Synthesis, spectroscopic characterization, and mechanism studies
The Cy7Gal probe was prepared following a two-step synthetic procedure shown in Fig. 1a. First, 2,3,4,6-tetra-O-acetyl-α-D-galactopyranosyl bromide was reacted with 4-hydroxyisophthalaldehyde in anhydrous acetonitrile yielding compound 1. Then, a Knoevenagel condensation between 1 and 2,3,3-trimethyl-1-(4-sulfobutyl)indolium (2) yielded the Cy7Gal probe. Besides, the Cy7 fluorophore was synthesized by protecting the hydroxyl group of 4-hydroxyisophthalaldehyde with t-butyldimethylsilyl chloride followed by a Knoevenagel condensation with 2 and the subsequent deprotection of the hydroxyl group (Supplementary Information, Scheme S1). Cy7Gal and Cy7 were fully characterized by 1H-NMR, 13C-NMR, and HRMS (see Supplementary Information). PBS (pH 7) solutions of Cy7 (1.0 x 10− 5 M) show an intense broad emission band centered at ca. 660 nm (ΦCy7 = 0.43) when excited at 580 nm (Fig. 1c), whereas as a clear contrast, PBS (pH 7) solutions of Cy7Gal are poorly fluorescent at the same excitation wavelength (ΦCy7Gal = 0.0062) (Fig. 1c). Besides, the emission intensity of Cy7 fluorophore and Cy7Gal remains unchanged in the 5–10 pH range (Supplementary Information, Figure S1). The hydrolysis of Cy7Gal in PBS (pH 7) solutions in the presence of the β-galactosidase enzyme was studied by HPLC (Supplementary Information, Figure S2a). The obtained chromatograms show the progressive disappearance of the Cy7Gal peak with the subsequent appearance of Cy7 signal in the presence of the enzyme, whereas Cy7Gal remains stable in PBS in the absence of β-galactosidase.
Moreover, specificity and selectivity of the probe to β-galactosidase was demonstrated after incubation of Cy7Gal with different interfering species such as cations, anions, small peptides and enzymes (Supplementary Information, Figure S2b) for 0.5 h. Of all the species tested, only β-galactosidase induced a marked emission enhancement at ca. 660 nm due to the hydrolysis of Cy7Gal which generates the Cy7 fluorophore. Besides, a more marked emission enhancement was observed when Cy7Gal was incubated in the presence of both esterase and β-galactosidase enzymes. This greater enhancement, when compared to that observed when only β-galactosidase was present, is ascribed to the hydrolysis of the acetate moieties in Cy7Gal by esterase and then to the rupture of the O-glycosidic bond by β-galactosidase yielding free Cy7.
Cy7Gal monitors senescence induction in mouse mammary tumour cells in vitro
In order to validate the ability of Cy7Gal to monitor senescence induction in cells, we used the 4T1 mouse mammary tumour cell line, which is considered a murine model of triple negative-breast cancer cells.25,26 Cells were treated with palbociclib (5 µM) for two weeks. Palbociclib is a CDK4/6 inhibitor which has been reported to induce cell cycle arrest and senescence in 4T1 cells.27 Cellular senescence induction was corroborated by the increase in SA-β-Gal activity assay using the X-Gal test (Fig. 2a and 2d). Besides, western blot experiments demonstrated the overexpression of β-galactosidase in 4T1 senescent cells (Supplementary Information, Figure S3a). Treatment with palbociclib was also accompanied by an increase in cell size and in the number of intracellular vesicles, in agreement to morphological features described for senescent phenotypes.28 Subsequently, 4T1 control and senescent cells were treated with Cy7Gal (20 µM) and analyzed by confocal microscopy 2 h post-treatment. Both control (Fig. 2b) and palbociclib treated 4T1 cells (Fig. 2e) did not exhibit any noticeable fluorescence signal in the absence of the probe when excited at 580 nm. Similarly, control non-senescent 4T1 cells displayed poor fluorescence emission after exposure to 20 µM of Cy7Gal (Fig. 2c). In contrast, senescent 4T1 cells treated with Cy7Gal showed intense red emission (Fig. 2f) which is attributed to the hydrolysis of Cy7Gal into Cy7 by increased SA-β-Gal activity in senescent cells. Quantification of the fluorescence signal associated with Cy7 from confocal images was determined for all treatments. A significant increase in the fluorescence signal (ca. 3.2 fold) was observed in palbociclib-treated 4T1 cells treated with Cy7Gal when compared with control 4T1 cells (Fig. 2m).
To further confirm the β-galactosidase-catalysed cleavage of Cy7Gal, we found (Fig. 2g-2l) that a marked reduction in the emission intensity (ca. 60%) was obtained after pre-incubation of senescent 4T1 cells with D-galactose (a specific inhibitor of β-galactosidase, 5 mM) for 0.5 h and then treated with Cy7Gal, when compared to senescent cells administered only with the probe. Besides, the response of Cy7Gal probe was demonstrated to be dependent on the increased lysosomal β-galactosidase activity in senescent cells. For this purpose, two siRNAs were used to knock-down the expression of GLB1,19 the gene that encodes lysosomal β-galactosidase in 4T1 cells. As shown in the Supplementary Information (Figure S3b), hs.Ri.GLB1.13.3 siRNA efficiently downregulated the transcription of GLB1 at 48 h post-transfection and resulted in a significantly decreased number of SA-β-Gal-positive cells, when compared to a scrambled siRNA. Confocal microscopy images of control and senescent 4T1 treated with Cy7Gal showed a marked decrease in the emission in the red channel (ca. 54 %, Supplementary Information, Figure S3d) for cells transfected with hs.Ri.GLB1.13.3 siRNA, when compared to non-transfected cells or cells transfected with scrambled siRNA (Supplementary Information, Figure S3c). Finally, viability assays indicated that the probe was innocuous for both normal and senescent cells (Fig. 2o). Thus, Cy7Gal is an appropriate probe to monitor senescence induction in cell cultures.
Renal clearance of Cy7 fluorophore allows cell senescence burden evaluation in vivo in a 4T1 breast cancer model treated with senescence-inducing chemotherapy
Once assessed the activation of Cy7Gal in 4T1 senescent cells, the probe was validated in a chemotherapy-induced senescence triple-negative breast cancer mouse model. For tumour generation, 4T1 cells were injected subcutaneously into the left mammary fat pad of young female BALB/cByJ mice. Mice were subsequently treated for 7 days with palbociclib at concentrations of 10 mg/Kg, 50 mg/Kg or 100 mg/Kg by daily oral gavage in order to induce different degrees of senescence burden in the tumours. Tumour volume was measured every two days with a calliper. Figure 3a shows that tumours from mice treated with 10 mg/Kg of palbociclib grew similarly to the untreated ones, while tumours from mice treated with 50 and 100 mg/Kg of palbociclib displayed a reduction in their size. Immunohistochemical staining of the proliferation biomarker Ki67 in autopsy samples revealed that samples from mice treated with 50 or 100 mg/Kg of palbociclib exhibited a significantly lower proportion of Ki67 positive nuclei than those from untreated mice or mice treated with the lowest palbociclib dose (Figs. 3b and 3c). Moreover, X-Gal staining in cryosections of mice treated or not with palbociclib revealed the specific induction of senescence (SA-β-Gal activity) in the tumours but not in organs, such as liver or kidney (Supplementary Information, Figure S4). This is indicative of an enhancement in cellular senescence in tumours when palbociclib treatment increases.29
In order to test the ability of probe Cy7Gal for cellular senescence detection in vivo, BALB/cByJ mice bearing 4T1 tumours and treated with palbociclib at different concentrations (i.e. 10 to 50 and 100 mg/Kg) were anesthetized and the Cy7Gal probe was intraperitoneally (i.p.) administered (2.5 µmol). Ex vivo studies demonstrated that the Cy7Gal probe allows to detect cellular senescence in palbociclib treated tumours. In fact, ex vivo IVIS images from mice treated with Cy7Gal revealed a strong fluorescence signal in tumours from mice administered with 100 mg/Kg of palbociclib, while the fluorescent signal decreases as a function of the reduction in palbociclib dose (Fig. 3f). In addition, no significant fluorescent signal in the liver, lungs, heart, spleen, or kidney from mice injected with Cy7Gal was found and no auto-fluorescence was observed for palbociclib treated mice at 100 mg/Kg dose in any organ (Fig. 3g).
Interestingly, fluorescence analysis in vivo using an IVIS® spectrum 15 min after Cy7Gal administration, clearly showed fluorescence accumulation in the bladder suggesting a rapid renal clearance of the Cy7 fluorophore (Supplementary Information, Figure S5). Urine was, therefore, collected after mice recovered from anesthesia and analyzed by IVIS. The emission in urine was higher in mice treated with increasing amounts of palbociclib (Fig. 3d). The emission of Cy7 in the urine was also measured in a fluorimeter and the amount of the fluorophore determined from a calibration curve indicating a correlation between palbociclib doses and µmoles of Cy7 in urine (Fig. 3e).
To further study the renal clearance of the Cy7Gal probe, we synthesized and characterize (Supplementary Information, Scheme 2) an analogous molecule lacking the sulfonic groups (WOS-Cy7Gal) that make the Cy7Gal a more hydrophobic dye disfavouring its accumulation in bladder.30
In a comparative experiment Cy7Gal or WOS-Cy7Gal were i.p. administered (2.5 µmol) to BALB/cByJ mice bearing 4T1 tumours induced by palbociclib (100 mg/Kg). Both probes gave a strong positive signal in the tumours and, interestingly, we detected signal accumulated in the bladder, but not the kidney of animals injected with the Cy7Gal, but not the WOS-Cy7Gal probe (Supplementary Information, Figures S6a and S6b). The concentrations of Cy7 and WOS-Cy7 fluorophores were measured in plasma and in the excreted urine. In agreement with a renal clearance of the Cy7 fluorophore, we found a detectable concentration in the urine (2.34 µmol) while the plasma concentration was significantly lower (0.33 µmol) (Supplementary Information, Figure S6c). In contrast, concentrations of WOS-Cy7 fluorophore were basically undetectable in plasma (0.02 µmol) and urine (0.01 µmol). These results were also corroborated by measuring the average radiance intensity of Cy7 and WOS-Cy7 in urine (Supplementary Information, Figure S6d) and indicate the crucial role of the sulfonic acid moieties in the renal clearance of the Cy7 fluorophore.
In vivo cell senescence burden evaluation in naturally aged BALB/cByJ mice
Aging is a universal physiological state with a progressive functional decline, which is accompanied by the development of age-related diseases.31 Senescence has been associated with age-dependent organismal changes, and the progressive increase of senescent cells with time is known to contribute to the functional impairment of different organs.32 In fact, strong correlations have been reported between aging and certain phenotypes such as mitochondria dysfunction,33 epigenetic changes,34 and an increase in cellular senescence.35 In this scenario, tools to measuring easily the pace of aging are of interest.36 Regarding this matter, we tested the ability of probe Cy7Gal to evaluate the burden of senescence in healthy old and young individuals. For this purpose, 2 and 14 months old BALB/cJyB mice were i.p. injected with the Cy7Gal probe. IVIS images of the anesthetized mice 15 min post Cy7Gal injection revealed fluorescence accumulation in the bladder of 14 months old mice (Supplementary Information, Figure S7). Urine collected in an Eppendorf tube after recovery from anesthesia was analyzed by IVIS (Fig. 4a). A stronger fluorescent signal (ca. 5.4 fold) in urine from old mice treated with Cy7Gal was observed when compared with the fluorescence in the urine of young mice also administered with the probe. The amount in µmol of Cy7 in the urine was also calculated and a significantly large amount of Cy7 in urine (8.1 fold) from old animals was found compared with young individuals (Fig. 4b). Moreover, 2 and 14 months old control mice not treated with Cy7Gal showed negligible fluorescence in urine (Fig. 4a and 4b). A simple mass balance of the amount of Cy7Gal injected and that of Cy7 in urine allows calculating that, on average, 49% of injected Cy7Gal was excreted in urine as Cy7 in old mice whereas for young mice this was only 6.1% (Fig. 4b). These differences are in agreement with a larger burden of cellular senescence in old animals.37 Then, mice were sacrificed and the bladder, brain and lungs were studied by IVIS imaging (Fig. 4c (i), (ii) and (iii) respectively). Control mice not treated with Cy7Gal showed negligible fluorescence in these organs at any age. In contrast, quantification of the emission intensity in IVIS images revealed an increase of 5.2 in the bladder of aged mice. We also observed increases of 2.3 and 3-fold for brain and lungs respectively in aged mice when compared to young animals (Fig. 4d (i), (ii) and (iii) respectively). This observation is in agreement with increases in cell senescence incidence in lungs reported with aging.38,39 In the brain, SA-β-gal activity is not a specific marker of neuronal senescence, as many healthy neurons have large lysosomal compartments with increased levels of β-galactosidase.40 Interestingly, however, the results suggest the capacity of the probe to permeate the blood-brain barrier.
In vivo senescence burden evaluation in a senescence-accelerated mouse model
In addition to natural aging, we next decided to test the probe in a strain of accelerated senescence mice (SAM). Inbreeding of AKR/J mice and selection for the early appearance of features such as hair loss, skin coarseness, and short life span, led to the isolation of senescence-prone (P) and senescence-resistant (R) series of mice which were crossed separately to establish the inbred SAMP and SAMR strains. Relative to their genetic background-controls (SAMR1 mice), SAMP8 mice exhibit several traits that are known to occur during aging at earlier physiological ages and, therefore, constitute a suitable model to test aging phenotypes.41
The SAMP8 model has been widely used in aging research to study immune dysfunction,42 osteoporosis43 or brain atrophy.44 Because the phenotypic age-related differences between SAMP8 and SAMR1 mice begin to be evident after approximately 6 months of age,45–47 2 and 14 months old SAMR1 and SAMP8 mice were i.p. injected with the Cy7Gal probe. As in the models above, urine was collected after mouse recovery from anesthesia and analyzed by IVIS (Fig. 5a).
The amount of Cy7 in the urine was also determined from a calibration curve (Fig. 5b). Even though urine collection from SAMP8 and SAMR1 was more challenging than in the BALB/cByJ mice models describe above due to the general lower amount of urine collected, the results still clearly show a stronger fluorescent signal in the urine (due to a larger amount of Cy7) in old mice treated with the probe when compared with the emission in urine from young mice. Thus, 14-month-old SAMR1 mice excreted approximately 2.0-fold µmols of Cy7 fluorophore (after Cy7Gal hydrolysis) compared to young SAMR1 mice, while a 3.1-fold Cy7 amount increase in urine is observed for SAMP8 mice aged 14 months with respect to young animals. Moreover, a larger Cy7 quantity (1.7 fold) in urine is found for 14 months old SAMP8 when compared with 14 months old SAMR1 mice in agreement with the reported large burden of cell senescence in SAMP8 animals.
Cy7Gal is a sensitive and specific probe for senescence burden evaluation regardless of the murine model
Finally, to determine the accuracy of the Cy7Gal probe as a tool for evaluation of senescence we divided mice in those with a low (LS) and high (HS) burden of senescence regardless of the murine models used. Mice in the first group include 2 months old BALB/cJyB, SAMR1 and SAMP8 mice, and BALB/cByJ mice bearing 4T1 tumours (not treated with palbociclib), whereas the second group includes 14 months old BALB/cJyB, SAMR1 and SAMP8 mice, and BALB/cByJ mice bearing 4T1 tumours treated with 50 and 100 mg/Kg of palbociclib (Fig. 5c). The plot revealed a mean urinary signal increase of ca. 4.8 fold in the HS group relative to LS mice. Besides the rate of true positives and false positives (one-specificity) by receiver operating characteristic (ROC) curve shows that the probe discriminated well HS and LS groups with an area under the curve of 0.95 (Fig. 5d, P < 0.0001).