The role of genomics and proteomics in degenerative diseases and liver fibrosis

Genetic diseases can be classified as chromosomal abnormalities (for example, trisomy 21), Mendelian disorders (single gene alterations with typical inheritance patterns, like autosomal dominant/recessive or X-linked), and complex diseases that are influenced by many genetic and environmental components. Degenerative diseases like liver fibrosis are complex illnesses [3]. The genetic contributions to these disorders are not attributable to a single gene alteration, but rather to a host of genetic susceptibilities defined by single nucleotide polymorphisms (SNPs) that predispose an individual to a disease. The susceptibility to an accumulation of environmental influences is either enhanced or reduced by genetic factors, thereby defining an individual's disease risk. Studies investigating these genetic traits are complicated, because there are many genes that influence the risk for complex diseases, yet the impact of each single genetic variant by itself is small. Therefore, large numbers of subjects are needed to provide sufficient statistical power to yield robust conclusions. Currently, there are almost 13 million SNPs catalogued in the NCBI human SNP database. Approaches to identify SNPs that are linked with a specific disease range from efforts to sequence specific disease-causing genes to genome scans requiring sequencing of large numbers of known SNPs that may or may not be associated with the disease.

Genomic and proteomic screening methods are often used to identify classes of genes that are differentially expressed in disease. These classes provide the investigator with potential pathways that could be involved in the regulation of this disease, thereby narrowing the focus of subsequent studies to uncover disease mechanisms and potential targets for therapy. By not limiting the study to pathways already associated with a disease, many new pathways not previously implicated are now under investigation [4].

This prospect is especially compelling in degenerative diseases like liver fibrosis, as these conditions often develop slowly, and therapies may not be necessary for all subclasses of patients [5]. Regardless of the etiology of liver fibrosis, some people progress rapidly towards cirrhosis, whereas others never develop fibrosis in the first place, or have slow progression of their fibrosis. It seems very unlikely that this phenomenon can be attributed solely to environmental influences. Identifying which patients are unlikely to ever progress to cirrhosis may prevent overtreatment of many patients.

Another use for these screening methods is to identify expression profiles, or patterns of expression of many genes, that in aggregate correspond with disease outcome or the response to therapy. By analyzing these different gene/proteome classes, investigators may ultimately predict an individual patient's response to therapy more accurately, or identify those who are at greatest risk of progression, thereby refining treatment decisions.

The need for non-invasive markers in the assessment of fibrosis progression

The need to assess hepatic fibrosis progression is becoming more important as the incidence of advanced liver disease continues to rise, especially due to viral hepatitis and fatty liver disease. Liver biopsies are currently the 'gold standard' for determining the presence and progression of liver fibrosis. They are performed percutaneously, via transjugular access, or during abdominal surgery. Liver biopsies have a low complication rate; however, significant hemorrhage may rarely occur, requiring hospitalization in approximately 2% of the patients; moreover, some patients have contraindications for liver biopsy [6–8]. Additionally, analysis of liver biopsy is often undermined by sampling error and inter-observer variation [ [6, 9–15]. Studying mRNA and protein disease profiles in liver biopsies is challenging, as tissue sample sizes are small, especially for protein analysis, and the biopsies contain a mix of many different cell types, including hepatocytes, hepatic stellate cells, Kupffer cells, endothelial cells, lymphocytes, red blood cells and bile duct epithelial cells. To study a single cell population from tissue sections requires laser-capture micro-dissection, which yields even less mRNA and protein than biopsy [15].

Close follow-up of liver disease by serial biopsy is not practical due to its invasive nature, fueling the need to find alternatives for liver biopsy. A less invasive and more accurate diagnostic tool would aid in diagnosis of fibrosis, follow-up of established therapies and evaluation of experimental therapies. Moreover, hepatocellular carcinoma (HCC) is one of the main complications of cirrhosis, and early detection markers that distinguish HCC from cirrhosis could significantly reduce morbidity and mortality. In the past ten years, genomic, transcriptomic and proteomic studies have provided candidate biomarkers for fibrosis assessment. Thus, in combination with standard laboratory serum analysis, these have led to novel fibrosis assessment tools using indirect and direct fibrosis markers, like FibroSpect (Prometheus Laboratories; San Diego, CA, US), FibroTest (BioPredictive;Paris, FR) and Fibroscan (Echosens; Paris, FR), ActiTest (BioPredictive;Paris, FR) and APRI (public domain) [16–20]. For a more exhaustive overview of currently used serum biomarkers, see a recent review by Smith et al. [21].

These fibrosis assessment tests have shown promising results, but a number of pitfalls are encountered, including limited availability, inter-laboratory variation, cost, insufficient resolution to differentiate between intermediate stages of fibrosis and false positive results due to other conditions, such as increased venous pressure or steatosis [22–29].

Because current methods for non-invasive assessment of liver fibrosis lack resolution, particularly in the intermediate fibrosis stages, more research is being directed towards genomics and proteomics in the search of biomarkers that provide a close association with fibrosis stage [30–35].

Genomics and transcriptomics

Genomics comprises the study of the genetic information (DNA, or RNA in certain viruses) of an organism. Transcriptomics aims to elucidate the transcripts of the genome or gene expression levels on the RNA level under varying conditions as changes in mRNA expression do not necessarily correspond with changes in protein expression, since alternative splicing, protein production, degradation and post-translational modifications influence protein stability and function [36, 37].

The cDNA microarray is the most frequently used high-throughput screening method for gene expression profiling. These 'lab-on-a-chip' assays contain specified DNA oligonucleotide probes in spots on a chip that hybridize with oligonucleotides from tissue or cells providing a quantitative assessment of the transcribed genes [42–44]. Thousands of genes can be analyzed simultaneously with cDNA microarray and variants on this technique include specialized arrays that can detect SNPs and alternative splice variants (exon junction arrays) [38–41].

Serial analysis of gene expression (SAGE) analyzes the transcriptome by cloning strings of short cDNA fragments into bacteria, sequencing the cDNA fragments and then counting the number of cDNA fragments. This technique can yield information on both gene expression and alternative splicing and can help identify previously unknown genes [42, 43].

Target genes that are revealed by the genomics/transcriptomics screens are generally validated by reverse transcription quantitative polymerase chain reaction (RT-qPCR), which is more quantitative and more reproducible than microarray, requires less mRNA and is more sensitive for genes that are expressed at low levels.

Proteomics

Proteomics is the study of the proteome in a cell compartment, tissue or organism comprising all proteins that are encoded for in the genome. Mass spectrometry (MS) is the key technique in proteomics. Many variants of mass spectrometers are currently in use, but they all rely on the same concept: to determine the accurate mass of a protein by measuring the mass-to-charge (m/z) ratio. In short, mass spectrometers consist of an ion source, a mass analyzer and a mass detector. After ionization, proteins and peptides travel through the mass analyzer, which evaluates their ratio of charge (z) versus mass (m). The mass detector counts the number of molecules per charge-to-mass ratio, and can provide the user with an output as a mass spectrum, with the m/z ratio on the x-axis and the molecule count per m/z on the y-axis [44].

MS is preceded by a separation step of proteins, either through 2D-polyacrylamide gel electrophoresis (2-DE/2D-PAGE), liquid chromatography (LC) or gel electrophoresis (DIGE) [45–47]. Once the protein spots are excised from the gel and digested, they can be processed through one mass analyzer (MS) to identify peptide mass, or two mass analyzers (MS/MS) to determine amino acid sequence.

MS requires ionization of proteins and peptides and two soft-ionizing techniques have been developed: matrix-assisted laser desorption/ionization (MALDI) and Electrospray ionization (ESI). Both techniques charge molecules that are subsequently analyzed by one or two mass analyzers (MS or versus tandem MS/MS). Time-of-flight (TOF) mass analyzers deduct peptide mass by measuring the time it takes for a charged peptide to travel through a vacuum tube in an electric field and can also be combined with one (Q-TOF) or two quadrupoles (Q-Q-TOF), which uses an oscillating electric field to selectively allow for molecules with a specific range of m/z ratios to proceed without collision [48].

When extremely high resolution is required, Fourier transform-ion cyclotron resonance (FT-ICR) MS or orbitrap can be used. The charged molecules are injected into a Penning ion trap where detectors measure the signal of molecules that pass. The m/z ratio determines the frequency with which the ion passes over the detector [49–51].

Surface-enhanced laser desorption/ionization (SELDI) analyzes protein mixtures that selectively bind to a biochip with the characteristics of choice. It needs only a small amount of crude sample (for example, serum or small liver biopsies), and is suitable for profiling multiple low-molecular-weight proteins. SELDI does not provide direct identification of the proteins. However, the peak profiles generated by SELDI can be useful by themselves in diagnostics, or in predicting prognosis and/or treatment response. Since manual labor is minimal, SELDI is by itself a suitable technique for high-throughput screening, especially for serum biomarkers; however, the cost is still too high for application on a large clinical scale [52–54]. For an excellent overview of the different types of mass spectrometers and their individual strengths and weaknesses, see a review by Domon and Aebersold [55].

A different type of high-throughput screening of proteins is the protein microarray. It functions much like a cDNA microarray, but probes are constructed out of proteins, antibodies or DNA constructs containing protein binding sequences that can capture proteins on a chip [56].

Technical pitfalls

Technical variations are largely responsible for the low reproducibility between studies by different research groups. Differences in patient characteristics can significantly alter genomic and proteomic profiles. Proteome studies are especially difficult because proteins cannot be amplified, unlike DNA and mRNA. Therefore, scientists have to rely on fractionation, enrichment/depletion and solubilization protocols prior to MS to prevent overload with the most abundant proteins present in the serum urine

In vivo and in vitro animal models of fibrosis

Not all human liver diseases can be reproduced accurately in animal models, and thus surrogate models are being developed to study mechanisms of chronic liver injury and fibrosis.

Rodent studies reviewed below have analyzed a variety of fibrosis progression and resolution models, and have evaluated liver tissue, cell isolates, serum and urine proteomes. Considerable additional effort will be required to systematically compare proteomes between different disease models, and to accomplish the translational step of comparing these proteomic changes to human liver diseases.

Chronic hepatitis C viral infection (HCV)

In patients with HCV, early knowledge of which patients have rapidly progressing fibrosis could be very beneficial in the decision-making process of whether to treat patients aggressively with antiviral therapy. A recent review by Walters and Katze explored the relationship among the gene expression profile of HCV patients, viral clearance and treatment response, shedding light on potential virus-host interactions, which may emerge as future therapeutic targets [71].

Novel biomarker candidates to predict hepatic fibrosis in hepatitis c identified by serum proteomics

Fibrosis progression has also been related to galectin-3-binding protein (G3BP). Cheung et al. found increased levels in patients with hepatitis C-related cirrhosis when compared with mild and moderate fibrosis in 76 serum and 20 tissue human samples [80].

Chronic hepatitis B viral infection (HBV)

Approximately 400 million patients worldwide are chronically infected with hepatitis B virus (CHB). Hepatitis B virus infection causes chronic hepatitis in about 10% of infected patients, and increases the risk of developing HCC about 100-fold versus in the uninfected population. An excellent review on the role of genomics, transcriptomics and proteomics in chronic HBV-associated liver disease, recently published by Sun et al., described determinants of susceptibility to persistent HBV infection, disease progression and HCC development; their review also describes the use of these technologies to understand the pathogenic mechanism of HBV and to identify biomarkers that could aid in early detection of HCC [94].

Recent high-throughput proteomic studies have identified protein patterns that correlate with fibrosis stage in HBV patients. Zhang et al. described a novel approach to evaluate the protein expression profile of plasma membrane and analyzed human liver samples from HBV patients in different fibrosis stages. Using 2-DE, IPGphor isoelectronic focusing system and Bio-Rad Protein II electrophoresis to the investigators identified positive correlation between fibrosis grade and annexin A2 levels [95]. Lu et al. analyzed plasma of 7 healthy controls and 27 HBV patients with different stages of fibrosis by 2-DIGE and identified the up-regulation of peroxiredoxin 2 as a potential biomarker of HBV related liver fibrosis [96]. Although not liver-specific, plasma gelsolin protein has been pointed out as a potential predictor of HBV progression. C Marrocco et al. found by 2-DE a repressed expression of the protein in plasma of eight cirrhotic HBV patients, compared to eight chronic HBV infected patients [97]. Mohamadkhani et al. performed a 2DE and mass spectrometry analysis in patients with either chronic hepatitis B or HBV-related cirrhosis and healthy controls. Their data suggest that progression of HBV-related liver injury is associated with a down-regulation of apolipoprotein A1 and an increase in myeloperoxidase levels [98]. However, these results still need to be validated, and their ability to distinguish intermediate fibrosis stages is not clear.

Poon et al. using SELDI ProteinChip arrays and the Significance Analysis of Microarray (SAM) algorithm generated a model able to predict fibrosis (Ishak score ≥ 3) and cirrhosis (Ishak score ≥ 5) with an accuracy of > 90% [99]. Liver histology (METAVIR score) of HBV-patients was compared with three sets of serum markers by Myers et al.: aminotransferases (ALT, AST), Fibrotest (α₂-macroglobulin, apolipoprotein A1, haptoglobin, gamma-glutamyl-transpeptidase (GGT) and total bilirubin) and Actitest (ALT). Aminotransferases and Actitest were good predictors of activity (stage A2-F3) and Fibrotest accurately predicted stage F2-F4 fibrosis, especially in the ranges of ≤ 0.20 or > 0.80. Limiting liver biopsy to patients with intermediate Fibrotest scores could prevent biopsies in about half of the patients, without affecting accuracy [100].

Wang et al. performed an in vitro study to investigate proteome differences between stably HBV transfected HepG2.2.15 versus wild type HepG2 cells by 2-DE followed by LC-ESI-MS/MS. Furthermore; they compared HepG2.2.15 cells with and without interferon-alpha (IFN-α) treatment. Proteins that were differentially expressed in HepG2.2.15 versus HepG2 cells could be classified as being involved in, for example, cytoskeletal matrix, heat shock stress, signal transduction and protease/proteasome components. IFN-α treatment induced expression of interferon-stimulated gene 15 (ISG15) and prohibitin, among others, providing potential pathways that could lead to tumor suppression and defense against viral infection. Although these results have not been validated in vivo, they nevertheless establish that proteomics and cell culture can be a powerful combination to uncover differences in the proteome, and can be useful for evaluating potential treatments [101].

Alcoholic liver disease (ALD)

Several genomic and proteomic studies have been published on patients with alcoholic liver disease. However, the majority of studies focus on either diagnostics or genes/proteins that differentiate alcoholic hepatitis from alcoholic steatosis, not in predicting progression towards fibrosis and cirrhosis. Nevertheless, alcoholic steatohepatitis is clearly a precursor for developing fibrosis. For example, 586 genes were differentially expressed between alcoholic hepatitis and alcoholic steatosis, and 211 genes were differentially expressed between alcoholic hepatitis and non-disease controls when Seth et al. compared liver tissue RNA from eight patients with alcoholic hepatitis, nine with alcoholic steatosis, and seven non-disease controls using microarray analysis. Induced genes included pathways controlling cell adhesion, immune response, oncogenesis, signal transduction and embryogenesis. The 111 down-regulated genes included pathways for protein synthesis, cell growth and maintenance, transcription, signal transduction and transport [102].

Non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH)

The prevalence of NAFLD and NASH is on the rise, which makes these increasingly important causes of liver fibrosis, especially in countries with a Western diet containing a high fat and/or carbohydrate content. For NAFLD and NASH there is also a lack of genomic/proteomic studies specifically examining fibrogenesis and fibrosis progression, much like in alcoholic liver disease. Most genomic/proteomic studies focus on distinguishing NASH from NAFLD, and not on progression to liver fibrosis and cirrhosis. The ability to distinguish NASH from NAFLD is very relevant, however, as NASH clearly predisposes to fibrosis. Therefore, we reviewed recent publications on this topic. However, studies directed at identifying proteins related to the speed of fibrosis progression in NASH could further select patients who could benefit from treatment, improve understanding of fibrosis pathogenesis in fatty liver, and identify targets for treatment of NASH-related fibrosis.

Autoimmune liver diseases

Primary biliary cirrhosis (PBC), as well as related autoimmune disorders, is more prevalent in certain families and thus seems to have a complex genetic component, with many factors probably adding to the risk of developing PBC. Some of these genes associated with PBC are major histocompatibility complex genes and common autoimmune genes [110]. Shackel et al. discovered that PBC liver tissue had increased expression of Th1 and Th2 type proteins as well as connective tissue growth factor and TGF-β3 when they analyzed their cDNA microarray from liver tissue from patients with PBC (n = 4), PSC (n = 4) and non-disease controls (n = 8). Moreover, Wnt and Notch signaling pathways were induced. PSC patient samples had increased expression of apoptosis associated molecules [111]. Serum samples from 44 PBC patients and 75 controls, analyzed using WCX magnetic beads and MALDI-TOF-MS, provided 69 potential protein biomarkers. Li et al. used these to construct a diagnostic model for PBC using the m/z peaks of protein biomarkers 3445, 4260, 8133, and 16290, which showed a sensitivity of 92.9% and a specificity of 82.4% [112].

Tahiri et al. noticed an induction of six known potential plasma membrane expressions: liver arginase, cytokeratins 8/18, heatshock proteins (HSP) 60/70/90, and valocin-containing protein (VCP) in autoimmune hepatitis when they investigated serum samples from 65 autoimmune hepatitis type 1 patients and 90 controls consisting of healthy blood donors (n = 40) and patients with systemic diseases (n = 20) or other liver diseases (n = 30). They proposed that these proteins could be targets for auto-antibodies in autoimmune hepatitis [113]. Using protein microarray, Song et al. analyzed serum samples from patients with autoimmune hepatitis (n = 44), PBC (n = 50), HCV (n = 41), HBV (n = 43), systemic lupus erythematosus (SLE) (n = 11), primary Sjögren syndrome (n = 11), rheumatoid arthritis (n = 2) and 50 healthy controls. Initially they found 11 differentially expressed genes between autoimmune hepatitis and other samples. They then produced an autoimmune hepatitis-specific protein chip, and found three new antigens, ribosomal protein S20 (RPS20), Alba-like, and dUTPase, that could be used as highly specific biomarkers for autoimmune hepatitis, and validated this with ELISA [114]. Bowlus et al. have recently analyzed by in situ MALDIMS (matrix-assisted laser desorption/ionization mass spectrometry) in PBC (n = 29), PSC (n = 11), AIH (n = 7) and healthy controls (n = 10) and found a promising pattern of protein expression in bile ducts, inflammatory infiltrates and hepatocytes that may represent a feasible way to identify novel targets in these diseases [115]. Song et al. created a protein microarray containing 5,011 recombinant human liver proteins and were able to identify three new antigens, RPS20, Alba-like and dUTPase, as highly autoimmune hepatitis-specific biomarkers [114].

Although these results are promising, the number of high-throughput genomic and proteomic studies in autoimmune liver diseases is still very limited.

Hepatocellular carcinoma

While beyond the scope of this review, there have been significant advances in early diagnosis, treatment and pathway analysis of HCC and hepatocarcinogenesis through genome, transcriptome and proteome high-throughput analysis, especially in the setting of liver cirrhosis. For more information on this topic we refer to recent reviews [116–118].