Non-Invasively Distinguishing Progress of Liver Fibrosis by Visualizing Hepatic Platelet Derived Growth Factor Receptor-Beta Expression with an MRI Modality in Mice


 Background: Activated hepatic stellate cells are the most critical cell responsible for liver fibrosis. In liver fibrogenesis, platelet-derived growth factor is the most prominent mitogen for hepatic stellate cells. This study aims to explore the potential of gadolinium (Gd)-labeled cyclic peptides (pPB) targeted to platelet-derived growth factor receptor-β (PDGFR-β) as a magnetic resonance imaging (MRI) radiotracer to identify the progress of liver fibrosis by imaging hepatic PDGFR-β expression. Results: Hepatic PDGFR-β expression level was found to be paralleled with the severity of liver fibrosis, which was increased with the progression of fibrosis and reduced with the regression. Majority of cells expressing PDGFR-β was determined to be activated hepatic stellate cells in fibrotic livers. Culture-activated human hepatic stellate cells expressed abundant PDGFR-β, and FITC-labeled pPB could bind to human hepatic stellate cells in a concentration and time dependent manner. With Gd-labeled pPB as a tracer, an MRI modality demonstrated that the relative hepatic T1-weighed MR signal value was increased progressively along with severity of hepatic fibrosis and reduced with the remission. Conclusion: Hepatic PDGFR-β expression reflects the progress of hepatic fibrosis, and MR imaging using Gd-labeled pPB as a tracer may distinguish different stages of liver fibrosis in mice.


Introduction
Liver brosis is a wound-healing response to various liver injuries, such as alcohol, viruses and hepatotoxic materials, and may be protective to a certain extent for brosis has been found to be associated with increased resistance of hepatocytes toward these injurious stimuli. 1 However, sustained or repeated liver injuries result in excessive brotic scars which inevitably damage hepatic lobular structure and distort hepatic vascular architecture, eventually resulting in liver dysfunction. For a long time, the prognosis of chronic liver disease (CLD) has been considered to largely depend on the severity and progression of liver brosis. 2 More importantly, liver brosis is strongly associated with hepatocellular carcinoma (HCC), because the majority of human HCCs are found to occur in the setting of advanced brosis or cirrhosis. 3 Thus, it seems imperative for the management of CLD to diagnose liver brosis as early as possible and stage the extent of brosis as accurately as possible.
Hepatic stellate cells (HSCs) are liver-speci c mesenchymal cells, which are located in the space of Dissue between hepatic sinusoidal endothelial cells and hepatocytes. HSCs display two cellular phenotypes, the quiescent and activated state. In the quiescent state, HSCs contain abundant vitamin A lipid droplets; whereas in the activated state, they lose vitamin A and transdifferentiate into myo broblast-like cells which express α-smooth muscle actin (α-SMA) and produce extracellular matrix (ECM). 4 When the liver is injured due to various etiological factors, the signals from the damaged hepatocytes and immune cells provoke quiescent HSCs transdifferentiating to an activated state. In CLD where the liver injury persists or recurrent episodes, HSCs become continuous or repeated activation, which subsequently produce massive ECM and inevitably result in liver brosis and even cirrhosis. Thus, activated HSCs are generally considered to be the most critical cell responsible for liver brosis. 5,6 In liver brogenesis, platelet-derived growth factor (PDGF) is acknowledged to be the most prominent mitogen for HSCs. At present, PDGF signaling network is found to be comprised of four ligands, PDGF A-D, and transduce their signals through two transmembrane receptors, PDGFR-α and -β. 7 It has been shown that PDGF receptor-β (PDGFR-β) is abundantly expressed on activated HSCs in vitro and in vivo, and deleting PDGFR-β on HSCs impairs their brogenic potential in vivo and decreases hepatic expression of α-SMA and collagen α1. 8,9 The cyclic peptide C*SRNLIDC* (pPB) contains arginine (R) and isoleucine (I) amino acids, which are the receptor-binding moieties of PDGF B-chain, and can bind speci cally to PDGFR-β. 10 After modi ed with pPB, carriers including human serum albumin, liposome or adenovirus carriers have been shown to have a speci c a nity for activated HSCs. [10][11][12] In this study, we investigated the association between hepatic PDGFR-β expression level and the extent of liver brosis. Then after pPB was modi ed with gadolinium (Gd) and used as a magnetic resonance imaging (MRI) agent, we tried to develop a molecular imaging modality to distinguish the process of liver brosis by visualizing hepatic PDGFR-β expression.

Experimental Animals
Male C57BL/6J mice were from the Department of Experimental Animals, Fudan University (Shanghai, China). The study was approved by Institutional Ethical Committee of Animal Experimentation, and all experiments were performed according to the governmental and international guidelines on animal experimentation.
Synthesis of pPB Cyclic Peptide and its derivatives As we described previously, 11 the linear octapeptide CSRNLIDC was synthesized by the Boc-protected solid-phase peptide synthesis method and cyclized by the linkage of two cysteine residues to gain cyclic peptide (pPB). Then pPB was incubated with FITC for 2 hours in the presence of triethylamine to gain FITC-labeled pPB cyclic peptide (pPB-FITC).
Maleimido-mono-amide-DOTA (20 mg) was dissolved in 1.0 mL N,N-Dimethylformamide and then added into pPB solution. After stirred for 2 hours at room temperature, the mixed solution was ltered, puri ed and freeze-dried to gain pPB-DOTA. Next, pPB-DOTA (60 mg) was dissolved in 20.0 mL pure water, and 32 mg of GdCl 3 ·6H 2 O was dissolved in 4.0 mL pure water. Then the two kinds of solutions were mixed, and the mixed solution was raised to a PH of about 6.0 by NaOH. After being vibrated overnight at 37 °C, the mixed solution was ltered, puri ed and freeze-dried to gain Gd-labeled pPB cyclic peptide (pPB-DOTA-Gd).
The purity of pPB and its derivatives was examined by analytical reverse phase high performance liquid chromatography, and their molecular weight was examined by electrospray ionization mass spectrometry.

Expression of PDGFR-β on Human HSCs
Human HSC-LX2 (from Cell Bank of Chinese Academy of Sciences, Shanghai, China) were cultured in the cell medium supplemented with 5% heat-inactivated fetal bovine serum (FBS) and 1% penicillin/streptomycin in 5% CO 2 humidi ed atmosphere at 37 °C. After cultured for 24 hours, most of the cells were veri ed to be activated by examining α-SMA expression with immuno uorescent cytochemistry.

Binding Characteristics of pPB Cyclic Peptide to Human HSCs
Human HSC-LX2 were incubated with the primary antibodies against α-SMA at 4 °C overnight. After the cells were incubated with Texas red-conjugated secondary antibodies (Molecular Probes), they were incubated with 1 µmol/L of pPB-FITC solution for 24 hours at 37 °C in the dark. Then the cells were rinsed and stained the nuclei with DAPI. After mounted, they were observed with a uorescence microscope.
In order to assess the binding e ciency of pPB cyclic peptide at different concentrations and different incubation durations to HSCs, human HSC-LX2 were incubated respectively with pPB-FITC solution at concentrations of 0, 0.04, 0.2, 1, 5, 25 and 125 µmol/L for 1 hour, or with 1 µmol/L solution for 0, 15, 30, 60, 90, 120 minutes and 24 hours at 37 °C in the dark. After incubation, these cells were washed by centrifugation at 1100 rpm in 4 °C, xed with 4% paraformaldehyde and immediately analyzed for the uptake rate by a FACS scan ow cytometer (FACSCalibur, USA) with CellQuest software (BD Biosciences).

Mice Model of Liver Fibrosis Induced by Bile Duct Ligation (BDL)
Liver brosis was induced in C57BL/6J mice by the ligation of common bile duct. 13 One week or four weeks after the treatment, treated mice were used for further experiments (referred to as BDL-1W and BDL-4W mice). And the mice subjected to sham were used as the control group.

Mice Model of Liver Fibrosis Induced by CCl 4 Treatment
Liver brosis was also induced in C57BL/6J mice by carbon tetrachloride (CCl 4 ) treatment (CCl 4 in olive oil, 1:9 (v/v), 2 ml/kg by intraperitoneal injection twice weekly). 14 Eight weeks or twelve weeks after the treatment, treated mice were used for further experiments (referred to as CCl 4 -8W and CCl 4 -12W mice).
Additionally, eight CCl 4 -8W mice were randomly selected to discontinue CCl 4 -treatment for another 4 weeks for the spontaneous regression of brosis (referred to as CCl 4 -8W + S4W mice). Mice treated with the same dosage of olive oil for 12 weeks served as the control group.

Histological Analysis of Hepatic Fibrosis
After xed in neutralized formalin, liver sections were stained with hematoxylin and eosin or Masson.
Extent of liver brosis was semi-quantitatively scored according to the Isake staging criteria. 15 For a morphometric analysis of liver brosis, the Masson staining ( brotic) areas were measured as previously described. 16 In addition, serum alanine aminotransferase (ALT) levels and liver hydroxyproline content were determined using assay kits (JianCheng, Nanjing, China).

Quantitative real-time PCR (qRT-PCR) analysis
The hepatic mRNA level of PDGFR-β was quantitated using qRT-PCR analysis as described. 16 The forward primer for PDGFR-β was AATATAAGAGGAAGAGTTG, and the reverse primer was TATACCCAAGGATTTCTA. GAPDH was used as the control, and its forward primer was TCCCTCAAGATTGTCAGCAA, and the reverse primer was AGATCCACAACGGATACATT.

Immunohistochemistry Analysis
After washed by PBS, liver sections were performed heat mediated antigen retrieval in sodium citrate buffer (pH 6.0) by using microwave. After cooled and blocked with 10% serum for 1 hour, the sections were incubated with rabbit anti-mouse PDGFR-β antibody (1:100 in blocking solution, Millipore, Massachusetts, USA) overnight at 4 °C. A Biotin conjugated goat polyclonal anti-rabbit IgG (1:100 in blocking solution, Abcam) was used as secondary antibody. Ten elds (200×) from each section were randomly selected and recorded. The PDGFR-β positive-staining areas were respectively quanti ed with NIN Image 1.62 software, and the percentage of positive-staining areas in each eld was calculated.
Ten randomly selected amplifying elds (400×) in each section were assessed.
In Vivo MRI Studies MR imaging was performed by using a Biospec 70/20 MRI scanner (7.0 T) to assess the accumulation of pPB-DOTA-Gd in livers after the control mice and CCl 4 -treated mice (n = 3 per group) were injected intravenously with pPB-DOTA-Gd at a dose of 0.05 mmol/kg [Gd 3+ ] in a total of 0.25 mL PBS solution, as previously described. 17 Brie y, dynamic T1-weighed MR images of the liver were collected prior to and 30, 60, 90 and 120minutes after pPB-DOTA-Gd was injection. Coronal section images of the liver were acquired with a fast low-angle shot (FLASH) sequence. Hepatic signal intensity (T1 (liver)) and muscle signal intensity (T1 (muscle)) were measured from the region of interest and the relative hepatic signal intensity was denoted as T1 (liver) / T1 (muscle).

Statistical Analysis
Data were presented as the mean ± standard deviation (SD) and analyzed by a one-way analysis of variance followed with least signi cant difference (LSD) test. SPSS 16.0 statistical software (Chicago, USA) was used. A p value less than 0.05 was considered statistically signi cant.

Expression of PDGFR-β in Fibrotic Livers Induced by BDL
After mice were treated with BDL for 1 weeks (BDL-1W), marked cholestasis and neutrophil in ltration were observed in hepatic lobules, and serum ALT level was signi cantly increased compared to the control mice ( Fig. 1A and 1B). And after mice were treated with BDL for 4 weeks (BDL-4W), the normal architecture of hepatic lobule was damaged, and visible brotic septa and extensive bridging brosis were observed (Fig. 1A). Compared to the control mice, the Masson staining ( brotic) area and the hydroxyproline content in liver tissue were both increased in BDL-treated mice, which was the highest in BDL-4W mice ( Fig. 1C and 1D). These results indicated that liver brosis developed and progressed after mice were treated with BDL.
As shown in Fig. 1E, hepatic mRNA level of PDGFR-β was both signi cantly higher in BDL-treated mice than in the control mice, which was the highest in BDL-4W mice. After hepatic slices were stained by immunohistochemistry, PDGFR-β positive-staining area was hardly visible in the control mice. But as liver brosis developed and progressed, PDGFR-β was found to be massively expressed in livers of BDL-treated mice, especially in BDL-4W mice (Fig. 1F). The results indicated that hepatic PDGFR-β expression was signi cantly increased with the development and progression of liver brosis induced by BDL.

Expression of PDGFR-β in Fibrotic Livers Induced by CCl 4treatment
Liver brosis was also induced in mice by CCl 4 -treatment in order to further investigate hepatic PDGFR-β expression in mice with the progression and regression of liver brosis. After mice were treated with CCl 4 for 8 weeks (CCl 4 -8W), massive hepatocytes balloon-like degenerative change and extensive brotic septa were observed. When CCl 4 -treatment was extended to 12 weeks (CCl 4 -12W), hepatic brosis was markedly aggravated and hepatic lobular architecture was observed to be extensively damaged. But when CCl 4 -treatment was withdrawn for 4 weeks after CCl 4 intoxication for 8 weeks (CCl 4 -8W + S4W), hepatocytes balloon-like degenerative changes signi cantly regressed and brotic septa markedly resolved ( Fig. 2A). Compared to the control mice, serum ALT level, the Masson staining area and the hydroxyproline content in liver tissue were all signi cantly increased in CCl 4 -treated mice. However, when CCl 4 -treatment was withdrawn for 4 weeks, these indexes were all markedly reduced ( Fig. 2B-D), which indicated that liver brosis signi cantly regressed after CCl 4 -treatment was withdrawn.
Compared to the control mice, hepatic mRNA level of PDGFR-β was signi cantly increased in CCl 4treatment mice, which was the highest in CCl 4 -12W mice, but was markedly reduced in CCl 4 -8W + S4W mice (Fig. 2E). As shown in Fig. 2F, after immunohistochemistry staining, extensive PDGFR-β positivestaining area was observed in livers of CCl 4 -treatment mice, especially in CCl 4 -12W mice. However, there was far fewer PDGFR-β positive-staining area in livers of CCl 4 -8W + S4W mice. These results indicated that hepatic PDGFR-β expression is increased with the development and progression of liver brosis, and reduced with the regression of brosis.
Additionally, no matter in BDL mice or in CCl 4 -treated mice, the PDGFR-β positive-staining area was observed to be mostly located in the brotic septa ( Fig. 1F and Fig. 2F).
Because α-SMA are thought to be the marker of activated HSCs, the cardinal cells expressing PDGFR-β in brotic livers are considered to be activated HSCs.

Expression of PDGFR-β in Activated HSCs in vitro
Double immuno uorescent staining was performed to investigate the expression of PDGFR-β on human activated HSC in vitro. After being sub-cultured for 24 hours, HSC-LX2 transformed into an activated cell type which was positive for α-SMA staining. Most of these culture-activated HSCs were observed to be positive for PDGFR-β staining (Fig. 4A). The result demonstrated that PDGFR-β was abundantly expressed on activated human HSC in liver brogenesis.
Binding Characteristics of pPB Cyclic Peptide to Activated HSCs pPB cyclic peptides were prepared and labeled with FITC (pPB-FITC). The purity of pPB and its derivative was above 95%. The molecular weight was 873 for pPB and 1300 for pPB-FITC.
At rst, the binding of pPB-FITC to HSC-LX2 was determined by confocal microscopic imaging. The uorescent signal (green) was clearly visible in the culture-activated HSC-LX2 after incubated with 1 µmol/L of pPB-FITC for 24 hours (Fig. 4B), which indicated that pPB-FITC bound to activated HSCs.
Second, when activated HSC-LX2 were incubated with pPB-FITC in a series of increasing concentration from 0.04 µmol/L to 125 µmol/L for 1 hour, their uorescence intensity was accordingly increased to approximately 6.0 to 500.0-fold (Fig. 4C). Additionally, when activated HSC-LX2 were incubated with 1 µmol/L of pPB-FITC for 15 minutes to 24 hours, an approximately 9.0 to 60.0-fold increase in their uorescence intensity was noted accordingly (Fig. 4D). These results indicated that the binding of pPB-FITC to activated human HSCs was concentration-dependent and time-dependent.
Imaging the Progress of Liver Fibrosis by MRI pPB cyclic peptides were labeled with Gd through DOTA (pPB-DOTA-Gd) and its molecular weight was 1554.
A dynamic T1-weighed MR imaging approach was used to assess the deposition of pPB-DOTA-Gd in the livers after the radiotracers were intravenously injected in the control and CCl 4 -treated mice. In CCl 4 -12W mice, the relative liver T1-MR signal intensity was gradually intensi ed until 60 minutes post-injection of pPB-DOTA-Gd, and then gradually reduced (Fig. 5A). There was no obvious change of hepatic signal intensity in the control mice and CCl 4 -8W + S4W mice prior to and 60 minutes after pPB-DOTA-Gd was injected; whereas hepatic signal intensity was markedly intensi ed in CCl 4 -12W mice and moderately intensi ed in CCl 4 -8w mice at 60 minutes post-injection of pPB-DOTA-Gd (Fig. 5B). Compared to the control mice, the relative liver T1-MR signal intensity was signi cantly increased in CCl 4 -treated mice, which was the highest in CCl 4 -12W mice (Fig. 5C). However, there was no signi cantly difference in the relative liver T1-MR signal intensity between the control mice and CCl 4 -treatment withdrawing mice.
These results indicated that hepatic deposition amount of pPB-DOTA-Gd 60 minutes after injection was gradually increased in parallel with the development and progression of liver brosis, and markedly reduced with the regression of brosis. The correlation was further assessed between the extent of liver brosis and the relative liver T1-MR signal intensity 60 minutes after pPB-DOTA-Gd were injected. The relative liver T1-MR signal intensity showed a strong positive correlation with Ishake stage (r = 0.858, p < 0.01, Fig. 5D) and Masson-staining area (r = 0.878, p < 0.01, Fig. 5E).

Discussion
At present, liver brosis can be assessed by invasive biopsy and non-invasive approaches including imaging techniques based on measuring liver stiffness (such as transient elastography) and serum markers. 18 Liver biopsy is still widely acknowledged as the gold standard for the diagnosis and stage of liver brosis, but it is an invasive procedure with a risk of rare but potentially life-threatening complications, which limit its use in the long-time follow-up and wide-scale screening of CLD. 19 Over the past decade, there has been tremendous advance in the development of non-invasive approaches to assess liver brosis. Most of these non-invasive approaches show good work at identifying the extremes of brosis but is undesirable in accurately differentiating intermediate stages. [18][19][20] In addition, liver brosis is a dynamic process, and the total amount of brous tissue depends on the synthesis, deposition, accumulation and degradation of ECM. 21 More importantly, in the past decades, it has been convincingly demonstrated that after the etiologies are successfully removed, the brotic liver can revert to a less brotic or even normal architecture in both rodent model of liver brosis and patients with CLD, especially at the early stage of brosis. 22,23 Hence, it may be more rational for the management of CLD to identify liver brosis at an advance stage or at a remission stage than to diagnose and stage brosis simply, which is crucial for clinicians to decide whether to take more aggressive measures to deal with CLD. So far, however, neither liver biopsy nor the existing non-invasive approaches can accurately predict the progress of liver brosis.
The signal transduced by PDGFR-β is acknowledged to be the most potent among the mitogenic pathways in HSCs. Kocabayoglu P et al. found that hepatic brosis accumulation was increase at least in part through increased HSC numbers by PDGFR-β activation and the lack of PDGFR-β on primary HSCs led to decreased expression of collagen I. 9 In this study, both in BDL-induced and in CCl 4 -induced mouse model of liver brosis, hepatic expression level of PDGFR-β was found to be signi cantly increased with the development and progression of brosis. The two kinds of mouse model with liver brosis mimic different pathogenesis of brosis, which is induced by cholestasis and the other is by toxin-induced hepatocellular injury. More importantly, when CCl 4 -treatment was withdrawn for 4 weeks after CCl 4 intoxication and liver brosis was found to be obviously regressed, hepatic expression level of PDGFR-β was observed to be markedly reduced. Thus, hepatic PDGFR-β expression was veri ed to be parallelly associated with the process of liver brosis.
Activated HSCs are the major ECM-producing cells in the liver, and it seems crucial for predicting the prognosis of CLD to distinguish the continuous or repeated activated state of HSCs. To date, there is still no non-invasive approach to distinguish the status of HSCs, and pathological examination is the only mean to identify the status of HSCs. In the study, the positive staining of PDGFR-β was observed to be mainly overlapped with α-SMA staining no matter in livers with mild brosis or with advanced brosis, so activated HSCs were the cardinal cells expressing PDGFR-β in brotic livers. In vitro, culture-activated human HSCs (LX2) were also observed to express abundant PDGFR-β. Thus, hepatic expression level of PDGFR-β was strongly associated with the number of activated HSCs. The continuous or repeated activation of HSCs leads to the development and progression of liver brosis. 21,22 Additionally, for the reversibility of liver brosis, it has been demonstrated to be critical to reduce the number of activated HSCs by apoptosis and deactivation. [22][23][24][25] Hence, it was feasible to identify the progression or regression of liver brosis by re ecting the number of activated HSCs with visualizing hepatic expression level of PDGFR-β.
pPB cyclic peptides can bind speci cally to PDGFR-β and show a well targeting property to activated HSCs. In our previous study, the surface modi cation with pPB was found to enhance the target effect of interferon-γ liposomes to activated HSCs, and increase the anti-brotic effect of interferon-γ. 11 In the present study, pPB labeled with FITC was observed to bind to activated HSC-LX2 and the binding was in a concentration-dependent and time-dependent manner. In this study, MRI was used to image the livers for it is acknowledged to be the highest spatial resolution to soft tissues among the existing imaging modalities in clinic. After pPB was labeled with Gd and used as an MRI tracer, mice with different extent of liver brosis showed signi cantly differences in the relative liver T1-MR signal intensity, which was increased with the development and progression of brosis and reduced with the regression of brosis. Thus, the molecular imagining approach can be helpful to evaluate whether CLD would continue to aggravate or get improvement after certain treatments by predicting the progression or regression of brosis.

Conclusion
In the present study, hepatic expression level of PDGFR-β was observed to be closely related with the progress of liver brosis, which was signi cantly increased with the development and progression of brosis and reduced with the regression of brosis. Activated HSCs were found to be the cardinal cell expressing PDGFR-β in brotic livers. After pPB cyclic peptides were labeled with Gd and used as an MRI tracer, the progress of liver brosis was identi ed by visualizing hepatic PDGFR-β expression with an MR imaging modality.

Declarations
Ethics approval and consent to participate All animal procedures were performed under the guidelines of the institutional review board and the ethics committee of Zhongshan Hospital, Fudan University.

Consent for publication
All the authors have approved the manuscript and agree with submission to your esteemed journal.

Availability of data and materials
All data generated or analyzed during this study are included in this published article.  Expression of PDGFR-β in brotic livers treated with CCl4. Liver brosis was induced in mice by treatment with CCl4 (CCl4 in olive oil, 1:9 (v/v), 2 ml/kg by intraperitoneal injection twice weekly) for 8 and 12 weeks