In Vitro and In Vivo Fluorescence Imaging of Antibody-Drug Conjugate-Induced Tumor Apoptosis Using Annexin V-EGFP Conjugated Quantum Dots.

Antibody-drug conjugates (ADCs) are conjugates of a monoclonal antibody and a cytotoxic drug that induce tumor apoptosis. The evaluation of ADC-induced tumor apoptosis is crucial for the development of ADCs for cancer therapy. To evaluate the efficacy of ADCs, we present in vitro and in vivo fluorescence imaging techniques for ADC-induced tumor apoptosis using annexin V-EGFP (EGFP: enhanced green fluorescent protein) conjugated quantum dots (annexin V-EGFP-QDs). This probe emits visible (VIS) and near-infrared (NIR) dual fluorescence at 515 nm (EGFP emission) and 850 nm (QD emission), which can be used for the detection of tumor apoptosis at the cellular and whole-body levels. By using annexin V-EGFP-QDs, we achieved VIS and NIR fluorescence imaging of human epidermal growth factor receptor 2-positive breast tumor apoptosis induced by an ADC, Kadcyla (trastuzumab emtansine). The results show that the in vitro and in vivo fluorescence imaging of ADC-induced tumor apoptosis using annexin V-EGFP-QDs is a useful tool to evaluate the efficacy of ADCs for cancer therapy.

Introduction

Antibody–drug conjugates (ADCs) are conjugates composed of a monoclonal antibody and a cytotoxic drug via a chemical linker.[1−3] ADCs are used as the antitumor drugs for cancer therapy. To date, ADCs have received market approval and over 70 are being investigated in various stages of clinical development. For the development of ADCs for cancer treatment, the examination of the efficacy of ADCs that induce tumor apoptosis is crucial.[4,5] For the apoptosis detection, we can employ several imaging modalities such as single photon emission computed tomography, positron emission tomography, magnetic resonance imaging, and optical imaging.[6−11] Among these imaging modalities, fluorescence imaging has a high sensitivity and a spatial resolution (∼μm) sufficient to perform molecular imaging at the cellular and whole-body levels.[11] Herein, we present visible (VIS) and near-infrared (NIR) fluorescence imaging techniques for the visualization of ADC-induced tumor apoptosis in vitro and in vivo.

In the early stages of apoptosis, the morphological change in the cancer cell membrane occurs due to the externalization of phosphatidylserine (PS).[12,13] To detect PS molecules on the surface of apoptotic cells, annexin V,[14] an endogenous protein with a binding ability to PS, is widely used as an apoptosis detection probe.[14−17] Although various types of apoptosis detection optical probes[18−36] [e.g., fluorescein isothiocyanate (FITC)-annexin V,[31−34] Cy5-annexin V,[35,36] and 800CW-annexin V[29] have been developed, there are no robust fluorescent probes that enable the long-term optical imaging of apoptosis in vivo. The disadvantage of fluorescent-dye-labeled annexin V probes is the instability of photobleaching caused by the irradiation of excitation lights. To overcome this disadvantage, we synthesized a quantum dot (QD)-based robust probe (annexin V–EGFP–QD) that consists of a fusion protein, annexin V and enhanced green fluorescent protein (annexin V–EGFP),[37] and an NIR-emitting CdSeTe/CdS QD.[38,39] This probe emits VIS and NIR dual fluorescence at 515 nm (EGFP emission) and 850 nm (QD emission), which can be used for the detection of tumor apoptosis at both the cellular and whole-body levels.

In this paper, we present in vitro and in vivo fluorescence imaging of tumor apoptosis induced by an ADC, Kadcyla (ado-trastuzumab emtansine or T-DM1),[40,41] using annexin V–EGFP–QD. Kadcyla is a conjugate of a humanized monoclonal anti-HER2 (human epidermal growth factor receptor 2)[42] antibody and emtansine, a highly potent microtubule polymerization inhibitor.[40,41] Kadcyla is used as an anticancer drug against HER2-positive solid tumors.[43−45] For HER2-positive breast tumor cells (KPL-4[46,47]), we detected Kadcyla-induced apoptosis by the VIS fluorescence of EGFP. We also detected the Kadcyla-induced apoptosis of the HER2-positive breast tumor in mice by the NIR fluorescence of QDs. The NIR fluorescence imaging enabled a long-term (24 days) observation of the tumor shrinking induced by Kadcyla-induced apoptosis. We demonstrate the utility of VIS and NIR fluorescence imaging using annexin V–EGFP–QDs for the detection of ADC-induced tumor apoptosis in vitro and in vivo.

Results and Discussion

Design of the VIS and NIR Emitting Apoptosis Detection Probe

To detect apoptotic cells in vitro, VIS fluorescence imaging is usually used because of the high sensitivity of the VIS light (400–700 nm) in conventional fluorescence microscopies.[31−34] However, VIS fluorescence imaging is not suitable for the in vivo detection of apoptosis because of the strong absorption and scattering of VIS light. In contrast, NIR light (700–900 nm) shows high permeability and low scattering in living tissues. Thus, we employ EGFP and NIR-emitting QDs for apoptosis detection at the cellular and whole-body levels, respectively.[29,35,36]

For in vitro and in vivo apoptosis detection, we designed a QD-based VIS and NIR dual emitting probe, annexin V–EGFP–QD (Figure a). Although several types of QD-based apoptosis detection probes have been developed,[25,28,48−52] there are no dual-emitting QD probes for the fluorescence imaging of apoptosis at the cellular and whole-body levels. Annexin V–EGFP is a histidine-tagged fused protein and directly binds to the surface of the Cd–S layer of glutathione-coated CdSeTe/CdS QDs (GSH–QDs) via the histidine tags of the protein.[53−55] This is due to the high affinity of histidine molecules to Cd2+ ions at the surface of a GSH–QD.[53−55] Annexin V molecules presented at the QD surface can bind the PS molecules of the apoptotic cell membrane in the presence of Ca2+ ions (Figure b). Since annexin V–EGFP–QD emits VIS (EGFP) and NIR (QD) dual fluorescence, in vitro and vivo fluorescence imaging of apoptotic cells can be performed using a single fluorescent probe, annexin V–EGFP–QD.

Figure 1

(a) Synthetic method for the preparation of an annexin V–EGFP–QD by conjugation of a GSH-coated CdSeTe/CdS QD (GSH–QD) with a recombinant protein, annexin V–EGFP. (b) Schematic representation for the binding of an annexin V–EGFP–QD to PS molecules at the surface of the apoptotic cell membrane in the presence of Ca2+ ions.

Characterization of Annexin V–EGFP–QDs

The binding of annexin V–EGFP protein to GSH–QDs can be confirmed by agarose gel electrophoresis (Figures and S1). The mobility of an NIR-emitting QD band decreased with the increasing amount of annexin V–EGFP, indicating the formation of an annexin V–EGFP–QD conjugate. The image of agarose gel electrophoresis for the mixture of annexin V–EGFP and GSH–QDs indicates that ca. five molecules of annexin V–EGFP bind to one QD particle (Figure ). Annexin V–EGFP–QDs can be purified by dialysis (membrane, MWCO: 300,000) to remove unconjugated annexin V–EGFP (66.7 kDa).

Figure 2

Agarose gel electrophoresis of GSH–QDs and mixtures of the QDs and annexin V–EGFP. The molar ratio of annexin V–EGFP–QD was changed from 0 to 7.5. The fluorescence bands of QDs were detected at the wavelength of 830 nm.

The fluorescence spectrum of annexin V–EGFP–QDs showed VIS and NIR dual emission resulting from EGFP and QDs, respectively (Figure a). The fluorescence quantum yields of annexin V–EGFP and CdSeTe/CdS QDs were 75 and 59%, respectively. The intensity of EGFP emission in annexin V–EGFP–QDs decreased by 6 times compared to that in annexin V–EGFP (Figures a and S2). This decrease in the fluorescence intensity of the EGFP emission can be explained by the intramolecular fluorescence resonance energy transfer (FRET) from EGFP to the QD in an annexin V–EGFP–QD conjugate. The fluorescence decay measurements confirmed the FRET from EGFP to the QD in the conjugate. The average fluorescence lifetime[56] (τ) of EGFP in annexin V–EGFP was significantly decreased from 2.7 to 0.53 ns by complexation with GSH–QDs (Figure b). The FRET efficiency from EGFP to the QD was determined to be 76 and 80% from the fluorescence intensities and lifetimes of EGFP, respectively.[57]

Figure 3

(a) Fluorescence spectra of annexin V–EGFP (green line) and annexin V–EGFP–QDs (red line) in PBS. A black dotted line shows the absorption spectrum of GSH–QDs. The inset shows the TEM image of annexin V–EGFP–QDs. Scale bar: 20 nm. (b) Fluorescence decay curves of EGFP emissions in annexin V–EGFP and annexin V–EGFP–QDs in PBS. An excitation pulse (at 475 nm) is shown as a black line.

Transmission electron microscopy (TEM) showed that annexin V–EGFP–QDs are monodispersed nanoparticles with a diameter of 4.4 ± 0.68 nm (inset in Figures a and S3). The hydrodynamic diameter of annexin V–EGFP–QDs was evaluated by using fluorescence correlation spectroscopy (FCS).[56] Since the diffusion times (DTs) of GSH–QDs and annexin V–EGFP–QDs were calculated to be 0.35 ± 0.031 and 0.51 ± 0.045 ms, respectively (Figure a), the hydrodynamic diameters of GSH–QDs and annexin V–EGFP–QDs were estimated to be 5.4 ± 1.5 and 7.9 ± 2.2 nm, respectively.[58−60] This result shows that the diameter of annexin V–EGFP–QDs increased 1.5 times compared to that of GSH–QDs after the binding of annexin V–EGFP to the surface of the QD.

Figure 4

(a) Fluorescence autocorrelation curves for GSH–QDs, annexin V–EGFP–QDs, and fluorescent beads (14 nm in diameter) in 10 mM Na2CO3. The inset graph shows the average DT of the particles: (1) GSH–QDs, (2) annexin V–EGFP–QDs, and (3) fluorescent beads. (b) Size-exclusion column chromatography for GSH–QDs (red line) and annexin V–EGFP–QDs (green line). An inset plot shows the relationship between the molecular weights of standard proteins and their retention times. Molecular weights: 670 kDa for thyroglobulin, 450 kDa for ferritin, 145 kDa for Herceptin, 79.5 kDa for transferrin, and 66 kDa for bovine serum albumin.

The size of GSH–QDs and annexin V–EGFP–QDs was also evaluated by using size-exclusion column chromatography (Figure b). Using the relationship between the molecular weights of standard proteins and their retention times (inset in Figure b), apparent molecular weights of GSH–QDs and annexin V–EGFP–QDs were determined to be 200 and 563 kDa, respectively. As the molecular weight of annexin V–EGFP is 66.7 kDa, the number of annexin V–EGFP protein bound to one QD particle was calculated to be 5.4. This finding is consistent with the result obtained from the agarose gel electrophoresis experiment (Figure ).

Binding Activity of the Annexin V–EGFP–QD to PS

The binding activity of the annexin V–EGFP–QD to PS was examined using FCS. FCS is a very sensitive method to detect changes in the hydrodynamic diameter of fluorescent particles.[57−60] We measured the DTs of annexin V–EGFP–QDs before and after the addition of liposomes containing 10% PS [liposome (PS+)], whose size was 43 nm in diameter (Figure S4). When the liposome (PS+) was added to the solution of annexin V–EGFP–QDs, the DT of annexin V–EGFP–QDs increased from 0.65 to 2.9 ms (Figures S5 and 5a). This change indicates that the hydrodynamic size of annexin V–EGFP–QDs increased 4.5-fold upon the binding of the PS liposomes. In contrast, the addition of liposomes containing no PS [liposome (PS−)] to the solution of annexin V–EGFP–QDs did not change the DT of annexin V–EGFP–QDs (Figures S5 and 5a).

Figure 5

Fluorescence autocorrelation curves for (a) annexin V–EGFP–QDs and (b) GSH–QDs in the absence and presence of soybean lecithin liposomes containing 0% PS (liposome PS−) and 10% PS (liposome PS+). To 10 μL of 10 nM annexin V–EGFP–QD or 10 nM GSH–QD solutions, 5 μL of a liposome solution (0.1 mg/mL, 0.1 mM Ca2+) was added.

A control experiment using GSH–QDs showed that the DT of GSH–QDs did not change upon the addition of liposomes (PS+) and (PS−) to the solutions of GSH–QDs (Figures S5 and 5b). This finding shows that there is no specific binding between GSH–QDs and liposomes (PS+) and (PS−). Thus, the change in the DT of annexin V–EGFP–QDs in the presence of liposome (PS+) indicates the specific binding of annexin V–EGFP–QDs to the PS molecules of liposomes via the annexin V moiety of the QDs.

In Vitro Imaging of Tumor Apoptosis

For fluorescence imaging of tumor cell apoptosis, we used a human breast tumor cell line, KPL-4, that overexpresses HER2[45,46] on the cell surface. The apoptosis of KPL-4 cells was induced by the treatment with an ADC, Kadcyla (a conjugate of Herceptin and emtansine).[40,41] Kadcyla acts as an anticancer drug against HER2-positive cancer.[42−44] Since Herceptin is ineffective to the treatment of KPL-4 cells,[62] we used Herceptin as a control drug to confirm the cytotoxic effect of Kadcyla.

To detect the tumor apoptosis induced by Kadcyla, we first used a traditional apoptosis-detection probe, FITC-annexin V. FITC-annexin V is a widely used a VIS-emitting fluorescent probe for the detection of apoptosis.[31−34] We observed that the KPL-4 cells treated with Kadcyla for 72 h emit intense VIS fluorescence from FITC-annexin V (lower image in Figure a). The image of the control cells treated with no drugs shows only weak emissions from a few cells (upper image in Figure a). The KPL-4 cells treated with an anti-HER2 monoclonal antibody (Herceptin)[61] also show weak emissions from a few cells (middle image in Figure a). These findings indicate that Kadcyla can induce the apoptosis of KPL-4 cells, while Herceptin cannot induce the apoptosis of the KPL-4 cells.[62,63]

Figure 6

Fluorescence imaging of breast tumor cells (KPL-4) treated with and without Herceptin and Kadcyla. Herceptin and Kadcyla (10 nM) were added to cell culture dishes and maintained at 37 °C for 72 h. The cell suspensions were dispensed and incubated with (a) FITC-annexin V (48 nM) or (b) annexin V–EGFP–QDs (33 nM) for 15 min at room temperature. Fluorescence emissions of FITC and EGFP were observed at 525 nm. Scale bar: 50 and 25 μm for magnified images.

As annexin V–EGFP–QDs emit VIS fluorescence from EGFP, we tested whether annexin V–EGFP–QDs can be used as an apoptosis detection probe similar to FITC-annexin V. For the KPL-4 cells treated with Kadcyla, we observed the VIS emission of EGFP (lower image in Figure b). For the control KPL-4 cells (upper and middle images in Figure b), we could not observe significant EGFP fluorescence emissions resulting from the apoptosis of the cells. These imaging patterns were similar to those obtained by using FITC-annexin V (Figure a). This finding indicates that annexin V–EGFP–QDs can be used as an apoptosis detection probe similar to FITC-annexin V for the detection of apoptotic cells.

In Vivo Imaging of Tumor Apoptosis

We conducted in vivo NIR fluorescence imaging of tumor apoptosis using breast tumor-bearing mice. The tumor-bearing mice were prepared by the transplantation of KPL-4 cells to a ventral region of nude mice. Annexin V–EGFP–QDs were intravenously injected to the tumor mice, and then, the NIR fluorescence of the mice was detected. The VIS fluorescence of EGFP could not be detected owing to the strong tissue absorption and autofluorescence (Figure S6).

For the control mouse treated with no antitumor drugs, NIR fluorescence resulting from the tumor accumulation of annexin V–EGFP–QDs was not observed (Figure a). For the mouse treated with Herceptin, NIR fluorescence from the breast tumor was also not observed (Figure b). In contrast, the mouse treated with Kadcyla showed intense NIR fluorescence emission from the breast tumor (Figure c),[63] indicating the accumulation of annexin V–EGFP–QDs to the tumor. These results were consistent with the results obtained by the cellular imaging using annexin V–EGFP–QDs (Figure b). The cellular and whole-body fluorescence imaging (Figures and 7) shows that Kadcyla (10 nM for the cell and 100 μL of 1 mg/mL for the mouse) induces the apoptosis of KPL-4 cells, and the resulting apoptotic cells can be detected by VIS and NIR fluorescence using annexin V–EGFP–QDs.

Figure 7

In vivo NIR fluorescence imaging (a–c) of breast-tumor-bearing mice. Ex vivo images are shown at the right side of each image. Three days after the injection of Herceptin and Kadcyla, annexin V–EGFP–QDs (100 μL of 1 μM solution) were intravenously injected into the mice via the tail vein. The dotted white circles show the positions of tumors. (a) Control mouse: no antitumor drug was injected. (b) Herceptin-injected mouse: 100 μL of Herceptin (1 mg/mL) was injected. (c) Kadcyla-injected mouse: 100 μL of Kadcyla (1 mg/mL) was injected. The fluorescence emissions at 830 nm were detected 3, 4, and 7 days after the injection of Herceptin and Kadcyla. Seven days after the injection of tumors and organs were isolated from the mice. 1: heart, 2: spleen, 3: kidney, and 4: liver. Scale bar: 10 mm.

Long-Term In Vivo Imaging of Tumor Apoptosis

Long-term imaging for a Kadcyla-treated breast tumor-bearing mouse (100 μL of 1 mg/mL Kadcyla injection) was performed to observe the shrinking of a breast tumor. Annexin V–EGFP–QDs were intravenously injected to the Kadcyla-treated mouse 3 days after the injection of Kadcyla (Figure a). NIR fluorescence images of the mouse were taken at 4, 6, 13, and 24 days after the injection of Kadcyla. The accumulation of annexin V–EGFP–QDs to the breast tumor was observed for the tumor-bearing mouse one day after the injection of annexin V–EGFP–QDs (Figure b). The change in the tumor size was evaluated from the NIR fluorescence image of the tumor. The size of the breast tumor was decreased by 4 times during the imaging experiment, showing the effect of Kadcyla on the shrinking of the tumor (Figures c and S7). Twenty-four days after the injection of Kadcyla, the significant shrinking of the tumor (from 10 to 3 mm in diameter) was observed (Figure b). The average NIR fluorescence intensity of the breast tumor was almost constant 6–24 days after the injection of annexin V–EGFP–QDs (Figure d), although the size of the tumor gradually decreased with time. The stable NIR emission from the tumor can be attributed to the robustness of QDs, which have high resistance to photobleaching.

Figure 8

(a) Time course of the experimental procedure for the administration of Kadcyla and annexin V–EGFP–QDs to a breast-tumor-bearing mouse. One hundred microliters of Kadcyla (1 mg/mL) and 200 μL of annexin V–EGFP–QDs (1 μM) were injected into the mouse. NIR fluorescence imaging was performed 4, 6, 13, and 24 days after the injection of Kadcyla. (b) NIR fluorescence images (at 830 nm) of a tumor-bearing mouse. The white dotted circles show the positions of tumors. (c) Time course of the change in the tumor size in a breast-tumor-bearing mouse treated with Kadcyla. The size was determined from the NIR fluorescence images of the breast tumors. (d) Time course of the average NIR fluorescence intensity of a breast tumor after the injection of annexin V–EGFP–QDs.

Finally, we checked the effect of the cytotoxicity of annexin V–EGFP–QDs on the viability of KPL-4 cells. The result showed that the annexin V–EGFP–QDs under the concentration of less than 1 nM did not affect the viability of KPL-4 cells (Figure S8). From the NIR fluorescence image of the mouse injected with annexin V–EGFP–QDs, we observed that the average concentration of annexin V–EGFP–QDs in the mouse was not higher than 1 nM (Figure S9). This finding indicates that the cytotoxicity of annexin V–EGFP–QDs did not affect the apoptosis of KPL-4 cells during the in vivo imaging in this study.

Conclusions

In this paper, we present in vitro and in vivo dual-fluorescence imaging of tumor apoptosis by using a conjugate of a recombinant protein, annexin V–EGFP, and a CdSeTe/CdS QD. We demonstrate that the conjugate, annexin V–EGFP–QDs, can be used for VIS and NIR dual-fluorescence imaging of tumor cell apoptosis. Since the NIR fluorescence emission of annexin V–EGFP–QDs is very bright and stable, long-term imaging of ADC-induced tumor cell apoptosis can be achieved. Although many types of fluorescence-labeled annexin V have been reported, there are a very few probes that can be used for the fluorescence imaging of tumor apoptosis in vitro and vivo. We demonstrate a long-term NIR fluorescence imaging technique of Kadcyla-induced tumor apoptosis using annexin V–EGFP–QDs. The presented imaging technique using annexin V–EGFP–QDs will greatly contribute to the study of the action of ADCs at the cellular level as well as the whole-body level.

Experimental Section

Materials

Glutathione-coated CdSeTe/CdS QDs (GSH–QDs) and annexinV–EGFP were prepared according to the previously reported methods (Supporting Information).[64,65] The anti-HER2 monoclonal antibody (Herceptin) and Kadcyla (trastuzumab emtansine) were purchased from Chugai Pharmaceutical Co. Ltd. (Japan). The FITC-annexin V apoptosis detection kit and MTT cell counting kit were purchased from Nacalai Tesque (Japan). Fluorescent beads (size: 14 nm in diameter, latex, FluoSpheres, carboxylate-modified and red fluorescent) were purchased from Molecular Probes, Inc. Soybean lecithin was purchased from Nacalai Tesque. l-α-Phosphatidyl-l-serine (soybean) was purchased from Sigma. All other regents were of analytical grade and were used as received without further purification. Breast tumor cells (KPL-4) were kindly provided by Dr. J. Kurebayashi (Kawasaki Medical School). Nude mice (5 week old female BALB/c nu/nu) were purchased from Nihon SLC Inc. (Japan).

Preparation of Annexin V–EGFP–QDs

One hundred microliters of annexin V–EGFP [1 mg/mL, phosphate-buffered saline (PBS)] was added to 200 μL of an aqueous solution of GSH–QDs (1 μM, 10 mM Na2CO3 solution). Then, the solution buffer was dialyzed (MWCO: 300,000) for 1 h using 10 mM Na2CO3 to remove unconjugated EGFP–annexin V. The purified annexin V–EGFP–QDs were preserved at 4 °C.

Preparation of the Liposome

Liposome PS (−): 10 mg of soybean lecithin in 10 mL of PBS was sonicated with a tip-type sonicator (Branson Sonifier-150) for 5 min. Then, the liposome suspension was passed through a 0.45 μm membrane filter. The average diameter of the resulting liposomes was 32 nm. Liposome PS (+): 9 mg of soybean lecithin, 1 mg of l-α-phosphatidyl-l-serine (soybean), and 10 mL of PBS were added to a glass tube, and the mixture was sonicated with a tip-type sonicator for 5 min. Then, the liposome suspension was passed through a 0.45 μm membrane filter. The average diameter of the resulting liposomes was 43 nm.

Agarose Gel Electrophoresis

Ten microliters of aqueous solutions of GSH–QDs and annexin V–EGFP was processed with 1% agarose gel in Tris-acetate buffer (pH 8.0) at 100 V for 20 min. Fluorescence emissions of QDs and EGFP were monitored at 830 and 525 nm, respectively.

Optical Measurements

The absorption spectra were recorded with a spectrophotometer (V-670, Jasco). The fluorescence spectra were recorded using a photonic multi-channel analyzer (C10027, Hamamatsu Photonics). The fluorescence decay curves of EGFP emissions were measured by excitation at 485 nm using a time-correlated single-photon counting kit (Horiba Fluoro Cube). The fluorescence autocorrelation curves were measured on a compact FCS system (C9413-01MOD, Hamamatsu Photonics, Japan) at an excitation of 473 nm using a laser-diode-pumped solid-state laser. The size of the pinhole was 25 μm, and the spectral range of the detection wavelengths was 500–900 nm. For the determination of the concentration of GSH–QDs, the number of QD particles in a 10 μL solution was measured by using FCS, and the QD concentration was estimated by using a 20 nM solution of rhodamine 6G as a reference. Fluorescence quantum yields were measured using an absolute quantum yield measurement system (C9920, Hamamatsu Photonics). Excitation wavelengths were set to be 475 nm for annexin V–EGFP and 488 nm for GSH–QDs.

Transmission Electron Microscopy

The morphologies of GSH–QDs and annexin V–EGFP–QDs were observed by TEM using a Hitachi H-800 microscope operating at an acceleration voltage of 200 kV. The TEM sample (ca. 1 μM QDs in PBS) was prepared by dropping the sample solution onto a copper grid.

Size-Exclusion Column Chromatography

Size-exclusion column chromatography using a high-performance liquid chromatography (HPLC) system (Elite LaChrom, Hitachi) was performed using a TSK-gel G4000SW column (7.8 mm × 30 cm, Tosoh). The mobile phase was 10 mM PBS (pH 7.2–7.4), and the flow rate was adjusted to 1 mL/min. Standard proteins of thyroglobulin (670 kDa), ferritin (450 kDa), Herceptin (145 kDa), transferrin (79.5 kDa), and bovine serum albumin (66 kDa) were used to draw a calibration curve. The HPLC chromatographs of GSH–QDs and annexin V–EGFP–QDs were obtained by monitoring the absorption at 600 nm.

Cellular Imaging of Tumor Apoptosis

KPL-4 cells were seeded to 35 mm cell culture dishes (3 × 105 cells per dish) and incubated in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum overnight at 37 °C. Then, the cells were exposed to Herceptin and Kadcyla (10 nM or none) for 72 h at 37 °C. For each dish, the cells floating in the medium and the cells detached during PBS washing were collected, and the cells attached to the dish were carefully detached by trypsin treatment to combine all the cell suspensions into one. The cells were washed once with PBS and then resuspended in 1 mL of the annexin V binding buffer (Nacalai Tesque). The cell suspensions were dispensed in 100 μL volumes and incubated with FITC-annexin V (5 μL; FITC-annexin V apoptosis detection kit, Nacalai Tesque) or annexin V–EGFP–QDs (final concentration, 33 nM) for 15 min at room temperature. The stained cell suspensions were then diluted with 400 μL of binding buffer, from which 100 μL was transferred to a plastic dish for the observation of cells. Fluorescence images were acquired with a fluorescence microscope (BZ-X700, Keyence Corp., Japan) with emission filters. The emission filter for FITC and EGFP was 525 ± 25 nm. The emission filter for QDs was 832 ± 18 nm.

Preparation of Tumor-Bearing Mice

A suspension of KPL-4 cells (107 cells per mouse) was transplanted to the ventral side of 5 week old female BALB/c nu/nu mice. After several weeks, we selected a mouse bearing a tumor less than 10 mm in diameter for imaging. Mice maintenance and animal experiments were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of RIKEN and approved by the Animal Ethics Committee of RIKEN (QA2015-01-8).

In Vivo Imaging of Tumor Cell Apoptosis

NIR fluorescence images were obtained using an in vivo fluorescence imaging system (Bruker, MS FX PRO). An aqueous solution (100 μL) of Kadcyla (1 mg/mL) was intravenously injected via a tail vein of a tumor-bearing mouse. After 3 days, 100 μL of annexin V–EGFP–QDs was injected into the mouse. In vivo NIR fluorescence images of the tumor were obtained after the injection of annexin V–EGFP–QDs. Four days after the injection of the probe, ex vivo images of the breast tumor and organs were taken. In long-term in vivo imaging, images were obtained up to 24 days after the injection for Kadcyla. The NIR fluorescence of annexin V–EGFP–QDs was observed at 830 nm by the excitation at 760 nm. The exposure time was 30–60 s, and the excitation light (400 W xenon lamp) power was 30 μW/cm2 at the ventral side of the mouse.