*2.5. Characterizations*

The hydrodynamic size of NPs were tested by dynamic light scattering (DLS) using a Zetasizer Nano instrument (Malvern Instruments, Malvern, UK). The morphology of NPs was investigated by scanning electronic microscopy (SEM) taken on a JSM-7500dF (JEOL, Tokyo, Japan) at 15 kV. The UV-vis absorption and steady-state emission spectra were measured on a Model V-550 spectrophotometer (Jasco, Tokyo, Japan) and a Model F-4500 fluorescence spectrometer (Hitachi, Tokyo, Japan), respectively. The metallopolymer structure was analyzed by proton nuclear magnetic resonance (1H NMR) measurements, which performed with a 600 MHz high-resolution NMR spectrometer (AVANCE 600 MHz FT NMR, Bruker, Bremen, Germany). The time-resolved photoluminescence experiments were carried out on an FLS 980 PL spectrometer (Edinburgh, UK) excited with laser diodes.

#### **3. Results and Discussion**

#### *3.1. Preparation, Characterization and Properties of Ru-Tb NPs*

A Tb-containing metallopolymer (Tb-Poly) was firstly synthesized by coordinating the precursor Tb(acac)3 3H2O with bipyridine-branched polymer PS−PBPyA (see the Supporting Information for synthesis and 1H NMR data, Figures S1 and S2). Figure 3a shows excitation spectra of Tb(acac)3 and Tb-Poly in DMF solution. Compared to Tb(acac)3, the excitation band of Tb-Poly blue shifts 33 nm due to the formation of a coordinate bond between Tb3+ and bpy ligand. Under the excitation of a 300 nm light, Tb-Poly exhibit the typical intra-f-f transitions of Tb3+ ion (Figure 3b, similar to free Tb(acac)3 in Figure S3). The multiplet transitions 5D4 → 7F6,5,4,3 are clearly observed, as well as a moderate emission at 545 nm. Time-resolved fluorescence of Tb-Poly was also measured by monitoring the 545 nm emission (Figure S4), and the phosphorescence lifetime was obtained to be 1.183 ms by a biexponential fitting. It can be derived that the lifetime of Tb-Poly is increased in comparison to that of Tb(acac)3 3H2O (~0.8 ms) [17] as a result of the replacement of coordinated H2O by bpy ligands. These results indicate that Tb(acac)3 is successfully chelated to the side chain of PS-PBPyA. The hydrophobic metallopolymers Tb-Poly could be easily formed into NPs with a modified nanoprecipitation method [10]. After the formation of NPs, however, a weak fluorescence emerges ranging from 350 nm to 450 nm, as indicated in Figure 3b. Since free Tb-Poly in solution is nonfluorescent in this range, it is reasonable to attribute this blue emission to coordinated bpy ligands of Tb-Poly in NPs, i.e., the back donation from energy level 5D4 of Tb3+ to chelated bpy in solid matrix [18]. This postulation can be further demonstrated by the decreased luminescence lifetime, from 1.183 to 0.94 ms (Figure S4), and lowered quantum yield, from 25.57% to 3.76% (Figure S5), of Tb-Poly in NPs.

**Figure 3.** (**a**) Excitation spectra of Tb(acac)3 3H2O (black) and Tb-Poly (red) in DMF solution. (**b**) Excitation and emission spectra of Tb-Poly in DMF (dash line) and in NPs (solid line), respectively.

The Ru-containing polymer (Ru-Poly) was similarly synthesized according to the procedures in the literature [10]. Based upon Ru-Poly and Tb-Poly, Ru-Tb NPs were prepared in combination with PS−PEG-COOH by the same nanoprecipitation method, in which the latter polymer is used to PEGylate the NPs. The resultant Ru-Tb NPs have a typical size of ~100 nm in diameter (Figure 4a). The shape and size are shown in the SEM image (Figure 4b). The doping ratio of Ru-Poly to Tb-Poly has been optimized to render the comparable emission intensity in air-saturated environments (see Figure S6 for more details). A weight ratio of 7:3:2 (Ru-Poly: Tb-Poly: PS−PEG-COOH) is finally adopted to prepare Ru-Tb NPs. It needs to point out that Ru-Tb NPs are based on a two-wavelength excitation, corresponding to the absorption of bpy ligands in Tb-Poly and metal-to-ligand charge-transfer (1MLCT) in Ru-Poly, respectively. Ru-Tb NPs are very stable in various media (like water and DMEM culture) due to the stabilization of PEG, which can be long-term stored without obvious changes in size (Figure S7).

**Figure 4.** (**a**) Dynamic light scattering and (**b**) scanning electron microscopy image of Ru-Tb NPs.

#### *3.2. Oxygen Sensitivity and Calibration of the Ratiometric Ru-Tb Nanoprobes*

The luminescence response of the ratiometric Ru-Tb NPs toward oxygen was investigated by purging the aqueous suspension with a gas mixture of O2/N2. As showed in Figure 5a, the 610 nm emission of Ru-Poly is strongly quenched with the increase of dissolved oxygen, while the 545 nm emission of Tb-Poly are kept rather constant. In small-sized nanoprobes, luminescence oxygen quenching can be expressed by the linear Stern−Volmer equation, as in the case of the homogeneous system,

$$R\_0/R - 1 = K\_{\rm SV} \,\mathrm{[O\_2]}\tag{2}$$

where *R*0 is the intensity ratio (red emission at 610 nm versus green emission at 545 nm) in the absence of oxygen, *R* the ratio in the presence of oxygen at a given concentration, *K*SV as the Stern−Volmer quenching constant, and [O2] as the concentration of dissolved oxygen. The Stern−Volmer plot of the intensity ratio of Ru-Tb NPs versus oxygen concentration is depicted in Figure 5b. The data were fitted quite well by the linear function (the correlation coefficient >0.997), which is important for practical applications. It needs to point out that the luminescent intensity of nanoprobes is decreased abnormally in the case of oxygen-saturated solution (Figure 5a, 43 ppm) for both the green and red emission. This may be caused by the loss of Ru-Tb NPs due to long-time purging by gas. If the detection modality is based on single intensity, this experimental data would deviate from the linear Stern−Volmer plot considerably. However, the robust ratiometric approach keeps the data still following a linear function relationship.

**Figure 5.** Oxygen sensitivity of Ru-Tb NPs in aqueous solution: (**a**) Emission spectra under 300 nm and 460 nm excitation at various oxygen concentrations (from top to bottom is 0 (nitrogen saturated), 2.15, 4.3, 8.6, 10.75, and 43 ppm (oxygen saturated) in sequence). (**b**) Ratio of red and green luminescent intensity-based Stern−Volmer plot. The experimental data (scatter) were calculated from the ratio luminescence intensities at 610 and 545 nm and linearly fitted (solid line).

#### *3.3. Intracellular Imaging of the Ratiometric Ru-Tb Nanoprobes*

The cytotoxicity of Ru-Tb NPs was tested by MTT assay on living HeLa cells (see Figure S8). The results show that a dosage of <33 μg mL−<sup>1</sup> renders slight growth inhibition of cells (>85% cell viability). Afterwards, the cellular uptake and distribution of Ru-Tb NPs were explored with confocal microscopy with a dosage of 30 μg mL−<sup>1</sup> (Figure 6). It can be deduced from the fluorescence images (red and green channels) and differential interference contrast (DIC) image that the PEGylated nanoprobes can be efficiently uptaken by live cells. Because of the limitation of available lasers in common confocal microscopy, the green channel is obtained under the excitation of a mercury lamp rather than a 300 nm laser. As a result, the quality of the acquired image is not very clear, and a 720 nm light is thus used in the following, based on the mechanism of two-photon excitation.

**Figure 6.** Images of HeLa cells treated with Ru-Tb NPs (30 μg mL−1): (**a**) red channel, excitation at 488 nm and emission at 585–625 nm; (**b**) green channel, excited by mercury lamp and emission at 525–565 nm; (**c**) overlay of red and green channels, together with the DIC image. In a merged picture, colocalization of red and green signal results in orange areas. Scale bar = 50 μm.

Recently three-dimensional multi-cellular tumor spheroids (3D MCTSs) have been popularly studied to mimic the real situation of solid tumor. MCTSs are a valid intermediate between monolayer in vitro cells and in vivo tissue that are formed by heterogeneous cell aggregation [19]; the hypoxic center of MCTSs provides an ideal place for monitoring oxygen. In this experiment, 300 μm MCTSs in diameter can be obtained after 7 days' cultivation from single Hela cell. For staining experiment, 30 μg mL−<sup>1</sup> Ru-Tb NPs in DMEM medium were incubated with MCTSs for 12 h. Cells were visualized by confocal microscopy immediately (Figure 7, and for the whole image, see the Figure S9). The Z-stack confocal images are taken under two-photon excitation. Interestingly, red and green channels could be monitored simultaneously under two-photon excitation at 720 nm, although the emission intensity is weaker than excited under one-photon. The emission spectra of Ru-Tb NPs were measured under excitation at 360 nm in aqueous solution, and the green and red emission peaks could be observed clearly (Figure S10).

**Figure 7.** Z-stack of two-photon microscopy images of MCTs. The images were taken every 14.04 μm section from the top to bottom of intact MCTs. The green and red channel was collected at 525–565 nm and 585–625 nm, respectively. The scale bar = 200 μm.

From Figure 7 and Figure S9, it can be seen that the intensity of red and green emission both increase gradually with the Z-stack section depth (<~54 μm), indicating the efficient uptake of Ru-Tb NPs by MCTSs. When the section depth is above 54 μm, the intensity of red emission decreases while the green emission is rather constant. Obviously, the microenvironment of interest has a high oxygen concentration which seriously quenches the luminescence of Ru-Poly. To display the distribution of dissolved oxygen within MCTSs, the intensity ratio of red emission to green emission of Ru-Tb NPs at the depth of −36 μm is displayed in pseudocolor, and particularly, the spatial distribution of dissolved concentration is roughly quantified by recording the intensity ratio along profiles spanning the section with the confocal software (Figure 8). From the above results, the normoxic brim, and hypoxic core of MCTSs can be discerned clearly.

**Figure 8.** (**a**) Pseudocolor ratiometric intensity images (IF610/IF545) of sectional MCTs at depth of −36 μm and (**b**) intensity ratio along with profiles spanning labeled cells, as indicated by the lines (yellow). The color bar from blue to orange corresponds to the increase of intensity ratio from 0 to 200. Scale bar = 200 μm.
