Photophysical properties of BTP-PEGm in solution and medium of cultured cellsWe first investigated the photophysical properties of BTP-PEGm in saline. All of the PEGylated derivatives of BTP showed high solubility in aqueous media and exhibited similar absorption and phosphorescence spectra in saline (Fig. S1). They also exhibited relatively large molar absorption coefficients of (8.6–9.2) × 103 dm3 mol−1 cm−1 at 488 nm and phosphorescence spectra with a maximum at 644 nm and sufficiently high quantum yield of ~ 0.17 under degassed conditions (Table 1). The phosphorescence decay profiles followed a single exponential decay pattern with phosphorescence lifetimes varying from ~ 4.0 μs to ~ 0.9 μs in degassed and aerated saline (Fig. S2, Table 1). These results demonstrate that BTP-PEGm exhibits excellent photophysical properties as an O2 probe in aqueous medium.Table 1 Photophysical properties of BTP-PEGm in saline at 37 °C.As intravascular O2 probes must not interact with proteins such as albumin or be taken up by cells when administered into the blood, we next investigated the photophysical properties of BTP-PEGm in fetal bovine serum (FBS). The phosphorescence spectra of BTP-PEGm in FBS (Fig. S3) were almost identical to those in saline, except that in FBS, autofluorescence was superimposed around 500–600 nm when excited at 488 nm. The phosphorescence decays could be fitted with a single-exponential function, although the residuals were slightly distorted for BTP-PEG2000 (Fig. S4), probably due to the effect of albumin present in the FBS. These results revealed that interaction between the phosphorescent group BTP and albumin molecules was sufficiently suppressed in BTP-PEGm complexes other than BTP-PEG2000.The cell permeability of BTP-PEGm was examined by adding the complex to medium of cultured HCT-116 cells and then determining the emission intensity and acquiring PLIM images. The results for BTP-PEG10000 are shown in Fig. 2. Autofluorescence images for each cell (Fig. 2a, left) revealed the position of that cell in the medium. The phosphorescence intensity images showed that the luminescence of BTP-PEG10000 was almost exclusively limited to the culture medium, indicating that very little probe had entered the cells. PLIM images taken under culture conditions of 21% and 2.5% O2 and distribution histograms of probe lifetime demonstrated that the phosphorescence lifetime of BTP-PEG10000 depends significantly on the concentration of O2 in the medium. The other derivatives, BTP-PEG5000 and BTP-PEG20000, gave results similar to BTP-PEG10000, whereas BTP-PEG2000 also exhibited emission from inside the cells (Figs. S5–S7). These data indicate that some proportion of BTP-PEG2000 passes through the cell membrane and remains within the cell.Figure 2(a) Fluorescence and phosphorescence intensity images and PLIM images of HCT-116 cells incubated in culture medium containing BTP-PEG10000 (10 μM) for 30 min at 37 °C under 21% and 2.5% O2 conditions. Excitation wavelength: 488 nm; collected emission wavelength range for fluorescence intensity images: 510–560 nm; and collected emission wavelength range for phosphorescence intensity and PLIM images: > 620 nm. Scale bar: 50 μm. (b) Distribution histograms of phosphorescence lifetimes for the PLIM images shown in (a).Characterization of BTP-PEGm as an intravascular O2 probeThe results of experiments using cultured cells revealed that the various BTP-PEGm probes, other than BTP-PEG2000, were not taken up into cells but rather remained in the medium and exhibited oxygen-sensitive phosphorescence. Application of BTP-PEGm as an intravascular oxygen probe requires that the complex remains in the bloodstream long enough and retains its photophysical properties. Therefore, we investigated the behavior of BTP-PEGm in renal tissues by injecting the probe into the tail vein of an anesthetized mouse. A schematic illustration of the renal surface is depicted in Fig. S8, together with an illustration of the typical renal surface of a mouse administered BTPDM1 and FITC-lectin, as reported in our previous paper32. In the kidney, the glomeruli first filter waste products from the blood and produce primary urine, and then the necessary components in the primary urine are reabsorbed in the proximal tubules (Fig. S8a). Areas of intense fluorescence (Fig. S8b), due primarily to FITC-lectin, indicate the locations of peritubular capillaries, and areas of intense phosphorescence (Fig. S8c) indicate BTPDM1 taken up by tubular cells. These images provide a clear visualization of the microarchitecture of the renal surface tissue, including tubular cells, peritubular capillaries, and the urinary space.Figure 3 shows images of phosphorescence intensity in renal tissues (ca. 10 μm from the surface) of mice administered 100 nmol of BTP-PEGm (100 μL saline solution). By comparing the images in Fig. 3a with those of Fig. S8, we find that the retention of BTP-PEGm in the bloodstream varies greatly depending on the length of the PEG chains. Images of phosphorescence intensity taken 10 min and 30 min after administration showed the presence of BTP-PEG2000 not only in peritubular capillaries but also in tubular cells and the urinary space. This indicates that intravenously administered BTP-PEG2000 is filtered by the glomeruli within approximately 10 min and transferred from the blood to the primary urine. A similar trend was observed with BTP-PEG5000, although the migration rate was much slower than that of BTP-PEG2000. By contrast, Fig. 3b demonstrates that BTP-PEG10000 and BTP-PEG20000 are retained in the blood without being filtered by the glomeruli, even 4 h after administration. A possible reason why BTP-PEG10000 and BTP-PEG20000 are not filtered in the glomerulus lies in their high molecular weight (MW) and molecular size. The threshold MW for glomerular filtration is estimated at approximately 30–50 kDa, depending on molecule shape, molecular structure, and flexibility37,38. For example, albumin and hemoglobin, which both have a molecular weight of ~ 67 kDa, are not filtered by the glomerulus. The reason why BTP-PEG20000 does not pass through the glomerular capillary wall can be explained by its average molecular weight (~ 63 kDa), similar to that of albumin and hemoglobin. Although the MW of BTP-PEG10000 (~ 32 kDa) is near the threshold for filtration, the glomerular barrier is also known to be highly size selective39. Dynamic light scattering measurements indicated that the hydrodynamic radius of albumin is 3.37 nm40. As the calculated Stokes radius of one PEG10000 unit is ~ 2.9 nm41, the estimated radius of the entire BTP-PEG10000 molecule would be greater than that of albumin and also that (~ 4 nm) of the glomerular pores37. We also acquired phosphorescence intensity images of the surface of hepatic tissue of mice administered 100 nmol of BTP-PEG10000 or BTP-PEG20000 (100 μL saline solution). As shown in Fig. S9, the sinusoid network could be clearly visualized 30 min and 4 h after administration, indicating that neither BTP-PEG10000 nor BTP-PEG20000 was taken up by hepatic cells.Figure 3(a) Emission intensity images of the surface of the kidney of a mouse at 10 and 30 min after administration of BTP-PEG2000 (100 nmol). (b) Emission intensity images of the surface of the kidneys of mice at 30 min and 4 h after administration of BTP-PEG5000, 10000, or 20000 (100 nmol). Excitation wavelength: 488 nm; collected emission wavelength range: > 620 nm. Scale bars: 50 μm.These results indicate that BTP-PEG10000 and BTP-PEG20000 exhibit excellent properties as intravascular O2 probes. However, considering the ease of administration to mice, BTP-PEG10000 is preferable because it has a lower molecular weight and lower viscosity when dissolved in saline. Therefore, we used BTP-PEG10000 as an O2 probe in subsequent experiments.Before proceeding to quantitative measurements, however, we investigated the photostability and the O2 response of BTP-PEG10000 in peritubular capillaries. The photostability of BTP-PEG10000 in peritubular capillaries was evaluated by acquiring 10 PLIM images of the surface of the kidney in the same field of view. The decrease in relative emission intensity was < 5% (Fig. S10), indicating that BTP-PEG10000 has sufficient photostability for successive PLIM measurements of the same field of view. We then examined the O2 response of BTP-PEG10000 in peritubular capillaries by changing the inspired O2 concentration of mice, which was adjusted using O2/N2 mixed gas. First, we acquired PLIM images of kidney tissue at a normal O2 concentration (21%), and after 2.5 min, the O2 concentration was reduced to 15% and PLIM images acquired. After 7.5 min, the O2 concentration was changed back to 21%, and PLIM images were again acquired. Figure 4a,b show the PLIM images of renal tissues and the average probe lifetimes, respectively, along with the time of measurement. Histograms showing probe lifetime for each image are shown in Fig. 4c. These data demonstrate that the phosphorescence lifetime of BTP-PEG10000 reversibly reflects the subtle changes in O2 concentration that occur in peritubular capillaries.Figure 4(a) PLIM images of the surface of the kidney of a mouse under different inspired pO2 conditions: 0–2.5 min (21% O2), 2.5–7.5 min (15% O2), > 7.5 min (return to 21% O2). BTP-PEG10000 was administered prior to the PLIM measurements. Excitation wavelength: 488 nm; collected emission wavelength range: > 620 nm. Scale bar: 50 μm. (b) Changes in average phosphorescence lifetimes for the PLIM images shown in (a). (c) Distribution histograms of phosphorescence lifetimes for the PLIM images shown in (a).The biocompatibility of BTP-PEG10000 is an important factor for potential in vivo application. To examine biocompatibility, saline (100 μL) or a stock saline solution of BTP-PEG10000 (100 μL) was injected into the tail vein of mice. No probe-induced abnormal behavior or weight loss was observed during the observation period (7 days) in any of the mice (Fig. S11), suggesting that BTP-PEG10000 is not acutely toxic in mice.Intravascular O2 imaging in tumor tissues using BTP-PEG10000As it was confirmed that the phosphorescence lifetime of BTP-PEG10000 is sensitive to changes in the O2 concentration in the renal capillaries, we quantitatively investigated the relationship between the phosphorescence lifetime of BTP-PEG10000 and pO2 using mouse plasma. The lifetime (\({\tau }_{\text{P}}^{0}\)) of BTP-PEG10000 in deoxygenated plasma was almost constant at three different temperatures close to the body temperature of mice under anesthesia (Table S1). Stern–Volmer plots of \({\tau }_{\text{P}}^{0}/{\tau }_{\text{P}}\) versus pO2 for BTP-PEG10000 in blood plasma exhibited straight lines with differing slopes at these temperatures (Fig. S12a). The quenching rate constant (\({k}_{\text{q}}\)) derived from the plot shown in Fig. S12a increased at higher temperatures (Fig. S12b), because the molecular diffusion rate increases with increasing temperature. The pO2 at each pixel in a PLIM image can be determined using the lifetime (\({\tau }_{\text{P}}\)) of BTP-PEG10000 according to the equation pO2 = (1/\({k}_{\text{q}}\))(1/\({\tau }_{\text{P}}\) − 1/\({\tau }_{\text{P}}^{0}\)), as derived from the Stern–Volmer equation. In the experiments discussed below, we used the average \({\tau }_{\text{P}}^{0}\) at three temperatures and the \({k}_{\text{q}}\) value at the body temperature of the mouse during the PLIM measurements.First, we administered BTP-PEG10000 (100 nmol in saline) to tumor-bearing mice under anesthesia, removed a section of skin to expose the tumor, and then acquired PLIM images of the tumor tissue. A typical PLIM image and a partially magnified image of tumor tissue measured using 4 × and 10 × objectives are shown in Fig. 5 along with corresponding intensity images. The phosphorescence of BTP-PEG10000 enabled clear visualization of the tumor microvasculature, with its characteristic chaotic and tortuous morphology. Such heterogeneity in tumor vascular architecture is thought to be closely associated with hypoxia in solid tumors42,43. However, few studies have simultaneously visualized in detail the tumor vasculature and O2 distribution in vivo. Therefore, we calculated the pO2 distribution of tumor vessels from PLIM images. The decay curves calculated for different locations in Fig. 5 followed single-exponential functions and indicated markedly different lifetimes (Fig. S13), thus revealing the heterogeneity of oxygen status within the tumor tissue (Table S2). Figure 5b,c show the lifetime distribution in PLIM images acquired with 4 × and 10 × objectives and the pO2 distributions calculated from the lifetimes, respectively. Figure 5 shows that the blood vessels near the center of the image have a narrow and intricate structure and that the O2 concentrations are much lower compared with the surrounding thicker blood vessels. We acquired similar PLIM measurements at different locations in the tumor (Fig. S14). These images showed that the oxygen level varied depending on the hypoxic heterogeneity in the tumor microenvironment (TME) with a sparse vascular network. In contrast, the pO2 distribution of normal tissue (e.g., kidney) was narrower than that of tumors due to the nearly uniform thickness of the capillaries in normal tissue (Fig. S15).Figure 5(a) Emission intensity and PLIM images of tumor tissue of a mouse administered BTP-PEG10000 (100 nmol). Upper images: 4 × objective lens. Lower images: 10 × objective lens. Excitation wavelength: 488 nm; collected emission wavelength range: > 620 nm. Scale bar: 500 μm (upper images), 200 μm (lower images). (b,c) Distribution histograms of phosphorescence lifetimes and pO2 values for PLIM images shown in (a).Intracellular O2 imaging in tumor tissues using BTPDM1Using hydrophobic BTPDM1 as an intracellular O2 probe, we previously reported that the O2 levels in tumor tissue are considerably lower than those in normal tissue43. Using the same probe, BTPDM1, we also performed high-resolution imaging of tissue oxygenation in the kidney32 and liver33 of living mice. In the previous section, we found that intravascular O2 levels in tumor tissue are inhomogeneous, probably due to the irregular vascular structure of tumor tissues. We therefore investigated the relationship between vasculature and intracellular O2 levels within tumor tissue using BTPDM1 as an O2 probe. As BTPDM1 emits red phosphorescence with a similar spectrum as BTP-PEG10000, we used FITC-dextran (500 kDa), which emits green fluorescence, for vascular imaging. The absorption and phosphorescence spectra and phosphorescence decay of BTPDM1 in aerated and de-aerated acetonitrile are shown in Fig. S16. The phosphorescence spectrum of BTPDM1 in acetonitrile exhibited a vibrational pattern, but the wavelength range was almost the same as that of BTP-PEG10000. The lifetime of BTPDM1 internalized in HCT-116 cells at 37 °C varied with the pO2 in the incubator, as shown in Fig. S17a, and the quenching rate constant was determined for three different temperatures from Stern–Volmer plots (Fig. S17, Table S3).As we were able to determine the relationship between the lifetime of BTPDM1 in HCT-116 cells and the pO2 in the incubator, we next performed PLIM/FLIM analyses of tumor tissues using BTPDM1 as an intracellular probe and FITC-dextran as a vasculature imaging reagent. FITC-dextran (100 μg, 100 μL in saline) was administered to a tumor-bearing mouse approximately 1 h after administration of BTPDM1 (250 nmol, 150 μL in saline containing 10% DMSO), followed by PLIM measurements. A typical PLIM image of a tumor in the mouse administered BTPDM1 is shown in Fig. 6, along with an image of FITC-dextran fluorescence intensity, which reveals the vasculature (Fig. S18). It can be seen from these images that a pO2 gradient exists, such that the intracellular O2 levels decrease as distance from relatively large blood vessels increases. Examining the distance dependence of intracellular O2 levels from the nearest blood vessel within the region of interest indicated in Fig. 6b, we found that the pO2 decreased from approximately 30 mmHg to < 10 mmHg over a distance of approximately 150 μm. This agrees well with the generally expected pO2 gradient with distance from blood vessels13,44.Figure 6(a) PLIM image of tumor tissue of a mouse administered BTPDM1 (250 nmol). Excitation wavelength: 488 nm; collected emission wavelength range: > 590 nm. Scale bar: 100 μm. (b) Average phosphorescence lifetimes (black solid line) and average pO2 values (red solid line) along the white dotted area in the PLIM image.Figure 7 shows FLIM/PLIM images taken using a low-magnification objective to observe the intracellular pO2 distribution over a broader area. The FLIM image depicts a non-uniformly distributed and tortuous vascular network, similar to that shown in Fig. 5. Furthermore, it can be noted from the PLIM image and lifetime distribution that the pO2 distribution is heterogeneous (i.e., some regions exhibit anomalously low pO2). Comparing the FLIM and PLIM images reveals that tumor cells in areas with less-developed vascular networks undergo hypoxia. This hypoxic microenvironment is closely associated with cancer cell proliferation, therapy resistance, and the tumor immune response45,46. In the future, it may be possible to use this method to track the dynamic oxygenation status of chronic and acute hypoxia, which can fluctuate over a period of several minutes.Figure 7(a) Fluorescence intensity and FLIM images and (b) phosphorescence intensity and PLIM images of tumor tissue of a mouse administered FITC-dextran (average molecular weight: 500 kDa; 1 mg/mL, 100 μL) and BTPDM1 (250 nmol). Excitation wavelength: 488 nm; collected emission wavelength range for fluorescence intensity and FLIM images: 510–560 nm; and collected emission wavelength range for phosphorescence intensity and PLIM images: > 590 nm. Scale bars: 500 μm.
