In vivo covalent cross-linking of photon-converted rare-earth nanostructures for tumour localization and theranostics

The development of precision nanomedicines to direct nanostructure-based reagents into tumour-targeted areas remains a critical challenge in clinics. Chemical reaction-mediated localization in response to tumour environmental perturbations offers promising opportunities for rational design of effective nano-theranostics. Here, we present a unique microenvironment-sensitive strategy for localization of peptide-premodified upconversion nanocrystals (UCNs) within tumour areas. Upon tumour-specific cathepsin protease reactions, the cleavage of peptides induces covalent cross-linking between the exposed cysteine and 2-cyanobenzothiazole on neighbouring particles, thus triggering the accumulation of UCNs into tumour site. Such enzyme-triggered cross-linking of UCNs leads to enhanced upconversion emission upon 808 nm laser irradiation, and in turn amplifies the singlet oxygen generation from the photosensitizers attached on UCNs. Importantly, this design enables remarkable tumour inhibition through either intratumoral UCNs injection or intravenous injection of nanoparticles modified with the targeting ligand. Our strategy may provide a multimodality solution for effective molecular sensing and site-specific tumour treatment.

RE(CH 3 CO 2 ) 3 (RE = Y, Yb, Er and Nd), 3 mL oleic acid and 7 mL 1-octadecene were added in a 50 mL three-neck flask. The ratio of Yb/Er/Nd is 18/0.5/1% and the total lanthanide amount is 0.4 mmol. The mixture was heated to 150 °C for 60 min before cooling down to room temperature. Subsequently, a methanol solution (6 mL) containing NH 4 F (59.3 mg) and NaOH (40.0 mg) was added and stirred for 30 min at 50 °C. Then methanol was evaporated and the solution was kept at 290 °C for 1.5 h under nitrogen atmosphere. After cooling down to room temperature, the core particles were precipitated using ethanol and collected through centrifugation several times after ethanol washing. Finally, the core particles were re-dispersed in 4 mL hexane. 4 :Yb/Er/Nd (18/0.5/1%)@NaYF 4 :Nd (20%) core-shell UCNs: Typically, 3 mL oleic acid and 7 mL 1-octadecene were added in a 50 mL three-neck flask, then 2 mL methanol containing 0.4 mmol RE(CH 3 CO 2 ) 3 (RE = Y/Nd (80/20%)) was also added in the solution. The mixture was heated to 150 °C for 60 min before cooling down to room temperature. The as-synthesized NaYF 4 :Yb/Er/Nd core particles in 4 mL hexane were added along with a methanol solution (6 mL) containing NH 4 F (59.3 mg) and NaOH (40.0 mg). The resulting mixture was stirred for 30 min at 50 °C. Then methanol was evaporated and the solution was kept at 290 °C for 1.5 h under nitrogen atmosphere. After cooling down to room temperature, the core-shell particles were precipitated using ethanol and collected through centrifugation several times after ethanol washing. The absolute quantum yield of core-shell UCN was 0.22 % under excitation at 808 nm (80 W cm -2 ) according to the standard process 2 .

1.3) Synthesis of PAA@UCNs:
The PAA@UCNs synthesis was carried out following a literature protocol reported previously 3 . Briefly, 300 mg of PAA (1800 Da) and 30 mL DEG were added to a three-neck flask. The mixture was heated to 110 °C to form a clear solution. Then a hexane solution containing 100 mg core-shell UCNs was added slowly and the mixture was maintained at 110 °C for 1 h under nitrogen protection. After heating at 240 °C for 1.5 h, the resultant solution was cooled down to room temperature and precipitated with ethanol. The obtained PAA@UCNs was recovered by centrifugation (18000 rpm for 10 min) and washed three times with ethanol/water (1:1 v/v).

1.4) Synthesis of PEI/PAA@UCNs:
The PAA-capped UCNs were covalently encapsulated with branched PEI as reported previously 4 . Briefly, 30 mg of PAA@UCNs in 5 mL water were activated by EDCHCl (50 mg) and NHS (5 mg) to form the active succinimidyl ester for 30 min. Then the mixture was added slowly to 10 mL PEI (2 mg mL -1 ) in PBS (pH 7.4) with 30 min sonication. After 24 h magnetic stirring, the PEI/PAA@UCNs was purified by centrifugation (18000 rpm for 10 min) and washed three times with ethanol/water (1:1 v/v) to remove excess reactant.

1.5) Synthesis of enzyme responsive cross-linking peptide sequence and negative control peptide:
All peptide sequences were synthesized by using standard Fmoc-chemistry based solid phase peptide synthesis (SPPS) with 2-Chlorotrityl Chloride Resin. HBTU/DIPEA was used for coupling reaction at room temperature and 20% piperidine/DMF solution was used to deprotect Fmoc group in the terminal of peptide. The synthetic route of enzyme responsive cross-linking peptide sequence Ac-FKC(StBu)AC(SH)-CBT was shown in supplementary Fig. 3 as the following process 5 .

1.5.1) Synthesis of substrate 1:
The isobutyl chloroformate (IBCF, 23.4 mg, 0.17 mmol) was added the mixture of Fmoc-Cys(Trt)-OH (72.9 mg, 0.12 mmol) and 4-methylmorpholine (MMP, 34.6 mg, 0.34 mmol) in 3 mL anhydrous THF at 0 °C under N 2 atmosphere and the mixture was stirred for 20 min. Then 2-cyano-6-aminobenzothiazole (CBT, 20.0 mg, 0.11 mmol) was added to the mixture for further stirring at 0 °C for 2 h before overnight at room temperature. After evaporating the THF solvent under reduced pressure, 10 mL water was added to the reaction mixture with extracting with 10 mL ethyl acetate for three times. The organic phase was combined and dried using Na 2 SO 4 . The mixture was concentrated under reduced pressure and subsequently a flash column chromatography with eluent ethyl acetate and hexane is performed to obtain a white product (53.1 mg, 65%). 1

1.5.2) Synthesis substrate 2:
Substrate 1 (40.0 mg, 0.05 mmol) of was added into 5 % piperidine dissolved in 3 ml of DMF and allowed to stir under ambient temperature for 20 min. Next, 5 mL of water was added to the reaction mixture and extracted with 10 mL ethyl acetate for three times. The organic phase was combined and dried using Na 2 SO 4 . After concentrating the mixture under reduced pressure, the sample was purified using flash chromatography with choice of eluent as ethyl acetate and petroleum ether to obtain substrate 2 (19 mg, 73%). 1

1.5.3) Synthesis substrate 3 and enzyme responsive cross-linking peptide of Ac-FKC(StBu)AC(SH)-CBT:
Substrate 2 (17.3 mg, 0.033 mmol) was dissolved in 1 mL of DMF solution containing Ac-Fk(Boc)C(StBu)A-OH peptide (28.9 mg, 0.041 mmol). EDC (10.3 mg, 0.066 mmol) of and NHS (10.1 mg, 0.066 mmol) were then added into the reaction mixture. Finally, DIPEA (14.5 µl, 0.083 mmol) was added into the reaction flask and allowed to stir over night under ambient condition. After monitoring the starting materials has been consumed in the reaction through TLC, the reaction mixture was extracted using 10 ml of ethyl acetate, water and brine and finally dried over anhydrous sodium sulphate. After the reaction mixture has been concentrated under reduced pressure. The crude mixture of 3 obtained was dissolved in 1ml of DCM and subsequently trifluoroacetic acid (TFA) and triisopropylsilane (TIPS) were added into the reaction flask. The mixture was then purified using reverse phase HPLC to obtain enzyme responsive cross-linking peptide (28.3 mg, 42% in two steps). 1

1.5.4) Synthesis of control non-cross-linking peptide:
The control peptide sequence Ac-FKC(StBu)AC without cross-linking property was obtained using the conventional solid phase peptide synthesis (SPPS) with 2-Chlorotrityl Chloride Resin. HBTU/DIPEA was used for coupling reaction at room temperature and 20% piperidine / DMF solution was used to deprotect Fmoc group in the terminal of peptide. After washing with DMF and DCM several times, 95% TFA was used to cleave the peptide from the resin and remove Boc and Trt protecting group under stirring for 2 h. The negative control peptide was purified using reverse-phase HPLC and then frozen-dried with lyophilizer. 1  After stirring for 24 h, the mixture was precipitated by centrifugation and washed with water thrice, and then it was dissolved in 10 mL PBS (10 mM, pH=7.4) containing enzyme responsive cross-linking peptide sequence Ac-FKC(StBu)AC(SH)-CBT (20.6 mg) in 200 μL DMSO. Accordingly, the controlled non-cross-linking particle conjugate (NCRUN) was prepared following the same protocol with Ac-FKC(StBu)AC. The final products CRUN and NCRUN were obtained by centrifugation and washed with water three times. The successful formation of enzyme responsive cross-linking peptide and Ce6 modified UCNs conjugate (CRUN) was confirmed by UV-Vis spectroscopic characterization ( Supplementary Fig. 4). The loading percentage of Ce6 on UCNs was determined by UV-Vis absorbance peak at 665 nm which is comparable other reported values 3,6 . The amount of enzyme responsive cross-linking peptide on UCN was determined by monitoring the fluorescence emission of CBT group (λ ex = 320 nm, λ em = 450 nm) as shown in supplementary Fig. 5.

2.1) Enzyme triggered cross-linking reaction of CRUN in buffer:
The reaction scheme of enzyme triggered cross-linking was shown in supplementary Fig. 6. Briefly, the stock solution of CtsB protease enzyme was dissolved in 25 mM of acetate buffer solution (pH 5.0) with 1 mM EDTA. The upconverted luminescence (UCL) spectrum in quartz cuvette was recorded after incubating CtsB (55 nM) and CRUN (1.5 mg mL -1 ) in working buffer (0.1 M sodium phosphate, 0.05 M NaCl, 1mM EDTA, pH 6.0) at 37 °C containing TCEP (5 mM, a commonly used reductive agent) for different time 7 . In the negative control experiment, the typical CtsB inhibitor 8 , antipain hydrochloride (100 μM), was pretreated with CtsB for 2 h at 37 °C and then incubated with CRUN for different time as above. After enzyme triggered cross-linking reaction, the CRUN were separated by centrifugation and re-dispersed in ethanol for TEM test (supplementary Fig. 7).
The UV-Vis absorption and luminescence spectra of CRUN in the presence of CtsB and anitipain inhibitor were measured after 4 hours incubation. In order to confirm the possibility of the cross-linking reaction, the formation of the covalent cross-linking product, firefly luciferin, was monitored by the fluorescence change at 489 nm (320 nm light as excitation wavelength) 5 (supplementary Fig. 8).
In the mean time, the lanthanide UCL of CRUN with CtsB treatment was also measured with 808 nm excitation at different time intervals (supplementary Fig. 9). The luminescence photographs of different CRUN samples in the absence of CtsB, in the presence of CtsB, or with both CtsB enzyme and inhibitor treatment were recorded in the dark, respectively. The relevant fluorescence intensity was measured by Image J software (supplementary Fig. 10). The luminescence decay curves of lanthanide dopants (Er 3+ , Nd 3+ and Yb 3+ ) in CRUN were measured in the absence and presence of CtsB enzyme after 4 h incubation (supplementary Fig. 11). After enzyme triggered cross-linking reaction (4 h ), the CRUN were separated by centrifugation and the precipitates were re-dispersed in 0.1 mL DMSO solution. The resonance light scattering (RLS) spectra were obtained by scanning synchronously with the same excitation and emission wavelength (λ ex = λ em ) from 808 to 900 nm (supplementary Fig. 12) 9 . Both the excitation and emission slit widths were kept at 10.0 nm.

2.2) Quantification of cathepsin B and folate receptor expression in HT-29 and NIH/3T3 cells:
The human colorectal adenocarcinoma cell line (HT-29, cat. no. HTB-38) and mouse embryonic fibroblast cell line (NIH/3T3, cat. no. CRL-1658) were provided from American-type culture collection (ATCC) and checked for mycoplasma contamination. These cells are not listed by ICLAC as misidentified cell lines (October 3rd, 2014). Both these two cells were cultured in high-glucose DMEM containing 10% FBS, penicillin (100 I.U. mL -1 ) and streptomycin (100 μg mL -1 ) at 37 °C in a humidified atmosphere with 5% CO 2 . The quantification of Cathepsin B (CtsB) in HT-29 cells (40.1±4.4 ng mg -1 protein) and NIH/3T3 cells (2.2±1.4 ng mg -1 protein) were performed by Human Cathepsin B ELISA (Enzyme-linked immunosorbent Assay) kit according to the procedure described by the manufacturer (CUSABIO, USA) 10 . The protein content was determined by using bicinchoninic acid (BCA) method after protein extraction from 1×10 5 cells based on the standard protocol (Thermo Scientific, USA) 11 . The quantification of folate receptor levels in HT-29 cells (3.8±0.3×10 5 folate receptors per cell) was also performed by Human FOLR1/Folate Receptor Alpha ELISA kit according to the procedure described by the manufacturer (CUSABIO, USA) 12 .
Western blot assay was also performed to compare the difference expression levels of CtsB in these two cell lines (Supplementary Fig. 13). Briefly, 1.2×10 6 HT-29 or NIH/3T3 cells were lysed with 50 μL 2×SDS sample buffer and boiled at 90˚C for 5 min. Then 20 μL of each samples were analyzed by 10% SDS-PAGE and transferred to 0.22 μm PVDF membrane (100 V, 110 min, 4 ˚C). The membrane was blocked with 5% milk (PBST) for 1h, and incubated with rabbit anti-Cathepsin (ab125067, Abcam) antibody (1:750 dilutions) at 4 ˚C overnight. The membrane was washed 3 times with PBST, then incubated with goat anti-rabbit 2 nd antibody (1: 10000 dilutions) for 1h at RT and washed 3 times with PBST. The samples were detected with LAS-4000 machine using enhanced chemiluminescence (ECL) exposure buffer. After that, another 20 μL (the same amount as the above) of each samples were analyzed in 10% SDS-PAGE and transferred to 0.22 μm PVDF membrane (100 V, 110 min, 4˚C). The membrane was incubated with rabbit anti-GAPDH antibody (sc-25778, Santa Cruz, 1:1000 dilutions) for 1 h at RT, washed 3 times with PBST, then incubated with goat anti-mouse 2 nd antibody (1: 10000 dilutions) and revealed by ECL.

2.3) Fluorescence imaging of enzyme triggered particle cross-linking in living cells:
Cathepsin B (CstB) over-expressing HT-29 cells and cathepsin B deficient NIH/3T3 cells were seeded with a cell density of 1×10 5 in an ibidi confocal μ-dish (35 mm, plastic bottom) in 1 mL DMEM media which was incubated for 24 h at 37 °C. The seeded cells were subsequently treated with CRUN (50 μg mL -1 ) and incubated for 4 h. The cell nuclei were stained by DAPI (1 μg mL -1 ) for 20 min and then the cells were washed with PBS (pH 7.4) for three times. As controls, the HT-29 cells and NIH/3T3 cells were treated with CRUN in the present of CtsB inhibitor. These two cells were also treated with non-cross linking NCRUN conjugates, which were used as another control. For the cell inhibition assay, the cells were first pretreated with CtsB inhibitor (antipain hydrochloride, 100 μM in DMEM) for 2 h at 37 °C follow by the addition of CRUN and incubated for a further 4 h. Fluorescent imaging of the samples were then carried out under confocal microscopy system with a continuous-wave 980nm laser as excitation wavelength (EINST Technology Pte Ltd). The cellular imaging data of enzyme triggered CRUN cross-linking in HT-29 and NIH/3T3 cells were shown in Fig. 2g and supplementary Fig. 14. The fluorescence analysis of enzyme triggered particle cross-linking in these two cells were measured by Image J (supplementary Fig. 15).

2.4) Singlet oxygen analysis in buffer solution:
Singlet oxygen ( 1 O 2 ) was detected based on a highly sensitive probe singlet oxygen sensor green (SOSG) according to the protocol reported previously 13 (Fig. 2i and Supplementary Fig. 16). Briefly, SOSG (100 μg per tube) was dissolved in 33 μL methanol as stock solution (5 mM). Different samples containing Ce6 (1 μM) were mixed with SOSG (2.5 μM) in PBS (10 mM, pH 7.4) and then irradiated by an 808 nm or 660 nm laser (0.4 W cm -2 ) for different periods of time in the dark.
In addition, a 2-mm thick pork tissue (adipose tissue) was used to mimic living condition for further investigation of the different penetration between 808 nm (NIR) and 660 nm light irradiation (Fig. 2i). The generation of singlet oxygen was based on the enhanced fluorescence of SOSG at 530 nm compared with background signals under the excitation wavelength at 494 nm.

2.5) In vitro photodynamic cytotoxicity assay of CRUN in HT-29 and NIH/3T3 cells:
The HT-29 and NIH/3T3 cell lines were seeded in a 96-well plate (1×10 4 cells per well in 100 μL DMEM) and cultured for 24 hours at 37 °C. The cells were incubated with CRUN at various concentrations for 4 h and then washed with fresh medium to remove free nanoparticles. For in vitro PDT treatment, cells were exposed to 808 nm or 660 nm irradiation at the same power density of 0.4 W cm -2 for 60 min in the dark (10 min break after 10 min irradiation). Similar control experiments were employed as following: cells treated with CRUN in the absence of laser illumination; with CRUN in the presence of CtsB inhibitor (antipain hydrochloride, 100 μM); with controlled NCRUN conjugates in the presence of laser irradiation; and cells treatment by NIR light irradiation only but without particle incubation.
Moreover, to demonstrate the deep-tissue penetration advantage of 808 nm (NIR) light for PDT treatment, CRUN was incubated with enzyme over-expressed HT-29 cells and treated with 660 nm or 808 nm light irradiation in the absence or presence of 2 mm thick pork tissue for 1 h under the same optical power density. After PDT treatment, the cells were incubated for another 48 h in the dark. The cell viabilities were evaluated by in vitro toxicology assay kit (TOX8, resazurin based) according to standard manufacturer's protocol 14 . Each experiment was repeated three times and the average values were used for analysis (supplementary Fig. 17 and  18).

2.6) Photoacoustic (PA) imaging experimental protocol and parameters:
Phantom in vitro and in vivo photoacoustic (PA) imaging experiments were performed using MSOT imaging system 15 . Typically, the optical excitation was provided by an optical parametric oscillator (OPO) with a tunable NIR wavelength which ranges from 680 nm to 980 nm that is in turn pumped by a Q-switched Nd:YAG laser with an average pulse duration of about 10 ns and repetition rate of 10 Hz. The PA signals were acquired using a 128-element concave transducer array spanning a circular arc of 270°. This transducer array has a central frequency of 5 MHz, which is used to provide a transverse spatial resolution in the range of 150-200 mm. During the process of image acquisition, the sample is being translated via the transducer array along its axis across the volume region of interest (ROI). In the process of in vivo imaging, the ultrasound gel was applied on the surface of the mouse skin and subsequent measurements were acquired in temperature-controlled water for a good acoustic coupling. For the PA experiments of i.v. injection, a similar PA imaging experiment was conducted under the same condition of acoustic signals were acquired using a 256-element tomographic ultrasound detector array spanning a circular arc of 270° with a central frequency of 5 MHz in the same system whole body imaging were operated under conditions of 0.3 mm step distance along the long axis of animal and 10 repeat pulse per position in designed time point. PA signal intensity of tumor was measured with region of interest (ROI) assay. Considering the multiple absorption peaks of lanthanide doped upconversion nanostructures due to their specific f-f transitions 16 , in our study, the multi-wavelength PA signals were recorded from 680 nm to 980 nm illumination, and the maximum contrasted PA signal can be obtained when the excitation wavelength is 680 nm during PA imaging experiments.

2.7) In vitro PA imaging of enzyme triggered particle cross-linking in buffer solution:
The stock solution of CRUN (1.5 mg mL -1 ) and CtsB (55 nM) were first incubated in 200 µL working buffer (0.1 M sodium phosphate, 0.05 M NaCl, 1 mM EDTA, pH 6.0) at 37 °C containing TCEP (5 mM) as reduction agent for different time. Then the solution was loaded into a PA phantom containing two-channel polyurethane cylindrical for PA testing, one for holding the control medium (working buffer) and the other for holding CRUN after incubation with enzyme for different time. The PA signal was recorded using a 128-element concave transducer array spanning a circular arc of 270° with the optimal excitation wavelength at 680 nm. In the negative control experiment, a typical inhibitor, antipain hydrochloride (100 μM) was first pretreated with CtsB for 2 h at 37 °C and then incubated with CRUN for different time as same protocol.

2.8)
In vitro PA imaging of enzyme triggered particle cross-linking in living cells: HT-29 cells and NIH/3T3 cells were cultured in a 65 mm dish with a density of 4 million cells per well in 2 mL DMEM media and incubated for 24 h at 37 °C. The cells were incubated with CRUN (1.5 mg mL -1 ) for 4 h at 37 °C incubator. In the control experiment, the cells were pretreated with CtsB inhibitor (antipain hydrochloride, 100 μM in DMEM) for 2 h at 37 °C before the addition of CRUN for 4 h incubation. Then cells were harvested using trypsin and washed three times with PBS. The cells were re-dispersed in 0.2 mL PBS for PA imaging test in two-channel cylindrical.

3.1) In vivo animal studies:
All animal experimental procedures were performed in accordance with the protocol #120774 approved by the Institutional Animal Care and Use Committee (IACUC). Female Balb/c nude mice (~6-8 week old) were purchased from Charles River Laboratories (Shanghai, China). Xenograft mice models were established by injecting 0.1 mL of tumor cells suspension (PSB/matrigel, BD biosciences 1:1, v/v) containing 3×10 6 HT-29 into both flanks or only right side of mouse. When the tumor volume reached a palpable size, the mouse was used for further studies.

3.2) In vivo PDT treatment through intratumoral injection of enzyme triggered cross-linking of CRUN in tumor bearing mice:
Xenograft mice models were treated when the tumor volumes approached to 3-5 mm in diameter. For in vivo PDT treatment, the tumor-bearing mice were randomized into four groups (n = 8, each group) and treated by intratumoral injection of particle samples. In group 1, CRUN (3 mg in 100 µL saline) was directly injected into implanted tumors, followed by an 808 nm laser irradiation after 4 h injection. In group 2, same amounts of controlled NCRUN conjugate was injected with subsequent laser exposure. Control experiments with injection of CRUN but no NIR light irradiation (group 3) and saline with laser treatment (group 4) were employed, respectively. The effective exposure time for each mouse was 45 min (5 min interval after 5 min irradiation with total treatment time of 90 min) with power density at 0.4 W cm -2 (fluence 1080 J cm -2 ). A second and third dose injection followed with PDT treatment described above was repeated at 4 d and 8 d after the first dose injection, respectively. Tumor size was measured three times a week using a vernier caliper upon PDT treatment. The tumor volume was calculated using the following equation: tumor volume (V) = length × width 2 /2. Relative tumor volume was calculated as V/V 0 (V 0 was the initial tumor volume before PDT treatment).

3.3) In vivo PA imaging of enzyme triggered cross-linking in tumor region:
For in vivo PA imaging of intratumoral injection of CRUN and NCRUN (3 mg in 100 µL saline), the tumor bearing mice was anaesthetized with 3% isoflurane and ultrasound gel was applied on the surface of the mouse skin. Next, the distribution of the probe in tumor region was monitored over a period of time using PA imaging. In order to obtain an image acquisition, several slices with a step size of about 0.3 mm spanning from the liver all the way to abdomen was manually selected by live MSOT images with 8 laser excitation wavelengths from 680 nm to 980 nm (50 nm interval). Multispectral imaging was then performed with 10 signal averages per wavelength per transverse slice before injection and 4, 8 and 24 h post-injection (supplementary Fig. 19).

3.4) Histology examination of tumor tissues after treatment:
Mice were sacrificed after the multiple intratumoral injections with NIR light mediated PDT treatment. The tumor tissues were collected and fixed with 4% paraformaldehyde solution, processed routinely into paraffin and sectioned into 8 microns thick slices. After stained with hematoxylin & eosin (H&E) followed by standard protocol 17,18 , the slices were examined by a digital optical microscope (BX 51, Olympus, USA). Damaged cells in tumor region was performed using the terminal deoxyribonucleotide transferase-mediated nick-end labeling (TUNEL) assay using an in situ damaged cell detection kit (Roche, Basel, Switzerland) according to the manufacturer's instructions 19 . The extent of damaged cells was evaluated by counting the number of TUNEL-positive (brown-stained) cells. The damaged index was calculated as the number of TUNEL-positive cells divided by the total number of cells in 10 randomly selected high-power fields, as shown in supplementary Fig. 20.

3.5) Synthesis of tumor affinity ligand conjugated CRUN (FA-PEG@CRUN):
The illustration of synthetic scheme of FA-PEG@CRUN and PEG@CRUN were shown in supplementary Fig.  21. First, the FA-PEG 5000 -SH conjugate was synthesized according to previous study 20,21 . Briefly, FA (44.2 mg) was activated with EDC (48.0 mg) and NHS (28.8 mg) in DMSO (6 mL) under N 2 for 30 min. The resulting activated FA solution was added dropwise to the DMSO solution (2 mL) of NH 2 -PEG 5000 -SH (Mw 5000, 100 mg) with triethylamine (TEA, 145.7 μL) under vigorous magnetic stirring in the dark at room temperature for 24 h. Then the reaction mixture was dialyzed in water using a dialysis membrane with molecular weight cut-off (MWCO) of 3500 Da for 3 days and followed by lyophilization to obtain the product FA-PEG 5000 -SH. Second, the as-prepared CRUN (40.0 mg) was activated by Mal-dPEG TM 2 -NHS (20.0 mg) ester linker in PBS (10 mM, pH = 7.4) for 12 h. After remove the excess reactants, the Mal-activated UCN complex was re-dissolved in PBS (10 mM, pH = 7.4) and then FA-PEG 5000 -SH (25.0 mg) was added into the solution with vigorous magnetic stirring in the dark at room temperature for 24 h. Accordingly, the PEG modified CRUN particles but without FA as affinity ligand (PEG@CRUN) were prepared as controlled conjugate following the same protocols with NH 2 -PEG 5000 -SH (25.0 mg) modification. The final products FA-PEG@CRUN and PEG@CRUN were obtained by centrifugation and washed with water for three times. The luminescence analysis of FA-PEG@CRUN (1.5 mg mL -1 ) at different time intervals after CtsB enzyme treatment (55 nM) was followed the same protocol as previous CRUN platform without FA modification (supplementary Fig. 22).

3.6) In vivo targeted PDT treatment with intravenous injection of enzyme triggered FA-PEG@CRUN:
Xenograft HT-29 tumor model in the right side of female Balb/c mice were first developed as described above. For in vivo tumor targeting PDT treatment, the tumor-bearing mice have been randomly arranged into six different groups (n = 8, each group) to perform a series of intravenous injection of saline (group 1), PEG@CRUN (group 2) and FA-PEG@CRUN (group 3-6) respectively (3 mg in 100 µl saline). After 4 h of i.v. injection, laser treatment was performed on groups 1-6 by irradiation the tumor region with a continuous 808 nm or 660 nm laser. The effective exposure time for each mouse was 45 min (5 min interval after 5 min irradiation with total treatment time of 90 min) with power density at 0.4 W cm -2 (fluence 1080 J cm -2 ). The tumor in group 4 and 6 were covered by 2 mm thick pork tissue and then irradiated with 808 nm or 660 nm lights under the same condition 22,23 . The PDT treatment with laser irradiation was repeated every two days. A second and third dose injection of the PDT treatment described above was repeated 4 d and 8 d after the first dose injection, respectively. Tumor size was measured and calculated as described above.

3.7) In vivo fluorescence imaging in living mice:
For in vivo luminescence imaging of live animals, the stock solution of PEG@CRUN and FA-PEG@CRUN were prepared by dissolving the substrates in saline (3 mg in 100 µl saline). The tumor bearing mice was anaesthetized with isoflurane and imaged immediately after tail vein injection at the indicated time points of 0, 4, 8 and 24 h post-injection. The optical imaging of Ce6 molecules (λ em = 670 nm) loading at UCNs were collected using an IVIS Lumina II imaging system (Caliper Life Sciences, Roissy, France) with the excitation wavelength of 640 nm and analyzed with the living Image 2.11 software package. For quantification of measured signals, regions of interest were drawn over the tumor region with blue circle and final values were expressed by averaging the obtained signals from the all the mice scanned in each experiment (supplementary Fig. 23).

3.8) In vivo PA imaging of enzyme triggered cross-linking in tumor region:
For in vivo PA imaging of intravenous injection of FA-PEG@CRUN and PEG@CRUN (3 mg in 100 µL saline), the tumor bearing mice was anaesthetized with 3% isoflurane and ultrasound gel was applied on the surface of the mouse skin. The location of the probe in tumor region was monitored over a period of time using PA imaging. In order to obtain an image acquisition, several slices with a step size of about 0.3 mm spanning from the liver all