Article Text

Download PDFPDF
Original article
Vitamin D counteracts fibrogenic TGF-β signalling in human hepatic stellate cells both receptor-dependently and independently
  1. Anja Beilfuss1,
  2. Jan-Peter Sowa1,
  3. Svenja Sydor1,
  4. Mechthild Beste1,
  5. Lars P Bechmann1,
  6. Martin Schlattjan1,
  7. Wing-Kin Syn2,
  8. Inga Wedemeyer3,
  9. Zoltan Mathé4,
  10. Christoph Jochum1,
  11. Guido Gerken1,
  12. Robert K Gieseler1,5,
  13. Ali Canbay1
  1. 1Department of Gastroenterology and Hepatology, University Hospital, University Duisburg-Essen, Essen, Germany
  2. 2The Institute of Hepatology, Regeneration and Repair Group, London, UK
  3. 3Institute for Pathology, University Hospital Cologne, Cologne, Germany
  4. 4Department of General, Visceral and Transplantation Surgery, University Hospital, University Duisburg-Essen, Essen, Germany
  5. 5Rodos BioTarget GmbH, Medical Park Hannover, Hannover, Germany
  1. Correspondence to Professor Ali Canbay, Department of Gastroenterology and Hepatology, University Hospital, University Duisburg-Essen, Hufelandstr. 55, Essen 45147, Germany; ali.canbay{at}


Objective Non-alcoholic fatty liver disease (NAFLD) is closely linked to obesity and constitutes part of the metabolic syndrome, which have been associated with low serum vitamin D (VD). Due to known crosstalk between VD and transforming growth factor (TGF)-β signalling, VD has been proposed as an antifibrotic treatment.

Design We evaluated the association between VD, the vitamin D receptor (VDR) and liver fibrosis in primary human hepatic stellate cells (phHSC) and 106 morbidly obese patients with NAFLD.

Results Treating phHSC with VD ameliorated TGF-β-induced fibrogenesis via both VDR-dependent and VDR-independent mechanisms. Reduction of fibrogenic response was abolished in cells homozygous for GG at the A1012G single nucleotide polymorphisms within the VDR gene. Compared with healthy livers, NAFLD livers expressed higher levels of VDR mRNA and VDR fragments. VDR mRNA was lower in patients homozygous for GG at A1012G and expression of pro-fibrogenic genes was higher in patients carrying the G allele.

Conclusions VD may be an antifibrotic treatment option early in the onset of fibrosis in specific genotypes for VDR. Known polymorphisms of the VDR may influence the response to VD treatment.


Statistics from

Significance of this study

What is already known on this subject?

  • Vitamin D (VD) is reduced in obese subjects.

  • Non-alcoholic fatty liver disease (NAFLD) and obesity are connected via the metabolic syndrome.

  • VD signalling is antagonistic towards the transforming growth factor (TGF)-β pathway in various models.

What are the new findings?

  • Serum VD is reduced in patients with NAFLD and correlates with a fibrogenic state in the liver. Liver fibrosis is associated with reduced full-length VD receptor (VDR) protein expression, but increased VDR protein fragments.

  • VD exerts a VDR-independent effect on early-TGF-β-induced SMAD activation and protects VDR from trypsin digestion in primary human hepatic stellate cells.

  • Reduction of TGF-β-induced profibrogenic gene expression by VD does not occur in cells homozygous for minor allele genotypes of the VDR.

  • Profibrogenic mRNA expression is partially dependent on known polymorphisms of the VDR in patients with NAFLD.

How might it impact on clinical practice in the foreseeable future?

  • The currently ineffective therapy of fibrotic processes by patients might benefit from a targeted approach to individuals without certain variants of the VDR gene. Targets within the TGF-β pathway might provide opportunities for patients with detrimental VDR single nucleotide polymorphisms.


Liver fibrosis and cirrhosis are not only a hallmark of chronic liver disease, but also comprise severe risks for development of hepatocellular carcinoma and acute-on-chronic liver failure. The functional basis for fibrosis and cirrhosis during chronic liver injury is activation of non-parenchymal cells, such as hepatic stellate cells (HSC). These cells secrete extracellular matrix, which serves as a scaffold for cellular reconstitution.1 Activated HSC produce collagen as well as tissue inhibitors of metalloproteinases that inhibit the degradation of extracellular matrix proteins.2 ,3 They further secrete transforming growth factor (TGF)-β, which expands the profibrotic microenvironment by activating neighbouring HSC in paracrine and autocrine fashion.4 Some studies demonstrated a beneficial effect for vitamin D (VD) as it inhibits TGF-β-induced fibrogenesis in vitro5 ,6 and in vivo.7 Accumulating evidence points to close interaction of VD and TGF-β signalling in animal models of fibrosis and in LX-2 cells.8

Increasing numbers of individuals worldwide are overweight or obese.9 A cluster of conditions, commonly referred to as the metabolic syndrome, occurs in association with obesity. In the liver, the metabolic syndrome manifests as non-alcoholic fatty liver disease (NAFLD),10 which is a major risk factor for type-2 diabetes mellitus, progressive alcoholic liver disease and drug hepatotoxicity.11 ,12 Some transplantation centres exclude NAFLD livers as potential allografts,13 although slightly steatotic organs may have an improved regenerative capacity.14 Nevertheless, NAFLD and, in particular, non-alcoholic steatohepatitis (NASH) (NAFLD with cell death and inflammation) are major health concerns, as individuals with NASH are likely to develop fibrosis, cirrhosis and hepatocellular carcinoma, and the complex pathogenic mechanisms that drive disease progression remain to be elucidated. The increased level of free fatty acids (FFA) is one of the major drivers of lipotoxicity;15 ,16 FFAs induce the formation of reactive oxygen species17 which have been shown to promote NAFLD progression. Although the specific role of VD in NAFLD progression remains unclear, obese individuals (who are at risk for developing NAFLD) exhibit repressed levels of serum VD.18 The low levels of VD in obese individuals may be an added risk factor for development of fibrosis and cirrhosis, as there is a loss of anti-TGF-β effects.

In the present study, we evaluated (1) if, and how low VD concentrations promote NAFLD progression; (2) if alterations in the vitamin D receptor (VDR) signalling modulate fibrogenic responses in human NAFLD and primary human HSC (phHSC); (3) if FFAs and VD interact in the fibrogenic response of phHSC; (4) if VD supplementation may inhibit the profibrogenic response by phHSC.

Materials and methods


Data of liver samples collected from 106 morbidly obese patients undergoing bariatric surgery with biopsy-proven NAFLD were analysed and compared with 10 healthy liver samples. All enrolled patients were physically and ultrasonographically examined, and a complete set of laboratory parameters as well as a liver biopsy were obtained.

The study protocol conformed to the ethical guidelines of the revised Declaration of Helsinki (2000, Edinburgh) and was approved by the Institutional Review Board (Ethics Committee) at the University Hospital of Essen. All patients provided written informed consent before enrolment. Patients’ general characteristics are given in table 1.

Table 1

Clinical and biochemical characteristics of patients with non-alcoholic steatohepatitis (NASH) and pre-bariatric surgery for obesity, individuals with NAFL and of healthy controls

Isolation of phHSC

For isolating phHSCs, explanted liver grafts or partially resected liver segments were perfused with Hank's Balanced Salt Solution (HBSS) and subsequently with HBSS containing 440–450 U/mL collagenase to digest the connective tissue. The obtained cell suspension was filtered through a 4 µm mesh and washed three times in HBSS. The supernatant was collected and subjected to density gradient centrifugation, prepared as 12.5% iodixanol solution (Optiprep, Axis-Shield, Wädenswil, Austria), overlayed with a 9% iodixanol solution and overlayed with GBSS (Sigma, Steinheim, Germany). The upper layer, between GBSS and 9% density, was carefully transferred into magnetic-activated cell sorting (MACS) buffer and washed twice. The cell suspension was incubated with CD133 Microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany), and the CD133 HSC were separated by MACS.19

SNP PCR and sequencing

Genomic DNA was extracted from peripheral blood mononuclear cells (patients with NAFLD) or directly from frozen cell culture pellets (phHSC) with QIAamp DNA mini Kit (250) for genomic DNA (Qiagen, Hilden, Germany). PCR amplification and nuclease restriction was done according to protocols individually designed for each target single nucleotide polymorphism (SNP). A detailed overview of sequences, enzymes and conditions is given in extended experimental procedures (online material). Randomly chosen samples were sequenced to verify the correct restriction site. Sequencing was performed with Big Dye Terminator v1.1 Sequencing Kit on an ABI PRISM 3100-Avant Genetic Analyzer (ABI Applied Systems, life Technologies, Carlsbad, California, USA).


All data of patients with NAFLD are expressed as means±SEM. Data of cell culture experiments represent at least n=4 independent measurements and are expressed as means±SEM unless specified otherwise. Statistical significance (p<0.05) was assessed by a two-tailed unpaired or paired Student t test, respectively. All statistical analyses were performed using GraphPad Prism (V.5.03, GraphPad Software, San Diego, California, USA).

For detailed experimental protocols, measurements and methods, please refer to the online supplementary methods.


Serum VD levels are decreased and hepatic VDR expression is elevated in obese patients with NAFLD

To assess the impact of VD serum levels on NAFLD progression, general characteristics, standard liver parameters and serum concentrations of FFAs, hyaluronic acid and VD were collected from 106 morbidly obese (NAFLD on liver biopsy) patients. In liver biopsies obtained from these patients, the expressions of collagen as well as the VDR were analysed.

Despite histological evidence of NAFLD, 84% of the patients exhibited near-to-normal serum transaminases (<50 U/mL for alanine transaminase and for aspartate aminotransferase). Table 1: please also refer to this table for general patient characteristics. According to histological scoring, 97.5% exhibited early fibrosis while 2.5% had advanced fibrosis (see online supplementary table S1 for detailed breakdown). Compared with healthy individuals, patients with NAFLD had significantly higher serum FFA but lower serum VD levels, consistent with previously published reports18 ,20 ,21 (figure 1A, B). No effect of the acquisition date (time of the year) on VD levels was found (data not shown). In addition to histological fibrosis assessment, increased serum levels of hyaluronic acid were found (figure 1C), which was accompanied by enhanced deposition of collagen fibres in the liver detected by Sirius red staining and western blot (figure 1D–F). In order to assess whether low serum VD concentrations correlate with hepatic VDR expression, mRNA and protein expression were analysed. While the hepatic VDR gene expression increased, the levels of full-length VDR protein were repressed compared with healthy livers (figure 2A, B). Intriguingly, VDR fragment in liver tissue of patients with NAFLD was significantly increased compared with controls (figure 2C).

Figure 1

Serum parameters and fibrosis evaluation in patients with non-alcoholic fatty liver disease (NAFLD). The NAFLD cohort showed increased serum concentrations of free fatty acids (FFA) (A) and reduced vitamin D levels (B). Serum hyaluronic acid (C) as a surrogate marker for collagen production, Sirius red staining (D, Fi-Fiii) and collagen1-α protein expression (E) were found elevated. Representative stainings with Sirius Red are shown in panel F, depicting a control liver with slight collagen deposition around vessels (i), liver tissue from a patient with NAFL with increased collagen in the portal area (ii), and from a patient with non-alcoholic steatohepatitis (NASH) with abundant collagen deposition spreading into the periportal areas (iii). Patients with NAFLD were grouped into NAFL or NASH according to the M30 threshold of 275 U/L (<275 U/L=NAFL; ≥275 U/L=NASH), as described by Feldstein et al.23

Figure 2

Vitamin D receptor expression in non-alcoholic fatty liver disease (NAFLD) liver. Within the liver tissue of patients with NAFLD, an increase of vitamin D receptor (VDR) mRNA was detected (A). Although the full-length VDR protein (B) was reduced, the fragment of proteasomal VDR degradation was elevated (C). Patients with NAFLD were grouped into NAFL or non-alcoholic steatohepatitis (NASH) according to the M30 threshold of 275 U/L (<275 U/L=NAFL; ≥275 U/L=NASH). Protein expressions of full length VDR or VDR fragment were normalised to glyceraldehyde 3-phosphate dehydrogenase as loading control. See online supplementary figure S1.

Degradation of the VDR protein by the proteasome usually generates a 20 kDa fragment of the actual VDR.22 To mimic physiological VDR digestion, protein isolates generated from phHSC were exposed to trypsin with or without VD. In all preparations, a decrease in the full-length VDR signal was observed when trypsin was added (see online supplementary figure S1). In the presence of VD, the full-length VDR signal was significantly stronger than without VD. Trypsin-catalysed VDR fragmentation was diminished in the presence of VD. The aggregate data confirm the presence of fibrogenesis in obese individuals with NAFLD. Low serum VD and changes in VDR protein expression suggest that availability of VD to liver cells may be severely impaired in NAFLD, possibly diminishing functioning VDR due to unchecked proteasomal digestion.

FFAs do not enhance fibrogenesis in phHSC

As described above, NAFLD slowly progresses to fibrosis and in the presented cohort, signs for ongoing HSC activation and fibrogenesis were found. Among other mechanisms, FFAs are considered putative ‘hits’ in the progression of NAFLD. As FFAs influence various metabolic pathways (ie, bile acid metabolism24), they might affect VDR expression, thereby leading to a reduction of bioavailable VD in liver tissue. To ascertain whether elevated FFA concentrations, in fact, altered the expression of VDR and/or enhanced fibrogenesis, phHSC cultures were pretreated with FFAs for 1 h, with or without VD supplementation for 24 h (total incubation time with FFA was thus 25 h). FFA treatment did not enhance fibrogenesis in phHSC (see online supplementary figure S2A–F). By contrast, FFA and VD supplementation appeared to repress profibrogenic genes. Consequently, at least in this in vitro system, fatty acids do not activate phHSC. As an additional known stimulant for HSC proliferation, platelet-derived growth factor (PDGF)-BB was employed. Though, in phHSC, no PDGF-BB-induced activation was observable via the analysed mRNA expressions (see online supplementary figure S2G, H).

VD supplementation counters TGF-β-induced fibrogenic response

TGF-β is a major HSC activator and a potent profibrogenic stimulus. To investigate whether VD can ameliorate TGF-β-induced fibrogenesis in primary human cells to a similar extent as described in various models,5 ,7 ,8 phHSC were stimulated with TGF-β for 24 h and optionally treated with VD. Upon TGF-β treatment, mRNA expression of TGF-β, α-smooth muscle actin (SMA), PDGF, collagen1-α, and VDR were significantly upregulated by phHSC (figure 3A–C, see online supplementary figure S3A, B). Additional treatment with VD repressed PDGF gene expression to almost basal levels (figure 3C) and significantly attenuated the increases in TGF-β (figure 3A) and α-SMA (figure 3B). Protein expression of TGF-β, collagen1-α, and VDR, however, were hardly affected (see online supplementary figure S3C–E). Protein expression of SMAD2, a central regulator of TGF-β-induced processes, was similarly unaffected by TGF-β or VD treatment (figure 3E). As expected, phosphorylation of SMAD2 was upregulated considerably upon TGF-β stimulation and was slightly diminished by VD treatment (figure 3F). It has been suggested that VD effects could be mediated by PDIA3 in addition to VDR.25 Upon stimulation with VD or TGF-β alone, no effect on PDIA3 mRNA was observable. Cotreatment with both VD and TGF-β led to a significant increase of PDIA3 expression (figure 3G). Changes in VDR expression on activated HSC and/or the timing of treatment in HSC cultures may modulate the extent of VD-mediated reduction in TGF-β-induced effects.

Figure 3

Fibrogenic and activation-related gene expression in primary human hepatic stellate cells (phHSC). phHSC from patients with non-viral liver diseases were isolated and treated with vitamin D (VD) and/or transforming growth factor (TGF)-β. TGF-β-induced expressions of the mRNAs of TGF-β (A), α- smooth muscle actin (SMA) (B) and platelet-derived growth factor (PDGF) (C) were diminished by VD. However, VD treatment only affected the TGF-β-induced protein expression of α-SMA (D). We found no effect of TGF-β or VD on the amount of SMAD2 protein, a central adaptor of TGF-β signalling (E). In contrast, SMAD2 phosphorylation mediated by TGF-β treatment was ameliorated by the addition of VD (F). Expression of the putative VD membrane receptor PDIA3 was increased only upon costimulation with VD and TGF-β (G). All protein expressions were normalised to glyceraldehyde 3-phosphate dehydrogenase as loading control. *, ***=p Versus untreated <0.05, <0.01 or <0.0001, respectively. See online supplementary figures S2 and S3.

VDR knockdown abolishes VD-mediated anti-TGF-β effect

To evaluate if TGF-β signalling is affected by suppressed hepatic VDR expression, as observed in patients with NAFLD, RNA interference for VDR expression was performed in phHSC. Under optimal conditions, the knockdown efficiency was satisfactory (see online supplementary figure S4A, B) and did not lead to reduced expression of collagen1-α (see online supplementary figure S4C, D) or other tested parameters. Under VDR knockdown, effects of VD on TGF-β were blunted (figure 4A). By contrast, absence of VDR significantly enhanced TGF-β-induced expression of α-SMA (mRNA and protein, figure 4B, D) and pSMAD2 (figure 4F), irrespective of VD treatment. Of note, expression of total SMAD2 was not changed (figure 4E).

Figure 4

Vitamin D receptor counteracts transforming growth factor (TGF)-β-induced gene expression. By means of siRNA knockdown, primary human hepatic stellate cells (phHSC) were deprived of the vitamin D receptor (VDR). Stimulation of cells with TGF-β induced mRNA and protein expressions of TGF-β (A and C) and α- smooth muscle actin (SMA) (B and D). This effect was independent of VDR knockdown. TGF-β-induced TGF-β mRNA expression was reduced by vitamin D (VD) treatment only in the presence of VDR (A). VDR knockdown led to an increase of the TGF-β-induced expression of α-SMA (B and D). Furthermore, SMAD2 expression was slightly increased in cells with VDR knockdown compared with wild-type phHSC, independent of stimulation with TGF-β or VD (E). TGF-β-enhanced phosphorylation of SMAD2 in wild-type phHSC was stronger in cells with VDR knockdown (F). All protein expressions were normalised to glyceraldehyde 3-phosphate dehydrogenase as loading control. *, **, ***=p versus untreated wildtype <0.05, <0.01 or <0.0001, respectively. #, ##, ###=p Versus untreated VDR siRNA <0.05, <0.01 or <0.0001, respectively. See online supplementary figure S4.

Taken together, our data suggest that (1) unimpaired VD–VDR signalling pathway modulates TGF-β signalling, while (2) VDR depletion results in the loss of VD activity (ie, antifibrotic actions). Since some VD–VDR effects might occur early during phHSC activation, we also checked whether VD could influence the initiation of HSC activation.

VDR-independent early effects of VD on SMAD signalling

PhHSC with or without VDR knockdown were treated with TGF-β or TGF-β and VD for 15 min, 30 min or 1 h, respectively. For the entire treatment period, SMAD2 protein expression was initiated and increased in cells irrespective of VDR knockdown (figure 5A, B). Independent of a VDR knockdown, VD treatment significantly reduced SMAD2 expression at 30 min after stimulation only. An even more prominent reduction by VD treatment was observed for phosphorylated SMAD2 at 15 min and 30 min (figure 5C, D). These findings demonstrate that VD is able to counteract TGF-β-induced SMAD2 expression and phosphorylation in a VDR-independent fashion.

Figure 5

Rapid SMAD2 phosphorylation is receptor-independently reduced by vitamin D. Protein expression of the central transforming growth factor (TGF)-β pathway adaptor SMAD2 increased after TGF-β stimulation from 15 min to 1 h postadministration (A). Treatment with vitamin D (VD) slightly decreased TGF-β-induced SMAD2 expression. The absence of vitamin D receptor (VDR) did not affect this outcome (B). SMAD2 phosphorylation strongly increased due to TGF-β treatment from 15 min to 1 h postadministration (C). Again, and independently of the presence of VDR, VD was able to reduce the degree of SMAD2 phosphorylation (D). All protein expressions were normalised to glyceraldehyde 3-phosphate dehydrogenase as loading control.

Human stellate cell response to TGF-beta and vitamin D is partially dependent on allele distribution of the VDR

Since SNPs within the VDR gene are associated with liver stiffness,26 a possible sign for liver fibrosis/cirrhosis, we investigated if the interaction of VDR and TGF-β in phHSC may also be associated with known SNPs of the VDR gene. To this end, in a subset of the experiments described above (see figure 3), known SNPs within the VDR gene were analysed (for details, please refer to online supplementary methods). Gene expressions under TGF-β and VD cotreatment (as shown in figure 3: condition TGF-β and VD) were categorised by individual genotype of phHSC for each SNP and evaluated for differential effects by 1-way analysis of variance with Tukey's Multiple Comparison Test. There were four known SNPs, where genotype influenced the reaction to VD treatment after TGF-β stimulation. For each of these four SNPs, there was one (homozygous) genotype, where mRNA expression of the respective target gene was not affected by VD treatment after TGF-β dosage (figure 6). Specifically, after stimulation with TGF-β, VD treatment did not lead to any reduction in TGF-β mRNA (corresponding to figure 3A), if cells harboured homozygous genotypes for A1012G (GG; figure 6A), BsmI G/A (AA; figure 6B) and TaqI T/C (TT; figure 6C). Similarly, TGF-β-induced expression of collagen1-α mRNA (corresponding to online supplementary figure S3A) was unaffected in cells from homozygous donors for Apl A/C (CC; figure 6D) when treated with VD.

Figure 6

Single nucleotide polymorphisms (SNPs) in the vitamin D receptor (VDR) gene affecting fibrogenic mRNA expression. In a subset of the primary hepatic stellate cells (phHSC) stimulated with transforming growth factor (TGF)-β and treated with vitamin D (VD) (see figure 3 condition TGF-β and VD for complete results) known SNPs of the VDR gene were analysed. TGF-β-induced TGF-β mRNA expression (corresponding to figure 3A) was not reduced due to VD treatment in phHSC with the genotype GG at the A1012G SNP (A), AA at the BsmI SNP (B) and TT at the TaqI site (C). Cells homozygous for CC at the ApaI SNP exhibited no reduction in collagen1-α mRNA expression (D; corresponding to online supplementary figure S3A) when cotreated with TGF-β and VD. See online supplementary figure S5. Genotypes are depicted in white for homozygous major allele, grey for heterozygous and black for homozygous of the minor allele, respectively.

Differential expression of profibrogenic genes depending on VDR polymorphisms in patients with NAFLD

As VD-mediated changes to TGF-β-induced gene expressions were significantly affected by VDR polymorphisms in phHSC, we aimed to connect possible genetic variants of VDR to the results found in patients with NAFLD. The same SNPs were analysed, as described above for phHSC. Distribution of genotypes for the analysed SNPs was not equal between phHSC and patients with NAFLD (see online supplementary figure S5).

No associations of the analysed SNPs with fibrosis (METAVIR Score) or with the NAFLD activity score were found. Though, for three SNPs (A1012G, BsmI, Tru9 I) significant differences of VDR mRNA or profibrogenic mRNA expressions were detected (figure 7). In detail, homozygous carriers of the GG allele of the A1012G SNP exhibited significantly lower VDR mRNA than AA homozygous or heterozygous patients (figure 7A). Conversely patients homozygous for the major allele (AA) exhibited significantly lower TGF-β (figure 7B) and α-SMA expression (figure 7C) compared with heterozygous/homozygous GG carriers. Additionally, patients homozygous for AA (at A1012G) exhibited a significant correlation of the mRNA expression of VDR and collagen1-α, which did not occur in the whole patient cohort or other genotypes (figure 7D). Similarly, patients homozygous for the major allele of BsmI (GG) had significantly lower TGF-β mRNA levels in comparison with heterozygous/homozygous AA carriers (figure 7E). For Tru9I, only patients either homozygous for the major allele (GG) or heterozygous (AG) were present in the cohort, probably due to the very low incidence of the minor allele. Homozygous GG carriers exhibited significantly higher mRNA expression of VDR (figure 7F) and α-SMA (figure 7G).

Figure 7

Effects of known polymorphism in the vitamin D receptor (VDR) gene on profibrotic gene expression in patients with non-alcoholic fatty liver disease (NAFLD). Expression levels of profibrotic genes in liver biopsies from patients with NAFLD were stratified according to genotypes of nine known single nucleotide polymorphisms (SNPs) within the VDR gene. Patients homozygous for the minor allele of the A1012G SNP (GG) had significantly lower VDR mRNA expression (A). By contrast, patients homozygous for the major allele of this polymorphism (AA) exhibited lower TGF-β (B) and α-smooth muscle actin (SMA) (C) mRNA levels. In the subgroup of these patients homozygous for A1012G AA, mRNA expressions of VDR and collagen1-α were significantly correlated. This effect did not occur in the whole patient cohort or other genotype combinations. NAFLD patients homozygous for the major allele of the BsmI polymorphism (GG) exhibited significantly lower TGF-β mRNA expression (E). Homozygous GG carriers for the Tru9I SNP had significantly higher mRNA expressions of VDR (F) and α-SMA (G).

In summary, some known polymorphisms of the VDR gene affect mRNA expression of the VDR itself and of profibrogenic genes in HSC in vitro and in an NAFLD patient cohort. Response to VD treatment may be dependent on the genotype of certain VDR SNPs.


HSC are the main cell type responsible for extracellular matrix deposition, which involves activation of HSC and the TGF-β pathway. Earlier studies in intrauterine tumour cells5 and HL-60 cells6 focused on the crosstalk between TGF-β signalling as a hallmark of fibrogenesis and vitamin D metabolism. Abramovitch et al7 demonstrated in a rat model of liver fibrosis and rat HSC, that VD can counter fibrosis. More recently, Ding et al8 elegantly unravelled a close genetic interaction of VDR and TGF-β signalling via enhanced VDR-binding to VD response element in the presence of SMAD3 on the same binding site. Chen et al27 added another piece of information, demonstrating interaction of VDR with the putative VD membrane receptor PDIA3. The present study corroborates these findings by showing that VD modulates TGF-β effects in cultured phHSC and that these effects occur during VDR deficiency and is partially dependent on VDR-SNPs.

Our aggregate data show that VD inhibits TGF-β-associated effects in phHSC. However, the weak effects observed might be explained by possible HSC preactivation and, thus, refractoriness to VD and/or VDR signalling. The individual aetiology of liver disease may further affect the response of isolated phHSC to VD treatment. For example, recent findings of Skoien et al28 suggest multiple fibrogenic mechanisms. This may be also indicated by our preliminary studies, which showed that phHSC isolated from patients with viral hepatitis did not respond or responded only marginally to VD treatment (data not shown). Furthermore, it appears that an optimal therapeutic effect may only be achieved in the early stages of disease.

Interaction between VD and TGF-β may also depend on the time of stimulation and/or cellular uptake/binding of VD or TGF-β, respectively. While it has been known for some time that VD induces rapid responses (see review by Yang et al29), current studies have shown that these may be mediated by the membrane receptor PDIA3.25 Moreover, the same group showed that PDIA3 and VDR interact on levels of gene expression and receptor binding of VD.27 This is consistent with our results, as short-term suppression of SMAD2 phosphorylation was independent of VDR expression. PDIA3 mRNA expression itself was induced only upon costimulation by VD and TGF-β in our experiments, which may indicate a crosstalk between PDIA3 and TGF-β signalling. Future work will be needed to ascertain whether early changes in TGF-β-induced effects are mediated by VD-PDIA3 activation. We hypothesise that (very) early VD supplementation may have a beneficial antifibrotic effect. This notion is supported by Abramovitch et al,7 who demonstrated a reduction of thioacetamide-induced liver fibrosis in rats only when VD was administered simultaneously with injury.

Earlier studies reported reduced serum VD levels in liver disease,30–32 but these focused on individuals with chronic viral hepatitis while neglecting the role of VD in NAFLD. In obese patients with NAFLD, the reduced levels of serum VD may be related to its aberrant storage in adipose tissue.24 VDR is itself degraded by proteasomal digestion, thereby amplifying a VD-deficient state within HSC.22 In liver tissue of patients with NAFLD, we found increased VDR mRNA levels, but a reduced amount of protein, which was associated with elevated amounts of VDR fragments. Our data show that VD supplementation may reduce VDR degradation and attenuate TGF-β-induced HSC fibrogenesis. By contrast, VDR knockdown resulted in enhanced TGF-β-induced effects in phHSC, suggesting that VDR—independent of VD—may keep TGF-β signalling in check. This would imply diminished full-length VDR protein in patients with NAFLD to facilitate a profibrotic status of HSC. The recent findings connecting genetic regulation of TGF-β via SMAD3 and VDR by enhanced binding activity of VDR on SMAD3-bound sites additionally demonstrate a close, though highly complex interaction of these signalling pathways.8 SMAD3 bound to certain regulatory elements facilitated binding of the VDR and, thus, counter regulation of TGF-β activation in LX-2 cells and rat HSC. VD may be a valuable antifibrotic supplement, but probably in a rather limited time window during early fibrogenesis or in very high doses33 to counteract permanently upregulated HSC activity. However, recent publications have described novel pharmaceutical VDR agonists,34 ,35 which might overcome this problem by stronger binding to the VDR and enhanced activation of the VDR pathway. Moreover, combined treatment with VD and S-Farnesylthiosalicylic acid has been shown to reduce PDGF-induced HSC proliferation in vitro.36 This synergistic effect may enhance the utility of VD treatment, especially in early stages of liver fibrosis. Further tests will have to demonstrate the efficacy of these components on possible cotreatment options in ongoing fibrosis and cirrhosis.

Differences in the protein stability under low VD conditions may indicate the risk of individual progression from simple steatosis (NAFL) to NASH. Additionally, certain VDR SNP(s) may pinpoint positive responders to VD treatment. Indeed, Grünhage et al26 found that liver stiffness is associated with certain VDR SNPs. Among a French cohort with type 2 diabetes, VDR SNP in intron 8 and exon 9 were associated with obesity.37 In the current NAFLD cohort, mRNA expression of VDR itself and of profibrogenic genes was significantly affected by presence of homozygous alleles for A1012G (major/minor), BsmI (major) and Tru9I (major). The A1012G polymorphism is located in the promoter region (at a putative GATA-3 binding site), possibly affecting VDR mRNA expression. Strikingly, this SNP also affected the influence of VD on TGF-β-treated phHSC. VD could not ameliorate TGF-β-induced TGF-β expression in phHSC homozygous for GG at the A1012G site. Another SNP affecting fibrogenesis in patients with NAFLD and phHSC response to VD treatment was BsmI. BsmI is considered a silent SNP without change of the amino acid sequence, though possibly affects mRNA stability of the VDR. However, we found reduced TGF-β expression in patients with NAFLD with the major allele for BsmI (GG) and no effect of VD on TGF-β-induced TGF-β expression in phHSC homozygous for the minor allele (AA). The Tru9I polymorphism is located in intron 9 and has currently no clear effect, but may influence translation. Patients with NAFLD with homozygous major allele for Tru9I (GG) exhibited higher VDR as well as α-SMA mRNA expression, again suggesting a complex interaction of these pathways. The reduction of TGF-β-induced fibrogenesis by VD in phHSC was affected by two additional SNPs. The incidence of some SNPs differed to a significant extent between the phHSC and the NAFLD cohort. This might explain why different SNPs affect the outcome in NAFLD and phHSC. Still, A1012G had similar impact on the VDR-TGF-β interaction in both phHSC and patients with NAFLD and may thus be at a crucial site within the VDR gene. Taken together it becomes clear, that the genetic basis affects expression and function of the VDR in patients with NAFLD. This has to be taken into account when discussing or investigating VD as possible antifibrotic therapy. There may indeed be patients, who could benefit from such a treatment, while others could lack a genetic basis for effectively countering fibrosis via the VD-VDR axis.

Our findings in freshly isolated phHSC and in patients with NAFLD corroborate earlier reports suggesting a potential therapeutic effect of vitamin D on the progression of this disease entity. Future studies will be needed to elucidate how VD interacts with putative receptors, VDR and PDIA3, and their effects on TGF-β signalling, and how genetic VDR variants influence fibrogenic outcomes. The combination of genetic and mechanistic insights could provide us with a rational approach to treating NAFLD.


We are grateful to Lena Wingerter for her expert assistance in cell preparation and sample acquisition; and Mathias Schlensak, MD (General and Visceral Surgery, Alfried Krupp Hospital, Essen, Germany) for his close collaboration and timely supply of liver samples.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • AB are J-PS contributed equally.

  • Contributors Study concept and design: AB, JPS, RKG, AC. Acquisition of data: AB, SS, MB, MS, IW. Analysis and interpretation of data: AB, JPS, LPB, CJ, RKG, AC. Drafting of the manuscript: JPS, RKG, WKS, AC. Critical revision of the manuscript for important intellectual content: JPS, WKS, RKG, AC. Statistical analysis: AB, JPS, LPB. Obtained funding: GG, AC. Administrative, technical, or material support: SS, MB, LPB, MS, IW, ZM, CJ, RKG. Study supervision: GG, RKG, AC.

  • Funding This work was supported by the Deutsche Forschungsgemeinschaft (DFG, grant 267/8-1 and 267/6-1) and the Wilhelm Laupitz Foundation to AC.

  • Competing interests None.

  • Ethics approval Ethics Committee at the University Hospital Essen.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles