Systemic inhibition of BMP1-3 decreases progression of CCl₄-induced liver fibrosis in rats

Abstract

Liver fibrosis is a progressive pathological process resulting in an accumulation of excess extracellular matrix proteins. We discovered that bone morphogenetic protein 1-3 (BMP1-3), an isoform of the metalloproteinase Bmp1 gene, circulates in the plasma of healthy volunteers and its neutralization decreases the progression of chronic kidney disease in 5/6 nephrectomized rats. Here, we investigated the potential role of BMP1-3 in a chronic liver disease. Rats with carbon tetrachloride (CCl₄)-induced liver fibrosis were treated with monoclonal anti-BMP1-3 antibodies. Treatment with anti-BMP1-3 antibodies dose-dependently lowered the amount of collagen type I, downregulated the expression of Tgfb1, Itgb6, Col1a1, and Acta2 and upregulated the expression of Ctgf, Itgb1, and Dcn. Mehanistically, BMP1-3 inhibition decreased the plasma levels of transforming growth factor beta 1(TGFβ1) by prevention of its activation and lowered the prodecorin production further suppressing the TGFβ1 profibrotic effect. Our results suggest that BMP1-3 inhibitors have significant potential for decreasing the progression of fibrosis in liver cirrhosis.

Keywords:

• Bone morphogenetic protein 1–3
• liver fibrosis
• integrin
• BMP7
• TGFβ1

Introduction

Fibrosis is an excessive deposition of extracellular matrix (ECM) proteins into tissues and constitutes a uniform response of tissue to chronic injury leading to scar formation with a subsequent progressive loss of organ function and disruption of normal tissue architecture (Rockey et al., 2015). Current therapeutic options for tissue fibrosis are limited and often implemented too late in the disease process so the organ transplantation is the only effective treatment for the end-stage disease, making fibrosis one of the major cause of morbidity and mortality worldwide (Wynn, 2008).

Liver fibrosis (LF) is a progressive pathological process that includes multiple cellular and molecular events leading to the hepatocyte apoptosis, hepatic stellate cell (HSC) activation and deposition of excess matrix proteins in the extracellular space, particularly fibrillar collagen type I. Collagen type I is synthesized as a procollagen form, with the mature protein folding after enzymatic cleavage by bone morphogenetic protein 1 (BMP1) at the C-terminal end of the procollagen. Change of organ architecture and functional units by deposition of ECM components leads to liver cirrhosis, portal hypertension, and liver failure (Bataller and Brenner, 2005; Friedman, 2008; Gressner et al., 2007; Liu et al., 2006; Rockey, 2008).

BMP1 and its BMP1-3 isoform (also called mammalian tolloid mTLD) are alternatively spliced products of the Bmp1 gene (Takahara et al., 1994) that are essential for the tissue patterning and ECM assembly by biosynthetic processing of a wide range of ECM precursors. Specifically, enzymatic cleavage of procollagen type I by BMP1 at the C-terminal end of the procollagen molecule finally leads to collagen deposition in liver cirrhosis associated with liver failure (Bataller and Brenner, 2005; Friedman, 2008; Gressner et al., 2007; Liu et al., 2006; Rockey, 2008). We recently isolated the endogenous full-length gene transcript of BMP1-3 isoform from the human and rat plasma as an active enzyme without the prodomain (Grgurevic et al., 2007; Grgurevic et al., 2011; Hulmes et al., 1997). Using a rat model of chronic kidney disease (CKD), pro-fibrotic effects of BMP1-3 were demonstrated via increased ECM expression and deposition, while therapeutic administration of specific polyclonal antibodies (Ab) inhibited the circulating BMP1-3 isoform activity and resulted in the significant reduction of renal fibrosis, improvement of kidney function, and prolonged rat survival. Since the role of BMP1 in chronic liver diseases was not systematically explored, we assessed the role of BMP1-3 in a model of liver fibrosis with the aim to translate the findings obtained on kidney to the liver. Our results clearly show that BMP1-3 inhibition prevents the progression of fibrosis and maintains the liver function.

Methods

Cell culture and treatment

Human HSC line LX-2, established from primary human HSC (Wang et al., 2013; Xu et al., 2005; Mosmann, 1983), was a generous gift from Dr. Scott L. Friedman, Mount Sinai School of Medicine, NewYork, NY. LX-2 cells were cultured in Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F12; Sigma, St. Louis, MO, D8437), supplemented with 10% foetal bovine serum (FBS; Sigma, F7524) at 37 °C in 5% carbon dioxide (CO₂) atmosphere. In experiments, LX-2 cells were seeded at 1 × 10⁵ cells per well in 12-well plates and serum was deprived for the duration of the experiment. Cells were stimulated with transforming growth factor beta 1 (TGFβ1) (10 ng/mL) for 20 h and then additionally treated with BMP7 (100 ng/mL) or BMP1-3 Abs (1 µg/mL), for the following 24 h.

Cell viability

Cell viability was determined using colorimetric MTT assay (Promega, Madison, WI G3580) (Livak & Schmittgen, 2001). The assay was performed according to manufacturer’s instructions and the absorbance at 490 nm was measured on Victor2 microplate reader (Perkin Elmer, Waltham, MA).

Antibodies

Monoclonal antibody against BMP1-3 (BMP1-3 mAb) was generated in mice immunized with a specific synthetic peptide: (aa 972-986; RYTSTKFQDTLHSRK) and were affinity purified (Multiple Peptide Systems, San Diego, CA). The neutralizing effect of the antibody was demonstrated by the inability of BMP1-3 to process dental matrix protein 1 (DMP-1) and procollagen I as previously described (Grgurevic et al., 2011). Antibodies produced were used in vivo and in vitro, and for immunohistochemistry. Liver sections were also stained by a murine monoclonal antibody to BMP7 (12G3, Genera Research Laboratory, Kalinovica, Croatia) and decorin (sc-22613, Santa Cruz Biotechnology, Dallas, TX), respectively. CD31/PECAM-1 antibody (AF3628-SP, R&D Systems, Minneapolis, MN) was used as a marker of liver sinusoidal endothelial cells. Pharmacokinetic experiments revealed a relatively short half-life of the BMP1 Abs in the blood returning to normal values after 72 h after intravenous injection (data not shown). Non-imune IgG of the same isotype and concentration as the primary monoclonal antibody was used as a negative control.

Proteins

Recombinant human BMP7 (rhBMP7) (Genera Research Laboratory, Zagreb, Croatia) was used for the systemic in vivo therapy and for in vitro studies. TGFβ1 (R&D Systems, 240-B) was used in vitro for LX-2 cell line activation experiments.

Animals

Study protocols for liver fibrosis were conducted in 96 male Sprague Dawley rats, with body weight between 300 and 350 g, 12 weeks old, from our own breeding, in a registered animal facility (School of Medicine, Zagreb; Directorate of Veterinary; Reg. No: HR-POK-001). Laboratory rats were acclimated for five days before experiment started and were randomly assigned to their respective treatment group. Laboratory animals were housed in polycarbonate cages in conventional laboratory conditions. Standard diet and fresh water were provided ad libitum according to good laboratory practice.

Ethical principles of the study ensured compliance with European Directive 010/63/E, the Law on Amendments to Animal Protection Act (Official Gazette 37/13, the Animal Protection Act (official Gazette 135/06), Ordinance on the protection of animals used for scientific purposes (Official Gazette 55/13), and FELASA recommendations and recommendations of the Medical School Ethics Committee.

Rat model of liver fibrosis

For the induction of LF a 10% solution of carbon tetrachloride (CCl₄) in olive oil was applied to laboratory rats intraperitoneally (i.p.) at 1 mL/kg twice a week (Constandinou et al., 2005; Isaka et al., 1996; Iredale et al., 2007; Jang et al., 2008). Treatment efficacy was monitored at three and eight weeks, respectively. Animals were randomly assigned into five groups (n = 12) per time point: (1) sham, olive oil i.p. twice-weekly; (2) control, 10% CCl₄ i.p. 1 mL/kg + control antibody; (3) 10% CCl₄ i.p. 1 mL/kg + BMP1-3Abs intravenously (i.v.) 20 μg/kg; (4) 10% CCl₄ i.p. 1 mL/kg + BMP1-3Abs i.v. 50 μg/kg; and (5) 10% CCl₄ ip 1 mL/kg + rhBMP7 i.v. 150 μg/kg. The BMP1-3Abs dose was chosen based on previous studies in a rat model of CKD (Grgurevic et al., 2011). At each time point animals were euthanized and their blood was drawn for serum isolation and the liver tissue was collected for further analyses. All data related to the in vivo model (biochemistry, immunohistochemistry/histology, and gene expression) were acquired throughout the experiment: each group had 12 animals and gene expression was done for each animal in the group once (in duplicates).

Liver preparation

After opening the abdominal cavity three parts of the liver were excised. One portion was fixed in 10% formalin for histological analyses, second part was conserved in Tri Reagent (Ambion, Austin, TX) for RNA isolation, and the third was weighted for hydroxyproline (HP) assay. The remaining tissue along with the spleen was placed in 50 mL plastic containers containing degassed 0.9% saline solution at room temperature (20–22 °C) for elastography analysis.

Liver and spleen stiffness measurement

Prepared liver samples were transferred to the shear wave elastography (SWE) examination within 1 h after liver removal from abdominal cavity. SWE is a two-dimensional real time ultrasound based elastographic method used for assessment of an organ’s stiffness (Bamber et al., 2013), accepted in clinical hepatology as a noninvasive indicator of fibrosis for the liver and spleen (Cosgrove et al., 2013). Stiffness was calculated using integrated software based on the Young’s modulus of elasticity equation and expressed in kPa (Bamber et al., 2013). Five elastographic measurements of the liver or spleen were performed on Supersonic Imagine Aixplorer® Ultrasound system, by linear transducer SuperLinear™ SL15-4 (4–15 MHz).

Biochemical analyses

The level of alanine aminotransferase (ALT) and albumin was measured from animal serum. One milliliter of blood was drawn into empty vacutainer tubes (BD Biosciences, Franklin Lakes, NJ) and the serum was collected by centrifugation at 1000 × g for 20 min. Analyses were performed using the clinical chemical analyzer Roche Cobas 6000 (Roche Diagnostics, Basel, Switzerland). All the original reagents, standards, and controls were purchased from Roche. Plasma levels of TGFβ1 were determined in sham, CCl₄-injected rats and in animals treated with a high dose (50 µg/kg) of anti-BMP1-3 monoclonal antibody by Quantikine TGFβ1 ELISA (R&D Systems).

Hydroxyproline assay

Hydroxyproline (HP) liver content was measured by a commercial assay (Sigma, MAK008) according to manufacturer’s instructions. All samples were run in duplicates and the absorbance was measured at 562 nm using EL808 microtiter plate reader (BioTek Instruments, Winooski, VT). HP concentration was calculated according to the standard curve using a linear regression expressed in µg/µL.

Gene expression analyses

Total RNA was isolated from LX-2 cells or liver tissue using Tri Reagent. 1 µg of total RNA was transcribed into cDNA using high-capacity cDNA reverse transcription kit (Applied Biosystems, Thermo Fisher Scientific, Waltham, MA) and oligo-dT primers for mRNA transcription. For gene expression analyses quantitative real-time polymerase chain reaction (PCR) was performed by Light Cycler (Roche) using SYBR Premix Ex Taq II (Takara, Japan) (Pauk et al., 2015). Sequences of primers shown are presented in Table 1. The comparative CT method (ΔΔCT) was used for relative quantification of gene expression (Livak and Schmittgen, 2001). Results were presented as relative changes in gene expression compared to the control sample with glyceraldehyde 3-phosphate dehydrogenase (Gapdh) used as a normalizer gene (Smits et al., 2009).

Table 1. Sequences of primers used in gene expression analyses.

Gene	Short name	Forward primer (5')	Reverse primer (3')
Human
Collagen type 1	COL1A1	GTCGAGGGCCAAGACGAAG	CAGATCACGTCATCGCACAAC
Glyceraldehyd 3-phosphate dehydrogenase	GAPDH	CATGAGAAGTATGACAACAGCCT	AGTCCTTCCACGATACCAAAGT
Rat
Bone morphogenetic protein 1	Bmp1	CAACGCCCTGTCCCAATGAC	GACACGACCATACTGATACACC
Transforming growth factor β1	Tgfb1	TCAAAGATTTAACATCTCCAACC	GGTGATCCTTTTTTTTGTCACCCT
Collagen type 1	Col1a1	TGTGTGCGATGACGTGCAAT	ACATCCTCAGCTCCCTGGG
α smooth muscle actin	Acta2	CGATAGAACACGGCATCATC	CATCAGGCAGTTCGTAGCTC
Connective tissue growth factor	Ctgf	TAGCAAGAGCTGGGTGTGTG	TTCACTTGCCACAAGCTGTC
Decorin	Dcn	GAGGCCTCTGGCATAATCCCTTACG	TGCCTCTGGGCTGATCTTGCTGAT
Integrin β1	Itgb1	GTTGCCCCAGCCTGTGCAGG	ACAATTCCGGCAACCACGCCT
Integrin β6	Itgb6	ACCCACCTGTCCGGAGCTGG	CGTCTGGCGGACCTGCACTT
Glyceraldehyd 3-phosphate dehydrogenase	Gapdh	ATGATTCTACCCACGGCAAG	CTGGAAGATGGTGATGGGTT

Histology and morphometric analyses

Liver samples were paraffin embedded, sliced to 5 μm thick sections, and stained with Sirius red (Chen et al., 2013; Ge & Greenspan, 2006a). To assess the extension of liver fibrosis histological scoring system and quantification of collagen amount were used (Nanji et al., 1989). According to Nanji et al.( 1989) liver fibrosis is classified into four histological stages ranging from 1 (initial pericentral fibrosis and several thin fibrous septa) to 4 (cirrhosis) (Calvaruso et al., 2009).

The amount of collagen in histological sections stained by Sirius red was quantitatively assessed by morphometric analysis. Within each histological stage the connective tissue amount can vary substantially, which has been proven to significantly influence the course of the disease (Smits et al., 2009). Each slide was analyzed by an optical microscope (Olympus Provis microscope, Campbell, CA) and photographed (magnification 4 × lens 4×/0.13) at four different locations, covering the entire slide surface with exclusion of larger blood vessel walls since their collagen content might influence the result. The computed morphometric image analysis was performed using a Sform software (Vams, Croatia) with fibrosis amount expressed as a ‘collagen proportionate area %’ (CPA%) per slide (Calvaruso et al., 2009; Ge & Greenspan, 2006a).

Immunohistochemistry

5 μm-thick sections were deparaffinized and rehydrated in the descending series of ethanol with final 1 × PBS incubation. To eliminate endogenous peroxidase activity, the sections were pretreated at room temperature with 3% hydrogen peroxide in methanol for 10 min. Sections were then incubated with the primary antibody against BMP1-3, BMP7, CD31, or decorin diluted 1:150 in 1 × PBS for 120 min at 37 °C in a moist chamber. The reaction was detected using Histostain SP kit (Invitrogen) while staining was visualized using DAB chromogen for BMP1-3/BMP7/CD31 and AEC chromogen for decorin detection. Slides were counterstained in hematoxylin and mounted using ClearMount (Invitrogen, Carlsbad, CA). Each slice was analyzed in its entire area using Olympus Provis microscope with 10× and 40× magnification (lens 10x/0.30, UPIanFI Ph1).

Data analysis

Data are summarized by treatment and time-point. In order to follow the dynamics of liver fibrosis, values determined at different times after CCl₄ treatment were considered to represent changes over time. For statistical analysis, liver HP concentrations, relative fibrosis area (by morphometry), and liver stiffness (by ultrasound elastography) were ln-transformed. General linear (mixed in the case of elastography to account for repeated measurements on the same organ) models were fit to data with treatment, time and treatment × time interaction effects. Treatment differences are expressed as geometric means ratios (GMR):GMR = exp[0Ln(A)-0Ln(B)], or as percent (%) differences derived from GMRs. Gene expression data were analyzed by the comparative CT method (ΔΔCT) by fitting general linear mixed models (GLMMs) to account for potential correlations between simultaneously analyzed genes. Adjustment for multiple comparisons was done by the simulation method. Analyses were performed using SAS 9.3 for Windows software (SAS Inc., Cary, NJ).

Results

Anti-BMP1-3 antibody attenuates TGFβ1-induced HSC activation

LX-2 is a human stellate cell line with features comparable to primary HSCs (Wang et al., 2013), so it was used here as an in vitro model for testing the efficacy of the BMP7 and BMP1-3 antibody treatment. TGFβ1 is a potent profibrogenic cytokine that stimulates the expression of collagen type I in primary HSCs. TGFβ1 treatment of LX-2 cells significantly increased the expression of Col1a1 mRNA, while anti-BMP1-3 (mAb reduced the upregulation of Col1a1 mRNA induced by TGFβ1.Addition of BMP7 also attenuated the TGFβ1 effect (Figure 1). LX-2 cell viability was not affected by any of the treatments (Supplementary Figure S1).

Figure 1. The expression of Col1a1 mRNA in LX-2 cells. Cells were stimulated with TGFβ1 (10 ng/mL) for 20 h and treated with BMP7 (100 ng/mL) and BMP1-3 Ab (1 µg/mL) in parallel to TGFβ1 treatment, for an additional 24 h. Col1a1 mRNA expression was measured by RT-PCR and expressed here as a fold change relative to the control (without any treatment). The experiment was performed three times in triplicates (n = 3). Analysis of Variance was used to assess differences between samples and the data are presented as mean value ± SEM.

CCl₄-induced liver fibrosis and the effect of therapy with BMP1-3 antibodies

To further explore the effect of BMP1-3 in liver fibrosis, rats were injected with CCl₄. Hepatotoxic effect of CCl₄ in rats was confirmed by a significant increase in the ALT expression, while no change in the albumin level was detected between the groups (Figure 2). Both ALT and albumin are increased in cirrhosis as a consequence of liver injury. The treatment with both, a higher dose of the anti-BMP1-3 mAb and BMP7 protein (150 μg/kg/day), retained the serum ALT level lower than in control animals.

Figure 2. Serum ALT (A) and albumin (B) values during eight weeks in all experimental groups. *p < .05 Control vs. BMP1-3Ab high; **p < .05 Control vs. BMP7.

HP concentration in extracellular matrix produced by activated HSCs is also present in serum and urine comprising correctly the rates and progression of liver fibrogenesis (Gabr et al., 2016; Sristastava et al., 2016).

Elevated concentrations of the liver HP were detected at three weeks and more considerably at eight weeks following injection of CCl₄ (Figure 3(A)). A similar increasing pattern was observed regarding the morphometrically assessed extent of fibrosis (Figure 3(B)), as well as liver stiffness determined by the ultrasound elastography at week 3 and 8 (Figure 3(C)). In CCl₄-injected rats treated with anti-BMP1-3 mAb (50 μg/kg/day), the level of HP was almost similar to control rats, while the fibrosis was reduced by 56%. BMP7 also reduced the liver HP content by 53% (Figure 4). At three weeks, the treatment with the anti-BMP1-3 mAb at a lower dose of 20 μg/kg/day reduced the CCl₄-induced increase in liver HP, while ALT and albumin level remained increased (Figure 3(A–C)). Regarding the extent of fibrosis at week 3, CCl₄ caused a progressive fibrosis over time, while anti-BMP1-3 therapy consistently reversed the fibrosis by 36–38% (Figure 3(D–F)). BMP7 apparently reversed the CCl₄ effect by around 34% at three weeks, but not at eight weeks. Similarly, liver stiffness and fibrosis were increased in rats injected with CCl₄ and antagonized by BMP1-3 inhibition (Figure 3(E,F)). Importantly, the higher anti-BMP1-3 mAb dose (50 μg/kg/day) consistently inhibited the CCl₄ effect on all outcomes at three and eight weeks (Figures 2 and 3). Inhibition of the CCl₄ effect revealed a decrease of HP at three weeks (41 ± 24), liver fibrosis (24 ± 15), and liver stiffness (7 ± 4) and at eight weeks HP (15 ± 9), liver fibrosis (12 ± 7), and liver stiffness (8 ± 5), respectively with the low anti-BMP1-3 mAb dose (20 ?g/kg/day). Aforementioned fibrosis extent was measured morphometrically on histological slides in a similar surface area of the liver middle lobe. Figure 4 summarizes relative differences of the BMP1-3 inhibition and BMP7 effect on the liver cirrhosis depicted in Figure 3. Both BMP1-3 inhibition and BMP7 had, at three weeks, similar effects on the fibrosis area and stiffness, but at eight weeks, the effect of BMP1-3 mAb on the liver fibrosis and stiffness were surprisingly increased by 58 and 20%, respectively, as compared to the BMP7 administration (Figure 3(E,F)).

Figure 3. (A–C) Indicators of CCl₄-induced LF and the effect of BMP1-3 Abs. (A) Liver HP concentrations. (B) Extent of fibrosis (‘relative fibrosis area’) as assessed morphometrically. (C) Liver stiffness assessed by ultrasound (US elastography). Data are summarized as mean ± SD (n = 6–8 per group). To assess between-group differences, for each outcome a general linear model (mixed model in the case of stiffness to account for repeated measures on the same organ) with treatment, week, treatment x week interaction effects was fit to ln-transformed data, and contrasts of interest (each treatment vs. Sham and each active treatment vs. control) were estimated. (D-F) Size of the effect of anti-BMP1-3 antibody high dose (50 μg/kg/day) and of BMP7 (150 μg/kg/day) on CCl₄-induced indicators of liver fibrosis. (D) Hydroxyproline concentration. (E) Extent of fibrosis (‘relative fibrosis area’). (F) Liver stiffness. Effect of CCl₄ is illustrated by the difference between control (CCl₄ + vehicle) and sham animals (no CCl₄ treatment – represented as a dashed line at geometric means ratio of 1, i.e. the ‘reference value’). Effect of CCl₄ in the presence of anti-BMP1-3 Ab high or BMP7 is illustrated by the differences between the respective treatment groups vs. sham animals. The effects of anti-BMP1-3 Ab high and of BMP7 are expressed numerically as percent (%) inhibition of the CCl₄ effect. All contrasts are derived from the models depicted in Figure 2 legend. Data (symbols, numbers) are point estimates with 95% confidence intervals. All intervals and p values are adjusted for multiple comparisons by the simulation method. ^ap < .001 vs. sham (no CCl₄ treatment); ^bp < .05 vs. control (CCl₄ + vehicle); ^cp < .001 vs. control.

Figure 4. Relative difference (Δ %) in the inhibitory effect on the CCl₄-induced indicators of LF between the BMP1-3 Abs high dose (50 μg/kg/day) and BMP7 (150 μg/kg/day). Differences are derived from the models depicted in Figure 2 legend and put into relationship the effects of two therapeutic treatments here, which were calculated as follows: (a) the effect of CCl₄ = (b) the inhibitory effect of BMP1-3 Abs = ; (c) the inhibitory effect of BMP7 = ; (d) relative difference between two inhibitory effects (BMP1-3Abs – BMP7) = . If the effect was greater with BMP1-3Abs it was given a positive sign, if it was greater with BMP7, it was given a negative sign. Data (symbols, numbers) are point estimates with 95% confidence intervals. All intervals and p values are adjusted for multiple comparisons by the simulation method. BMP1-3 Ab: polyclonal antibodies against BMP1-3; BMP: bone morphogenic protein. *LF i.e. liver fibrosis refers to morphometrically assessed relative fibrosis area and liver stiffness was assessed by ultrasound elastography.

Localization of BMP1-3 in the healthy and cirrhotic liver

BMP1-3 in the healthy liver was mostly found in sinusoidal endothelial cells (SEC) and to a lesser extent in hepatocytes, while in the fibrotic liver BMP1-3 was stained with a similar intensity in both hepatocytes and SEC (Figure 5(A,B)). BMP7 in the normal liver was not detected while a strong staining intensity was found in the fibrotic rat liver (Figure 5(C,D)).

Figure 5. Immunohistochemistry of CD31, BMP1-3, and BMP7 in sections of healthy and fibrotic livers. (A) CD31 was expressed in sinusoidal endothelial cells (SEC), (B) BMP1-3 was expressed in healthy liver mostly in SEC, (C) Expression of BMP1-3 was localized in fibrotic liver hepatocytes and SEC, (D) BMP7 expression was undetectable in healthy rat liver, (E) BMP7 was expressed in fibrotic liver mostly in hepatocytes, and (F) non-immune IgG of the same isotype (IC) in concentration as the primary monoclonal antibody was used as a negative control. HL:healthy liver; FL: fibrotic liver. Arrows indicate SEC and arrowheads indicate hepatocytes (40×). Scale bar length is 50 µm.

Effects of anti-BMP1-3 therapy on extracellular matrix deposition and plasma levels of TGFβ1

Histomorphometric analyses of the collagen content on liver sections stained with Sirius red revealed significant differences between treated animals. Compared to the CCl₄ control rats, the amount of collagen found in anti-BMP1-3 mAb treated rats was significantly lower (p < .05) and a similar trend was observed in rats treated with BMP7 (Figure 6(A,B)). A marked increase in the collagen deposition in control rats was observed around the portal triad and nodules of hepatocytes, while in anti-BMP1-3 treated rats a thinner bands of collagen accumulation without complete nodal formation were observed at three and eight weeks after CCl₄ administration (Figure 6). These findings were biochemically confirmed by a similar trend of HP content (Figure 4). Rats treated with anti-BMP1-3 mAbs had lowered plasma levels of TGFβ1 when compared to CCl₄-control rats (Figure 6(C)).

Figure 6. (A) Treatment of rats with induced liver fibrosis with a low and high dose of a BMP1-3 antibodies decreases accumulation of ECM and preserves liver structure. Histomorphometric analyses of rat liver sections stained with Sirius red three and eight weeks after treatment with 10% CCl₄ i.p. 1 mL/kg + i.v. control antibody; 10% CCl₄ i.p. 1 mL/kg + BMP1-3Abs i.v 20 μg/kg; 10% CCl₄ i.p. 1 mL/kg + BMP1-3Abs i.v 50 μg/kg, and 10% CCl₄ i.p. 1 mL/kg + rhBMP7 i.v 150 μg/kg. Arrows indicate ECM localization. (B) Morphometric analysis was performed using a Sform software (Vams, Croatia) with fibrosis amount expressed as a proportion of positive colored area (Collagen proportionate area (CPA%)) per slide. Results are presented as mean ± SEM. *p < 0.05 versus control rats. (C) TGFβ1 plasma values in rats with liver fibrosis are decreased after therapy with a high dose (50 µg/kg) of BMP1-3 monoclonal antibody. Data is presented as mean ± SD (n = 12 per group). *p < .05 vs. sham, #p < .05 vs. control. Scale bar length is 500 µm.

Effects of BMP1-3 inhibition on the liver gene expression and immunolocalization of decorin

Liver gene expression data are summarized in Figure 7. CCl₄ injection resulted in increased Tgfb1 expression of six- and eight-fold at three and eight weeks, respectively. At eight weeks, the higher dose of anti-BMP1-3 mAb (50 µg/kg) reduced this effect by 48%. Further, CCl₄ stimulated a marked increase in Col1a1 expression, which was reduced at three weeks by 52 and 71% by both anti-BMP1-3 mAb doses (20 and 50 µg/kg), respectively. BMP7 therapy also reduced the Col1a1 expression at both three and eight weeks (Figure 7). Increased Acta2 expression by CCl₄ at three (20-fold) and eight weeks (100-fold), was reduced by BMP1-3 inhibition for 60% at eight weeks. CCl₄ injection reduced the Bmp1 expression by 3.5-fold at three weeks and 4-fold at eight weeks, while anti-BMP1-3 antibody (50 µg/kg) attenuated this effect towards sham values. CCl₄ increased the expression of Ctgf by 15-fold at three weeks and by 70-fold at eight weeks. The high anti-BMP1-3 mAb dose (50 µg/kg) further increased the Ctgf expression at eight weeks. CCl₄ increased the Itgb6 expression by 180-fold at three weeks and by 40,000-fold at eight weeks. BMP1-3 inhibition further increased the Itgb6 expression by 557 to 793% at three weeks, while at eight weeks it was reduced by 81 to 84%. Itgb1 expression was increased in CCl₄-injected rats by two-fold at three weeks and by 3.8-fold at eight weeks, while at three weeks BMP1-3 inhibition increased the Itgb1 expression (Figure 7).

Figure 7. The effects of CCl₄ treatment and of antibodies against BMP1-3 (BMP1-3 Abs) at a low (20 μg/kg/day) or at a high dose (50 μg/kg/day) and of BMP7 (150 μg/kg/day) on CCl₄-induced liver gene expression. A general linear mixed model was fitted to normalized Ct (ΔCt) data to obtain adjusted treatment effect estimates accounting for correlations between simultaneously assayed genes, time and treatment*time interaction. Adjustment for multiple comparisons was by the simulation method. Data are shown as ‘fold expression over sham’ (animals not treated with CCl₄ – indicated as ‘fold expression =1’ with a dashed line) for each treatment group and were obtained as antilog (base 2) of adjusted ΔΔCt values: symbols identifying treatments (as per legend) are point-estimates and vertical bars are 95% confidence intervals (symmetrical on the log-scale). p values for each treatment group (identified by its respective symbol) vs. sham (dashed line) are given below the dashed line either as ‘vs. Sham all p < .001’, or as individual p values (below the lower limit of the confidence intervals) were this was not the case (Bmp1, Itgb6). Horizontal arrows depict comparisons of each active treatment vs. control (CCl₄ + vehicle-treated animals) with differences given as ‘+’ or ‘−’ percentage differences (derived from ratio of ratios) and corresponding p values.

CCl₄ administration also increased the Dcn expression, four-fold at three weeks and 50-fold at eight weeks. BMP1-3 Ab inhibition further increased the Dcn expression at both three and eight weeks. After three weeks, in anti-BMP1-3 mAb treated rats the intensity of decorin staining increased in subendothelial and stromal compartment as well as in the cytoplasm of hepatocytes (Figure 8(A–J)). The prodecorin is primarily cleaved by BMP1 gene enzymatic isoforms, therefore, inhibition of BMP1-3 followed by increased decorin staining suggested indeed an accumulation of the prodecorin form in the liver.

Figure 8. Immunolocalization of decorin in the liver after three and eight weeks of therapy with low (20 µg/kg/day) and high dose (50 µg/kg/day) of BMP1-3 antibody and BMP7 in rats with induced LF. After three weeks of therapy in all treatment groups, the intensity of cytoplasmic decorin staining increased in hepatocytes. (A) Liver three weeks after induction of fibrosis; (B) BMP1-3Ab low dose after three weeks; (C) BMP1-3Ab high dose after three weeks; (D) BMP 7 after three weeks; (E) Healthy rat liver; (F) Liver eight weeks after induction of fibrosis; (G) BMP1-3Ab low dose after eight weeks; (H) BMP1-3Ab high dose after eight weeks; (I) BMP7 after eight weeks; and (J) non-immune IgG of the same isotype (IC) in concentration as the primary monoclonal antibody was used as a negative control. Dark arrows indicate decorin cytoplasmic staining in hepatocytes and arrowheads indicate decorin in SEC (magnification ×40). Scale bar length is 50 µm.

Discussion

In a model of CCl₄-induced rat liver disease, we evaluated the effect of BMP1-3 on the liver fibrosis in comparison to already proven antifibrotic properties of BMP7 (Sugimoto et al., 2007; Bi et al., 2012; Tacke et al., 2007). Liver fibrosis is driven primarily by the inflammation and activation of HSCs to a myofibroblast-like phenotype through multiple cell mediators (Constandinou et al., 2005; Grgurevic et al., 2016; Vukicevic and Grgurevic, 2016) which are reduced by administration of BMP1-3 mAbs, targeting thus specific pathways in the liver disease.

To test the potential antifibrotic effect of BMP1-3 mAbs, we first established the HSC culture system and measured the Col1a1 expression in LX2 cells. Treatment with TGFβ1 activated the LX2 cells and induced an overexpression of Col1a1 gene, while the treatment of activated cells with both, BMP1-3 mAbs and BMP7, reduced the Col1a1 expression. Different conditions of the in vitro and in vivo systems used, although not properly reflecting genetic reprograming of tissue HSC activation (De Minicis et al., 2007; Gressner and Gao, 2014), mirrored the effect from the in vivo results in support of an antifibrotic role of BMP1-3 mAbs.

The presence of BMP1-3 was immunohistochemically demonstrated in both the healthy and cirrhotic liver suggesting that at least a part of circulating BMP1-3 is produced in the liver. Administration of BMP1-3 mAbs resulted in an effective inhibition of the fibrosis progression comparable to that of BMP7. However, BMP1-3 inhibition is similar in some aspects to the BMP7 therapy which is significant finding, since administration of a growth factor from the TGFβ superfamily is not a therapeutic option, while administration of an anti-BMP1-3 mAb is potentially safe and requires significantly less frequent injections. Apart from blocking the BMP1-3 activity, which is important for the collagen maturation, administration of BMP1-3 mAbs was accompanied by a reduced expression of Tgfb, Itgb6, and Acta2 gene expression.

In the CCl₄ model of LF, HSC as liver pericytes are the major source of myofibroblasts which are Col1 + α-smooth muscle (αSMA) positive cells that produce the ECM scar in the liver (Kisseleva et al., 2012). In our study, the expression of Acta2 was inhibited with a BMP1-3 mAb at week 8, via inhibition of the activation of quiescent HSC and fibrosis progression, as confirmed by histomorphometry. Decreased accumulation of collagen without a complete nodal formation was observed in the same time interval in anti-BMP1-3 treated groups, as compared to control and sham rats. Due to liver necrosis and subsequent fibrosis after CCl₄ injection, the liver enzyme ALT in serum and HP tissue content were elevated and then reduced by the BMP1-3 inhibition. A direct proportionality was found between the degree of fibrosis and the liver stiffness (Bamber et al., 2013), using the shear-wave elastography as a quick, noninvasive and reliable method for quantification of fibrosis. Additionally, we showed a decreased expression of Tgfb1 and Itgb6 after inhibition of BMP1-3 that is essential for TGFβ1 signaling (Dong et al., 2017; Grgurevic et al., 2011; Ha, 2017). Interaction of BMP1-3 with αvβ6 integrin is potentially a major mechanism of decreased TGFβ activation during LF prevention. BMP1/TLD proteinases activate TGFβ1 as a fibrogenic master cytokine (Gressner & Gao, 2014; Grgurevic et al., 2011; Liu & Yang, 2006; Rockey et al., 2015) by a direct cleavage of the latent TGFβ-binding protein (LTBP), resulting in the release of the large latent complex (LLC) from the ECM, and consequent matrix metalloproteinase (MMP)-dependent latency-associated protein (LAP) cleavage (Ge & Greenspan, 2006a, 2006b). We suggest that BMP1-3 inhibition reduces the processing of LTBP1 and consequently TGFβ1 release from its latent form. The cleaved TGFβ1 prodomain acts as a cage (Ha, 2017) which, following BMP1-3 mAb therapy, further stabilized via an increased expression of integrin β6 subunit containing the active TGFβ1. Cells might have recognized this effect and subsequently decreased the expression of β6 integrin and TGFβ1-encoding genes. Mechanistically, plasma levels of TGFβ1 were decreased in rats treated by a high BMP1-3 antibody dose (50 µg/kg) by almost 60% as compared to CCl₄-control rats, suggesting a BMP1-3 mAb-mediated suppression of TGFβ1 activation.

BMP7 was used as a positive control in reduction of LF development because previous studies demonstrated that BMP7 reduced the LF in mice inhibiting epithelial to mesenchymal trans differentiation of hepatocytes (Zeisberg et al., 2007). BMP7 has anti-apoptotic and anti-inflammatory properties, stimulates proliferation (Grgurevic et al., 2016; Sugimoto et al., 2007; Vukicevic and Grgurevic, 2016), and accelerates liver regeneration after partial hepatectomy in mice (Bataller and Brenner, 2005; Bi et al., 2014; Chen et al., 2013; Wang et al., 2013; Zhong et al., 2013). As a previously proven molecule with antifibrotic activity, BMP7 also reduced Tgfb1, Itgb1, Itgb6, and Acta2 gene expression,

Following BMP1-3 inhibition, endogenous expression of Dcn mRNA and protein was increased. Decorin is a small leucine-rich proteoglycan and downregulates the expression and the biological activity of TGFβ1 (Fullár et al., 2016; Yamaguchi et al., 1990). It has been demonstrated that three BMP1 isoforms (BMP1-1, BMP1-3, and BMP1-5) can effectively remove the pro-peptide from the human prodecorin resulting in a well-established mature proteoglycan (von Marschall and Fisher, 2010). Previous studies indicate the efficacy of decorin in antagonizing the TGFβ1 effect on HSCs by suppression of ECM production (Fullár et al., 2016), migration of fibroblasts, trophoblasts, and differentiation of myocytes (Baghy et al., 2012; Droguett et al., 2006; Fischer et al., 2001). Decorin has an evident anti-fibrotic activity in the liver after injury with CCl₄ (Cai et al., 2014) and Dcn silencing caused an increased activation of HSCs both in vivo and in vitro (Baghy et al., 2011). A possible role of prodecorin in inhibition of fibrosis in kidney and heart injury models could be a consequence of unprocessed prodecorin tissue accumulation (Isaka et al., 1996; Li et al., 2009; Yan et al., 2009). Similarly in these studies BMP1-3 inhibition supported the increased prodecorin production and prevention of LF progression via further inhibition of TGFβ1 profibrotic activity in the liver cirrhosis.

Conclusion

These results suggest that the development of BMP1-3 inhibitors possesses a great potential for expanding attractive therapeutic/preventive targets for antifibrotic intervention(s) in preventing the progression of the fibrotic process in liver cirrhosis.

Notes on contibutors

Lovorka Grgurevic conducts research in the Center for Translational and Clinical Research (CETRI), University of Zagreb School of Medicine, on formulation development and toxicology testing. She explores the structure and function of bone morphogenetic proteins (BMPs) in biological fluids, has discovered circulating osteogenic proteins and associated molecules in the plasma, and then investigated their effectiveness in bone healing and in models of acute and chronic renal failure. She contributed significantly to the discovery of a new carrier for BMPs and tested its efficacy in animal models of bone defects. Dr Grgurevic discovered novel biomarkers for bone repair, breast, and prostate cancer prognosis. She received a new laboratory installation grant from the Croatian Science Foundation and was awarded by the Croatian Academy of Sciences and Arts for scientific achievements.

Igor Erjavec is a postdoctoral fellow at the Laboratory for mineralized tissues and Proteomic center at the Center for Translational and Clinical Research (CETRI), University of Zagreb School of Medicine. Scientific field of his research are animal models, bone remodeling, and regeneration with emphasis on bone morphogenetic proteins and the study of kidney and liver regeneration. Member of Croatian and European Calcified Tissue Society. Co-authored 15 publications in Medline database.

Ivica Grgurevic is an Assistant Professor of Medicine and Board certified Consultant in Internal Medicine, Gastroenterology and Hepatology. He graduated at University of Zagreb School of Medicine and attended postgraduate clinical fellowship in Hepatology at Institute for Liver and Digestive Health, University College London, Royal Free Hospital, UK. Currently works at Department of Gastroenterology, Hepatology and Clinical Nutrition, University hospital Dubrava Zagreb. He was the first person in Croatia to introduce new methods in Hepatology: Quantitative ultrasound elastography (in 2009); Contrast-enchanced ultrasound (2014); and invasive measurement of portal pressure (HVPG) (2015). His principal scientific interests are in the field of Hepatology and Ultrasound. He was awarded the best poster in the category of Liver imaging at United European Gastroenterology Week in Vienna 2016. Member of EASL, EFSUMB, IASGO, and former Secretary general of the Croatian Society of Gastroenterology 2013–2016, currently serves as the Vice-president of the Croatian Society for the Ultrasound in Medicine and Biology.

Ivo Dumic-Cule is now a resident in Radiology at Clinical Hospital Dubrava, Zagreb. For five years he has been working as assistant in Laboratory for mineralized tissues at University of Zagreb School of Medicine, investigating the role of bone morphogenetic proteins in various biological systems.

Jelena Brkljacic is a cell biologist, now working at Xellia R&D. For eight years she has been working in Laboratory for mineralized tissues at University of Zagreb School of Medicine, investigating the activity of bone morphogenetic activity in vitro, focusing on the effect of bone morphogenetic protein 6 (BMP6) on the prevention of heparin induced secondary osteoporosis. She contributed to the production of rhBMP by developing the in vitro assay for BMP activity testing.

Donatella Verbanac works as an Assistant Professor at the School of Medicine University of Zagreb since 2009, leading Department for Intracellular Communication within the Centre for Translational and Clinical Research. In parallel, she participates in several scientific projects, as well as in education activities for the graduate and post-graduate students. She is the Coordinator of translational project ‘Assessment of Microbiota, Inflammatory Markers, Nutritional, and Endocrinological Status in IBD Patients’ funded by Croatian Science Foundation (2014–2018). Her previous experience was within pharmaceutical industies where she worked as Program Leader and Project Manager. For the achievements in the drug development and research projects she has been awarded with PLIVA’s Pharmaceutical Industry and GlaxoSmithKline awards. She has published more than 40 scientific and professional papers. Co-inventor of one patent and coauthor of several chapters in the books, author of few professional books on nutrition and one text-book. She acts also as a reviewer, monitor, and vice-chair for the projects funded by EC, as well as for many distinguished journals in the field of medicinal chemistry and life sciences. She is member of Croatian Biochemical and Molecular Biology Society, Croatian Society of Enteral, and Parenteral Nutrition and British Biochemical Society.

Mario Matijasic is a researcher in the Center for Translational and Clinical Research (CETRI), University of Zagreb School of Medicine, with a decade of experience in anti-inflammatory drug discovery in pharmaceutical industry. Current interests include bone morphogenetic proteins (BMPs), biomarkers, and microbiota research.

Hana Cipcic Paljetak is a researcher in the Center for Translational and Clinical Research (CETRI), University of Zagreb School of Medicine, with almost a decade of experience in drug discovery in pharmaceutical industry in the anti-infective and anti-inflammatory therapeutic areas. Current interests include bone morphogenetic proteins (BMPs), biomarkers, and microbiota research.

Rudjer Novak is a research associate at the Department for Proteomics in the Center for Translational and Clinical Research (CETRI), University of Zagreb School of Medicine. He specializes in the area of mass spectrometry and participates in scientific projects in the area of regenerative medicine and cancer proteomics. His main scientific interest is exploring the function of bone morphogenetic proteins.

Mihovil Plecko is a student at University of Zagreb School of Medicine. He is on his last year of study. He is a volunteer at Laboratory for mineralized tissues since 2012, and his topics of interest are bone morphogenetic proteins and bone metabolism.

Jadranka Bubic Spoljar, PhD, Senior advisor for laboratory animals, Head of animal and breeding facility, designated veterinarian of Animal facilities University of Zagreb, School of Medicine, manage person responsible for the monitoring of preclinical studies, coordination with the Croatian Ministry, and responsible person for animal welfare, quarantine, and health status of animals. Member of the Ethics Committee at the School of Medicine.

Dunja Rogic is a medical biochemistry specialist, professor at Zagreb University School of Pharmacy and Biochemistry, and head of Department of Laboratory Diagnostics, University Hospital Center Zagreb. She is a member of the EFLM Working Group for Postanalytical Phase, of the National Health Council, Croatian Ministry of Health, and of the Ethics Committee, University Hospital Center Zagreb. Her research interests encompass evidence-based medicine, point-of-care testing, critical care, laboratory parameters of renal diseases, and organization and management of a clinical laboratory. She is the author/coauthor of about 50 manuscripts, six book chapters, an editor of several textbooks, a lecturer and organizer of continuous education courses for physicians and medical biochemists, and a member of scientific and organizing committees and an invited speaker at national and international conferences.

Vera Kufner has seventeen years of working experience in biomedical research. She is actively involved in number of scientific projects developing recombinant proteins and anti-osteoporotic therapies. She has extensive experience and expertise in biochemistry, cellular biology, and chemistry, with broad know-how in affinity and ion-exchange chromatography techniques and in training students and young researchers in laboratory practice. Besides scientific work, she is involved in teaching one course of postgraduate program within School of Medicine. She is a team member of on-going HRZZ Research project Assessment of Microbiota, Inflammatory Markers, Nutritional, and Endocrinological Status in IBD Patients and of FP7 project OSTEOGROW.

Martina Pauk works as a senior research assistant in the Laboratory for mineralized tissues, Center for Translational and Clinical Research (CETRI), University of Zagreb School of Medicine. Her research is focused on mechanisms of hepcidin regulation and involvement of other BMPs in the prevention of hemochromatosis and subsequent osteopenia in Bmp6−/− mice. She participated in research projects funded by Croatian Science Foundation named ‘Novel anabolic targeted therapy for osteoporosis: BONE6-BIS Consortium’ (2012-2015) and ‘Newly Discovered Circulating Biomarkers and Therapeutic Goals for Human Diseases’ (2014–2017). Also, she is actively involved in the Framework Program-7 (FP-7) HEALTH program entitled ‘OSTEOGROW‘– Novel Bone Morphogenetic Protein 6 Biocompatible Carrier Device for Bone Regeneration in bioanalytical method validations, writing Standard Operating Procedures (SOP) for work in conditions of GLP (Good laboratory practice), pharmacokinetics of therapeutic proteins, toxicology studies, analysis of clinical trial samples, and stability studies by optimized C2C12-BRE-Luc cell bioassay. She is also involved in teaching activities for students at postgraduate studies in the field of Bone morphogenetic proteins in bone and cartilage regeneration at University of Zagreb School of Medicine.

Tatjana Bordukalo Niksic works as a research associate on several projects in the Laboratory for mineralized tissues in the Center for Translational and Clinical Research, University of Zagreb School of Medicine. She performs molecular biology and biochemical analyses related to the role of bone morphogenetic proteins (BMPs) in various biological systems and actively participates in writing of scientific papers.

Slobodan Vukicevic is a full professor and head of Laboratory of Mineralized Tissues and Proteomic Center at the Center for Translational and Clinical Research, University of Zagreb School of Medicine. Scientific interests comprise bone and cartilage morphogenetic proteins and development of drugs for regeneration of bone, kidney, pancreas, and heart muscle. Is an invited speaker at international conferences and universities. Received international awards for achievements in science. Chairman and organizer of several international conferences on calcified tissues. Member of EMBO, World Academy of Arts and Sciences (WAAS), and Croatian Academy of Sciences and Arts (CASA). Author of more than 190 manuscripts which is cited more than 6600 times, (h-index 44). Editor of four books on BMPs and inventor on 35 patents. Founder of Genera Research, innovation-based biotechnology company developing a novel regenerative therapy for bone defects via coordinating the collaborative FP7 program OSTEOGROW.

Acknowledgements

We thank Durdica Car and Mirjana Marija Renic for providing animal care and assisting in rat experiments and Ivancica Bastalic and Lucija Kuc ko for administrative support. We are grateful to Vladimir Trkulja for the data analyses and the graphical presentation of results.

Disclosure statement

The authors declare that they have no conflict of interest.

Additional information

Funding

This work was in part supported by the Hrvatska Zaklada za Znanost (Croatian Science Foundation), under Grant BMP1-IsoFor [UIP-09-2014-3509] and Scientific Center of Excellence for Reproductive and Regenerative Medicine (project ‘Reproductive and regenerative medicine - exploration of new platforms and potentials‘, Grant Agreement [KK01.1.1.01.0008] which is funded by the European Union through the European Regional Development Fund).