The study was approved by our Institutional Ethics Committee (Laboratory of Stem Cell Research and Application, University of Science, Vietnam National University (VNU)-Ho Chi Minh City (HCMC)). Healthy, 8-week old male Balb/c mice from Pasteur Institute in HCMC (Vietnam) were kept in a stable environment of 12 hours light-dark cycle in the micro-ventilation cage system (THREE-SHINE Inc., Korea) with ad libitum access to food and water, and were acclimated for 1 week prior to the operation.

In this study, the BDL procedure was performed as described by Carmen Tag et al. 23. The common bile duct was ligated between sutures and divided with care to avoid injury to the portal vein and the pancreas. Intramuscular (i.m.) administration of 10 mg/kg Ilium xylazil-20 (Troy Laboratories, Australia) was used as a sedative, followed by 7 mg/kg of Zoletil (Virbac, France). Lincomycin (Vemedim, Vietnam) at 20 mg/kg x 2 doses/day was given for 3 days i.m. to prevent post-surgery infections. BDL mice with visible jaundice at 7 days post-operation (d.p.o) were divided into three treatment groups (n=5):

(1) Sham-surgery treated with 200 μl/day of NaCl (0.9% solution);

(2) BDL with Placebo as 200 μl/day of NaCl (0.9% solution); and

(3) BDL + GCSF (10 µg/kg/day; Neupogen® Syringe (Filgrastim) (Roche, Switzerland)).

All treatments were administered subcutaneously (s.c.) starting on d.p.o 8 for 5 consecutive days, and mice were humanely sacrificed the day after the last dose of GCSF (on day 13 post surgery).

Blood samples were taken from the facial vein. Leukocyte and neutrophil counts were calculated by trained and blinded observers using the blood film technique. In brief, a drop of the blood sample was smeared upon a glass slide, air-dried for 3 minutes, fixed in absolute methanol for 5 minutes, and then stained with Giemsa's azure eosin methylene blue solution (Merck Millipore, Germany). Serum aspartate aminotransferase, alkaline phosphatase, and albumin levels were assayed using the GOT (ASAT) IFCC mod. liquiUV kit (HUMAN Diagnostics, Germany), QuantiChrom™ Alkaline Phosphatase Assay Kit (BioAssay Systems, USA), and QuantiChrom™ BCG Albumin Assay Kit (BioAssay Systems, USA), respectively. Data were acquired and analyzed by the DTX 880 Multimode Detector system (Beckman Coulter, USA).

Total RNA was extracted from the median lobe of the liver with easy-spin (DNA-free) Total RNA Extraction Kit (iNtRON, Korea). The cDNA was synthesized using SensiFAST™ cDNA Synthesis Kit (Bioline, UK). Real-time RT-PCR analyses were performed using SensiFAST™ SYBR® Hi-ROX Kit (Bioline, UK), and then data were acquired and analyzed on the LightCycler® 480 Instrument II (Roche Life Science, USA) in reference to the Livak's 2^-ΔΔCt method 24. Primers used in this study are shown below:

For immunohistochemistry and Western blot analysis, the primary antibodies used were anti-cytokeratin 7 (ab181598) and smooth muscle actin-a lpha (ab15734) (both from Abcam, MA, USA), β-actin (13E5, Cellsignaling, MA, USA). The secondary antibody was horseradish peroxidase-conjugated secondary goat-anti-rabbit (ab6721, Abcam, MA, USA). Liver tissue used in these analyses was from the median lobe.

The IHC procedure was previously described 25. In brief, slides were blocked in goat serum blocking buffer (2% Goat serum, 1% BSA, 0.1% Triton X100, 0.05% Tween-20 in phosphate-buffered saline, pH 7.2-7.4, with 0.05 % sodium azide) before incubating with primary α-smooth muscle actin (α-SMA) (1:200) or anti-cytokeratin 7 (CK7) antibody (1:5000). Slides were incubated with peroxidase blocking buffer, i.e. 3% hydrogen peroxide (H 2 O 2 ) in PBS, for 10 minutes at RT to block intracellular peroxidase, and then incubated with secondary antibody (1:500) in Tris-buffered saline (TBS) solution with 1% bovine serum albumin (BSA) for 1 h at RT. Immunocomplexes were developed using the ACE kit (Sigma-Aldrich, St Louis, MO). Hematoxylin-Gill III (Merck Millipore, Germany) was used for counterstaining. To quantify the CK7-positive ductal structures around the portal area, 10 random images/slide x 3 slides/mice were captured at x200 magnification. The CK7-positive bile duct densities were calculated as the number of CK7-positive bile ducts divided by tissue surface area.

Liver tissue was instantly frozen in liquid nitrogen and preserved at -80^o C until use. Ten mg of tissue was digested and then protein concentration was identified by Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, USA) using DTX 880 Multimode Detector system (Beckman Coulter, USA). Protein samples were subjected to SDS-PAGE on the SE 250 Minivertical Unit (GE Healthcare Life Sciences, USA) at 100 V for 3 h. T he protein bands were transferred into the Immun-Blot® PVDF Membrane (BIO-RAD, Singapore) in the Western blot blotting tank (GE Healthcare Life Sciences, USA) at 100 V for 3 h. The membrane was pre-blocked with TBS (containing 0.1% Tween 20), 5% skimmed milk for 30 minutes before incubation with the appropriate antibodies (primary α-SMA (1:1000) or CK7 (1:5000) or beta-actin (1:6000)) dissolved into TBS (containing 0.1% Tween 20) and 0.5% skimmed milk at 4 o C for 12 h. Signals were developed using Clarity™ Western ECL Kit (BIO-RAD, USA) and visualized by Azure c300 Chemiluminescent Western Blot Imaging System (LifeGene, Israel). The band intensity was measured by the free software Image Studio™ Lite (LI-COR Biosciences, USA).

Mouse liver tissue was obtained from the median lobe, fixed in 4% paraformaldehyde (Merck Millipore, Germany), and sectioned from the core of the tissue mass. The tissue was paraffin-embedded and cut into 4 µm slices, with each cross-section approximately 1 mm apart.

Hematoxylin-Eosin staining : Liver slides were stained with Papanicolaou's solution 1a (also called Harris' hematoxylin solution) (Merck Millipore, Germany) and Eosin Y (0.5 %) aqueous solution (Merck Millipore, Germany) according to the protocols afore mentioned.

Picrosirius red staining and relative-quantification of the collagen-positive area: Liver slides were stained with Picrosirius Red Stain Kit (Polysciences, PA, USA) according to the manufacturer’s instructions. Collagen accumulation around the portal areas and inside the liver parenchyma (excluding the blood vessels) of the liver lobule (at 10 lobules/slice x 5 slices/mouse) was determined by switching the captured images into the green channel in the RGB stack mode to highlight the red-stained collagen (ImageJ 1.51r software). A trained and blinded observer analyzed the area percentage of positive sites.

Data in this study were analyzed and performed by the software GraphPad Prism 6 (GraphPad Inc ., USA). For survival analysis, the Log-rank (Mantel-Cox) test was used. Differences were considered as statistically significant if p-value ≤ 0.05. For boxplot comparison, the difference between medians (DBM) and overall visible spreading (OVS) was calculated. The two groups were considered significantly different if DBM/OVS was ≥ 50%.

The calculations were as follows: difference between medians (DBM)= $\left|median\left(1\right)-median\left(2\right)\right|$; overall visible spread (OVS)=Q3-Q1 in which Q3 is the larger 3^rd quartile value and Q1 is the smaller 1^st quartile value.

At 7 days, BDL mice showed jaundice and ruffling (Figure 1a), and visible bile duct obstruction with full gallbladder at sacrifice (Figure 1b). The serum activity of aspartate transaminase (AST) significantly increased (p=0.0022), with a decline in albumin (ALB) concentration, when compared to those of sham mice (Figure 1c) (p=0.03).

Figure 1

After 5 days of GCSF, consistent with GCSF ’s effect on mobilizing bone marrow cells, the total serum leukocyte count in GCSF-treated mice reached 32.72±13.49 x 10⁶ cells/ml (n=5), while that of the placebo was 10.46 ± 4.99 x10⁶ cells/ml (n=4) (p<0.05) (Figure 2a). The neutrophil count in GCSF-treated mice was 27.83±14.41x10⁶ cells/ml (n=5), compared to 6.48±2.87 x 10⁶ cells/ml (n=4) in the placebo group (p<0.05) (Figure 2b). The corresponding median survival of GCSF-treated BDL mice (n=10 for each group) was 13.0 days, as compared to 10.5 days for placebo-BDL mice, with a hazard ratio (log -rank) of 1.88 (Figure 2c) (p= 0.08 nearing statistical significance).

There was a decrease in serum aspartate aminotransferase (AST) concentration in BDL mice treated with GCSF (median serum AST: 116.33 IU/L), compared to those BDL mice treated with placebo (median serum AST: 297.76 IU/L) (DBM/OVS=51.23%) (Figure 3a). This was not seen with serum ALP or albumin values (Figure 3b, c). Though positive results were seen, these indicators did not fully reflect the disease condition in treated mice and were not significant different (Figure 3, p>0.05); this was expected as bile duct obstruction was fixed26.

Figure 3

Down-regulation of mRNA levels of pro-fibrotic genes in GCSF-treated BDL mice were observed, with notable decrease in the levels of TGF-β mRNA (2.61 fold-change, DBM/OVS=85.99%), α-SMA (2.46 fold-change, DBM/OVS=22.83%), and col1a1 (3.28 fold-change, DBM/OVS=69.41%) (Figure 4). A parallel reduction in the amount of collagen deposit by 1.79-fold (p=0.006) was seen around and across portal areas (bridging fibrosis) of the GCSF treated BDL mice compared to placebo (Figure 5).

Figure 5

Similarly, consistent with the down regulation of α-SMA expression, Western blot data showed a 10.46±0.36 and 5.56±0.81 fold-change for BDL and GCSF-BDL groups, respectively, compared to that for sham mice (p<0.0001) (Figure 6). Immunostaining confirmed a decreased periportal α-SMA expression in GCSF-treated liver of BDL mice (Figure 7).

Figure 6

Figure 7

There was significant improvement in the liver histology of GCSF-treated BDL mice compared to that of BDL mice, with fewer infarction zones (Figure 8) from BDL bile necrosis. Moreover, as previously mentioned, there was less bridging fibrosis (Figure 5d: -1.79-fold change, p=0.006) around the expansion areas of ductal structures (Figure 9d: -1.77-fold change, p<0.0001) in the GCSF-treated BDL mice.

Figure 9

In compensation to hepatic injury, bile duct progenitor cells proliferate to expand the ductal structures 27. This was seen in the portal areas of BDL liver with a high number of cytokeratin-7 (CK-7) positive newly formed ductal structures (Figure 9b) (41.12 ± 1.1 bile ducts/mm²). GCSF treatment reduced the number of CK7 positive bile ducts (23.2 ± 2.75 bile ducts/mm²) (1.77-fold change, p<0.0001), as well as decreased whole liver CK7 protein levels (2.33-fold reduction, p<0.0001) (Figure 10).

Figure 10

In this study, bile duct ligation (BDL) leads to obstructive jaundice, hepatomegaly, increased hepatic transaminase levels, and decreased liver functions (such as albumin levels). Since Balb/c mice are more vulnerable to hepatic fibrosis 28 and liver injury, with a high mortality by 14 days of surgery compared to other strains of mice, this study was limited to a 12-day course after mice exhibited liver injury by the 7^th day and after completion of the 5 doses of GCSF treatment. In this study, administration of 5 doses of 10 µg/kg/day to BDL Balb/c mice increased the leukocyte and neutrophil counts in peripheral blood of these mice, compared to that for placebo-treated mice (p<0.05). Our results showed that administration of GCSF improved the mortality in GCSF-treated BDL mice compared to the placebo (hazard ratio=1.88)27. In this study, GCSF-treated BDL mice had low AST activities compared to placebo-treated BDL mice, suggesting a less injured liver 21,27. Furthermore, GCSF diminished the liver fibrosis via down regulation of pro-fibrotic genes, such as TGF-β1, α-SMA, and col1a1. This was consistent with reduction of α-SMA protein level (confirmed by IHC and Western blot) and collagen deposit (confirmed by Sirius red staining) 21.

This model of liver injury adds to the list of other experimental models of liver injury, including partial hepatectomy in rats 22, radiation 29, acute failure from thioacetamide (TAA) 30, D-galactosamine31, CCl₄ 32 and NAFLD 33 publications show the beneficial effects of GCSF on the attenuation of hepatic injury (adjust references). Underlying the attenuated hepatic injury are various phenomena, including the reduction of the inflammatory response, the promotion of endogenous repair 21, increase in hepatocyte proliferation 34,35 with increased PI3K/Akt pathways 33, development of oval cell proliferation 22, and reduction in the apoptotic drive33.

During liver injury, a correlation between ductular reaction and fibrogenesis, presumably from the extracellular synthesis of the newly forming bile ducts, have been reported 36,37,38. Inhibition of the ductular response reduced liver fibrosis in MDR2 ^-/- mice 39. In this study, G-CSF significantly reduced cytokeratin-7 (CK7) expression and the number of CK7-positive ductal structures in BDL mice, providing a novel effect of GCSF on improving liver injury by diminishing the ductular reaction. Future studies into the mechanism of GCSF protective effects in liver injury should target the relationship between GCSF and the ductular response. It is worth noting that the histopathological features of biliary atresia include portal fibrosis and inflammation, but also ductular proliferation 39. Therefore, these evidence provides a rationale for the use of GCSF as an intervention for BA.

In this experimental model of intermediate-term liver damage by obstructive BDL injury, post-injury treatment with GCSF improved the hepatic outcome of bile duct-ligated mice, notably a reduction in hepatic fibrosis and its association with down regulation of the ductular reaction. GCSF may be useful as a treatment for liver fibrosis in BA disease.

ALB: Albumin

ALP: Alkaline photphatase

AST: Aspartate transaminase

BA: Biliary Atresia

BDL: Bile duct ligation

CK7: Cytokeratin-7

GCSF: Granulocyte colony stimulating factor

HSC: Hematopoietic stem cell

IHC: Immunohistochemistry

MELD: Model for End-stage Liver Disease

NAFOSTED: National Foundation for Science and Technology Development

αSMA: alpha Smooth Muscle Actin

TGFβ: Transforming Growth Factor beta

Huy Quang Do conducted the experiments and composed the manuscript. Trinh Van Le, Minh Thanh Dang, Tien-Trieu Pham-Le and Luan Van Chan carried out the experiments and analyzed the liver function, gene-expression and histology. Khon Chan Huynh performed and analyzed the protein expression. Ai-Xuan Le Holterman is senior researcher, advices on the design of experimental models and on data analyses and revise the manuscript. Nhung Hai Truong made substantial contributions to the ideas, experiment design , interpretation of data, and submission.

The authors declare that they have no competing interests.

This research is funded by Vietnam National Foundation for Science & Technology Development (NAFOSTED) under grant number 108.05-2017.30 and Science & Technology Incubator Youth Program, managed by the Center for Science and Technology Development, Ho Chi Minh Communist Youth Union.