Comparison of phosphorescent agents for noninvasive sensing of tumor oxygenation via Cherenkov-excited luminescence imaging

Abstract. Cherenkov emission generated in tissue during radiotherapy can be harnessed for the imaging biochemistry of tissue microenvironments. Cherenkov-excited luminescence scanned imaging (CELSI) provides a way to optically and noninvasively map oxygen-related signals, which is known to correlate to outcomes in radiotherapy. Four candidate phosphorescent reagents PtG4, MM2, Ir(btb)2(acac), and MitoID were studied for oxygen sensing, testing in a progressive series of (a) in solution, (b) in vitro, and (c) in subcutaneous tumors. In each test, the signal strength and response to oxygen were assessed by phosphorescence intensity and decay lifetime measurement. MM2 showed the most robust response to oxygen changes in solution, followed by PtG4, Ir(btb)2(acac), and MitoID. However, in PANC-1 cells, their oxygen responses differed with Ir(btb)2(acac) exhibiting the largest phosphorescent intensity change in response to changes in oxygenation, followed by PtG4, MM2, and MitoID. In vivo, it was only possible to utilize Ir(btb)2(acac) and PtG4, with each being used at nanomole levels, to determine signal strength, lifetime, and pO2. Oxygen sensing with CELSI during radiotherapy is feasible and can estimate values from 1 mm regions of tissue when used in the configuration of this study. PtG4 was the most amenable to in vivo sensing on the timescale of external beam LINAC x-rays.


Introduction
Extent of oxygenation in tumors is a known indicator of the success of radiation therapy, partly due to the oxygen enhancement ratio, 1 as well as due to oxygen being a surrogate marker for other features of tumor aggressiveness. 2 Therefore, monitoring of tumor oxygenation during fractionated radiation therapy would be advantageous to gauge the likelihood of response or even to estimate if treatment plan alternations such as a boost to hypoxic areas might be beneficial. Oxygen electrodes have been utilized to measure hypoxia. [3][4][5] While this method allows direct measurement of oxygen levels in tumors, it is invasive and is limited to tumors that are easily accessible. Tumor biopsies with indirect measurements for hypoxia, including pimonidazole staining and immunostaining for HIF-1α, provide valuable information of individual tumor microenvironments, 6-10 but these methods are also invasive and do not provide real-time measurement of tumor oxygenation levels. Blood oxygen level-dependent magnetic resonance imaging (BOLD-MRI), 11,12 electron paramagnetic resonance oximetry, 13,14 and near-infrared spectroscopic tomography 15 provide real-time information based on hemoglobin saturation. Positron emission tomography also measures real-time levels of hypoxia, utilizing 18 F-labeled nitroimidazole derivatives, whose luminescence are dependent on oxygen levels. 16,17 Despite the potential value, there has not been any clinical convergence on a method for imaging tumor oxygenation that is noninvasive, precise, and quantitative. In this study, oxygen-related mapping in tumors is demonstrated with Cherenkov-excited luminescence scanned imaging (CELSI), which uses the inherent delivery of radiation to get maps of oxygen-dependent luminescence signals from injected chemical sensors. Current reagents utilized for phosphorescent sensing of oxygen that can be utilized for noninvasive determination of oxygen content [PtG4, MM2, IrðbtbÞ 2 ðacacÞ, and MitoID] are evaluated.
Radiation treatment with MV photons or MeV electrons causes the production of Cherenkov light in tissue. This optical emission occurs when charged primary or secondary electrons pass through the dielectric medium, such as tissue, at a velocity greater than the speed of light. This optical signal has been imaged to visualize surface dose in radiation therapy. [18][19][20] In addition, early pilot studies in tissue phantoms and individual animals have shown utilization of Cherenkov emission as the excitation source in imaging applications, including the detection of fluorophores and phosphors in conjunction with radiation therapy. 21 The potential to sense tumor oxygenation through CELSI with the phosphorescent reagent PtG4 has been shown by us, because the PtG4 phosphorescence is quenched in the presence of oxygen, reducing the observed lifetime of emission. In the presence of ambient oxygen pressure (pO 2 ), the PtG4 phosphorescence lifetime is 16.9 μs, and in low pO 2 it is 47 μs. 22,23 In this study, this agent was compared to other oxygenquenched chemical agents, and each was examined for its potential to sense tumor hypoxia by directly measuring pO 2 levels in vivo with mice bearing subcutaneous tumors. 24 At the end of the study, the spatial confidence in mapping pO 2 was estimated, using doses typical of fractionated radiotherapy.

Cell Culture
PANC-1 cells and MDA-MB-231 luc-D3H2LN cells were grown in DMEM with 10% FBS in an incubator at 37°C with 5% CO 2 , and 100% humidity and divided utilizing 0.05% trypsin when desired confluency was reached.

Fluorescence Assay In Vitro
PANC-1 cells were seeded in black 96-well plates with a clear bottom at 5000 cells∕well. MM2, IrðbtbÞ 2 ðacacÞ, and MitoID (10, 20, 30 μg∕mL in PBS) were added to the cells and allowed to incubate for 24 h at 37°C. Media containing the reagents were removed and cells were rinsed with PBS. 10% FBS in PBS with or without glucose oxidase catalase oxygen scavenger (100 nM glucose oxidase, 1.5 μM catalase, and 56 mM glucose) was added after rinsing and the fluorescence intensity was analyzed with an excitation of 380 nm and emission of 660 nm. For limits of detection studies, 20, 2, 0.2, and 0.02 μg∕mL of MM2 and IrðbtbÞ 2 ðacacÞ were utilized in the same fashion as described above. A student's t-test was utilized to assess differences in fluorescence between oxygenated versus oxygen scavenged conditions.

Cherenkov Imaging
Cherenkov emission was induced by a linear accelerator (Varian LINAC 2100CD, Palo Alto, California). The imaging system consisted of a time-gated intensified CCD camera (ICCD, PI-MAX4 1024i, Princeton Instruments) and a commercial lens (Canon EF 135 mm f/2L USM). The camera was focused on a mirror that reflected the imaging field ∼1 m away. The time-gated ICCD camera was synchronized to the radiation pulses (∼3.25 μs duration and 360 Hz repetition rate) with the intensifier set as ×100 and turned on at a 4.26-or 1000-μs gate delay following each radiation pulse for phosphorescence or background measurement, and luminescence generated during 85.60-μs gate width (PtG4 and MM2), or 7.77 μs [IrðbtbÞ 2 ðacacÞ] was integrated via this ICCD. Images of the luminescence at different delay times between LINAC pulse and phosphorescence emission were acquired to construct emission lifetime. To maximize signal and minimize background interference, the room lights were shut off throughout these studies, and all lights in the room were masked off with black cloth and black tape.

CEL Studies In Vitro
Concentrations of 20, 2, 0.2, and 0.02 μg∕mL of MM2, IrðbtbÞ 2 ðacacÞ, MitoID, and PtG4 were utilized 96-well plate as above. For solution studies, 300 μL of each solution with and without oxygen scavenger were utilized. For cellular studies, PANC-1 cells were seeded in black 96-well plates with a clear bottom at 5000 cells∕well. The reagents were added to the cells and allowed to incubate for 24 h at 37°C. Media containing the reagents were removed and cells were rinsed with PBS. 10% FBS in PBS was added after rinsing.

General Animal Imaging
All animal procedures were approved by the Institutional Animal Care and Use Committee, and the studies here were carried out in compliance with these approved procedures.

In Vivo Imaging
Briefly, 10 5 MDA-MB-231 luc-D3H2LN cells were injected under the skin on the flank of a nude mouse (two tumors per mouse, eight tumors per reagent studied). After ∼3 weeks of growth, animals were chosen for use when their tumor was ∼6 × 6 × 3 mm 3 in size. The CELSI scan was completed vertically. Images of the luminescence at different delay times between LINAC pulse and phosphorescence emission were acquired to construct emission lifetime. Under general anesthesia of inhaled isofluorane, 50 μL of 50 μM of each reagent was directly injected into the tumors and the animal was imaged alive and then again at 30 min after euthanize, which allows determination of emission lifetime at ambient pO 2 (alive) low pO 2 environments (dead), respectively.

Statistical Analysis
The tissue pO 2 was determined utilizing the Stern-Volmer relationship. The differences between the live and dead conditions, as shown in Fig

Results
The set of four phosphorescent oxygen sensors were directly compared to determine which of these reagents would be ideal for use in CELSI in tumors. These agents were chosen based upon published data, and they represented the most promising mix of agents for use in vivo with longer lifetime emission from phosphorescence. The absorbance and emission spectra of PtG4 (green), MM2 (gray), IrðbtbÞ 2 ðacacÞ (blue), and MitoID (red) are shown in Fig. 1. 25 A murine experiment to sense tissue oxygenation was completed in mice with subcutaneous MDA-MB-231luc-D3H2LNtumors. The setup for CELSI is shown in Figs. 4(a) and 4(b). Briefly, gantry head of the linear accelerator that delivers the radiation is positioned below the mouse, which is located on the treatment couch. A mirror is placed that redirects the photons to the ICCD camera for imaging. A total of four tumors per compound were imaged, with local injection of 50 μL of each reagent [50 μM PtG4, 2.5 nmol, 1.75 mg per tumor and 50 μM IrðbtbÞ 2 ðacacÞ, 2.5 nmol, 35 μg per tumor]. Each mouse was imaged while alive and then repeated 30 min after euthanasia to capture both normoxic and hypoxic conditions, respectively. In the euthanized animal, the drop in blood circulation and respiration causes a dramatic decrease in pO 2 values. Temperature was controlled using a heating pad under each animal, throughout the study, and the CELSI scan was completed vertically for these cases. Images of the luminescence were captured at different delay times after the LINAC pulse to determine emission lifetimes for each reagent. The maximum intensity projection images of CELSI from both PtG4 and IrðbtbÞ 2 ðacacÞ at a delay time of 4.26 μs are shown in Fig. 4(c). A comparison of the intensity differences between PtG4 phosphorescence at various delay times in an alive and dead mouse are shown in Fig. 5(a). The emission lifetimes were mapped for the alive and euthanized mouse for PtG4 and are quantified and displayed via a box and whiskers plot [ Fig. 5(b)]. Using the Stern-Volmer equation, we estimated the average tissue pO 2 with the reported quenching constant for PtG4 27 [Fig. 5(c)]. The phosphorescence at different delay times for a euthanized mouse is shown for IrðbtbÞ 2 ðacacÞ [ Fig. 5(d)]. In addition, we created a map of emission lifetimes and estimated the average pO 2 using the reported quenching constant for IrðbtbÞ 2 ðacacÞ using the Stern-Volmer relationship. 14

Discussion
Due to the redshifted emission of PtG4 (772 nm) in comparison to the other reagents, it was originally hypothesized that this reagent would be the most effective for imaging in tumors since this wavelength would achieve the most effective tissue penetration. However, we wanted to compare PtG4 with other commercially available phosphorescent oxygen sensors to identify the reagents with the best qualities for utilization in CELSI. Several factors are involved in the selection of ideal compounds. As with fluorescence imaging, the quantum yield of the compound is a significant factor that determines the efficiency of imaging. In CEL imaging, time gating is used to separate Cherenkov emissions from luminescence, so luminescence lifetime is also important to consider. Medical LINACs have a 4-μs radiation pulse, and so this moderate pulse time limits the time gating to the microsecond regime, and without significant change to the LINAC or acquisition, nanosecond lifetimes would not be possible to measure. The luminescent agents that are available with microsecond lifetimes are typically phosphorescent. It has not been feasible to detect fluorescent agents that have nanosecond lifetimes, other than through wavelength filtering and continuous wave detection. 21 The typical timing used for image acquisition that is coupled to the LINAC is shown in Fig. 6(a). Experimentally, the gate width is generally set between 5 and 10 times the deoxygenated lifetime of the reagent of interest (τ 0 ). The CEL is then detected at multiple delay times in order to determine the emission lifetime of each phosphorescent agent. The time between radiation pulses is one factor that will limit the maximum lifetime when choosing a phosphorescent compound. The LINAC generates radiation pulses at a repetition rate between 2.7 and 17 ms (600 to 100 MU∕ min). Shorter lifetimes can be difficult to discriminate from the radiation-induced Cherenkov light. The LINAC radiation pulse does not immediately turn off, and both stray charge and Cherenkov emissions may be detected during the transition phase. Phosphorescent compounds with lifetimes in this range (<4.5 μs) will be difficult to detect. A plot of the decay curves depicting lifetime measurements as a function of quantum yield for each compound used in the study is shown in Fig. 6(b). IrðbtbÞ 2 ðacacÞ has been utilized in LED applications and has a high quantum yield of 0.33, 26 which should also impart ideal imaging properties; however, its shorter emission lifetime of 5.8 μs 26 may limit its application in CELSI. While PtG4, MM2, and MitoID have lower quantum yields, their longer emission lifetimes should be advantageous for signal detection. In addition, MM2 and MitoID are nanoparticles with multiple copies of oxygen-sensitive porphyrins, 25 which should provide enhanced signal-to-noise ratio. With optimization in mind, a comparison of commercially available oxygen sensors was completed in vitro, comparing signals in ambient oxygen to that in the presence of oxygen-scavenging GODCAT. These conditions were employed since phosphorescent oxygen sensors exhibit maximum luminescence in the absence of oxygen. PtG4 was not employed in this study since the maximum cutoff for the fluorescence emission of the plate reader is shorter than the PtG4 fluorescence emission. As expected, these reagents show a signal enhancement in the presence of GODCAT conditions, indicating a response to change in oxygenation levels. MM2 and IrðbtbÞ 2 ðacacÞ gave the most robust responses in sensing oxygen changes via steady state fluorescence [ Fig. 2(a)]. At the highest concentration utilized (30 μg∕mL), the signal was attenuated in comparison to the 20 μg∕mL samples, which could be attributed to self-quenching, although further investigation of this was not done. Due to these observations, since MM2 and IrðbtbÞ 2 ðacacÞ gave the best signal, the next focus was to determine the lowest concentration of each sensor that could be utilized to detect oxygen levels in vitro. We found that IrðbtbÞ 2 ðacacÞ still elicited a luminescence signal in response to oxygen changes in PANC-1 cells at 0.2 μg∕mL. While MM2 exhibited a statistically significant response to oxygen scavenging conditions at a lower concentration (0.02 μg∕mL), the responses to oxygenation changes were not robust at concentrations lower than 2 μg∕mL [ Fig. 2(b)]. We investigated CEL both in solution and loaded into PANC-1 cells of the previously studied PtG4 as well as MM2, IrðbtbÞ 2 ðacacÞ, and MitoID. Both PtG4 and MM2 exhibited strong phosphorescence at 20 μg∕mL in solution, particularly with oxygen scavenging, when excited by Cherenkov emission, with phosphorescence intensities of 1349 and 1269, respectively, whereas IrðbtbÞ 2 ðacacÞ and MitoID did not display significant phosphorescence in solution, with intensities of 250 and 150, respectively [ Fig. 3(a)]. We have previously seen excellent results with CELSI with PtG4, 19,21,22 and since MM2 contains multiple copies of a porphyrin grafted to a nanoparticle, enhanced signal in response to oxygen was anticipated for both of these reagents. As expected, we found that the intracellular phosphorescence intensity was markedly decreased in comparison to the phosphorescence measured in solution.
Interestingly, IrðbtbÞ 2 ðacacÞ exhibits a much better phosphorescence signal in cells versus solution, giving the largest detectable signal in comparison to the other reagents [ Fig. 3(b)]. This is likely due to the fact that this complex binds to albumin or another cellular protein, thereby enhancing permeability. However, previous reports indicate that it is no longer sensitive to oxygen when bound to albumin in vitro. Our results and other cell studies with IrðbtbÞ 2 ðacacÞ that show oxygen sensitivity indicate that binding to albumin is not the mechanism that imparts enhanced signal in cells versus solution. 26 The performance of MM2 was diminished intracellularly, possibly due to lower cell permeability than IrðbtbÞ 2 ðacacÞ. In fact, IrðbtbÞ 2 ðacacÞ was found to penetrate cells far more rapidly than MM2 (30 min versus 24 h). PtG4 gave a robust response to oxygenation changes in cells; however, the intensity was somewhat attenuated in comparison to IrðbtbÞ 2 ðacacÞ. One critical advantage of PtG4 is that the dendrimer shell prevents its interaction with other biomolecules that could potentially perturb its emission lifetime and thus pO 2 estimations. This is considered to be a key factor in making this agent a linear reporter of tissue pO 2 , and in fact PtG4 and similar reagents have been calibrated appropriately in extensive studies. [19][20][21][22][23][24]27 While MitoID displays an enhanced signal in cells versus solution, it still elicits the weakest phosphorescent response to changes in oxygen concentration. PtG4, IrðbtbÞ 2 ðacacÞ, and MM2 seemed to have promise for use in CELSI. Therefore, we investigated these reagents for sensing oxygenation levels in subcutaneous tumors. We were unable to detect significant phosphorescence with MM2 in vivo. However, we were able to sense emission lifetime and as such calculate tissue oxygenation utilizing CELSI for both PtG4 and IrðbtbÞ 2 ðacacÞ at the nmol level [Figs. 5(b)-5(d)]. Unfortunately, we were not able to measure fluorescence lifetime or tissue pO 2 in live mice with IrðbtbÞ 2 ðacacÞ. This is likely because the luminescence lifetime in a fully oxygenated sample of IrðbtbÞ 2 ðacacÞ is shorter than we can detect since the shortest delay time feasible for imaging after the radiation pulse is 4.26 μs. We also observed quenching with IrðbtbÞ 2 ðacacÞ at concentrations higher than 50 μM in subcutaneous tumors, as we did in the fluorescence plate reader experiments.
A summary of the results as well as a comparison of the pros and cons of the oxygen sensors we utilized for this study are provided in Table 1. Briefly, both PtG4 and IrðbtbÞ 2 ðacacÞ are stable indefinitely, whereas MM2 and MitoID nanoparticles have a short shelf life of only 1 to 2 weeks once resuspended. PtG4 has a dendrimer shell that prevents its interaction with biomolecules, allowing accurate determinations of fluorescence lifetime and pO 2 calculations. IrðbtbÞ 2 ðacacÞ and MitoID appear to interact with biomolecules in some way, given the difference in phosphorescent intensities discovered in solution versus cells. Finally, we have observed that MM2, MitoID, and IrðbtbÞ 2 ðacacÞ exhibit self-quenching at higher concentrations, where we have not seen this effect with increasing concentrations of PtG4.

Conclusion
When used in vitro in PANC-1 cells at the same concentrations, the signal strengths suggest that the optimal agents would be IrðbtbÞ 2 ðacacÞ, followed by PtG4, and then MM2 and MitoID. However, in vivo, only PtG4 and IrðbtbÞ 2 ðacacÞ were found to be measurable at nmol doses in the tumors, and assessment of tissue oxygenation accomplished. On the timescale of LINAC produced x-ray imaging, PtG4 is the more ideal agent because of the lifetime on the scale of tens of microseconds versus IrðbtbÞ 2 ðacacÞ, which has a lifetime Experimentally, the gate width is generally set between 5 and 10 times the deoxygenated lifetime (τ 0 ) of the reagent of interest. Luminescence is detected at a series of delay times (red blocks) after the Cherenkov emission (blue column) to assess the emission lifetime of phosphorescent agents. The LINAC generates radiation pulses at a repetition rate between 2.7 and 17 ms (600 to 100 MU∕ min). (b) Plot of emission decay curves as a function of quantum yield. of a few microseconds, which hinders detection at higher tissue pO 2 values. This work represents: (1) a unique way to harness Cherenkov emission for imaging purposes in conjunction with radiotherapy, (2) a noninvasive determination of tumor pO 2 , and (3) a direct comparison of phosphorescent sensors available to probe tissue oxygenation.

Disclosures
The authors report no conflicts of interest.