TGF-β and Iron Differently Alter HBV Replication in Human Hepatocytes through TGF-β/BMP Signaling and Cellular MicroRNA Expression

The nature of host-virus interactions in hepatitis B virus infection is incompletely understood. Since soluble factors, e.g., cytokines and metals, may exacerbate liver injury in chronic hepatitis, we considered that defining the effects of receptor-mediated signaling upon viral replication will be significant. Consequently, we studied effects of iron or TGF-β-induced TGF-β/BMP signaling in the HepG2 2.2.15 cell model of hepatitis B virus replication. We found iron and TGF-β increased hepcidin mRNA expression or TGF-β receptor kinase activity, respectively, which indicated that 2.2.15 cells responded appropriately to these substances. However, iron increased but TGF-β decreased hepatitis B virus mRNA and DNA expression. TGF-β induced expression at the mRNA level of multiple TGF-β/BMP pathway genes. This change was not observed in iron-treated cells. On the other hand, presence of SMAD proteins in iron or TGF-β-treated cells, including of SMAD4, did confirm convergence of TGF-β/BMP signaling pathways under these conditions. Since transcription factors in TGF-β/BMP signaling pathways could not have directly targeted hepatitis B virus itself, we studied whether iron or TGF-β exerted their effects through alternative mechanisms, such as by involvement of antiviral cellular microRNAs. We discovered cellular microRNA expression profiles were significantly different in iron or TGF-β-treated cells compared with untreated control cells. In many cases, exposure to iron or TGF-β changed microRNA expression in opposite directions. Introduction in cells of sequences representing such differentially expressed microRNAs, e.g., hsa-miR-125a-5p and -151-5p, even reproduced effects on virus replication of iron- or TGF-β. We surmised that TGF-β/BMP pathway members, i.e., SMADs, likely governed iron or TGF-β-induced microRNA expression. Iron may have mediated Drosha/DGCR8/heme-mediated processing of microRNAs. In turn, cellular microRNAs regulated replication of hepatitis B virus in iron or TGF-β-treated cells. This knowledge should advance studies of mechanisms in viral-host interactions, hepatic injury, and therapeutic developments for hepatitis B.


Introduction
Intra-and extracellular soluble signaling molecules are involved in hepatitis virus replication but these interactions are not well understood. For instance, inflammatory cytokines affect hepatitis B virus (HBV) replication by recruiting more than one signaling pathways. Among these, interleukins (e.g., IL12, IL18) may inhibit HBV replication, including with recruitment of interferon (IFN)-c released from NK or T cells [1,2]. Interferon-a has widely been used for treating HBV with JAK/STAT signaling serving intermediary roles [3]. The role of these intracellular signaling pathways in transducing antiviral effects of interferon is far from complete and new information is still emerging [4]. Other cytokine pathways off note include tumor necrosis factor-a, which suppressed HBV replication [5]. Also, transforming growth factor (TGF)-b inhibited HBV replication [6], presumably with involvement of TGF-b signaling through SMAD-2 and -3 [7]. The canonical TGF-b signaling pathways involve SMADs -2 and -3 compared with bone morphogenetic protein (BMP) signaling via SMAD-1, -5, and -8. Nonetheless, after activation of TGF-b-or BMP receptors leads to heteromeric complexing between SMADs, followed by engagement with the common-mediator, SMAD-4, which is required and sufficient for regulation of nuclear transcription, and in this way, brings together TGF-b/BMP signaling pathways. How these diverse intracellular signaling pathways may regulate replication of HBV (or other viruses) is yet to be clarified.
Disease-modifying cofactors, e.g., iron, are capable of altering HBV replication. In clinical studies, elevated hepatic iron content has been associated with higher prevalence of HBV infection [8], as well as worse outcomes in chronic hepatitis [9]. However, the molecular basis by which iron may alter HBV replication is unknown. Recently, hepatic release of hepcidin was found to be important in iron homeostasis, by decreasing intestinal iron absorption as well as hepatic iron uptake. As hepcidin exerts its intracellular effects by TGF-b/BMP signaling [10], a relationship emerged between this molecule and other intracellular mediators of cytokines. Unexpected signaling mechanisms were found to regulate hepcidin expression, e.g., epidermal growth factor, and also hepatocyte growth factor, which transduced their effects on hepcidin through PI3 kinase or MEK/ERK pathways [11]. Interestingly, iron-induced hepcidin expression altered HCV replication in cultured cells [12].
Therefore, intracellular signaling pathways could regulate hepatitis virus replication in many ways. More knowledge in this area will be significant for virus-host interactions and hepatic injury. Also, cytokines, chemokines and receptors expressed during host-viral interactions have elicited interest as targets for antiviral therapies [13].
Here, we reasoned that study of intersections in signaling by cytokines, e.g., TGF-b, on the one hand, and signaling by small molecules, e.g., iron, on the other hand, would help clarify mechanisms in HBV replication. One possibility was that iron and TGF-b exerted their effects by mechanisms completely independent of one another. Another possibility was that these molecules shared common intermediaries but that were regulated differently.
We used HepG2 2.2.15 cells for our studies since these express stably transduced HBV genomes with virus replication and have been suitable for mechanistic studies in vitro [14,15]. This cell line model permitted us to characterize the effects of TGF-b and iron on HBV replication, to demonstrate activation of intracellular signaling along TGF-b and BMP pathways, and to establish potential effector mechanisms in regulation of HBV replication.

Cells
HepG2 2.2.15 cells were derived from HepG2 cells stably transfected with full-length HBV genomes (14). HepG2 cells were originally from American Type Culture Collection (ATCC, Manassas, VA). Cells were cultured in Dulbecco's Modified Eagle's medium with 10% fetal bovine serum, 1% glutamine, 1% nonessential amino acids, and antibiotics (DMEM). After doseranging studies, we used 20 ng/ml TGF-b, 100 mM iron or 480 ng/ml TKI with cell culture for up to 48 h. For assays, cells were lysed in 100 mM Tris-HCl, pH 8.0, 0.2% Nonidet P-40 and protease inhibitors (Calbiochem Cocktail set III, EMD4Biosciences, Darmstadt, Germany). Protein content was measured by Bradford Reagent (Pierce Research Products, Rockford, IL). To evaluate cell viability or changes in cell numbers, MTT assays were performed, as described previously [15].

HBV Assays
To demonstrate hepatitis B core antigen (HBcAg), cell lysates were resolved in 1.2% native agarose gels, followed by transfer to nylon membranes (Ambion, Life Technologies, Carlsbad, CA). HBcAg was detected with anti-HBc (Dako, Carpinteria, CA), as described previously [14]. For southern blotting, protein-detergent complexes were precipitated in 2.5 M KCl, viral pellet was dissolved in Tris-EDTA, and incubated with proteinase K (Ambion) for 1 h at 37uC. DNA was extracted with phenolchloroform and precipitated by ethanol. 20 mg DNAs were resolved in 1.2% agarose gels and transferred to nylon membranes (Ambion). To analyze HBV RNA with northern blotting, 20 mg total cellular RNAs were extracted by Trizol Reagent (Life Technologies), resolved in 1.0% formaldehyde-agarose, and transferred to nylon membranes (Ambion). Nonradioactive HBV probe was prepared with psoralen-Biotin labeling kit (Ambion), as described by the manufacturer, with full-length HBV DNA from EcoRI-linearized pCP10 plasmid [16]. Prehybridization and hybridization of blots was as described previously [15]. Northern blots were rehybridized with human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe, as described previously [15]. Replicate blots were quantified by densitometry.

MicroRNA Studies
Profiles of microRNA expression were obtained in total cellular RNA samples by microfluidics approach with arrays (Sanger databases, versions 9.1 and 11.0; LC Sciences, Dallas, TX). Background was subtracted with regression-based mapping on 5-25% of lowest intensity data excluding blank spots. Transcripts with undetectable signals below 36 background, incorporating spot analysis parameters, were excluded. Later, we excluded transcripts with ,500 arbitrary signals as these are not identified by qRT-PCR. RNA hybrids were examined in HBV subtype ayw (Genbank accession number 665257.1) with miRanda program using minimal free energy (mfe) below 212 kcal/mol and miRanda score of over 140 as selection criteria.
We did not determine the effects of TKI on TGF-b-induced gene expression because it successfully blocked TGF-b response in 2.2.15 cells. However, we studied whether TKI would alter cellular responses to iron. In cells treated by iron plus TKI, we found no increases in expression of TGF-b superfamily cytokines or receptors, although expression of Nodal, INHA and LTBP4 declined by 2-3-fold.

SMAD Protein Expression in Cells
We immunostained cells for phosphorylated SMAD-2, -3 and SMAD-4 proteins to confirm changes in TGF-b/BMP signaling. In untreated cells, SMAD-2 was observed occasionally in cell cytoplasm (Fig. 3A). Iron did not alter SMAD-2 expression. This was different in TGF-b-treated cells, since SMAD-2 was now expressed in 82% of cells, p,0.05, along with translocation of the protein in many cells to nuclei. By contrast, iron plus TKI did not affect SMAD-2 expression. SMAD-3 was expressed well under basal conditions and also in iron-or TGF-b-treated cells (Fig. 3B). Expression of SMAD-4 changed prominently with significant increases from basal levels in iron-(53%) or TGF-b-treated cells (57%), p,0.05. In cells treated with iron plus TKI, increases in SMAD-4 expression were limited (Fig. 3C). These findings indicated that TGF-b signaling contributed to SMAD-4 expres-

Iron or TGF-b Altered Cellular miRNA Expression
To identify potential transducers of TGF-b/BMP signaling in HBV replication, we analyzed miRNA expression. We chose to study this mechanism because expression of HBV genes should not have been directly regulated by SMADs activated by TGF-b/ BMP signaling. On the other hand, SMADs can regulate expression of cellular miRNA, which, in turn, could have altered HBV replication (see below).

Discussion
These studies established 2.2.15 cells were appropriately responsive to iron and TGF-b. In case of iron, this was evidenced by greater hepcidin expression, which is transcriptionally regulated via BMP signaling [20], and requires the cofactors, hemojuvelin and neogenin [21]. Responsiveness to TGF-b was evidenced by receptor kinase activity blocked by TKI. TGF-b changed expression of TGF-b/BMP pathway genes and targets, along with increases in phosphorylated SMAD-2, -3 and -4, as was anticipated. Iron did not alter expression of TGF-b/BMP pathway genes at mRNA levels but SMAD-4 expression linked TGF-b and BMP signaling pathways in iron-treated cells [7]. Despite association of iron with unfavorable outcomes in people with HBV [8,9], underlying mechanisms directing HBV replication were unknown. Previously, depletion of iron in HepG2 cells decreased HBV production [22]. This was consistent with increased HBV replication here in iron-treated HepG2 cells. We came across another study of TGF-b and HBV replication in HepG2 cells [6], where TGF-b decreased HBV replication, similar to our results. However, underlying mechanisms in how TGF-b directed HBV replication were unknown.
The role of TGF-b-induced SMADs in transcriptional controls was one possibility. Upon ligand-binding, the constitutionally active type II TGF-b receptor phosphorylates type I TGF-b receptor, followed by recruitment of SMAD-2 and -3, which complex with SMAD-4 for transcriptional activation [7]. BMP receptors activate SMAD-1, -5 and -8, which too complex with SMAD-4 for transcriptional activation. Therefore, SMAD-4 connects TGF-b and BMP signaling. Under some situations, TGF-b-or BMP-specific SMADs may be simultaneously activated [23]. We considered that cellular miRNAs could have served intermediary roles in transducing the effects of TGF-b/ BMP signaling on HBV replication. Recently, cellular miRNAs attracted much attention as ubiquitous antiviral regulators, including for HBV [18,19,24]. miRNAs negatively regulate gene expression by sequence-specific binding of mRNAs followed by mRNA degradation or interference in mRNA translation. The identification of promoter sites in genomic regions encoding miRNAs, including SMAD-binding elements (SBE) [25,26], was noteworthy to us. Multiple miRNA were shown to contain SBE, e.g., hsa-let-7a, -7b, -7c, -7d, -7e, -7f, -21, -23b, etc. [26]. These miRNA were expressed in iron-or TGF-b-treated cells in our study. However, miRNA expression was different in iron-or TGF-b-treated cells. This difference was not explained by SMADs alone. Also, this difference was not explained by transcription factors regulating iron homeostasis, e.g., hemojuvelin or neogenin, since these are not incriminated in miRNA transcription.
Another explanation was provided by mechanisms in processing of pri-miRNA to mature miRNA [27]. All canonical pri-miRNA undergo sequential cleavages in nucleus and cytoplasm by Dicer or Drosha RNAse III enzymes for maturation. Drosha requires the RNA-binding protein, DiGeorge Critical Region 8 (DGCR8), for pre-miRNA cleavage. Recently, heme was identified as a constituent of Drosha-DGCR8 complex, with the ionic state of heme-bound iron determining whether the complex will be capable of pri-miRNA cleavage [27]. In the ferric state, Drosha-DGCR8-heme complex was highly stable and actively processed pri-miRNA to mature miRNA. By contrast, transition to ferrous state abolished pri-miRNA cleavage by Drosha-DGCR8-heme complex. Thus, alterations in intracellular environment, leading to ferric or ferrous state of iron in Drosha-DGCR8-heme complex, likely accounted for differential miRNA expression in iron-or TGF-b-treated cells.
Our miRNA expression studies showing modulation of the effects of iron-or TGF-b on HBV replication were instructive. As hsa-miR-125a-5p and -151-5p mimics prevented iron-induced increases in HBV replication, while their antagonists prevented TGF-b-induced decreases in HBV replication, this established miRNA were the principal intracellular effectors of TGF-b/BMP signaling.
Differentially expressed miRNA in iron-treated cells, and for that matter, TGF-b-treated cells, target genes widely, including pathways in cell stress and toxicity, cell proliferation, immune responses, metabolism, etc., which will be relevant for chronic hepatitis. Iron plays major roles in oxidative stress and cytotoxicity, which coupled with perturbations in HBV replication, may alter natural history of hepatitis. TGF-b is especially significant for hepatic growth control and also for fibrosis. Therefore, regulation of miRNA expression by iron and TGF-b will provide opportunities for understanding disease mechanisms in HBV-related liver injury and fibrosis. Moreover, mechanisms directing antiviral miRNA expression in iron-or TGF-b-treated cells should be helpful for drug development, according to considerations discussed previously by other investigators [27].