HCV-Mediated Apoptosis of Hepatocytes in Culture and Viral Pathogenesis

Chronic Hepatitis C Virus (HCV) infection is associated with progressive liver injury and subsequent development of fibrosis and cirrhosis. The death of hepatocytes results in the release of cytokines that induce inflammatory and fibrotic responses. The mechanism of liver damage is still under investigation but both apoptosis and immune-mediated processes may play roles. By observing the changes in gene expression patterns in HCV-infected cells, both markers and the causes of HCV-associated liver injury may be elucidated. HCV genotype 1b virus from persistently infected VeroE6 cells induced a strong cytopathic effect when used to infect Huh7.5 hepatoma cells. To determine if this cytopathic effect was a result of apoptosis, ultrastructural changes were observed by electron microscopy and markers of programmed cell death were surveyed. Screening of a human PCR array demonstrated a gene expression profile that contained upregulated markers of apoptosis, including tumor necrosis factor, caspases and caspase activators, Fas, Bcl2-interacting killer (BIK) and tumor suppressor protein, p53, as a result of HCV genotype 1b infection. The genes identified in this study should provide new insights into understanding viral pathogenesis in liver cells and may possibly help to identify novel antiviral and antifibrotic targets.


Introduction
Hepatitis C Virus (HCV) is a small enveloped virus belonging to the family Flaviviridae, genus Hepacivirus. The single positive-sense RNA genome encodes a polyprotein that is cleaved by viral and cellular proteases into 10 different proteins [1][2][3][4]. Due to its high genetic variability, HCV has been classified in six genotypes that are differentiated based on nucleotide sequence in HCV infection and subsequent liver injury. These data reveal new insights into the molecular and immunologic mechanisms governing HCV-mediated apoptosis, induced by genotype 1b, and will be useful for the identification of novel antiviral or antifibrotic targets.

Preparation and quantitation of LB-piVe virus
LB-piVe were created by transfecting VeroE6 cells with a plasmid (pVA) encoding adenovirusassociated VA RNAI, as described previously [28]. Briefly, VeroE6 cells were transiently transfected with pVA, a plasmid encoding adenovirus VA RNA I , to inhibit interferon pathways. At 24 hr post-transfection, the cells were infected with patient plasma, identified as LB, a genotype 1b strain. Seven days post-infection (dpi), the cells were divided 1:6 and re-transfected with pVA. The passage and transfection process was repeated weekly for 20 weeks, without re-infection [28]. After 20 weeks, the cells were passaged weekly without transfection. Two years later the cells were expressing virus and were passaged weekly for a total of 4 yrs more. Persistently infected cells (LB-piVe) were used to grow virus for infectious stocks. Briefly, flasks (T75) containing LB-piVe cells and culture medium were frozen (at -80°C) and thawed 3 times. The cell debris was removed by centrifugation and subsequent filtering (0.45 μm), and filter-clarified culture supernatants were obtained and used as LB-piVe virus stock. LB-piVe viral titers were measured by titration on naïve Huh7.5 cells, as described [28]. The 50% tissue culture infectious dose (TCID 50 ) was calculated using the method of Reed and Muench [29]. UV-inactivated virus was obtained from the same virus stock in conditioned medium. After exposure to UV light (254-nm) for 60 seconds, using a Stratalinker 2400 (Agilent Technologies, Santa Clara, CA), the medium served as the "mock infection", a non-infectious control in the same conditioned medium. Serial dilutions of the UV-inactivated virus were tested for infectivity in naïve Huh7.5 cells as described [28]. Therefore, the conditioned media contained virus even though it was not infectious. All LB-piVe infections were carried out by infecting naïve Huh7.5 cells at MOI = 0.01 unless indicated.

Electron microscopy
LB-piVe infected cells were harvested four days post-infection, washed twice with PBS, and fixed by incubation for 48 hours in 4% paraformaldehyde and 1% glutaraldehyde in 0.1 M phosphate buffer pH 7.2. Next, cells were washed in PBS and post-fixed by incubation for 1 hour with 1% osmium tetroxide (Electron Microscopy Biosciences, Hatfield, PA, USA) and dehydrated in a graded series of ethanol solutions. Cell pellets were embedded in Epon resin (Sigma-Aldrich Corp., St. Louis, MO, USA) which was allowed to polymerize for 48 hours at 60°C. Ultrathin sections were cut, stained with 5% uranyl acetate and 5% lead citrate, and deposited on electron microscopy grids coated with collodion membrane, for examination under a Jeol 1230 transmission electron microscope [(TEM); Tokyo, Japan] connected to a Gatan digital camera driven by Digital Micrograph software (Gatan, Pleasanton, CA, USA).

Apoptosis induction
Apoptotic Huh 7.5 cells were produced by treatment with 100 nM Staurosporine (R&D Systems Inc., Minneapolis, MN, USA) or 3μM Gambogic Acid [30;31] for 16 hours at 37°C. For the immunoblotting experiments, apoptosis was induced with 1μM gambogic acid or 200 ng/ ml of actinomycin D for 24 hr.

Flow cytometry analyses
All flow cytometry studies were performed on a FACS Calibur cytometer (BD Biosciences, San Diego, CA, USA) with data analysis conducted using FlowJo 7.6.3 software (Tree Star, Inc., Ashland, OR, USA).

Phycoerythrin (PE) Annexin V staining
To determine expression levels of phosphatidylserine at the cell surface, cells were stained with PE Annexin V in a buffer containing 7-Amino-Actinomycin D (7-AAD) following manufacturer's protocol (PE Annexin V Apoptosis Detection Kit I, BD Biosciences, San Jose, CA, USA). Cells were harvested and washed twice with ice cold PBS. 10 5 cells were incubated with 5 μl of PE Annexin V and 5μl of 7-AAD for 15 minutes at room temperature.
concentration and purity of the extracted RNA was determined using a NanoDrop 1000 Spectrophotometer (Thermo Fisher Scientific Inc., Waltham, MA). One μg of purified RNA was used per PCR Array. Genomic DNA elimination was performed by incubating samples for 5 min at 42°C according to manufacturer's recommendations (Qiagen, Valencia, CA). Next, samples were placed on ice for >1 minute. The Reverse transcription mix was added and the reaction proceeded for 15 min at 42°C. The reaction was stopped by incubating at 95°C for 5 min. After adding RNase-free water to each reaction, samples were mixed with the RT 2 SYBR Green master mix and aliquoted into the PCR Array plates (Human Apoptosis RT2 Profiler PCR Array, Qiagen, Valencia, CA, USA). Plates were sealed with optical adhesive films and placed in the real-time cycler (Applied Biosystems 7700, Applied Biosystems Foster City, CA). Cycling conditions were as follows: 10 minutes at 95°C and then 40 cycles consisting of 15 seconds at 95°C and 1 minute at 60°C. The CT values were used to analyze the data. Relative expression of genes was determined using data from the real-time cycler (CT values) and the ΔΔCT method (RT 2 Profiler PCR Array Handbook, Sabiociences, 11/ 2011). Fold changes in gene expression were calculated by comparing gene expression in LB-piVe infected cells to that in UV-inactivated LB-piVe infected cells using the Web-Based PCR Array Data Analysis from Sabiociences (Valencia, CA, USA).

Pathway analysis
Pathway analysis was performed using the Pathway Studio 9 Software (Ariadne Genomics, Rockville, MD). Protein interaction maps were generated and colored according to the level of gene expression (red: up-regulated genes; blue: down-regulated genes; grey: no change).

Accession numbers
Sequences can be accessed from GenBank through the NCBI website: LB-piVe (FJ976045, FJ976046 and FJ976047).

Ultrastructural changes in LB-piVe infected Huh7.5 cells show markers of apoptosis
We previously reported that an HCV genotype 1b virus present in the plasma of an infected patient established a persistent infection in VeroE6 cells, demonstrating a mild cytopathic effect (CPE). Persistently infected VeroE6 cells, termed LB-piVe cells, were passaged weekly for two years and maintained in culture for more than 6 years [28]. The virus rescued from LB-piVe cells induced strong CPE when incubated with the HCV-permissive Huh7.5 human hepatoma cell line, while UV-inactivated LB-piVe supernatant did not [28]. To examine the effect of HCV infection at the sub-cellular level, Huh7.5 cells were infected (MOI = 0.01) with LB-piVe or with a mock infection, UV-inactivated LB-piVe (UV-LB-piVe), for 4 days. We verified the presence of viral RNA by RT-PCR (data not shown) and intracellular expression of HCV NS5A by flow cytometry (Fig 1A). We found that approximately 5.50% of the Huh7.5-LB-piVe infected cell population showed an increase in fluorescence intensity ( Fig 1A, center Fig 1C). Interestingly, signs of apoptosis, such as membrane blebbing, budding apoptotic bodies ( Fig 1D, 1E and 1F, green arrows) and crescent-shaped chromatin aggregation ( Fig 1G, green arrow) could also be visualized. No signs of apoptosis were visualized in UV-LB-piVe -infected cells ( Fig 1H).

Membrane expression of phosphatidylserine in LB-piVe infected cells is revealed by positive staining for Annexin V
Changes in the plasma membrane are one of the first characteristics of the apoptotic process detected in living cells. Upon induction of apoptosis, rapid alterations in the organization of phospholipids in most cell types occur, leading to exposure of phosphatidylserine (PS) on the cell surface. PS can be detected through high affinity binding of fluorochrome-labeled Annexin V. The combination of phycoerythrin (PE) Annexin V and 7-Amino-Actinomycin D (7-AAD), which has a high DNA binding constant and is efficiently excluded by intact cells, enables the differentiation between apoptotic and necrotic cells. In this context, viable cells are not stained with Annexin V or 7-AAD; apoptotic cells are positive for Annexin V and negative for 7-AAD, and necrotic cells stain positive for both. To further explore the process of LB-piVe induced apoptosis, virus-infected cells were harvested, and then incubated with PE Annexin V in a buffer containing 7-AAD (Fig 2).
Flow cytometry analysis revealed that UV-LB-piVe-treated cells were primarily negative for PE Annexin V and 7-AAD staining, demonstrating that 81.4% of the cells were viable (I) as indicated in Fig 2A (left panel), and Fig 2B. A minor population of cells were found to be PE Annexin V and 7-AAD positive indicating that they were already dead (IV; necrotic, 4.7%) or early/late apoptotic (II or III; 13%). When cells were treated with the apoptosis-inducer, Staurosporine (Fig 2A, center panel and Fig 2B), only 64.7% were live cells with 27.3% undergoing apoptosis. Interestingly, 49% of the LB-piVe-infected Huh7.5 cells remained alive at 5 dpi (Fig  2A, right panel and Fig 2B), with a large percentage undergoing apoptosis (37.7%). This suggests that virus infection, replication and/or spread are responsible for apoptosis in hepatocytes because the UV-LB-piVe did not induce apoptosis (13%) to the same extent as the infectious virus. The infectious virus induced cell death by necrosis (9.2%; Fig 2A, right panel and Fig 2B) two-fold over the UV-LB-piVe-infected cells (4.7% ; Fig 2A, left panel and Fig 2B). This may represent cells that have lysed due to late phase apoptosis or release of cytokines into the supernatant. However, induction of TNF is probably not responsible for necrosis as treatment of LB-piVe-infected Huh7.5 cells with TNF did not induce apoptosis or necrosis (data not shown).

Gene profiling analysis of LB-piVe infected cells using a human apoptosis PCR array
Chronic HCV infection is associated with progressive liver injury and subsequent development of fibrosis and cirrhosis. The mechanism of liver damage induced by HCV is still under investigation and may comprise apoptosis and immune-mediated processes involving necrosis or autophagy [32][33][34]. To characterize the mechanism of apoptosis induced by LB-piVe in Huh7.5 cells, we used a human apoptosis PCR array to screen differentially expressed genes implicated in the apoptotic signaling pathways. The array consisted of 84 key genes involved in programmed cell death, 5 housekeeping genes, a genomic DNA control, reverse-transcription controls, and positive PCR controls. Genes that showed a 2-fold or higher change in expression in two independent experiments were considered to be significantly upregulated by viable LB-piVe virus. No genes were found to be significantly down-regulated.
A list of the genes analyzed in this study, grouped according to their biological functions, together with the fold change in expression, is provided in Tables 1 and 2. LB-piVe-infected cells demonstrated significant changes in the expression of several important host genes that are involved in both the extrinsic apoptosis pathway activated through the death receptors, and the intrinsic apoptosis pathway that is dependent on mitochondrial dysfunction [25]. Among the genes with the greatest induction were pro-apoptotic proteins such as BCL2-interacting killer (BIK; 44.5 fold), BCL2-like 11 (BCL2L11; 14.9 fold), Death-associated protein kinase 1 (DAPK1; 8.6 fold), Tumor protein p73 (TP73; 7.3 fold), Lymphotoxin alpha (LTA; 6.3 fold), Tumor necrosis factor (TNF; 6.25 fold), BCL2-associated agonist of cell death (BAD; 4.1 fold), TNFRSF1A-associated via death domain (TRADD; 3.8 fold), and TNF receptor superfamily member 6 (Fas, CD95) ( Table 1).

Functional analysis of regulated genes in LB-piVe infected cells
To confirm that the differentially expressed genes were also up-regulated at the protein level, we randomly selected some of those genes to explore their expression using flow cytometry analysis. Since the PCR array showed that Fas receptor (CD95) was up-regulated twofold in virus-infected cells (Table 1), we stained both LB-piVe-infected and UV-LB-piVe-treated Huh7.5 cells with a PE-mouse anti-human Fas antibody. We observed that 29.1% of the virusinfected cells expressed Fas as shown in Fig 3A (right panel), and Fig 3B. The level of Fas expression was comparable to our positive control for apoptosis, the Staurosporine-treated cells (29.7%; Fig 3A, center panel and Fig 3B). In contrast, only 15.9% of the UV-LB-piVetreated cells stained positive for Fas (Fig 3, left panel and Fig 3B), which is consistent with the PCR array ( Table 1).
The Fas pathway can potentially induce apoptosis by activation of caspase-8 and caspase-3. Because Fas expression was up-regulated in LB-piVe-infected cells (Table 1 and Fig 3), it was of interest to determine if caspase proteins were similarly activated in these cells. Total intracellular caspase activation was examined using a cell-permeable, FITC-conjugated pan-caspase inhibitor (ApoStat, R@D Systems, Inc., Minneapolis, MN, USA). Huh7.5 cells were infected with either LB-piVe or treated with UV-LB-piVe, or Staurosporine. Cells were then harvested, washed and stained with ApoStat. It should be noted that ApoStat only binds to active caspases and the unbound ApoStat is washed away. When assayed by flow cytometry, 76% of the LB-piVe-infected cells expressed active caspases (Fig 4A, right panel, and Fig 4B) compared to 51% of the UV-LB-piVe cells (Fig 4A, left panel). Active caspases were stained in 62% of the staurosporine-treated cells (Fig 4A, center panel and Fig 4B). Further characterization of caspase 3 activity was observed in HCV-infected cells, using an inhibitor of caspase 3 activity (Z-VAD-FMK) to show specificity. Cell staining with a rabbit anti-active caspase-3 antibody showed that caspase-3 activity was four times higher in LB-piVe-infected cells (29.2%; Fi5, left bottom panel) than in the UV-LB-piVe-treated cells (7.01%; Fig 5, left top panel). In this experiment, gambogic acid was used to induce apoptosis and activate caspase 3 for use as a positive control. Gambogic acid-treatment yielded 68.2% of cells expressing activated caspase-3 (Fig 5, left center panel). The addition of the broad caspase inhibitor Z-VAD-FMK caused a decrease in the amount of active caspase 3 in all cases (Fig 5,  right panels), as expected. Some residual caspase 3 activity remained in both Gambogic acidtreated cells (4%, Center, right) and LB-piVe-treated cells (11.7%), suggesting that caspase 3 was activated rapidly after infection and before Z-VAD-FMK treatment.
Active caspases cleave human poly (ADP-ribose) polymerase (PARP1) to generate an 89-KDa fragment [35;36]. Caspase 3 was present in Huh7.5 cells, but the activated form of caspase 3 was detected only in cells that were infected with live virus. Therefore, we investigated whether cleaved PARP1 also was present in LB-piVe-infected cells. The 89-KDa fragment was detected by immunoblot in infected cells both at 48 and 72 hours post-infection, but not in naïve Huh7.5 or UV-LB-piVe-treated cells (Fig 6). Fas and TNF lead to caspase activation in the extrinsic pathway. To expand our analysis of the intrinsic apoptosis pathway [25], we studied expression of two additional proteins in HCVinfected cells: Bcl-2-interacting killer (BIK, 20 KDa) and Bcl-xL (30 KDa). BIK was only detected in cell lysates of LB-piVe-infected cells but not in naïve Huh7.5 or UV-LB-piVetreated cells (Fig 6). Down-regulation of the anti-apoptotic Bcl-xL, an isoform of BSL2, was observed in LB-piVe-infected cells, consistent with the PCR array observations (Table 1).

Protein interactions and pathway analysis
To further understand the interactions among the multiple genes regulated during LB-piVe infection of Huh7.5 cells, we analyzed the protein network generated by the PCR array datasets using the Pathway Studio software. Potential biological interactions, cross-talk, and modifications among these differentially expressed proteins are shown in Fig 7. The caspase cascade seems to play an important role in the process, by linking the death receptor (extrinsic) and mitochondrial (intrinsic)-mediated pathways [37][38][39]. The observation of TNF and Fas upregulation, two key players in the initiation of apoptosis, together with up-regulation of other genes including, BAD and BIK, located at the mitochondrial membrane, clearly shows that LB-piVe infection triggers an extremely complex process.
The pathway analysis taken with protein expression studies (Figs 2-6) demonstrate that LB-piVe infection is associated with cell apoptosis that may be triggered through both the extrinsic and intrinsic pathways. Furthermore, these data suggest that LB-piVe infection, which results in cytopathic effect, is marked by the induction of apoptosis in Huh7.5 cells. Our data suggests that apoptosis in HCV-infected cells occurs by activation of the extrinsic pathway via Fasligand binding and subsequent activation of caspase-3 through a caspase cascade that leads to the cleavage of PARP1, which is known to cause programmed cell death [35;36].

Discussion
In the last decade, a shift toward therapies that inhibit viral or host proteins essential for HCV replication has become a major focus in drug development research [40;41]. A better understanding of the HCV life cycle and the host processes that trigger HCV-associated pathology is imperative to ensure the development of successful antiviral therapies. Host immune mechanisms and direct effects of HCV proteins have been implicated in hepatitis C viral pathogenesis and apoptosis has been linked to the progression of HCV-associated liver injury [23;26;42-44]. However, the molecular processes and specific cellular factors that determine viral persistence and pathogenesis remain to be elucidated.
Liver fibrosis due to viral infection is one of the leading reasons for liver dysfunction, including liver failure associated with HCV infection. Assessment of liver damage is performed by liver biopsy; the most common method for detection and evaluation of fibrosis. Liver biopsy is invasive, painful and associated with serious complications. Apoptosis of hepatic stellate cells is associated with the regression of fibrosis. Therefore, it is important to delineate the mechanisms involved in cell death and evaluate biomarkers that lead to the progression of viralinduced fibrosis.
In this study, we investigated the changes in human hepatoma gene expression patterns induced by LB-piVe [28], an HCV plasma isolate that adapted to grow in VeroE6 cells, wherein it induced a mild cytopathic effect. To evaluate the strong cytopathic effect of LB-piVe and correlate its eruption with apoptosis, we studied the alterations in cell structures induced by the virus. Huh7.5 cells were infected with LB-piVe virus. We observed that approximately 5.50% of the Huh7.5-LB-piVe infected cells expressed NS5A (Fig 1A, center panel). Interestingly, many LB-piVe-infected cells (Fig 1B, left and center panels), showed signs of gross cell death when visualized by light microscopy. We suspect that we could only detect 5.50% of NS5A-positive cells because cells undergoing apoptosis could not be stained due to massive cell destruction.
Ultrastructural analysis of Huh7.5-LB-piVe infected cells by transmission electron microscopy (TEM) elaborated the presence of a membranous web composed of multi-membranes vesicles (MMV) and large lipid droplets (Fig 1C, 1D and 1E). Our findings are consistent with previous reports which have shown that HCV induces changes in the host-cell membranes, resulting in a membranous web after activation of the autophagy pathway [45][46][47]. In addition, HCV infection of Huh7 cells induces autophagocytic vacuoles [48;49] and the autophagy machinery is required for HCV replication [50]. Accumulation of autophagosomes is triggered via activation of the unfolded protein response pathway and concomitant up-regulation of the autophagosome marker, LC3 [48]. It has been proposed by Chu et al. that autophagosomes may provide a hidden reservoir for HCV replication; protecting the virus from host innate immune responses [51]. While the autophagy pathways are out of the scope of this study, based on previous studies, we presume that both autophagic and apoptotic pathways are activated and involved in cross-talk, induced by HCV.
However, membrane blebbing, budding apoptotic bodies (Fig 1D, 1E and 1F, green arrows), crescent-shaped chromatin aggregation (Fig 1G, green arrow) could also be observed. These ultrastructural observations, as well as LB-piVe induced changes in the plasma membrane of infected cells revealed that LB-piVe-infected cells were undergoing cell death, suggesting a direct link between virus infection and apoptosis (Fig 2A). Gene expression analysis of infected cells is a powerful tool that can be used to characterize host-cell antiviral responses, including those that may contribute to liver pathology. Previously, the impact of HCV replication on host cell functions has been studied using microarray expression profiling in cells that were transfected with RNA encoding either individual HCV genes, HCV subgenomic or full-length replicons, and in cells infected with the HCV J6/JFH-1 chimera, JC1 [16;18-24]. These studies demonstrated that HCV regulates host genes involved in oxidative stress, apoptosis, lipid metabolism, immunity, proliferation, and intracellular transport [24]. A recent study using an apoptosis-specific PCR array [52] identified BCL2-interacting killer (BIK) mRNA expression as up-regulated, consistent with our results, reported here (Table 1). They showed that expression was also upregulated at the protein level and revealed its role in HCV replication through interaction with NS5B.
Gene expression profiles in liver biopsy tissue from patients with fibrosis and cirrhosis resulting from HCV genotype 3a infection identified significant changes in the expression of genes involved in cell signalling, kinase activity, protein metabolism, protein modulation, cell structure/cytoskeleton, and transcriptional regulation [18]. Most notably, up-regulation of caspase-9 and epithelial membrane protein 1 (EMP1) was associated with fibrosis and Bcl-2-related proline-rich protein (BCL2L12) and programmed cell death protein 1 precursor (PDCD1) was associated with cirrhosis. Here, we report results of the first expression profile screening of hepatocytes infected with HCV genotype 1b using a human apoptosis PCR array containing 84 key genes involved in programmed cell death. Both anti-apoptotic and pro- apoptotic genes; BCL2-interacting killer (BIK; 44.5 fold), Tumor necrosis factor (TNF; 6.25 fold), TNF receptor superfamily member 6 (Fas; 2 fold), BCL2-like 11 apoptosis facilitator (BCL2L11; 14.96 fold), caspase-3 (CASP3; 5.36 fold), caspase-6 (CASP6; 4.36 fold) and caspase-9 (CASP9; 3.67) were among the genes that were upregulated (Tables 1 and 2).
To verify the PCR array results at the protein level, we measured their expression in functional assays. Of special interest was the Fas pathway since it is known to trigger cell apoptosis by activating caspase-9 and then caspase-3 more downstream. Up-regulation of Fas and caspase-3 activities in LB-piVe-infected cells was confirmed by flow cytometry (Figs 3 and 5), consistent with previous findings [23 ;53].
After activation, caspases can cleave human poly (ADP-ribose) polymerase (PARP1) to generate an 89-kDa fragment [35;36]. Immunoblot analysis revealed the presence of cleaved PARP1 which has a role in DNA repair and maintenance of genome integrity. The presence of cleaved PARP1 is indicative of DNA damage, a key feature of apoptosis. Significant downregulation of the anti-apoptotic protein Bcl-xL was observed in LB-piVe infected but not in UV-LB-piVe treated cells (Fig 6), suggesting a role for HCV in regulating programed cell death, and consequently determining the fate of infected hepatocytes. The protein network generated during this study (using the PCR array data) clearly shows the association of the two key players in the initiation of apoptosis; TNF and Fas with the caspase cascade and with genes such as BAD and BIK, located at the mitochondrial membrane (Fig 7).
Collectively, the PCR array, the gene expression and functional analyses described here, strongly indicate that LB-piVe infection is associated with programmed cell death. More importantly, the data suggest that both the extrinsic and intrinsic apoptotic pathways are involved in HCV-1b-mediated apoptosis. Further studies will be needed to define all the factors involved in this complex process and to better understand the mechanism of apoptosis in hepatocytes and their role in the progression of fibrosis. The genes identified in this study will provide the basis for new insights into HCV pathogenesis and may possibly help to identify novel therapeutic targets.