The ancient Greek myth of the eagle feeding on the liver of Prometheus, which regrows after each assault, is frequently used as a metaphor to illustrate the remarkable regenerative potential of this complex organ. Currently, it is clear that multiple liver-resident cell types linked by intricate signaling networks coordinate liver function, homeostasis, and regeneration. Signals from nonparenchymal liver cells tune specific hepatocyte cell functions and may have profound consequences for viruses replicating in hepatocyes. To date, we know relatively little about how such regulatory processes influence the course of viral infections in the liver. This is, in part, because of a lack of robust small animal models for hepatotropic viruses, for example, hepatitis C virus (HCV). Moreover, the development of adequate cell-culture models reflecting the morphology and functions of polarized, differentiated hepatocytes in the context of additional liver-resident cell types has proven difficult.
In this issue of Hepatology, Rowe et al. report on a set of elegant experiments involving coculture of primary human liver sinusoid endothelial cells (LSECs) and liver cells to explore how their interplay influences HCV. They show that bone morphogenetic protein 4 (BMP4), which is secreted from LSECs, promotes HCV RNA replication within hepatocytes.
In a first set of experiments, Huh-7.5 cells, a hepatocarcinoma cell line highly permissive for HCV infection and replication, were incubated with conditioned medium derived from primary human LSECs. By using this approach, the researchers observed increased permissiveness of Huh-7.5 cells for infection by cell-culture–derived HCV and for propagation of HCV replicons. The replicons used were HCV subgenomes lacking viral structural genes and were transfected as RNA into cells. Therefore, the researchers' results highlight that a soluble factor within the LSEC-conditioned medium facilitated viral RNA replication, rather than cell entry or virus particle production.
To identify the culprit with proviral activity, Rowe et al. followed a rational screening approach. Considering the previously reported observation that hepatocyte-derived vascular endothelial growth factor A (VEGF-A) protects liver cells from injury (possibly also including viral infection) by acting on LSEC, they reasoned that VEGF-A may interfere with secretion of the elusive proviral factor from LSECs. Therefore, they added VEGF-A to LSECs and indeed observed less proviral activity of LSEC-conditioned medium in a VEGF-A dose-dependent fashion. Suspecting that they were on the right track, they conducted a microarray analysis of messenger RNAs (mRNAs) down-regulated by VEGF-A treatment in primary LSECs. Filtering these primary hits for transcripts that encode secreted proteins, four possible candidates remained. Of those, only BMP4 matched the molecular-weight range of the proviral factor, which was established by filtration studies. Subsequent experiments clearly established that VEGF-A repressed BMP4 mRNA and protein expression in LSECs. Moreover, recombinant BMP4 increased HCV infection in Huh-7.5 and primary human hepatocytes. Collectively, these results highlight that LSEC-derived BMP4 promotes HCV replication in hepatocytes, and that secretion of BMP4 from LSEC is suppressed by VEGF-A. This conclusion was further confirmed by dissection of the VEGF-A-induced signaling cascade that regulates BMP4 expression and, in turn, HCV replication. Specifically, the researchers show that VEGF-A and also VEGF-E triggered phosphorylation of vascular endothelial growth factor receptor 2 (VEGFR-2), which signaled through p38 mitogen-activated protein kinase (p38 MAPK) to repress BMP4 mRNA expression. Finally, a specific inhibitor of the p38 MAPK, designated SB203580, not only prevented down-regulation of BMP4 mRNA by VEGF-A, but also rescued the proviral activity of LSEC-conditioned medium in the presence of VEGF-A. Thus, the circuit of evidence revealing a paracrine-signaling network between hepatocytes and LSECs, which involves VEGF-A, VEGFR-2, p38 MAPK and BMP4 in regulation of HCV replication in hepatocytes, was completed.
Notably, previous reports from the same researchers and other groups had revealed that HCV infection promotes VEGF-A expression, which causes hepatocyte depolarization and facilitates HCV cell entry.[6-8] However, these data were obtained with hepatocyte monocultures, and therefore a possible influence of BMP4 regulation by VEGF-A did not factor into these observations. Although both BMP4 and VEGF-A are proviral, as reported previously (and by Rowe et al. in this issue of Hepatology), and act on different steps of the HCV replication cycle (i.e., RNA replication and cell entry, respectively), the negative regulation of BMP4 expression by VEGF-A could profoundly influence the net outcome on HCV replication. Consequently, this regulatory circuit could play an important role in the liver in vivo.
To partially mimic the in vivo situation, the researchers monitored HCV replication in a direct coculture of primary LSECs and Huh-7.5 cells. In these cocultures, the addition of antibodies against VEGF-A increased HCV infection. Thus, at least in this nonpolarized in vitro system, inhibition of VEGF-A was proviral, arguing that the BMP4-linked suppressive effect of VEGF-A on HCV replication may dominate over its HCV entry-enhancing effect. Ultimately, expression of BMP4 and its regulators was monitored in healthy and inflamed livers. Remarkably, abundance of BMP4 was clearly increased in livers from patients with chronic hepatitis C or with alcoholic liver disease. Surprisingly, the total amount of VEGF-A and VEGFR-2 was increased as well. Although these findings, at first glance, seemed to challenge the model that VEGF-A-dependent signaling down-regulates BMP4, there was no significant difference in the abundance of phosphorylated VEGFR-2. However, the researchers additionally show elevated expression of endothelial cell marker proteins, such as vascular endothelial cadherin (VE-cadherin) and CD31, in the diseased livers. This latter finding may reflect elevated angiogenesis and vascularization in the inflamed liver. When determining the ratio of phosphorylated VEGFR-2 and VE-cadherin to estimate the activation of VEGF-A relative to the amount of endothelial cells, it became clear that indeed relative VEGFR-2 phosphorylation was reduced in the inflamed liver samples. Therefore, these ex vivo expression analyses support the model of the negative regulation of BMP4 expression by VEGF-A signaling.
The elegant study by Rowe et al. raises a number of interesting questions that should be addressed in the future. First, which factors positively regulate BMP4 expression and what is the role of elevated BMP4 in the inflamed liver? The researchers cite recent observations that implicate BMP4 in liver development and regeneration[10, 11] and speculate that HCV has adapted to exploit this endogenous function to facilitate its propagation. Second, it remains open how BMP4 promotes HCV RNA replication. Again, the researchers provide a possible link: BMP4 belongs to the transforming growth factor beta superfamily, which signals through mothers against decapentaplegic (Smad)-dependent and -independent pathways. Consistently, a genome-wide RNA interference screen revealed that knockdown of Smad5 significantly reduces HCV replication. Possibly, a microarray-based expression analysis of BMP4-treated hepatocytes may clarify which host factors are regulated by this growth factor to foster HCV RNA replication. Alternatively, BMP4-dependent signaling may directly influence the function of viral factors.
Last, future work should explore whether BMP4 levels influence the course of chronic HCV infection or the response of HCV to antiviral therapies. Because BMP4 seems to be elevated in the inflamed liver, it is possible that high abundance of BMP4 and its proviral effect contributes to the decreased response rates to antiviral therapies in patients with advanced liver disease. If so, BMP4 may be a biomarker with predictive value for therapy outcome. Of note, in a clinical study with hepatocellular carcinoma patients, VEGF-signaling blockage by sorafenib did not alter HCV RNA viral load, although when hepatocyte monocultures were treated with sorafenib, in vitro HCV replication was reduced. This strengthens the notion that interactions of parenchymal and nonparenchymal cells in the liver need to be taken into account to better predict the role of host pathways in HCV infection. Finally, coculture models of liver cells and LSECs similar to the one described by Rowe et al. may uncover novel, clinically relevant aspects of replication by other hepatotropic viruses.
Gisa Gerold, Ph.D.
Thomas Pietschmann, Ph.D.
Institute of Experimental Virology TWINCORE, Center for Experimental and Clinical Infection Research Hannover, Germany