In Vivo Visualization of Tumor Metabolic Activity with a Tetra Glucose-Conjugated Zinc-Phthalocyanine Photoacoustic Contrast

Abstract

Tissue metabolic alterations are associated with tumor progression and serve as clinical biomarkers. Molecular imaging methods can provide a noninvasive assessment of this altered metabolic activity. Currently, in the clinic, nuclear medicine using ¹⁸F-FDG, a radioactive analogue of glucose, is the gold standard for visualizing metabolically active tumors. However, the accompanying ionizing radiation and accumulated radiation dosage limit its unchartered use. Noninvasive imaging of tissue metabolic activity without incorporating any radioactive isotope or another additional anatomical imaging is a promising alternative to nuclear medicine. Here, we introduce the first-of-its-kind tetra glucose-conjugated molecular photoacoustic (PA) contrast agent, a water-soluble and biocompatible small molecule based on the Zn-phthalocyanine scaffold. Although the Zn-phthalocyanine core is hydrophobic, the conjugation of four glucose units through their anomeric carbon ensured the water solubility of this agent, thereby aiding in its potential translation for in vivo studies. In addition, such a conjugation contributed to the high cellular uptake of this molecule in two aerobic cancer cell lines, as demonstrated using flow cytometry and epifluorescence microscopy studies. Importantly, with live metabolic assays, we elucidated the mechanism through which the contrast agent could be utilized as a glucose antagonist in nutrient-starved cells. Finally, with real-time in vivo PA tomography studies in a 4T1 mouse tumor model, we showed maximum agent accumulation within 4 h and tumor washout within 12 h post intravenous administration. Noninvasive molecular PA imaging of metabolic tumors with this probe offers a promising alternative to nuclear medicine, especially in assessing therapy response with the requirement of shorter intervals for follow-up in the clinic.

1. Introduction

Metabolic reprogramming is a key hallmark of cancer that dictates dynamic alterations in the levels of metabolite precursors such as glucose, pyruvate, glutamine, and fatty acids within tumor tissue. − Noninvasive techniques for visualizing and quantifying these metabolic changes can accurately diagnose and differentiate hypermetabolic cancer tissues from normal tissues. , Additionally, monitoring real-time changes in tissue metabolic activity before and after treatment is crucial for assessing therapy and evaluating treatment response. ,

Magnetic resonance (MR) methods such as MR spectroscopy and hyperpolarized MRI with ¹³C-labeled pyruvate offer advanced metabolic imaging abilities and quantification of metabolite conversion. − However, the low sensitivity and lengthy acquisition and analysis times associated with these methods limit their routine use in clinical practice for metabolic imaging. Positron emission tomography (PET) is the current gold standard in the clinic, and ¹⁸F-fluoro 2-deoxy glucose (¹⁸F-FDG) is one of the radiotracers routinely used. ¹⁸F-FDG exploits the Warburg phenomenon and demonstrates a stronger PET signal intensity, due to higher accumulation of the tracer in metabolically avid tumor tissues compared to the normal tissue. − Although PET offers exceptional diagnostic sensitivity, it is costly and relies on radioisotopes with short half-lives. Additionally, the low spatial resolution of PET inevitably requires a complementary anatomical imaging method, such as computed tomography (CT). This combination makes longitudinal imaging of patients, especially those requiring frequent follow-ups, challenging due to the risk of accumulated radiation exposure from repeated PET/CT scans. − An alternative that avoids ionizing radiation while providing both anatomical and functional metabolic information within the same modality can potentially address this limitation.

Photoacoustic (PA) tomography emerges as a compelling alternative and can provide anatomical three-dimensional images at a resolution of ∼100 μm and at tissue depths of ∼3–5 cm. − Along with anatomical information, it can indirectly evaluate functional changes such as oxygen utilization within tissues by unmixing tissueoxy and deoxy hemoglobin PA signatures that indirectly assess the tissue metabolic demand. − However, label-free PA imaging has a low contrast and poor specificity. Introducing smart contrast agents can circumvent this caveat. − Indeed, PA contrast agents developed recently have successfully evaluated important disease biomarkers such as tumor pH, oxidative stress in atherosclerotic plaques, reactive nitrogen species, and upregulated enzyme activity in plaques and macrophages. − In this context, it is important to note that most of the recent advances in the development of PA contrast agents have heavily focused on nanoparticles, and there has been limited progress in the design and development of new, rational small molecule agents. − A substantial portion of small molecule dyes studied as PA contrast agents relied on the scaffold of fluorescent or optical agents, namely, boron dipyrromethene (BODIPY), indocyanine green (ICG), or IRDye and their modifications. − In this study, we chose the phthalocyanine construct rather than the fluorescent dye scaffolds for several reasons. Phthalocyanines belong to the class of second-generation porphyrin-based photosensitizers. Compared to optical dyes such as BODIPY, ICG, and IRDye800, phthalocyanines hold higher potential for PA contrast, primarily because, in the latter case, there is preferential relaxation of energy from the excited to the ground state through nonradiative (thermal or internal conversion) rather than radiative emission, which is the route of energy dissipation for optical agents. Indeed, a comparison of PA properties within several small molecule photosensitizers, including methylene blue, chlorins, phthalocyanine, protoporphyrin IX, and squaraine, undoubtedly established phthalocyanine as the most promising photosensitizer with PA abilities. Previously, one optical agent and a small molecule, IRDye800-2DG (Licor Biosciences), were used for PA imaging in metabolic tumors, but as primarily an optical agent, the sensitivity and uptake in certain tumors were relatively poor. The challenges with PA dyes often are insufficient solubility, low molar extinction coefficient, and their tendencies to aggregate and photobleach. Indeed, even phthalocyanines, despite their promising rich PA contrast and exceptional photostability, remain underexplored for PA biomedical applications, primarily due to insolubility issues. Considering these limitations of previous probes, we developed a novel small-molecule, tetraglucose-conjugated PA agent to noninvasively assess metabolically active tumors. This agent features the phthalocyanine scaffold with low fluorescence quantum yield (implying high PA yield) and also incorporates four glucose moieties attached through their anomeric carbon, which is expected to greatly enhance the solubility and cellular uptake. We anticipated that the high degree of molecular symmetry, along with the extended π conjugation exhibited by phthalocyanines, would also hold for the tetrasubstituted phthalocyanine that we constructed, facilitating higher nonradiative emission, leading to its PA signal. The modulation of radiative and nonradiative emission in phthalocyanines, correlating to their fluorescent or PA property, is achieved through variation in the central metal ion or the ring substitutions, and electron-withdrawing ring substituents, reducing the fluorescence quantum yield of the ring. Therefore, the presence of four glucose moieties connected through the oxygen atom on anomeric carbon could potentially improve the PA contrast. Finally, the addition of four glucose moieties not only aids in retaining the symmetry of the Zn-phthalocyanine but also adds 16 hydroxyl groups making the probe readily water-soluble.

Our findings highlight the agent’s advantageous photophysical properties, including significant absorbance in the near-infrared (NIR) range, a high molar extinction coefficient, and an improved PA quantum yield compared to existing scaffolds. We further demonstrate the in vitro and in vivo efficacy of this small-molecule contrast agent using cell lines and wild-type mouse models. Overall, this study presents a photostable, biocompatible, water-soluble small-molecule PA contrast agent that can be employed to noninvasively image metabolically active tumors, serving as a potential surrogate imaging contrast agent in certain superficial cancers where PA tomography is approved for clinical use.

2. Materials and Methods

2.1. Synthesis of Photoacoustic Contrast Agent

All synthetic procedures and relevant characterization are provided in the Supporting Information.

2.2. Photophysical Properties

Absorbance measurements for GPc in the different solvent systems and at different concentrations were performed with a UV–vis spectrophotometer (UV-1800, Shimadzu, Japan). Fluorescence emission spectra for GPc (DMSO) and Zn-phthalocyanine tetrasulfonate (ZnPcS4), used as a standard for the measurement of fluorescence quantum yield, were acquired in a multimode plate reader (BioTek Synergy H1). All experiments were conducted at room temperature. The fluorescence quantum yield of GPc in DMSO was obtained from the following equation

$${\phi }_{\mathrm{G}\mathrm{P}\mathrm{c}}={\phi }_{\mathrm{Z}\mathrm{n}\mathrm{P}\mathrm{c}\mathrm{S}4}[({\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}}_{\mathrm{G}\mathrm{P}\mathrm{c}}/{\mathrm{s}\mathrm{l}\mathrm{o}\mathrm{p}\mathrm{e}}_{\mathrm{Z}\mathrm{n}\mathrm{P}\mathrm{c}\mathrm{S}4})]\times [({{\eta }_{\mathrm{D}\mathrm{M}\mathrm{S}\mathrm{O}}}^2)/({{\eta }_{\mathrm{D}\mathrm{M}\mathrm{S}\mathrm{O}}}^2)]$$

where φ_ZnPcS4 is the reported fluorescence quantum yield of the standard ZnPcS4 in DMSO. The slopes are obtained from the plot of the integrated fluorescence intensity vs absorbance for each of them, and η is the solvent’s refractive index. PA spectra of GPc in DMSO and in PBS at the respective concentrations were acquired using a multispectral optoacoustic tomography scanner (iTHERA Medical, MSOT inVision 256TF, Germany). The phantom studies were performed in 1.5% Agar (Sigma) and 1 mL of Intralipid 20% (Fresenius Kabi, China) per 50 mL, which was added to minimize the light scattering effect. In a selected phantom volume, scans were acquired from a 660 nm–800 nm wavelength with 5 nm increments from 670 nm, with 3 averages per wavelength. The built-in temperature sensor in the scanner was used to equilibrate the water bath temperature to the desired value for the temperature dependence studies on GPc.

2.3. Plasma Stability Experiments

Plasma was extracted from mouse blood, and a 100 μM concentration of GPc was prepared in plasma using 1 mM GPc stock (1× PBS) and used for PA image acquisition. The PA acquisition parameters were kept the same as above, and scans were acquired at different time intervals from time of preparation, 0 to 12 h.

2.4. Cytotoxicity and Protein Expression Studies

4T1 cells were procured from ATCC and OVCAR 3, and hTERT FT 282 cells were obtained as a gift from Prof. Ramray Bhat (Department of Developmental Biology, IISc). Both the cell lines were maintained in RPMI 1640 media (HiMedia) supplemented with 20% FBS (Gibco) and 1% Penicillin/Streptomycin (Lonza) in a 5% carbon dioxide, 37 °C temperature humidified incubator.

Cytotoxicity studies were performed with a standard sulforhodamine B (SRB) assay. Briefly, 20,000 cells were seeded in a 96-well plate, and after 24 h, different concentrations of GPc in 1× PBS (1 μM–2000 μM) were added to the plate in triplicates. 48 h post incubation with GPc, cells were fixed with 30% trichloroacetic acid, washed with 1% acetic acid, and dried overnight. SRB dye was then added the following day following the assay protocol, and absorbance was recorded at 510 nm to quantify the cell viability. The assay was performed for 3 biological repeats.

2.5. Cellular Uptake Experiments

For flow cytometry studies, both 4T1 and OVCAR 3 cells were seeded in 6 well plates, and 12 h postseeding, cells were incubated with 100 μM GPc (from a 10 mM stock of GPc in 1× PBS) for the following time periods0 h, 1 h, 2 h, 4 h, and 6 h. Postincubation cells were washed three times with PBS, and single-cell suspension was prepared using standard procedures. The single-cell suspension in polystyrene tubes was centrifuged at 450 RCF for 5 min and at 4 °C. The supernatant was discarded, and the cells were resuspended in 200 μL PBS or PBS + propidium iodide (PI) (2 μL/1 mL of 1 mg/mL PI stock), as required. We recorded >30,000 events per sample (BD FACS Celesta, BD Biosciences). Flow cytometry data was analyzed with FCS express software. The experiments were performed in 3 individual repeats.

Epifluorescence microscopy studies were performed with both 4T1 and OVCAR 3 cells, which were plated on glass coverslips, and the cells were incubated with 100 μM GPc for 0, 1, 2, 6, and 8 h, following which they were fixed with 4% PFA and permeabilized with 0.5% triton-X100. Blocking was performed with 2% BSA; then cells were stained with GLUT1 (E4S6I #73015) rabbit primary antibodies with a dilution of 1:50 in 1% BSA and incubated at 4 °C overnight followed by secondary antibody incubation at room temperature for 2 h (antirabbit IgG Alexa Fluor 488 # 4412, 1:300). Immunofluorescence images were acquired (Carl Zeiss Axio Observer 7, NA = 0.4). Images were acquired using a PCO edge 4.2 camera (Excelitas Technologies). The samples were imaged in three different channels for visualizing DAPI (λ_ex = 350 nm, λ_em = 465 nm), GPc (λ_ex = 663 nm, λ_em = 691 nm), and GLUT1 (λ_ex = 488 nm, λ_em = 496 nm). A magnification of 63× was used for image acquisition. Fluorescence intensity and colocalization were performed using ImageJ software.

2.6. Metabolic Assays

A real-time seahorse glycolytic stress assay was first performed to determine the bioenergetic profile of the 4T1 and OVCAR3 cell lines. Briefly, 4T1 and OVCAR3 cells were seeded in 8 well seahorse mini culture plates (Agilent Technologies) with complete medium and incubated in a 37 °C CO₂ incubator overnight. The assay cartridge plate (Agilent Technologies) was hydrated overnight in a 37 °C non-CO₂ incubator the day prior to the assay. On the day of the assay, fresh DMEM XF assay medium was prepared, and the pH was adjusted to 7.4. Following the assay guidelines, the medium in the mini cell culture plate was replaced with the freshly prepared assay medium, and the water in the assay cartridge plate was replaced by the calibrant solution, and both the plates were incubated for 1 h in a 37 °C non-CO₂ incubator. The appropriate concentration of oligomycin and 2-DG was prepared in the assay media, and GPc concentrations ranging from 10 to 500 μM were also prepared in the assay media. Finally, during the assay with the glucose-starved OVCAR 3 cells, glucose/GPc, oligomycin, and 2-DG were added to the injection ports A, B, and C in the respective concentrations as required. Once the assay was completed, the cell plate was taken out, and an SRB assay was performed to account for live cell count normalization. Finally, the Seahorse data was analyzed using Wave (Agilent Technologies). The assay was repeated for 3 biological repeats.

For the competition assay, an induced glycolytic stress assay was performed with the OVCAR 3 cells. The cell plate and cartridge were prepared in a similar manner prior to the assay. In this case, after the initial 15 min run to acquire basal glycolysis values, glucose was injected into port A at different concentrations (0 μM, 250 μM, 500 μM, or 1 mM), and the assay was run for 30 min, following which GPc was added through port B at a concentration of 250 μM or 0 μM (only PBS) and following port B injection, the assay was run for another 30 min. Finally, through port C and port D, oligomycin and 2 DG were added in a manner similar as above; post assay completion, SRB assay was immediately performed to assess the live cell count, and data was normalized to the number of live cells in the WAVE analysis software.

2.7. Animal Experiments

All animal studies were performed per the institutional guidelines outlined by the Institutional Animal Ethics Committee, Indian Institute of Science (protocol no. CAF/ethics/065/2024). Female athymic nude mice, 4–6 weeks old, were used in this study. 4T1 flank tumor models were made by injecting 1.5 × 10⁶ cells suspended in PBS/Matrigel (1:1) in the left flank of the mice. Mice were housed at the Central Animal facility (CAF, IISc) with food and water and adhered to the 12 h dark/light cycle. Tumor volumes were longitudinally assessed with vernier calipers, and imaging was performed when the tumor size was approximately 350–400 mm³. Animals were fasted before imaging for 8–12 h.

2.7.1. PA Imaging

On the day of imaging, mice were transported from CAF to the imaging room and left to acclimatize for 2 h. Mice were restrained, and 150 μL of 150–200 μM of the GPc in PBS was injected through the tail vein. Following injection of the GPc probe, the mouse was secured to the holder under anesthesia (2.5–3% isoflurane + O₂), oriented supine in the scanner, and imaged at 30 min, 2 h, and 4 h postadministration of GPc. The temperature of the water bath in the MSOT scanner was maintained at 35–36 °C throughout the scan, and the mouse was imaged at 1% O₂ and 3.0% isoflurane (Isifrane 250, Abbott). A volume was chosen such that transverse slices of the tumor, kidney, and liver were included, and a step size of 0.5 mm was selected. The speed of sound was adjusted prior to acquisition to obtain a good preview of the anatomy. We chose wavelengths of 660, 680, 700, 720, 740, 760, 780, 800, 820, 840, 860, and 880 nm for each position, using 5 averages per wavelength with an acquisition time of 10 ms per frame to minimize the influence of animal movement in the images. The concave transducer array, comprising 256 elements spanning a circular arc of 270°, was employed to detect PA signals at a central frequency of 5 MHz. The approximate scan time was 15 min for each scan. Animals were euthanized at 6 h (Cohort 1), 12 h (Cohort 2), and 24 h (Cohort 3) postadministration of GPc using CO₂ asphyxiation, and the tumor, liver, kidney, spleen, and intestines were excised, washed with cold water, and coated with ultrasound gel, and PA imaging of the ex vivo tissue was performed using the same holder and the same acquisition parameters.

2.7.2. Serum Biochemistry and Ex Vivo Histology

Athymic nude mice (6 weeks old, female) were injected with saline (N = 2, 150 μL, intravenously) or GPc (N = 2, 150 μL, 200 μM, intravenously, double dosing on day 1 and day 3), and blood was collected through the abdominal aorta after 24 h of the second dose. 500 μL of blood was collected from each mouse and held straight in an Eppendorf for 30 min at room temperature, after which it was stored at 4 °C and given to the Rohana Veterinary Diagnostic (RVD) lab, Bengaluru, Karnataka, India, for the assessment of the following parameters: SGPT, SGOT, ALP, BUN, creatinine, serum albumin, and total protein.

For histology, excised organs (liver, kidney, and spleen) were washed with PBS and fixed in 4% paraformaldehyde for 6 h. Fixed tissues were stored in sodium azide solution and given to RVD lab for sectioning (5 μm) and hematoxylin and eosin (H&E) staining of paraffin-embedded sections. Images of H&E-stained slides were taken in an Olympus-colored microscope (model number IX73) under 20× magnification.

2.8. Image Processing

All frames in the acquired PA image were subjected to a standard reconstruction using a standard FBP (field back projection) algorithm included with analysis software (iTHERA Medical, Germany). The reconstructed images were then postprocessed for fluence correction and spectral unmixing (Hb, HbO₂, and GPc spectra). Fluence correction was performed by drawing a mask around the tumor and another mask around the entire anatomy. For both fluence-corrected images, spectral unmixing for the three chromophores was performed to give individual signatures of each chromophore without much interference from the other chromophores. Tumor GPc uptake was quantified from the mean PA intensity on the spectrally unmixed images by drawing a region of interest (ROI) on the tumor slices or any other organ of interest.

2.9. Statistical Analysis

For all experiments, three independent repeats were performed, unless mentioned otherwise. Ordinary one-way or repeated measures (RM) ANOVA were performed as statistical tests, and the specific tests were mentioned along with the respective figures. The significance is represented as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Data are presented as arithmetic mean ± standard deviation.

3. Results and Discussion

3.1. Design and Photophysical Properties of GPc

First, we investigated optical and PA small molecule dyes available in the literature and suitable positions in the glucose moiety for conjugation to such a dye. Taking cues from reports − and to override the current limitations of small molecule PA contrast agents, we initiated the rational design of a tetra-glucose conjugated small molecule PA dye. We designed an agent to overcome the high background signals typically visualized with optical agents. Here, we chose a unique scaffold with a low fluorescence quantum yield compared to the known NIR optical dyes such as ICG and IRDye800. We focused on the Zn-phthalocyanine molecule, and synthesis of the final GPc agent was achieved in three steps with slight modifications of reported procedures (Figure A). , In the first step, nucleophilic substitution of 4-nitro phthalonitrile with 2,3,4,6-tertraacetyl glucopyranose furnished the protected precursor (I) for the phthalocyanine synthesis. The acetylated form of the protected precursor (I) was then subjected to base hydrolysis to yield the deprotected precursor (II). In the final step, (II) underwent a cyclocondensation reaction in the presence of zinc acetate to generate the Zn-phthalocyanine ring conjugated to four glucose units through their anomeric carbon (III). The detailed synthesis and characterization data of the reaction products are included in the Materials and Methods and Supporting Information (Figures S1–S9).

1.

Synthesis and photophysical characterization of the PA contrast agent, GPc. (A) The three step synthesis of the tetra glucose conjugated Zn-phthalocyanine small molecule. (B) Absorbance spectra of GPc in DMSO at various concentrations, generating a molar extinction coefficient (ε) = 37,049 M^–1 cm^–1. (C) Fluorescence emission spectra of GPc in DMSO at different concentrations and at λ_em at 690 nm and λ_ex at 400 nm. (D,E) Mean PA intensity of GPc in DMSO and in PBS at different concentrations, demonstrating the linear relationship between concentration and PA signal intensity, and solvent effect on PA sensitivity of GPc. (E) % PA intensity of GPc, acquired over 12 h, in a mouse serum sample (n = 3).

Next, the UV/vis, fluorescence, and PA spectral properties of GPc were evaluated in both DMSO and PBS. In the UV/vis spectrum of GPc, we observed the characteristic narrow Q-band at 680 nm and B-band at 352 nm of phthalocyanines. We recorded the molar extinction coefficient ε₆₈₂ = 37,049 M^–1 cm^–1 in DMSO (Figure B). One of the main hypotheses in the design of our agent was to overcome the limited solubility of phthalocyanine scaffolds, thereby debilitating the in vivo translation of these agents. For comparison, Zn-phthalocyanine is insoluble in water and exhibits poor solubility even in DMF (2.7 mg/mL, MedChemExpress, USA), wherein ultrasonic simulators or heat are required for complete solubilization. We indeed observed excellent water solubility for GPc, probably due to the presence of a high number of free hydroxyl groups from the four glucose moieties. As reported in the literature, we also observed a significant effect of the solvent on the molar extinction coefficient (ε) of the probe, with a much lower molar extinction coefficient in PBS. Although the extinction coefficient is lower in PBS than in DMSO, the probe maintains excellent stability in PBS over 24 h (Supporting Information, Figure S10) without any significant observable change in the absorption maxima. Fluorescence emission experiments conducted at different excitation wavelengths between 400 and 500 nm showed maximum emission at the wavelength (λ_em) of 690 nm at an excitation wavelength (λ_ex) of 400 nm, and emission spectra acquired at different concentrations of GPc revealed a linear relation of emission intensity and probe concentration (Figure C). The introduction of 16 hydroxyl groups to the agent seemed to affect the fluorescence emission properties of the agent, and we observed that the fluorescence quantum yield was quite lower (Φ_F = 0.08, DMSO) (Supporting Information, Figure S11). Compared to the NIR agents, ICG (Φ_F = 0.12) and IRDye800 (Φ_F = 0.15) but similar to the commercially available ZnPcS4 (Φ_F = 0.07, DMSO), which was also used as the standard in our quantum yield studies (Supporting Information, Figure S11). More importantly, this sufficiently low fluorescence quantum yield for GPc suggests a much stronger nonradiative decay through thermoelastic expansion, resulting in a higher PA quantum yield than that of ICG and IRDye800.

The PA absorbance and spectrum of GPc in DMSO were in excellent agreement with the absorbance spectra, and the PA intensity at different concentrations of 10 μM, 18 μM, 50 μM, and 100 μM suggested a linear relationship between concentration and PA signal intensity (Figure D). More importantly, GPc displayed excellent PA sensitivity, as observed from the strong signal intensity at lower concentrations of 10 μM. As observed with the absorbance studies, the PA signal intensity of GPc was lower in the case of PBS when compared to DMSO (Figure E), the reduced ε in PBS compared to DMSO is attributed to the hydrophobic interactions of the glucose-substituted phthalocyanine with water molecules. Although the signal intensity is weaker in PBS, a similar linear correlation of signal intensity to concentration was observed. Since the amplitude of the PA signal (A) is proportional to parameters such as the light fluence, F, an acoustic attenuation coefficient, ξ, the optical absorption coefficient of the imaging target, μa, and also a temperature-dependent coefficient, Γ, known as Grüneisen parameter, as noted in eq (eq ). We explored the temperature dependence of the GPc PA intensity at two different temperatures of 25 and 36 °C in the same sample, and within this temperature range, the PA signal varied by ±30 MSOT au (Supporting Information, Figure S12).

$$A=F·\xi ·\mu a·\mathrm{\Gamma }$$ 1

Collectively, the encouraging photophysical and PA properties of the new small molecule, tetra-glucose substituted ZnPc (GPc), warranted in vitro biological assessment to validate its translational potential for noninvasive live animal imaging within metabolic active tumors.

3.2. In Vitro Stability, Cytotoxicity, and Cellular Uptake Studies

The PA behavior and stability of the GPc agent within a biologically relevant system was studied by observing the change in PA signal intensity over time in mouse serum over a period of 12 h. While buffer solutions can provide a general idea of probe stability, mouse serum mirrors a physiological environment and pH. GPc incubated in mouse serum showed no significant change in PA signal intensity (%) over 12 h, signifying excellent probe stability in biological systems (Figure F).

Next, we evaluated the toxicity of GPc in ovarian cancer (OVCAR 3) and breast cancer (4T1) cell lines and in normal hTERT FT 282 immortalized cells to evaluate the biocompatibility of the contrast agent. A standard colorimetric SRB assay based on correlating the protein content of live cells to the viable cell population was performed to assess the cell viability after incubation with GPc for 48 h, over a concentration range of 1 μM to 2 mM of GPc. The three cell lines had slightly variable LD₅₀ values, suggesting that the agent did not have the exact same toxicity effect on the two cell lines, which is not unexpected as the metabolic phenotype is different for the three cells. The LD₅₀ in OVCAR 3 was 1.02 mM; in 4T1 cells, it was 361.8 μM, and in the normal hTERT FT 282, it was 492.3 μM (Figure A,B, Supporting Information, Figure S13). Notably, these results imply that the agent has low toxicity at doses below 200 μM, and hence at the lower concentrations used for in vivo studies, the agent should not cause any detrimental effect on other organs.

2.

Cytotoxicity analysis of GPc in OVCAR 3 (A) and 4T1 (B) cells using the SRB colorimetric assay, performed in 3 biological repeats. Flow cytometry analysis of OVCAR 3 (C) and 4T1 (D) cells incubated for different times with 100 μM GPc (n = 3).

After establishing good biocompatibility and excellent biostability of GPc, we explored the uptake in both cell lines. We first assessed the expression of GLUT transporters in OVCAR 3 and 4T1, mainly GLUT1 and GLUT3, as these proteins are integral in transporting glucose and glucose-conjugated small molecules into cells. GLUT transporters are often overexpressed in cancer cells, , and under a hypoxic microenvironment, their expression is reportedly amplified to meet the energy demand of tumorigenic cells. We observed that under normoxic conditions both the OVCAR 3 and 4T1 cells expressed GLUT1 and GLUT3. Under hypoxic conditions, we observed that the expression of both transporters although higher (4T1), was not significant (Supporting Information, Figure S14). Next, using the flow cytometry technique, we directly quantified the time-dependent uptake of GPc in OVCAR 3 and 4T1 cells. We chose to perform the studies with a 100 μM GPc concentration as saturating concentrations of GPc could also result in self-quenching, as observed with other fluorescent dyes such as 2-NBDG, a fluorescent d-glucose analogue. The uptake of GPc was rapid in both cell lines, as measured using the median fluorescence intensity (MFI) of cells (Figure C,D, and Supporting Information, Figure S15). The control cells without any GPc incubation had very low MFI, and by 1 h post incubation, MFI was significantly higher in the case of OVCAR 3 (Figure C). In the 4T1 cell line, the MFI was significantly higher at 6 h postincubation (Figure D). The signal augmentation observed until 6 h indicated that signal saturation was not attained by that time. Of specific interest with respect to translation, the steady increase in signal up to 6 h suggested that the rate of uptake of GPc was much higher than the decomposition of the agent within the first 6 h. These results demonstrated promise for in vivo imaging for longer time points postinjection of the contrast agent.

Additional validation of the cellular uptake studies was also performed with epifluorescence microscopy experiments. Although the low fluorescence quantum yield of the GPc posed a hurdle for these studies, our results agreed with the inferences from flow cytometry studies. Both 4T1 and OVCAR 3 cells were incubated with GPc for durations of 0, 1, 2, 6, and 8 h. In both cell lines, GPc uptake was observed at 2 h, and cellular fluorescence showed a gradual increase in cellular uptake over time in both cells (Figure A,B, Supporting Information, Figure S16). GLUT1 is primarily an extracellular membrane protein, and we did not see GPc colocalized to GLUT1 on the cell surface, highlighting that our GPc contrast agent could enter tumor cells efficiently.

3.

Epifluorescence microscopy images of 4T1 (A) and OVCAR 3 (B) cells with different incubation times of GPc (red), costained with glucose transporter membrane protein, GLUT1 (green) show cellular uptake of GPc within both cells by 2 h and retention until 6 h; scale bars: 20 μm.

3.3. Mechanistic Studies to Assess the Function of GPc as a Metabolic Substrate or Glucose Antagonist

Considering the impressive properties of this small molecule as a PA contrast agent, we first proposed elucidating the possible mechanistic evaluation of this molecule to serve as a metabolic precursor. We took note of the fact that there are only a few examples of small moleculeseither with a substitution in the glucose itself, such as ¹⁸F-FDG or 2-DG or its fluorescent analogue 2-NBDG, or a dye conjugated to the glucosesuch as in IRDye800-2DG. Of these molecules, ¹⁸F-FDG, 2-DG, and 2-NBDG are known glucose antagonists; however, the behavior of the optical probe IRDye800-2DG to this effect has not been studied or reported. Hence, GPc could behave as either an agonist or an antagonist, such as 2-DG and 2-NBDG, respectively. Bioenergetic profiling revealed that both the cell lines of our choice had similar aerobic energy phenotypes as observed from the basal glycolytic rates reflected by the extracellular acidification rate (ECAR) and mitochondrial oxygen consumption rates (OCRs) suggestive of oxidative phosphorylation (OXPHOS) (Figure A, Supporting Information, Figure S17). To further characterize the behavior of GPc in metabolically active cells, we performed the Seahorse glycolytic stress assay with OVCAR 3 cells. First, we validated the changes in glycolytic rates and glycolytic capacity when, instead of a saturating concentration of glucose (10 mM), a much lower concentration of 1 mM glucose was administered to the glucose-starved cells. We observed that although the ECAR was lower, indicating that the glycolytic rate and reserve were lower than those of 10 mM, the cells functioned in the same manner. That is, upon addition of 1 mM glucose, an increase in ECAR was observed, followed by a slight increase in ECAR upon shutting down the electron transport chain with oligomycin, and then a final decrease in ECAR upon adding 2 DG, the glucose antagonist (Figure B). Since the LD₅₀ of GPc in OVCAR 3 was 1.02 mM, it was imperative to perform the assay with lower concentrations of GPc ranging between 50 and 500 μM. When the assay was performed by switching the substrate from glucose to GPc, we found that the ECAR prominently and transiently increased upon adding 250 and 500 μM GPc and then rapidly decreased to the baseline level before the second step involving the addition of oligomycin (Figure C, Supporting Information, Figure S17). The transient rise in ECAR seemed to initially indicate that the GPc did not entirely behave as a glucose antagonist like 2 DG or ¹⁸F-FDG, as that would have resulted in no increase from basal levels of ECAR upon injection of GPc into the first assay port. Additionally, the observation that the ECAR decreases after the initial jump could infer that although it partly can act as a metabolic source like glucose, a much higher concentration of the GPc would be required to sustain the cells, as observed when we add 10 mM glucose to the starved cells in the initial assay design. To probe further into this and ascertain whether cells could utilize GPc similar to glucose as a source of nutrients, we first checked if GPc and glucose both were substrates for the GLUT1 transporter. When we performed the assay using BAY 786, a known inhibitor of GLUT1, we observed a complete shutdown of the glycolytic pathway and no elevation of ECAR, when glucose was added post incubation with BAY 786 (Supporting Information, Figure S17). This validated the concentration of BAY 786 used (10 μM) and also ascertained that the incubation time with the inhibitor was sufficient to inhibit all GLUT1-mediated transport. Interestingly, in the same assay, with BAY 786, when GPc was used instead of glucose, we observed only a very minute drop in ECAR (Supporting Information, Figure S17). This implied that GPc did not primarily enter the cells through GLUT1 and was not a substrate that would compete with glucose for entry through GLUT1.

4.

(A) Metabolic phenotypic profiling of 4T1 and OVCAR 3 cells through Seahorse glycolytic stress assay, where ECAR and OCR were measured following the sequential addition of saturating concentrations of glucose (10 mM), oligomycin, and 2-DG demonstrating both cell lines to be aerobic. (B) Changes in ECAR in OVCAR 3 cells when the stress assay is performed with a less saturated (more dilute) 1 mM concentration of glucose (blue) compared to the assay standard of 10 mM glucose (red). (C) ECAR rates in OVCAR 3 cells, when GPc is added at 500 and 250 μM instead of glucose as the nutrient substrate to glucose starved cells. (D) ECAR rates in OVCAR 3 cells when both glucose (port A) and GPc (port B) were added in a sequential manner followed by oligomycin and 2 DG. (E) Glycolytic capacitythe maximum ECAR following inhibition of mitochondrial ATP production subtracted from the nonglycolytic acidification rate and (F) glycolytic reserve for OVCAR 3 for the competition assay.

We further ventured to perform competitive assays with glucose and chose the live seahorse method due to the high sensitivity of this assay combined with the display of poor fluorescence by GPc, which discouraged us to study competition through fluorescence-based methods such as flow cytometry, the results from the competition assays (Figure D–F, Supporting Information, Figure S17) shed interesting insights into the pharmacodynamics of GPc. When 20-fold or 40-fold lower concentrations of glucose were injected into port A (500 or 250 μM compared to the standard 10 mM saturating concentration), we observed that the ECAR rates and also the glycolytic capacity and % glycolytic reserves lowered as expected (Figure E,F); however, the glycolysis process did not entirely shut down, with ECAR values diminishing to baseline levels over time, as observed when only GPc was injected into port A (Figure C). More importantly, when we assess the difference in ECAR with and without injection of GPc (250 μM) in port B, following the injection of glucose (1 mM) in port A, we observed a significant decrease in the ECAR, the glycolytic capacity, and the % glycolytic reserve only when GPc is injected (Figure E,F, Supporting Information, Figure S17b,d). This suggests an antagonistic role that GPc is playing, creating a definite constraint in the glycolytic pathway that leads to these decreased metabolic parameters in a consistent manner across concentrations of GPc. These metabolic evaluations for the small molecule contrast agent GPc suggested a similar antagonistic nature to ¹⁸F-FDG, where the probe is readily taken up in metabolically active cells but gets trapped once inside and does not metabolize readily like glucose.

3.4. In Vivo PA Imaging with GPc in 4T1 Tumor-Bearing Mice

Given the promising attributes of GPc as a PA contrast agent for imaging metabolic activity, we aimed to evaluate this probe in a 4T1 mouse tumor model. We performed longitudinal and cross-sectional multispectral optoacoustic tomography (MSOT) imaging when the tumor size was approximately 300–400 mm³ in a series of animals (Figure A). Detailed tomography images were acquired for 12 wavelengths (660 nm–880 nm) with 20 nm wavelength intervals and 5 frames per wavelength within a selected volume that included the tumor, liver, and kidney. A single wavelength (800 nm) image of the 2D slice shows the tumor (Figure B, Supporting Information, Figure S18) within the anatomy, and at 870 nm, the anatomical resolution is best due to the compromise between tissue light penetration and absorption of water, which maximizes after 900 nm. Reconstruction of the 2D slices, at the postprocessing step, generated an image with a final spatial resolution of 150 μm and spectral unmixing following linear regression models. Longitudinal imaging (Cohort 1) post intravenous (i.v) tail-vein administration of GPc in the 4T1 model, showed a gradual increase in agent accumulation from 30 min to maximum accumulation of the GPc within 2 to 4 h postinjection (n = 5) (Figure B, Supporting Information, Figure S18). The PA intensity corresponding to the GPc concentration in the tumor obtained from an overlay of the GPc map on the grayscale anatomy image exhibits excellent distribution and localization of the agent inside the tumor, in contrast to IRDye800-2DG, which showed elevated uptake at the outer tumor surface but minimal uptake inside the A431 tumor. We also evaluated the endogenous chromophores Hb (deoxygenated hemoglobin), HbO₂ (oxygenated hemoglobin), and HbT (total hemoglobin), and observed a positive correlation of GPc uptake to the total hemoglobin content (HbT), thereby signifying the ability of the contrast agent to image a hypermetabolic tumor (Figure B, Supporting Information, Figure S18). Additionally, the core of the tumor displayed a stronger signal for Hb compared to that of HbO₂, implying a poor oxygen distribution in the core. However, impressively, the agent can permeate into the tumor, and the low oxygen gradient is not a deterrent to the uptake of the agent in hypoxic regions of the tissue (Figure B, Supporting Information, Figure S18).

5.

Experimental design and timeline for the in vivo and ex vivo imaging with (A) details of 4T1 tumor implantation in athymic nude mice in different mice for serial imaging (Cohort 1) and cross-sectional imaging (Cohort 2 and 3) and ex vivo imaging of tissues post excision and in vivo studies. (B) In vivo PA images of a 2D slice displaying the anatomy in x–y plane with the tumor visible (circled in yellow), uptake of GPc (jet) overlaid on the anatomy (gray) in two mice at 30 min and 2 or 4 h post intravenous administration of GPc, and an overlay of the intrinsic contrast of deoxyhemoglobin (Hb, blue), and oxyhemoglobin (HbO₂, red) and total hemoglobin (HbT) in the same mice.

3.5. Biodistribution and Clearance of GPc through In Vivo and Ex Vivo Imaging

From the serial in vivo imaging studies, we observed the agent uptake in the tumor, liver, and kidneys from 30 min to 4 h postadministration (Figures B and A,B). The metabolic fate of GPc in the body postadministration evaluates partially the pharmacokinetics (PK) of the probe, and our observation of the highest PA signal at 4 h provided evidence that a substantial amount of the administered GPc remained undegraded within the cell or did not get transported out of the cell, contributing to the PK of GPc. To evaluate the biodistribution and clearance of the agent, we focused on assessing the distribution of the agent in the nontarget clearance organs, such as the liver and the kidney, from the real-time in vivo imaging data (Figure B). As expected, once GPc is injected i.v. through the tail vein, it perfuses quickly through the vasculature into the organs, such as the liver and kidney, and into the tumor. In the initial time points (30 min and 2 h), the uptake is similar in the liver, kidney, and the tumor; however, by 4 h, the tumor retains a much higher amount of the GPc (Figure C,D). We followed the clearance of the agent by quantifying the GPc PA signal in a separate subset of 4T1 mice at 12 h (n = 3) and 24 h (n = 2) postadministration (Figures A and D), wherein we observed the complete washout of the probe from the tumor within 12 h (Figure A, Supporting Information, Figure S19). Quantification of the in vivo uptake supported this claim, as it showed that the concentration of GPc is significantly lower in the tumor at 12 and 24 h compared to 4 h (p = 0.008, p = 0.01) (Figure C). In the clearance organs, mainly the liver and kidney, quantification of PA signals at 12 and 24 h also suggested the probe getting cleared from the body, and the in vivo data at 12 and 24 h from these two organs could not clearly validate the preferred route of clearance (Figure D). Further, through ex vivo PA imaging of the excised tumor, liver, spleen, and kidney tissues at time points of 6 h (n = 4), 12 h (n = 3), and 24 h (n = 2) post-GPc administration, we visualize the agent is getting cleared out within 24 h. The ex vivo images suggest the agent’s retention to be higher in the liver and spleen than the kidneys, suggesting favorable elimination through the hepatobiliary rather than the renal route (Figure E). Although small molecules are preferably cleared through the kidneys, the structure of GPc has a core phthalocyanine unit that renders a certain extent of hydrophobicity to the molecule, and hydrophobic small molecules often are metabolized and excreted through the liver. Additionally, GLUT expression is elevated in the liver compared to other tissues. This could be a plausible reason for the higher retention of GPc in the liver if its entry is partially GLUT-mediated, as observed for IRDye800-2 DG. There is no significant GPc signal in the ex vivo tumor tissue after 12 and 24 h, implying the washout of the agent from the tumor site. This highlights the feasibility of using this small molecule as a contrast agent for imaging metabolic tumors.

6.

(A) The pharmacokinetics of uptake and washout of GPc in and from the tumor through in vivo real-time PA imaging longitudinally at different time points from 30 min to 12 h postadministration of GPc (jet) and (B) the biodistribution profiles in the upper and lower abdominal region, especially in the liver and kidney (marked on anatomical slice, yellow dash) across time, the images to the left of the white dashed line indicate images from the same mouse, whereas the 12 and 24 h in vivo biodistribution profile are from mice that were cross-sectionally imaged (Cohort 2 and Cohort 3). (C) The quantification of GPc uptake and washout in and out of the tumor from the mean PA intensity of the tumor volume of interest (VOI). (D) Quantification of GPc uptake and retention in the clearance organs of liver and kidney from the in vivo images across time. (E) PA contrast images of GPc (jet) overlaid on the grayscale anatomical image of the excised tissues at 6 h, 12 h, and 24 h post GPc administration.

3.6. Serum Biochemistry and Histology

To evaluate the safety and any potential toxicity of the agent to off-target organs, we conducted serum biochemistry analysis and histology studies. We used six additional nontumor-bearing athymic nude mice for these studies. For histology analysis, we also used the excised liver, kidney, and spleen from two 4T1 mice used in our imaging studies (Cohort 3). H&E staining of the liver, kidney, and spleen was assessed to evaluate any visible damage to these organs upon GPc retention and circulation in the body for long periods of time. No signs of damage were observed in the liver tissues from control mice, mice that received one and two doses of GPc, and 4T1 tumor-bearing mice with one dose of GPc followed by imaging at 24 h. In all cases, the liver exhibited a normal lobular architecture with hepatocytes arranged in hepatic cords radiating from a central vein and well separated by blood sinusoids (Figure A). Similarly, the kidney of GPc injected mice did not show any signs of histopathological changes in the glomeruli and tubules as compared to the control mouse (Figure B). The spleens of both control and GPc injected mice also showed a monotonous sheet of well-organized red pulp and sinusoids of similar morphology (Figure C). The results from serum biochemistry show similar values for the important enzyme parameters and the total protein in both control mice and the mice that received repeated two doses of GPc (day 1 and day 3). These results suggested no major alterations in organ function upon GPc administration (Figure D).

7.

Histopathology results show the H&E staining of the liver (A), kidney (B), and spleen (C) from the mice. The control group (saline), nontumorigenic mice with GPc single (+) and double dosing (++), and GPc injected 4T1 tumor-bearing with the dosage used in biodistribution studies. CV: central vein, PV: portal vein, S: sinusoids, H: hepatocytes, G: glomerulus, BC: Bowman’s capsule, T: tubule, RP: red pulp, F: follicle. N = 2 per group, magnification (20×), scale bar 25 μm. (D) Serum biochemistry profiles in 2 control mice and 2 nontumorigenic mice that received two doses of GPc. ALP = alkaline phosphatase, AST = aspartate aminotransferase, and ALT: alanine aminotransferase.

4. Conclusions

In summary, we demonstrated the successful development of a small-molecule, water-soluble PA contrast agent, GPc, that is readily taken up in metabolically active tumors and can promote the noninvasive visualization of highly metabolic sites without the involvement of any radioactivity or ionizing radiation. Importantly, for the first time, we elucidated the mechanism of action of a glucose-conjugated contrast agent with live metabolic assays. We validated through careful Seahorse experiments and subsequent analysis of glycolytic function and changes in ECAR with and without glucose addition that GPc functions as an antagonist similar to the clinical PET tracer, ¹⁸F-FDG, as it severely constrains the glycolysis process and utilization of glucose once it is administered, leading to its accumulated entrapment inside cells. Although our studies have provided ample evidence of GPc entering cells, further studies are warranted to understand the exact mechanism of its entry into cells. Our experiments in the presence of GLUT1 inhibitor BAY-786 also validated that the major mode of entry is not through GLUT1.

However, apart from GLUT1, there are 11 other GLUT transporters expressed in various tissues, and most importantly, GLUT3 is another important transporter for which glucose as well as ¹⁸F-FDG is a substrate. Therefore, more studies are needed to validate if the entry of GPc into cells could be partially mediated by other GLUTs or mediated through passive diffusion. Another interesting aspect that could be evaluated in future studies is the fate of GPc once it enters the cells. This would explain why and how GPc transiently increases the ECAR before completely restricting the glycolytic pathway. Our in vivo and ex vivo studies in a metabolically active 4T1 mouse tumor model demonstrated excellent uptake and retention of GPc at the tumor site until 6 h and finally washout from the tumor by 12 h, thereby achieving high translatable potential for precise in vivo imaging of metabolic activity. One limitation in our in vivo biodistribution profiles is that we were not able to observe the spleen in our in vivo PA images distinctly, so we have only analyzed slices that show the liver and the kidney profiles. Due to the exceptional PA contrast using this molecule, we hypothesize that it will enable imaging of enhanced metabolic activity in smaller lesions in future studies. PA tomography can resolve metabolic signatures from endogenous hemoglobin contrast, as shown in our findings as well; therefore, a combined multiplexed approach of molecular imaging of glucose metabolism with both endogenous and exogenous contrast offers a promising strategy for live metabolic imaging up to a tissue depth of ∼5 cm. Combining these functional data with anatomical information from the same image could be a potential platform for longitudinal noninvasive imaging of metabolic lesions.

Glossary

Abbreviations

2DG

2-deoxy-d-glucose

¹⁸F-FDG

¹⁸F-fluoro 2-deoxy glucose

ATCC

American type culture collection

ATP

adenosine triphosphate

ALP

alkaline phosphatase

ALT

alanine aminotransferase

AST

aspartate aminotransferase

BP

back projection

BSA

bovine serum albumin

BUN

blood urea nitrogen

CT

computed tomography

ECAR

extracellular acidification rate

ESI

electrospray ionization

FACS

fluorescence-activated cell sorting

FBP

field back projection

FBS

fetal bovine serum

FSC

forward scatter

HPLC

high-performance liquid chromatography

HRP

horseradish peroxidase

i.v

intravenous

ICG

indocyanine green

MFI

median fluorescence intensity

MRI

magnetic resonance imaging

MSOT

multispectral optoacoustic tomography

NIR

near infrared

OCR

oxygen consumption rate

OXPHOS

oxidative phosphorylation

PET

positron emission tomography

PFA

paraformaldehyde

PI

propidium iodide

RM

repeated measures

ROI

region of interest

RP

reverse phase

SRB

sulforhodamine B

SSC

side scatter

TLC

thin layer chromatography

TOF

time of flight

VOI

volume of interest

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c00151.

• Details of the synthesis procedures for compounds I, II, and GPc; ¹H and ¹³C NMR spectra and HRMS characterization as well as the final HPLC chromatogram of GPc; experimental details of photophysical characterization; stability studies of GPc (PBS) through absorbance, calculation of GPc’s FQY, and variable-temperature PA studies; experimental details of biochemical stability studies of GPc in mouse serum; cytotoxicity analysis from the SRB assay on hTERT FT282 cells; experimental details of Western blot and images and analysis; flow cytometry studies of GPc with OVCAR 3 and 4T1 cells including the overlay plots; additional images of the epifluorescence microscopy studies including the additional time points 1 and 8 h and the nuclear staining; additional experimental detail of metabolic assays, including general Seahorse stress assay profile, Seahorse experiment with GLUT1 inhibitor BAY 786, and data on other concentrations for the competition assay; and in vivo PA imaging data on all mice (PDF)

S.S. and P.P. conceived the research and P.P., A.S.V., S.K.P., C.S., and V.V. performed the experimental work. P.P., A.S.V., S.K.P., A.S., and V.V. analyzed the data and discussed the results. S.S. and P.P. wrote the final manuscript. All authors have approved the final version of the manuscript. CRediT: Pooja Annasaheb Patkulkar data curation, formal analysis, methodology, project administration, validation, visualization, writing - review & editing; Arjun S.V. data curation, formal analysis, methodology, software, validation, visualization, writing - review & editing; Ananya Sharma software, validation, visualization; Suvam Panda data curation, validation, writing - review & editing; Vinay V data curation, methodology; Chandan Shringi methodology, resources; Sanhita Sinharay conceptualization, funding acquisition, investigation, project administration, resources, supervision, validation, visualization, writing - original draft, writing - review & editing.

S.S. acknowledges the support from the Indian Institute of Science and the Mazumdar Shaw endowment for the RI Mazumdar Young Investigator in Bioengineering. S.S. also acknowledges funding from the Science and Engineering Research Board (SERB), Department of Science and Technology, DST, India, Grant Number SPG/2021/002503 and Scheme for Translational and Advanced Research in Sciences, STARS, Ministry of Human Resource Development (MHRD), India, Grant Number: SP/MOES/23-0855.

The authors declare no competing financial interest.