Real-time imaging of senescence in tumors with DNA damage

Detection of cellular senescence is important not only in the study of senescence in various biological systems, but also in various practical applications such as image-guided surgical removal of senescent cells, as well as the monitoring of drug-responsiveness during cancer therapies. Due to the lack of suitable imaging probes for senescence detection, particularly in living subjects, we have developed an activatable near-infrared (NIR) molecular probe with far-red excitation, NIR emission, and high “turn-on” ratio upon senescence-associated β-galactosidase (SABG) activation. We present here the first successful demonstration of NIR imaging of DNA damage-induced senescence both in vitro and in human tumor xenograft models.

We have developed an activatable NIR molecular probe with far-red excitation and NIR emission (708 nm), and a 130-fold enhanced fluorescence intensity upon β-gal activation. We present here the first demonstration of NIR imaging of DNA damage-induced senescence in xenograft human tumors in living mice.

Results and Discussion
Design, synthesis and activation of the NIR probe by recombinant β-gal. The fluorescence of the probe is designed to "turn-on" in the presence of senescence-associated β-gal (SABG) -the probe that remains non-fluorescent without SABG fluoresces when SABG is expressed (Fig. 1). A hemicyanine derivative ( Fig. 1) was selected as the fluorophore for its excellent stability, low toxicity, and fluorescence emission in NIR range upon activation 19 . The fluorophore has demonstrated its suitability for fluorescence imaging of various biomolecules, such as the detection of β-Lactamase in Staphylococcus aureus 20 , the detection of nitroxyl (HNO) 21 and palladium 22 in live cells, the detection of nitroreductase 23 , γ-glutamyl transpeptidase 24 and tyrosinase 25 in zebrafish. More recently, the hemicyanine NIR dyes have been used for in vivo detection of cysteine 26,27 , alkaline phosphatase in tumor models [28][29][30] , superoxide radical anion 31 , hydrogen sulphide 32 , hydrogen polysulfides 33 and γ-glutamyl transpeptidase 34 in mice models.
Therefore, the probe NIR-BG (NIR probe for β-gal) was efficiently synthesized by glycosylation of the hemicyanine dye and a protected galactosyl bromide, followed by a deprotection step (Figs S3-10). The fluorescence of the NIR-BG was initially at its background due to the reduced electron-donating strength of the oxygen atom upon glycosylation; upon cleavage of the galactose, the fluorescence of the hemicyanine restored (turned-on) by regaining its zwitterionic resonance form 19 .
We then evaluated the "turn-on" efficiency of the NIR-BG probe using recombinant β-gal from E. coli. The intensity of the fluorescence emission of the probe with and without the enzyme was compared (Figs 2a, S1). While the fluorescence of the probe in PBS buffer remained at baseline intensity, the reaction of the probe with β-gal produced a strikingly more intense fluorescence signal peaked at 708 nm (NIR range), with a fluorescence on/off ratio of 130-fold within 5 min (Fig. 2a). The fluorescent product had identical optical properties as the uncaged hemicyanine dye. With constant amount of β-gal, the fluorescence intensity is linearly proportional to the probe concentration, with a linear correlation coefficient of determination R 2 of 0.9988 (Fig. 2b). High-performance liquid chromatography (HPLC) analysis was conducted to confirm the enzymatic hydrolysis process (Fig. S2). Our probe itself eluted at 11.87 min; when the probe was exposed to β-gal, a new peak at 11.65 min appeared, which was attributed to the enzymatic hydrolysis product, based on the same retention time, UV-vis absorption, and mass spectrometry as the standard hydrolyzed/uncaged hemicyanine dye. This result confirmed that our probe was efficiently activated by β-gal. Also, as shown in the Fig. 2c, the fluorescence intensity dropped with the increase of the β-gal inhibitor (IPTG, isopropyl thio-β-D-galactoside) concentration, further confirming the probe activation is through galactose hydrolysis by β-gal, and our probe may work as a tool for β-gal inhibitor screening. The stability assay showed the probe is stable with 30 times of laser excitation or 48 hours incubation with β-gal (Fig. S11).
Kinetics and detection limit of β-gal by the NIR-BG probe. With excess amount of enzyme, the NIR-BG probe could be fully activated within 3 minutes (Fig. 3a,b). To quantify the kinetic parameters, time-dependent fluorescence intensity was measured in the presence of β-gal and probes at different concentrations (Fig. 3c). A good linear equation was obtained at various probe concentrations: 1 μM, R 2 = 0.9943; 2.5 μM, R 2 = 0.9939; 5 μM, R 2 = 0.9988; 10 μM, R 2 = 0.9994; 20 μM, R 2 = 0.9984. The reaction rate was found to increase with probe concentrations (Fig. 3d). The Lineweaver-Burke plot of the probe hydrolysis by β-gal was generated based on the slope of each line in Fig. 3C. The reaction rate was calculated using Lineweaver-Burk equation, where [E 0 ] is the concentration of β-gal. The kinetic parameters K M , K cat and K cat /K M, were calculated to be 106 μM, 3.3 s −1 , and 0.31 μM −1 •s −1 , respectively. The Michaelis constant (K M ) of our probe was much lower than that of commercial X-Gal. The value of K cat /K M of probe was 16-fold higher than that of X-Gal (K cat /K M = 8.29 × 10 −3 μM −1 •s −1 ), suggesting the high activation efficiency of our probe by β-gal.
In order to evaluate the detection limit of our NIR-BG probe for β-gal, the probe was incubated with β-gal at various concentrations (Fig. 4). Within 3 min, the probe fluorescence could be detected with a good linear

Detection of β-gal in LacZ(+) live cells.
In order to see whether our NIR-BG probe can be used for β-gal detection in live cells, we chose an engineered cell line that stably express LacZ(+), the gene that encodes β-gal. CT26.CL25, a mouse colon fibroblast carcinoma cell line that express β-gal, was therefore used in our initial live cell experiments. CT26.WT, the wild type cell line that does not express β-gal, was used as a control. Our probe did not appear toxic in neither cell culture after 24-hour incubation (Fig. S12). Cell samples after incubating with our NIR-BG probe were analyzed using flow cytometry ( Fig. 5a,b) and confocal microscopy ( Fig. 5c-f). Flow cytometry showed obvious fluorescence at 1 hour and 4 hours in CT26.CL25 cells incubated with our probe, whereas only background signal was detected in CT26.WT cells that received the same treatment ( Fig. 5a,b). Confocal fluorescence microscopy images showed consistent results. Fluorescence signal was obvious in samples with 1-hour probe treatment, and the intensity increased with time in CT26.CL25 cells ( Fig. 5c-e); while no fluorescence in the NIR channel was detected in the CT26.WT cells even after 4-hour incubation (Fig. 5f). The NIR signal in CT26.CL25 cells were highly colocalized with the β-gal expression (Fig. S13). Clearly, the fluorescence detected in CT26.CL25 cells using both flow cytometry and confocal microscopy was from our NIR-BG probe after hydrolysis of the galactose by β-gal, while no fluorescence was detected in CT26.WT due to the lack of β-gal. The β-gal expression was confirmed by colorimetric X-gal staining. Intensive blue stain was observed in CT26. CL25 cells, indicating high β-gal expression compared with the wild type cell line (Fig. S14A).

Imaging of senescent state in drug or radiation treated human cancer cells.
With the confidence of our probe being able to detect β-gal in live cells, we moved further to evaluate whether it could differentiate senescent cells and normal cells. Human cervical cancer cells (HeLa) and metastatic breast cancer cells (MCF7), were used in all four groups of cells evaluated. Senescence in HeLa and MCF7 cells were induced using camptothecin (CPT), an inhibitor of DNA topoisomerase I that could cause lethal DNA strand breaks 37 . Interestingly, when the cells were incubated with both CPT and cycloheximide (CHX), an experimental blocker of protein synthesis, cellular senescence could be alleviated 38 . Radiation therapy is another effective method for cancer treatment in clinic; it can also induce cell cycle arrest leading to senescence. All four groups of cells, with no drug treatment, with CPT treatment, with both CPT and CHX treatment, or with 10 Gy radiation treatment, were then incubated with our NIR-BG probe, followed by flow cytometry (Fig. 6). Importantly, CPT or radiation treated cells had significantly higher fluorescence compared to the untreated control cells; the higher fluorescence presumably resulted from the expressed senescence-associated β-gal (SABG), which turned on the fluorescence of the NIR-BG probe. Meanwhile, the fluorescence in cells treated with both CPT and CHX was significantly reduced comparing to the CPT-treated group, suggesting the suppressed senescence due to CHX treatment. Same trend was observed for both HeLa and MCF7 cells (Fig. 6). Since SABG is localized in the lysosome, we then performed confocal microscopy of the normal and senescent HeLa or MCF7 cells (Fig. 7). Fluorescence from the NIR-BG probe (NIR-BG channel) in CPT or radiation treated cells was strikingly higher than that of the control cells without treatment. More importantly, the red fluorescence from our probe overlapped mainly with the lysosomes (LAMP-1 channel), suggesting the high probability that our probe was activated in lysosomes, where SABG was expressed. This was further confirmed by co-staining the cells with anti-β-gal antibody (β-gal channel); our probe signal in senescent cells, induced either by chemotherapeutics or by radiation, was well colocalized with both lysosome and β-gal (Fig. 7). This feature was observed in both HeLa cell groups and MCF7 cell groups. These results further confirmed that the NIR-BG could image chemotherapy and radiation therapy induced senescence in different types of cancer cells. X-gal staining also verified the existence of SABG in cells after chemotherapy and radiation therapy (Fig. S14B).
Cell cycle inhibitors p21 and p16 are often expressed by senescent cells 1 . Therefore, we examined the cell lysates of all cell samples from the imaging studies using Western blot (Figs 8 and S15). Western blot indicated that tumor suppressors p16 and p21 both increased in HeLa cells treated with CPT or radiation compared with normal HeLa cells (Fig. 8a,b), confirming the cell samples we used were indeed senescent, and the induction of senescence by CPT or radiation was at least partially through p53 pathway, as p21 is typically induced directly by p53. In addition to HeLa cells, MCF7 cells also expressed higher levels of p16 and p21 upon CPT treatment (Fig. 8c,d). The Western blot results confirmed the senescent state of these drug or radiation treated cells used in the imaging experiments. Interestingly, β-gal levels in drug or radiation treated HeLa or MCF7 cells were all elevated compared with the control cells, further suggesting the activation of β-gal enzymatic activity during senescence (Fig. 8).
Real-time imaging of senescence in tumor models. Successful detection of cellular senescence in vitro allowed us to further examine the capability of NIR-BG to visualize senescence in tumors in living mice. It is important to note that unactivated NIR-BG's absorption peaks around 640 nm and emission peaks around 660 nm, while the activated probe NIR-BG has the maximal absorption and emission at 680 nm and 710 nm  (Fig. S1). The imaging instrument IVIS spectrum can take advantage of this major difference between the unactivated and activated probe, therefore we examined our animals using two different filter settings (Ex640 nm/Em680 nm for unactivated probe and Ex675 nm/Em720 nm for the activated probe). In a preliminary experiment, we used the genetically modified mice colon cancer cell line CT26 to determine whether NIR-BG could differentiate tumors with and without active β-gal (Fig. 9a). The LacZ(+) CT26.CL25 tumors showed significantly higher signals of NIR-BG than CT26.WT tumors at 1 h after probe injection using 675 nm excitation and 720 nm emission filter setting; while their difference was minimal when using 640 nm excitation and 680 nm emission setting. This result suggested the unactivated form of NIR-BG could distribute to tumor with little influence by the β-gal expression, and the probe could be activated in β-gal expressing tumors therefore was detected upon 675 nm excitation and 720 nm emission. We also confirmed β-gal expression in resected whole tumor and tumor slides (Fig. S17). Immunofluorescent staining of the resected tumor tissues also showed homogenous distribution of probe NIR-BG which overlapped with β-gal in the CT26.CL25, but not in the CT26.WT tumors (Fig. S18).
We then moved on to HeLa xenografted mice models to see whether probe NIR-BG could visualize chemotherapy-induced senescence in tumors. Tumor-bearing mice were first systemically treated with CPT or saline before the probe NIR-BG was injected. With the treatment of CPT, the tumor growth was significantly inhibited compared to the mice group receiving saline only (Fig. S16). We were delighted to find that NIR-BG behaved similarly in the drug-treated mice (with drug-induction triggered expression of β-gal) to the CT26 tumor models (with genetically expressed β-gal) (Fig. 9). HeLa tumors receiving either CPT treatment or saline showed similar signal of NIR-BG while using the 640 nm excitation and 680 nm emission filter setting, however, there was a significantly higher signal from the activated form of NIR-BG (observed using 675 nm excitation and 720 nm emission filter) in the HeLa tumors with CPT treatment (Fig. 9b). The result was also quantified in Fig. 9. Resected intact whole tumor staining, histochemical staining with X-gal (Fig. S17) and immunofluorescent staining of tumor tissue with antibodies against β-gal (Fig. S19) further confirmed the probe activation in senescent tumors.

Conclusion
The importance of detecting cellular senescence lies not only in biological studies of senescence in various biological systems, but in various practical applications such as image-guided surgical removal of senescent cell, as well as the monitoring of drug-responsiveness during cancer therapies. Due to the lack of suitable imaging probes for  senescence detection, particularly in living subjects, we have developed an activatable NIR-BG molecular probe with far-red excitation, NIR emission, and high "turn-on" ratio upon β-gal activation. We present here the first demonstration of NIR imaging of DNA-damage induced senescence in xenograft human tumor models. In the in vitro work, we observed fluorescence in cells with knocked-in LacZ as well as senescent cells induced by drug or radiation treatment. The fluorescence signal co-localized with lysosomes in senescent cells, suggesting the presence of SABG in lysosome, one of the key features in cellular senescence. Cell cycle inhibitors p16 and p21 had elevated expression in cells with enhanced fluorescence signal, confirming the induction of cellular senescence in the cell studies. We finally examined our probe in mice bearing either LacZ(+) tumor or drug-treated xenograft human tumor models, and the probe NIR-BG was able to detect both genetically expressed β-gal or drug-induced SABG, suggesting the potential applicability of using NIR-BG to evaluate the efficacy of cancer therapeutics.

Materials and Methods
Materials. Our probe was prepared following a literature procedure (also see Supporting Information) 23,39 .
Induction of cellular senescence. HeLa and MCF7 cells were seeded in a 24-well plate at a density of 2 × 10 3 cells/well. Then cells were cultured in the presence of 7.5 nM camptothecin (CPT) or CPT together with 0.15 μM cycloheximide (CHX) for a week. And for radiation therapy induced cellular senescence, cells were irradiated with 10 Gy X-ray (2 Gy/min) and then were cultured for another 48 hours.
After washing, all cover slides were mounted by ProLong Gold Antifade Mountant (Catalog #P36934, Invitrogen, USA), and images were acquired using a confocal laser scanning microscopy (Leica TCS SP8, Germany).
X-gal staining. Cells were seeded in a 24-well plate coated with cover slides at a density of 3 × 10 4 cells/well. Western blot analysis. Proteins from both treated and untreated cells were extracted with radioimmunoprecipitation assay (RIPA) buffer. Total protein was quantifed using the Pierce ™ BCA (bicinchoninic acid) Protein Assay Kit (ThermoFisher Scientific). Samples (25 μg) were separated on 10% SDS-PAGE (sodium dodecyl sulfate polyacrylamide gel electrophoresis) gels and transferred to PVDF (polyvinylidene fluoride) membranes. After being incubated in 5% skim milk, the PVDF membranes were incubated with the primary antibody at 4 °C overnight and then with the secondary antibody for 1 hour at room temperature. Immunodetection was performed using enhanced chemiluminescence (ECL), and the protein bands were quantified using ImageJ software. Relative protein levels were calculated by normalization to β-actin.
In vivo animal imaging. In vivo imaging was performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of the University of New Mexico following the National Institutes of Health guidelines for animal care. 2 × 10 6 HeLa cells or 5 × 10 4 CT26.WT or CT26.CL25 cells were injected subcutaneously to athymic female nude mice (Harland Laboratories) to establish tumor models. For the CT26. WT and CT26.CL25 tumor models, NIR-BG (10 nmol) was injected through tail vein when tumor size reached 150 mm 3 . The animals were then subjected to fluorescence imaging at 1 h post injection. For the HeLa xenografted mice, when tumor size reached 100 mm 3 , one group of the mice (5 mice per group) were treated with CPT (2 mg/ kg every two days for 4 times, total dosage at 8 mg/kg) while the control group (5 mice per group) was treated with the same volume of saline. Two days after the last treatment, mice were injected with the NIR-BG probe (10 nmol) for fluorescence imaging at 1 h post injection. Animal imaging was performed using an IVIS spectrum (PerkinElmer, USA) optical imaging system with two filter settings (Ex640 nm/Em680 nm for unactivated probe and Ex750 nm/Em720 nm for activated probe). After live animal imaging, the mice were euthanized and tumors were resected for immunohistochemistry staining and imaging.
Immunohistochemistry staining. The tumors were immersed in optimal cutting temperature (OCT) compound and were frozen on dry ice immediately after resection. The tissues were then sectioned into 4 μm slides and were fixed with 4% paraformaldehyde for 5 min. The slides were then counter-stained with anti-β-gal antibody (1:250) and anti-LAMP1-lysosome antibody (1:250) for immunofluorescent histology imaging and counter-stained with eosin and X-gal (for β-gal).

Statistical analysis.
Values are reported as the mean ± standard deviation unless otherwise noted. Student's t-test and two-way analysis of variance (ANOVA) were used to determine the statistical significance with probability values less than 0.05. All statistical calculations were performed using Prism 7.0 (GraphPad Software).