Chorioretinal Hypoxia Detection Using Lipid-Polymer Hybrid Organic Room-Temperature Phosphorescent Nanoparticles

Abstract

Ischemia-induced hypoxia is a common complication associated with numerous diseases and is the most important prognostic factor in retinal vein occlusions (RVO). Early detection and long-term visualization of retinal tissue hypoxia is essential to understand the pathophysiology and treatment of ischemic retinopathies. However, no effective solution exists to evaluate extravascular retinal tissue oxygen tension. Here, we demonstrate a lipid-polymer hybrid organic room-temperature phosphorescence (RTP) nanoparticle (NP) platform that optically detects tissue hypoxia in real-time with high signal-to-noise ratio. The fabricated NPs exhibit long-lived bright RTP, high sensitivity toward oxygen quenching, and desirable colloidal and optical stability. When tested as a hypoxia imaging probe in vivo using rabbit RVO and choroidal vascular occlusion (CVO) models via intravitreal and intravenous (IV) injection respectively, its RTP signal is exclusively turned on where tissue hypoxia is present with a signal-to-noise ratio of 12.5. The RTP NP platform is compatible with multimodal imaging. No ocular or systemic complications are observed with either administration route. The developed organic RTP NPs present a novel platform approach that allows for biocompatible, non-destructive detection of tissue hypoxia and holds promise as a sensitive imaging tool to monitor longitudinal tissue oxygen levels and evaluate various hypoxia-driven vascular diseases.

Graphical Abstract

1. Introduction

Ischemia-induced hypoxia is a common complication that can lead to neovascularization and severe vision impairment in several diseases, including RVO,^1,2 proliferative diabetic retinopathy (PDR), sickle cell retinopathy (SCR), retinopathy of prematurity (ROP), and choroidal ischemia.^3–5 RVO is the 2^nd most common retinal vascular disorder and represents a major cause of vision loss, affecting more than 16 million people worldwide.^6–8 In RVO, the retina within the affected occluded retinal vasculature can become ischemic and thereby become hypoxic.⁹ RVO is a very heterogeneous disease with highly variable visual acuity outcomes. It is essential that physicians be able to prognosticate outcomes with patients to counsel them on treatments and set appropriate expectations. Therefore, it is critical to detect ischemia-induced hypoxia, the most important pathogenic and prognostic factor of RVO, to better understand RVO pathogenesis.⁸ In addition, long-term visualization and quantification of retinal hypoxia are strongly desirable and can lead to a better understanding of the pathophysiology of ischemic retinopathies, including RVO, PDR, SCR, and ROP.

Recently, several techniques have been developed to monitor hypoxia such as oxygen-sensitive microelectrodes, magnetic resonance imaging (MRI), flow oximetry system, dual wavelength retinal oximetry, and fluorescence and phosphorescence lifetime imaging.^10–13 However, these methods each have their limitations. Oxygen-sensitive microelectrode is very positionally-dependent. ^14–23 In tissue with focal hypoxia surrounded by physoxia (normal physiologic oxygen tension), multiple measurements are required, and hypoxic regions can be missed. In addition, this technique is a destructive procedure that requires the implantation of microelectrodes, limiting its clinical utility. Although MRI is a minimally invasive approach and provides a large field of view and depth information, it is unable to provide enough resolution to identify small areas of focal hypoxia (in-plane pixel size = 0.39 × 0.39 mm²).¹² Retinal oximetry is a commercially available, non-invasive method to measure the percentage of hemoglobin oxygen saturation within the large retinal vasculature.¹³ However, it is unable to measure choroidal oximetry, which provides oxygen to the central vision or fovea along with the metabolically active photoreceptors, and it is unable to measure the perivascular oxygen tension within tissue, where there can be hypoxia due to impaired ability of the tissue to extract oxygen from the vasculature in many disease states.^24–28 Fluorescence and phosphorescence lifetime imaging are minimally-invasive, optical approaches based on oxygen-dependent quenching of fluorescence or phosphorescence, which can be used to image and measure oxygen tension within retinal vessels.^10,29 One disadvantage of this technique is that it cannot provide long-term visualization of hypoxia due to rapid clearance of the injected small molecule dyes from the body. While eye drops are non-invasive and minimize systemic side effects and are commonly used to treat the anterior portion of the eye, they have limited ability to penetrate to the posterior segment of the eye given the numerous barriers to entry, and thus in clinical practice IV or intravitreal injections are often necessary to achieve sufficient concentration of nanosensors to the posterior segment of the eye including the retina and choroid.³⁰ Thus, there is a critical clinical need for an effective, non-destructive method to measure oxygen tension in the tissue microenvironment rather than strictly within retinal blood vessels, and no effective solution exists to this problem.

Metal-free purely organic phosphors are an emerging class of RTP materials with unique properties such as long lifetime (milliseconds to seconds) and large Stokes shift (wavelength difference between the absorption and emission peak maxima).^31–35 These features endow organic phosphorescence-based sensors several advantages over traditional optical sensor design for hypoxia detection and imaging in biological systems. First, purely organic phosphorescence signal from the long-lived triplet excited state is highly susceptible to molecular oxygen quenching through triplet energy transfer,^36–41 whereas conventional fluorescent probes have short-lived emission (nanoseconds), and thus are typically insensitive to oxygen tension change. Second, the large Stokes shift effectively eliminates the interference of the excitation light source or the background autofluorescence by wavelength-based deconvolution, enabling high signal-to-noise ratio measurements. Last, unlike conventional inorganic or organometallic-based RTP materials such as Oxyphors⁴² containing precious rare-earth and transition metals with potential toxicity or stability issues in bio-applications,⁴³ purely organic phosphors are more cost-effective, robust, and biocompatible.

There are several key design considerations to achieve bright organic RTP, including molecular motion suppression^44–46 through doping in rigid hosts such as solid-state crystalline structures or rigid polymer films. However, these strategies limit the practical applicability of organic phosphors toward in vivo hypoxia detection since good oxygen diffusivity, excellent aqueous dispersibility, and biocompatibility also need to be integrated in material design and processing. Consequently, despite the great potential, there are very few organic RTP material systems exploited as in vivo oxygen sensors.^47–50

This report describes a lipid-polymer hybrid core-shell, metal-free organic RTP NP platform that can optically visualize chorioretinal tissue hypoxia in real-time with high signal-to-noise ratio. Specifically, through a facile one-pot self-assembly protocol, the oxygen-sensitive organic phosphor is encapsulated in a rigid, oxygen-permeable polymer matrix core, which is further encased in a layer of phospholipid shell. Such formulation of organic phosphors into nanoprobes represents an effective strategy to enhance their biocompatibility and spatiotemporal resolution for in vivo bioimaging. The fabricated hybrid NPs exhibit long-lived bright RTP with high sensitivity toward oxygen quenching and good long-term stability in vitro, making them promising tissue hypoxia imaging agents for pre-clinical studies. As a proof of concept, the in vivo tissue oxygen-sensing efficacy and biosafety of these RTP NPs are assessed via intravitreal injection in a rabbit RVO model and IV injection in a rabbit CVO model. To the best of our knowledge, this work represents the first non-destructive method to longitudinally visualize oxygen tension in the chorioretinal tissue rather than in hemoglobin within the retinal vasculature. The reported hybrid RTP NP platform could enable quantitative mapping of oxygen gradient and measure the degree of tissue ischemia with high spatiotemporal resolution.

2. Results and Discussion

2.1. Self-assembly of lipid-polymer hybrid organic RTP NPs

A facile and versatile assembling approach for organic RTP nanomaterials is required for practical biosensing and bioimaging applications. Nanoprecipitation is an efficient and convenient method allowing large scale and rapid production of homogenous NPs by utilizing a sharp solubility change of host materials in miscible dissimilar solvents.^51,52 Here we developed lipid-polymer hybrid, core-shell RTP NPs, termed Br6A-LPS4Br NPs, with the metal-free purely organic phosphorescent molecule dispersed in the polymer matrix core, the surface of which was further passivated with an amphiphilic lipid layer. Poly(4-bromostyrene) (PS4Br) was selected as the matrix polymer, and Br6A, previously reported by our lab,⁵³ as the metal-free purely organic phosphor for the NP fabrication (Figure 1). The transparency, rigidity, and oxygen permeability of PS4Br are essential properties for the nanosensor design to achieve bright and sensitive RTP for tissue hypoxia detection. The high rigidity of the matrix polymer is required to effectively suppress the vibrational emission quenching of the metal-free phosphor so as to achieve bright RTP emission.⁴⁴ Good oxygen permeability of the polymer matrix is necessary for the embedded organic phosphor to respond to oxygen tension change in the surrounding environment in real-time. Styrene-based polymers have desirable rigidity and good oxygen permeability, and have been exploited in sensors for dissolved oxygen detection.^41,52,54,55 Additionally, we envision that the emission intensity of Br6A in PS4Br can be enhanced by external halogen bonding effect of the brominated polystyrene matrix. The halogen bonding between the carbonyl oxygen of Br6A and the neighboring bromine atom of PS4Br can suppress the collisional emission quenching as well as enhance spin–orbit coupling and intersystem crossing of Br6A.⁵³

Figure 1. Materials design strategy for biocompatible oxygen-sensing.

a, Synthetic procedure of lipid-polymer hybrid core-shell, room-temperature phosphorescent NPs, Br6A-LPS4Br NPs. The phosphorescence signal of NPs aqueous suspension is susceptible to oxygen quenching. b, Simplified Jablonski diagram showing the mechanism of phosphorescence quenching by molecular oxygen through triplet energy transfer. S₀, S₁, and T₁ are the singlet ground state, first excited singlet state, and excited triplet state of the organic phosphor, respectively.

Lipid coating on the solid polymeric core was achieved by introducing an anionic phospholipid, 1,2-dimyristoyl-sn-glycero-3-phosphate (DMPA) in the aqueous outer phase during the nanoprecipitation process. As the main components of cellular membrane, phospholipids have excellent biocompatibility.⁵⁶ Its amphiphilic and anionic structure can help increase the NPs’ stability in water by preventing their aggregation through electrostatic repulsion or hydration,^57–61 particularly at high concentrations and for long-term storage. It is important to note that injection of highly concentrated polymer solution (10 mg/mL PS4Br) without DMPA to the aqueous outer phase resulted in massive aggregation instead of discrete NPs. Stable Br6A-PS4Br NPs can be produced only at a low polymer concentration (1 mg/mL). Therefore, DMPA lipids have an important role in achieving homogeneous dispersion of the NPs in the aqueous phase. Dynamic light scattering (DLS) data indicate fairly narrow-dispersed Br6A-LPS4Br NPs at an encapsulation ratio of 5 wt% Br6A (with respect to PS4Br), with an average hydrodynamic diameter of 163.9 ± 2.7 nm (mean ± S.D., n = 3) and polydispersity index (PDI) of 0.134 ± 0.008 (mean ± S.D., n = 3) in Milli-Q water (Figure 2a). The more negative zeta potential (ζ) of the resulting Br6A-LPS4Br (−44.7 ± 0.5 mV) in comparison to bare Br6A-PS4Br (−32.1 ± 0.8 mV) implies the successful passivation of NP surface with anionic DMPA lipid (Figure 2b). The size and spherical shape of the resulting NPs were further confirmed by scanning electron microscope (SEM, Figure 2c).

Figure 2. Chemophysical and photophysical characterizations of Br6A-LPS4Br NPs.

a, Hydrodynamic size (diameter, nm) distribution of Br6A-LPS4Br NPs measured by dynamic light scattering (DLS). Well-dispersed NPs in aqueous solutions with an average hydrodynamic diameter of 163.9 nm (PDI: 0.134) were fabricated from a single-step nanoprecipitation process. b, Surface zeta potential change with and without lipid DMPA coating. Compared to bare Br6A-PS4Br NPs, Br6A-LPS4Br exhibited a relatively more negative value, indicating its surface was successfully coated with anionic DMPA. c, Morphology of Br6A-LPS4Br NPs observed by scanning electron microscopy. Scale bar: 300 nm. d, Temporal stability of Br6A-LPS4Br NPs stored in Milli-Q water at room temperature. NPs size (orange line) and polydispersity (blue line) were monitored over the course of 13 weeks using DLS. No obvious aggregation was observed, suggesting long-term stability. All error bars indicate S.D. (n = 3). e, Steady state photoluminescence excitation, emission, and delayed emission (delayed for 500 μs) spectra of Br6A-LPS4Br NPs dispersed in Argon (Ar)-purged, anoxic aqueous solution. f, Phosphorescence lifetime of RTP NPs in an anoxic aqueous solution monitored at 530 nm (λ_ex = 365 nm). Photophysical properties of Br6A-LPS4Br NPs confirmed that the green, 530 nm emission is indeed of phosphorescent nature. g, Oxygen sensitivity calibration. Steady state photoluminescence emission of Br6A-LPS4Br NPs suspension at various O₂ saturation levels (0– 21%). λ_ex = 365 nm.

The colloidal stability of RTP NPs is an important criterion for their long-term in vivo oxygensensing applications. We studied the temporal storage stability of Br6A-LPS4Br NPs by monitoring the change in their size and polydispersity over 13 weeks. The stock solution (2.5 mg/mL) was stored in Milli-Q water under ambient conditions. There was no sign of aggregation of the RTP NPs suspension over 13 weeks, as shown in Figure 2d. Therefore, this simple one-step nanoprecipitation method is suitable for robust production of lipid-stabilized, polymer-supported organic RTP NPs.

2.2. Phosphorescence properties of organic RTP NPs

The RTP nature of the emission from the fabricated Br6A-LPS4Br NPs was then confirmed. As expected, the NPs aqueous suspension exhibited bright green emission when it was excited by 365 nm light after the removal of dissolved oxygen by argon purging (photograph in Figure 1). Steady state photoluminescence spectroscopy showed the excitation and emission spectra maxima at 360 nm and 530 nm, respectively. The small shoulder peak around 425 nm corresponds to the fluorescence emission of Br6A. The gated emission spectrum obtained with a 500 μs delay time overlapped very well with the steady state emission spectrum, which indicated that the green emission of NPs is indeed phosphorescent (Figure 2e). Lifetime (τ) measurement of the green emission monitored at 530 nm resulted in 4.0 milliseconds (ms) (Figure 2f), further corroborating the phosphorescent emission. Quantum efficiency of Br6A-LPS4Br NPs in anoxic aqueous suspension Φ_p (455–675 nm) was measured to be 16.9 ± 3.0%. Therefore, these photophysical characteristics of Br6A-LPS4Br NPs are in good agreement with those of Br6A in crystalline state⁵³ or in isotactic poly(methyl methacrylate) solid solution⁴⁶ reported previously, indicating the successful incorporation of Br6A in the polymer matrix by nanoprecipitation method. We then tested the total emission intensity of Br6A-LPS4Br NPs aqueous suspension with various partial pressures of oxygen (pO₂ from 0–21%). The green phosphorescence emission is highly responsive to small changes in pO₂, with gradually quenched phosphorescence signal as the pO₂ increased, whereas the fluorescence emission at 420 nm remained the same (Figure 2g). This confirms our hypothesis that the phosphorescence emission of Br6A-based RTP NPs can be applied for sensitive hypoxia detection. Though measurements with more precise control over pO₂ would be needed to calculate the hypoxia detection range, it is reasonable to estimate that these RTP NPs will be able to distinguish differences of approximately 3% in pO₂, and hence with great potential to determine different degrees of tissue ischemia. In addition, we studied the optical stability of Br6A-LPS4Br NPs and found that the total emission intensity of the NPs suspension upon Argon purging remained the same after 13 weeks (Supplementary Information, Figure S1).

2.3. In vivo imaging of organic RTP NPs in rabbit RVO models

To evaluate the efficiency of organic RTP NPs as in vivo nanosensors for the detection of tissue hypoxia, the synthesized Br6A-LPS4Br NPs were administrated intravitreally into six rabbits with laser-induced hemi-RVO as described previously to obtain localized hypoxia.^62,63 After acquiring baseline images one week post laser-induced RVO (Supplementary Information, Figure S2), 50 μL of Br6A-LPS4Br NPs at a concentration of 2.5 mg/mL was administrated to the rabbits via intravitreal injection. Longitudinal distribution and phosphorescence signal of the NPs was monitored for up to 7 days post-injection using color fundus photography and fundus phosphorescence imaging (Figure 3). The phosphorescence intensity was determined using the regions of interest (ROI) analysis method. Figure 3 shows in vivo longitudinal visualization of hypoxia and the surrounding retinal vasculature pre- and post-administration of Br6A-LPS4Br NPs at different time points such as 15 min, 1, 2, 4, 8, and 24 hours, and 2, 4, and 7 days. Figure 3a and 3c illustrate the color fundus images of two different sides of the same rabbit eye: the hypoxic RVO side and the physoxic control (untreated) side. There was no phosphorescent signal observed before the injection of Br6A-LPS4Br NPs and the NPs were clearly visualized starting at 1 h post-injection (Figure 3b). These images demonstrate dynamic changes of the RTP signal of Br6A-LPS4Br NPs over time. The location of NPs is clearly visualized at 1–2 h post-injection and is still visible up to 7 days on the hypoxic side. This contrast indicates the ability of these organic RTP NPs to track tissue hypoxia in vivo over multi-day period without having to do reinjection. As expected for the untreated side, no phosphorescence signal was detected over time given the normal tissue oxygen tension (Figure 3d). By applying an image segmentation algorithm to separate the contrast derived from the distribution of Br6A-LPS4Br NPs, average phosphorescence intensity (API) was quantified for each time point (Figure 3e). This quantification shows that API significantly increased post-injection compared to pre-injection of Br6A-LPS4Br NPs. The API in the hypoxia increased by 4.97-fold over the first hour postinjection from 8.58 ± 1.46 (a.u.) pre-injection to 42.67 ± 0.07 (a.u.) (p < 0.001) and reached a peak at 2 h post-injection (API = 45.13 ± 1.31 (a.u.)), yielding a SNR of 12.5 (First Standard Deviation method). Although the API then decreased over time, it was still 168 % higher at day 7 (API = 14.45 ± 8.86 (a.u.)) compared to pre-injection.

Figure 3. Longitudinal phosphorescence imaging of intravitreal Br6A-LPS4Br NPs in living rabbit retinal hypoxia and control over 7 days (λ_ex = 365 nm, λ_em = 530 nm).

a,c, Color fundus photographs of the hypoxic (a) and the physoxic control (c) side of the same rabbit before and after intravitreal administration of Br6A-LPS4Br NPs (50 μL, 2.5 mg/mL) at different time points (15 min, 1, 2, 4, 8 h and day 1, 2, 4, and 7). The color fundus photographs show rabbit fundus features such as retinal vessels (RVs) (yellow arrow), choroidal vessels, optic nerve (turquoise arrows), and distribution of NPs in the vitreous. b,d, Phosphorescence images of the hypoxic (b) and the physoxic control (d) side before and up to 7 days after intravitreal administration of Br6ALPS4Br NPs. Black dotted circles (a) and white dotted circles (b) indicate the position of NPs. Red arrow (b) indicates the location of phosphorescence signal. e, Average phosphorescence intensity measured from the hypoxic side (blue line) and the control side (red line). Error bars show the standard deviations of three independent measurements. The phosphorescence signal increased significantly on the hypoxic side by 1 h post-injection, peaked at 2 h post-injection, and persisted for at least 7 days.

The in vivo reproducibility of retinal hypoxia detection by organic RTP NPs was verified in other rabbits with laser-induced RVO two weeks prior to intravitreal injection of Br6A-LPS4Br NPs with the same dose and concentration (50 μL, 2.5 mg/mL) (Supplementary Information, Figure S3). Hypoxia was monitored for 7 days. Phosphorescence images showed high contrast and peaked at 2 h post-injection, reconfirming the in vivo hypoxia detecting and tracking capability of these organic RTP NPs for different degrees of tissue ischemia. To further validate that the phosphorescence signal is activated by local hypoxia, Br6A-LPS4Br NPs (50 μL, 2.5 mg/mL) was injected into a normal rabbit and imaged with color fundus photography and fundus phosphorescence imaging at different time points over a period of 17 days (Supplementary Information, Figure S4). Phosphorescence signal was not detected either in the NP-injected side or in the non-injected side (Supplementary Information, Figure S4b,d), and ROI analysis shows that API did not change over time (Supplementary Information, Figure S4e). These results confirmed that the RTP signal of Br6A-LPS4Br NPs is quenched under normal oxygen tension in a healthy retina, and hence the developed RTP NPs can selectively detect hypoxia.

2.4. In vivo multimodal imaging of organic RTP NPs

We then sought out to confirm that the signal generated in the hypoxic area of the RVO model is truly arising from the phosphorescence emission of Br6A in RTP NPs. A near-infrared fluorescent dye (IR-780) was co-encapsulated with the organic phosphor Br6A, yielding narrow-dispersed (PDI: 0.108 ± 0.012) Br6A-IR780-LPS4Br NPs with an average hydrodynamic diameter of 130.5 ± 1.7 nm (Figure 4a), to allow for dual phosphorescence and fluorescence imaging. Since the fluorescent signal of IR-780 is not affected by oxygen tension change, co-localizing the fluorescence and phosphorescence signals allows us to track the post-injection distribution of RTP NPs. Intravitreal injection of Br6A-IR780-LPS4Br NPs (50 μL, 2.5 mg/mL) into the RVO rabbit showed co-localization of the fluorescence signal of IR780 and the phosphorescence signal (Figure 4b,c), indicating the hyperphosphorescence signal in the RVO model originates from Br6A in the RTP NPs in response to tissue hypoxia.

Figure 4. In vivo multimodal fluorescence and phosphorescence images of intravitreal Br6A-IR780-LPS4Br NPs and retinal hypoxia in living rabbits.

a, Hydrodynamic size distribution of Br6A-IR780-LPS4Br NPs measured by dynamic light scattering. Narrow-dispersed (PDI: 0.108 ± 0.012), fluorescent dye (IR-780) and organic phosphor (Br6A) co-encapsulated RTP NPs with an average hydrodynamic diameter of 130.5 ± 1.7 nm were synthesized using a similar nanoprecipitation method. b,c, Fundus photography, fluorescence, and phosphorescence images obtained before (b) and after (c) intravitreal administration of 50 μL of Br6A-IR780-LPS4Br NPs at a concentration of 2.5 mg/mL. White dotted lines show the distribution of RTP NPs in the vitreous post-injection, the corresponding fluorescence emission from IR-780, and the corresponding phosphorescence emission from Br6A. The results show close correlation between the fluorescent and phosphorescent signal, indicating that the phosphorescent signal in the rabbit RVO model is emanating from RTP NPs.

2.5. In vivo imaging of organic RTP NPs in rabbit CVO models

To realize minimally invasive delivery of these organic RTP NPs for in vivo hypoxia imaging, we identified an RTP NPs formulation suitable to be administrated intravenously. Nanotherapeutics via IV delivery are usually formulated with average sizes of 10–150 nm in diameter to reduce reticuloendothelial system (RES) clearance and evade the 5 nm renal filtration cut off.^64,65 Unlike normal blood vessels with endothelial tight junctions and blood–retina barrier, occluded retinal vessels tend to be disorganized and demonstrate hyperpermeability.^66,67 With these design criteria in mind, we hypothesized that smaller RTP NPs in the sub-100 nm range will more effectively extravasate and accumulate at occluded sites, leveraging the enhanced permeability and retention (EPR) effect. They will also more likely to have higher diffusivity, allowing for more efficient penetration into the hypoxic chorioretinal tissue. Therefore, we used a more hydrophilic diblock copolymer, polystyrene-b-poly(4-vinylpyridine) (PS4VP, PS block MW 35.5 kD and P4VP block MW 4.4 kD, Figure 5a), as the host matrix to construct NPs’ polymeric core, which leads to smaller RTP NPs via bulk nanoprecipitation. Similar to polystyrene, P4VP is highly rigid at room temperature, and is oxygen-permeable.⁶⁸ The low block ratio of P4VP in the overall diblock copolymer allows PS4VP to have similar characteristics in its rigidity and oxygen permeability compared to polystyrene. Hence, the resulting RTP NPs have brightness and sensitivity comparable to those of Br6A-LPS4Br NPs toward hypoxia detection. IR-780 was co-encapsulated for particle tracking in blood vessels and the retina. To improve RTP NPs’ biocompatibility and pharmacokinetics for systemic delivery, a FDA-approved lipid-PEG conjugate, 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[carboxy(polyethylene glycol)-2000] (DSPE-PEG2000), was used instead of DMPA to create a “stealth” shell on the surface of the organic RTP NPs.⁶⁹ By means of a similar nanoprecipitation procedure as described before, RTP NPs Br6A-IR780LPS4VP-PEG with an average hydrodynamic diameter of 46.1 ± 0.6 nm and PDI of 0.134 ± 0.005 were successfully synthesized (Figure 5b).

Figure 5. In vivo phosphorescence images of Br6A-IR780-LPS4VP-PEG NPs and chorioretinal hypoxia in living rabbits.

a, Chemical structure of polymer PS4VP and phospholipid DSPE-PEG2000. b, Dynamic light scattering measurement of Br6A-IR780-LPS4VP-PEG NPs fabricated from PS4VP and DSPE-PEG2000. Fairly narrow-dispersed (PDI: 0.134 ± 0.005), fluorescent dye (IR-780) and organic phosphor (Br6A) co-encapsulated RTP NPs with an average hydrodynamic diameter of 46.1 ± 0.6 nm were synthesized using a similar nanoprecipitation method. c–e, Color fundus photography (c), fluorescence (d), and phosphorescence (e) images before and after IV injection of 4 mL Br6A-IR780-LPS4VP-PEG NPs at a concentration of 2.5 mg/mL. Color fundus images in c show the healthy retinal vessels (RVs), choroidal vessels (CVs) as well as the location of laser injured sites (white spots). Fluorescence and phosphorescence images obtained before and post injection at different time points, demonstrating the accumulation of RTP NPs at laser lesions (white arrow) and the corresponding phosphorescent signal from the RTP NPs (red arrows) detecting tissue hypoxia. f, Quantification of mean phosphorescence intensities at laser lesions over 7 days. Phosphorescent signal was not visible before the injection of RTP NPs or on the physoxic control side. In contrast, the phosphorescent signal increased significantly in the hypoxic areas at 15 min post-injection, peaked at 2 h post-injection, and gradually decreased after that. Error bars show the standard deviations of three independent measurements.

The choroid is a vascular-rich tissue immediately deep to the retina which supplies oxygen to the outer half of the retina including the fovea, or central vision, and thus it is critical to understand tissue oxygen tension within both the retina and choroid. Using an established rabbit model of CVO with laser photocoagulation⁷⁰ (Supplementary Information, Figure S5), we found that intravenously administrated Br6A-IR780-LPS4VP-PEG NPs were able to effectively accumulate at the laser lesions (~ 300 μm in diameter) in the choroid and detect focal tissue hypoxia in a “turn-on” modality with a peak RTP signal at 2 h post-injection (Figure 5d–f). While the phosphorescence signal wasn’t clearly visualized until 1 h post intravitreal injection in rabbits with RVO, it was clearly visible here at 15 min post IV injection. This is likely due to the different initial oxygen levels in RTP NPs. For intravitreal injection, the NPs suspension is initially saturated with atmospheric oxygen (~ 21% O₂). It takes time to deplete the dissolved oxygen injected along with the NPs at the hypoxia site before the phosphorescence signal can be turned on. In contrast, NPs after IV injection is immediately diluted with high volume of blood and hence can be considered at physoxia, i.e. much lower levels of oxygen universally found in blood and normal tissues, approximately 5–13% and 4–7.5%, respectively.⁷¹

To confirm the phosphorescent signal detected is attributed to RTP NPs selectively sensing hypoxia rather than tissue autofluorescence, NPs encapsulated with IR-780 only and no Br6A (IR780-LPS4VP-PEG NPs) was injected intravenously into rabbits with CVO. Consequently, only the fluorescent signal showing NPs location was observed, and no hyperphosphorescence at the occluded hypoxic laser lesions (Supplementary Information, Figure S6).

To increase the clinical translatability, the in vivo reproducibility of choroidal hypoxia detection by Br6A-IR780-LPS4VP-PEG NPs via IV injection was verified in multiple Dutch Belted rabbits (Supplementary Information, Figure S7). These RTP NPs initially clustered at the rim of laser lesions (15 min) and gradually penetrated the center region. At 4 h, patches of RTP NPs accumulated at the center regions of laser lesions can be clearly visualized on the fluorescence channel (Supplementary Information, Figure S7b). Strong fluorescence signal was still detectable in both retinal and choroidal blood vessels even at 48 h post-injection, suggesting that Br6A-IR780-LPS4VP-PEG NPs exhibit prolonged blood circulation.

2.6. In vivo toxicity and biosafety analysis of organic RTP NPs

To test the potential toxicity of IV and intravitreal administration of RTP NPs in these rabbits, several biosafety analyses were performed, including body weight analysis, liver function tests (LFT), kidney function tests (KFT), TUNEL assay, and histopathology. All rabbits used in this study had their weight measured daily (Figure 6a and Supplementary Information, Figure S8a). Body weight of intravenously treated, intravitreally treated, and untreated groups increased appropriately over 7 days, demonstrating that these RTP NPs do not induce negative systemic impact on living rabbits. Hematoxylin and eosin (H&E) staining demonstrates preserved normal retinal cellular morphology and nuclei without fragmentation or extracellular debris (Figure 6b,c and Supplementary Information, Figure S8b,c). TUNEL assay analysis demonstrates no evidence of cells undergoing apoptosis (Figure 6d and Supplementary Information, Figure S8d). To inspect potential local inflammations post-intravitreal administration, slit lamp examination was performed and there was no evidence of intraocular inflammation, either anterior uveitis or vitritis. Serum blood tests were performed to examine the acute toxicity of RTP NPs to liver or kidney function of the animals. As shown in Table 1 for IV injection and Supplementary Information Table S1 for intravitreal injection, all LFTs and KFTs are within normal range at 14 days after RTP NPs administration in all animals, indicating normal liver and kidney function with no systemic toxicity. According to a National Institute for Occupational Safety and Health report on ocular ultraviolet effects from 295 nm to 400 nm in the rabbit eye, the effect on the lens abruptly drops at 313 nm.⁷² Therefore, the excitation wavelength of 365 nm for only a short period of exposure for imaging will not likely to cause harmful effect, and indeed in our study no ocular or systemic complications were observed in the rabbit models after imaging.

Figure 6. Biosafety evaluation in living rabbits.

a, Body weight increase measured daily for 7 days from three different groups: untreated control and intravenously treated CVO models in White New Zealand and Dutch Belted pigmented rabbits, showing no evidence of systematic toxicity through equivalent, appropriate weight gain in both RTP NPs treated and untreated control groups. b,c, H&E staining of tissues obtained 1 month post-IV injection of Br6A-IR780-LPS4VP-PEG NPs (b) compared to the untreated control group (c), demonstrating preserved cellular morphology and nuclei without fragmentation or extracellular debris from dead cells in all of the evaluated organs, including eye, heart, kidney, lung, liver, and spleen. d, TUNEL assay analysis at 1-month post-IV injection of Br6A-IR780-LPS4VP-PEG NPs. DAPI (blue) indicates cell nuclei. Green color stained with FITC evaluates for any potentially apoptotic cells, which are not noted. Scale bar: 75 μm.

Table 1.

Liver and Kidney Function tests at 14 days post-administration for untreated control and intravenously treated CVO model in New Zealand white and Dutch Belted pigmented rabbits demonstrating normal liver and kidney function tests

			Br6A-IR780-LPS4VP-PEG				Merge
Liver Function Tests (LFT)	Normal Range	Unit	Rabbit #1	Rabbit #2	Rabbit #3	Average	Control #1	Control #2	Control #3	Average
Albumin	2.7–5	g/dL	3.7	4.1	4.2	4.0 ± 0.3	3.6	4.2	4.8	4.2 ±0.6
Total Protein	5–7.5	g/dL	5.3	6	6.2	5.8 ± 0.5	5.7	6.1	7.2	6.3 ±0.8
Alanine aminotransfera se (ALT)	25–65	U/L	35	43	41	39.7 ± 4.2	37	25	64	42.0 ±20.0
Alkaline phosphatase (ALP)	10.0– 86.0	U/L	76	86	66	76.0 ± 10.0	26	58	78	54.0 ±26.2
Total Bilirubin (TBIL)	0.2–0.5	mg/dL	0.4	0.3	0.2	0.3 ± 0.1	0.5	0.5	0.4	0.5 ±0.1
Kidney Function Tests (KFT)
Blood urea nitrogen (BUN)	5.0–25.0	mg/dL	15	21	21	19 ± 3.5	19	22	24	21.7 ±2.5
Creatinine (CREA)	0.5–2.6	mg/dL	0.67	0.68	1.11	0.8 ± 0.2	1.2	1.14	0.9	1.1±0.2
Calcium	5.6–12.1	mg/dL	12.0	11.4	12.9	12.1 ± 0.8	11.7	12.2	11.7	11.9±0.3
Glucose	74–148	mg/dL	137	111	189	145.7 ± 39.7	138	160	127	141.7±16.8

3. Conclusions

In this work, we have successfully developed a versatile lipid-polymer hybrid assembly platform to generate efficient metal-free organic RTP NPs for longitudinal optical visualization of chorioretinal tissue hypoxia in living rabbits in a non-destructive fashion. The facile nanoformulation involves (1) a rigid and oxygen-permeable polymer core for the effective activation of bright RTP from the embedded organic phosphor in hypoxic environments, and (2) an amphiphilic phospholipid shell, allowing for excellent water dispersity, biocompatibility, and colloidal stability. The phosphorescent signal of the fabricated organic RTP NPs demonstrates milliseconds decay time and is highly responsive toward oxygen quenching, which enables them to be exploited as “turn-on” imaging probes for chorioretinal tissue hypoxia. When tested in vivo using rabbit RVO and CVO models, both oxygen-sensing efficacy and biosafety of the organic RTP NPs were demonstrated. The phosphorescence signal was exclusively generated with high SNR in the RVO side or laser lesions in the CVO side of the rabbit eye where tissue hypoxia was present. Specifically, when administrated intravenously in rabbit CVO model, the RTP NPs exhibited prolonged blood circulation and were able to effectively accumulate at target laser lesions in the choroid and detect focal tissue hypoxia. The fluorescent dye and organic phosphor co-encapsulated NPs showed co-localization of fluorescence and phosphorescence signals via multimodal imaging, confirming the hyperphosphorescence in both RVO and CVO models originates from organic RTP NPs in response to tissue hypoxia. No ocular or systemic complications or toxicity were noted after either IV or intravitreal administration of RTP NPs. These data provide proof-of-concept that the developed nanoformulation of organic RTP materials allows for biocompatible, non-destructive detection of hypoxia in chorioretinal tissue, currently unachievable by other methods.

While further studies will be conducted to assess this RTP nanosensor’s hypoxia detection window and in vivo oxygen tension determination, the results of this work provide a strong indication that the developed organic RTP NPs hold significant promise as an advanced non-invasive imaging tool for long-term visualization of tissue oxygen levels and evaluation of various hypoxia-driven vascular diseases. Since the light-generating mechanism of the organic phosphor Br6A in response to oxygen tension is at molecular scale, and RTP particles are at nanoscale, it is reasonable to anticipate high spatial resolution (nanoscale) for in vivo hypoxia imaging, though in practice the resolution will be limited to around 5 μm due to optical resolution in the eye without adaptive optics. This study lays a solid foundation demonstrating that this organic RTP nanosensor has a great potential to serve as a general tissue hypoxia tracking probe via minimally invasive systemic delivery. This has both scientific and translational impact at many levels and will enable improved understanding of the pathophysiology, and early diagnosis and prognosis research of diseases involving ischemia-induced hypoxia, beyond the retina and choroid.

4. Experimental Section/Methods

Materials:

Reagents used to synthesize the organic phosphor Br6A were purchased from Millipore Sigma. Phospholipids 1,2-dimyristoyl-sn-glycero-3-phosphate (DMPA) and 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[carboxy(polyethylene glycol)-2000] (DSPE-PEG2000) were purchased from Avanti Polar Lipids. Polystyrene-b-poly(4-vinylpyridine) (PS4VP) was purchased from Polymer Source. Other chemicals and materials such as IR-780 iodide, poly(4-bromostyrene) (PS4Br, M_w ~65,000), solvents, and Amicon^® Ultra-4 centrifugal filters (MWCO 30 kD) were also purchased from Millipore Sigma. Malvern Panalytical folded capillary zeta cells were purchased from Fisher Scientific.

Preparation of lipid-polymer hybrid organic RTP NPs for intravitreal injection:

The metal-free organic phosphor, Br6A was synthesized according to a published procedure from our lab,⁵³ and its purity was confirmed by ¹H NMR. The lipid-polymer hybrid NPs containing the organic phosphor were fabricated by a single-step nanoprecipitation method. To prepare Br6A-LPS4Br NPs, phospholipid DMPA was dispersed into 4 wt% ethanol-water solution at a concentration of 0.15 mg/mL (15 wt% to PS4Br) as the aqueous outer phase. The dispersion became completely transparent when it was heated up to 65 °C, which ensures the added DMPA lipids were in the liquid phase. A stock solution of 10 mg/mL PS4Br mixed with Br6A (5 wt% to PS4Br) was prepared in tetrahydrofuran (THF) as the organic internal phase. Then 1 mL of this THF solution was rapidly injected into the aqueous phase with vigorous stirring, followed by subsequent sonication in a bath sonicator (Bransonic 2510-DTH) at a frequency of 42 kHz and power of 100 W for 10 min. Additional stirring for 30 hours at room temperature allows for the complete evaporation of THF and ethanol. Washing the NPs in Milli-Q water followed by filtering through Amicon^® Ultra-4 centrifugal filters with a molecular weight cutoff of 30 kD three times was the last purification step for NP preparation. The purified Br6A-LPS4Br NPs were then re-suspended in Milli-Q water to make 2.5 mg/mL stock concentration. The fluorescent dye co-encapsulated Br6A-IR780-LPS4Br NPs were synthesized using a similar procedure except when mixing PS4Br with organic phosphor Br6A, IR-780 iodide (stock solution prepared in acetone at 1mg/mL) was also added at 2 wt% to PS4Br.

Preparation of lipid-polymer hybrid organic RTP NPs for IV injection:

To prepare Br6A-IR780-LPS4VP-PEG NPs for IV administration, PS4VP was used as the polymer matrix instead of PS4Br, and DSPE-PEG2000 (30 wt% to PS4VP) was used as the phospholipid instead of DMPA. To make 1 mL of the organic internal phase containing polymer, organic phosphor (5 wt% to polymer), and fluorescent dye (2 wt% to polymer), 400 μL THF was used to dissolve 10 mg of PS4VP, followed by adding 100 μL of Br6A stock in THF at 5 mg/mL and 200 μL of IR-780 iodide stock in acetone at 1mg/mL. Then 300 μL of acetone was added to the mixture to adjust the volume ratio of THF and acetone (1:1). The rest of the steps were similar to the nanoprecipitation procedure described above.

Characterization of lipid-polymer hybrid organic RTP NPs:

Hydrodynamic diameter (nm), size distribution, and surface charge (ζ potential, mV) of the lipid-coated and phosphor-embedded polymer NPs in Milli-Q water were analyzed by means of Malvern Zetasizer Nano ZSP (Model number: ZEN5600) using disposable cuvettes and folded capillary zeta cells, respectively. All measurements were conducted at a backscattering angle of 173° (NIBS default) at 25 °C. Size and size distribution data were obtained using Malvern Zetasizer software (Ver. 7.11). The average hydrodynamic diameter was determined using the peak means of size distribution plots by intensity. Particle size and morphology of the NPs were further examined by SEM (Thermo Fisher Nova 200 Nanolab) after carbon coating. Photophysical properties including steady-state and delayed emission spectrum, excitation spectrum, emission lifetime, and absolute quantum yield were measured using a Photo Technologies International (PTI) QuantaMaster spectrofluorometer equipped with an integrating sphere. Anoxic aqueous suspensions of RTP NPs were prepared by purging oxygen with argon for 30 min through sample solution in quartz cuvette with a rubber septum cap. Then to achieve various pO₂ (0–21%) in the aqueous suspension, different amount of air was refilled back into the anoxic sample. All experiments were performed in three freshly prepared samples (n = 3). Data are expressed as the mean ± standard deviation.

Animal studies ethics oversight:

All animal studies were implemented under the guidelines of the Association for Research in Vision and Ophthalmology (ARVO) Statement on the Care and Use of Laboratory Animals in Ophthalmic and Vision Research. The experimental protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Michigan (Protocol number: PRO00008566, PI: Y. Paulus).

Animal model preparation:

A total of 11 rabbits were used in this study. Nine New Zealand white rabbits (2–4 months and 2.5–3.0 kg) were obtained from the Center for Advanced Models and Translational Sciences and Therapeutics (CAMTraST) at the University of Michigan Medical School, and two Dutch Belted rabbits (3 months and 1.3–1.5 kg) were purchased from Covance. The animals were divided into three groups: control, RVO, and CVO. In the RVO control group, the animals received intravitreal injection of organic RTP NPs (50 μL, 2.5 mg/mL) and had normal, healthy retinas. In the RVO group, the animals received hemi- RVO by Rose Bengal dye-enhanced photochemical thrombosis laser photocoagulation. In the CVO group, the rabbits were treated with laser photocoagulation without administration of Rose Bengal. Throughout experiments and recovery, the animal condition including mucous membrane color, body temperature, heart rate, and respiratory rate was recorded and documented every 15 minutes. To induce anesthesia, a dose of ketamine (40 mg/kg) and xylazine (5 mg/kg) was injected intramuscularly (IM). The rabbit’s pupils were dilated using tropicamide 1% and phenylephrine 2.5% ophthalmic. A drop of 0.5% tetracaine was applied for topical anesthesia before experiments. In addition, lubricant (Systane, Alcon Inc., TX, USA) was provided every minute to avoid corneal dehydration during experiments. To maintain the animal’s body temperature, a water-circulating heating blanket was used.

RVO model generation:

RVO model was generated using Rose Bengal dye-enhanced photochemical thrombosis as described in detail in previous studies.^62,63,73 Briefly, a 532 nm green light laser mounted on a slit lamp was used to create the RVO model (Vitra 532 nm, Quantel Medical, Cournon d’Auvergne, France). To visualize the target retina vessels, a contact lens (Volk H-R Wide Field, laser spot 2x magnification, Volk Optical Inc, Mentor, OH, USA) was placed on the cornea. Rabbits under anesthesia were injected IV with Rose Bengal (50 mg/mL). 5–10 seconds after the injection, 20 spots of 532 nm laser at a power of 150 mW, aerial spot diameter of 75 μm, and pulse duration of 500 ms were illuminated at the same position at a distance of one-half to one disc diameter from the optic nerve to avoid optic neuropathy.⁷⁴ To avoid reperfusion, 20 laser spots were further applied with a power of 300 mW. The treated position on the veins was carefully selected to prevent damaging the adjacent arteries or optic nerve.⁷⁵

CVO model generation:

To induce CVO model, a contact lens was placed on the cornea of the rabbit eye. Gonak Hypromellose Ophthalmic Demulcent Solution 2.5% was placed on the surface of the contact lens for coupling. The rabbit eye was irradiated with a 532 nm green light laser at a power of 450 mW, aerial beam diameter of 150 μm, and pulse duration of 500 ms using a Zeiss SL 130 slit lamp (Carl Zeiss Meditec, Jena, Germany), to which the Vitra photocoagulator was connected. Fifteen shots of the laser were illuminated into the eye at different positions. To create CVO model in Dutch Belted rabbits, the laser power was reduced from 450 mW to 300 mW due to melanin absorption. Twelve spots were illuminated on the retina with aerial spot size of 150 μm and pulse duration of 500 ms. After taking color fundus photographs, fluorescein angiography (FA) and indocyanine green angiography (ICGA) were performed to evaluate the vasculature and confirm vascular occlusion.

Follow-up RVO and CVO evaluation:

All rabbits with RVO model were examined fifteen minutes after the laser treatment, and at day 7 and day 14 post-photocoagulation. The rabbit models were assessed by color fundus photography, fluorescein angiography (FA), indocyanine green (ICG) angiography, and phosphorescence photography.

Color fundus photography:

All retinal vessel network and laser-induced hypoxia were imaged using a custom-modified 50-degree color fundus photography (Topcon 50EX, Topcon Corporation, Tokyo, Japan). The digital images were captures by EOS 5D camera with a resolution of 5472×3648 pixels with a pixel size of 6.55 μm². Color fundus images were obtained using the maximum 50-degree angle of coverage centered at five different positions of the eye: the optic nerve, the superior retina above the optic disc, the inferior retina below the optic disc, the temporal medullary ray, and the nasal medullary ray. Color fundus montages were created using the I2K Retina software (Topcon Corporation, Tokyo, Japan).

Fluorescein angiography and indocyanine green angiography:

Fluorescein angiography (FA) and indocyanine green angiography (ICGA) were performed on the Topcon 50EX camera by changing the camera’s appropriate internal excitation and emission filters for each. For FA, a dose of 0.2 mL fluorescein sodium at a concentration of 10% fluorescein (Akorn, Lake Forest, IL, USA) was intravenously injected into the rabbit via the marginal ear vein. For ICGA, 2.0 mL of ICG solution at concentration of 2.5 mg/mL (Akorn, Lake Forest, IL, USA) was injected intravenously via the marginal ear vein. FA and ICGA images were subsequently acquired after fluorophore injection, and late phase FA and ICGA images were acquired at every minute for a period of at least 20 minutes.

Phosphorescence photography examinations:

To evaluate the potential of organic RTP NPs as an oxygen sensor for detection of hypoxia in vivo, phosphorescence photography was evaluated on rabbit models using a custom-modified Topcon 50EX camera with custom-made filters (excitation filter with the bandpass wavelengths of 335 to 379 nm (FF01–357/44, Semrock, NY, USA) and barrier filter with a bandpass of 498 to 542 nm (FF01–520/44, FF01–357/44, Semrock, NY, USA) and UV-excitation light source (4.4 × 10⁻⁴ W/cm²) with center wavelength of 365 nm and bandwidth of 9 nm (M365LP1, Thorlabs, USA). All rabbits with the RVO model received intravitreal injection of 50 μL organic RTP NPs at a concentration of 2.5 mg/mL. Phosphorescent photography was acquired immediately after the injection and follow-ups for different time points: 15 min, 1, 2, 4, 8, 24 h and day 2, 4, and 7. For IV injection, the rabbits with CVO model were injected with 4 mL of RTP NPs at a concentration of 2.5 mg/mL. Phosphorescent imaging was acquired immediately after the injection and follow-ups for different time points: 15 min, 2, 4, 8, 16, 24 h, 48 h and day 7. It took about 2 seconds to capture each fundus phosphorescence image of the rabbit models. The control groups were monitored over a period of 17 days post-injection. The dynamic changes of phosphorescent signal over time were determined by region of interest (ROI) using ImageJ software.

Biosafety analysis:

Biosafety analysis was performed on the treated animals using different methods such as body weight analysis, hematoxylin, and eosin (H&E) staining, TUNEL assay, liver function test (LFTs), and kidney function tests (KFTs). Body weight was measured daily post administration of RTP NPs for each group over a period of 7 days. At day 14 after injection of RTP NPs, 400 μL of blood sample was collected from each rabbit for LFTs and KFTs. Then, the rabbits were euthanized by IV injection of euthanasia solution (0.22 mg/kg) via the marginal ear vein (Beuthanasia-D Special, Intervet Inc., Madison, NJ, USA). The organs and eye tissues were harvested and fixed with 10% neutral buffered formalin (VWR, Radnor, PA, USA). To prevent retinal detachment, eye tissues were fixed with Davidson’s fixative solution (Electron Microscope Sciences, PA, USA) for 24 h. Afterwards, the samples were placed in 50% alcohol solution for 8 h and then replaced with 70% alcohol solution and kept at room temperature for 24 h. The fixed tissues were embedded in paraffin, sectioned into 6 μm thick sections, and stained with hematoxylin and eosin (H&E) for histopathological examination. TUNEL assay analysis were performed using TUNEL in situ Cell Death Detection Kit protocol (Sigma-Aldrich, USA). The stained slides were analyzed using DM6000 microscope. H&E images were captured using the BF450C camera and TUNEL fluorescence images were obtained using the FF363x camera (DM600, Leica Biosystems, Nussloch, Germany).

Data Availability

The authors declare that the main data supporting the findings of this study are present within the article and its Supplementary Information file. All data are available from the corresponding authors upon reasonable request.