyses of HeLa cells treated with LPS followed by BODIPY-perox for 24, 48, and 72 h. (E) Western blot analysis of reactive oxygen species (ROS)-related proteins (BCL2 and Bax) in HeLa cells treated with LPS or vehicle. β-Actin was used as a control. (F) Corresponding quantitative comparison of the BCL2 and Bax protein levels and β-actin control after LPS treatment.

Phorbol 12-myristate 13-acetate (PMA) generates H₂O₂ at the cellular level via superoxide [30,31]. HeLa cells treated with PMA (1 μg/mL) for 60 min exhibited higher BODIPY-perox fluorescence than control cells as determined by CLSM (Figure S29). Moreover, adding DPI as a broad-spectrum inhibitor or ebselen as a general antioxidant quencher of H₂O₂ inhibited the H₂O₂-induced enhancement of the BODIPY-perox fluorescence (Figure S29). As noted above, the cellular H₂O₂ levels could be changed by subjecting to LPS treatment (100 nM/mL). Further support for this conclusion came from western blot analyses of two recognized apoptosis proteins (BCL2 and Bax) (Fig. 6E). Increased production of H₂O₂ during LPS treatment results in the activation of apoptosis, increased Bax protein expression, and decreased BCL2 expression. Bax protein expression increased by approximately 4.5-fold after 72 h of LPS treatment, whereas BCL2 protein expression decreased by approximately 3.8-fold (Fig. 6F). Actin was used as a control.

3.5. In vivo and ex vivo imaging of HeLa cell tumor

To assess the potential of BODIPY-perox as a tumor-targeting H₂O₂-activatable fluorescence/MRI probe for bioimaging, we conducted experiments using a HeLa cell tumor xenograft mouse model. After establishing the tumor model, BODIPY-perox (5 mg/kg) was administered intravenously (IV) to the mice through their tail veins. CEST MR imaging with pre-saturation using an RF pulse (single hard pulse of B₁ = 3.6 μT, 1 s) was carried out using a 4.7 T animal MRI system (Biospec 47/40, Bruker) before and after the IV injection of BODIPY-perox. T₂-W, CEST, and merged images are displayed in panels A and B of Fig. 7, showing the state of the HeLa tumor xenograft mice before and 20 h after IV treatment with BODIPY-perox. After injection of this probe, we observed CEST signal enhancement at 2.5 ppm, primarily at the tumor sites (indicated by the yellow circle). This observed enhancement is thought to reflect the selective activation of the boronic ester group within BODIPY-perox at the tumor site. The MTR_asym (%) value at 2.5 ppm within the tumor region increased approximately 2.5-fold 20 h after the injection of BODIPY-perox (Fig. 7C and D). These findings are taken as evidence that the CEST signal was enhanced following probe injection and that tumor-derived H₂O₂ actively promotes the “switching-on” of probe BODIPY-perox.

Fig. 7.

In vivo CEST imaging of BODIPY-perox in a HeLa xenograft mouse model. T₂-weighted (T₂-W) and CEST MR images obtained (A) before and (B) 20 h after intravenous (IV) injection of probe BODIPY-perox (5 mg/kg, 100 μL) into the HeLa tumor-bearing mice (n = 5). Each CEST MR image recorded at 6 ppm is merged with the corresponding T₂-W MR image. (C) MTR_asym value measured with saturation offset from water before and after injection of BODIPY-perox in the HeLa xenograft mouse model. (D) Quantitative comparison of the mean MTR_asym (%) values at 2.5 ppm for the tumor region (yellow dotted elipsoid). The data represent the average of four measurements with error bars showing the standard deviation. p values were determined using a nonpaired student’s t test: **, p<0.01.

Inspired by the CEST imaging results, we performed a time-dependent in vivo fluorescence imaging study of BODIPY-perox at 730 nm in xenograft HeLa tumor models and other organs. This study was carried out using an optical imaging system (IVIS Spectrum, PerkinElmer). The tumor-bearing nude mice were IV injected with BODIPY-perox. Imaging was then carried out at various times. Enhanced fluorescence intensity was observed at the tumor site 6 h after injection, a finding interpreted in terms of the accumulation of activated BODIPY-perox at the tumor site (Fig. 8A). Quantitative analysis revealed that the fluorescence intensity in the BODIPY-perox group at the tumor site increased in a time dependent manner and consistently exceeded that of the background over the 6–48 h time window post-injection (Fig. 8B). At 24 h, for instance, the fluorescence intensity at the tumor site was 4-fold higher than background.

Fig. 8.

Dual imaging (fluorescence and CEST) in vivo tumor detection in a HeLa xenograft mouse model using BODIPY-perox as a probe. (A) Time-dependent in vivo optical imaging of xenograft HeLa tumor-bearing mice following intravenous (IV) injection of BODIPY-perox (5 mg/kg, 100 μL), and (B) corresponding quantitative comparison of fluorescence intensity obtained in the tumor region concerning body signal (background) (n = 5). (C) Complementary, time-dependent in vivo CEST imaging of probe BODIPY-perox in a HeLa xenograft mouse model. (D) Quantitative comparison of the MTR_asym (%) values at 2.5 ppm for the tumor region at different time points following injection of BODIPY-perox (nonpaired Student’s t-test). (E) Representative ex vivo fluorescence images obtained from several organs, including the heart (1), liver (2), spleen (3), kidney (4), and tumor (5), extracted after the final in vivo imaging at 20 h, for the BODIPY-perox group compared to the PBS group. (F) Corresponding comparison of the fluorescence intensities measured in organs and tumors. (G) Ex vivo fluorescence images of tumor tissues (n = 5) and quantitative comparison (H) of fluorescence intensities for the BODIPY-perox and PBS groups. Error bars show standard deviations (nonpaired Student’s t-test). *, p<0.05; **, p<0.01; and ***, p<0.001.

The positive results of the time-dependent fluorescence imaging encouraged us to perform complementary time-dependent CEST imaging. We recorded CEST images of mice before BODIPY-perox injection, which served as a control. We observed extensive CEST stimulation between 16 and 20 h after injection, a finding interpreted in terms of progressive activation of BODIPY-perox (Fig. 8C and D). The MTR_asym (%) values displayed a gradual and statistically significant 2.1-fold increase at the tumor site 20 h after injection (Fig. 8D). In line with the fluorescence data, these CEST results revealed a higher signal intensity at the tumor site 20 h after injection.

The biodistribution of BODIPY-perox was studied in mice bearing HeLa tumors 20 h after injection (Fig. 8E). Phosphate buffered saline (PBS) was used as a control. After sacrifice, selected organs and tumor tissues were harvested from the mice and used to compare their respective fluorescence intensities. Solid tumors from animals treated with PBS showed no noticeable fluorescence, whereas obvious fluorescence was seen in the tumors and kidneys of mice treated with BODIPY-perox.

The strong signal observed in the mouse kidneys during these ex vivo experiments is attributed to the renal excretion of BODIPY-perox. We detected relatively weaker ex vivo fluorescence signals in the liver and none in the heart or spleen. Quantitative analysis revealed a 4.1-fold higher fluorescence intensity for BODIPY-perox relative to PBS in the tumor region (Fig. 8G). This increase was recapitulated in the tumors collected from two groups (BODIPY-perox or PBS-injected), where again a 4-fold increase in the fluorescence intensities for BODIPY-perox was seen relative to the PBS control (Fig. 8F and H). BODIPY-perox accumulation within the tumor is explained on the basis of the enhanced permeability and retention (EPR) effect [33,45–47]. Importantly, our probe, BODIPY-perox, exhibited no negative side effects during or after treatment. These results underscore the effectiveness of our approach to activate selectively probes (BODIPY-perox in the present instance) within tumors using H₂O₂ as a trigger. More specifically, our findings serve to confirm that tumors can be imaged effectively in vivo using the fluorescence/CEST signal emanating from activated BODIPY-perox.

4. Conclusions

In summary, we developed a highly sensitive and selective tumor-targeting probe, BODIPY-perox, that permits the noninvasive detection of intracellular hydrogen peroxide (H₂O₂) in cancer cells. This probe relies on the incorporation of a biocompatible NIR fluorophore BODIPY moiety into an H₂O₂-sensing boronic ester. Using a combination of flow cytometry, confocal microscopy, fluorescence imaging, and CEST MRI, we demonstrated that BODIPY-perox may be used to monitor accurately the variations in H₂O₂ levels produced by endogenous sources in various cell types and repeatative stimuli. The ability to monitor H₂O₂ levels in cells allows cancer cells to be distinguished effectively from normal cells. In vitro studies revealed that BODIPY-perox displays acceptable biocompatibility and that it may be used to distinguish cancer cells from normal cells. Follow-up in vivo tests using a xenograft mouse cancer model revealed that BODIPY-perox selectively accumulates in solid tumors and produces maximal fluorescence and CEST signal intensity roughly 48 h after injection. No adverse events or side effects were seen. We thus suggest that BODIPY-perox or analogous H₂O₂-sensitive probes could prove useful in the early detection and treatment of cancer.

Biographies

Sanu Karan received his Ph.D. degree in Analytical Science from Chungnam National University, Korea, in 2022. Presently, he serves as a postdoctoral researcher at the Korea basic Science Institute in Cheongju, Korea. His research focuses on drug delivery and bio-imaging, as well as the development of endogeneous stimulated fluorescent probes.

Jee-Hyun Cho received her master’s degree in physical chemistry (with a focus on NMR) from Seoul National University, Korea, and her Ph.D. in analytical chemistry (specializing in MRI) from Chung-Ang University, Korea. Her research is primarily centered on analyzing disease characteristics and advancing diagnostic techniques using MRI methodologies, including multiple quantum MRI, CEST MRI, and MR spectroscopy. Within her laboratory, she has access to state-of-the-art imaging equipments, including 7 T and 3 T human MRI, as well as a 9.4 T animal MRI scanner. Her research encompassess a wide range of studies involving diseases in various animal models, including mice, rats, monkeys, and humans.

Jonathan L. Sessler did his undergraduate work at the University of California before receiving his Ph.D. from Stanford University in 1982. He is presently the Doherty-Welch Chair in Chemistry at The University of Texas at Austin. His research interests include drug discovery, sensor design, anion recognition, critical elements, expanded porphyrins, and supramolecular chemistry. He has published more than 900 papers and is an inventor of record on over 80 issued U.S. patents. He is a member of a number of learned societies, including the U.S. National Academy of Sciences, the American Academy of Arts and Sciences, the European Academy of Science, and the Chinese Academy of Science (Foreign Member).

Kwan Soo Hong earned Ph.D. degree in the Department of Physics from Seoul National University, Korea, in 1998. Currently, he serves as the Director of the Ochang Institute of Biological and Environmental Science at the Korea Basic Science Institute, Korea. Additionally, he holds a professorial position in the Department of Chemistry at ChungAng University, Korea. His research focuses on bionano-materials, organic optical biosensors for theranostic applications, and drug delivery systems.

Data availability

Data will be made available on request.