Conjugated Polymer Amplified Far‐Red/Near‐Infrared Fluorescence from Nanoparticles with Aggregation‐Induced Emission Characteristics for Targeted In Vivo Imaging

Fluorescence‐amplified far‐red/near‐infrared (FR/NIR) nanoparticles (NPs) are synthesized by co‐encapsulation of conjugated polymer donor (poly[9,9‐bis(2‐(2‐(2‐methoxyethoxy)ethoxy)ethyl)fluorenyldivinylene]; PFV) and a fluorogen acceptor (2‐(2,6‐bis((E)‐4‐(phenyl(4′‐(1,2,2‐triphenylvinyl)‐[1,1′‐biphenyl]‐4‐yl)amino)styryl)‐4H‐pyran‐4‐ylidene)malononitrile; TPE‐TPA‐DCM) with aggregation‐induced emission (AIE) characteristics using biocompatible bovine serum albumin (BSA) as the encapsulation matrix. The good spectral overlap and close proximity between PFV and TPE‐TPA‐DCM in BSA NPs result in a 5.3‐fold amplified TPE‐TPA‐DCM emission signal via fluorescence resonance energy transfer (FRET). The obtained PFV/TPE‐TPA‐DCM co‐loaded BSA NPs are spherical in shape with a large Stokes shift of ∼223 nm and low cytotoxicity. The BSA matrix allows further functionalization with arginine‐glycine‐aspartic acid (RGD) peptide to yield fluorescent probes for specific recognition of integrin receptor‐overexpressed cancer cells. The advantage of PFV amplified FR/NIR signal from TPE‐TPA‐DCM is further demonstrated in cellular and in vivo imaging using HT‐29 colon cancer cells and a murine hepatoma H22 tumor‐bearing mouse model, respectively. The high FR/NIR fluorescence and specific cancer targeting ability by RGD surface functionalization make the PFV/TPE‐TPA‐DCM co‐loaded BSA‐RGD NPs a unique FR/NIR fluorescent probe for cellular imaging and in vivo tumor diagnosis in a high contrast and selective manner.


Introduction
Development of effi cient fl uorescent probes with intense far-red/near-infrared (FR/NIR) emission ( > 650 nm) and large Stokes shifts is of great importance in cancer research. [ 1 ] The fl uorescence in FR/NIR region is a unique interrogation window for cellular and in vivo bioimaging with low interferential absorption and limited biological autofl uorescence. [ 2 ] The large Stokes shift can minimize the interference between excitation and emission, as well as shift the emission spectrum away from the sample autofl uorescence to result in high detection sensitivity. [ 3 ] So far, organic fl uorophores [ 4 ] and inorganic semiconductor quantum dots (QDs) [ 5 ] have been widely used as FR/NIR fl uorescent probes for in vitro and in vivo cancer diagnosis. Organic FR/NIR fl uorophores often have poor photostability and small Stokes shifts. [ 6 ] In addition, water-dispersible FR/ NIR emitters generally have planar aromatic structures which make them vulnerable to aggregate via π − π stacking or upon interaction with bioanalytes. This can lead to non-radiative processes, which will signifi cantly reduce their fl uorescence, known as aggregation-caused quenching (ACQ). [ 7 ] As compared to organic fl uorophores, QDs could provide higher fl uorescence, improved photostability and larger Stokes shifts. [ 8 ] However, the oxidative degradation of the heavy metal components of QDs leads to the release of heavy metal ions, which are highly toxic to biological species. [ 9 ] The increasing interest in bioimaging and cancer diagnosis motivates us to pursue alternative FR/NIR fl uorescent probes that can overcome the limitations of poor photostability, small Stokes shift, ACQ effect and high toxicity.
Recently, we have developed a new category of organic luminogens, which have the opposite characteristics to ACQ effect. These luminogens show unique aggregation-induced emission (AIE) feature. [ 10 ] AIE luminogens are often propellershaped. The dynamic rotations of their aromatic rotors nonradiatively deactivate their excited states in the solution state to yield weak fl uorescence. In the aggregate state, the propeller shape of the molecules prevents π − π stacking and blocks the ACQ pathway. The restriction of the intramolecular rotations (RIR) opens radiative decay channel and the luminogens are highly fl uorescent when aggregated. [ 10 ] AIE luminogens can be chemically modifi ed to be water-soluble, which require multiple synthetic steps and long reaction time. As compared to the sophisticated modifi cation of AIE luminogen structures, two strategies have been developed to directly transform organic soluble AIE luminogens into aqueous media for bioimaging applications. One is to form nanoaggregates by direct addition of water to a stock solution of AIE luminogens in watermiscible organic solvents, such as THF and DMSO. [ 11 ] The other is to encapsulate AIE luminogens using block copolymers or biocompatible proteins as the matrix to form nanoparticles (NPs). [ 3 , 12 ] Although these nanomaterials show good photostability and low cytotoxicity in cellular imaging, very few have shown FR/NIR emission and their applications in targeted in vivo imaging remain unexplored. Additionally, probes with high brightness are always desirable to achieve good imaging contrast at minimum dosage. [ 13 ] As compared to sophisticated chemical modifi cations to improve the brightness of each fl uorophore, fl uorescence resonance energy transfer (FRET) is a powerful spectroscopic tool to enhance fl uorescence signals and increase apparent Stokes shift of fl uorescent probes for biological sensing and imaging. [ 14 ] It takes advantage of the light-harvesting property of the donor to enhance the acceptor fl uorescence when they are within close proximity. [ 15 ] In this regard, conjugated polymers (CPs) are effective light-harvesting energy donors due to their π -conjugated backbones and large absorption coeffi cients, [ 16 ] which have been reported to enhance the fl uorescence of various fl uorescent materials, such as organic fl uorophores, [ 17 ] fl uorescent proteins, [ 18 ] QDs [ 19 ] and gold nanoclusters. [ 20 ] In addition, recent studies have revealed that CP-based nanoprobes show high brightness, good photostability and low cytotoxicity, which make them ideal candidates for in vitro and in vivo imaging applications. [ 2c , 13,21 ] By virtue of these desirable features, a CP is selected in this study to pair with an FR/NIR AIE luminogen to constitute the fi rst signal amplifi ed FR/NIR probe for live-animal imaging and cancer diagnosis.
As FRET requires the donor and acceptor chromophores to be located within close proximity, AIE luminogens are ideal fl uorescent acceptors as they show bright fl uorescence in the aggregated states. So far, successful examples of fl uorescence amplifi ed AIE probes for bioimaging applications have been rarely exploited and there is no demonstration of their application in targeted in vivo imaging. [ 3 ] This study thus opens up new opportunities for the development of a new generation of promising probes for the advancement of bioimaging.

Synthesis and Characterization of PFV/TPE-TPA-DCM co-Loaded BSA NPs
Scheme 1 shows the chemical structures of TPE-TPA-DCM and PFV. TPE-TPA-DCM was synthesized by conjugating a paradigm of AIE luminogens, tetraphenylethene (TPE), to 2-(2,6-bis(( E )-4-(diphenylamino)styryl)-4 H -pyran-4-ylidene)malononitrile (TPA-DCM). [ 12 ] The TPE-TPA-DCM could form nanoaggregates upon  gradual addition of water into the luminogen solution in tetrahydrofuran (THF) and the fl uorescence intensity of TPE-TPA-DCM is signifi cantly increased with increasing the water fractions ( f w ) in THF/water mixtures from 50% to 90%, demonstrating the unique AIE feature of TPE-TPA-DCM ( Figure S1 in the Supporting Information (SI)). Figure 1 shows the absorption and photoluminescence (PL) spectra of PFV and TPE-TPA-DCM in THF. PFV has two absorption maxima at 425 and 455 nm. The emission spectrum of PFV has a maximum at 467 nm and a shoulder at 498 nm. On the other hand, TPE-TPA-DCM shows two absorption bands centered at 350 and 486 nm with an emission maximum at 633 nm. Noteworthy is that the emission spectrum of PFV overlaps well with the absorption spectrum of TPE-TPA-DCM, indicating that they may form a good donor-acceptor pair for FRET. It is expected that FRET could occur when PFV and TPE-TPA-DCM are brought into close proximity when they are co-encapsulated into NPs.
Bovine serum albumin (BSA) was selected as the polymer matrix for NP formulation due to its biocompatibility and nontoxicity. [ 22 ] The PFV/TPE-TPA-DCM co-loaded BSA NPs were synthesized through a modifi ed desolvation method. [ 23 ] As shown in Scheme 2 , the NP synthesis started with the addition of a THF solution containing TPE-TPA-DCM and PFV into BSA aqueous solution under sonication, which was followed by cross-linking with glutaraldehyde and THF removal. During NP formation, the TPE-TPA-DCM aggregates tend to be entangled with PFV and the hydrophobic domains of BSA to afford the hydrophobic interiors of the NPs. The negative zeta potentials of the obtained NPs in aqueous solution ( ∼ − 30 mV) indicate that the outermost layer of the PFV/TPE-TPA-DCM co-loaded BSA NPs contains ionized carboxylic groups, which stabilize the NPs through electrostatic repulsion. The similar strategy has also been used to synthesize BSA NPs containing TPE-TPA-DCM or PFV only. As compared to copolymerization of donor and acceptor monomers or covalent conjugation of donor and acceptor molecules to achieve FRET, this strategy of donor/acceptor co-encapsulated into one NP is simple to operate, which minimizes the tedious synthetic efforts.
To optimize the FR/NIR emission, PFV/TPE-TPA-DCM coloaded BSA NPs with various donor/acceptor molar ratios were synthesized and characterized. Table S1 in the SI summarizes the encapsulation effi ciencies (EEs) and average sizes of the PFV/TPE-TPA-DCM co-loaded BSA NPs synthesized at various feeding concentrations of PFV. The feeding concentrations of TPE-TPA-DCM and BSA are fi xed at 0.01 and 1.5 mg/mL, respectively. As shown in Table S1     both TEM and FESEM, which is smaller than the hydrodynamic diameter (159 nm) measured by LLS due to the dry sample state in TEM and FESEM observations.
The UV-vis absorption spectra of PFV-loaded, TPE-TPA-DCM-loaded, and PFV/TPE-TPA-DCM co-loaded BSA NPs in water are depicted in Figure S2 in the SI. PFV-loaded NPs share two absorption bands centered at 430 and 460 nm, respectively. These bands are red-shifted by ∼ 5 nm, respectively, as compared to those of PFV in THF ( Figure 1 ). TPE-TPA-DCM-loaded NPs have two absorption maxima at 358 and 510 nm, which are redshifted by 8 and 24 nm, respectively, as compared to those of TPE-TPA-DCM in THF. As expected, PFV/TPE-TPA-DCM coloaded BSA NPs possess the absorption peaks corresponding to PFV-loaded and TPE-TPA-DCM-loaded NPs, respectively.  1) were also synthesized and used as controls. In addition, as RGD peptide can specifically target integrin receptors overexpressed in many tumor cells, [ 24 ] the PFV/TPE-TPA-DCM co-loaded BSA NPs were further modifi ed with positively charged RGDKKKKKK peptide (isoelectric point ∼ 11.2) at pH 7.4 through electrostatic interaction (Scheme 2 ). The content of peptide in PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs with [PFV RU]/[TPE-TPA-DCM] = 18.8:1 was 6.9 μ mol/g NPs measured by high performance liquid chromatography (HPLC), which corresponded to ∼ 9700 RGD peptide immobilized on each NP. A detailed calculation is described in the SI.

Cellular Imaging
Cellular imaging based on PFV/TPE-TPA-DCM co-loaded BSA NPs with and without RGD functionalization was investigated by confocal laser scanning microscopy (CLSM). HT-29 colon cancer cells that have overexpressed integrin receptors were used as the target cells. [ 25 ] Figure S4 in the SI). Little fl uorescence signal from PFV-loaded NP-stained HT-29 cells (Figures S4A-B) and weak fl uorescence from TPE-TPA-DCMloaded NP-stained HT-29 cells (Figures S4C-D) are observed. The much higher fl uorescence intensity of HT-29 cancer cells shown in Figure 4 C as compared to that in Figure 4 A as well as Figures S4A and S4C, demonstrates that the PFV amplifi ed TPE-TPA-DCM fl uorescence is maintained in cells.  Figure S5 in the SI) is observed in Figures 4 E-F. Quantitative analysis using Image Pro Plus software indicates that the average fl uorescence intensity from PFV/TPE-TPA-DCM coloaded BSA-RGD NP-stained HT-29 cells is ∼ 2-fold higher than that from HT-29 cells in Figure 4 C. These results suggest that more PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs are internalized by HT-29 cancer cells due to specifi c binding between RGD and integrin receptors overexpressed in HT-29 cells. [ 25 ] The specifi c targeting ability of PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs to HT-29 colon cancer cells was further assessed using MCF-7 breast cancer cells with low expression of integrin receptors in cell membrane as a control. MCF-7 cells after incubation with PFV/TPE-TPA-DCM co-loaded BSA NPs with and without RGD functionalization show no obvious difference in fl uorescence intensity ( Figure S6 in the SI). Moreover, quantitative studies of both PFV/TPE-TPA-DCM co-loaded BSA-RGD NP-stained cells suggest that the average fl uorescence intensity from HT-29 cells in Figure 4 E is ∼ 1.9-fold higher than that from MCF-7 cells in Figure S6B. These results further verify the specifi c targeting ability of PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs to integrin receptor-overexpressed cancer cells.
To evaluate the cytotoxicity of PFV/TPE-TPA-DCM co-loaded BSA NPs, the metabolic viability of HT-29 cancer cells after incubation with the NPs was studied at different NP concentrations. As shown in Figure 5 , the cell viability remains > 90% within 48 h even at the highest concentration tested (100 μ g/mL), suggesting low cytotoxicity of the PFV/TPE-TPA-DCM co-loaded BSA NPs.

In Vivo Live-Animal Imaging
The amplifi ed FR/NIR signal within PFV/TPE-TPA-DCM coloaded BSA NPs was also studied using a Maestro EX in vivo fl uorescence imaging system.    and 6 C, intense fl uorescence is observed in the liver area of the mice at 1.5 h post-injection because NPs generally have a tendency to undergo reticuloendothelial system (RES) organ uptake, especially the liver tissue. [ 28 ] The signifi cant decrease in fl uorescence intensity from the liver area of the mice over time suggests that the NPs can be excreted from the body through the biliary pathway from the liver to bile duct, intestine, and feces. [ 29 ] This is confi rmed by the obvious fl uorescence signals in the feces of mice treated with both NPs.
The mice treated with PFV/TPE-TPA-DCM co-loaded BSA NPs with and without RGD functionalization were sacrifi ced at 24 h post-injection and major tissues including tumor were isolated for ex vivo fl uorescence imaging. In consistent with the results of non-invasive in vivo fl uorescence imaging, the fl uorescent signals are mainly observed in tumor and liver tissues and the effi cient tumor targeting of RGD-functionalized NPs is also demonstrated ( Figure 7 ).

Conclusions
In summary, we have developed a fl uorescence-amplifi ed FR/ NIR probe through co-encapsulation of a CP donor (PFV) and an AIE fl uorogen acceptor (TPE-TPA-DCM) into biocompatible BSA NPs. The TPE-TPA-DCM emission was amplifi ed by up to 5.3-fold at [PFV RU]/[TPE-TPA-DCM] = 18.8:1. In addition to the high FR/NIR fl uorescence, the PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs showed large Stokes shift, spherical morphology and low cytotoxicity. Both in vitro and in vivo experiments revealed bright FR/NIR signals and specifi c targeting effect of PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs for both cancer cell and live animal imaging. As probes with high FR/NIR fl uorescence are of great importance for in vivo imaging and cancer diagnosis, this study provides fundamental guidelines to yield bright FR/NIR fl uorescent probes using a FRET strategy and an AIE luminogen. The successful example of amplifi ed FR/NIR emission for cellular and in vivo cancer imaging in a high contrast and selective manner will inspire more exciting research in developing novel FR/NIR fl uorescent probes for improved bioimaging applications. Because of the good performance of PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs in targeted in vivo imaging, we will continue to work on the detailed in vivo studies of the NPs, such as their blood circulation half-life and clearance curve of NPs via the biliary pathway.
imaging results obtained from PFV/TPE-TPA-DCM co-loaded NPs as well as TPE-TPA-DCM-loaded and PFV-loaded NPs in water, respectively, upon excitation at 455 nm. Using the same spectral unmixing algorithm, the acceptor fl uorescence signals (from 600 to 900 nm with maximum at 660 nm, Figure S7A in the SI) were separated from the background. As shown in Figure 6 A, the fl uorescence intensity from PFV/TPE-TPA-DCM co-loaded NPs is much higher than that of TPE-TPA-DCMloaded NPs and no obvious signal is observed from PFV-loaded NPs, which further indicate the signal amplifi cation of TPE-TPA-DCM fl uorescence by PFV.
In vivo FR/NIR fl uorescence imaging and tumor detection based on PFV/TPE-TPA-DCM co-loaded BSA NPs with and without RGD functionalization were further studied on a tumor-bearing mouse model with the Maestro EX in vivo fl uorescence imaging system. In these experiments, murine hepatic H 22 cancer cells were subcutaneously inoculated into the left axillary space of ICR mice, affording H 22 tumor-bearing mice. As H 22 tumor is demonstrated to be integrin positive, [ 26 ] H 22 tumor-bearing mice can be used to evaluate the application of PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs in in vivo targeted imaging of integrin positive tumors. After intravenous injection of PFV/TPE-TPA-DCM co-loaded BSA and BSA-RGD NPs into the mice, respectively, the H 22 tumor-bearing mice were imaged by the Maestro system at various time points. The in vivo non-invasive fl uorescence images were taken upon excitation at 455 nm and the mouse autofl uorescence was removed by the same spectral unmixing algorithm ( Figure S7 in the SI).  Figure S8 in the SI) was also taken and analyzed under the same conditions for Figure 6 B. Noteworthy is that a much higher fl uorescence intensity is observed from the mouse at 24 h post-injection shown in Figure 6 B as compared to that in Figure S8, revealing that PFV amplifi ed TPE-TPA-DCM signal could be used for in vivo imaging in a high contrast manner.
Moreover, as shown in Figure 6 B, the fl uorescence intensity in the tumor area located at the left axillary of the mouse increases over time during 24 h, indicating the accumulation of PFV/TPE-TPA-DCM co-loaded BSA NPs in tumor tissue. The elevating accumulation of NPs into tumor tissue is because of the enhanced permeability and retention (EPR) effect resulting from the tumor microenvironment such as leaky vasculature and poor lymphatic drainage, which is well known as passive tumor targeting of the NPs. [ 27 ] In comparison, the fl uorescence intensity from the tumor site of mice treated with PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs (Figure 6 C) is higher as compared to that in Figure 6 B at all tested time points. Quantitative analysis using the Maestro software indicates that the RGD functionalization of NPs endows a ∼ 2-fold increase in fl uorescence intensity of tumor tissue at all time points (Figure 6 D). These results suggest that PFV/TPE-TPA-DCM co-loaded BSA-RGD NPs can achieve active tumor targeting through specifi c RGD-integrin recognition, making them an effective FR/NIR probe for in vivo fl uorescence imaging and tumor detection in a high contrast manner. In addition, as shown in Figures 6 B After incubation for 2 h, the cells were washed three times with 1 × PBS buffer and then fi xed with 75% ethanol for 20 min, which were further washed twice with 1 × PBS buffer. The stained cells were subsequently imaged by CLSM (Zeiss LSM 410, Jena, Germany) with imaging software (Olympus Fluoview FV1000). The fl uorescent signals from the NPs were collected upon excitation at 405 nm (0.2 mW) or 532 nm (0.2 mW) with a 650 nm longpass barrier fi lter. MCF-7 cancer cells incubated with PFV/ TPE-TPA-DCM co-loaded NPs with and without RGD functionalization were also studied following the same procedures.
Cytotoxicity study: Cytotoxicity of the PFV/TPE-TPA-DCM co-loaded BSA NPs against HT-29 colon cancer cells was evaluated by MTT assay. In brief, HT-29 cancer cells were seeded in 96-well plates (Costar, IL, USA) at a density of 4 × 10 4 cells/mL. After 24 h incubation, the cells were exposed to a series of doses of PFV/TPE-TPA-DCM co-loaded BSA NPs with [PFV RU]/[TPE-TPA-DCM] = 18.8:1 at 37 ° C. To eliminate the UV absorption interference of the PFV/TPE-TPA-DCM co-loaded BSA NPs at 570 nm, the cells were incubated with the same series of doses of the NPs as the control. After the designated time intervals, the sample wells were washed twice with 1 × PBS buffer and freshly prepared MTT solution (0.5 mg/mL, 100 μ L) in culture medium was added into each sample well. The MTT medium solution was carefully removed after 3 h incubation in the incubator for the sample wells, whereas the control wells without addition of MTT solution were washed twice with 1 × PBS buffer. DMSO (100 μ L) was then added into each well and the plate was gently shaken for 10 min at room temperature to dissolve all the precipitates formed. The absorbance of individual wells at 570 nm was then monitored by the microplate Reader (GENios Tecan). The absorbance of MTT in the sample well was determined by the differentiation between the absorbance of the sample well and that of the corresponding control well. Cell viability was expressed by the ratio of the absorbance of MTT in the sample wells to that of the cells incubated with culture medium only.
In Vivo Fluorescence Imaging: All the animal studies were performed in compliance with the guidelines set by the Animal Care Committee at Drum-Tower Hospital. Murine hepatic H 22 cancer cell suspension containing 5-6 × 10 6 cells (0.1 mL) were injected subcutaneously to the ICR mice (average body weight of 25 g) at the left axilla. When the tumor volume reached a mean size of about 300 mm 3 , the mice were intravenously injected with PFV/TPE-TPA-DCM co-loaded BSA NPs ([PFV RU]/[TPE-TPA-DCM] = 18.8:1, 1 mg/mL, 200 μ L) with and without RGD functionalization, respectively. The mice were anesthetized and placed on an animal plate heated to 37 ° C. The time-dependent biodistribution in mice was imaged using a Maestro EX in vivo fl uorescence imaging system (CRi, Inc., Woburn, USA). The light with a central wavelength of 455 nm was selected as the excitation source. In vivo spectral imaging from 500 to 900 nm (with 10 nm step) was conducted with an exposure time of 150 ms for each image frame. The mouse autofl uorescence was removed using spectral unmixing software, leaving the pure NP fl uorescence. After removing the autofl uorescence, the average NP fl uorescence intensity in tumor tissue was calculated from the unmixed signal image using the region-of-interest (ROI) function of the Maestro software. Scans were carried out at 1.5 h, 4 h, 8 h and 24 h post-injection. In addition, the H 22 tumor-bearing mice were also sacrifi ced at 24 h post intravenous injection of PFV/TPE-TPA-DCM co-loaded BSA NPs with and without RGD functionalization, respectively. The organs including brain, stomach, kidneys, liver, spleen, heart, lung, intestine and tumor were excised and imaged by the Maestro system for ex vivo fl uorescence imaging.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.
Characterization: UV-vis spectra were recorded on a Shimadzu UV-1700 spectrometer. Emission spectra were recorded on a Perkin-Elmer LS 55 spectrofl uorometer. Average particle sizes of the NPs were determined by LLS with a 90 Plus particle size analyzer (Brookhaven Instruments Co., USA) at a fi xed angle of 90 ° at room temperature. Zeta potential of the NPs was measured using a zeta potential analyzer (ZetaPlus, Brookhaven Instruments Co., USA) at room temperature. Morphology of the NPs was investigated by FESEM (JSM-6700F, JEOL, Japan) at an accelerating voltage of 10 kV. Sample was fi xed on a stub with a double-sided sticky tape and then coated with a platinum layer using an autofi ne coater (JEOL, Tokyo, Japan) for 60 s in a vacuum at a current intensity of 10 mA. Morphology of the NPs was also studied by TEM (JEM-2010F, JEOL, Japan).
Fabrication of PFV/TPE-TPA-DCM co-Loaded BSA NPs: The PFV/TPE-TPA-DCM co-loaded BSA NPs were prepared by a modifi ed desolvation method. Briefl y, BSA (7.5 mg) was dissolved in Milli-Q water (5 mL). Subsequently, THF (8 mL, desolvation agent) containing predetermined amounts of PFV (changed from 125 to 500 μ g) and TPE-TPA-DCM (fi xed at 50 μ g) were added dropwise into the BSA aqueous solution at room temperature under sonication using a microtip probe sonicator (XL2000, Misonix Incorporated, NY, 18 W output), which resulted in the formation of PFV/TPE-TPA-DCM co-loaded BSA NPs. Glutaraldehyde solution (50%, 5 μ L) was subsequently added to cross-link the obtained NPs at room temperature for 4 h. THF was removed by rotary evaporation under vacuum. The cross-linked NP suspension was fi ltered through a 0.45 μ m microfi lter, which was then washed with Milli-Q water. The amounts of PFV/TPE-TPA-DCM loaded into the BSA NPs were determined from the absorption spectra with reference to a calibration curve of PFV and TPE-TPA-DCM in THF. The encapsulation effi ciency is defi ned as the ratio of the amount of PFV or TPE-TPA-DCM loaded into the NPs to the total amount of PFV or TPE-TPA-DCM in the feed mixture.
Surface Functionalization: The RGD peptide functionalization of PFV/ TPE-TPA-DCM co-loaded BSA NPs was performed according to the literature. [ 30 ] In brief, PFV/TPE-TPA-DCM co-loaded BSA NPs (1.5 mg) were fi rst dispersed in 0.1 × PBS (1 mL). RGDKKKKKK solution (10 − 3 M, 20 μ L) was then added into the NP suspension and gently mixed for 2 h. Subsequently, The NPs were washed three times with Milli-Q water and the supernatant was collected to determine the free RGD peptide by high performance liquid chromatography (HPLC, Shimadzu LC-ESI spectrometer). The amount of RGD peptide on each NP was determined as the ratio of the total number of immobilized peptide to the total number of NPs, which is described in detail in the SI.
Cell Culture: HT-29 colon cancer cells, MCF-7 breast cancer cells and murine hepatic H 22 cancer cells were cultured in Dulbecco's Modifi ed Eagle's Medium (DMEM) containing 10% fetal bovine serum and 1% penicillin streptomycin at a constant temperature of 37 ° C in a humidifi ed environment containing 5% CO 2 . Prior to the imaging experiments, the cells were precultured until confl uence was reached.
Cell Imaging : HT-29 cancer cells were cultured in chamber (LAB-TEK, Chambered Coverglass System, Rochester, USA) at 37 ° C. After 80% confl uence, the medium was removed and the adherent cells were