In vivo O2 imaging in hepatic tissues by phosphorescence lifetime imaging microscopy using Ir(III) complexes as intracellular probes

Phosphorescence lifetime imaging microscopy (PLIM) combined with an oxygen (O2)-sensitive luminescent probe allows for high-resolution O2 imaging of living tissues. Herein, we present phosphorescent Ir(III) complexes, (btp)2Ir(acac-DM) (Ir-1) and (btp-OH)3Ir (Ir-2), as useful O2 probes for PLIM measurement. These small-molecule probes were efficiently taken up into cultured cells and accumulated in specific organelles. Their excellent cell-permeable properties allowed for efficient staining of three-dimensional cell spheroids, and thereby phosphorescence lifetime measurements enabled the evaluation of the O2 level and distribution in spheroids, including the detection of alterations in O2 levels by metabolic stimulation with an effector. We took PLIM images of hepatic tissues of living mice by intravenously administrating these probes. The PLIM images clearly visualized the O2 gradient in hepatic lobules with cellular-level resolution, and the O2 levels were derived based on calibration using cultured cells; the phosphorescence lifetime of Ir-1 gave reasonable O2 levels, whereas Ir-2 exhibited much lower O2 levels. Intravenous administration of NH4Cl to mice caused the hepatic tissues to experience hypoxia, presumably due to O2 consumption to produce ATP required for ammonia detoxification, suggesting that the metabolism of the probe molecule might affect liver O2 levels.

In vivo O 2 imaging in hepatic tissues by phosphorescence lifetime imaging microscopy using Ir(III) complexes as intracellular probes Kiichi Mizukami 1 , Ayaka Katano 1 , Shuichi Shiozaki 1 , Toshitada Yoshihara 1* , Nobuhito Goda 2 & Seiji Tobita 1* 3 Ir (Ir-2), as useful O 2 probes for PLIM measurement. These small-molecule probes were efficiently taken up into cultured cells and accumulated in specific organelles. Their excellent cell-permeable properties allowed for efficient staining of three-dimensional cell spheroids, and thereby phosphorescence lifetime measurements enabled the evaluation of the O 2 level and distribution in spheroids, including the detection of alterations in O 2 levels by metabolic stimulation with an effector. We took PLIM images of hepatic tissues of living mice by intravenously administrating these probes. The PLIM images clearly visualized the O 2 gradient in hepatic lobules with cellular-level resolution, and the O 2 levels were derived based on calibration using cultured cells; the phosphorescence lifetime of Ir-1 gave reasonable O 2 levels, whereas Ir-2 exhibited much lower O 2 levels. Intravenous administration of NH 4 Cl to mice caused the hepatic tissues to experience hypoxia, presumably due to O 2 consumption to produce ATP required for ammonia detoxification, suggesting that the metabolism of the probe molecule might affect liver O 2 levels.

Phosphorescence lifetime imaging microscopy (PLIM) combined with an oxygen (O 2 )-sensitive luminescent probe allows for high-resolution O 2 imaging of living tissues. Herein, we present phosphorescent Ir(III) complexes, (btp) 2 Ir(acac-DM) (Ir-1) and (btp-OH)
Molecular oxygen (O 2 ) is essential for cellular function and is continuously supplied to whole tissues in the body to maintain homeostasis. Oxygen deprivation (hypoxia) in tissues is associated with the pathophysiology of various diseases such as cerebral infarction, fatty liver disease, chronic kidney disease, diabetic retinopathy, and cancer [1][2][3] . Understanding the cellular and molecular mechanisms underlying hypoxia-associated diseases requires O 2 imaging technology capable of detecting the tissue oxygen status in real time and with high resolution. In vivo O 2 detection methods that have been developed so far include oxygen electrodes 4,5 , blood oxygenation level dependent magnetic resonance imaging (Bold MRI) 6,7 , positron emission tomography (PET) with a hypoxia tracer 8 , electron paramagnetic resonance (EPR) oximetry 9,10 , hypoxia markers [11][12][13] , and optical imaging [14][15][16] . These methods can work in vivo, but they have advantages and limitations in terms of applicable targets, spatial resolution, tissue permeability, convenience, reversibility, etc. Of these known methods, optical imaging utilizing O 2 quenching of probe phosphorescence has great advantages in that high-resolution O 2 images can be obtained in a reversible manner, although it is limited to the detection of a penetration depth through tissues of ~ 1 cm even under near-infrared excitation 17 .
To visualize tissue O 2 levels with high sensitivity and high stability, various phosphorescent metal complexes have been developed over the past few decades [14][15][16]18,19 . The most common compounds used as luminescent O 2 probes include Pt(II) and Pd(II) porphyrins, Ru(II) complexes, Pt(II) complexes, and Ir(III) complexes, which give intense phosphorescence in the visible to near-infrared wavelength regions with reasonably long lifetimes (> 1.0 μs). Oxygen probes targeting various biological tissues of interest have been developed by modifying these O 2 -sensing luminophores: cell-penetrating conjugates of Pt(II)-porphyrins 20,21 , dendritic Pt(II)-and Pd(II)porphyrins [22][23][24] , Ru(II) complex derivatives targeting the cell nucleus 25 , ratiometric O 2 probes based on Ir(III) complexes 26,27 , etc. With the use of phosphorescent probes, wide-field luminescence lifetime measurements using a gated CCD camera have enabled whole-body O 2 imaging including tumor tissues, ocular fundus, and specific organs [28][29][30][31] . In contrast, phosphorescence lifetime imaging microscopy (PLIM) 32,33 allows high-resolution O 2 imaging at the cellular level, providing precise information on microenvironment along with real-time changes in O 2 levels. The PLIM method has been applied to O 2 imaging of giant cells 34 , cell spheroids 20,24,35,36 , neurospheres 37,38 and the epithelium of rat and human colon tissues 38 using cell-permeable small-molecule and nano-particle probes. By using dendrimerized Pt-porphyrin probes, the PLIM method has been successfully applied to microvascular and interstitial O 2 imaging of the brain 22,39,40 , bone marrow 41,42 , and retinal tissue 43 in vivo.
In contrast, few studies have reported tissue oxygen distributions in vivo using an intracellular O 2 probe 44 . We have recently developed small-molecule O 2 probes based on Ir(III) complexes 18,19 . Among them, BTPDM1 ((btp) 2 Ir(acac-DM); Ir-1) (Fig. 1A) allows for the measurement of O 2 levels in hypoxic tumors 45,46 and highresolution O 2 imaging in renal cortex in vivo 47,48 . To improve the reliability of the O 2 level as measured with a small-molecule probe, it is necessary to quantify O 2 levels in different tissues in vivo and investigate various issues: localization and stability of probes in tissues, calibration of emission lifetime in tissues (cell-specific calibration), the influence of probe administration on the target tissues, etc.
In this study, we established an in vivo O 2 imaging method using small-molecule probes by investigating the O 2 distribution in the hepatic lobules of living mice by PLIM measurements with two structurally related Ir(III) complexes, Ir-1 and (btp-OH) 3 Ir (Ir-2) (Fig. 1A). The hepatic lobule is a fundamental structural unit of the liver with a hexagonal shape (Fig. 1B) 49,50 . In the lobule, blood runs unidirectionally from the vertices of the hexagon called portal triads, through microcirculatory networks known as sinusoids, to a central vein (CV) in the middle of the lobule. The portal triad contains two distinct inlets for blood flow: a well-oxygenated hepatic artery and a poorly-oxygenated portal vein, the latter of which constitutes ~ 70% of total liver blood flow. Therefore, even under physiological conditions, liver parenchymal cells and hepatocytes surrounding the portal triad are exposed to relatively low oxygen concentrations, and pericentral hepatocytes experience much lower oxygen concentrations, forming a zonal heterogenous distribution of oxygen along the sinusoids in the hepatic lobule. These intralobular oxygen distributions, however, are considered to be prerequisite for normal liver metabolic functions in a zone-specific manner. In addition, this metabolic zonation is often disrupted in disease conditions with aberrant oxygen distribution in the hepatic lobules. To deepen our knowledge about the dynamic alterations in liver metabolism in both physiological and pathophysiological conditions, real-time imaging of intracellular oxygen tension in situ is a promising method with the development of oxygen-sensitive probes.
We first compared the photophysical properties and oxygen sensitivity of Ir-1 and Ir-2 in solution and cultured cells. Next, we investigated the feasibility of these probes for the measurements of O 2 distribution in cell spheroids. Based on these results, we performed PLIM measurements of hepatic tissues of mice using Ir-1 and Ir-2 to reveal the probe performances.  (Table 1). Using the solubility of oxygen in MeCN at an oxygen partial pressure (pO 2 ) of 0.21 atm 51 , the quenching rate constant can be converted from pressure units to concentration units. The converted k q values (6.38 × 10 9 M −1 s −1 for Ir-1 and 6.91 × 10 9 M −1 s −1 for Ir-2) were close to the diffusioncontrolled rate of bimolecular reactions in MeCN 51 , indicating that oxygen has very high quenching ability to the excited triplet state of both complexes. We recently revealed that the phosphorescence quenching of Ir(III) complexes by molecular oxygen occurs not only by energy transfer but also by charge transfer from the triplet Ir(III) complex to O 2 52 . As a result, Ir(III) complexes show much larger k q values compared with metalloporphyrins and Ru(II) complexes. The O 2 sensitivity is slightly higher in Ir-2 than Ir-1 owing to the longer intrinsic lifetime and larger k q value of Ir-2. The intrinsic lifetimes (5-10 μs) of Ir-1 and Ir-2 are shorter than those (10-100 μs) of Pt-porphyrins, but longer than those of typical Ru(II) complexes used as O 2 probes 18 , and thus the O 2 sensitivity of Ir-1 and Ir-2 is considered to be comparable with Pt-porphyrins and higher than Ru(II) complexes.

Characterization of intracellular O 2 probes.
To evaluate the cellular uptake efficiencies of Ir-1 and Ir-2, we compared the emission intensities of human colorectal adenocarcinoma (HT-29) cells incubated with these complexes (5 μM, incubation time: 2 h) using a plate reader. Here, we used their prototype complex BTP ((btp) 2 Ir(acac); btp = benzothienylpyridine, acac = acetylacetone) 45 as a reference compound. The emission intensities of each complex under 21% and 2.5% O 2 conditions were corrected for the number of HT-29 cells in each well and the ε values (4700, 5400, and 2800 M −1 cm −1 for BTP, Ir-1 and Ir-2) at 488 nm. The relative emission intensities after this correction (Fig. S1) suggest that the cellular uptake efficiency was greatly improved with Ir-1 and Ir-2 compared with BTP by introducing a hydrophilic dimethylamino and hydroxyl group, respectively.
The subcellular localization of Ir-1 and Ir-2 investigated by costaining experiments with organelle-specific trackers (LysoTracker Green and ERTracker Green) using HeLa cells showed preferential accumulation to lysosomes for Ir-1 and endoplasmic reticulum (ER) for Ir-2 (Fig. S2). Both complexes taken up into cells are presumed to accumulate in the organelle membrane because of their lipophilicity. The cytotoxicity of Ir-1 and Ir-2 evaluated with WST assay (incubation time: 24 h under the presence of the probe) revealed that the median lethal doses (LC 50 ) of Ir-1 and Ir-2 for HT-29 cells were 10-15 μM and ~ 30 μM (Fig. S3A). In addition, analysis of mitochondrial membrane potential using JC-1 dye (Fig. S3B) showed that the membrane potential was almost unaffected by Ir-1 or Ir-2 loading (1 μM, 2 h). In typical experiments in this study, the cells were incubated with Table 1. Photophysical parameters of Ir-1 and Ir-2 in MeCN at room temperature. a . τ 0 p and 0 p denote the phosphorescence lifetime and quantum yield taken in degassed solutions, and τ p and p denote the phosphorescence lifetime and quantum yield taken in aerated solutions. Probe max abs (nm) max phos (nm) τ 0 p (µs) τ p (ns) τ 0 p /τ p φ 0 p φ p k q (10 4 mmHg −1 s −1 ) www.nature.com/scientificreports/ the probes (500 nM or 1 μM) for 2 h prior to microscopic measurements. Therefore, under these experimental conditions, the effect of the probes on cell activity is considered to be sufficiently small. Photostability is another important characteristic required for an intracellular O 2 probe. We investigated the photostability of Ir-1 and Ir-2 in cells by taking emission images of HT-29 cells under continuous 488 nm laser pulse irradiation with our PLIM system. The cells were incubated with Ir-1 or Ir-2 (1 μM) for 2 h under 21% or 2.5% O 2 conditions, and PLIM images of the HT-29 cells were taken at every 50 scans until 550 scans (Fig. S4). During this irradiation time, the phosphorescence lifetimes of Ir-1 and Ir-2 were almost constant, although the phosphorescence intensity changed slightly by 488 nm light irradiation. Also, no significant change was observed in the cell morphology. In this study, the signals accumulated with 50 scans were usually averaged to obtain a single PLIM image. Therefore, Ir-1 and Ir-2 were found to have sufficient photostability for obtaining clear PLIM images.

Quantification of O 2 levels.
To quantify the oxygen levels in tissues based on lifetime measurements, we need a calibration curve that represents the relationship between lifetime and pO 2 . Since Ir-1 and Ir-2 are likely to accumulate in specific organelle membranes, the pO 2 dependence of the phosphorescence lifetime in cells is expected to be different from that in solution. Therefore, we calibrated the lifetime by measuring PLIM images of HT-29 cells incubated with Ir-1 or Ir-2 under 21, 15, 10, 5, and 0% O 2 conditions at 37 °C ( Fig. 2A,B); we added 10 µM antimycin A (AntA) to the medium prior to the PLIM measurements to suppress the oxygen consumption by cellular respiration, and in the experiments under N 2 saturated conditions we added Na 2 SO 3 (500 mM) into the medium to remove oxygen remaining in the culture. The average lifetime of an entire image was plotted against the pO 2 according to the Stern-Volmer equation, τ p 0 /τ p = 1 + k q τ p 0 pO 2 . A linear relationship was obtained for both complexes (Fig. 2C), and the k q values for Ir-1 and Ir-2 were derived to be 4.22 × 10 3 mmHg −1 s −1 and 5.36 × 10 3 mmHg −1 s −1 along with the τ p 0 , 5.20 µs and 6.18 µs, respectively. These values were used to quantify oxygen levels in cell spheroids from phosphorescence lifetimes (τ p ).

Evaluation of O 2 distribution in cell spheroids.
In the past decade, three-dimensional (3D) spheroid systems have received much attention in fields such as drug discovery, cancer research, and toxicology. They provide a more physiologically-relevant environment and organ-like microarchitecture compared with conventional 2D cell cultures and better mimic the crucial tumor tissue properties and microenvironment. The excellent cell-permeable properties of Ir-1 and Ir-2 suggested their potential for in-depth staining of 3D cell spheroids and thus the potential of these probes for visualization of the O 2 distribution within spheroids and living tissues.
So, we first acquired the PLIM images of HT-29 cell spheroids that were incubated with Ir-1 or Ir-2. Here, each probe (500 nM) was added to the medium after the spheroids were formed, and the spheroids were further incubated with the probe for 24 h prior to PLIM measurements. The bright-field images taken on different planes in the z direction (Fig. 3) show that spheroids with a diameter of 150-200 μm were almost uniformly stained, including the core, by Ir-1 and Ir-2. We found from the Z-stack phosphorescence lifetime images of an HT-29 spheroid stained with Ir-1 (Fig. 3B) and Ir-2 (Fig. 3C) that the cells closer to the center of the spheroid gave longer lifetimes, i.e. lower O 2 levels compared with those in the peripheral region. For quantitative evaluation of the O 2 gradient in a spheroid, we investigated the line profiles of the average phosphorescence lifetime on the plane approximately 20 μm from the bottom (Fig. 3D). We also measured the lifetimes along the central area in the z-direction (Fig. 3E), and derived the pO 2 distribution (Fig. 3D,E) based on the τ p 0 and k q obtained for HT-29 cultured cells. Both probes showed that the pO 2 is reduced near the center and bottom surface of the spheroids, because the oxygen supply from the bottom direction is cut off by the bottom glass, and also the peripheral cells consume oxygen carried by diffusion from the culture medium. The degree of hypoxia in the core depended on the spheroid size (Fig. S5) and the respiratory activity as will be described below.
We next investigated whether Ir-1 and Ir-2 can image changes in the oxygen status of spheroids caused by metabolic stimulus. We used FCCP (carbonyl cyanide p-(trifluoromethoxy)phenylhydrazone) and AntA to alter the oxygen status of HT-29 spheroids by stimulating metabolic processes; FCCP is an uncoupler of oxidative phosphorylation in mitochondria that disrupts ATP synthesis by transporting protons across the membrane and thus increases the O 2 consumption rate. AntA is known to inhibit the mitochondrial electron transport chain from cytochrome b to cytochrome c1 and suppress O 2 consumption. Metabolic stimulation of HT-29 cell spheroids by FCCP and AntA significantly changed the PLIM images (Fig. 4). The lifetime images of the spheroid plane ~ 30 μm from the bottom (Fig. 4B,C) demonstrated that upon addition of FCCP to the medium, the lifetimes of both probes quickly increased within ~ 10 min, especially in the core. The oxygen partial pressures calculated from the lifetimes based on the calibration in Fig. 2 indicate that the pO 2 decreased sharply to less than 20 mmHg by increased O 2 consumption, and then it slowly recovered. Conversely, the lifetimes of both probes rapidly dropped within 10 min upon metabolic stimulation using AntA (Fig. 4D,E), corresponding to the increase in pO 2 by suppression of O 2 consumption. The pO 2 changes of spheroids reflected the metabolic effects of FCCP and AntA, demonstrating that Ir-1 and Ir-2 can be used to track the oxygen status of cell spheroids. These results confirmed the high penetrating ability of Ir-1 and Ir-2 into spheroids and their feasibility as tissue O 2 probes.
In vivo O 2 imaging of hepatic tissues. Since we confirmed that Ir-1 and Ir-2 can be efficiently taken up into spheroids and evaluate the oxygen status, we next attempted to visualize the oxygen gradient of hepatic lobules by in vivo PLIM measurements with Ir-1 or Ir-2 as a probe. Each probe (250 nmol in 25 μL dimethyl sulfoxide, diluted using PBS to 175 μL) was administered intravenously to anesthetized mice. Approximately 10 min after probe administration, the abdomen was opened to expose the liver, and the phosphorescence lifetime images of hepatic tissues (~ 10 μm from the surface of the liver) were measured around the central vein (CV) using different magnifications. Both probes gave clear PLIM images of hepatic tissues with cellular-level www.nature.com/scientificreports/ www.nature.com/scientificreports/ resolution (Fig. 5). Here, the areas that appear black are the regions where the emission intensity from the probe is extremely low, and these mainly correspond to the nucleus and sinusoid that have low probe uptake. It can be seen from the PLIM images measured with a 40 × objective lens that Ir-1 and Ir-2 are internalized into hepatocytes that line up along sinusoids. The arrangement of hepatocytes and lifetime distribution indicate that these images correspond to a hexagonal hepatic lobule (Fig. 1B) and that CV exists in the area with the longest lifetime (displayed in orange). The region surrounding the CV is considered to be the portal vein (PV) because it has a much shorter lifetime, i.e. much higher oxygen level. The direction of blood flow seen in the video image of hepatic tissues (Video S1) also indicates that the phosphorescence lifetime in the vicinity of CV is longer than that in the surrounding PV. PLIM images obtained with Ir-1 and Ir-2 show a similar lifetime gradient that increases from the PV to CV. To quantify the O 2 levels in specific locations in the lobules from the phosphorescence lifetimes, we performed calibration of the phosphorescence lifetimes using an AML 12 (alpha mouse liver 12) cell line that was established from mouse hepatocytes (Fig. S6). Using the τ p 0 and k q of Ir-1 and Ir-2 as determined in a monolayer of AML 12 cells, we evaluated the average pO 2 of the regions of interest (ROIs) that can be attributed to CV, PV, and the intermediate region (IR); the phosphorescence lifetime of Ir-1 gave reasonable O 2 levels (24 ± 6.1 mmHg for CV, 32 ± 4.9 mmHg for IR, and 39 ± 4.2 mmHg for PV), whereas Ir-2 exhibited much lower O 2 levels (3 ± 1.8 mmHg for CV, 5 ± 1.8 mmHg for IR, and 7 ± 2.7 mmHg for PV) ( Table 2). These results indicated that PLIM images obtained with Ir-1 and Ir-2 clearly visualized the oxygen gradient from PV to CV. However, the O 2 levels derived from the phosphorescence lifetimes differed between Ir-1 and Ir-2.
To confirm that Ir-1 and Ir-2 retain their spectral properties in hepatic tissues after systemic administration, we measured the emission spectra of the livers of living mice. Each probe molecule (250 nmol) was administered through the tail vein, and the emission images and spectra of livers were taken using a fluorescence microscope with an excitation filter of 450-500 nm and emission filter of > 532 nm (Fig. S7). The observed emission spectra with maxima at 619 nm for Ir-1 and 607 nm for Ir-2 were slightly red-shifted compared with those in MeCN, www.nature.com/scientificreports/ but they were in good agreement with those in AML 12 cells, demonstrating that these probes maintain their spectral properties in hepatic tissue. The reason why Ir-2 gave an abnormally low O 2 level seems to be related to the function of the liver to metabolize xenobiotics. So, to clarify how clearance of harmful metabolic byproducts and detoxification of xenobiotics affect hepatic O 2 levels, we investigated the detoxification of ammonia in the liver, which has the function of converting toxic ammonia to urea by the urea cycle or to glutamine by glutamine synthesis in hepatocytes. With increased metabolic activity of the liver, oxygen consumption should be enhanced to produce ATP. We intravenously administered NH 4 Cl (0.13 g/kg) to anesthetized mice at ~ 30 min after 250 nmol Ir-1 injection into the tail vein. PLIM images of hepatic lobules measured 10 min after the administration of NH 4 Cl exhibited a marked increase in the phosphorescence lifetime of Ir-1 (Fig. 6A,B). The average phosphorescence lifetimes To clarify the difference in in vivo probe performance between Ir-1 and Ir-2, we performed O 2 imaging experiments on liver tissues using another Ir(III) complex, (btp) 2 Ir(acac-2OH) (Ir-3), which has a similar structure to Ir-1 (Fig. S8A). The τ 0 p and k q of Ir-3 in AML 12 cells were determined to be 4.80 µs and 4.82 × 10 3 mmHg −1 s −1 . The PLIM image (Fig. S8B) in the hepatic lobule of an Ir-3-administered mouse exhibited a clear lifetime gradient and gave an average pO 2 of 16 ± 3.2 mmHg in CV, 21 ± 3.4 mmHg in IR, and 24 ± 3.8 mmHg in PV (Fig. S8C). The hepatic O 2 levels obtained with Ir-3 were close to those obtained with Ir-1, although Ir-3 gave somewhat smaller pO 2 values. www.nature.com/scientificreports/ In contrast to the results in the liver tissue, Ir-1 and Ir-2 gave similar O 2 tensions in the kidney (Fig. S9). Intravenously administered probes rapidly migrated from the vasculature into the proximal tubule cells and were hardly excreted in the urinary space during the observation period. The average O 2 tension of tubular cells in each image was derived using calibration lines obtained for human kidney 2 (HK-2) cells to be 45 ± 9.8 mmHg for Ir-1 and 36 ± 4.3 mmHg for Ir-2 (Fig. S10).

Discussion
We first verified that Ir-1 and Ir-2 exhibit desired photophysical properties as O 2 probes in cells and cell spheroids. Although the τ p 0 values (5-10 μs) of Ir-1 and Ir-2 are much smaller than those of Pt(II)-and Pd(II)-porphyrins, their oxygen sensitivity is sufficiently high due to their large k q values. In addition, the moderately long lifetimes of the Ir(III) complexes have the advantage of increasing the counting efficiency in PLIM measurements and thus shortening the image acquisition time.
One key issue in applying small-molecule intracellular oxygen probes for in vivo O 2 measurements is the method of lifetime calibration. We quantified the tissue pO 2 based on the k q and τ p 0 determined using a cell line that is the same as or close to the cell type in the target organ in which the probe is accumulated. The O 2 quenching rate constants in HT-29 and AML 12 cells were reduced by an order of magnitude compared with those in solution (see Table 1). Since lipophilic Ir-1 and Ir-2 are likely to accumulate in organelle membranes after they pass through the plasma membranes of living cells, the much smaller k q values in cells can be attributed, at least in part, to the decrease in the diffusion rates of O 2 and Ir(III) complexes in organelle membranes. Further reductions in k q values can be caused by the binding of excited probe molecules to proteins in the organelle membrane in which they are incorporated. The τ p 0 values of Ir-1 and Ir-2 in cells were decreased by 10-15% compared with those in MeCN. One possible reason lies in the concentration quenching in organelle membranes. The τ p 0 of Ir-1 partitioned into DMPC liposomes tended to shorten with increasing probe concentration in solution (Table S1), suggesting that concentration quenching may have occurred. Cross-sensitivity to endogenous species other than O 2 cannot be ruled out, but its contribution appears to be relatively small because the τ p 0 values of Ir-1 and Ir-2 in cells are close to those in MeCN. It was also confirmed that phosphorescence lifetime is almost independent of pH and the presence of glutathione under physiological conditions (Tables S2 and S3).
One of the major advantages of Ir-1 and Ir-2 is their extremely high cellular uptake efficiencies. The O 2 imaging experiments using HT-29 cell spheroids (Fig. 3) demonstrated that Ir-1 and Ir-2 are efficiently internalized into the spheroids with a diameter of ~ 200 μm. The penetrative abilities of some molecular O 2 probes into cell spheroids reported so far are compared in Table S4. These probes include glucose conjugates of Pt(II)-mesotetrakis-(pentafluorophenyl)porphyrin 20 , Pt(II)-5, 10, 15, 20-tetrakis-(4-carboxyphenyl)porphyrin 36 , a Pt(II) complex bearing a cyclometalating 3-di(2-pyridyl)benzene-based moiety 35 , and click-assembled oxygen-sensing nanoconjugates with Pd(II) tetracarboxytetrabenzoporphyrin as a phosphorescent core 24 . Comparing each probe for its concentration in culture medium and the incubation time required for observation of clear luminescence images, it can be seen that Ir-1 and Ir-2 have excellent cell permeability. Furthermore, PLIM measurements with these probes made it possible to track oxygen levels and distributions in spheroids. These properties are very useful for investigating the oxygen status in tissues.
The oxygen gradient in the hepatic lobule has been demonstrated by applying the hypoxia marker 2-nitroimidazole to liver tissue sections 53 . High-resolution visualization of hepatic oxygen distribution in vivo has been performed by Paxian et al. 44 by intravital emission microscopy using [Ru(phen) 3 ] 2+ (Tris(1,10-phenanthroline) ruthenium(II)) as an O 2 probe. They observed a continuous increase in emission intensity of [Ru(phen) 3 ] 2+ from periportal to pericentral regions in rat liver, implying an O 2 gradient within the liver tissue. However, the emission intensity depends on the probe distribution in the tissue, so that intensity-based measurements cannot assess the precise pO 2 and O 2 gradient. As for hepatic tissue oxygen levels, Kietzmann and Jungermann 54,55 have reported the pO 2 to be 60-65 mmHg in the periportal blood and 30-35 mmHg in the perivenous blood, and they suggested the pO 2 level to be approximately 15 mmHg lower in periportal and perivenous cells. Tsukada and Suematsu 56 have investigated the average pO 2 in hepatic microcirculation by phosphorescence lifetime measurements of BSA (bovine serum albumin)-bound Pd-TCPP (Pd(II)-meso-tetra(4-carboxyphenyl)porphine) in hepatic tissues of mice. They obtained an average pO 2 of 59.8 mmHg in portal vessels and 38.9 mmHg in central venules. Our PLIM measurements with Ir-1 showed that the average pO 2 of hepatocytes near PV and CV were 39 ± 4.2 mmHg and 24 ± 6.1 mmHg, respectively. Although the hepatic oxygen levels obtained in our study are somewhat lower than those reported for intravascular pO 2 , these are considered to be reasonable values given the oxygen levels to which hepatocytes are exposed. To the best of our knowledge, the PLIM images shown in Fig. 5 are the first to visualize the oxygen concentration gradient in hepatic lobules in vivo using an intracellular O 2 probe. www.nature.com/scientificreports/ www.nature.com/scientificreports/ Imaging the spatiotemporal pO 2 changes within the liver microarchitecture provides useful information about the oxygen response and dynamics of the liver at the cellular level during physiological stimulation. Our PLIM measurements with Ir-1 showed that intravenous administration of NH 4 Cl resulted in a rapid decrease in hepatic tissue O 2 levels, and its recovery rate depended on the position in the lobule (Fig. 6). Hepatocytes are known to display considerable functional differences depending on their position along the porto-central axis of the liver lobule, and ammonia detoxification in the liver occurs by two pathways: consumption by urea synthesis in the periportal area and removal by glutamine synthesis in perivenous area. Since these processes require O 2 consumption, ammonia stimulation may produce hypoxic regions in the hepatic lobules depending on the O 2 consumption rate of hepatocytes.
In HT-29 cell spheroids, Ir-1 and Ir-2 showed almost the same O 2 levels and distribution, whereas in hepatic tissues of living mice, Ir-2 gave much lower oxygen levels: 7 ± 2.7 mmHg for the periportal region and 3 ± 1.8 mmHg for the perivenous region (Table 2). In kidneys, on the other hand, both probes accumulated in tubular cells and showed similar oxygen levels: 45 ± 9.8 mmHg for Ir-1 and 36 ± 4.3 mmHg for Ir-2 (Fig. S9). These results suggest that the administration of Ir-2 to mice causes hypoxia in hepatic tissues due to some specific metabolic processes in the liver. Since spectral measurements (Fig. S7) revealed that the luminescent properties of Ir-1 and Ir-2 were maintained in liver tissue, it is likely that Ir-2 is excreted in the bile through solubilization processes that do not affect the luminescent properties such as glucuronidation at the OH group. The luminescence spectrum of the extract from bile 6 h after probe administration showed that Ir-1 was excreted through the bile duct along with bile (Fig. S11). Considering that Ir-3, which has the same acac moiety as Ir-1, gave similar liver tissue O 2 levels as Ir-1, the lipophilic Tris-ligand structure of Ir-2 may be related to its metabolism in the liver.

Conclusions
Small-molecule O 2 probes, Ir-1 and Ir-2, were efficiently taken up into cells when added to the culture media of monolayer cells and cell spheroids, and this enabled reversible pO 2 measurements in the cells. PLIM measurements of mouse livers following intravenous administration of Ir-1 or Ir-2 allowed high-resolution O 2 imaging of hepatic tissues, exhibiting an O 2 gradient from the pericentral to periportal regions in hepatic lobules. The phosphorescence lifetime of Ir-1 gave reasonable hepatic O 2 levels after calibrating the lifetime using cultured AML 12 cells. Furthermore, Ir-1 allowed visualization of the pO 2 changes in hepatic tissues stimulated by ammonia. However, Ir-2 gave a much lower pO 2 compared with Ir-1, which may be due to the toxicity of Ir-2, which promoted detoxification in the liver. These results reveal that Ir(III) complexes allow imaging of spatiotemporal changes in oxygen levels within the tissue microarchitecture in vivo, but some complexes may influence oxygen consumption in the liver when used for oxygen imaging of hepatic tissues.

Methods
Materials. The Ir(III) complexes used in this study were synthesized and characterized according to the methods described in the Supplementary Information. Acetonitrile (MeCN; Kanto Chemical, spectroscopic grade) and NH 4 Cl (Kanto Chemical, special grade) were used as received. Antimycin A from Streptomyces sp. (Ant A) and carbonyl cyanide 4-(trifluoromethoxy) phenylhydrozone (FCCP, 98%) were purchased from Sigma-Aldrich. Cell Counting Kit-8 (CCK-8) was purchased from Dojindo Laboratories. 1 H-NMR spectra were recorded with a JEOL JNM-ECS400 (400 MHz) spectrometer in DMSO-d 6 . 1 H-NMR chemical shifts were referenced to tetramethylsilane. The apparent resonance multiplicity was described as s (singlet), d (doublet), dd (double doublet), t (triplet), and m (multiplet). ESI-MS spectra were recorded on an Applied Biosystems API 2000 mass spectrometer.
Photophysical properties in solution. Absorption spectra were recorded on a UV/Vis spectrophotometer (Ubest-V550, JASCO). Emission spectra were measured with a system consisting of a monochromatized Xe arc lamp, a sample holder, and a photonic multichannel analyzer (C10027-01, Hamamatsu Photonics), and were fully corrected for spectral sensitivity between 200-950 nm. Phosphorescence lifetimes in solution were measured with a lifetime measurement system (Quantaurus-Tau C11367G, Hamamatsu Photonics) based on the time-correlated single photon counting (TCSPC) method. Phosphorescence quantum yield was measured with an absolute photoluminescence quantum yield measurement system (C9920-02, Hamamatsu Photonics) that consisted of a Xe arc lamp, a monochromator, an integrating sphere, a multichannel detector, and a personal computer 57 . Cell and cell spheroid cultures. Human  In vivo PLIM measurements. All protocols for animal experiments were approved by the Ethical Committee on Animal Experiments of Gunma University , and all animal experiments were conducted in accordance with the institutional guidelines. In vivo PLIM measurements were carried out using six-to eight-week-old Balb/c male mice (CLEA Japan). After general anesthesia, probe solution (200 µL) containing 250 nmol Ir(III) complex in 10% DMSO/saline was injected into the tail vein, and the liver or kidney was exposed for ~ 10 min before PLIM experiments.
Calibration of the phosphorescence lifetime. The oxygen partial pressure in cells was obtained from the phosphorescence lifetime based on a calibration using cultured cells. HT-29, AML12, or HK-2 cells were seeded on glass-based dishes and stained with Ir-1 or Ir-2 (500 nM, 2 h). PLIM images were acquired for the stained cells at 21%, 15%, 10%, 5%, and 0% O 2 in an incubator at 37 °C. Here, in the 5-21% O 2 experiments, cells were incubated with Ant A (10 µM, > 0.5 h) to block cellular respiration, and in the 0% O 2 experiments, fresh medium containing Na 2 SO 3 (500 mM) was used to remove existing O 2 in the medium. The average phosphorescence lifetimes of the PLIM images taken under different pO 2 were plotted according to the Stern-Volmer equation, τ p 0 /τ p = 1 + k q τ p 0 pO 2 , to determine the bimolecular quenching rate constant, k q . The pO 2 of cells, spheroids, and tissues was obtained from the phosphorescence lifetime (τ p ) using the k q and τ p 0 . Subcellular localization of probes. HeLa cells were cultured in DMEM (Gibco) with 10% FBS, penicillin (50 units/mL), and streptomycin (50 μg/mL), and were grown under a 5% CO 2 atmosphere at 37 °C. The cells were seeded into glass bottom imaging dishes (Greiner) and allowed to adhere for 24 h, incubated with Ir-1 (500 nM) or Ir-2 (1 μM) for 2 h, washed 3 times with DMEM, and then the medium was replaced with DMEM (FluoroBrite, Gibco) without FBS. Luminescence microscopic images were obtained with an inverted microscope (IX71, Olympus) equipped with an electron multiplying CCD camera (Evolve 512, PHOTOMETRICS) driven by MetaMorph software.
Evaluation of cytotoxicity using CCK-8 assay. HT-29 cells (2.5 × 10 4 cells/well) were seeded into a 96-well flat bottom plates (Greiner) for 48 h. The cells were incubated with various concentrations of each probe for 24 h at 37 °C under a 5% CO 2 atmosphere. The medium was removed, and the cells were washed gently with McCoy's 5A medium without phenol red. Cell Counting Kit-8 reagent (CCK-8, Dojindo) was added to each well, and incubation was continued for 1 h 58 . The absorbance at 450 nm of each well referenced at 650 nm was recorded using a microplate reader (Infinite 200 PRO, Tecan). Cell viability (% of control) was evaluated as (A sample − A blank )/(A control − A blank ) × 100, where A sample is the absorbance of cells exposed to the probe, A control is the absorbance of cells without probe, and A blank is the absorbance of wells without cells.