Introduction

All organisms are predisposed to aging, which at the biological level is defined as the accumulation of molecular and cellular damage over time. The major culprits driving this phenomenon include impairment of antioxidant defense systems, as well as the overproduction of reactive oxygen species (ROS)1. Beyond canonical sources of oxidative stress, recent studies implicate that the dysregulation of redox-active metal ion (i.e., copper (Cu)) homeostasis may also be an important source of aberrant ROS2,3. However, tools to accurately study Cu in vivo remain limited, prompting the development of chemical probes. Under physiological conditions, the majority of cellular Cu binds tightly within the active sites of metalloproteins or associates with a protein-based exchangeable pool. When the capacity of this protein pool is exceeded, GSH, despite its lower binding affinity (KD = 9.13 × 10−12 M), interacts with Cu4,5,6. The latter is referred to as the ‘labile Cu pool’ owing to its relatively weak binding interactions within the cell7,8. However, these properties may be compromised as GSH readily oxidizes to yield the inert oxidized glutathione (GSSG) form.

Conventional approaches to measure analytes such as Cu involve quantification via inductively coupled plasma mass spectrometry (ICP-MS). However, in this technique the sample is permanently destroyed during analysis and therefore, crucial spatiotemporal information is lost. Moreover, ICP-MS is further limited by its inability to distinguish the labile Cu pool from the total Cu content present, as well as Cu(I) from the +2-oxidation state. Indeed, while various reports demonstrate total hepatic Cu does not change significantly as a function of aging, it is impossible to determine if this is true with regards to weakly bound Cu9,10,11. To overcome these limitations, we develop activity-based sensing (ABS) probes, which are powerful chemical biology tools tailored to visualize a given analyte’s chemical reactivity with minimal off-target interference from related biomolecules12,13. We accomplish this by matching chemical detection to the unique reactivity profile of a target analyte. Moreover, when ABS is coupled to photoacoustic (PA) imaging, a modality that involves the conversion of light into an ultrasonic signal, molecular features that would otherwise be elusive become visible. PA imaging is particularly advantageous because sound waves can travel with minimal perturbation through tissue, enabling the acquisition of high-resolution, three-dimensional images14.

Using the design principles of ABS, we recently developed PACu-1, a Cu-responsive PA probe featuring a tetradentate TPA (tris-(2-Pyridylmethyl)amine) trigger capable of detecting Cu(I) with exquisite oxidation state and metal ion selectivity15. In this instance, Cu activity refers to Cu(I) capable of mediating redox chemistry. Unfortunately, preliminary studies reveal that this first-generation molecule lacks the requisite sensitivity to visualize differential labile hepatic Cu activity in a murine model of aging.

Herein, we present the development of next-generation Labile Cu Probes (LCPs) with an optimized pentadentate Cu-responsive trigger, which enhances both sensitivity and selectivity for in vivo imaging. The enhanced performance resulting from adding a fifth coordinating atom is instrumental in enabling the discovery of an age-dependent increase of labile hepatic Cu activity. Guided by these findings, we design two longitudinal animal studies to examine the impact of restoring Cu homeostasis. Our results indicate that in response to treatment of aged mice with ATN-224, a selective Cu chelator, ALDH1A1 activity increases. Owing to its protective role in the liver, this suggests there is a possible hepatocyte-sparing effect.

Results

Development of pentadentate Cu-Responsive Triggers

As previously noted, our initial attempts to use PACu-1 to detect labile Cu activity in aging liver tissue were not successful, indicated by a lack of age-dependent variation in the PA signals. In retrospect, this outcome was anticipated considering PACu-1 was tailored for a Wilson’s disease model, which exhibits Cu levels up to 20 times greater than those in healthy animals16. Aging, unlike Wilson’s disease, does not involve mutation of the Cu export machinery, and therefore, any change due to aging is expected to be more subtle.

Our strategy to enhance the sensitivity of PACu-1 involved revisiting its structural design, particularly the Cu-binding component which is a tetradentate TPA trigger17,18,19 appended to a near infrared absorbing aza-BODIPY dye20 (Fig. 1a). Initial binding of Cu(I) to TPA precedes the ensuing oxidative cleavage event that is responsible for separating the ‘spent’ trigger from the latent imaging agent to afford a ‘turn-on’ PA response. By adding an extra binding site to this trigger, we aimed to improve the probe’s affinity for Cu and, consequently its sensitivity to detect subtle changes in Cu activity associated with aging.

Fig. 1: Development and characterization of NPy4 trigger.
figure 1

a Chemical structure of PACu-1. b Chemical structures of LCP-Green-1 to 7. c Fluorescence fold turn-on after treating LCP-Green-1 to 7 (5 μM) with 20 eq of Cu(I) or Fe(II) in the presence of 2 mM GSH for 60 min (n = 3 independent replicates). d Fluorescence fold turn-on after treating LCP-Green-1 and LCP-Green-4 (5 μM) with 0.05, 0.1, 0.2, 0.5, 1, 2, or 5 eq of Cu(I) in the presence of 2 mM GSH (n = 3 independent replicates). e Fluorescence fold turn-on after treating LCP-Green-1 and LCP-Green-4 (5 μM) with 20 eq of Cu(I) in the presence of 2, 5 or 10 mM GSH (n = 3 independent replicates). f Proposed mechanism of Cu(I)-mediated activation of the NPy4 trigger. g Chemical structure of LCP-PA. h Absorption spectrum of LCP-PA (blue line, λmax = 680 nm) and turned over product (red line, λmax = 767 nm). i PA signal of FEP (fluorinated ethylene-propylene) tubes filled with 50 μM LCP-PA (left) or the turned over product (right) in 1:1 v/v HEPES:DMSO in a tissue-mimicking phantom. Excitation = 780 nm. j Ratiometric PA turn-on of LCP-PA (5 μM) after treating with 20 eq of various metal ions in the presence of 2 mM GSH. (n = 3 independent replicates). All assays were performed in 50 mM HEPES buffer (pH = 7.4, 1:1 v/v HEPES:DMF). All data is reported as the mean ± standard deviation. Error bars = SD. Source data are provided as a Source Data file.

Full size image

To this end, we designed six chelators, which together with the parent TPA trigger, afforded Labile Copper Probe (LCP) Green-1 to 7 (Fig. 1b). We chose to initially modify the Tokyo Green fluorescent dye21, as opposed to the PA-active aza-BODIPY scaffold in order to facilitate in vitro assessment of trigger performance. First, we obtained LCP-Green-1 by equipping Tokyo Green with the original TPA trigger. In addition, we appended a hydroxymethyl group at the C-6 position of one pyridine ring or to the methylene linker to yield LCP-Green-2 and LCP-Green-3, respectively. A symmetrical bis-pyridine arm was obtained by replacing the hydroxymethyl moiety with a fourth pyridine group to afford LCP-Green-4. We replaced the unmodified pyridine arm of the LCP-Green-4 trigger with hydroxyethyl or 2-methoxybenzyl substituents to give LCP-Green-5 and LCP-Green-6, respectively. Finally, we substituted the bis-pyridine arm of LCP-Green-6 with a 6-hydroxymethylpyridine unit to obtain LCP-Green-7.

Evaluation of LCP-Greens and Synthesis of LCP-PA

To identify the most suitable trigger for our study, we first examined the responsiveness of each LCP-Green to Cu(I), as well as potential cross-reactivity with Fe(II). Previous work has demonstrated that it is possible to shift the reactivity profile of TPA away from Cu to other metal ions (e.g., Fe) by altering the ligand set22,23. Because the detection of hepatic Fe activity can confound our results24, it was critical for the chosen trigger to exhibit the highest Cu:Fe selectivity ratio. It is known that total Fe is approximately  one magnitude higher than total Cu in livers. We used this ratio as a guideline in designing our competition assays to assess potential interference from Fe(II) when detecting Cu(I)25,26. First, we incubated each probe with 20 equivalents of Cu(I) or Fe(II) in the presence of GSH (2 mM) for one hour. The ‘turn-on’ responses were assessed and the corresponding Cu:Fe ratios were calculated and summarized in Supplementary Fig. 9. Besides the ligand featuring four pyridine arms (NPy4) used in the construction of LCP-Green-4, no other triggers outperformed TPA (Fig. 1c), where the normalized Cu:Fe ratios for LCP-Green-1 and LCP-Green-4 were 5.2 ± 0.2 and 8.6 ± 0.2, respectively. This result demonstrates the inclusion of an additional pyridine ring enhances the preference for Cu by 1.64-fold. Moreover, we found the enhancement of fluorescence to be largest for LCP-Green-4 within the series ((38.2 ± 0.9)-fold cf. (14.8 ± 0.4)-fold for LCP-Green-1). To compare their responsiveness at sub-stoichiometric levels of Cu, we decreased the number of equivalents from 5 to 0.05 (Fig. 1d). While the change in fluorescence intensity of LCP-Green-1 was negligible when Cu(I) was below 0.2 equivalents, LCP-Green-4 showed a significant change in signal even with as little as 0.05 equivalents. Lastly, we increased the concentration of GSH present from 2 mM to 10 mM, which is up to levels typically found in the liver (Fig. 1e). We postulate that if the initial binding between Cu(I) and LCP-Green-4 is tighter than to LCP-Green-1, a smaller decrease in the fold turn-on would result when GSH is at 10 mM. Indeed, the fluorescence of LCP-Green-4 decreased by only 1.8-fold, whereas the change for LCP-Green-1 was 4.3-fold. Of note, attempts to directly measure binding affinities were unsuccessful because even residual amounts of oxygen present were sufficient to drive probe activation. The proposed mechanism is shown in Fig. 1f15,17,27,28,29. Full characterization of LCP-Green-4 can be found in Fig. S10.

After having identified the best performing LCP-Green, we installed the NPy4 trigger onto our aza-BODIPY to afford LCP-PA (Fig. 1g). LCP-PA absorbs maximally at 680 nm (ε = 2.7 × 104 M−1cm−1) but this shifts to 767 nm (ε = 3.7  × 104 M−1cm−1) upon reaction with Cu (Fig. 1h, i). Beyond testing for potential interference from Fe(II), we also subjected LCP-PA to all biologically relevant metal ions (Fig. 1j). Under no circumstance did we observe significant off-target activation. Likewise, the presence of these metal ions did not impact Cu-mediated probe activation (Fig. S11). Additional characterization of LCP-PA can be found in Supplementary Fig. 10.

Detection of labile Cu activity in live cells and mice

Next, we conducted a series of confocal imaging experiments on HepG2 liver cells to assess the efficacy of our developed NPy4 trigger compared to the existing TPA trigger, particularly within environments rich in endogenous chelators (Fig. 2a). To further this comparison, we pretreated two additional groups of cells with bathocuproine disulfonate (BCS), a strong Cu chelator with a dissociation constant of 1 × 10−20 M30, before applying each of the LCP-Green congeners (Fig. 2b). We hypothesize there will be a decrease in fluorescence intensity only if a probe is activated by labile Cu activity. Upon analysis, the fluorescence intensity in cells treated with BCS and LCP-Green-4 showed a 1.21-fold decrease, while the LCP-Green-1 treated cells displayed no notable change. In addition, when we employed deferoxamine (DFO), a selective Fe(II) chelator, no change in fluorescence of LCP-Green-4 was observed compared to vehicle controls. Of note, DFO has been reported to reduce Cu(II) in vitro, however it did not lead to a change in LCP-Green-4 signal since most labile Cu was already reduced by the reductive intracellular environment31. Likewise, supplementation with iron prior to probe treatment did not alter the fluorescence signal intensity (Figs. S21 & S22). Collectively, these results indicate that LCP-Green-4 is only activated by endogenous Cu activity.

Fig. 2: Molecular imaging of NPy4-based imaging agents.
figure 2

a Representative fluorescence images of HepG2 cells pretreated with a vehicle or BCS (300 μM for 24 h) prior staining with LCP-Green-1 or LCP-Green-4 (5 μM). Scale bar = 40 μm. b Quantified data for panel a (n = 15 - 23 biological replicates). c Cross-section schematic to reference the positioning of the liver. d Representative cross-sectional photoacoustic images of the liver. Spectral unmixing was applied to isolate the signal from the aza-BODIPY product after LCP-PA activation. Mice were pretreated with vehicle or CuCl2. Scale bar = 5 mm. White arrows indicate the black hair which was excluded from signal quantification. Dotted outline indicates the general ROI of the liver used for signal quantification. e Quantified data for panel (d). (n = 3 biological replicates). All data is reported as the mean ± standard deviation. Error bars = SD. Statistical analysis was performed using a two-tailed t-test. Source data are provided as a Source Data file.

Full size image

With these promising results, we turned our attention to in vivo testing of LCP-PA in C57BL/6 J mice. Our initial goal was to examine the biodistribution of the probe after systemic administration, as its effective localization to the liver is pivotal for our studies. Whole-body PA imaging revealed LCP-PA predominantly localized to the liver. We then captured an in vitro PA spectrum of the probe after it has reacted with its target, which is necessary for in vivo spectral unmixing (Fig. S12). This step is crucial to account for potential interference from endogenous PA-active absorbers in the body like oxyhemoglobin and deoxyhemoglobin32, as well as account for non-uniform signals that can result from natural variations in tissue composition, blood flow, and probe distribution. A liver cross-section schematic is provided for anatomical reference (Fig. 2c). Consequently, the residual signal in the liver, highlighted in green, can be confidently assigned to LCP-PA activated by hepatic Cu (Fig. 2d). A separate group of mice received an intraperitoneal injection of a CuCl2 solution (5 mg/kg) prior to undergoing PA imaging. The analysis of these images showed that the signal for the Cu-treated group was 1.41-fold higher than in the control group (Fig. 2e). In some images, black hair could be detected (indicated by white arrows). These signals were localized to the surface of the body and were excluded from signal quantification to avoid confounding the data. Collectively, these experiments confirm that both the fluorescent and PA properties of the LCP probes can reliably detect basal and elevated activities of Cu in HepG2 cells and C57BL/6 J mice.

LCP-PA imaging reveals age-dependent increase in labile hepatic Cu activity

Having validated LCP-PA in vivo, we focused our subsequent efforts on testing the hypothesis that labile hepatic Cu activity changes as a function of age. We performed PA imaging using C57BL/6 mice across four age groups: young (~7 weeks old), adult (~13 weeks old), middle-aged (~42 weeks old), and elderly (~82 weeks old) (Fig. 3a)33. We chose not to study mice older than 82 weeks, despite their median lifespan of 108 to 116 weeks, due to the well-documented risk that anesthesia necessary for PA imaging can substantially increase the chance of mortality in these animals34. To our delight, we discovered the activation of LCP-PA was indeed lowest in the young group and highest in the elderly group. The results in ascending order are 1.06-fold, 1.14-fold, 1.29-fold, and 1.50-fold for the young, adult, middle-aged, and elderly mice, respectively (Fig. 3b, c).

Fig. 3: in vivo imaging of hepatic copper and evaluation of cellular copper activity relative to GSH level.
figure 3

a Table summarizing the age of mice in weeks and the corresponding age of humans in years for the four age groups utilized in our study. b Representative cross-sectional photoacoustic images of the liver from young, adult, middle-aged and old mice. Spectral unmixing was applied to isolate the signal from aza-BODIPY after activation of LCP-PA. Signal outside of this ROI is from black hair and was not quantified. Scale bar = 5 mm. c Quantified data for panel a (n = 4 biological replicates). d Representative fluorescence images of HepG2 cells pretreated with vehicle, GSH ethyl ester (5 mM for 8 h) or NEM (500 μM for 8 h) prior to staining with LCP-Green-4 (5 μM). Scale bar = 40 μm. e Quantified data for panel (d) (n = 20 biological replicates). All data is reported as the mean ± standard deviation. Error bars = SD. Statistical analysis was performed using a two-tailed t-test (no adjustments were made for multiple comparisons). Source data are provided as a Source Data file.

Full size image

Subsequently, we investigated the potential relationship between labile hepatic Cu and GSH activity. We used LCP-Green-4 in confocal imaging experiments with HepG2 cells, which were first treated with a PBS vehicle control to establish baseline fluorescence levels. Additional sets of HepG2 cells were pre-treated with GSH ethyl ester (5 mM) or NEM, a broad-spectrum thiol scavenger (500 μM) (Fig. 3d). The GSH ethyl ester treatment led to a 0.4-fold reduction in fluorescence compared to the vehicle, indicating decreased probe activation. In contrast, NEM treatment resulted in a small but significant 10% increase in fluorescence, suggesting heightened probe activation (Fig. 3e).

In Vivo PA Imaging Shows GSH Activity Decreases as a Function of Age

To measure GSH activity in the murine model of aging we employed above, we relied on the use of PACDx, a GSH-responsive PA imaging probe developed in-house by our group35. This probe features a custom 4-nitro-2-fluorobenzenesulfonyl trigger installed onto a hemicyanine dye platform. Upon reaction with GSH via SNAr chemistry, the dye product is liberated (Fig. 4a), which is accompanied by an increase in both the PA signal, as well as fluorescence (Fig. 4b). Prior to in vivo PA imaging, we tested its performance in HepG2 cells. In comparison to controls pretreated with a vehicle, HepG2 cells that were incubated with GSH ethyl ester (5 mM) for two hours prior to staining with PACDx were 25% more fluorescent (Fig. 4c, d). Additionally, we treated a separate set of cells with BSO, an inhibitor of γ-glutamylcysteine synthetase, to diminish cellular GSH levels. Relative to non-treated cells, the extent of PACDx activation was indeed attenuated (Fig. 4c, d).

Fig. 4: Measurement of GSH activity via PACDx imaging.
figure 4

a Reaction mechanism of GSH-mediated activation of PACDx to afford the O-HD product. b Fluorescence spectra of PACDx (in black) and O-HD (in red). c Representative fluorescence images of HepG2 cells pre-treated with a vehicle (water), GSH ethyl ester (5 mM for 8 h), or BSO (500 μM for 24 h) prior to staining with PACDx (2 μM). Scale bar = 100 μm. d Quantified data for panel c (n = 20 biological replicates). e Representative cross-sectional photoacoustic images of the liver. Spectral unmixing was applied to isolate the signal from O-HD after activation of PACDx. Scale bar = 3 mm. f Quantified data for panel e (n = 4 biological replicates). All data is reported as the mean ± standard deviation. Error bars = SD. Statistical analysis was performed using a two-tailed t-test (no adjustments were made for multiple comparisons). Source data are provided as a Source Data file.

Full size image

Next, we performed in vivo PA imaging using C57BL/6 mice belonging to the same four age groups as described above (Fig. 4e). Interestingly, we discovered an inverse correlation to Cu activity where the activation of PACDx was highest in the young group and lowest in the elderly group, indicating a decrease of GSH activity during aging. The results in descending order are 2.41-fold, 2.16-fold, 1.86-fold, and 1.08-fold for the young, adult, middle-aged, and elderly mice, respectively (Fig. 4f).

Altered Cu and GSH activity attenuates ALDH1A1 activity

One of the potential adverse effects of elevated labile hepatic Cu activity is the overproduction of ROS via Fenton-like chemistry. Likewise, the observed age-dependent decrease in hepatic GSH activity could also contribute to oxidative stress because it is involved in antioxidant systems where GSH is used as a substrate (e.g., GSH peroxidase) and as a ROS scavenger. Because Cu-induced cytotoxicity is well-documented to be linked with ROS generation, we focused on examining the effect of decreasing GSH activity on ROS production in liver cells. To this end, we incubated HepG2 cells with BSO, to disrupt GSH biosynthesis, and then stained cells with dichlorodihydrofluorescein diacetate (DCFH-DA), a general ROS sensor (Fig. 5a). Compared to control cells, the fluorescence of the BSO-treated cells was 5.3-fold higher which indicates greater extents of ROS (Fig. 5b, c).

Fig. 5: Measurements of ROS production and ALDH1a1 activity and protein expression levels.
figure 5

a Schematic showing ROS-mediated activation of DCFH2-DA. b) Representative fluorescence images of HepG2 cells pretreated with a vehicle (water) or BSO (500 μM) for 24 h prior to staining with DCFH2-DA (10 μM). Scale bar = 40 μm. c Quantified data for panel (b) (n = 20 biological replicates). d Schematic showing ALDH1A1-catalyzed activation of AlDeSense. e Representative fluorescence images of HepG2 cells pretreated with a vehicle (water) or CuCl2 (100 μM) for 24 h prior to staining with AlDeSense (2 μM). Scale bar = 40 μm. f Quantified data for panel e (n = 20 biological replicates). g Schematic to show the procedure for mass spectrometric analysis of liver samples from young and elderly mice. h Quantified data from mass spectrometric analysis of ALDH isoforms of pooled samples from young and old mice groups. Percentage change refers to change of protein expression in elderly mice compare to that in young mice. All data is reported as the mean ± standard deviation. Error bars = SD. Statistical analysis was performed using a two-tailed t-test. Source data are provided as a Source Data file.

Full size image

An increase in ROS is intriguing because low cellular levels are important to the general health of hepatocytes, such as hepatic stem cells (SCs) that become quiescence lose their self-renewal capacity36. In the context of SCs, it has been proposed that a decrease in GSH activity is linked to an increase of SC proliferation and differentiation, senescence, and occurrence of apoptosis37. To examine this in liver cells, we employed AlDeSense, which is a fluorogenic ABS probe we developed to visualize aldehyde dehydrogenase 1a1 (ALDH1A1) activity (Fig. 5d)38,39. Beyond its crucial protective role, this enzyme is a universal marker of SCs owing to its role in mediating proliferation and differentiation via the retinoic acid signaling pathway40. Thus, a decrease of ALDH1A1 activity can be used as an indicator of less cellular protection and in the case of SCs, a loss of stemness. To examine the effect of Cu on ALDH1A1 activity, we supplemented HepG2 cells with Cu, stained with AlDeSense, and then performed confocal imaging (Fig. 5e). We observed a significant decrease in fluorescence when comparing Cu-treated cells to the vehicle controls (Fig. 5f).

Further, we conducted a mass spectrometric analysis where liver samples from groups of young (n = 3) and elderly (n = 3) mice were pooled to assess in vivo ALDH1A1 expression (Fig. 5g). Results from this preliminary experiment revealed that ALDH1A1 levels were 79% lower in the elderly mice sample, suggesting a loss of hSCs (Fig. 5h). In contrast, other ALDH isoforms either remained stable or increased, which are employed by the cell to detoxify aldehydes produced by lipid peroxidation due to elevated ROS levels41.

Chelation therapy restores hepatic Cu homeostasis in middle-aged mice

Because elevated labile hepatic Cu activity and the loss of ALDH1A1 activity may be coupled, an intriguing proposition is whether restoring Cu homeostasis can increase the activity of this enzyme (Fig. 6a). To answer this, we designed a longitudinal study where middle-aged mice were subjected to metal ion chelation therapy using ATN-224, a high affinity Cu chelator (Fig. 6b). This specific chelator was selected for this experiment because it does not bind to Fe and is safe for use in humans as a potential chemotherapeutic for Wilson’s disease treatment42,43,44. Over a period of 60 days, animals received a daily dose of ATN-224 (0.5 mg/kg) or a vehicle control via oral gavage. The effect of chelation on labile hepatic Cu activity was monitored with LCP-PA via PA imaging. Prior to commencement of treatment, the average PA intensity between the two groups were indistinguishable (1.94 ± 0.37). However, the signal of the ATN-224-treated mice decreased to 1.24 ± 0.27 by day 20 and to 1.13 ± 0.08 by day 40. No further changes were observed beyond this point (Fig. 6c and Fig. S13). This PA imaging experiment indicates that labile hepatic Cu activity has returned to levels comparable to young mice. In contrast, the activation of LCP-PA became more pronounced throughout the study for the control group, increasing to 1.99 ± 0.19 by day 20, 2.26 ± 0.37 by day 40, and 2.81 ± 0.28 by day 60 (Fig. 6d). At the completion of this study, all animals were sacrificed, and their livers were collected for mass spectrometric analysis aimed at quantifying the effects of Cu chelation on ALDH1A1. While the ALDH1A1 levels were on average 1.29-fold higher in the group treated with the chelator compared to the control, the variability in the data prevented these results from reaching statistical significance (Fig. 6e).

Fig. 6: Schematic of copper chelation therapy.
figure 6

a Schematic showing copper induced hepatic stem cell depletion. b Schematic showing the first copper chelation therapy experiment. c Hepatic PA signal intensity for mice treated with vehicle in the first copper chelation therapy experiment. Mice are imaged at day 0, 20 40 and 60 with LCP-PA (n = 3 biological replicates). d Hepatic PA signal intensity for mice treated with ATN-224 in the first copper chelation therapy experiment. Mice are imaged at day 0, 20 40 and 60 with LCP-PA (n = 3 biological replicates). e Normalized ALDH1A1 levels from the mass spectrometric analysis for the mice in the first copper chelation therapy study (n = 3 biological replicates). f Schematic showing the second copper chelation therapy experiment. g Normalized ALDH1A1 level from the mass spectrometric analysis for the mice in the second copper chelation study (n = 5 biological replicates for vehicle and n = 4 biological replicates for ATN-224). h Immunohistostaining of ALDH1A1 in mouse liver sample (ALDH1A1 is stained brown). Scale bar = 100 μm. All data is reported as the mean ± standard deviation. Error bars = SD. Statistical analysis was performed using a two-tailed Mann Whitney U test. Source data are provided as a Source Data file.

Full size image

Restoring Hepatic Cu Activity Increases ALDH1A1 Activity

Based on these results, we designed a second longitudinal study with several key modifications (Fig. 6f). First, the treatment duration was extended from 60 days, as in the initial study, to 140 days to explore whether a longer period is necessary for restoration of ALDH1A1 activity. Of note, mice received ATN-224 daily for the first 20 days and subsequently every three days until the end of the experiment. This regimen was designed to ensure that the total number of ATN-224 treatments was not a confounding variable. Second, the application of LCP-PA to monitor labile hepatic Cu activity was omitted. Lastly, in addition to harvesting liver tissue for mass spectrometric analysis of ALDH1A1 expression as before, we also propose to perform complimentary immunohistochemical (IHC) staining for ALDH1A1. Consistent with the results from our first study, mass spectrometric analysis revealed increased ALDH1A1 levels in animals subjected to Cu chelation therapy, with the magnitude of change between treated animals and controls increasing by approximately 35% (p = 0.063) (Fig. 6g). The IHC staining confirmed a more pronounced expression of the ALDH1A1 protein in the ATN-224-treated samples (Fig. 6h and Fig. S14). Furthermore, IHC staining for 8-OHdG, a biomarker of DNA oxidative damage, indicated that ATN-224 treatment significantly reduced oxidative stress in the mice’s liver (Fig. S15). Finally, analysis of liver markers in the blood confirmed that this dosing regimen did not adversely impact the function of the liver (Fig. S16).

Collectively, these results demonstrate that Cu chelation therapy can restore labile hepatic Cu activity, leading to a decrease in ROS production, and consequently influencing ALDH1A1 activity. Beyond its role in stemness, ALDH1A1 has been shown to promote hepatocyte proliferation and reduce hepatic necrosis, oxidative stress, ECM remodeling, and inflammation during drug-induced liver damage45. Therefore, interventions that increase ALDH1A1 activity may represent a strategy to reverse the effects of aging on liver function.

Discussion

The liver is arguably our most vital organ, safeguarding against harmful metabolic byproducts and xenobiotics. It is renowned for its ability to regenerate, often completely recovering after a significant loss of tissue46,47, unless compromised by conditions such as liver cancer or aging48. As we age, our susceptibility to chemical exposure increases, potentially due to the gradual loss of hepatocyte viability. For instance, hepatic SCs are particularly sensitive to oxidative damage49. Recent landmark studies have demonstrated that redox active metal ions such as Cu, which were once thought to be inert static cofactors buried within metalloproteins, can be mobilized to mediate a variety of cellular functions50,51,52,53,54,55. However, when Cu homeostasis, which includes uptake, efflux, chelation, and transport, becomes dysregulated, this metal ion can trigger the overproduction of ROS. Our work focused on designing highly sensitive probes to track labile hepatic Cu activity, previously unmonitorable in aging models. The development of biocompatible labile copper probes (LCPs) (Fig. S17), especially LCP-PA, has enabled us to non-invasively observe an increase in labile hepatic Cu activity with aging.

While our findings show both increased Cu activity and decreased GSH levels occur as a function of age, various mechanisms can explain this observation that do not directly link the two. Age-related changes in redox protein cofactors or the upregulation of MT1/MT2 storage proteins could independently influence Cu distribution and GSH activity, indicating that these events may be concurrent but not linked56,57,58. Additionally, it is important to consider that prolonged Cu chelation could have broad metabolic effects which may influence our findings. Regardless whether these events are linked or occur in parallel, it is evident that these changes contribute collectively to oxidative damage, endangering hepatocytes such as hSCs that require low ROS levels to maintain their undifferentiated state. Our findings indicate that the activity of the ALDH1A1 enzyme decreases as mice age which may compromise liver defense and impact repair mechanisms.

Our research underscores the possibility of maintaining ALDH1A1 activity through metal ion chelation therapy. Our findings reveal that labile hepatic Cu levels can be normalized within a matter of weeks, a change that is accompanied by an increase in ALDH1A1 levels. The implications of these findings are profound: by managing hepatic Cu levels, we could significantly enhance liver health. Interventions might include dietary modifications to reduce overall Cu intake59 or to increase the consumption of antioxidants60, providing potential strategies for promoting liver longevity and combating the effects of aging. Ongoing work is now focus on deciphering the mechanism of how ALDH1A1 activity is being altered by Cu activity.

Methods

Ethical statement

This research complies with all relevant ethical regulations and was conducted in accordance with ethical guidelines established by the Institutional Animal Care and Use Committee (IACUC) of the University of Illinois at Urbana–Champaign (protocol #21223, IRB00007613), following the principles outlined by the American Physiological Society on research animal use.

Housing conditions: Light: 6am to 6 pm. Ambient temperature: 72 Fahrenheit degree. Humidity: 27%.

Statistical analysis and reproducibility

Statistical analyzes were performed in Microsoft Excel. Sample sizes in all experiments were chosen based on similar pilot studies performed in our lab or through power calculations to detect a p value <  0.05. All data were analyzed using Student’s t tests (two-tailed) or Mann Whitney U test (two-tailed). Data are expressed as mean ± SD. Group variances were similar in all cases. Sample sizes for Fig. 2c ranged from 15 to 23 due to some of the wells having different numbers of cells or cells being out of plane or out of focus. This variability was accounted for in the statistical analysis. The experiments were not randomized. The Investigators were not blinded to allocation during the experiments and the outcome assessment.

In vitro buffer preparation

1:1 v/v HEPES:DMF buffer solutions were prepared by mixing reduced glutathione, freshly made HEPES buffer (25 mL) and DMF (25 mL) to a final concentration of 50 mM HEPES and 2 mM glutathione (except for glutathione dependent assay). For LCP-Green assays, buffers are made without DMF. Adjustments to the desired pH value was done via addition of 1 M HCl or 1 M NaOH. pH values were determined using a Mettler-Toledo SevenCompact pH meter calibrated using pH 4.0, 7.0 and 10.0 standard buffers at 25 °C.

Tissue phantom preparation

Tissue phantoms were prepared by suspending agarose LE (750 mg) in deionized water (50 mL). The suspension was heated for one minute in a microwave oven until a viscous, translucent gel was produced. Immediately, 1 mL of 2% milk was added. The hot gel was poured into a cylindrical mold containing two plastic straws (3 mm diameter) and cooled at room temperature for at least 20 min. After cooling, the straws were removed and the gel was removed from the mold, yielding a tissue phantom with two parallel channels for the placement of FEP tubes containing sample solutions.

Field of view selection

The field of view for all experiments performed on the MSOT inVision 128 imaging system (iTheraMedical) was selected by using the built-in, cross-sectional anatomy guide to correctly position the animals to image the liver.

Image processing of background signal

An initial scan prior to the administration of LCP-PA or vehicle was performed for all PA imaging experiments in mice. This is used to account for any background signal that originates from endogenous pigments such as hemoglobin found in blood. After the final scan is taken, a ratio (PAFinal/PAInitial) is calculated to determine the change in PA intensity. Furthermore, we used the built-in spectral unmixing feature to distinguish between signals coming from LCP-PA, the turned over product, and blood based on the absorbance spectra of oxy-hemoglobin (HbO2) and hemoglobin (Hb).

LCP-Green series selectivity assay

The initial fluorescence signal (500 to 650 nm) of the LCP-Greens (5 µM in HEPES buffer, pH 7.4) were measured before the addition of a panel of metal ions (100 µM). After addition, the cuvette was sealed and incubated for one hour. Final measurements were recorded, and the fluorescence fold turn-on was calculated by dividing the final fluorescence intensity by the initial fluorescence intensity. All metal solutions were prepared in deionized water from their chloride salt except for Mohr’s salt and tetrakis(acetonitrile)copper(I) hexafluorophosphate which were dissolved in MeCN.

Cu-dose dependent assay

The initial fluorescence signal (500 to 650 nm) of the LCP-Greens (5 µM in HEPES buffer with 2 mM glutathione, pH 7.4) were measured before the addition of different concentrations of tetrakis(acetonitrile)copper(I) hexafluorophosphate. After addition, the cuvette was sealed and incubated for one hour. Final measurements were recorded, and the fluorescence fold turn-on was calculated by dividing the final fluorescence intensity by the initial fluorescence intensity.

Glutathione dependent assay

The initial fluorescence signal (400 to 550 nm) of the LCP-Greens (5 µM in HEPES buffer with 2-, 5-, or 10-mM glutathione, pH 7.4) were measured before the addition of tetrakis(acetonitrile)copper(I) hexafluorophosphate (100 µM). After addition, the cuvette was sealed and incubated for one hour. Final measurements were recorded, and the fluorescence fold turn-on was calculated by dividing the final fluorescence intensity by the initial fluorescence intensity.

LCP-PA in vitro selectivity assay

The initial absorbance (400 to 800 nm) of LCP-PA (5 µM, 1:1 v/v DMF/HEPES, pH 7.4) was measured before the addition of a panel of metal ions (100 µM). These initial measurements were used to determine the initial ratio770/680 via UV-vis spectroscopy. After addition, the cuvette was sealed and incubated for one hour. Final measurements were recorded, and the ratiometric fold turn-on was calculated by dividing the final ratio770/680 by the initial ratio770/680.

Cytotoxicity assay

96-well plates were seeded with 10,000 HepG2 cells per well (100 μL of 20,000 cells/mL) and incubated at 37 °C with 5% CO2 for 24 h. Media was removed and fresh EMEM containing 0, 1, 2, 5, 10, 20 μM of PACDx, LCP-Green-1, LCP-Green-4 or LCP-PA were added (0.1% DMSO final v/v). The media was removed after 24 h incubation and replaced with 100 μL of a 20:1 mixture of serum-free EMEM and (3-(4,5- dimethylthiazol2-yl)-2,5-diphenyl-tetrazolium bromide (MTT, 5 mg/mL stock in PBS). The cells were incubated for two hours under the same condition and then the medium was removed and replaced with DMSO (500 μL/well). The absorbance of each well was recorded at 555 nm on a microplate reader. Viability was calculated by the absorbance relative to the vehicle control.

Live cell imaging

The HepG2 cell line used in this study was obtained from the American Type Culture Collection (ATCC, HB-8065).

For labile copper imaging: HepG2 cells were plated in a poly-L-lysine coated microwell plate in EMEM (10% FBS) media for 24 h. HepG2 cells were then incubated with or without 100 µM CuCl2 in EMEM (10% FBS) for 24 h at 37 °C. After removing CuCl2, cells were further incubated with LCP-Green-1 or LCP-Green-4 (5 μM, 1% DMSO:EMEM) for 30 min at 37 °C. Cells were imaged using a Zeiss LSM700 confocal microscope to evaluate the copper sensing ability of both probes.

To examine the ability of both probes to detect endogenous levels of copper, HepG2 cells were incubated with 300 µM BCS for 24 h in a poly-L-lysine coated microwell plate. After removing BCS, cells were further incubated with LCP-Green-1 or LCP-Green-4 (5 μM, 1% DMSO:EMEM) for 30 min at 37 °C and imaged using a Zeiss LSM700 confocal microscope.

To examine the relationship between reduced GSH levels and labile copper activity, HepG2 cells were incubated with 500 µM NEM for eight hours or 5 mM glutathione ethyl ester for one hour in a poly-L-lysine coated microwell plate. After removing the incubation media, cells were further incubated with LCP-Green-4 (5 μM, 1% DMSO:EMEM) for 30 min at 37 °C and imaged using a Zeiss LSM700 confocal microscope to evaluate labile copper activity change.

For ALDH1A1 imaging: HepG2 cells were plated in a poly-L-lysine coated microwell plate in EMEM (10% FBS) media for 24 h. HepG2 cells were then incubated with or without 100 µM CuCl2 in EMEM (10% FBS) for 24 h at 37 °C. After removing CuCl2, cells were further incubated with AldeSense-AM (2 μM, 1% DMSO:EMEM) for 15 min at 37 °C. Cells were imaged using a Zeiss LSM700 confocal microscope to evaluate the ALDH1A1 activity.

For glutathione imaging: HepG2 cells were plated in a poly-L-lysine coated microwell plate in EMEM (10% FBS) media for 24 h. HepG2 cells were then incubated with 500 µM BSO for 24 h or 5 mM glutathione ethyl ester for one hour at 37 °C. After removing the incubation media, cells were further incubated with PACDx (2 μM, 1% DMSO:EMEM) for 60 min at 37 °C. After incubation, PACDx was replaced with PBS and cells were imaged using an EVOS FL epifluorescence microscope with a Cy5 filter cube to evaluate the levels of GSH.

For ROS imaging: HepG2 cells were plated in a poly-L-lysine coated microwell plate in EMEM (10% FBS) media for 24 h. HepG2 cells were then incubated with 100 µM CuCl2 for 24 h or 500 µM BSO for 24 h at 37 °C. After removing the incubation media, cells were further incubated with DCFH-DA (10 μM, 1% DMSO:EMEM) for 30 min at 37 °C. After incubation, DCFH-DA was replaced with PBS and cells were imaged using a Zeiss LSM700 confocal microscope to evaluate the ALDH1A1 activity.

All cell imaging experiments results were quantified relative to fluorescent intensity for each relevant control experiment using ImageJ 1.53k.

In vivo PA imaging

The mice were continuously anesthetized using isoflurane and placed in the prone position in the animal holder for imaging immediately before injection and one hour post injection. The temperature of the imaging chamber was set to 34 °C and the animal was allowed to equilibrate to the temperature for 15 min before imaging. Cross-sectional images were acquired at the abdomen of the mouse for liver imaging with a step size of 0.3 mm. The imaging position was guided by the built-in anatomy atlas in the MSOT InVision 128 system and was kept consistent for all scans. The excitation wavelengths used were selected based on the absorbance of the turnover product of LCP-PA and endogenous absorbers. 10 frames were recorded at every imaging wavelength. Subsequently, spectral unmixing, an automated function in the viewMSOT software, was used to separate signals coming from hemoglobin and deoxyhemoglobin.

PA imaging of hepatic Cu

Mice (C57BL/6) were purchased from the Jackson Laboratory and allowed to acclimate for at least one week before imaging. Mice were anesthetized using isoflurane (1.5 – 2.0%) and hair was removed from around the abdomen before imaging was performed. PA images of the liver were acquired prior injection of probe and 60 min post-injection. For initial testing of LCP-PA, mice were administered with either vehicle or CuCl2 (5 mg/kg), then imaged. To image hepatic copper as a function of age, four different age groups of mice (~7-, ~13-, ~42-, and ~83-weeks old) were given a systemic injection of LCP-PA (50 µM, 150 µL in 10% DMSO/PBS). Mice undergoing chelation therapy via oral gavage with ATN-224 were imaged every 20 days with systemic injection of LCP-PA (1 mg/kg in 10% DMSO/PBS).

PA imaging of hepatic GSH

Mice (C57BL/6) were purchased from the Jackson Laboratory and allowed to acclimate for at least one week before imaging. Mice were anesthetized using isoflurane (1.5 – 2.0%) and hair was removed from around the abdomen before imaging was performed. PA images of the liver were acquired prior injection of probe and 60 min post-injection. To image hepatic GSH as a function of age, four different age groups of mice (~7-, ~13-, ~42-, and ~83-weeks old) were given a hydrodynamic tail vein injection of PACDx (0.3 mg/kg in 10% DMSO in saline).

First Cu chelation therapy and PA monitoring

Female C57BL/6 mice (~40-weeks old) were separated into two experimental groups. The treatment group was administered ATN-224 (prepared by dissolving ATN-224 in drinking water at a dose of 0.5 mg/kg body weight) via the oral gavage technique every day for 60 days. The control group received the same volume of drinking water where no ATN-224 was added. Labile hepatic copper was monitored through the application of LCP-PA (i.p. injection) and PA imaging at the start of the study, and every 20 days after. No ATN-224 was administered on days when PA imaging was performed. At the end of treatment, mice were euthanized and their blood was collected in PBS with protease inhibitor for mass spectrometric analysis.

Second Cu chelation therapy

Female C57BL/6 mice (~41-weeks old) were separated into two experimental groups. The treatment group was administered ATN-224 (prepared by dissolving ATN-224 in drinking water at a dose of 0.5 mg/kg body weight) via the oral gavage technique every day for 20 days, and once every three days for 120 days. The control group received the same volume of drinking water where no ATN-224 was added. In the end of treatment, mice were euthanized, and their blood and liver were collected. Half of the liver were fixed by formalin and submitted for immunohistostaining, the other half were collected in PBS with protease inhibitor and submitted for mass spectrometric analysis. Blood samples were submitted for the liver function test.

Liver function test and mass spectrometric analysis

For liver function tests, samples were submitted to and processed by Veterinary Teaching Hospital, College of Veterinary Medicine, University of Illinois Urbana-Champaign.

Serum was collected from mice following euthanasia. Blood samples were allowed to clot at room temperature and were centrifuged at 3000 x g for 10 min to separate serum. Samples were stored at 2–8 °C and protected from light until analysis. All procedures were conducted in compliance with institutional animal care guidelines. Biochemical analyzes were performed using a Beckman Colter AU analyzer. Specific reagents and methods used included the following:

Albumin (ALB): Quantified using the bromocresol green (BCG) dye-binding method at pH 4.2. The absorbance of the albumin-BCG complex was measured at 600 and 800 nm.

Alkaline Phosphatase (ALP): Measured based on the conversion of p-nitrophenyl phosphate to p-nitrophenol at pH 10.4 in the presence of 2-amino-2-methyl-1-propanol. Absorbance was monitored at 410 and 480 nm, proportional to ALP activity.

Alanine Aminotransferase (ALT): Activity was determined by a modified method, monitoring the consumption of NADH at 340 nm in the presence of alanine and α-oxoglutarate, proportional to ALT activity.

Total Bilirubin (TBILI): Quantified using the stabilized diazonium salt (DPD) method. Bilirubin coupled with DPD in the presence of caffeine and surfactants to form azobilirubin, measured at 570 and 660 nm.

Cholesterol (CHOL): Determined via enzymatic hydrolysis of cholesterol esters by cholesterol esterase (CHE), followed by oxidation of free cholesterol by cholesterol oxidase (CHO). The resulting H2O2 reacted with 4-aminoantipyrine and phenol to form a red chromophore, measured at 540 and 600 nm.

Aspartate Aminotransferase (AST): Activity was determined by a modified method, monitoring the consumption of NADH at 340 nm in the presence of aspartate and α-oxoglutarate, proportional to ALT activity.

Gamma-glutamyl Transferase (GGT): Measured based on the conversion of gamma-glutamyl-3-carboxy-4-nitroanilide to glycylglycine and 5-amino-2-nitrobenzoate. Absorbance of 5-amino-2-nitrobenzoate was monitored at 410 and 480 nm, proportional to GGT activity.

Glucose (GLUC): Glucose is phosphorylated by hexokinase (HK) in the presence of adenosine triphosphate (ATP) and magnesium ions to produce glucose-6-phosphate (G-6-P) and adenosine diphosphate (ADP). Glucose-6-phosphate dehydrogenase (G6P-DH) specifically oxidizes G-6-P to 6-phosphogluconate with the concurrent reduction of nicotinamide adenine dinucleotide (NAD + ) to nicotinamide adenine dinucleotide, reduced (NADH). For Beckman Colter AU/DxC AU analyzers, the change in absorbance at 340/660 nm is proportional to the amount of glucose present in the sample.

Total Protein (TP): Cupric ions in an alkaline solution react with proteins and polypeptides containing at least two peptide bonds to produce a violet colored complex. The absorbance of the complex at 540/660 nm is directly proportional to the concentration of protein in the sample.

Urea Nitrogen (BUN): Urea is hydrolyzed enzymatically by urease to yield ammonia and carbon dioxide. The ammonia and α-oxoglutarate are converted to glutamate in a reaction catalyzed by L-glutamate dehydrogenase (GLDH). Simultaneously, a molar equivalent of reduced NADH is oxidized. The rate of change in absorbance at 340 nm, due to the disappearance of NADH, is directly proportional to the BUN concentration in the sample.

Each assay included quality control samples analyzed daily to verify performance. All results were recorded as U/L for enzymatic activity or mg/dL for concentrations, with conversions to SI units as required. The analytical measurement ranges and interference limits for each assay were consistent with the manufacturer’s instructions.

Proteomic analyzes were carried out in the Proteomics Core of the Roy J. Carver Biotechnology Center at the University of Illinois Urbana-Champaign. Liver tissue was homogenized in PBS buffer containing 6 M guanidine hydrochloride, 0.1% sodium deoxycholate, 10 mM tris(2-carboxyethyl)phosphine (TCEP), and 40 mM 2-chloroacetamide. The samples were boiled for 30 min to fully denature the proteins and promote reduction and alkylation of disulfide bonds. Proteins were chloroform-methanol precipitated and then resuspended in 50 mM ammonium bicarbonate buffer. Protein concentration was determined via BCA assay, and the proteins were then digested with trypsin 1:50 w/w (enzyme:substrate) at 37 C. On the following day, the tryptic peptides were desalted using C18 Sep Paks (Waters) and dried in a vacuum centrifuge.

The dried peptides were resuspended in 0.1% formic acid (FA) with 2% acetonitrile (ACN) and injected into an UltiMate 3000 RSLCnano system with a) a 50 cm Acclaim PepMap C18 column and a Q Exactive HF-X mass spectrometer or b) a 50 cm uPAC C18 column and an Orbitrap Fusion mass spectrometer (Thermo). Reversed phase separation over 60 min with mobile phases of 0.1% FA and 0.1% FA in 80% ACN was done for all samples. For samples analyzed by the HF-X, MS1 and MS2 scans were collected at 120k and 15k, respectively. The isolation window was 1.2 m/z, and the top 15 most abundant peptides were fragmented with HCD at 30 NCE. For samples analyzed by the Fusion, MS1 scans were collected in the Orbitrap at 120k resolution while the MS2 scans were acquired in the ion trap. The Fusion was operated in the top speed mode with a total cycle time of 3 s. Precursors selected for MS2 were isolated with a 1.6 m/z window and fragmented with CID at 35 NCE.

All raw LC-MS data was processed with Mascot Distiller and searched with Mascot v2.8.3 (Matrix Science) against the Uniprot Mus musculus reference proteome. For HF-X analyzes, a precursor tolerance of 10 ppm and a fragment mass tolerance of 0.02 Da was used. For Fusion analyzes, a precursor tolerance of 10 ppm and a fragment mass tolerance of 0.3 Da was used. A static carbamidomethylation modification was added to all searches along with variable modifications of methionine oxidation and protein N-terminal acetylation. Quantitation was accomplished with the Average [MD] algorithm in Mascot.

Immunohistochemical staining

5 μm paraffin-embedded sections were deparaffinized, placed in heated citrate buffer for 15 min for antigen retrieval, permeabilized in 3% H2O2 for 10 min, placed in normal goat serum (Invitrogen #31873) for 1 h, incubated overnight at room temperature in primary antibody at 1:200 (8-OHdG polyclonal antibody, product #bs-1278R Bioss Antibodies), incubated in goat anti-rabbit IgG secondary antibody at 1:400 for 30 min at room temperature (Invitrogen #31460), stained with diaminobenzidine (DAB) for 15 min (Fisher Scientific #AC112080250), and counterstained with 1% methyl green (Fisher Scientific #48-003-018) for 5 min to provide contrast.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.