Gli1 activation and protection against hepatic encephalopathy is suppressed by circulating transforming growth factor β1 in mice

Abstract

Background & aims: Hepatic encephalopathy (HE) is a neurologic disorder that develops during liver failure. Few studies exist investigating systemic-central signalling during HE outside of inflammatory signalling. The transcription factor Gli1, which can be modulated by hedgehog signalling or transforming growth factor β1 (TGFβ1) signalling, has been shown to be protective in various neuropathies. We measured Gli1 expression in brain tissues from mice and evaluated how circulating TGFβ1 and canonical hedgehog signalling regulate its activation.

Methods: Mice were injected with azoxymethane (AOM) to induce liver failure and HE in the presence of Gli1 vivo-morpholinos, the hedgehog inhibitor cyclopamine, Smoothened vivo-morpholinos, a Smoothened agonist, or TGFβ-neutralizing antibodies. Molecular analyses were used to assess Gli1, hedgehog signalling, and TGFβ1 signalling in the liver and brain of AOM mice and HE patients.

Results: Gli1 expression was increased in brains of AOM mice and in HE patients. Intra-cortical infusion of Gli1 vivo-morpholinos exacerbated the neurologic deficits of AOM mice. Measures to modulate hedgehog signalling had no effect on HE neurological decline. Levels of TGFβ1 increased in the liver and serum of mice following AOM administration. TGFβ neutralizing antibodies slowed neurologic decline following AOM administration without significantly affecting liver damage. TGFβ1 inhibited Gli1 expression via a SMAD3-dependent mechanism. Conversely, inhibiting TGFβ1 increased Gli1 expression.

Conclusions: Cortical activation of Gli1 protects mice from induction of HE. TGFβ1 suppresses Gli1 in neurons via SMAD3 and promotes the neurologic decline. Strategies to activate Gli1 or inhibit TGFβ1 signalling might be developed to treat patients with HE.

Keywords: Acute liver failure; Azoxymethane; Hepatic encephalopathy; SMAD3; Sonic hedgehog; TGFβ1.

Fig. 1. Acute liver insult led to cortical Gli1 activation

(A) Time course of neurological decline present in the AOM HE model. Neurological score is a measure of 6 reflexes scored between 0 and 2 with a lower score indicating greater neurological decline. The time points of specific stages of neurological decline are identified on the graph. (B) H&E histochemistry in vehicle and AOM-treated livers at the indicated timepoints. (C) Cortical Gli1 mRNA expression after AOM treatment at the indicated stages of neurological decline. (D) Gli1 immunostaining in the cortex at the various neurological stages following AOM injection. Subcellular location of immunoreactivity is indicated by a black arrow (cytoplasmic staining) or red arrow (nuclear staining). Quantitation of Gli1 immunohistochemistry including relative field staining intensity (left axis) and percentage of Gli1-positive cells that undergo nuclear translocation (right axis). (E) Gli1 expression in vehicle and AOM cortex at coma was assessed by immunofluorescence and counterstained with the neuron marker NeuN or the astrocyte marker GFAP. (F) Gli1 immunohistochemistry in patients from controls or HE patients. Quantification of Gli1 staining intensity and percent of nuclear translocated Gli1 positive cells are indicated. For mRNA analyses and quantitative IHC, data were reported as mean±SEM (n=4). *p<0.05 compared to vehicle cortex.

Fig. 2. Suppression of Gli1 expression exacerbated neurological decline

Mice were infused with Gli1-VM or mismatched-VM into the cortex 3 days prior to AOM injection. (A) Cortical Gli1 knockdown was validated by Gli1 immunofluorescence (red) near the infusion site (white arrow) with 4',6-diamidino-2-phenylindole (DAPI) (blue) used as a counterstain. (B) Diagram of Gli1-VM knockdown area from coronal and horizontal views of whole mouse brain. Diffusion of morpholino was 1.5 mm and is indicated by green circle. (C) Neurological decline was assessed using neurological score as previously described. Gli1-VM treatment significantly worsened neurological decline in AOM mice compared to mismatched-VM mice (p=.0435). (D) Gli1-VM infused AOM-treated mice entered coma significantly earlier than mismatched-VM infused, AOM-treated mice. (E) Liver damage was assessed in Gli1-VM and mismatched-VM mice by H&E histochemistry. Data were reported as mean±SEM (n=8). *p<0.05 compared to mismatched-VM cortex.

Fig. 3. TGFβ1 expression was upregulated in the liver and increased in the brain following HE

(A) Liver TGFβ1 expression was assessed at coma in vehicle and AOM-treated mice by RT-PCR (B) and immunohistochemistry. (C) TGFβ1 (red) and albumin (green) colocalization in vehicle and AOM livers at coma. DAPI (blue) was used as a nuclear stain. (D) Serum TGFβ1 in vehicle and AOM-treated mice at coma by immunoblotting with albumin used as a loading control. (E) Cortical TGFβ1 protein expression at coma in vehicle and AOM-treated mice. (F) Cortical TGFβ1 mRNA expression in vehicle and AOM-treated mice at coma. For protein and mRNA analyses, data were reported as mean±SEM (n=4).

Fig. 4. Treatment of AOM mice with TGFβ neutralizing antibodies was neuroprotective

A) Mice pretreated with TGFβ neutralizing antibodies (anti-TGFβ) prior to AOM injection had significantly lessened neurological decline (p=.0140). Neurological decline was assessed using the neurological score as previously described. (B) Time to coma for AOM and AOM + anti-TGFβ mice. (C) Percent of total brain water in cortex from vehicle, AOM, AOM + anti-TGFβ, or anti-TGFβ mice at coma. (D) Relative fluorescence of phospho-SMAD3 in vehicle, AOM, AOM + anti-TGFβ, or anti-TGFβ cortex at coma (E) and control or HE patient cortex. (F) Cortical Gli1 mRNA expression in vehicle, AOM, AOM + anti-TGFβ, or anti-TGFβ mice at coma. For ELISA and mRNA analyses, data were reported as mean±SEM (n=4). For time to coma and neurological decline analyses, data were reported as mean±SEM (n=12). *p<0.05 compared to vehicle liver or vehicle cortex. #p<0.05 compare to AOM cortex.