Fibrogenic Gene Expression in Hepatic Stellate Cells Induced by HCV and HIV Replication in a Three Cell Co-Culture Model System

Retrospective studies indicate that co-infection of hepatitis C virus (HCV) and human immunodeficiency virus (HIV) accelerates hepatic fibrosis progression. We have developed a co-culture system (MLH) comprising primary macrophages, hepatic stellate cells (HSC, LX-2), and hepatocytes (Huh-7), permissive for active replication of HCV and HIV, and assessed the effect of these viral infections on the phenotypic changes and fibrogenic gene expression in LX-2 cells. We detected distinct morphological changes in LX-2 cells within 24 hr post-infection with HCV, HIV or HCV/HIV in MLH co-cultures, with migration enhancement phenotypes. Human fibrosis microarrays conducted using LX-2 cell RNA derived from MLH co-culture conditions, with or without HCV and HIV infection, revealed novel insights regarding the roles of these viral infections on fibrogenic gene expression in LX-2 cells. We found that HIV mono-infection in MLH co-culture had no impact on fibrogenic gene expression in LX-2 cells. HCV infection of MLH co-culture resulted in upregulation (>1.9x) of five fibrogenic genes including CCL2, IL1A, IL1B, IL13RA2 and MMP1. These genes were upregulated by HCV/HIV co-infection but in a greater magnitude. Conclusion: Our results indicate that HIV-infected macrophages accelerate hepatic fibrosis during HCV/HIV co-infection by amplifying the expression of HCV-dependent fibrogenic genes in HSC.


Results
Establishment of a novel, co-culture model system consisting of three cell types involved in hepatic fibrosis, and supporting HCV and HIV co-infection. Currently, an in vitro model system that represents the hepatic microenvironment permitting active HCV/HIV co-infection is not available. In an effort to determine the role of these viral replications on hepatic fibrosis progression, we have developed a three-cell co-culture system consisting of HCV-infected hepatocytes (Huh-7, human hepatocellular carcinoma derived cell line widely used in HCV research field for its high permissiveness to HCV infection 22 ), HIV-infected primary macrophages (Mϕ), and hepatic stellate cells [LX-2, an immortalized line of human primary HSC 23 ] as schematically shown in Fig. 1A. In brief, primary human monocyte-derived Mϕ were infected with HIV 24 and then co-culture was established by addition of Huh-7 cells, with or without HCV infection, as well as LX-2 cells. These cells (Mϕ, LX-2 and Huh-7 or MLH co-culture) were maintained in 2% human serum in EMEM (Eagle's Minimum Essential Medium) up to 9 days, since longer duration of cultures caused cell death. We determined the survival of all three cell types during 9 day co-culture period by performing fluorescence-activated cell sorting (FACS) analysis (Fig. 1B,C). To facilitate detection of LX-2 cells, these cells were labeled with the Carboxyfluorescein N-hydroxysuccinimidyl ester (CFSE, fluorescent cell staining dye) [(LX-2(CFSE)]. We first verified the specific detection of LX2(CFSE) and CD68-immunostained Mϕ by using FACS detectors FL1 and FL4, respectively, using each of individual cell types (Fig. 1B). Then we detected the LX-2(CFSE) and CD68-immunostained Mϕ as well as non-fluorescent Huh-7 cells on day 9 of co-culture by FACS analysis (Fig. 1C). These results indicate that all three cell types in MLH co-culture could survive up to 9 day of co-culture. Importantly, we detected the replication of HIV and HCV as evidenced by detection of HIV p24 and HCV core antigen for the duration of MLH co-culture (Fig. 1D,E).
Activation of HSC is one of the central mechanisms of hepatic fibrosis development. Therefore, in order to determine the HSC-specific phenotypes induced by HCV and/or HIV replication, we separated LX-2 cells from Mϕ and Huh-7 cells by adapting the MLH co-culture system in a transwell system as shown in Fig. 2A,B. Previous literature indicated that culturing the HSC on a plastic surface promoted their spontaneous activation, while culturing them on matrigel-coated surface allowed them to maintain their quiescence state 23,25 . Thus, we coated the LX-2 cell-culturing tissue culture plate surfaces with matrigel to maximize the detection of viral replication-dependent HSC activation phenotypes. Then we assessed the roles of HCV and or HIV infection on HSC activation phenotypes including their morphology changes, migration, proliferation and fibrogenic gene expressions 26,27 . In some cases, we used CFSE-labeled LX-2 cells to facilitate the detection of their morphology changes and proliferation.
HIV or HCV infection promoted morphological changes in LX-2 cells and significantly enhanced their migration within 24 hr of MLH co-culture. We determined the effect of HCV and/or HIV replication under transwell MLH co-culture condition on LX-2 cell morphology by using a fluorescence microscope as schematically shown in Fig. 2A top panel, in which LX-2 cells were placed at the matrigel-coated bottom well while Mϕ and Huh-7 cells were placed in the transwell insert. Results showed that mono-or co-infection of HCV and HIV in MLH co-culture induced distinct morphological change in LX-2 cells resulting in "stellate morphology" with elongated cytoplasmic processes, on day 1 of MLH-co-culture, compared to those lacking virus exposure, which showed flat morphology ( Fig. 2A, bottom panel). The stellate morphology of LX-2 cells likely indicate their quiescent state as shown in previous report 28 , since we detected reduced proliferation of LX-2 cells showing stellate morphology (see below, Fig. 3C). Ikeda et al. showed that quiescent HSC migration was associated with their stellate morphology 29 . Therefore, we determined the effect of HIV and HCV replication on LX-2 cell migration efficiency. LX-2 cell migration was determined in a transwell configuration mimicking normal liver environment for HSC migration, in which the top of porous transwell membrane were coated with matrigel and bottom with collagen I to mimic space of Disse and fibrillary matrix, respectively, as described by Hu et al. 30 (Fig. 2B, top panel). We placed LX-2 cells on top of matrigel coated transwell insert and counted the number of LX-2 cells migrated to the bottom of transwell within a day of their co-culture with Mϕ and Huh-7 cells placed at the bottom well, in the presence and absence of HCV and/or HIV infection (Fig. 2B,  HCV/HIV co-infection of MLH culture for 7 days did not affect overall LX-2 cell proliferation. We have measured the rate of LX2(CFSE) cell proliferation, by determining the fluorescence intensity of CFSE levels, to be ~three divisions during 7-day culture period, regardless of whether LX-2 cells were cultured singly or under MLH co-culture condition in a transwell setting (Fig. 3A). Next, we determined the effect of HIV or HCV infection, or HCV/HIV co-infection, on LX2(CFSE) cell proliferation and morphology change under MLH co-culture/transwell setting following 7-day co-culture. As shown in Fig. 3B, overall LX-2 cell proliferation was not affected by HCV and HIV mono-or co-infection of MLH

HIV/HCV co-infection augmented the HCV infection-dependent upregulation of selected fibrogenic genes in LX-2 cells under MLH co-culture conditions.
To assess the effect of HCV and HIV mono-or co-infection on fibrogenic gene expression in LX-2 cells, we isolated total RNA from LX-2 cells cultured in transwell MLH co-culture conditions in the presence and absence of HCV and/or HIV infection, and then subjected these RNAs to human fibrosis microarray analysis (Fig. 4A). Initial assessment indicated that the effects of viral infection in MLH co-culture on fibrogenic gene expressions in LX-2 cells became evident by day 7 post-co-culture ( Supplementary Fig. S1A). Therefore, we performed all subsequent microarray experiments by using LX-2 cell RNAs collected on day 7 to 8 of MLH co-cultures (generated by using 7 different primary Mϕ) in the presence or absence of HCV and or HIV infection. The results of these experiments were following: After HCV mono-infection, five genes, including C-C motif chemokine ligand 2 (CCL2), interleukin 13 receptor subunit alpha 2 (IL13RA2), interleukin 1α (IL1A), interleukin 1β (IL1B) and matrix metalloprotease 1 (MMP1) were upregulated, on average, more than 1.9 fold in LX-2 cells during MLH co-culture. (Fig. 4B, red fill). However, their upregulations were not statistically significant as shown in volcano plot (Fig. 4D, red dots). Interestingly, these same genes were also upregulated by HCV/HIV co-infection but in greater magnitude (Fig. 4B, green fill) and statistically significant manner (Fig. 4E, green dots). On the contrary, HIV mono-infection showed no impact on fibrogenic gene expression in LX-2 cells under equivalent conditions (Fig. 4B, black fill. See also Fig. 4C). Clustergram shown in Supplementary Fig. S1B summarizes the relatedness between HCV infection-and HCV/ HIV co-infection-mediated changes in fibrogenic gene expression patterns in LX-2 cells from MLH co-culture, compared to no infection, as well as lack of obvious changes by HIV infection. In general, primary Mϕ from different donors had minimal to moderate variation in most fibrogenic gene expression levels in LX-2 cells during MLH co-cultures regardless of viral infection (Fig. 5A, see also Supplementary Fig. S2). However, profound variations in the expression levels of above-mentioned, five selective genes in LX-2 cells were detected following infection of HCV or HCV/HIV to different MLH co-cultures differing only by their primary Mϕ content (Fig. 5A,B). These results suggest that HCV infection is the main driver of selective fibrogenic gene upregulation in LX-2 cells under MLH co-culture condition and HIV-co-infection augmented this HCV-mediated gene upregulation. Importantly, these results suggest that variation in primary Mϕ, which was the only variable within MLH co-culture constituents, determined the extent of selective fibrogenic gene upregulation in LX-2 cells under MLH co-culture infected by HCV or HCV/HIV.  6A) and FACS analysis (Fig. 6B), respectively. In addition, immunofluorescence analyses indicate that only small proportion of LX-2 cells expressed detectable level of αSMA protein under our MLH co-culture condition on day 7 (Fig. 6C).

Discussion
In this study, we have developed an in vitro co-culture system (MLH) consisting of three major cell types in the liver involved in hepatic fibrosis development, including primary Mϕ, HSC (LX-1) and hepatocytes (Huh-7), permissive for active replication of HCV and HIV. We have determined the effects of HCV and HIV on early phenotypic changes in HSC to understand the viral mechanisms triggering HSC activation and resulting in hepatic fibrosis. Our results indicated that HCV and/or HIV replication in MLH co-cultures trigger morphological changes in HSC and enhanced the invasive potential of HSC (Fig. 2). Importantly, HIV mono infection in MLH co-culture had no impact on fibrogenic gene transcriptions in HSC, while co-infection of HCV/HIV significantly augmented the expression of HCV-responsive fibrogenic genes in HSC, including CCL2, IL1A, IL1B, IL13RA2 and MMP1 (Figs 4 and 5), providing mechanistic insight into enhanced fibrogenesis in HCV/HIV co-infected patient population compared to those infected with HCV alone. Use of different primary Mϕ in our MLH co-culture system provided us with a perspective regarding the role of different Mϕ for upregulation of fibrogenic genes in HSC upon HCV and/or HIV infection. Interestingly, expression of the majority of fibrogenic genes in HSC showed limited variations between MLH cultures generated by using Mϕ from different donors, regardless of HCV and/or HIV infection under our experimental condition (Figs 5A and S2). However, in the case of five selective genes upregulated 1.9 fold or more on average by HCV or HCV/HIV co-infection, the magnitude of these gene inductions in HSC varied substantially depending on different primary Mϕ used (Fig. 5A,B). It is likely that differences in individual phenotypes of Mϕ contributed to this variation, since previous studies suggested that HCV-dependent induction of M2-polarized Mϕ promoted HSC activation and hepatic fibrosis 25,31,32 . Interestingly, we detected no correlation between HIV replication levels (represented by p24 levels in Fig. 1D) and altered fibrogenic gene expression folds in LX-2 cells following HIV infection of different MLH co-culture (Fig. 5A). These results suggest that divergent primary Mϕ phenotypes following HIV infection, not the HIV replication efficiency per se, determined the magnitude of HCV/HIV co-infection-dependent fibrogenic gene expression in LX-2 cells under MLH co-culture condition. Future studies are needed to determine the Mϕ phenotypes that will trigger high level of fibrogenic gene expression in HSC following HCV or HCV/HIV co-infection. Such information could be useful to identify patients at higher risk of developing hepatic fibrosis following these viral infections.
Theoretically, HIV-infected Mϕ could have directly modulated fibrogenic gene expression in HSC regardless of HCV infection or induced a different set of genes from those induced by HCV in the coinfection condition and thereby accelerate hepatic fibrosis in HCV/HIV co-infection cases. However, by performing microarray analysis of 84 genes involved in human fibrosis using HSC specific RNA samples obtained from MLH co-culture system, our study eliminated either of these possibilities. Instead, our data showed that HCV/HIV co-infection augmented fibrogenic gene expression in HSC compared to HCV mono infection, consistent with previous report 18 . However, different from this previous study 18 , we did not detect the effect of HIV on fibrogenic gene expression in HSC. Differences in experimental strategies such as HSC/hepatocytes exposed to HIV in the previous study versus MLH that include primary Mϕ infected with HIV in our study, may have caused this discrepancy. Under our experimental conditions, well-known fibrogenic genes, including αSMA (ACTA2), isoforms of collagenase (COL1A2 and COL3A1), transforming growth factor-beta (TGF-β), tissue inhibitor of metalloproteinase (TIMP, except TIMP1), and matrix metalloproteinase (MMP, except MMP1) were not induced in LX-2 cells by HCV and HIV mono-and co-infection of MLH co-cultures (Figs 6 and S2). Despite this, we detected myofibroblast-like morphological changes in fraction of HSC from HCV infected or HCV/HIV co-infected MLH co-culture at 7 days post co-culture indicating that those HSC may have undertaken an activation process (Fig. 3C).
It is notable that substantially upregulated fibrogenic genes in HSC by HCV or HCV/HIV infection of MLH co-culture belong to inflammatory cytokines, including CCL2, IL1A, IL1B and IL13RA2. The CCL2, also called monocyte chemoattractant protein 1 (MCP1), is a proinflammatory cytokine secreted by HSC and Kupffer cells in the liver, and promotes hepatic fibrosis by stimulating the recruitment of monocytes to the injured liver 33,34 . Based on our data, we propose that significant upregulation of CCL2 in HSC under HCV/HIV co-infection condition likely contributed to enhanced fibrosis development, in addition to previously suggested immuno-pathological mechanism of CCL2 for accelerating HCV/HIV co-infection-mediated liver fibrosis 35 .
IL1A and IL1B constitute two forms of the proinflammatory cytokine interleukin-1. Previous studies suggested IL1A as an early responder of inflammatory response and IL1B as a late responder, recruiting neutrophils and Mϕ, respectively, to the site of injury 36 . IL1 receptor knockout mice were protected from hepatic fibrosis development indicating the critical roles of these two cytokines in this process 37 . Our data indicate that significant upregulation of IL1A and IL1B from quiescent HSC due to HCV/HIV co-infection might be an early inflammatory response contributing to enhanced HSC activation leading to hepatic fibrosis.
IL13RA2 was shown to be overexpressed in activated HSC, and blocking the IL13 receptor reduced hepatic fibrosis development caused by non-alcoholic steatohepatitis (NASH) 38 . Our data indicate that HCV infection could promote IL13RA2 expression in HSC, which could be further upregulated by HIV co-infection, supporting the role of this factor in accelerated hepatic fibrosis by HCV/HIV co-infection.
Matrix metalloproteinases (MMPs) degrade extracellular matrix and play critical roles in tissue repair and remodeling. Different MMPs were implicated to function differently based on their substrate specificity, and environment in which they are expressed, not only in ECM remodeling but also in immune responses 39 . Over expression of MMP-1, MMP-8 and MMP-13 was shown to reduce the number of activated HSC and attenuate the hepatic fibrosis when transiently over expressed in the liver [40][41][42][43] . These findings suggest the protective role of MMPs during liver injury. However, evidence for profibrotic roles of MMPs also exist 44 . Since HCV-dependent MMP-1 induction in HSC was augmented by HIV co-infection, we speculate that MMP1 induction in HSC could have a potential to play a profibrotic role.
HCV-infected hepatocytes and HIV-infected Mϕ were separated from HSC via transwell during our three cell co-culture condition. Therefore, soluble factors from HCV and HIV infected hepatocytes and Mϕ must have contributed to the induction of fibrogenic genes in HSC. One such candidate is TGF-β1, which plays a pivotal role during tissue fibrosis development 45 , since HCV replication in Huh-7 cells was shown to induce this cytokine [46][47][48] . However, the study by Schulze-Krebs indicates that TGF-β1accounted for only ~50% of profibrogenic activity derived from HCV replicating cells, suggesting the presence of additional mediators induced by HCV replication 48 . Interestingly, recent literature idicates the important role of TGF-β2 in hepatic stellate cells activation and liver fibrogenesis, potentially even more so than the role of TGF-β1 in this process [49][50][51] . Chida et al. showed that silencing of TGF-β2 in HCV-infected Huh-7 cells reduced fibrogenic phenotypes in the human hepatic stellate cell line TWNT4 in two cell co-culture study. Importantly, they showed that serum TGF-β2 levels in HCV-infected patients positively correlated with hepatic fibrosis stages F0-F2. Supporting this finding, a recent study by Abd el-Meguid et al. also demonstrated positive correlations between HCV infection-associated hepatic fibrosis and elevated TGF-β2 level in serum and peripheral leucocytes 51 . Our preliminary data showed the significantly higher TGF-β2 levels in supernatant of MLH co-culture following HCV/HIV-co-infection compared to no infection ( Supplementary Fig. S3). These results suggest that TGF-β2 may be one of the main drivers of HCV/HIV-co-infection mediated upregulation of fibrogenic genes in HSC. However, further studies are necessary to establish the exact role of TGF-β2 in HCV/HIV co-infection mediated acceleration of hepatic fibrosis development. Also effort should be directed to identify any additional, soluble profibrogenic factors responsible for activating HSC during HCV/HIV co-infection, since such biomarkers could serve as a diagnostic tool to detect hepatic fibrosis induction or a potential target of therapeutic interventions to inhibit fibrosis development following these viral infections.
One of the limitations of this study is that our in vitro system does not include other residential liver components, such as sinusoidal endothelial cells, which were shown to contribute to hepatic fibrosis development 52 . Additional limitations may be the small number of primary Mϕ that we used to generate MLH co-cultures, which limited the representation of Mϕ characteristics from free-range, out bred human population. Despite these limitations, the three-cell MLH co-culture system, for the first time, allowed us to determine the effects of active replication of HCV and HIV in hepatocytes and Mϕ, respectively, on fibrogenic gene expression in hepatic stellate cells associated with hepatic fibrosis development.
In conclusion, we showed that HCV infection in hepatocytes trigger key fibrogenic factors in HSC, including proinflammatory cytokines as well as factors involved in tissue remodeling leading to activation of HSC. HIV infection in Mϕ, while having no impact on fibrogenic gene expression in HSC, specifically augmented HCV-dependent fibrogenic factor induction in HSC. We believe that our data provided mechanistic insight into accelerated hepatic fibrosis by HCV/HIV co-infection, which is HIV-dependent amplification of HCV-mediated HSC activation.  Transwell invasion assay. LX2 were seeded in the hanging insert of transwell plate (0.8 µm pore size), which were pre-coated with collagen (100 µg/ml) and matrigel (50 µg/ml). Macrophages (+/−HIV) and Huh-7 cells (+/−HCV) were seeded in the lower compartment. After incubating the assembled transwell plates at 37 °C for 24 hrs, migrated LX2 cells were counted by staining cell nuclei with Hoechst. HCV RNA Electroporation. Genotype 1a H77S RNA was electroporated into Huh-7 cells as previously described 56 . HCV infection was determined by detecting HCV core using indirect immunofluorescence.

HIV-1 infection of macrophages.
Gene expression analysis by RT 2 PCR profiler Array. The RT 2 PCR Profiler (SA Biosciences, Qiagen) was used to examine the expression patterns of 84 genes involved in human fibrosis, according to the manufacturer's instructions by using RNA isolated from LX-2 following MLH co-cuture with or without HCV and HIV. The Real-time RT-PCR was performed in a Bio-Rad PCR machine (model CFX96). Gene expression fold difference was analyzed for those genes whose Ct value was less than 34 by using the web-based software RT 2 Profiler PCR Array Data Analysis (Qiagen) (see Supplementary data set file for detail).
Statistical Analysis. The one-way analysis of variance (ANOVA) was performed by using GraphPad Prism 6 software to determine the significance in differences between non-infected and HCV and/or HIV infected samples. Flow Cytometry. CFSE pre-labeled LX2 cells were stained with αSMA in PBS supplemented with 3% BSA and 0.1% Triton X-100 for 1 hr following fixation with 4% paraformaldehyde and subjected to fluorescence-activated cell sorter (FACS) analysis by using an Accuri ™ C6 Cytometer (BD Biosciences).

Data Availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.