Granulocyte-Macrophage Colony-Stimulating Factor Protects Dimethylnitrosamine-Induced Rat Liver Fibrosis By Inhibiting Transforming Growth Factor-β1 Signaling Pathway

Granulocyte-macrophage colony-stimulating factor (GM-CSF) exerts several therapeutic pharmacological effects but its role in liver brosis has not yet been studied. The current study investigates the inhibitory effects of GM-CSF on dimethylnitrosamine (DMN)-induced liver brosis in rats. In this study, liver brosis was induced in Sprague-Dawley rats by intraperitoneal injections of DMN (10 mg/kg of body weight) for three consecutive days per week for four weeks. To see the inhibitory effects on disease onset, GM-CSF (50 µg/kg of body weight) was injected for 2 consecutive days per week for 4 weeks along with DMN, while to see the therapeutic effects on disease progression, the GM-CSF injection was set forth at 4 weeks after the DMN injection. We found that DMN administration produced characteristics of molecular and pathological manifestations of liver brosis in rats including increased expressions of collagen I, alpha-smooth muscle actin (α-SMA), and transforming growth factor beta 1 (TGF-β1), and decreased PPAR-γ expression. Similarly, elevated serum levels of aspartate aminotransferase (AST), total bilirubin level (TBIL), and decreased albumin level (ALB) were observed. Treatment with GM-CSF improved the pathological liver conditions and signicantly inhibited the elevated AST and TBIL, and increased ALB serum levels to normal. GM-CSF signicantly decreased collagen I, α-SMA, and TGF-β1 expression and increased peroxisome proliferator-activated receptor gamma (PPAR-γ) expression. In conclusion, GM-CSF reduced the DMN-induced rat liver brosis by inhibiting TGF-β1 signaling pathway. our previous study showed that GM-CSF inhibited the TGF-β-induced Rho-ROCK pathway and also rescued excessive collagen production in vivo 28 . In the present study, we developed a DMN-induced liver brosis in rats and examined the antihepatobrotic effects of GM-CSF on liver brosis. The underlying molecular mechanisms of GM-CSF in reducing liver brosis are discussed.


Introduction
Liver brosis -caused by chronic liver injuries -is one of the major causes of mortality worldwide 1 . A global study has reported that liver brosis accounted for 2.2% of death worldwide 2 . Liver brosis is a wound-healing response to chronic liver injuries resulting in excessive accumulation of extracellular matrix (ECM) proteins in hepatic tissues. Progressive liver brosis advances to cirrhosis, liver failure, and portal hypertension eventually leading to hepatic dysfunction 3 . Various etiologies such as chronic viral hepatitis, fat accumulation, alcoholic and non-alcoholic hepatic injuries, and toxin/drug-induced metabolic or autoimmune diseases that trigger reiterative damage to liver tissue have been implicated in liver brosis 4,5 . Dimethylnitrosamine (DMN) is a well-known hepatotoxin for inducing liver brosis in rats 6 . The DMN-induced liver brosis model closely resembles liver damage development in humans which includes nodular regeneration, ascites, ECM deposition, biochemical alterations, and histopathological manifestations 7 .
Pathophysiologically, chronic hepatic injuries initiate the production of various brogenic cytokines in the liver tissue. Continued production of brogenic cytokines, such as target transforming growth factorbeta (TGF-β), connective tissue growth factor (CTGF), and platelet-derived growth factor (PDGF), transdifferentiates the quiescent nonparenchymal HSCs into brogenic myo broblast-like cells 8 . Phenotypic transformation of HSCs into active myo broblast-like proliferative state triggers the production of a massive amount of ECM proteins, primarily brillar collagens, bronectin, and α-SMA; and ultimately leads to hepatic brosis [9][10][11] . Activated HSCs are also responsible for the proliferation and migration of phenotypically transformed broblasts and the binding of TGF-β1 to its receptor that triggers the migration of such broblasts 12 . The binding of TGF-β1 with the type II receptor results in recruitment and phosphorylation of the type I receptor and phosphorylates Smad2 or Smad3 downstream. Phosphorylated Smad2 and Smad3 bind to Smad4 to form a heterotrimeric complex. This complex translocates to the nucleus and transcribes genes involved in ECM synthesis and deposition 13 . Owing to the importance of TGF-β signaling, ECM synthesis, and HSCs transformation in liver brosis pathophysiology, several anti-brotic strategies that target the ECM 14 and HSCs 15 , and stem cell-based therapies 16 that target TGF-β1 17 and enhance anti-brotic e cacy 18 are the state-of-the-art therapeutic modalities for the treatment of liver brosis.
Granulocyte-macrophage colony-stimulating factor (GM-CSF) is a multipotent cytokine synthesized by many cell types, including macrophages, lymphocytes, broblasts, and endothelial cells 19 . GM-CSF has been implicated in a multitude of biological functions such as chemotaxis of in ammatory cells to wound sites 20 ; proliferation and differentiation of early hematopoietic progenitor cells 21 , epithelial regeneration 21 , and wound-healing and neovascularization 22 . GM-CSF plays a complex tissue-dependent role in brosis 23 , and has antiviral and immunoregulatory effects in patients with chronic hepatitis B 24 .
GM-CSF has shown hepatic regeneration after 70% hepatectomy by promoting the hepatocellular DNA synthesis in a rat model 25 .
Previously, we have demonstrated that GM-CSF inhibits glial formation and exhibits a long-term protective effect after spinal cord injury 26 . Additionally, we have shown that GM-CSF offers therapeutic potential for the remodeling of vocal fold (VF) wounds and the promotion of VF regeneration 20 , and in the mobilization of bone marrow mesenchymal stem cells stimulation 27 . Furthermore, our previous study showed that GM-CSF inhibited the TGF-β-induced Rho-ROCK pathway and also rescued excessive collagen production in vivo 28 . In the present study, we developed a DMN-induced liver brosis in rats and examined the antihepato brotic effects of GM-CSF on liver brosis. The underlying molecular mechanisms of GM-CSF in reducing liver brosis are discussed.

Experimental design
The study was carried out in compliance with the ARRIVE guidelines (https://arriveguidelines.org) . The  experimental protocols for the animal study were approved by the Inha University Institutional Animal   Care and Use Committee (INHA-IACUC, approval ID: INHA 170228-484-2) on their ethical procedures and scienti c care. All the animals were treated strictly following the approved protocols. Sixty male Sprague-Dawley rats (8 weeks, 300 g) were purchased from Orient-Bio (Gyeonggi-do, South Korea). Rats were housed in a pathogen-free animal facility under 12 h light/dark cycle at constant temperature and humidity throughout the experiment. All rats were fed with standard rat chow with access to tap water ad libitum. After a week of acclimatization, the rats were assigned randomly into six groups (n = 10 for each group). Liver brosis was induced in the rats by intraperitoneal (IP) injection of DMN (10 mg/kg body weight). DMN was administered for the initial four weeks (three alternative days per week) and sacri ced either after 4 or 8 weeks after DMN administration.
For GM-CSF treatment, a group of DMN-injected rats received GM-CSF (50 µg/kg body weight, IP injection) right from the day of DMN administration for four weeks (two alternative days per week). The other group received GM-CSF (50 µg/kg body weight, IP injection) only from the 5 th week after DMN administration. For the sham control group, an equal volume of saline (0.9%, IP injection) was injected. The experimental scheme is shown in Figure 1 where various groups were assigned as follows: 1. Control-4w (administered saline for four weeks; sacri ced at the end of 4 th week); 2. DMN-4w (administered DMN for four weeks; sacri ced at the end of 4 th week); 3. DMN+GM-4w (administered DMN and GM-CSF for four weeks; sacri ced at the end of 4 th week); 4. Control-8w (administered saline for the initial four weeks; sacri ced at the end of 8 th week); 5. DMN-8w (administered DMN for the initial four weeks; sacri ced at the end of 8 th week); . DMN+GM-8w (administered DMN for the initial four weeks; treated with GM-CSF from the 5 th week until the subjects were sacri ced at the end of the 8 th week).
The body weight of each subject was measured three times a week plus one time shortly before excising out the liver from the anesthetized rats.

Serum biochemical analysis
Blood samples were collected from venous and arterial blood vessels and heart chambers after 4 or 8 weeks of DMN administration. After leaving the samples at room temperature for 30 min, blood samples were centrifuged at 3000 rpm for 10 min. The supernatant was collected and stored at −70 o C before further analysis. Blood serum was used to analyze aspartate aminotransferase (AST), albumin (ALB), and total bilirubin (TBIL) by spectrometry using commercially available kits.

Histopathological examination
On the nal day of the experiment, the whole liver was excised from anesthetized rats and weighed. Tissue samples were xed with 10% neutral formalin solution. Histopathological slides of the tissue samples were prepared by a certi ed histopathologist. Brie y, xed liver samples were embedded in para n blocks and 5 µm thick sections were prepared. Para n-embedded sections were depara nized and processed for Sirius Red and hematoxylin-eosin (H&E) staining. H&E and Sirius Red staining have been employed to evaluate the progression of liver brosis 42,43 .

Immunohistochemical examination
Thin liver tissue sections were prepared and mounted on slides, depara nized in xylene, and rehydrated in alcohol. The level of collagen I, α-SMA, TGF-β1, and PPAR-γ were determined by immunohistochemical staining using the corresponding primary antibodies for collagen type I, α-SMA, TGFβ1, or PPAR-γ at recommended concentrations according to the manufacturer's instructions.

Western blot analysis
Total protein was isolated from a part of liver tissues according to the manufacturer's instructions. Liver tissue was homogenized in 300 µL RIPA buffer [0.5% Nonidet P-40, 20 mM Tris-Cl (pH 8.0), 50 mM NaCl, 50 mM NaF, 100 μM Na 3 VO 4 , 1 mM dithiothreitol, 50 μg/mL phenylmethylsulfonyl uoride] containing protease inhibitors. The homogenates were centrifuged at 13,200 rpm for 30 min at 4°C. Protein (30 µg) was separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS PAGE) and electrotransferred to polyvinylidene di uoride (PVDF) membranes. The PVDF membranes were blocked with non-fat milk solution in Tris-buffered saline containing 0.1% Tween-20 (TBST) for 1 h at room temperature. The membranes were then incubated with primary antibody (α-SMA, collagen type I, TGFβ1, or β-actin) at recommended concentrations overnight at 4°C. After washing, the membranes were incubated with horseradish peroxidase (HRP)-conjugated anti-rabbit secondary antibody directed against the primary antibody for 1 h at room temperature. The membranes were detected by an enhanced Bio-Rad Western blot detection system (Bio-Rad Laboratories, Hercules, CA, USA).

Image analysis
The prepared histopathological or immunohistological slides were examined under a microscope (DMi8, Leica Microsystems Inc., Buffalo Grove, IL, USA). The acquired images were analyzed quantitatively using ImageJ software 44 . Brie y, the acquired images (n = 5 for each group) were deconvoluted and threshold adjusted. The percentage expression of the appropriate color was calculated and the averaged value (± standard error of means, SEM) was presented.

Statistical analysis
Statistical analysis Statistical analyses were performed using SPSS software (IBM SPSS Statistics, IBM Corp., Armonk, NY, USA). Data for each experimental group was presented as the means ± SEM (standard error of means). Statistical signi cance. To test whether there were any statistically signi cant differences between the means (± SEM) of the independent groups, one-way analyses of variance (ANOVA) were performed followed by Tukey's test as a post hoc test. Statistical signi cance was represented as *** for p ≤ 0.001, ** for p ≤ 0.01 and * for p ≤ 0.05, and ns (not signi cant). Likewise, there was a 28.10 (± 1.16) vs. 2.71 (± 1.16) and 19.68 (± 1.34) vs. 2.69 (± 1.34) folds increase in collagen I expression in the DMN-vs. DMN+GM-CSF-treated groups for the 4-and 8-week treatment groups, respectively, as compared with the corresponding sham-treated control groups ( Figure 4B).

Results
Expression of α-SMA, too, was increased by 15.21 (± 1.15) and 11.84 (± 0.59) folds and in the 4-and 8week-DMN-treated groups respectively as compared with the corresponding control groups, which dropped down to 1.67 (± 0.09) and 1.48 (± 0.26) folds respectively in the GM-CSF-treated groups ( Figure  4C). The increase in collagen I and α-SMA expression levels in the DMN-treated groups vs. control groups were statistically signi cant (p ≤ 0.001), and the decrease in collagen I and α-SMA expression in DMN vs. DMN+GM groups were statistically signi cant for both the 4-and 8-week treatment groups (p ≤ 0.001).

GM-CSF reduced DMN-induced liver damage and inhibited hepatotoxicity
Qualitative visual inspection of the excised livers showed that the liver lobes of the sham-treated control rats were brown, smooth, and soft with an evident glossy surface ( Figure 5A). By contrast, the surface of the DMN-treated livers was rough, coarse, hard, and shrunken with dark discoloration. However, liver morphology of the GM-CSF-treated group showed a relatively smoother surface, an enhanced brown texture without scars, and rendered a similar liver texture as that of the sham-treated rats. Quantitatively, the 4-and 8-week DMN-injected rats showed a signi cant decrease in liver weight. The average weight of the liver was 3.91 (± 0.09) g in Control-4w vs. 2.94 (± 0.15) g in DMN-4w groups and 3.87 (± 0.06) g in Control-8w vs. 2.66 (± 0.23) g in DMN-8w groups. The percentage decrease in liver weight was 24.85 (± 1.13) % and 25.91 (± 1.28) % in the DMN-4w and DMN-8w groups, respectively, compared with the corresponding sham-treated control groups ( Figure 5B). GM-CSF signi cantly improved the DMN-induced liver weight loss. The average weight of the liver was 3.58 (± 0.1) g in DMN+GM-4w and 3.52 (± 0.12) g in DMN+GM-8w groups. The percentage decrease in liver weight was only 11.18 (± 1.12) % and 9.94 (± 1.08) % in the DMN+GM-4w and DMN+GM-8w groups, respectively, as compared with the corresponding sham-treated control groups. The reduction of liver weight loss by GM-CSF was statistically signi cant in both DMN-4w vs. DMN+GM-4w and DMN-8w vs. DMN+GM-8w groups.

GM-CSF improved survival rate and increased body weight of DMN-treated rats
Control-4w and Control-8w groups showed 100% survival ( Figures 7A and 7B), but 40% of the rats died in the DMN-4w group, and 70% of the rats died in the DMN-8w group during the drug intervention process. Interestingly, GM-CSF treatment substantially increased the survival of the DMN-treated rats only in the 4week group, but not in the 8-week group. The DMN+GM-4w group showed a 100% survival ( Figure  7A), while the DMN+GM-8w group did not show any increase in survival ( Figure 7B).
Additionally, the 4-week administration of DMN signi cantly decreased the bodyweight of rats. On average, DMN-4w group rats lost 11.38 (± 4.68) % of their body weight compared to the baseline, while the Control-4w group gained 19.46 (± 2.46) % weight over four weeks ( Figure 7C). However, GM-CSF treatment prevented DMN-induced body weight loss. DMN+GM-4w group showed an 8.78 (± 1.46) % increase in body weight over four weeks of GM-CSF treatment. The average bodyweight of the rats in the DMN+GM-4w group, 350.8 (± 5.19) g, was signi cantly higher than those in the DMN-4w group, 289.17 (± 9.51) g, while their baseline weights were comparable, 322.5 (± 2.8) g and 326 (± 2.74) g respectively, ( Figure 7C). Similarly, the 8-week DMN-injected group, too, showed a dramatic decrease, −11.33 (± 7.12) %, in body weight as compared to the baseline. GM-CSF signi cantly recovered the body weight lost by DMN injection in four weeks ( Figure 7D). There was an increase in body weight after GM-CSF administration, but the increase was not statistically signi cant.

GM-CSF inhibited DMN-induced TGF-β1 expression
Qualitative visual assessment of immunohistochemical slides showed that DMN treatment signi cantly increased TGF-β1 expression, and GM-CSF treatment substantially lowered the DMN-induced increase in TGF-β1 expression in both the 4-and 8-week treatment groups ( Figure 8A). Quantitatively, DMN administration increased TGF-β1 expression by 11.05 (± 0.16) and 11.34 (± 0.09) folds in the 4-and 8week treatment-groups, respectively. TGF-β1 expression levels in GM-CSF-treated groups were comparable to those in the corresponding control groups in both the 4-and 8-week treatment groups ( Figure 8B). The difference in TGF-β1 expression levels between Control-4w and DMN-4w; and that between DMN-4w and DMN+GM-4w were statistically signi cant (p ≤ 0.001), but the difference in TGF-β1 expression levels between Control-4w and DMN+GM-4w was not statistically signi cant. Similarly, for the 8-week treatment groups, the difference in TGF-β1 expression levels between Control-8w and DMN-8w; and between DMN-8w and DMN+GM-8w were statistically signi cant (p ≤ 0.001), but the difference in TGF-β1 expression levels between Control-8w and DMN+GM-8w was statistically not signi cant.
Western blotting con rms the effects of GM-CSF Western blotting results showed that DMN signi cantly increased the expression of Collagen I, α-SMA, and TGF-β1; and GM-CSF signi cantly decreased the DMN-induced expression of Collagen I, α-SMA, and TGF-β1 in both the 4-and 8-week groups ( Figure 9A). Compared with the corresponding control groups, the folds-increase in Collagen I ( Figure 9B), α-SMA ( Figure 9C), and TGF-β1 ( Figure 9D Figure 10A). Quantitatively, DMN administration decreased PPARγ expression by 0.71 (± 0.03) and 0.35 (± 0.02) folds in the 4-and 8-week treatment-groups, respectively. PPARγ expression levels in GM-CSF-treated groups were comparable to those in the corresponding control groups in both the 4-and 8-week treatment groups ( Figure 10B). The difference in PPARγ expression levels between Control-4w and DMN-4w; and that between DMN-4w and DMN+GM-4w were statistically signi cant (p ≤ 0.001), but the difference in PPARγ expression levels between Control-4w and DMN+GM-4w was not statistically signi cant. Similarly, for the 8-week treatment groups, the difference in PPARγ expression levels between Control-8w and DMN-8w, and that between DMN-8w and DMN+GM-8w were statistically signi cant (p ≤ 0.001), but the difference in PPARγ expression levels between Control-8w and DMN+GM-8w was not statistically signi cant. Discussion DMN, a potent liver-speci c toxin, has remained one of the popular chemicals to induce liver brosis in animal models. It has been shown that chronic administration of DMN to rats produced advanced liver brosis with diffuse nodularity, marked portal hypertension, and accumulation of ascites 29 . Repeated injections of DMN bring about abnormalities in the biochemical and pathological manifestations of liver injury leading to liver brosis. Liver disease produced by DMN has been shown to closely represent human liver brosis. In the current study too, DMN injection induced liver brosis in rats in both the 4-and 8-week treatment schemes. DMN-treated rat livers exhibited several crucial pathological features of liver brosis including hardened and shrunken dark livers with a rough surface, an abnormal arrangement of hepatic plates, in ltrated in amed cells, and collagen ber deposition. In addition, serum biochemical indicators for liver in ammation such as AST, ALB, and TBIL were released into the bloodstream in DMNtreated rat livers.
The development and progression of liver brosis is a consequence of highly coordinated cellular and molecular processes that occur following chronic liver injuries subsequently leading to activation of the hepatic stellate cells (HSC). Activation of HSCs from its quiescent state is considered one of the hallmarks of hepatic brosis which induces accumulation of collagen and ECM deposition in the liver 30 .
Further activation of HSCs, marked by highly upregulated α-SMA 31 , leads to myo broblast-like phenotype and deposit large quantities of extracellular matrix (ECM) components in the liver 32 . Chronic exposure of DMN also leads to the increment in the α-SMA deposition, a widely accepted marker of transactivation of HSCs to myo broblasts 33 . In the current study too, DMN injections increased expression of Collagen I and α-SMA proteins in liver tissue in both the 4-and 8-week treatment groups indicating HSC activation.
TGF-β1 is one of the important cytokines which directly activates HSCs and incites the brotic process 34 . Pathophysiologically, upon activation, TGF-β1 binds to its receptors and initiates the SMAD-dependent and/or SMAD-independent signaling pathways downstream resulting in the expression of pro brotic genes such as those encoding α-SMA, ECM proteins, and secreted cytokines and growth factors leading to brosis 35 . DMN injection has been shown to activate TGF-β1 in rat liver 36 . Our data also show that DMN injection markedly increased TGF-β1 expression in rat livers in both the 4-and 8-week groups.
Owing to the signi cance of TGF-β1 activation in liver brosis progression, inhibition of the TGF-β1 pathway has remained one of the therapeutic strategies for liver brosis 37 .
GM-CSF has been shown to exert several therapeutic pharmacological effects but, to our knowledge, its role in liver brosis has not yet been studied. Based on our previous ndings on GM-CSF-induced inhibition of TGF-β1-dependent collagen synthesis in vocal fold scarring in rabbits 28 . we hypothesized that GM-CSF might also inhibit TGF-β1-dependent liver brosis. Consistent with our hypothesis, GM-CSF substantially reduced TGF-β1 expression in rat liver induced by DMN injections in 4-and 8-week treatment groups. Con rmed by both immunohistochemical staining and Western blotting analyses, TGF-β1 expression was signi cantly lesser in GM-CSF-treated groups as compared with that in DMN-treated groups. Further con rming the antihepato brotic bioeffects of GM-CSF, we found that GM-CSF signi cantly recovered the DMN-induced decrease in PPARγ expression. Upregulation of PPAR-γ by GM-CSF is crucial because when PPAR-γ is activated/upregulated, it suppresses the in ammation and inhibits TGF-β1 signaling pathways leading to the inhibition of TGF-β1-induced increase in α-SMA and collagen type I expression levels and ultimately reduces the ECM deposition, which ameliorates hepatic brosis in the hepatic cells 38-40 .
It has been reported that the accumulation of ECM proteins disturbs the liver architecture by forming a brous scar and the subsequent development of cirrhosis with nodules of regenerating hepatocytes, which often leads to progressive loss of liver function 3 . Hence, anti brogenic therapy that suppresses the activation of HSCs has been preferentially considered as an attractive target to prevent the pathological progression to cirrhosis in chronic liver diseases 41 . Our present data show that the number of α-SMA positive cells in the liver increased by DMN treatment were suppressed by GM-CSF administration and also suppressed the increased collagen accumulation. Taken together, these ndings suggest that the anti brotic effect of GM-CSF may be due to suppression of HSC activation. Moreover, GM-CSF inhibited the DMN-induced liver damage and hepatotoxicity, and also increased body weights and survival rate.
The hepatoprotective effects of GM-CSF against DMN-induced hepatotoxicity were further con rmed by the prevention of liver weight loss resulted from DMN administration. Intraperitoneal injection of GM-CSF was able to keep the liver near to normal biomarker levels and reversed histological integrity.
It should, however, be noted that some rats died in the DMN + GM-8w group where DMN was injected alone for four weeks and GM-CSF was administered only from the beginning of the 5th week. But most of the rats that died were at the end of the 4th week following DMN administration which did not receive GM-CSF treatment. In the DMN + GM-4w group, on the other hand, where GM-CSF was administered together with DMN, GM-CSF considerably minimized the toxic effects of DMN.

Conclusion
In summary, our results con rm the inhibitory and therapeutic effects of GM-CSF against DMN-induced liver brosis in rats. The results showed that GM-CSF had some speci c therapeutic effects on pathological changes in the liver. We demonstrated that GM-CSF act as anti brogenic agent which signi cantly reduced the DMN-induced increase in collagen I and α-SMA expression by suppressing TGF-β1 expression and increasing PPAR-γ expression that ultimately suppressed the activation of HSCs. Hence, GM-CSF can be an attractive target to prevent/cure pathological progression to liver brosis and should studied further for its potent clinical applications.  Figure 1 Schematic diagram of the experimental protocol. Rats were divided into six groups. Each group has ten rats (n = 10). Hepatic brosis was induced by administration of DMN (10 mg/kg body weight, IP injection) three times per week for four weeks. GM-CSF was administered in rats (50 µg/kg body weight, IP injection) two every other day for four weeks in DMN+GM-4w group. The DMN+GM-8w group received GM-CSF (50 µg/kg body weight, IP injection) two alternative days per week for four weeks only from the 5th week after DMN administration. Control groups were administered equivalent saline alone (0.9%) three times a week for four weeks via IP injection. Rats of the 4-week groups were euthanized on 29th day and 8-week groups on 55th day after the initial DMN administration.

Figure 2
Effects of GM-CSF on DMN-induced histopathological changes in the 4-week group. Liver tissue was collected at the 29th day after the initial DMN administration and xed in 10% neutral formalin solution.
Thin sections (5 μm) were cut and stained with hematoxylin and eosin (H&E) and Sirius Red. Collagen I and α-SMA proteins were detected immunoshitochemically.

Figure 3
Effects of GM-CSF on DMN-induced histopathological changes in the 8-week group. Liver tissue was collected at the 57th day after the initial DMN administration and xed in 10% neutral formalin solution.
Thin sections (5 μm) were cut and stained with hematoxylin and eosin (H&E) and Sirius Red. Collagen I and α-SMA proteins were detected immunoshitochemically.   Effects of GM-CSF on DMN-induced hepatotoxicity in rats. Blood samples were collected from venous and arterial blood vessels and heart chambers after 4 or 8 weeks of DMN administration. Aspartate aminotransferase (AST), albumin (ALB), and total bilirubin (TBIL) were calculated from blood serum spectrometrically. Changes in serum AST (U/L) (A), TBIL (mg/dL) (B), and ALB (g/dL) (C) of rats treated with or without DMN ± GM-CSF for 4 and 8 week groups are given. Results are expressed as means ± SEM. Statistical signi cance was assigned as *** for p ≤ 0.001, ** for p ≤ 0.01, and * for p ≤ 0.05.

Figure 7
Effects of GM-CSF on DMN-induced changes in survival and body weights. Survival of rats are shown for the sham controls (n=10), DMN-treated (n=10), and DMN+GM-CSF-treated (n = 10) groups for 4-week (A) and 8-week groups (B). The death of rates was recorded every week and used to calculate survival rate.
Body weights of rats are shown for the sham controls (n=10), DMN-treated (n=10), and DMN+GM-CSFtreated (n = 10) groups for 4-week (C) and 8-week group (D) rats. Body weights were measured weekly throughout the study. Results are expressed as means (± SEM). Statistical signi cance for the controls vs. DMN-treated groups with or without GM-CSF are expressed as ## for p ≤ 0.01, # for p ≤ 0.05, and that for the DMN alone vs. DMN+GM-CSF-treated groups is expressed as *** for p ≤ 0.001.