1.

Introduction

The noninvasive measurement of tissue oxygenation is challenging and has been addressed by numerous research groups for decades.^1,2 Tissue hypoxia can be caused by a number of factors, e.g., low oxygen partial pressure (${\mathrm{pO}}_2$) in arterial blood (due to pulmonary disease or high altitude), reduced ability of blood to carry oxygen (as a result of anemia, methemoglobin formation, carbon monoxide poisoning, reduced tissue perfusion), inability of cells to use oxygen because of intoxication, or other factors.³ Tissue oxygenation is also relevant for some therapeutic approaches, such as radiotherapy, where microcirculation and adequate oxygen supply are major success factors:⁴ it was reported that nearly triple radiation dose is needed to kill hypoxic cells as compared to normal aerobic cells.⁵ Prognosis for survival and recurrence-free survival of patients with hypoxic tumors is significantly shorter⁶ than for patients with nonhypoxic tumors. The measurement of tissue oxygenation is, therefore, of great clinical interest not only to measure tumor oxygenation, but also as a predictive assay for future treatment development and diagnostics (radiotherapy, chemotherapy, photodynamic therapy).^7,8 It should be noted that measuring ${\mathrm{pO}}_2$ also provides valuable fundamental information regarding tissue respiration.⁹

During the last decades, two main methods have been used for ${\mathrm{pO}}_2$ measurements in tissue: Eppendorf polarographic needle electrodes and optical fiber–based sensors.^10,11 Both methods are based on the use of needles or interstitial catheters with diameters on the order of several hundreds of microns. Consequently, in vivo measurements with these methods induce tissue damage or modification. Following the development of sensitive and noninvasive time-resolved optical spectroscopic techniques based on the oxygen-dependent luminescence quenching of molecular probes, a third method is now available, whereby the level of oxygenation can be measured quantitatively and minimally invasively at selected sites.^12–14 Time-resolved luminescence detection techniques to measure ${\mathrm{pO}}_2$ are, in general, more reliable than luminescence intensity-based methods, because luminescence lifetime does not depend on the concentration of a probe.^15,16 Unfortunately, most of these luminescent oxygen probes are phototoxic, due to the production of singlet oxygen in the quenching process of their triplet state.¹⁷ This phototoxicity leads to tissue damage, including vascular damage, during the ${\mathrm{pO}}_2$ measurements.¹²

Luminescent complexes of transition metals, i.e., Ru(II), Os(II), Rh(III), Ir(III), and Pt(II), are very promising ${\mathrm{pO}}_2$ sensors. They display useful properties, such as long luminescence lifetimes, high luminescence quantum yields, high extinction coefficients in the visible range, and good chemical, thermal, and photochemical stabilities.^18,19 Following preliminary studies indicating that the ruthenium organometallic complex dichlorotris(1, 10-phenanthroline)-ruthenium(II) hydrate [Ru(Phen)] (see chemical structure in Fig. 1) presents both a limited phototoxicity and oxygen-dependent luminescence lifetimes, we decided to study it in vivo for ${\mathrm{pO}}_2$ measurements, thus expanding on prior studies in liquids.^20–22 Ru(Phen) presents three main absorption bands, shown in Fig. 1. The first and most intense band (262 nm) is due to the 1,10-phenanthroline ligand $\pi -\pi *$ transition, the broad band located between 350 and 500 nm is due to the $d_{\pi }-\pi *$ metal-to-ligand charge transfer, and the smallest one at 684 nm corresponds to the metal $d\text{-}d$ transition.^23,24 The broad emission of this complex at 600 nm (see Fig. 1) can be efficiently quenched by oxygen. It is assigned to a phosphorescence process resulting from a metal-to-ligand charge transfer.²⁴ This emission band is very bright and highly photostable.²⁵

Fig. 1

Absorption (solid line) and emission spectrum (dotted line, $\text{excitation}=470\mathrm{nm}$) of dichlorotris(1, 10-phenanthroline)-ruthenium(II) hydrate [Ru(Phen)] (1 and $0.01\mathrm{mg}/\mathrm{ml}$, respectively) in 0.9% NaCl isotonic solutions. Drawing of Ru(Phen)’s molecular structure is inserted.

Ru(Phen) is a complex cation, which does not have specific active groups. Biological activity of such compounds is a function of the cation as a whole and not of the metallic atom.²⁶ Antitumoral and virostatic action of Ru-based polypyridyl complexes was observed both in vivo and in vitro 50 years ago.^27–30 Ru(Phen) is modestly cytotoxic (${\mathrm{IC}}_{50}\sim 90\mu \text{mol}/\mathrm{l}$) and possesses relatively high clearance rate in vivo.^31–33 In general, in vivo toxicity strongly depends on the way of Ru(Phen) administration. Minimum lethal doses of Ru(Phen) in mice are $6.6\mathrm{mg}/\mathrm{kg}$, when administered by intraperitoneal injection.³² In another study, intravenous injection of Ru(Phen) between 4 and $16\mathrm{mM}/\mathrm{kg}$ in rats frequently led to a drop in heart rate and respiratory arrest. A fast decrease of Ru(Phen) concentration in the bloodstream due to a high renal clearance rate was also observed.³³ Interestingly, there was neither any toxic effect observed after per os administration of ${}^{103}\mathrm{Ru}$(Phen) in digestion studies with sheep, nor any disruption of metabolic activity of the rumen microbiota.³⁴ Apart from a few references to a minimal phototoxicity of Ru(Phen) during the measurements, no in vivo or in vitro phototoxicity tests have been published to the best of our knowledge.^35,36

In the present study, we have quantified the phototoxicity of Ru(Phen) in vivo after intravenous injection. The model we used for this purpose is the chicken chorioallantoic membrane (CAM), a model widely used in preclinical studies of the microvasculature.^37–39 In addition, we report the first in vivo measurement of tissue ${\mathrm{pO}}_2$ based on the time-resolved detection of Ru(Phen) luminescence. These measurements have been performed using a unique optical fiber-based, time-resolved spectrometer developed in our laboratory.⁴⁰ Our study demonstrates that the fluence necessary to perform a tissue ${\mathrm{pO}}_2$ measurement is lower than the phototoxic threshold by about two orders of magnitude. This avoids any possible photoinduced vascular effects. Finally, an illustrative application of this approach to measure intratumor oxygenation is presented in human tumors xenografted on CAM.

2.

Materials and Methods

2.2.

Chicken Embryo Chorioallantoic Membrane Preparation

Fertilized chicken eggs (Animalco AG, Switzerland) were transferred into an automatic turn incubator (FIEM snc, Italy). Eggs were incubated blunt end up during 3 days at 37°C, at 65% of relative humidity and at atmospheric oxygen pressure (155.4 mm Hg). On the third embryo development day (EDD), a small part of the shell on the pointed end was removed creating a hole ($\sim 3\mathrm{mm}$ in diameter), which was then covered by tape (Scotch® Magic™, St. Paul, Minnesota). Eggs were then returned into the incubator blunt end down in a static position until used. At EDD 11, the hole in the shell was enlarged to $\sim 2.5\mathrm{cm}$ in diameter, enabling easy CAM observation. Each treated CAM was placed either under an epifluorescence microscope Nikon eclipse E 600 FN (Nikon, Japan) or into the gas chamber (see Fig. 2) for further measurements and processing.

Fig. 2

Schematic representation of in vivo sample part of the time-resolved luminescence lifetime experimental setup for oxygen partial pressure (${\mathrm{pO}}_2$) measurement. Egg is inserted in the closed gas chamber with the optical fiber perpendicularly positioned 2 mm above chorioallantoic membrane (CAM). The gas exchange occurs through the inlet and the outlet of the gas. Irradiation (${\lambda }_{\mathrm{exc}}$) and detection (${\lambda }_{\mathrm{em}}$) is carried out with the same optical fiber from ${1\text{-}\mathrm{mm}}^2$ spot. Two main areas of interest are defined: (a) intravascular compartments ($200\mu \text{m}<\text{vessels}$), indicated with black circle, and (b) extravascular compartments ($\text{vessels}<100\mu \text{m}$), indicated with red circle (red circle appears gray in BW figure).

2.3.

CAM-Tumor Model Preparation

Human A2780 ovarian carcinoma cells were used to establish tumor xenografts onto the CAM, as described in previous studies.⁴¹ Briefly, tumor cells were suspended in an RPMI-1640 medium, GlutaMAX™ (Gibco, United Kingdom) supplemented with 10% fetal bovine serum, 1% penicillin/streptomycin, and containing 20% methocel (Sigma-Aldrich). Several 50-ml drops (${1\times 10}^6\text{cells}/\text{drop}$) were dispensed on the internal part of the lid of a Petri dish and were incubated in a humid atmosphere containing 5% ${\mathrm{CO}}_2$ at 37°C in the dark. After 24 h, spheroids were harvested and placed on the CAM (EDD 7). CAM were prepared as described above until EDD 7; the hole in the shell was enlarged in order to provide better access for spheroids placement. CAM with spheroids was incubated at 37°C, 65% relative humidity, and atmospheric oxygen pressure (155.4 mm Hg) until used. The size of tumor xenografts was estimated and $\sim 3\times 3\mathrm{mm}$ spheroids were used for further measurements (see Fig. 3).

Fig. 3

(a) Representative luminescence angiography of CAM with the tumor 10 min after Ru(Phen) ($10\mathrm{mg}/\mathrm{kg}$) intravenous injection and (b) before Ru(Phen) administration. (c) Ru(Phen) reciprocal lifetimes determined in the physiological and tumoral vessels. (**$p<0.001$, seven eggs). Physiological and tumoral vessels are denoted with corresponding arrows. (d) Low-magnification ($1.6\times$) white light image of the CAM presents a grafted tumor marked by the arrow.

2.5.

Ru(Phen) Pharmacokinetics and Biodistribution in CAM

Ru(Phen) pharmacokinetics measurements in CAM were performed as described in previous studies.³⁷ Briefly, 20 μl of solution containing $10\mathrm{mg}/\mathrm{kg}$ of Ru(Phen) were injected under the microscope (within 5 s) into the main vein of CAM. Ru(Phen) excitation was performed by a Hg-arc lamp (HBO $103\mathrm{W}/2$, Osram, Germany) filtered at $470\pm 20\mathrm{nm}$. Excitation and Ru(Phen) emission were separated by a 505-nm dichromatic mirror, and a long-pass emission filter at 520 nm was used to reject all excitation light. Images showing the Ru(Phen) luminescence were recorded 10 min after intravenous injection at regular 1-min intervals. These images of the CAM surface were recorded with a digital scientific camera (PCO.1300, PCO Imaging, Germany) and a low-magnification objective (4X/0.13, Plan Fluor $\infty /-$, Nikon, Japan) as a function of time, with 100-ms exposure. Ru(Phen) luminescence radiance inside the blood vessels ($I_{\text{in}}$) with respect to the extravascular tissues ($I_{\text{out}}$) enables determination of the normalized photographic contrast $C_{\mathrm{phot}}$ according to Eq. (1):

Eq. (1)

$$C_{\mathrm{phot}}=[I_{\text{in}}-I_{\text{out}}]/[(I_{\text{in}}+I_{\text{out}})\times C_{\mathrm{phot},\max }],$$

where $C_{\mathrm{phot},\max }$ represents the highest photographic contrast. This photographic contrast was determined for areas corresponding to the center and the walls of the vessels (see Fig. 4). Ru(Phen) luminescence radiance measurements were performed by analyzing the signal recorded along a profile perpendicular to the vessel axis using the Image J software (National Institutes of Health, Bethesda, Maryland). These measurements have been performed for vessels (mainly veins) presenting five different diameters ranging between 50 and 500 μm. One point presented in the pharmacokinetics curve (see Fig. 4) corresponds to the average value of measurements performed with three different eggs.

Fig. 4

Pharmacokinetics of Ru(Phen) during 10 min following intravenous injection defined by photographic contrast ($C_{\mathrm{phot}}$) in Eq. (1). An insert figured biodistribution of Ru(Phen) 10 min after injection (see Fig. 7). White circles correspond to $C_{\mathrm{phot}}$ assessed in the wall of the blood vessels as displayed in the inserted picture. Black circles correspond to $C_{\mathrm{phot}}$ estimated in the intravascular space (see insert). (*$p=0.0408$).

Vascular luminescence imaging and spectroscopy of photosensitive drugs is often strongly influenced by the hemoglobin contained in the bloodstream. This pigment strongly absorbs below 600 nm, with a maximum at 405 nm.⁴² We minimized this hemoglobin absorption effect in vivo by using an excitation wavelength at 470 nm, which is a compromise between light transmission trough small vessels and Ru(Phen) excitation (Fig. 1).

2.7.

Ru(Phen) Phototoxicity Assessment

On EDD 11, eggs were placed under the microscope and a treatment area was defined by placing a small plastic ring (8 mm in diameter; thickness: 1 mm) on the CAM membrane. The purpose of this plastic ring was to facilitate the identification of the illuminated spot 24 h after the treatment. Prior to injection of Ru(Phen), an autofluorescence image of the CAM surface was recorded. Subsequently, Ru(Phen) was injected in situ under the microscope at doses ranging between 1 and $20\mathrm{mg}/\mathrm{kg}$. Irradiation of the $2.2\text{-}{\mathrm{mm}}^2$ treated area was performed with the same microscope and filter set as described above in the biodistribution study, with an irradiance of $64\mathrm{mW}/{\mathrm{cm}}^2$. The fluence was 1, 2, and $10\mathrm{J}/{\mathrm{cm}}^2$ (corresponding to irradiation times of 16, 32, and 160 s), and the drug-light interval (DLI) was 1 or 10 min. Eggs were returned to the incubator and incubated in the dark until the following day.

The evaluation of Ru(Phen) phototoxicity was based on the vascular damages induced on the CAM membrane 24 h after irradiation. The status of the vessels was controlled by injecting 20 μl of FITC-dextran solution in the vein and performing a fluorescence angiography 1 min after the injection. The spectral design of the microscope was not changed for this angiography. One milliliter of black ink (Parker, France) was injected into the extra-embryonic cavity of CAM to screen the autofluorescence produced by deep-seated tissues. The diameter of the vessels was measured with the Image J software. This distance in pixels was recalculated and used as the scale displayed in Fig. 5. The diameter of the largest closed vessel was measured and used as an index of vascular damage range between 0 and 5, as described in detail in previous publications.^37,38 The minimal value “0” refers to no detectable photodamage to the vasculature. Closure of the smallest visible blood vessels (diameter 5 to 10 μm) was termed grade 1, closure of the vessels with a diameter ranging between 10 and 30 μm grade 2, closure of the vessels with a diameter ranging between 30 and 70 μm grade 3, and closure of the vessels with diameter 70 μm and partial closure of larger vessels grade 4. A total occlusion of the irradiated area was assigned grade 5. These data were plotted in three-dimensional graphs ($\text{vascular damages}\times \text{fluence}\times \text{drug dose}$). The average values and standard deviations obtained with four eggs are presented in such graphs (see Fig. 5).

Fig. 5

Phototoxic effects induced by Ru(Phen) on the CAM blood vessels for drug-light intervals (DLIs) of 1 and 10 min. Typical luminescence angiography of $20\mathrm{mg}/\mathrm{kg}$ Ru(Phen) performed (a) 1 min and (d) 10 min after injection before irradiation. The diameter of the illuminated area is 2 mm. Fluorescence angiography of fluorescein isothiocyanate-dextran performed 24 h after illumination with (b) $1\mathrm{J}/{\mathrm{cm}}^2$, DLI 1 min, index of vascular damage “0,” (c) $10\mathrm{J}/{\mathrm{cm}}^2$, DLI 1 min, index of vascular damage “3,” (e) $1\mathrm{J}/{\mathrm{cm}}^2$, DLI 10 min, index of vascular damage “0,” (f) $10\mathrm{J}/{\mathrm{cm}}^2$, DLI 10 min, index of vascular damage “3,” (excitation at $470\pm 20\mathrm{nm}$). (g) Three-dimensional graphs reporting the index of vascular damage induced by different light and Ru(Phen) doses ranging between 1 to $10\mathrm{J}/{\mathrm{cm}}^2$ and 1 to $20\mathrm{mg}/\mathrm{kg}$, respectively. Four eggs were treated per condition. Error bars represent the standard deviations of a confidence interval 68%. See also Sec. 2 for more details.

2.9.

${\mathsf{pO}}_2$ Measurements Based on Ru(Phen) Luminescence Lifetime

2.9.1.

Measurements in solution

On EDD 11, 300 μl of blood were carefully drawn from the vein of CAM and placed in a conical microcentrifuge tube with anticoagulant (ethylenediaminetetraacetic acid, Eppendorf®, Hauppauge, New York). A fraction of CAM blood (150 μl) was separated by centrifugation (1 min, 1200 rpm) to obtain blood cell-free serum fraction. Two types of solutions were prepared: (1) CAM blood and (2) CAM blood serum (hereinafter, serum). Fifty microliter of CAM blood or serum were mixed with 20 μl Ru(Phen) solutions (0.5 and $0.01\mathrm{mg}/\mathrm{ml}$) and placed in 150-μl plastic containers. These samples were stabilized during 1 h in the dark and installed in a chamber (0.28 l) with gas, temperature, and humidity controlled. The chamber is integrated in the lifetime measurement setup (Fig. 2). The fiber of the time-resolved spectrometer was positioned perpendicularly $\sim 2\mathrm{mm}$ above the solutions. This distance was defined in such a way that the luminescence was detected from a $1\text{-}{\mathrm{mm}}^2$ surface (Fig. 2). Ru(Phen) solutions were subjected to gas flow during 15 min with defined oxygen concentration (0, 5, 10, 15, and 21% of ${\mathrm{O}}_2$ in ${\mathrm{N}}_2\mathrm{M}/\mathrm{M}$) prior to measurement. The gas flow ($\sim 3\mathrm{l}/\mathrm{h}$) and humidity (100%) were controlled and measured with the gas mixer (The BRICK). Temperature was stabilized at 30°C with a thermostabilizer (The CUBE; Life Imaging Services GmbH, Switzerland). The temperature was kept the same as the CAM temperature during in vivo ${\mathrm{pO}}_2$ measurements. The reciprocal values of Ru(Phen) luminescence lifetimes were plotted for different ${\mathrm{pO}}_2$. This approach enables us to perform a linear fit of the data since the lifetime and ${\mathrm{pO}}_2$ are governed by the Stern-Volmer relation [Eq. (3)]:

Eq. (3)

$${\tau }_0/\tau =1+{\tau }_0{k}_q({\mathrm{pO}}_2),$$

where $\tau$ and ${\tau }_0$ are the Ru(Phen) luminescence lifetimes in the presence and absence of oxygen, respectively. $k_q$ is the bimolecular quenching constant and (${\mathrm{pO}}_2$) is oxygen partial pressure. Eight lifetime measurements were averaged to generate one point in Fig. 6.

Fig. 6

(A) ${\mathrm{pO}}_2$ resolution of [(a) and (b)] Ru(phen) or [(c) and (d)] PdTCPP represented by the ratio of the luminescence lifetimes measured in the absence (${\tau }_0$) and presence ($\tau$) of oxygen for different ${\mathrm{pO}}_2$ values of a gas mixture in biological liquids: (a) RuPhen in the serum, (b) Ru(Phen) in the blood ($p=0.6$, serum/blood), (c) PdTCPP in the serum, and (d) PdTCPP in the blood (*$p=0.03$, blood/serum). (two repetitions). (B) Stern-Volmer characteristics of Ru(Phen) reciprocal luminescence lifetimes measured in vivo as a function of the environmental ${\mathrm{pO}}_2$ (f) in the intravascular space and (e) in the extravascular space, and the ${\mathrm{pO}}_2$ calibration curves obtained (a) in the blood and (b) in the serum. The significantly measured values of differences $\mathrm{\Delta }{\mathrm{pO}}_2$ and $\mathrm{\Delta }1/\tau$ in our condition are displayed by black lines with arrows.

The same protocol and approach were applied to obtain the comparative Stern-Volmer relations by using standard oxygen sensor Oxyphor R0 ($0.5\mathrm{mg}/\mathrm{ml}$ PdTCPP at pH 7.4) in the CAM blood and serum [V (PdTCPP): $\mathrm{V}(\text{blood}/\mathrm{serum})=20\mu \text{l}$ : 50 μl).

2.9.2.

In vivo measurements

In vivo measurements of Ru(Phen) luminescence lifetimes were performed at EDD 11 and 12. An egg was placed into a chamber (0.28 l) with gas, temperature, and humidity controlled, integrated in the lifetime measurement setup (Fig. 2). Twenty microliters of Ru(Phen) solution corresponding to a dose of $1\mathrm{mg}/\mathrm{kg}$ were injected into a vein. The fiber of the time-resolved spectrometer was positioned perpendicularly $\sim 2\mathrm{mm}$ above the membrane. Two main areas of interest were defined: (1) intravascular compartments ($100\mu \text{m}<\text{vessels}$) and (2) extravascular compartments ($\text{vessels}<100\mu \text{m}$) (see Fig. 2). A delay of 10 min was observed before all measurements, while flushing the gas chamber with a flux of $\sim 3\mathrm{l}/\mathrm{h}$, after each change of ${\mathrm{O}}_2$ concentration. Oxygen concentrations of 0, 5, 10, 15, and 21% ${\mathrm{O}}_2$ in ${\mathrm{N}}_2\mathrm{M}/\mathrm{M}$ were applied. During the measurements, the temperature of CAM was measured with noncontact infrared thermometer (Raytek a Fluke Company, Santa Cruz, California) positioned 1 cm above the CAM and covering a $25\text{-}{\mathrm{mm}}^2$ surface. The mean value of CAM temperatures was 30°C. Ru(Phen) excitation and luminescence were performed on a $1\text{-}{\mathrm{mm}}^2$ surface. The 470-nm laser fluence applied for one measurement did not exceed $120\mathrm{mJ}/{\mathrm{cm}}^2$. The data were processed and analyzed as mentioned above by fitting the measurements with the mono-exponential function [Eq. (2)]. Reciprocal lifetimes of Ru(Phen) in intra/extravascular compartments were inserted in the Stern-Volmer relation [Eq. (3)]. Six eggs were investigated per set of experimental conditions.

3.

Results

3.1.

Ru(Phen) Pharmacokinetics and Biodistribution in CAM

Typical luminescence angiographies obtained after intravenous injection of Ru(Phen) solutions (dose of $10\mathrm{mg}/\mathrm{kg}$) for vessels presenting diameters of $\sim 70$ and 450 μm (see white arrows) are given in Fig. 7.

Fig. 7

Biodistribution of Ru(Phen) ($10\mathrm{mg}/\mathrm{kg}$, intravenous injection) within large and small CAM blood vessels as well as in the extravascular space: (a) typical CAM picture prior the Ru(Phen) and black ink administration, (b) 1 min after Ru(Phen) injection, (c) 10 min after Ru(Phen) injection. Profile plots represent local intensities of Ru(Phen) throughout the section of the blood vessel marked with white arrow.

Uptake of Ru(Phen) by the walls appears in the profile plot as two intense maxima [Fig. 7(c)]. Ru(Phen) intravascular luminescence intensity was high compared to the extravascular space in both small and large vessels shortly (20 s) after injection, suggesting that Ru(Phen) is inside the vessel and fully distributed throughout the entire vascular system already 1 min after the injection [Fig. 7(b)]. Profile plot of the vessel section shows increased fluorescence intensity in the area corresponding to the intravascular space during the initial minute after injection, thus confirming homogeneous intravascular distribution [Fig. 7(b)]. While the profile plot of the vessel section showed higher intensities in the extravascular space compared to intravascular space in Fig. 7(a), this observation is reversed in Fig. 7(b): intravascular intensity of Ru(Phen) was more than 100 times higher than extravascular intensity. Interestingly, luminescence angiography of Ru(Phen) performed 10 min after injection showed a high accumulation of Ru(Phen) in the walls of the large vessels and initial accumulation in the extravascular space [Fig. 7(c)]. It should be noted that the intensity ratio between intracellular and extracellular spaces rapidly decreased (see Fig. 4). This indicates that Ru(Phen) leaks out of the vessels. Therefore, the luminescence pharmacokinetics of Ru(Phen) was evaluated within 10 min. Normalized photographic contrast $C_{\mathrm{phot}}$ for intravascular space and vessel walls determined from Eq. (1) at different times following the injection is presented in Fig. 4. The intravascular $C_{\mathrm{phot}}$ rapidly decreases to 50% of its initial value, reaching a plateau at 0.55 after 5 min (Fig. 4). On the other hand, $C_{\mathrm{phot}}$ in the wall increases to the plateau at 0.8, which represents 80% of the initial value of intravascular $C_{\mathrm{phot}}$. As $C_{\mathrm{phot}}$ represents intravascular concentration of Ru(Phen), it means that $\sim 20\%$ of initial concentration is lost.

4.

Discussion

The biodistribution and further accumulation of a drug in a tissue depends on its hydrophobic or hydrophilic properties, as well as its affinity for biomacromolecular structures (e.g., subcellular organelles, membranes, proteins, etc.). Hydrophilic compounds, such as fluorescein, usually have a very fast body clearance, mainly via renal excretion.⁴³ In many cases, this fast clearance is advantageous because it limits the drug toxicity. A rapid decrease of Ru(Phen) concentration in the bloodstream was observed within 10 min after intravenous injection in rodents.³¹ Ru(Phen) was excreted in urine and no trace thereof was found in the nervous system.³¹ This is one of the reasons Ru(Phen) was, in an experiment to visualize ${\mathrm{pO}}_2$ in hepatic tissues, continuously infused in rats during intravital microscopy to reach a constant Ru(Phen) concentration in plasma.³³

We have also observed a rapid decrease in Ru(Phen) intravascular intensity during the first minutes after intravenous injection (Fig. 7). This can be due to a fast leakage of the molecules out of the vessels (faster leakage from small vessels than from larger vessels). For the sake of this study, we proposed to talk about intravascular measurements when the fiber was probing an area containing vessels with diameters $>100\mu \text{m}$, whereas the term extravascular was used when this area contained smaller vessels only. However, this clearance was not similar to that of a typical hydrophilic compound. Indeed, the temporal evolution of the photographic contrast after intravenous injection of hydrophilic Ru(Phen) in the CAM is very similar to that of the more lipophilic benzoporphyrin derivative monoacid ring A (BPD-MA),³⁷ whereas we would have expected a behavior similar to that of water-soluble fluorescein.³⁷ Additionally, we found an increasing Ru(Phen) accumulation in the walls of large ($\text{diameter}>100\mu \text{m}$) vessels (Fig. 7). These biodistribution and pharmacokinetics observations suggest that the ${\mathrm{pO}}_2$ can be realistically measured in three compartments (bloodstream, vessel wall, extravascular space) with a time-resolved microspectrophotometric setup and at appropriate times after injection of appropriate Ru(Phen) doses. It is interesting to note that this sets the Ru(Phen)-mediated measurement of ${\mathrm{pO}}_2$ apart from more standard, noninvasive techniques, such as pulse oxymetry, which can only measure intravascular changes of oxygen.

We were not able to distinguish subcellular localization of Ru(Phen) in vivo, but cell culture studies showed Ru(Phen)’s possible localization in the endocytic vesicles in the cytoplasm and in the membrane.³⁶ It was reported that even if Ru(Phen) is transported inside the cells via endocytic pathway, the content concentration of Ru(Phen) in the extracellular space remains higher than in the intracellular space.³⁶ Thus, any energy transfer between excited Ru(Phen) and oxygen may occur mainly near cellular membranes.

We demonstrated that Ru(Phen) has limited dark toxicity and phototoxicity. This sets it apart from a number of other molecules of relevance to photodynamic therapy (PDT) (porfimer sodium, verteporfin, and purpurins), which display high phototoxicity, and can cause vessel closure. Several studies report alterations of the vascular endothelium during irradiation of tissues in vivo, photosensitized with them.^12,44–47 At short DLIs, in vivo^12,44,48 endothelial dysfunction and blood flow stasis followed by vessel occlusion appear to be their predominant mechanisms of damage. Moreover, a direct correlation between the vascular changes and the amount of photosensitizer in circulation during photodynamic therapy has been suggested.⁴⁴

This also holds true for commonly used oxygen sensor PdTCPP bound to albumin, whose intravascular concentration remains nearly constant over several hours and has been shown to be highly phototoxic.¹² While PdTCPP-albumin complex is kept inside the vessels, Ru(Phen) weakly binds with proteins⁴⁹ and rapidly leaks out of the vessels to extravascular space (Fig. 7). Mechanisms of vascular damage induced by Ru(Phen) will probably be governed by pathways other than those of PdTCPP.

Photodamage is the result of the interaction of photoactivated molecules with oxygen due to singlet oxygen production. In contrast to approved photosensitizers used in photodynamic therapy, e.g., verteporfin [BPD-MA, quantum yield of singlet oxygen production ${\mathrm{\Phi }}_{\mathrm{\Delta }}$ in methanol is 0.78 (Ref. 50)], the quantum yield of singlet oxygen production (${\mathrm{\Phi }}_{\mathrm{\Delta }}$) of Ru(Phen) is very low [${\mathrm{\Phi }}_{\mathrm{\Delta }}$ in ${\mathrm{D}}_2\mathrm{O}$ is 0.24 (Ref. 16)], but higher than, e.g., fluorescein used for photodiagnosis in ophthalmology [${\mathrm{\Phi }}_{\mathrm{\Delta }}$ in ${\mathrm{H}}_2\mathrm{O}$ is 0.03 (Ref. 16)]. We found that we need only $0.2\mathrm{mg}/\mathrm{kg}$ of BPD-MA,⁵¹ but up to $20\mathrm{mg}/\mathrm{kg}$ of Ru(Phen) to reach photodamage with vascular index 3 after application of a fluence of $10\mathrm{J}/{\mathrm{cm}}^2$.

While photodynamic therapy requires enough cellular damage to kill a cell, ${\mathrm{pO}}_2$ measurements need to avoid this outcome. Highly sensitive detection systems, low Ru(Phen) concentration ($1\mathrm{mg}/\mathrm{kg}$) with high emission quantum yield, and fluence two orders of magnitude lower than phototoxic threshold ($120\mathrm{mJ}/{\mathrm{cm}}^2$) are an excellent combination for noninvasive ${\mathrm{pO}}_2$ assessment.

We carried out a comparative study of oxygen sensitivity of Ru(Phen) and PdTCPP, two water-soluble molecules, in a different microenvironment related to in vivo conditions: a blood and a serum sample. In these conditions, intact blood cells are present in the blood sample. The serum cell-free fraction consisted of a natural mixture of CAM serum proteins. Each modification of the environment can lead to modifications of the oxygen sensitivity of probes. PdTCPP is a highly pH-sensitive probe and well dissolved at alkaline pH. Similar to other porphyrin-based complexes, its primary subcellular targets are mitochondria.⁵² Thus, it is not surprising that PdTCPP’s $k_q$ value is higher in CAM blood samples than in the serum fraction, where it can be dissolved in red blood cell mitochondria. Ru(Phen) in contrast to PdTCPP is relatively pH independent,⁵³ and $k_q$ values of Ru(Phen) in both microenvironments are not significantly different (Table 1).

Both PdTCPP and Ru(Phen) luminescence lifetimes showed a linear Stern-Volmer ${\mathrm{pO}}_2$ dependence in CAM blood and serum solutions (Fig. 6). However, we have measured significantly shorter ${\tau }_0$ of Ru(Phen) in the blood samples than in serum. This is a surprising result, as we would have expected to measure the same Ru(Phen) luminescence lifetime in both microenvironments. A possible explanation could be that measurements were performed starting at physiological temperature and then going down to 30°C as the eggs stayed outside the incubator and cooled somewhat. Such a temperature does not suppress metabolism in the blood cells. In conditions of normoxia (ambient ${\mathrm{pO}}_2$), enzymatic process of glycolysis is inhibited by the presence of oxygen, and pyruvate in mitochondria is oxidized to ${\mathrm{CO}}_2$ and ${\mathrm{H}}_2\mathrm{O}$.⁵⁴ When ${\mathrm{pO}}_2$ decreases, hypoxia induces acceleration of glycolysis.⁵⁵ In this process, glucose is converted to lactate.⁵⁵ It was reported that erythrocytes in the hypoxic condition accelerate glucose consumption.⁵⁶ Production of lactic acid leads to pH drop. As it is known, erythrocytes deliver molecular oxygen to tissues through allosteric regulation of hemoglobin;⁵⁷ the oxygen affinity of hemoglobin decreases with lowering pH, which results in the release of oxygen from oxyhemoglobin.⁵⁸ This residual oxygen can be further involved in the luminescence quenching process and induce shortening of Ru(Phen) lifetimes in the blood samples. Thus, it could be envisioned that Ru(Phen) could be used to monitor not only ${\mathrm{pO}}_2$, but also glycolysis. Additionally, it was recently reported that a combination of Ru(Phen) and a glucose oxidase enzyme was used to monitor glucose in vitro in the macrophage³⁵ and when immobilized in polymeric bilayers as a component of glucose biosensors.^59,60

Oxygen-dependence of Ru[Phen] luminescence lifetimes in solutions was previously reported whether alone or in a combination with other sensors.^20–22 However, there are only few studies about Ru(Phen) oxygen sensitivity in vivo,^33,61,62 and none of them focuses on luminescence lifetimes. The vasculature of CAM creates the main access for gas exchange. Using Ru(Phen) luminescence lifetime measurements, we demonstrated that any fluctuation of external ${\mathrm{pO}}_2$ rapidly influences intravascular ${\mathrm{pO}}_2$ [see Fig. 6(B)]. Ru(Phen) luminescence lifetimes measured in the intravascular and extravascular space were governed by Stern-Volmer oxygen dependences and were not significantly different from each other [Fig. 6(B)]. It should also be noted that during our experiment, we had to change the measurement location on CAM (a heterogenous membrane) between each measurement, which adds real CAM-intrinsic fluctuations to the ${\mathrm{pO}}_2$ fluctuations. Measurements performed in one location only show much fewer variations (data not shown). Wilson et al. observed similar relationship of ${\mathrm{pO}}_2$ in the interstitial space and in the blood plasma on a mouse model using phosphorescence of oxyphors G3 and G2, respectively.⁶³ CAM plexus is well perfused and contains high numbers of erythrocytes for oxygen transport. Thus, Stern-Volmer relations obtained in vivo are closely related to calibration preformed in the blood sample [Fig. 6(B)].

Finally, we have illustrated the advantage of this ${\mathrm{pO}}_2$ measurement approach compared to invasive probes. It should be noted that this advantage is particularly striking when superficial measurements are carried out, i.e., to a depth of several hundred microns. The limited penetration of Ru(Phen) excitation light prevents ${\mathrm{pO}}_2$ measurements deeper in the tissues.⁴² Nevertheless, interstitial measurements of Ru(Phen) luminescence can also be performed with a fiber that can potentially be much thinner than that of the standard systems,^10,11 thus reducing the tissue damages. In our CAM-tumor model, tumoral tissues are less perfused than the physiological tissues.⁶⁴ As ${\mathrm{pO}}_2$ in hypoxic tumor decreases from the surface to the core,⁶⁵ our measurements at the surface of the CAM-tumor model can be affected by ambient ${\mathrm{pO}}_2$. In spite of this, significantly shorter reciprocal lifetimes were found in the tumor compared to physiological tissue. The ${\mathrm{pO}}_2$ fluctuations as a function of distance from the tumoral tissue can be precisely defined using luminescence lifetime imaging. In general, imaging approaches require high luminescent drug concentrations to decrease signal/noise ratio and to acquire a good-resolution image. Because Ru(Phen) possesses enough luminescence intensity for lifetime detection at low concentrations, it is a very promising oxygen probe suitable for lifetime imaging. Furthermore, low phototoxicity of Ru(Phen) offers a possibility to coadminister it in combination with photodynamic agents. Photodynamic therapy can be performed only in oxygenated tissues.⁶⁶ In the absence of oxygen, this therapy is not applicable. Thus, determination of tissue oxygenation with Ru(Phen) can indicate in which tissues PDT will be successful.

5.

Conclusions

In this study, we have evidenced Ru(Phen)’s intra/extravascular biodistribution, fast pharmacokinetics, and very low in vivo phototoxicity. Noninvasive, highly sensitive, optical, time-domain-based luminescence lifetime measurement technique was demonstrated to easily detect Ru(Phen) luminescence lifetimes in different microenvironments. Furthermore, we have demonstrated that Ru(Phen)’s luminescence lifetimes in various microenvironments present linear Stern-Volmer oxygen dependences not only in the biological liquids (CAM blood and serum) but also in vivo in the intra/extravascular space of CAM. Reliable and easy ${\mathrm{pO}}_2$ prediction was illustrated in CAM tumors with this approach.