Responsive Fluorinated Nanoemulsions for ¹⁹F Magnetic Resonance Detection of Cellular Hypoxia

Abstract

We report two highly fluorinated Cu-based imaging agents, CuL₁ and CuL₂, for detecting cellular hypoxia as nanoemulsion formulations. Both complexes retained their initial quenched ¹⁹F MR signals due to paramagnetic Cu²⁺; however, both complexes displayed a large signal increase when the complex was reduced. DLS studies showed that the CuL₁ nanoemulsion (NE CuL₁) had a hydrodiameter of approximately 100 nm and that it was stable for four weeks post-preparation. Hypoxic cells incubated with NE CuL₁ showed that 40% of the Cu²⁺ taken up was reduced in low oxygen environments.

Graphical Abstract

A highly fluorinated Cu2⁺ complex for ¹⁹F MR sensing of cellular hypoxia as nanoemulsion formulations.

Introduction

Hypoxia in solid tumor cancers results from inadequate O₂ delivery to these rapidly dividing cells, resulting in O₂ deficiency. This leads to increased levels of hypoxia inducible factor (HIF-1), which regulates the cells’ ability to adapt to the new environment.^{1, 2} If left untreated, the cells can become resistant to chemotherapy and cause malignant proliferation.^3–6 Therefore, early diagnosis of hypoxic cancer cells is vital for tumor excision to avoid metastasis and secondary malignant tumor growths. An additional effect of hypoxia is a more reducing intracellular environment, which can be used for selective targeting of therapeutic and diagnostic agents. Among the imaging agents that have been developed to target hypoxia,^{7–14 64}CuATSM (ATSM = diacetyl-bis(N⁴-methylthiosemicarbazone)) has been used as a positron emission tomography (PET) agent that functions via reduction of the Cu²⁺ complex and retention in hypoxic cells, however this agent requires the use of radioactive materials.^{15, 16}

Magnetic resonance imaging (MRI) is the most widely used imaging modality to diagnose cancer and can be used to image whole organisms with high depth penetration without employing ionizing radiation. However, early tumor detection is difficult as the spatial and contrast resolution between cancerous growths and surrounding normal tissue is poor. As an emerging alternative, ¹⁹F MRI can be used as there is no detectable fluorine in the body and, thus, any signal present will originate from exogenous agents. Moreover, the ¹⁹F nucleus provides comparable characteristics: 100% isotopic abundancy, nuclear spin of ½, 83% MR signal receptivity compared to ¹H, and further, the carbon-fluorine bond is biostable.¹⁷

Previously, we have demonstrated the use of fluorinated CuATSM imaging agents as “turn-on” probes for cellular hypoxia.^7–9 These agents use paramagnetic Cu²⁺, which serves as a powerful paramagnetic relaxation enhancement (PRE) source that attenuates ¹⁹F MR signal due to T₂ shortening.^{7–9, 17–21} Following reduction to Cu⁺ in hypoxic cells and subsequent demetallation, the ¹⁹F signal is fully restored, furnishing a signal turn-on in this environment. For these agents to be a viable option for in vivo studies, the need for elevated fluorine concentration on individual probes increases, to allow an overall brighter MRI signal. Unfortunately, the hydrophobic CuATSM scaffold and fluorine atoms decrease aqueous solubility. Therefore, synthesizing a highly fluorinated CuATSM complex also requires a new delivery formulation: nanoemulsions.

Nanoemulsions are nano-sized liquid particles that consist of an oil droplet core, which assists in the dissolution of hydrophobic molecules, and a layer of emulsifier, which is necessary to reduce excess interfacial energy and prevent aggregation (Figure 1).^{22, 23} Common emulsifiers are amphiphilic molecules such as phospholipids and pegylated molecules.²⁴ Importantly, a number of ¹⁹F MRI nanoemulsion formulations have been employed to deliver perfluorinated carbon-based agents for in vivo imaging.^25–29 By encapsulating the prepared Cu²⁺-based probes into nanoemulsions, we envisioned that the complexes would maintain high fluorine spin density while staying miscible within the aqueous environment. Herein, we present two CuATSM-derived complexes, CuL₁ and CuL₂ (Figure 1), with 18 and 36 equivalent fluorines, respectively. The CuL₁ complex forms stable nanoemulsions that display “turn-on” in ¹⁹F MR modalities upon reduction, and preferential signal turn-on in hypoxic cells.

Figure 1.

(A) Chemical structures for CuL₁ and CuL₂. (B) Oil-in-water- nanoemulsion composition.

Results and discussion

Structural characterization

CuL₁ and CuL₂ incorporate multiple perfluoro-tert-butyl units to confer high fluorine density and a glucosamine moiety to tune the overall lipophilicity of the whole complex. Their syntheses are described in full in the Supporting Information. Single crystals of CuL₁ were grown by evaporation of a concentrated ethanol solution at room temperature and brown, needle-like crystals were collected. The X-ray structure (Figure S1) confirmed a square-planar Cu²⁺ center embedded in the [N₂S₂] pocket, similar to the parent CuATSM complex.³⁰ Interestingly, hydrogen bonds were observed between two different glucosamine motifs within the crystal packing. The average distance between Cu²⁺ and ¹⁹F nuclei was measured as 8.4 Å, a distance at which Cu²⁺ would attenuate the ¹⁹F MR signal.¹⁷ Single crystals of CuL₂ did not form.

Solution state characterization

¹⁹F NMR characterization.

To characterize the effect of Cu²⁺ on the ¹⁹F NMR signal in our complexes, we obtained spectra and measured relaxation times of H₂L₁, CuL₁, H₂L₂, and CuL₂. The ¹⁹F NMR spectra (Figure S2) of CuL₁ and CuL₂ demonstrated that Cu²⁺ broadens the ¹⁹F signal in these complexes as compared to their respective ligands H₂L₁ and H₂L₂. The relaxation times (Table 1) of the fluorine atoms decreased by roughly 30-fold for T₁ and 100-fold for T₂. The larger fold of decrease in T₂ is consistent with the longer electronic relaxation time T_1e of the square planar Cu²⁺ center.^{31, 32} The singlet peak observed for all prepared ligands and complexes indicated that the fluorine atoms are all magnetically equivalent, which is a benefit for maximizing the signal-to-noise ratio (SNR) in ¹⁹F MR-based sensing.

Table 1 -.

¹⁹F NMR parameters for 3 mM H₂L₁, CuL₁, H₂L₂, and CuL₂ in d₆-DMSO at room temperature at 9.4 T.

	H2L1	CuL1	H2L2	CuL2
δ (ppm)	−70.0	−69.9	−70.0	−70.0
T1 (ms)	590	20.4	675	22.2
T2 (ms)	360	N/Aa	333	N/Aa
T2* (ms)	157	4.3	78.2	2.2

^aDue to the fast transverse relaxation for CuL₁ and CuL₂, the T₂ was too short to be measured.

Cyclic voltammetry.

To understand the redox properties of CuL₁ and CuL₂, the Cu²⁺/Cu⁺ redox potentials were determined by cyclic voltammetry in DMF (Figure S3). Measured half potentials of −0.62 V for CuL₁ and −0.60 V for CuL₂ were close to the reported values for the parent CuATSM complex (−0.63 V vs. SCE), revealing the potential for these complexes to target hypoxic cells.³³ Due to the presence of the quaternary carbon, the electron-withdrawing perfluoro-tert-butoxide and glucosamine moiety did not significantly change the reduction potential of the Cu²⁺ centers. This observation also indicates other R groups (instead of glucosamine) could be incorporated into this molecular scaffold to further functionalize or solubilize the complex. The ΔE_p values of CuL₁ and CuL₂ indicate that the reduction of CuL₁ is quasi-reversible while the reduction of CuL₂ is irreversible. Considering the different coordination preferences between Cu⁺ and Cu²⁺, the large bulkiness of the functional groups on H₂L₂ could result in a lower stability for [Cu⁺L₂] than [Cu⁺L₁], thus making the reduction process for CuL₂ less reversible than CuL₁.

Preparation of nanoemulsion

Formulation.

To improve incorporation of these complexes into aqueous media, an oil-in-water nanoemulsion formulation strategy was employed. H₂L₁, CuL₁, H₂L₂, and CuL₂ nanoemulsions (namely, NE H₂L₁, NE CuL₁, NE H₂L₂, and NE CuL₂) were prepared following published literature.²⁶ Lecithin, Milli-Q water, and safflower oil were mixed and heated at 80 °C to form the emulsion mixture. H₂L₁, CuL₁, H₂L₂, and CuL₂ were dissolved in DMSO and the hot, pre-made emulsion was added directly to the DMSO solution, vortexed to create a crude emulsion, and ultrasonicated at 0 °C to afford a stock nanoemulsion (Scheme S2). Unfortunately, upon reduction of NE CuL₂ with Na₂S₂O₄, both NE CuL₂ and the reduced NE CuL₂ gave similar T₂* values and chemical shifts, making them difficult to differentiate under MR settings. Therefore, all future studies were performed with NE CuL₁.

Size determination.

The size distribution of NE CuL₁ was evaluated by dynamic light scattering (DLS). In water and various buffered environments, NE CuL₁ displayed a hydrodynamic diameter of ~100 nm with PDI ≤ 0.20 (Table 2). These results are consistent with reported results on similar systems^{34, 35} and suggest effective formation of the nanoemulsion formulation. Transmission electronic microscopy (TEM) was employed to visualize the morphology of the prepared NE CuL₁, using neutral ammonium phosphomolybdate staining to improve contrast.²³ Therefore, as shown in Figure 2, the nanoemulsions appeared as bright sphere-like structures with a size distribution of 81 ± 19 nm, well correlated with DLS results. Energy-Dispersive X-ray Spectroscopy (EDS) of a single nanoemulsion revealed an even inner distribution of copper and sulfur elements, further confirming the successful preparation of a nanoemulsion containing the CuL₁ complex.

Table 2 –

Size distribution of NE CuL₁ in water and different buffers determined by dynamic light scattering (DLS).

Solvent/Buffer	Average Size (nm)	PDI
Water	93.3 ± 0.7	0.20
PBS	94.5 ± 0.3	0.18
HEPES	97.9 ± 0.2	0.20
RPMI (w/o FBS)	98.0 ± 0.4	0.19
DMEM (w/o FBS)	105.7 ± 0.8	0.16

Figure 2.

(A) Representative TEM image of NE CuL₁ negatively stained with neutral ammonium phosphomolybdate. Inset: Expanded view of a single nanoemulsion. (B) Size distribution of NE CuL₁ (n=104). (C) Copper and sulfur elemental profiling of a single NE CuL₁ particle by Energy-Dispersive X-ray Spectroscopy (EDS).

Nanoemulsion stability.

The stability of both ligand and complex nanoemulsions was assessed by DLS (size) and ¹⁹F NMR (fluorine content). The stability of the NE CuL₁ was also assessed by inductively coupled plasma-optical emission spectroscopy (ICP-OES) to evaluate the copper content within the nanoemulsion. As shown in Figure S4, at 100 μM [Cu²⁺], the prepared NE CuL₁ displayed great aqueous stability with marginal change (< 5%) in both its average size and the polydispersity index (PDI). However, at 5 mM [Cu²⁺], the NE CuL₁ is prone to aggregation and showed an increase in its average hydrodynamic diameter and a decrease in PDI. Interestingly, although the size distribution of 5 mM NE CuL₁ changed over the course of 28 days, the copper leaching was minimal (< 5%), similar to the 100 μM NE CuL₁. Additionally, a 100 μM sample of the NE H₂L₁ was subjected to DLS and ¹⁹F NMR analysis over the course of four weeks after preparation. Compared to the NE CuL₁, the leaching of H₂L₁ within NE H₂L₁ was much faster when studied by ¹⁹F NMR, especially at a higher ligand concentration (Figure S5). While the 0.5 mM NE H₂L₁ showed 6% leaching after two weeks, the 5 mM NE H₂L₁ leaching increased to 20% during the same time. The increased cargo leaching for NE H₂L₁, as compared to NE CuL₁, is likely due to a more polar nature of H₂L₁ that encouraged its escape from the nanoemulsion.

Relaxation time determination.

To understand the MR properties of CuL₁ and H₂L₁ within their nanoemulsion environments, ¹⁹F NMR spectra were taken for each nanoemulsion (Figure 3A). NE H₂L₁ displayed an intense singlet peak at −70.5 ppm. On the other hand, NE CuL₁ gave a very broad peak at −70.7 ppm. Fluorine relaxation times (T₁ and T₂) were measured for both nanoemulsions (Table 3). NE H₂L₁ had a T₁ of 380 ms and a T₂ of 7.0 ms, which was much shorter compared to that of the DMSO solution of H₂L₁. The large decrease in T₂ is likely due to the viscous safflower oil core of the nanoemulsion and potential intermolecular aggregation due to hydrogen bonding between the glucose motifs as observed in the crystal packing. The T₁ relaxation time of NE CuL₁ was too short to measure and the T₂* was 0.4 ms, consistent with its broad and nearly quenched signal.

Figure 3.

(A) ¹⁹F NMR spectra of 1.0 mM NE H₂L₁, NE CuL₁, and reduced NE CuL₁. (B) Phantom ¹⁹F MRI of 1.0 mM NE H₂L₁, NE CuL₁, and reduced NE CuL₁.

Table 3 –

¹⁹F NMR parameters of 0.5 mM NE H₂L₁, NE CuL₁, and reduced NE CuL₁ at room temperature.

	NE H2L1	NE CuL1	NE CuL1 + Na2S2O4
δ (ppm)	−70.5	−70.7a	−70.5
T1 (ms)	380	N/Ab	440
T2 (ms)	7.0	N/Ab	4.4
T2* (ms)	7.0	0.4	4.4

^aThe peak for NE CuL₁ was broad.

^bDue to the fast transverse relaxation for NE CuL₁, the T₁ and T₂ were too short to be measured.

Redox behavior.

To determine if the chemical reduction of CuL₁ would successfully happen inside the nanoemulsion, 1.0 mM test-tube reactions between Na₂S₂O₄ and NE CuL₁ were carried out (Figure 3A). Upon addition of Na₂S₂O₄, the characteristic orange-brownish color of CuL₁ disappeared immediately, and the whole nanoemulsion system turned off-white, consistent with reduction of Cu²⁺ to Cu⁺. The ¹⁹F NMR characteristics of the reduced system were similar to the NE H₂L₁ (δ = −70.5 ppm; T₁ = 0.44 s; T₂ = 4.4 ms) (Table 3). Therefore, while being encapsulated inside the nanoemulsion, CuL₁ was likely converted to H₂L₁ upon reduction. We note that there was no significant change to the ¹⁹F NMR of NE H₂L₁ in the presence of Na₂S₂O₄ (Figure S6).

The reduction of the Cu²⁺ center within the nanoemulsion was further confirmed via UV-vis absorption and EPR spectroscopy (Figures S7). Before reduction, the UV-vis absorption spectrum of NE CuL₁ displayed a characteristic Cu²⁺ d-d absorption peak at 480 nm³⁶ and the EPR spectral pattern was well correlated to a square planar Cu²⁺ center^{37, 38} with g_iso = 2.061 and A_Cu = 106 G. Post reduction, all these spectral features were lost, consistent with reduction of Cu²⁺. Importantly, the UV-vis absorption spectrum of NE CuL₁ was recovered upon exposing the reduced NE CuL₁ to air, indicating that the binding of Cu²⁺ by H₂L₁ was not perturbed by the presence of the nanoemulsion formulation.

Phantom MR images.

Phantom ¹⁹F MR imaging for 1.0 mM NE H₂L₁, NE CuL₁, and reduced NE CuL₁ were performed. Fast-low-angle-shot (FLASH) pulse sequence was applied to allow tracking of species with short T₂ values. As shown in Figure 3B, Cu²⁺ effectively quenches the ¹⁹F MR signal and the signal-to-noise ratio for NE CuL₁ was on the same level as noise (SNR = 2.2). On the other hand, an intense signal was captured for the NE H₂L₁ (SNR = 12) and the reduced NE CuL₁ (SNR = 7.6), each with the same ¹⁹F concentration as the NE CuL₁. These results demonstrated a “turn-on” response in the ¹⁹F MR imaging modality when CuL₁ is reduced and converted to H₂L₁ within nanoemulsion formulations and hold promise for potential ¹⁹F MR imaging-based studies.

Cell studies

Cytotoxicity and cell uptake.

To evaluate the nanoemulsion’s ability to act as a biological probe, cell studies were performed with MCF-7 breast cancer cells. Cytotoxicity of the NE CuL₁ was tested using a Live/Dead assay under both normoxic and hypoxic conditions. Fluorescence imaging data showed >95% viability of the nanoemulsion incubated cells in both normoxic and hypoxic environments (Figure S8). To track the copper content of the MCF-7 cells, cell uptake studies were performed on normoxic and hypoxic cells after 2, 4, and 6-hour incubation (ICP-OES). As shown in Figure S9, a gradual increase in copper uptake was observed for increasing incubation times. At 6 hours, the cellular copper level was 3.2 ± 0.1 fmol/cell, which should give a cellular fluorine content of ~60 fmol/cell. When comparing the copper uptake between normoxic and hypoxic conditions, we saw no differences. This similarity could be due to the fact that CuL₁ was encapsulated in the nanoemulsion and uptake and retention of the nanoemulsion is not dependent on the oxygen level. Uptake studies at 4 °C, a temperature at which energy-dependent active transportation is blocked, showed 9-fold less cellular copper content after 4 hours compared to studies at 37 °C, consistent with an energy-dependent cell uptake pathway. The energy-dependent cell uptake of the NE CuL₁ correlated well with the vesicular character of these formulations.²⁴

To visualize the uptake of NE CuL₁, MCF-7 cells were incubated with a fluorescent ZnSalen complex that has been reported to stain the hydrophobic interior of lipid droplets (Figure 4A).³⁹ MCF-7 cells were incubated with NE CuL₁ only, with the fluorescent dye only, and with both fluorescent dye and NE CuL₁, respectively. As expected, no fluorescence was observed within the cells when only NE CuL₁ was administered (Figure 4A top). With incubation of the fluorescent dye itself, cellular fluorescence was weak (Figure 4A middle). When the cells were treated with both NE CuL₁ and the fluorescent dye (Figure 4A bottom), an intense intracellular fluorescence was observed, exhibiting an enhancement of fluorescence by nearly 3-fold compared to cells incubated with the fluorescent molecule only (Figure 4B). A closer look at the image revealed dot-like red fluorescence inside the cytoplasm, corresponding to the vesicular character of the nanoemulsions. These results further confirmed efficient uptake of the nanoemulsions into cells.

Figure 4.

(A) Confocal images of MCF-7 cells incubated with NE CuL₁ (top row), the fluorescent dye (middle row), and both NE CuL₁ and the fluorescent dye (bottom row). (B) Quantitative comparison of intracellular fluorescence of ~100 cells at different incubation conditions. (C) Whole cell ¹⁹F NMR of NE H₂L₁ (top) and NE CuL₁ (middle) in normoxic (20% O₂) environment and NE CuL₁ (bottom) in hypoxic (0.1% O₂) environment.

Cellular hypoxia reduction studies.

NE CuL₁ was employed for hypoxia detection through ¹⁹F NMR spectroscopy (Figure 4C). As a positive control, normoxic MCF-7 breast cancer cells were first cultured with 100 μM NE H₂L₁ for 4 hours. The cells were collected, transferred into an NMR tube, and a ¹⁹F NMR spectrum was taken with 5-fluorocytosine (5FC) as external reference. A broad peak was recorded at −70.6 ppm. This result indicated that inside the cytosol, the NE H₂L₁ stayed intact and its fluorine signal was detectable. Additional MCF-7 cells were incubated with NE CuL₁ under both normal (20%) and low (0.1%) oxygen tension. Only under the low oxygen tension was a peak at −70.6 ppm observed, indicating the selective reduction of NE CuL₁ in the cells grown under hypoxic conditions. At 20% O₂, no ligand signal was detected; the trace signal of perfluoro-tert-butanol might have come from the slight decomposition of the complex inside the cells. To quantify the cellular reduction of copper inside the cells, the reduced copper content was estimated by ¹⁹F NMR spectroscopy (where the peak integration represents the fluorine content of the reduced complex and therefore the amount of copper follows the molar ratio between Cu and F with CuL₁, which is 1:18), and the total copper content quantified via ICP-OES. It was therefore determined that roughly 40% of CuL₁ was reduced in the cytosol under hypoxic conditions.

Conclusions

In summary, we demonstrated a CuATSM-based nanoemulsion sensor system that selectively displays ¹⁹F NMR signal in hypoxic cells. CuATSM derivative, CuL₁, was embedded within oil-in-water nanoemulsions which displayed ideal morphology and aqueous stability. Switching of the ¹⁹F MR signal via tuning of copper redox state was demonstrated by adding chemical reducing agents and the effective reduction of copper was verified by spectroscopic methods. Selective detection of cellular hypoxia was further achieved in breast cancer cells grown under both normoxic (20% O₂) and severe hypoxic (0.1% O₂) conditions, where ~40% of the Cu²⁺ uptaken (~ 1.3 fmol/cell) was estimated to be reduced by cellular machinery. These results thus provide solid evidence that responsive oil-in-water nanoemulsion systems can be used to detect changes in cellular environments using ¹⁹F magnetic resonance. Ongoing work includes synthesizing a complex with a larger T₂* and higher cellular uptake for in vivo applications.