Manganese-based type I collagen-targeting MRI probe for in vivo imaging of liver fibrosis

Abstract

Liver fibrosis is a common pathway shared by all forms of progressive chronic liver disease. There is an unmet clinical need for noninvasive imaging tools to diagnose and stage fibrosis, which presently relies heavily on percutaneous liver biopsy. Here, we explored the feasibility of using a novel type I collagen-targeted manganese (Mn)-based MRI probe, Mn-CBP20, for liver fibrosis imaging. In vitro characterization of Mn-CBP20 demonstrated its high binding affinity for human collagen (K_d = 9.6 µM), high T₁-relaxivity (48.9 mM⁻¹ s⁻¹ at 1.4 T and 27 °C), and kinetic inertness to Mn release under forcing conditions. We demonstrated MRI using Mn-CBP20 performs comparably to previously reported gadolinium-based type I collagen-targeted probe EP-3533 in a mouse model of carbon tetrachloride-induced liver fibrosis, and further demonstrate efficacy to detect fibrosis in a diet-induced mouse model of metabolically-associated steatohepatitis. Biodistribution studies using the Mn-CBP20 radiolabeled with the positron-emitting ⁵²Mn isotope demonstrate efficient clearance of Mn-CBP20 primarily via renal excretion. Mn-CBP20 represents a promising candidate that merits further evaluation and development for molecular imaging of liver fibrosis.

Introduction

Chronic liver disease (CLD) is a major global health burden that contributes to approximately 2 million deaths annually worldwide¹. Sustained insult resulting from conditions including hepatitis B or C virus infection, fatty liver disease, alcohol use, toxins, and autoimmune and genetic disorders, can result in chronic activation and dysregulation of tissue repair mechanisms leading to liver fibrosis^2,3. Regardless of the underlying cause, liver fibrosis is a common pathological feature of all progressive liver disease and is associated directly with poor outcomes, including cirrhosis, hepatocellular carcinoma, and liver failure^4,5. Meanwhile, the liver can potentially regress or reverse established fibrosis when the underlying cause is promptly identified and effectively eliminated⁶. In this regard, liver fibrosis assessment is key to prognosticating CLD outcomes and to evaluating response to therapy. Liver biopsy is the current gold standard for diagnosing and staging liver fibrosis, but the procedure is invasive and carries non-negligible risks of pain, bleeding, and infection, and samples online a small portion of the liver⁷. Serum biomarkers of liver function, extracellular matrix (ECM) production, and fibrogenesis-related cytokines, have been evaluated as potential noninvasive alternative to liver biopsy^8,9. However, many of these biomarkers lack sensitivity and/or specificity, risk false negative in end-stage CLDs, and can be confounded by a wide variety of extrahepatic disorders^10,11. Thus, there is an unmet need for noninvasive imaging to detect, tomographically map, quantify, and surveil the progression and resolution of liver fibrosis.

Clinical imaging modalities such as ultrasound and magnetic resonance (MR)-based elastography are widely used for the detection of liver fibrosis by measuring tissue stiffness^12–14. However, both methods are more effective for the detection of advanced fibrosis than the diagnosis and staging of earlier fibrosis. Molecular MR imaging using probes targeted to type I collagen, a protein that is abundantly present in the extracellular matrix of fibrotic tissue, has shown promise in experimental studies to detect and quantify liver fibrosis in animal models^15–18. For example, MRI using the gadolinium (Gd)-based type I collagen-targeted probe EP-3533 was shown to provide higher sensitivity than MR elastography to diagnose early-stage liver fibrosis in a rat model of diethylnitrosamine-induced liver fibrosis¹⁸. There are concerns over a rare but debilitating condition called nephrogenic systemic fibrosis related to Gd retention in renally impaired patients^19,20. Even in patients with normal renal function, long-term retention of Gd after linear or macrocyclic Gd-based contrast agents (GBCAs) is well documented despite unclear direct adverse effects^21–23, heightening the need for developing Gd-free alternatives.

In this study, we explored the feasibility of using a novel type I collagen-targeted manganese (Mn)-based MRI probe, Mn-CBP20, for liver fibrosis imaging. We demonstrate high affinity for type I collagen and high relaxivity similar to EP-3533, demonstrate how MRI using Mn-CBP20 performs comparably to EP-3533 in a mouse model of carbon tetrachloride (CCl₄)-induced liver fibrosis, and further demonstrate efficacy to detect fibrosis in a choline-deficient, l-amino acid-defined, high-fat diet (CDAHFD) mouse model of metabolically-associated steatohepatitis (MASH). While both models share the pathological feature of fibrosis, the underlying pathogenesis is different, with CCl₄ representing toxin-induced oxidative stress and injury and CDAHFD mimicking human MASH with respect to metabolic dysregulation and steatosis. Biodistribution studies using the Mn-CBP20 radiolabeled with the positron-emitting ⁵²Mn isotope demonstrate efficient clearance of Mn-CBP20 primarily via renal excretion.

Methods

Synthesis of Mn-CBP20

A batch of 0.204 g (0.091 mmol) collagen-binding peptide CM101P was stirred in 6 mL dimethylformamide (DMF). To this mixture, a few drops of N,N-diisopropylethylamine (DIPEA) were added so that pH check on wetted pH paper returned pH ≥8.5 and the solution turned from heterogenous to clear and homogenous. Next, a total of 149 mg (0.45 mmol) CDTA-monoanhydride was added in three equal portions (48, 52, and 49 mg each) roughly 1 h apart and the reaction was monitored by liquid chromatography-mass spectrometry (LC-MS). The addition of CDTA-monoanhydride initially resulted in a heterogenous white mixture, which becomes clear and homogenous upon conversion to the tri-acetylated intermediate. The DMF solution was added dropwise into 95 mL EtOAc, and the resultant white solids were isolated by centrifugation and decanting the supernatant. The crude solids were next taken up in 20 mL water, and then 90 mg MnCl₂•4H₂O (0.45 mmol) were added, and the solution was adjusted pH 7 using NaOH. Any solids were removed by filtration and the mixture was purified by reverse phase HPLC using a Teledyne IPSCO RediSep Gold C18 column. Mobile phase A: water adjusted to pH 7.0, Eluent B: Acetenitrile. Gradient program: 5% B for 2 column volumes, ramp from 5 to 50% B over 8 column volumes, hold at 50% B for 4 column volumes. Fractions containing pure Mn-CBP20 were pooled and lyophilized to obtain 110 mg (0.033 mmol, 36% yield) of Mn-CBP20 as white solids. ESI-MS: m/2z = 1693.2 [M + 5H]²⁺; calcd. 1693.1.

Relaxivity measurements

Relaxivity measurements were performed on a Magritek Spinsolve 60 MHz Multi-X Plus NMR spectrometer (1.41 T and 27 °C), a Siemens Biograph mMR scanner (3 T and 27 °C), and a Bruker Biospec scanner (4.7 T and 27 °C). At 1.41 T, longitudinal (T₁) relaxation was acquired via an inversion recovery measurement: repetition time: 10 s; maximum inversion time: 5 s; number of steps: 10; and transverse (T₂) relaxation was measured using a Carr–Purcell–Meiboom–Gill spin echo measurement: repetition time: 4 s; number of steps: 10; echo time: 0.2 ms; final echo time 200 ms. In addition, T₁ relaxation measurement was performed with an inversion recovery gradient echo sequence at 3 T using 8 different inversion times ranging from 100 to 1300 ms, and at 4.7 T with 14 different inversion times ranging from 12.5 to 5000 ms. T₁ values of various probe concentrations were measured by fitting the inversion recovery signal intensities as a function of the inversion time; and T₂ values of various probe concentrations were measured by fitting the signal intensities as a function of the echo time using algorithms written in MATLAB 2023b (MathWorks, Natick, MA). Relaxivity (r_1,2) was determined from the slope of the plot of 1/T_1,2 vs. probe concentration for at least four concentrations.

Collagen-binding affinity assay

Mn-CBP20 binding affinity to human type I collagen was evaluated by incubating varied concentrations of the probe with type I collagen deposited on a well plate or in a blank well for 2 h as described previously^16,24. The same experiment using EP-3533 was performed side by side for comparison. Mn and Gd concentration in the blank well and in the collagen-containing well were measured by ICP-MS to determine the total probe concentration and unbound probe concentration, respectively. The difference in total and unbound probe concentration corresponds to the concentration of the probe that is bound to collagen. To estimate binding affinity, the data were plotted and fit according to the following equation

$$\frac{\left[\mathit{bound}\right]}{\left[\mathit{collagen}\right]}=\frac{N\left[\mathit{free}\right]}{K_d+\left[\mathit{free}\right]}$$

Where N corresponds to the number of binding sites per collagen monomer, and K_d corresponds to the dissociation constant. Here N was fixed to 2.

[⁵²Mn]Mn-CBP20 radiolabeling

An aliquot of 70.3 MBq ⁵²MnCl₂ in 0.3 mL HCl (0.1 M) (t_1/2 = 5.6 days, obtained from the Cyclotron Facility at the University of Alabama, Birmingham, AL) was spiked to 1 mL stock solution of Mn-CBP20 (6 mM, pH 6-7). The pH was carefully adjusted to pH 3 by adding 1 M HCl and stirred at room temperature for 1 h, and then the pH was adjusted back to pH 6 to 7 by adding 1 M NaOH. The volume of added HCl and NaOH was recorded, and the total volume was then brought to 2 mL accordingly by adding sterile saline.

Animal models

All animal experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and in compliance with the ARRIVE guidelines²⁵ and approved by the MGH Institutional Animal Care and Use Committee. A total of 38 C57BL/6 mice (Charles River Laboratories, Wilmington, MA) were studied. All the mice were housed in an animal facility under controlled temperature and 12-h light/12-h dark cycle conditions.

CCl₄-induced liver fibrosis

Male C57BL/6 mice (6-week-old) were treated with an oral gavage of carbon tetrachloride for 6 weeks (N = 5) or 12 weeks (N = 5) (2–3 times per week, 0.1 mL of 20% CCl₄ in olive oil the first week, 30% the second week, and 40% from weeks 3 through 6 or 12). Control mice were treated with vehicle (olive oil) only for 6 weeks (N = 5) or 12 weeks (N = 3). In the 12-week CCl₄ group, there was 1 death due to accidental instillation of CCl₄ into the trachea, and 1 death in the 10th week, leaving 3 mice undergoing MRI and tissue analysis. Animals treated with CCl₄ or vehicle for 6 weeks were imaged with Mn-CBP20 (10 µmol/kg). Animals treated with CCl₄ or vehicle for 12 weeks were imaged with Mn-CBP20 (10 µmol/kg) and EP-3533 (10 µmol/kg) on 2 consecutive days in a random order.

CDAHFD model of MASH

To induce MASH, 6-week-old, male C57BL/6 mice were fed CDAHFD (N = 5) consisting of 60 kcal% fat and 0.1% methionine by weight (A06071302; Research Diets, New Brunswick, NJ) for 6 weeks. Another group of gender and age-matched mice (N = 3) were fed standard chow for 6 weeks.

Biodistribution and elimination

An additional group of 20 adult C57BL/6 mice (10–12 weeks old, 10 males and 10 females) received Mn-CBP20 radiolabeled with Mn-52. Each mouse was administered a cocktail dose of ^52/natMn-CBP20 (⁵²Mn complex: 2.3–3.1 MBq, ^natMn complex: 10 µmol/kg body weight). After injection, mice were housed in cages with bedding changed every 8 h to prevent reabsorption of excreted dose. At day 1 and day 7 post-injection, respectively, 5 male and 5 female mice underwent PET-CT and were euthanized to evaluate biodistribution and elimination of the probe.

MRI data acquisition and analysis

Animals were anesthetized with isoflurane (3%) and placed in a specially designed cradle with body temperature maintained at 37 °C. The isoflurane level was maintained at 1–2% for a respiration rate of 60 ± 5 breaths per minute. Imaging was performed using a 4.7 Tesla MRI scanner (Bruker, Billerica, MA) with a custom-built volume coil. The tail vein was cannulated for i.v. delivery of the MRI probe. A series of three-dimensional (3D) T1-weighted fast low-angle shot (FLASH) MR images were first acquired (repetition time/echo time = 15/2 ms; field of view = 48 mm × 30 mm×25 mm; matrix = 192 × 120 × 100; flip angle = 30°; acquisition time = 3 min), then a bolus of molecular probe was administered intravenously, and imaging was repeated with the 3D T1-weighted FLASH for 45 min. After the imaging session, animals were sacrificed (90 min after injection), and the liver tissue was subjected to histopathologic analysis.

The liver-to-muscle contrast-to-noise ratio (CNR) was calculated as the difference in signal-to-noise ratio between liver and muscle regions of interest (ROI). An ROI was manually traced, encompassing the liver parenchyma while avoiding major blood vessels. A second ROI was placed on the dorsal muscle visible in the same image slice to quantify the signal intensity in the muscle for comparison. The third ROI was placed in the field of view without any tissue (air) to measure the variation in the background signal. The same analysis was performed on the pre- and post-injection images acquired with the FLASH sequence. Image visualization and quantification were performed in AMIDE²⁶. CNR was calculated by subtracting the signal intensity (SI) in the muscle from that in the liver and normalizing to the standard deviation (SD) of the signal in the air outside the animal, CNR = (SI_liver − SI_muscle)/SD_air. ΔCNR was calculated by subtracting the CNR_pre from CNR_post, ΔCNR = CNR_post – CNR_pre. A blood ROI was positioned in the left ventricular blood pool. The dynamic MRI signal time courses of the liver or blood were calculated by measuring the percentage change in signal intensity between post-injection and pre-injection images.

PET-CT imaging

Mice were imaged in a MultiScan™ LFER 150 PET/CT scanner (Mediso USA, LLC; Arlington, VA, USA). Static PET data were acquired in list mode and were reconstructed using the 3D ordered subset expectation maximization (OSEM) image reconstruction algorithm with 30 iterations and 1 subset (Tera-Tomo 3D) and were corrected for scatter and attenuation. Image visualization and quantification were performed in Horos (Horosproject.org, Nimble Co LLC d/b/a Purview, Annapolis, MD, USA).

Tissue analysis

After imaging, the animals were sacrificed under anesthesia, and the liver tissues were harvested. One piece of the left lobe of the liver was fixed in 4% paraformaldehyde in PBS, dehydrated, embedded in paraffin, and then sectioned into 5-μm-thick slices at 100 μm intervals for later staining with Sirius Red and hematoxylin and eosin (H&E). Another piece of the left lobe was quickly frozen in liquid nitrogen for later hydroxyproline analysis. For the CCl₄ model (N = 8) and vehicle-treated (N = 8) animals, the collagen proportional area (CPA), as determined by the percentage area stained with Sirius Red, was quantified and averaged from the histologic images from five sections at different level per animal by using ImageJ^27,28. For the CDAHFD model (N = 5) and age-matched normal (N = 3) animals, hydroxyproline in the liver tissue weighing at ~200 mg was quantified with high-performance liquid chromatography analysis of tissue acid digests as previously described^29,30. Hydroxyproline is expressed as the amount per wet weight of tissue.

Ex vivo Mn-52 quantification

Ex vivo quantification of ⁵²Mn activity in each injected dose was determined using a Capintec CRC-15PET dose calibrator. Injected dose activity was determined from the difference in activity recorded in the loaded syringe and injection line prior to and after injection. This value was then adjusted by subtracting the activity measured from the injection site (tail). Quantification of Mn-52 activity in ex vivo tissue samples was performed using a Perkin Elmer 2480 Wizard2 gamma counting system. The percentage injected (%ID) dose in ex vivo tissue samples was determined by comparing tissue counts to a measured aliquot of the injected dose. Bone, blood, fat, and muscle were estimated to account for 13%, 7%, 13%, and 40% of the total body weight. All measurements were corrected to account for decay.

Statistical analysis

Data are reported as the means ± SD. Differences between the two groups were tested with a two-tailed paired or unpaired t test. Differences among more than two groups were tested with one-way analysis of variance (ANOVA), followed by Tukey’s post hoc test with P < 0.05. Correlations between CPA and MR imaging metrics (ΔCNR) of fibrosis were assessed with the Pearson correlation coefficient. P values less than 0.05 were considered to indicate statistical significance. All statistical analyses were performed using GraphPad Prism 10 (GraphPad software).

Results

Mn-CBP20 design and synthesis

We sought a Mn-based imaging probe construct that binds type I collagen specifically and with high avidity, is thermodynamically stable and kinetically inert with respect to Mn release, and can be manufactured in as few synthetic steps as possible. For targeting of type I collagen we utilized the peptide CM101P, which is based on collagen-specific cyclic decapeptide sequences identified through a phage display screen and has been further optimized for collagen binding²⁴. CM101P is the same disulfide bridged cyclic decapeptide with flanking amino acids that was employed in the previously reported collagen-targeted probes EP-3533, CM-101, and SNIO-CBP^16,24,31. To tightly incorporate Mn²⁺ as the MR signal-generating component, the peptide was conjugated with three CDTA-monoamide ligands. Prior studies have demonstrated how hexadentate ligands built from the rigid trans-1,2-diaminocylcohexane core bind Mn²⁺ with high-thermodynamic stability and are kinetically inert to Mn²⁺ de-chelation^32–37 Furthermore, the CDTA-monoanhydride synthon used for ligand conjugation can be generated cleanly and quantitatively from inexpensive materials in a single step.

The synthesis of Mn-CBP20 is outlined in Fig. 1. Briefly, CDTA-monoanhydride was synthesized from commercially available CDTA by stirring in acetic anhydride with pyridine added, as reported previously³⁸ and then conjugated with CBP via stirring in DMF with DIPEA. Next, the crude reaction product containing L-CBP20 was reacted with MnCl₂ and Mn-CBP20 isolated after purification RP-HPLC in 36% overall yield based on CM101P.

Fig. 1. Synthesis of Mn-CBP20.

(i) acetic acid, pyridine, RT, 38 h; (ii) CDTA-mA, DMF, DIPA, RT; (iii) MnCl₂•4H₂O, pH 7 H₂O.

Mn-CBP20 physical properties are well-suited for molecular MR imaging of collagen

Mn-CBP20 binds to human collagen with comparable affinity to the previously reported probe EP-3533^{15,18,24,28,39–43} (Fig. 2A). The collagen affinity of Mn-CBP20 was similar to that of EP-3533 with K_d values of 9.5 ± 2.5 µM and 5.5 ± 1.1 µM, respectively.

Fig. 2. In vitro assays for collagen binding and kinetic inertness.

A Plots of free vs collagen bound Mn-CBP20 (black circles) and EP-3533 (open green circles) demonstrate comparable affinity for human type I collagen at RT. B Kinetics of Mn²⁺ dissociation from Mn-CBP20 (black circles and Mn-CDTA (open black circles) under forcing conditions (0.5 mM Mn2 + , 10 mM Zn(OT)₂, pH 7.4 50 mM Tris: 18 mM citrate, 27 °C) monitored by r₂ change at 1.4 T and indicates Mn-CBP20 is more kinetically inert than Mn-CDTA. The dotted black line corresponds to the r₂ value of MnCl₂•4H₂O dissolved in the corresponding buffer mixture.

Mn-CBP20 possesses high relaxivity required for in vivo imaging of type I collagen deposition. The r₁ value of Mn-CBP20 in pH 7.4 solution (10 mM citrate) at 1.4 T and 27 °C is or 48.9 mM⁻¹ s⁻¹ per peptide construct (16.3 mM⁻¹ s⁻¹ per Mn²⁺ ion), which is comparable to previously reported r₁ of EP-3533 recorded under similar, although not identical, conditions, Table 1.

Table 1.

Comparisons of Mn-CBP20 and EP-3533 physical properties relevant to collagen molecular imaging.

	Kd (µM)a	Per molecule r1 (mM−1 s−1)			Per metal ion r1 (mM−1 s−1)			Per molecule r2 (mM−1 s−1)	Per metal ion r2 (mM−1 s−1)
		1.41 T	3 T	4.7 T	1.41 T	3 T	4.7 T
Mn-CBP20	9.5	48.9b	27.7d	17.1e	16.3b	9.23d	5.69e	117.9b	39.3b
EP-3533	5.5	48.3c	29.6d	21.4e	16.1c	9.85d	7.13e	77.7c	25.9c

^aKd determined in the presence of human type I collagen; ^b1.41 T and 27 °C, pH 7.4, 50 mM Tris: 10 mM citrate; ^c1.41 T and 37 °C, PBS; ^d3 T and 27 °C, PBS; ^e4.7 T and 27 °C, PBS.

To assess kinetic inertness to Mn²⁺ release, we evaluated Mn²⁺ dissociation kinetics upon challenge of 0.167 mM Mn-CBP20 (0.5 mM Mn²⁺) with 10 mM Zn(OTf)₂ in the presence of pH 7.4 50 mM Tris/18 mM citrate buffer (Fig. 2B). The presence of a weak chelator such as citrate is required to maintain homogenous reaction conditions during the course of the measurement, as treatment of Mn-CBP20 with even 1 mol equiv. Zn²⁺ in the absence of chelating buffer precipitates solids from the reaction mixture. Mn²⁺ dissociation was monitored by observing change in r₂. Under this set of assay conditions, the r₂ value associated with dissociated Mn²⁺ (16.3 mM⁻¹ s⁻¹) is approximately half that of Mn²⁺ bound within Mn-CBP20. As a positive control, we monitored Mn²⁺ dissociation from the kinetically inert complex Mn-CDTA (r₂ = 5.9 mM⁻¹ s⁻¹). Under these assay conditions, Mn²⁺ dissociates from Mn-CBP20 with half-life (t_1/2) of 22.5 h. Dissociation from Mn-CDTA is slightly faster, with t_1/2 of 15.5 h.

Mn-CBP20 enhanced MRI detects collagen deposition in the mouse model of CCl₄-induced liver fibrosis

Sirius Red staining and quantitation of liver collagen show that the percentage collagen proportional area (CPA) was significantly elevated in mice receiving 6-weeks CCl₄ treatment vs vehicle-treated control (2.8 ± 1.0% vs 0.4 ± 0.2%, respectively, P = 0.03) (Fig. 3A, B). CPA further increased after 12 weeks in CCl₄-treated mice compared to control (6.9 ± 2.4% vs. 0.8 ± 0.2%, respectively, P < 0.001; P = 0.002 vs 6-week CCl₄ treatment).

Fig. 3. Molecular MRI using Mn-CBP20 in mice treated with carbon tetrachloride (CCl₄) or vehicle (olive oil) for 6 or 12 weeks to induce liver fibrosis or to serve as controls.

A Representative histologic images of the liver tissues from CCl₄ and control mice, stained with Sirius Red. Scale bar: 250 μm. B Total collagen assessed by collagen proportional area (CPA) in Sirius Red stained slides as a fibrosis measure in CCl₄ and control groups. C T₁-weighted MRI images acquired before and 45 min post-injection of Mn-CBP20 (10 µmol/kg). D Comparisons of liver MRI signals in CCl₄ and control groups, quantified by changes in liver-to-muscle contrast-to-noise ratio (∆CNR) at 45 min post-injection of Mn-CBP20. One-way ANOVA, *P < 0.05, **P < 0.01, ***P < 0.001, data are shown as means $\pm$SD. E Correlation plot of ΔCNR (Mn-CBP20) versus %CPA (Sirius Red) in mouse livers showing linear correlation (r = 0.91, P < 0.001).

Increased liver collagen in CCl₄-treated mice was reflected in the molecular MR imaging data. Axial T₁-weighted liver images acquired prior to and 45 min after injection of Mn-CBP20 into mice receiving either 6-week or 12-week regimens of vehicle or CCl₄ treatment are shown in Fig. 3C. The images reflect near complete Mn-CBP20 washout from the liver of vehicle-treated mice but marked delayed liver enhancement in CCl₄-treated mice. The (pre-post) injection increase in liver-to-muscle contrast-to-noise ratio (ΔCNR) values recorded 45 min after Mn-CBP20 injection were significantly greater in mice receiving 6 weeks of treatment with CCl4 compared to vehicle-treated control (3.8 ± 0.3 vs. 2.3 ± 0.2, respectively, P = 0.03) (Fig. 3D). Similarly, the ΔCNR values recorded in mice receiving 12 weeks of CCl₄ treatment were significantly higher than mice receiving 12 weeks of vehicle (4.7 ± 1.5 vs. 2.4 ± 0.5, respectively, P = 0.009) (Fig. 3D). Furthermore, the ΔCNR values recorded following the injection of Mn-CBP20 correlated tightly with CPA (r = 0.91, P < 0.001) (Fig. 3E).

As expected from the in vitro characterization data, Mn-CBP20 performs similarly to EP-3533 for MR imaging of liver fibrosis. Figure 4A, B compares the time course of percentage liver signal change in mice receiving either 12 weeks of treatment with CCl₄ or vehicle control following injection of Mn-CBP20 or EP-3533, respectively. Both probes differentially generated greater liver enhancement in CCl₄ treated mice that is sustained from 1 min post-injection up to at least 55 min post-injection. There was no difference in blood half-life in the CCl₄ and vehicle-treated mice for EP-3533 (17.8 ± 2.5 min vs. 18.9 ± 2.6 min, P = 0.86) or Mn-CBP20 (13.5 ± 1.5 min vs. 12.2 ± 2.4 min, P = 0.76), as shown in Fig. 4C, D. Separate comparisons of ΔCNR generated 45 min after injection of either probe in CCl₄ vs. vehicle-treated mice indicate that Mn-CBP20 offers comparable sensitivity to EP-3533 for detection of liver fibrosis (Fig. 4E).

Fig. 4. Liver and blood MRI signal dynamics post-injection of collagen-targeted probes EP-3533 or Mn-CBP20.

Comparisons of liver percentage signal change (%SI) over time in 12-week CCl₄ mice and vehicle-treated control mice administered with A EP-3533 (10 µmol/kg) or B Mn-CBP20 (10 µmol/kg) (n = 3/group). Blood %SI over time from the same groups received C EP-3533 (10 µmol/kg) or D Mn-CBP20 (10 µmol/kg). E Both collagen-targeted probes, EP-3533 and Mn-CBP20 induced significantly greater ∆CNR at 45 min post-injection in the livers of 12-week CCl₄ mice as compared to controls. One-way ANOVA, *P < 0.05, **P < 0.01, data are shown as means $\pm$SD.

Mn-CBP20 enhanced MRI detects liver fibrosis in a nutritional mouse model of MASH

We further evaluated Mn-CBP20 molecular MRI as a marker of liver fibrosis using age-matched mice receiving either a 6-week regimen of CDAHFD or standard chow. CDAHFD results in steatohepatitis that recapitulates salient features of human MASH including liver fibrosis, as evidenced by the histologic data from one representative animal in each group shown in Fig. 5A, where severe steatosis and increased liver collagen are evident in specimens stained with H&E and Sirius Red, respectively. HPLC quantitation of liver hydroxyproline content, an amino acid that is highly abundant in type 1 collagen, indicates that levels are significantly higher in mice fed CDAHFD vs standard diet (638 ± 135 μg/g vs. 196 ± 61 μg/g, respectively, P = 0.002) also consistent with liver fibrosis (Fig. 5B).

Fig. 5. Molecular MRI using Mn-CBP20 in mice fed with choline-deficient L-amino acid-defined high-fat diet (CDAHFD) or standard chow (naïve) for 6 weeks.

A Histologic images of the liver tissues from a representative mouse in each group, stained with H&E and Sirius Red, respectively. Scale bar: 250 μm. B Total collagen assessed by hydroxyproline assay as a fibrosis measure in CDAHFD and naïve control groups. Unpaired Student’s t test, *P < 0.05, data are shown as means $\pm$SD. C T₁-weighted MRI images acquired before and 40 min post-injection of Mn-CBP20 (10 µmol/kg). D Comparisons of liver ∆CNR at 35-45 min post-injection of Mn-CBP20.

Figure 5C shows axial T₁-weighted images recorded in mice fed CDAHFD and standard diet prior to and 45 min after injection of Mn-CBP20. Markedly stronger liver enhancement is observed in CDAHFD, consistent with increased liver collagen content. The effect is also reflected in the quantitation of liver enhancement, with significantly greater ΔCNR in mice fed CDAHFD vs. standard diet (4.9 ± 0.8 vs. 3.1 ± 0., P = 0.02) (Fig. 5D). The liver ΔCNR showed a significant positive correlation with liver hydroxyproline content (r = 0.92, P = 0.001).

Mn biodistribution

⁵²Mn biodistribution was evaluated following injection of [⁵²Mn]Mn-CBP20 mixed with ^natMn-CBP20 at the same mass dose as used for MRI. This radiolabeling approach enables us to tomographically visualize whole-body biodistribution of administered Mn in vivo using PET, and to definitively discern administered Mn from endogenous Mn in tissues.

Whole-body PET-CT images recorded 1 and 7 days after [⁵²Mn]Mn-CBP20 injection are shown in Fig. 6A and ex vivo ⁵²Mn biodistribution data are shown in Fig. 6B, C and also tabulated in Supplementary Tables 1 and 2. The highest concentration of ⁵²Mn 1 day after injection is found in the kidneys, consistent with the expected renal elimination of the compound. Substantial activity is also found in the liver and organs of the gastrointestinal tract which could be related to partial elimination of the probe. However, Mn is also eliminated via the hepatobiliary path, and we cannot presently rule out the possibility that de-chelated ⁵²Mn also contributes to this signal. Other tissues with relatively high concentrations of ⁵²Mn 1 day after [⁵²Mn]Mn-CBP20 injection include the pancreas, heart, and salivary glands. By day 7, residual activity in the kidney, liver, gastrointestinal tract, pancreas, and heart decreased substantially, but relatively high levels of residual activity in the salivary glands persisted. This observation is consistent with prior studies in which persistent ⁵²Mn activity was observed in the salivary glands of mice treated with [⁵²Mn]Mn-DPDP³⁵. We note that salivary Mn levels have previously been evaluated as a biomarker of low-level Mn exposure⁴⁴.

Fig. 6. Biodistribution of Mn-CBP20 probe in healthy control mice.

A Whole-body PET-CT(color scale) images of normal mice imaged at 1 day and 7 days post-injection of [⁵²Mn]Mn-CBP20. PET imaging is reported as a percentage injected dose per cubic centimeter (% ID/cc). Ex vivo biodistribution of [⁵²Mn]Mn-CBP20 at B 1 day and C 7 days p.i., expressed as percent injected dose per gram of tissue (%ID/g) in various organs and tissue. n = 10 mice (5 males and 5 females)/timepoint. Data are shown as means $\pm$SD.

Discussion

Liver fibrosis is a hallmark feature of progressive CLD, and the fibrosis stage is a predictive marker for poor outcomes such as liver cancer or end-stage liver disease^45–47. Patients with advanced-stage liver fibrosis face a high risk of death from CLD and few available treatment options. Patients with mild to moderate fibrosis stand a better chance of benefitting from interventions to halt or even reverse CLD progression⁴⁸. However, moderate fibrosis often occurs in the absence of clinical signs and too often remains undetected until progression to a more advanced stage. Patients suffering from progressive CLD could benefit immensely from noninvasive tools to diagnose, stage, and track the progression or resolution of moderate liver fibrosis.

A prior study using a rat model of bile duct ligation indicated that molecular MR imaging using the experimental type I targeted probe EP-3533 in rats provided sensitivity to detect mild and moderate liver fibrosis that could not be detected using MR elastography¹⁸. Although the technology is promising, challenges to developing EP-3533 for clinical imaging are anticipated. EP-3533 uses Gd as the MR signal-generating component. Concerns over the presence of residual Gd that is retained by patients who receive contrast-enhanced MRI scans have been widely publicized and contributed to a challenging regulatory landscape for the development of new Gd-based imaging probes.

Mn-based imaging probes offer a potentially viable alternative to Gd-based imaging probes. The Mn²⁺ ion is a very effective T₁-relaxation agent, and Mn²⁺ is an endogenously present nutritional element⁴⁹. Mn-based MRI agents have an established history of clinical application. Mangafodipir trisodium (Teslascan®) was approved for liver imaging in the US and EU but is no longer marketed^50,51. Mn-DPDP enhances the hepatobiliary system via the transient accumulation of the dissociated Mn²⁺ in hepatocytes. Therefore, Mn-DPDP needs to be infused due to concerns about the cardiovascular safety of Mn^2+ 52. In recent years, novel Mn agents have advanced to evaluation in clinical trials^53,54. EVP-1001-1 (SeeMore™), a weakly chelated Mn agent, completed phase II trial for imaging myocardial viability⁵⁵. SN132D, a Mn-based nanoparticle for tumor imaging, recently completed phase I trial in patients with breast cancer⁵⁶. Oral Mn chloride tetrahydrate (LumenHance®) was approved for gastrointestinal MRI and recently, a new oral formulation of Mn chloride with L-alanine and vitamin D3 (Orviglance®) completed phase III trial for liver-specific imaging⁵⁷. RVP-001 is a stable manganese chelate that was designed to replace extracellular fluid GBCAs with equivalent imaging properties, and it is currently being evaluated in phase II trial (NCT06322342).

In this study, we synthesized and evaluated a Mn-based probe targeted to type I collagen termed Mn-CBP20. In vitro characterization showed that Mn-CBP20 binds to human type I collagen with comparable affinity to EP-3533, and also exhibits comparable relaxivity. Direct comparison of Mn-CBP20 to EP-3533 at the same dose in a mouse model of CCl4-induced liver fibrosis indicates that Mn-CBP20 is equally effective to EP-3533 for the detection of liver collagen deposition, and thus suggests that Mn-CBP20 may offer similar advantage over current clinical imaging for detection of mild to moderate liver fibrosis. Additional dose optimization including a dose escalation study to determine the optimal dose range for human use and the maximum tolerated dose is necessary for clinical translation of this Mn-based collagen-targeted agent. We also demonstrated Mn-CBP20 was effective for the detection of liver fibrosis in a CDAHFD mouse model of MASH, underscoring the potential use of what is both the most prevalent and most rapidly increasing CLD. Biodistribution studies of ⁵²Mn-labeled Mn-CBP20 demonstrate efficient clearance of the probe primarily via renal excretion. The non-negligible liver ΔCNR in the sham animals and liver uptake of ⁵²Mn-CBP20 in healthy mice could arise from the amount of native collagen in normal livers, as reflected by the total collagen quantified in hydroxyproline assay (Fig. 5B). Further studies to characterize the metabolism, pharmacokinetics, and toxicity of Mn-CBP20 are warranted for translational purposes.

There are limitations to the study. While the straightforward synthesis of Mn-CBP20 highlights the potential for low cost of goods manufacture, synthetic activities for this study focused narrowly on obtaining small amounts of compound for preliminary evaluation, rather than on process optimization. MR imaging data obtained using Mn-CBP20 was evaluated in the context of liver collagen content determined through ex vivo biochemical assays, which underscores collagen specificity and the potential to quantify liver collagen deposition. However, MR imaging using Mn-CBP20 was not evaluated in the context of the histologically assigned fibrosis stage, and our study did not compare against currently available serum biomarker panels or clinical imaging that can generally reliably detect advanced fibrosis. Our imaging studies were designed to demonstrate proof of concept, and utilized male mice which have been reported to have a stronger fibrotic response than females⁵⁸. However, future evaluation will include both male and female animals. We used EP-3533 as a positive control in the imaging studies and animals with no fibrosis as negative controls. A negative control probe was not used here but we have previously shown that untargeted probes do not detect liver fibrosis compared to EP-3533⁴⁰.

In conclusion, Mn-CBP20 represents a promising candidate that merits further evaluation and development for molecular imaging of liver fibrosis.

Author contributions

C.Z., H.M., N.J.R., and I.Y.Z. performed the animal experiments and imaging. D.D., E.M.G., H.D., and V.H. participated in synthesizing and characterizing the collagen-specific contrast agent. H.M. and N.J.R. carried out the radiolabeling and biodistribution experiments. C.Z., H.M., and P.P. participated in ex vivo experiments. C.Z., H.M., E.M.G., and I.Y.Z. analyzed the data. C.Z., E.M.G., and I.Y.Z. wrote the manuscript. P.C. and I.Y.Z. designed and supervised the study. All authors reviewed the manuscript.

Data availability

Data are provided within the manuscript or supplementary information files.

Competing interests

E.M.G. holds equity in Reveal Pharmaceuticals and serves as a consultant to both Reveal Pharmaceuticals and Collagen Medical LLC. V.H. is an employee of Collagen Medical LLC. P.C. holds equity in, and is a consultant to, Collagen Medical LLC; holds equity in Reveal Pharmaceuticals Inc.; and receives research support from Pliant Therapeutics, Canon Medical USA, and Transcode Therapeutics. I.Y.Z. receives research support from Reveal Pharmaceuticals Inc. and currently serves as an Associate Editor of npj Imaging but was not involved in the editorial review of or the decision to publish this article. The other authors have declared that no conflict of interest exists.

Supplementary information

The online version contains supplementary material available at 10.1038/s44303-025-00075-1.