OBES SURG (2010) 20:1698–1709

DOI 10.1007/s11695-010-0171-6

PHYSIOLOGY RESEARCH

Hepatic Gene Networks in Morbidly Obese Patients

With Nonalcoholic Fatty Liver Disease
Samer Gawrieh & Tesfaye M. Baye & Melanie Carless & James Wallace &
Richard Komorowski & David E. Kleiner & Deborah Andris & Bassem Makladi &
Regina Cole & Michael Charlton & Joanne Curran & Thomas D. Dyer &
Jac Charlesworth & Russell Wilke & John Blangero & Ahmed H. Kissebah &
Michael Olivier

Published online: 16 May 2010

# Springer Science+Business Media, LLC 2010

Abstract Methods Quantitative global hepatic gene expression

Background Genetic factors alter the risk for nonalcoholic analysis was performed on 53 morbidly obese Caucasian
fatty liver disease (NAFLD). We sought to identify subjects undergoing bariatric surgery (27 with NAFLD and
NAFLD-associated genes and elucidate gene networks 26 controls). After standardization of data, gene expression
and pathways involved in the pathogenesis of NAFLD. profiles were compared between patients with NAFLD and

This work was funded by a grant from the Biotechnology and

Bioengineering Center of the Medical College of Wisconsin (SG and
MO) and grant HL74168 from the Heart, Lung, and Blood Institute of
the National Institutes of Health (MO).
S. Gawrieh (*) M. Carless : J. Curran : T. D. Dyer : J. Charlesworth : J. Blangero
Department of Medicine, Division of Gastroenterology Department of Genetics,
and Hepatology, Medical College of Wisconsin, Southwest Foundation for Biomedical Research,
9200 W. Wisconsin Ave, San Antonio, TX, USA
Milwaukee, WI 53212, USA
e-mail: [email protected]
J. Wallace : D. Andris
R. Wilke Department of Surgery, Medical College of Wisconsin,
Division of General Internal Medicine, Milwaukee, WI, USA
Medical College of Wisconsin,
Milwaukee, WI, USA
R. Komorowski
B. Makladi : A. H. Kissebah Department of Pathology, Medical College of Wisconsin,
Division of Endocrinology, Medical College of Wisconsin, Milwaukee, WI, USA
Milwaukee, WI, USA

S. Gawrieh D. E. Kleiner
Zablocki VA Medical Center, Laboratory of Pathology, National Cancer Institute,
Milwaukee, WI, USA Bethesda, MD, USA

T. M. Baye : R. Cole : R. Wilke : A. H. Kissebah : M. Olivier

Human and Molecular Genetics Center, M. Charlton
Medical College of Wisconsin, Department of Gastroenterology and Hepatology, Mayo Clinic,
Milwaukee, WI, USA Rochester, MN, USA

T. M. Baye : R. Cole : M. Olivier

Biotechnology and Bioengineering Center, M. Olivier
Medical College of Wisconsin, Department of Physiology, Medical College of Wisconsin,
Milwaukee, WI, USA Milwaukee, WI, USA
OBES SURG (2010) 20:1698–1709 1699

controls. The set of genes that significantly correlated patients who share similar clinical risk factors for NAFLD
with NAFLD was further analyzed by hierarchical (e.g., obesity, diabetes, hypertriglyceridemia) widely vary
clustering and ingenuity pathways analyses. from normal to the entire histological spectrum of NAFLD
Results There were 25,643 quantitative transcripts, of [10, 11]. Family studies suggest that the risk for NAFLD is
which 108 were significantly associated with NAFLD partly heritable [12–14]. Ethnicity influences the frequency
(p<0.001). Canonical pathway analysis in the NAFLD- and severity of NAFLD [15, 16]. Finally, emerging data
associated gene clusters showed that the hepatic fibrosis link gene variants with NAFLD and its histological
signaling was the most significant pathway in the up- phenotypes [17–19].
regulated NAFLD gene cluster containing three (COL1A1, Hepatic gene expression in NAFLD has been analyzed
IL10, IGFBP3) significantly altered genes, whereas the in several human studies[11, 20–23], with reported
endoplasmic reticulum stress and protein ubiquitination differential expression of genes involved in lipid
pathways were the most significant pathways in the down- metabolism, mitochondrial function, oxidative stress,
regulated NAFLD gene cluster, with the first pathway insulin signaling, cell death, and hepatic fibrosis. There
containing one (HSPA5) and the second containing two was however little or no concordance among these studies
(HSPA5, USP25) significantly altered genes. The four on the genes described to be associated with NAFLD, and
primary gene networks associated with NAFLD were although their functional classifications were described,
involved in cell death, immunological disease, cellular the potential interactions among significantly altered
movement, and lipid metabolism with several significantly genes were not characterized.
altered “hub” genes in these networks. The aims of this study were to identify genes
Conclusions This study reveals the canonical pathways associated with NAFLD and elucidate gene networks
and gene networks associated with NAFLD in morbidly and pathways involved in the pathogenesis of NAFLD
obese Caucasians. The application of gene network using integrated bioinformatics analyses of global quan-
analysis highlights the transcriptional relationships titative hepatic expression in a morbidly obese Caucasian
among NAFLD-associated genes and allows identifica- cohort.
tion of hub genes that may represent high-priority candi-
dates for NAFLD.
Methods
Keywords Fatty liver . NAFLD . Genes . Gene networks .
Gene expression . Pathogenesis Recruitment and Phenotyping of Subjects

The study protocol has been reviewed and approved by

Introduction the Medical College of Wisconsin's Institutional Review
Board. Subjects gave written informed consent for
Nonalcoholic fatty liver disease (NAFLD) is the most participation in the study. Subjects were of northern
common liver disease in the United States [1–3]. It has a European descent, morbidly obese (BMI≥ 40 kg/m2 or
wide spectrum of phenotypes starting with simple steatosis >35 kg/m2 with significant comorbidities) with docu-
(SS) to steatosis with a variable mix of subphenotypes mented unsuccessful dietary attempts to lose and maintain
including ballooning, inflammation, and fibrosis, as diag- weight, and who underwent bariatric surgery at the MCW.
nosed on liver biopsy [4]. Nonalcoholic steatohepatitis A protocol intraoperative liver biopsy was performed on
(NASH) represents an advanced phenotype where most of all patients for histological phenotyping. Patients with
these subphenotypes are present [5]. Unlike SS, which rarely alcohol intake >20 g/d and those with other liver diseases
results in liver complications [6, 7], NASH can progress to (hepatitis B, hepatitis C, autoimmune hepatitis, primary
cirrhosis, liver failure, and hepatocellular carcinoma [6, 7]. biliary cirrhosis, Wilson's disease, alpha1 antitrypsin
The prevalence of NAFLD and NASH will probably deficiency, or hemochromatosis) based on positive
increase as part of the ongoing obesity epidemic and will disease-specific serological tests and suggestive liver
likely add a huge burden to the healthcare system [8]. histology were excluded. Patients using drugs associated
Although some of the molecular mechanisms involved with NAFLD (systemic glucocorticosteroids, tamoxifen,
in onset and progression of steatosis, necro-inflammation, tetracycline, amiodarone, methotrexate, valproic acid,
and fibrosis in NAFLD have been well characterized [9], anabolic steroids, estrogens at doses higher than those
little is known about the underlying complex gene used for hormone replacement, or other known hepato-
interaction networks involved in NAFLD. toxins) preceding the liver biopsy were also excluded. For
Several observations point to a role for genetic factors in all participants, clinical, and biochemical data were
the pathogenesis of NAFLD. Liver biopsy findings in collected.
1700 OBES SURG (2010) 20:1698–1709

Collection of Biological Material million beads are available to quantitate RNA levels for
each sample. The HumanWG-6 Expression BeadChips
All subjects had fasting blood samples for serum extraction were scanned on an Illumina BeadArray™ reader. The
collected in the morning of planned surgery. In addition to BeadStudio software package included with the Illumina®
the biopsy obtained for routine histological analysis and BeadStation 500GX system extracted gene expression data
phenotyping, an additional intraoperative liver biopsy was from images collected from the Illumina BeadArray Reader.
done, and tissue was snap-frozen immediately in the The application reports experiment performance based on
operating room and used for RNA studies. built-in controls that accompany each experiment.

Histological Evaluation and Phenotypes Definitions Identification of Detectable Transcripts

All liver biopsy samples were read by two expert For each sample, approximately 48,000 transcripts were
pathologists (R.K. and D.E.K.) to define the NAFLD quantified using the BeadChip supplied by Illumina. A
phenotype and semiquantitatively score the individual χ2 “tail” test was used to assess whether there is a
histological features/subphenotypes including steatosis, significant excess of samples with transcript-specific
lobular and portal inflammation, hepatocellular ballooning, expression values above the 95th percentile of the null
Mallory's hyaline, and fibrosis according to the scoring distribution. We used this test because it allows “detec-
system suggested by the NIH NASH Clinical Research tion” of even those RNA molecules that are clearly present
Network working group [24]. NAFLD was diagnosed above baseline levels in only a subset of individuals.
when ≥5% macrosteatosis was present. Subjects with 0 to Using a false discovery rate of 5% [25], we identified
5% macrosteatosis were diagnosed as non-NAFLD con- expression phenotypes that exhibited significant expres-
trols. Both pathologists confirmed the phenotype as sion by this criterion.
NAFLD or non-NAFLD control on all included subjects.
The analyses in this study focused on genes associated Standardization of Expression Values
with NAFLD.
For the analysis of transcriptional variation, we focused on
RNA Isolation from Liver Tissue those detectable transcripts that passed a tail test that
determined if there is sufficient quantitative signal over that
Frozen liver tissue samples were mixed with lysis buffer, expected by chance. To minimize the influence of overall
homogenized and used for total RNA isolation following signal levels, which may reflect RNA quantity and quality
the Qiagen RNeasy Mini Kit protocol. The isolated RNA rather than a true biological difference between individuals,
concentration (µg) and purity (260:280 nm) were measured abundance values of all retained transcripts were
using a NanoDrop spectrophotometer. The resulting RNA standardized by z-scoring within individuals (using decile
was reverse-transcribed, converted into cDNA, and subse- percentage bins of transcripts, grouped by average log-
quently amplified producing cRNA using the MessageAmp transformed raw signals across individuals), followed by
II RNA Amplification Kit (Ambion). This method has been linear regression against individual-specific average log-
shown to reliably amplify small amounts of messenger transformed raw signal and its squared value. For each
RNA (mRNA) obtained from limited tissue samples transcript, we directly normalized these residual expression
obtained by biopsy. scores by employing an inverse Gaussian transformation
across individuals to ensure that the assumptions
Transcriptional Profiling underlying variance components-based analyses were not
violated. The normalized phenotypes were comparable
We used commercially available Illumina HumanWG-6 between individuals and across transcripts. Current micro-
Expression BeadChips for whole genome expression arrays allow accurate detection of gene transcripts that
analysis. The BeadChips contain six arrays, each with correlate well with real-time PCR as shown in many studies
approximately 48,000 probes derived from human [11, 20, 21, 26, 27]. Furthermore, the high heritability of
genes in the National Center for Bioinformatic Infor- housekeeping gene transcripts traditionally used in RT-PCR
mation Reference Sequence (RefSeq) and UniGene validation (e.g., heritability for B-actin expression is 0.373,
databases. This system uses a “direct hybridization” p1:0  10 9 , and for cyclophilin-D is 0.559, p1:0  10 9 )
assay, whereby gene-specific probes are used to detect [28] make them less than ideal as references for internal
labeled RNAs. Each bead in the array contains a standardization. Lastly, the limited availability of tissue for
50-mer, sequence-specific oligo probe synthesized using mRNA extraction restricted our ability to run RT-PCR
Illumina's Oligator™ in-house technology. More than 1.6 analysis.
OBES SURG (2010) 20:1698–1709 1701

Detection of NAFLD-Associated Gene Transcripts relationship between two nodes is represented as an edge
(line). The connectivity of genes (nodes) is based on data in
The t test statistic was used to measure the difference the IPA knowledge base, which is a large repository of gene-
between the sample means in units of the standard phenotype associations, molecular interactions, chemical
deviation for which the difference can be tested using knowledge, and regulatory events, pulled from the full text
certain p value. We used a significance cut-point of of scientific publications by Ph.D.-level scientists. The node
p<0.001. color indicates the degree of up-regulation (red) or down-
regulation (green). Nodes are displayed using various shapes
Hierarchical Clustering that represent the functional class of the gene product. A
functional analysis of a network then identified the biological
This was performed with two groups of samples (NAFLD functions and/or diseases that were most significant to the
vs. controls) using Genesis software [29]. Cluster analysis genes in the network.
was performed to divide genes or samples into groups on
the basis of similarities among their gene expression
profiles, which is often indicative of similarity with respect Results
to function.
Characteristics of Study Subjects
Gene Network and Pathway Analyses
Fifty-three Caucasian subjects whose histological pheno-
To describe the interactions between the altered gene type was confirmed by the two study pathologists were
expression levels in NAFLD, the Ingenuity Pathways included in this analysis: 27 with NAFLD and 26 controls.
Analysis (IPA) software program (Ingenuity Systems, The clinical and laboratory characteristics of these subjects
version 6.1) was used. This software application allows are summarized in Table 1. As anticipated, subjects with
identification of biological and network/pathway interac- NAFLD had higher mean fasting glucose, insulin, trigly-
tions between genes. After uploading a list of genes that are cerides, and ALT levels. The total cholesterol, HDL, LDL,
significantly differentially expressed in our samples, the and alkaline phosphatase levels were similar between the
program was used to uncover any interactions between the groups. There were no significant differences in age, BMI,
genes. IPA identifies the biological functions and pathways percentage of females, or fractions of subjects with diabetes
that are most relevant to the experimental datasets. The or hypertension between the two groups.
altered genes identified by expression profiling analysis
were mapped to genetic networks available in the IPA Hepatic Transcriptional Profile in NAFLD
database, ranked by score, and presented as a graph
indicating the molecular relationships between genes. A total of 25,643 quantitative transcripts were identified in
liver using the proposed transcriptional profiling. Of these,
Ingenuity Pathway Analysis Procedures 108 were significantly associated with NAFLD (p<0.001).
The strongest 10 up- or down-regulated NAFLD-associated
From the llumina HumanWG-6 data, we uploaded the genes are shown in Table 2.
ProbeID and the associated expression value to upload into
IPA. Each ProbeID was mapped to its corresponding gene Cluster-Based Canonical Pathway Analysis
object in the Ingenuity Pathways Knowledge Base. These
genes, called focus genes, were overlaid onto a global To gain further insights into the pathogenesis of NAFLD,
molecular network developed from information contained in we analyzed the NAFLD-correlated genes to elucidate
the database. Networks of these focus genes were then dominant canonical pathways. Hierarchical clustering
algorithmically generated based on their types of interactions analysis was first performed to organize the 108 NAFLD-
(direct and/or indirect). Scores were generated (based on correlated transcripts based on similarity of function within
Fisher's test) to rank networks according to how relevant the up- or down-regulated clusters (Fig. 1). These two
they are to the genes in the input dataset. The score takes into clusters were then subjected to canonical pathway analysis,
account the number of focus genes (genes in our lists) in the which showed that hepatic fibrosis signaling was the most
network and the size of the network to approximate how significant pathway in the up-regulated NAFLD gene
relevant this network is to the original list of focus genes. cluster containing three (COL1A1, IL10, IGFBP3) signifi-
The network is then presented as a graph indicating the cantly altered genes. Endoplasmic reticulum (ER) stress and
molecular relationships between genes/gene products. Genes protein ubiquitination pathways were the most significant
or gene products are represented as nodes, and the biological pathways in the down-regulated NAFLD gene cluster, with
1702 OBES SURG (2010) 20:1698–1709

Table 1 Characteristics of
study subjects Variable Control (n=26) NAFLD (n=27) p

Age (years) 40.5±12.4 43.1±10.3 ns

Female, n (%) 25 (96) 24 (89) ns
BMI (kg/m2) 48.8±5.9 49.6±7.4 ns
Diabetes, n (%) 7 (26.9) 11 (40.7) ns
Hypertension, n (%) 13 (50) 19 (70.3) ns
Fasting glucose (mg/dL) 103.9±43.2 131.0±50.8 0.04
Fasting insulin (mu/mL) 20.2±13.5 35.8±20.0 0.001
Total cholesterol (mg/dL) 175.8±29.1 178.3±38.6 ns
Triglycerides (mg/dL) 123.6±37.3 172.9±80.3 0.006
HDL (mg/dL) 45.4±9.1 41.1±7.9 ns
LDL (mg/dL) 105.5±26.2 103.3±32.2 ns
ALT (U/L) 18.1±11.6 30.0±16.3 0.003
Total bilirubin (mg/dL) 0.4±0.1 0.5±0.3 ns
Data presented as frequency or Alkaline phosphatase (U/L) 74.3±31.1 73.3±23.2 ns
mean ± standard deviation Platelets (103/uL) 309.3±73.6 298.9±72.7 ns
unless otherwise indicated

the first pathway containing one (HSPA5) and the second Hepatic Gene Networks Analysis
containing two (HSPA5, USP25) significantly altered genes.
Remaining relevant pathways not reaching significance at p To determine significant biological functions and to reveal
<0.05 in each cluster are shown in Fig. 2. transcriptional correlations among genes associated with

Table 2 Strongest NAFLD-associated genes

Molecules Exp. Value Exp. Chart

Log ratio up-regulated

PPFIBP1 0.583
ZAK 0.527
WNK4 0.526
RGN 0.524
SMUG1 0.521
MLLT10 0.505
DCN 0.497
CYP4F22 0.497
CSN2 0.495
FABP4 0.491
Log ratio down-regulated
GABRB2 −0.638
CHERP −0.600
EBNA1BP2 −0.579
DYNC1I1 −0.570
RAB3IP −0.562
KLHL8 −0.546
TUBD1 −0.507
HEATR3 −0.491
OCM2 −0.480
MKX −0.476
Fig. 1 Hierarchal clustering of 108 NAFLD-correlated genes
OBES SURG (2010) 20:1698–1709 1703

Fig. 2 The canonical pathways

detected in each NAFLD-
associated gene cluster. a Up-
regulated pathways and b
Down-regulated pathways

NAFLD, the 108 significant genes were subjected to gene network analysis. Four significant hepatic gene networks
network analysis. were noted in NAFLD (Table 4). Prominent gene
The most significant cellular and molecular biological functions within these networks included cell death,
functions associated with NAFLD-correlated genes were immunological disease, cancer, cellular growth and pro-
cell death, cellular movement, antigen presentation, cell liferation, cellular movement, lipid metabolism, molecular
morphology, and cellular development. Genes in each of transport, and small molecule biochemistry.
these functional categories are listed in Table 3. To demonstrate the biological interactions of these
Of the 108 NAFLD-associated genes, 63 had connec- genes within a network and highlight “hub” genes that
tivity data in IPA knowledge base and were eligible for interact with other NAFLD-correlated genes, gene net-
1704 OBES SURG (2010) 20:1698–1709

Table 3 Gene biological functions associated with NAFLD

p Molecules

Cell death 3.43E-04-3.14E-02 SMUG1, DDIT3, DPP3, IL10, DCN, EEF1A2, CXCL12,
IGFBP3, ULBP2, RGN, MAFB, HSPA5
Cellular movement 5.46E-04-3.52E-02 TSPAN3, COL1A1, IL10, ITGA9, DCN, CXCL12,
IGFBP3, FABP4, MAFB
Antigen presentation 9.93E-04-3.13E-02 COL1A1, IL10, DCN, CXCL12, FABP4
Cell morphology 9.93E-04-3.48E-02 RAB3IP, COL1A1, PSPN, MLLT10, IL10, KPNA2,
CXCL12, NEFM, IGFBP3
Cellular development 1.57E-03-3.22E-02 CD164, COL1A1, DDIT3, IL10, DCN, CXCL12,
IGFBP3, MAFB, PRDM16

works involved in cell death, lipid metabolism, and interrogated transcripts. Despite these limitations, these
molecular transport are shown in Fig. 3. studies provided insights into the pathogenesis of NAFLD
and highlighted a large number of NAFLD-correlated
genes. They also demonstrated the need for additional
Discussion bioinformatics analytical methods that can both provide a
systematic approach to connect the data to biology and aid
Several human studies have analyzed hepatic gene expres- in prioritizing the often large number of NAFLD-associated
sion in NAFLD [11, 20–23]. These studies reported genes.
differential expression of genes involved in lipid metabo- In this study, we analyzed global hepatic gene expression
lism, mitochondrial function, oxidative stress, insulin in a clinically well characterized and ethnically homoge-
signaling, cell death, and hepatic fibrosis. There was neous cohort. Using gene networks and pathways analyses,
however little or no concordance among these studies on we confirmed a significant role of several previously
the genes described to be associated with NAFLD. This described biological processes in the pathogenesis of
may be due to small sample size, study of ethnically mixed NAFLD.
cohorts or cohorts of different ethnicities, or use of different There were several interesting genes in our study that
gene expression platforms with varying numbers of showed strong evidence of up-regulation including FABP4,

Table 4 The significant hepatic gene networks associated with NAFLD

ID Score Top functions Focus Molecules in network

molecules

1 39 Cell death, renal necrosis/cell death, 18 Akt, AMOTL2, Ap1, Ck2, COL1A1, Collagen(s), CSN2,
renal, and urological disease CXCL12, DCN, DDIT3, DGKZ, ERK, FABP4, FSH, IGFBP3, IL1,
IL10, Insulin, ITGA9, Jnk, LDL, MAFB, Mapk, NFkB
(complex), P38 MAPK, PI3K, PPFIBP1, RAB3IP, Ras, STAT5a/b,
STX4, Tgf beta, ULBP2, UNC5CL, ZAK
2 23 Immunological disease, cancer, cellular 12 CD164, CD274, CD33 (includes EG:945), CHI3L2, CTSZ
growth, and proliferation (includes EG:1522), DPP3, FOS, GPNMB, GPR109B, HRSP12,
HSP, IFNG, KPNA2, LZTS1, MEFV, MIRN17 (includes EG:406952),
MKX, MRPS6 (includes EG:64968), NAT6, NEFM, neuroprotectin
D1, NFE2L2, OCM2, PRKACA, PSPN, RBP3, RIF1, SLC15A3,
SLC1A4, TGFB1, TNF, TYRP1, USP4, USP25, VIM
3 21 Cellular movement, nervous system 11 ATF6B, CLDN2, CLDN3, CLDN11, DGKD (includes EG:8527),
development and function, organismal EEF1A2, ERP29, GCHFR, GPR37, HDLBP, HEATR3, HNF4A,
functions HSPA5, INS1, ITGB1, Mg2+, OS9, POLL, RBM42, RGN, SEL1L,
SFRS12, SLC2A4, SMAD2, SMUG1, SNRPD3, SSBP1, STXBP4,
THRAP3, TMBIM6, TRAF6, TSPAN, TSPAN3, VARS, WNK4
4 21 Lipid metabolism, molecular transport, 11 ALOX5AP, ASF1B, ATP1B1, BICD1, CCNE1, CEP76, CHERP,
small molecule biochemistry CLIP1, COIL, DOT1L, DYNC1I1, DYNLL2, EBNA1BP2, FAAH,
GABRB2, HIST1H3E, Histone h4, HTR1A, IL4, KLC2, KLHL8,
LEPROT, MLLT10, NFIL3, NOC2L, NR3C1, PPP5C, progesterone,
SLC1A7, SMARCB1, SS18, TGFBI, TNFSF4, TP53BP1, TP53I3
OBES SURG (2010) 20:1698–1709 1705

The protein encoded by liprin beta 1 (PPFIBP1) is a

member of the LAR protein-tyrosine phosphatase-
interacting protein (liprin) family. Tyrosine phosphorylation
of proteins is important in coordinating cellular responses
to extracellular stimuli. This protein was found to play a
role in altering tumor invasiveness and metastasis [34, 35].
The ZAK gene encodes a cytoplasmic protein, which is a
member of the MAPKKK family of signal transduction
molecules. The protein has proapoptotic activity, plays a
role in cell cycle checkpoint regulation, and mediates the
effects of tumor growth factor-β (TGF-β) in inducing
myocardial fibrosis [36–38]. Although TGF-β is a potent
inducer of hepatic stellate cells production of collagen [39],
the effects of its interaction with ZAK on hepatic fibrosis
are currently unknown.
The gene encoding regucalcin (RGN) was also among
the strongest up-regulated in our study. Regucalcin is a
calcium-binding protein, which regulates the calcium
effects on liver cell functions, modulates lipid metabolism
in the adipose and liver tissues in aging rats by interacting
with leptin or adiponectin, and suppresses apoptotic cell
death [40–42].
The SMUG1 (single-strand-selective monofunctional
uracil-DNA glycosylase 1) gene encodes a glycosylase that
removes uracil from single- and double-stranded DNA in
nuclear chromatin, thus contributing to base excision repair
[43, 44]. DNA damage is a well-described phenomenon in
NAFLD [45].
The DCN gene encodes for decorin, an extracellular
matrix proteoglycan that binds to type I collagen fibrils and
exerts several important biological effects. Decorin sup-
Fig. 3 Significant hepatic gene networks in NAFLD. a This gene presses TGF-β activation of hepatic stellate cells and thus
network is involved in cell death. Several significantly expressed hub influences matrix assembly and remodulation, and suppresses
genes such as COL1A1, IL10, and DCN can be identified. b This gene different cancer cell lines by inducing apoptosis [46–49].
network is involved in lipid metabolism, molecular transport, and
small molecules biochemistry. Red: up-regulated in NAFLD compared
The CYP4F22 (cytochrome P450, family 4, subfamily F,
to control; green: down-regulated in NAFLD compared to control polypeptide 22) gene encodes a member enzyme of the
cytochrome P450 superfamily that are involved in drug
metabolism and synthesis of cholesterol, steroids, and other
PPFIBP1, ZAK, RGN, SMUG1, DCN, CYP4F22, and lipids.
CSN2. The protein encoded by CSN2, casein beta, has a role in
The FABP4 gene has been also reported to be up- regulating protein stability and degradation via the ubiquitin–
regulated in a prior hepatic expression study in NAFLD proteasome system and is involved in cell transformation,
[23]. It encodes for fatty acid binding protein 4, adipocyte, tumorigenesis, and hepatic regeneration [50–52].
a member of the fatty acid-binding proteins. FABPs are Less is known about the strongest down-regulated genes
lipid chaperones with high affinity to binding long chain discovered in this study. The CHERP (calcium homeostasis
fatty acids and are important mediators of inflammation and ER protein) gene encodes a protein that modulate calcium
insulin signaling in glucophages and adipocytes [30, 31]. homoeostasis, cell growth and proliferation [53], and the
Genetic variation in FABP4 has been shown to influence DYNC1I1 (dynein, cytoplasmic 1, intermediate chain 1)
triglycerides levels and the risk for diabetes in humans [32]. gene encodes for a molecular motor protein that is
Indeed, mice deficient in FABP4 and another FABP (mal1) important for cell division and transport of intracellular
are protected from diet-induced obesity, insulin resistance, organelles and cell migration [54, 55].
diabetes, and fatty liver [33]. Therefore, FABP4 represents a Beyond the identification of individual genes, our
logical NAFLD candidate gene. analysis focused on the identification and characterization
1706 OBES SURG (2010) 20:1698–1709

of overarching biological functions associated with these and resulted in improvement in insulin sensitivity and
genes. The most significant biological functions involving hepatic triglycerides [72]. Based on these data, HSPA5
genes with significantly altered expression included cell represents a logical NAFLD candidate gene. The
death, cellular movement, antigen presentation, cell mor- ubiquitin-specific peptidase 25 (USP25) is protease
phology, and cellular development. These data are involved in the release of ubiquitin from degraded proteins
consistent with findings of other studies revealing the by disassembly of the polyubiquitin chains [73, 74].
biological significance of cell death [56, 57] and altered However, its biological functions are largely unknown.
immunological function [58, 59] in NAFLD. About 55% of our subjects with NAFLD had portal
Canonical pathways analysis revealed that the hepatic or lobular inflammation on liver biopsy. Of the
fibrosis pathway was the only significant up-regulated differentially expressed genes in NAFLD, FABP4 and
pathway in NAFLD, with COL1A1, IL10, and IGFBP3 IL10 have important roles in mediating inflammation and
being the most significantly altered in this cohort. Increased thus may contribute to the inflammatory findings in these
collagen 1 deposition results in progressive hepatic fibrosis, patients.
which may ultimately lead to cirrhosis. The type I pro- Although statistical significance of expression level
collagen, alpha 1 (COL 1A1) gene is up-regulated in changes may be one way to select a candidate gene for a
subjects with NAFLD. Whether variants in this gene are given disease, gene network analysis offers the advantage
associated with altered risk for increased fibrosis with of understanding the interaction of significant genes
NAFLD needs to be explored. Interleukin 10 (IL10) is a associated with a disease and the ability to find hub
cytokine with pleiotropic effects on immunoregulation and genes within a network that interact with several other
inflammation. It has been shown to improve insulin genes up- and downstream of them. The high intercon-
resistance and reduce hepatic steatosis [60] . The insulin- nectivity of hub genes with other correlated genes within
like growth factor binding protein 3 (IGFBP3) is one of six a biological network may imply functional and biological
high-affinity binding proteins for IGFs that reduce the importance of these genes [75]. This approach to
levels of free IGF and antagonize their insulin-like activity. identifying hub genes via gene network analysis has been
It interferes with adipocyte differentiation [61] and also applied to expression analysis of different tissues in other
interferes with insulin and glucose homeostasis resulting in rodent and human disease models, where hub genes were
impaired glucose tolerance and insulin resistance [62, 63]. proposed as potential biomarkers for the disease or targets
IGFBP3 has IGF-independent cellular activities such for drug therapy [76–78]. Thus, these hub genes may be
control of cell growth and apoptosis [64, 65]. It therefore viewed as high-priority disease candidates for further
represents another attractive NAFLD candidate gene. study. In this dataset, four significant gene networks were
The finding that the two significantly down-regulated associated with NAFLD (Table 4), and the IL10 and
canonical pathways in NAFLD were the ER stress and COL1A1 are examples of significantly expressed hub
protein ubiquitination pathways is interesting. This finding genes in network A.
highlights the importance of ER in the pathogenesis of The described functions of these top networks fit well
NAFLD, whether it is with the unfolded protein response to with the known major physiological events in NAFLD,
accumulating misfolded/unfolded proteins (ER stress) or such as lipid metabolism, organismal functions, and cell
the ubiquitination of these misfolded proteins as a quality- death. Similar to our findings in human NAFLD, network
control response [66, 67]. These data are in line with earlier analysis of hepatic gene expression in a rat model of
studies demonstrating a role for ER stress in inducing NAFLD showed importance of genes involved in cell
apoptosis in hepatocytes after accumulation of fatty acids death, immune function, and lipid metabolism [79].
[68] and with human data showing ER stress and Several points need to be noted: (1) the subjects
suboptimal unfolded protein response in NAFLD [69]. characterized within this study were morbidly obese and
There were two genes in these pathways that were undergoing bariatric surgery. Therefore, findings of this
significantly expressed in the dataset: HSPA5 and study may not be applicable to other patients with NAFLD
USP25. The heat shock 70-kDa protein 5 (HSPA5 also and lesser degrees of obesity. (2) Most of the subjects were
known as BIP or GRP78) is a member of the glucose- females, and thus, there may be male-specific changes in
regulated proteins (GRPs). It acts as an ER chaperone hepatic gene expression that our analysis would not have
and sensor of ER stress [70]. Although a recent human uncovered. (3) Although correlation of significantly
study of ethnically mixed cohort did not show a expressed genes levels with laboratory variables such as
significant difference in HSPA5 (BIP) between controls insulin and ALT levels are of significant clinical interest,
and subjects with NAFLD [69], HPSA5 has been shown the limited size of the current study cohort does not allow
to influence insulin biosynthesis [71] and its overexpres- for these analyses with sufficient statistical power. (4) It is
sion in mice reduced ER stress and SREBP-1c cleavage not possible from the current data to decipher which altered
OBES SURG (2010) 20:1698–1709 1707

gene transcripts are primary or secondary changes in 15. Browning JD, Szczepaniak LS, Dobbins R, et al. Prevalence of
hepatic steatosis in an urban population in the United States:
relation to NAFLD pathogenesis. Ultimately, combining
impact of ethnicity. Hepatology. 2004;40(6):1387–95.
genomics, proteomics, and transcriptomics methods may 16. Browning JD, Kumar KS, Saboorian MH, et al. Ethnic differences
help in dissecting these changes. This study, nonetheless, in the prevalence of cryptogenic cirrhosis. Am J Gastroenterol.
combined the strengths of recruitment of an ethnically 2004;99(2):292–8.
17. Miele L, Beale G, Patman G, et al. The Kruppel-like factor 6
homogeneous cohort with detailed clinical phenotypic
genotype is associated with fibrosis in nonalcoholic fatty liver
characterization, use of whole-genome quantitative tran- disease. Gastroenterology. 2008;135(1):282–91.
scriptional profiling, and application of integrated bioinfor- 18. Romeo S, Kozlitina J, Xing C, et al. Genetic variation in PNPLA3
matics analyses to allow immediate connection of confers susceptibility to nonalcoholic fatty liver disease. Nat
Genet. 2008;40(12):1461–5.
transcriptional data to NAFLD biology. 19. Yoneda M, Hotta K, Nozaki Y, et al. Association between
In summary, this study reveals the gene biological PPARGC1A polymorphisms and the occurrence of nonalcoholic
functions and canonical pathways associated with fatty liver disease (NAFLD). BMC Gastroenterol. 2008;8:27.
NAFLD. The application of gene networks analysis 20. Sreekumar R, Rosado B, Rasmussen D, et al. Hepatic gene
expression in histologically progressive nonalcoholic steatohepa-
highlights the transcriptional relationships among
titis. Hepatology. 2003;38(1):244–51.
NAFLD-associated genes and allows identification of 21. Chiappini F, Barrier A, Saffroy R, et al. Exploration of global
hub genes that may represent high priority candidates gene expression in human liver steatosis by high-density
for NAFLD. oligonucleotide microarray. Lab Invest. 2006;86(2):154–65.
22. Yoneda M, Endo H, Mawatari H, et al. Gene expression profiling
of non-alcoholic steatohepatitis using gene set enrichment
analysis. Hepatol Res. 2008;38(12):1204–12.
23. Greco D, Kotronen A, Westerbacka J, et al. Gene expression in
References
human NAFLD. Am J Physiol Gastrointest Liver Physiol.
2008;294(5):G1281–7.
1. Schwimmer JB, Deutsch R, Kahen T, et al. Prevalence of 24. Kleiner DE, Brunt EM, Van Natta M, et al. Design and validation
fatty liver in children and adolescents. Pediatrics. 2006;118 of a histological scoring system for nonalcoholic fatty liver
(4):1388–93. disease. Hepatology. 2005;41(6):1313–21.
2. Clark JM, Brancati FL, Diehl AM. The prevalence and etiology of 25. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a
elevated aminotransferase levels in the United States. Am J practical and powerful approach to multiple testing. J R Stat Soc
Gastroenterol. 2003;98(5):960–7. B. 1995;57:289–300.
3. Charlton M. Nonalcoholic fatty liver disease: a review of current 26. Elam MB, Cowan Jr GS, Rooney RJ, et al. Hepatic gene
understanding and future impact. Clin Gastroenterol Hepatol. expression in morbidly obese women: implications for disease
2004;2(12):1048–58. susceptibility. Obesity (Silver Spring). 2009;17(8):1563–73.
4. Sanyal AJ. AGA technical review on nonalcoholic fatty liver 27. Maiti AK. Gene network analysis of oxidative stress-mediated
disease. Gastroenterology. 2002;123(5):1705–25. drug sensitivity in resistant ovarian carcinoma cells. Pharmacoge-
5. Brunt EM. Nonalcoholic steatohepatitis. Semin Liver Dis. nomics J. 2010;10(2):94–104.
2004;24(1):3–20. 28. Goring HH, Curran JE, Johnson MP, et al. Discovery of
6. Matteoni CA, Younossi ZM, Gramlich T, et al. Nonalcoholic fatty expression QTLs using large-scale transcriptional profiling in
liver disease: a spectrum of clinical and pathological severity. human lymphocytes. Nat Genet. 2007;39(10):1208–16.
Gastroenterology. 1999;116(6):1413–9. 29. Eisen MB, Spellman PT, Brown PO, et al. Cluster analysis and
7. Ekstedt M, Franzen LE, Mathiesen UL, et al. Long-term follow- display of genome-wide expression patterns. Proc Natl Acad Sci
up of patients with NAFLD and elevated liver enzymes. U S A. 1998;95(25):14863–8.
Hepatology. 2006;44(4):865–73. 30. Furuhashi M, Hotamisligil GS. Fatty acid-binding proteins: role in
8. Lazo M, Clark JM. The epidemiology of nonalcoholic fatty liver metabolic diseases and potential as drug targets. Nat Rev Drug
disease: a global perspective. Semin Liver Dis. 2008;28(4):339–50. Discov. 2008;7(6):489–503.
9. Browning JD, Horton JD. Molecular mediators of hepatic 31. Furuhashi M, Fucho R, Gorgun CZ, et al. Adipocyte/Macrophage
steatosis and liver injury. J Clin Invest. 2004;114(2):147–52. fatty acid-binding proteins contribute to metabolic deterioration
10. Dixon JB, Bhathal PS, O'Brien PE. Nonalcoholic fatty liver through actions in both macrophages and adipocytes in mice. J
disease: predictors of nonalcoholic steatohepatitis and liver fibrosis Clin Invest. 2008;118(7):2640–50.
in the severely obese. Gastroenterology. 2001;121(1):91–100. 32. Tuncman G, Erbay E, Hom X, et al. A genetic variant at the fatty
11. Younossi ZM, Baranova A, Ziegler K, et al. A genomic and acid-binding protein aP2 locus reduces the risk for hypertriglycer-
proteomic study of the spectrum of nonalcoholic fatty liver idemia, type 2 diabetes, and cardiovascular disease. Proc Natl
disease. Hepatology. 2005;42(3):665–74. Acad Sci U S A. 2006;103(18):6970–5.
12. Struben VM, Hespenheide EE, Caldwell SH. Nonalcoholic 33. Maeda K, Cao H, Kono K, et al. Adipocyte/Macrophage fatty acid
steatohepatitis and cryptogenic cirrhosis within kindreds. Am J binding proteins control integrated metabolic responses in obesity
Med. 2000;108(1):9–13. and diabetes. Cell Metab. 2005;1(2):107–19.
13. Willner IR, Waters B, Patil SR, et al. Ninety patients with 34. Serra-Pages C, Medley QG, Tang M, et al. Liprins, a family of
nonalcoholic steatohepatitis: insulin resistance, familial tendency, LAR transmembrane protein-tyrosine phosphatase-interacting
and severity of disease. Am J Gastroenterol. 2001;96(10):2957–61. proteins. J Biol Chem. 1998;273(25):15611–20.
14. Schwimmer JB, Celedon MA, Lavine JE, et al. Heritability of 35. Kriajevska M, Fischer-Larsen M, Moertz E, et al. Liprin beta 1, a
nonalcoholic fatty liver disease. Gastroenterology. 2009;136 member of the family of LAR transmembrane tyrosine
(5):1585–92. phosphatase-interacting proteins, is a new target for the
1708 OBES SURG (2010) 20:1698–1709

metastasis-associated protein S100A4 (Mts1). J Biol Chem. 55. Gennerich A, Vale RD. Walking the walk: how kinesin and dynein
2002;277(7):5229–35. coordinate their steps. Curr Opin Cell Biol. 2009;21(1):59–67.
36. Yang JJ. Mixed lineage kinase ZAK utilizing MKK7 and not 56. Feldstein AE, Canbay A, Angulo P, et al. Hepatocyte apoptosis
MKK4 to activate the c-Jun N-terminal kinase and playing a role and fas expression are prominent features of human nonalcoholic
in the cell arrest. Biochem Biophys Res Commun. 2002;297 steatohepatitis. Gastroenterology. 2003;125(2):437–43.
(1):105–10. 57. Malhi H, Bronk SF, Werneburg NW, et al. Free fatty acids induce
37. Jandhyala DM, Ahluwalia A, Obrig T, et al. ZAK: a MAP3Kinase JNK-dependent hepatocyte lipoapoptosis. J Biol Chem. 2006;281
that transduces Shiga toxin- and ricin-induced proinflammatory (17):12093–101.
cytokine expression. Cell Microbiol. 2008;10(7):1468–77. 58. Li Z, Oben JA, Yang S, et al. Norepinephrine regulates hepatic
38. Cheng YC, Kuo WW, Wu HC, et al. ZAK induces MMP-2 innate immune system in leptin-deficient mice with nonalcoholic
activity via JNK/p38 signals and reduces MMP-9 activity by steatohepatitis. Hepatology. 2004;40(2):434–41.
increasing TIMP-1/2 expression in H9c2 cardiomyoblast cells. 59. Li Z, Soloski MJ, Diehl AM. Dietary factors alter hepatic innate
Mol Cell Biochem. 2009;325(1–2):69–77. immune system in mice with nonalcoholic fatty liver disease.
39. Friedman SL. Molecular regulation of hepatic fibrosis, an Hepatology. 2005;42(4):880–5.
integrated cellular response to tissue injury. J Biol Chem. 60. den Boer MA, Voshol PJ, Schroder-van der Elst JP, et al.
2000;275(4):2247–50. Endogenous interleukin-10 protects against hepatic steatosis but
40. Yamaguchi M. Role of calcium-binding protein regucalcin in does not improve insulin sensitivity during high-fat feeding in
regenerating rat liver. J Gastroenterol Hepatol. 1998;13(Suppl): mice. Endocrinology. 2006;147(10):4553–8.
S106–12. 61. Chan SS, Schedlich LJ, Twigg SM, et al. Inhibition of adipocyte
41. Izumi T, Yamaguchi M. Overexpression of regucalcin suppresses differentiation by insulin-like growth factor-binding protein-3. Am
cell death and apoptosis in cloned rat hepatoma H4-II-E cells J Physiol Endocrinol Metab. 2009;296(4):E654–63.
induced by lipopolysaccharide, PD 98059, dibucaine, or Bay K 62. Silha JV, Gui Y, Murphy LJ. Impaired glucose homeostasis in
8644. J Cell Biochem. 2004;93(3):598–608. insulin-like growth factor-binding protein-3-transgenic mice. Am
42. Fukaya Y, Yamaguchi M. Overexpression of regucalcin sup- J Physiol Endocrinol Metab. 2002;283(5):E937–45.
presses cell death and apoptosis in cloned rat hepatoma H4-II-E 63. Chan SS, Twigg SM, Firth SM, et al. Insulin-like growth factor
cells induced by insulin or insulin-like growth factor-I. J Cell binding protein-3 leads to insulin resistance in adipocytes. J Clin
Biochem. 2005;96(1):145–54. Endocrinol Metab. 2005;90(12):6588–95.
43. Matsubara M, Tanaka T, Terato H, et al. Action mechanism of 64. Oh Y, Muller HL, Lamson G, et al. Insulin-like growth factor
human SMUG1 uracil-DNA glycosylase. Nucleic Acids Symp Ser (IGF)-independent action of IGF-binding protein-3 in Hs578T
(Oxf). 2005;49(49):295–6. human breast cancer cells. Cell surface binding and growth
44. Pettersen HS, Sundheim O, Gilljam KM, et al. Uracil-DNA inhibition. J Biol Chem. 1993;268(20):14964–71.
glycosylases SMUG1 and UNG2 coordinate the initial steps of 65. Rajah R, Valentinis B, Cohen P. Insulin-like growth factor (IGF)-
base excision repair by distinct mechanisms. Nucleic Acids Res. binding protein-3 induces apoptosis and mediates the effects of
2007;35(12):3879–92. transforming growth factor-beta1 on programmed cell death
45. Seki S, Kitada T, Yamada T, et al. In situ detection of lipid through a p53- and IGF-independent mechanism. J Biol Chem.
peroxidation and oxidative DNA damage in non-alcoholic fatty 1997;272(18):12181–8.
liver diseases. J Hepatol. 2002;37(1):56–62. 66. Zhang K, Kaufman RJ. From endoplasmic-reticulum stress to the
46. Khetani SR, Szulgit G, Del Rio JA, et al. Exploring interactions inflammatory response. Nature. 2008;454(7203):455–62.
between rat hepatocytes and nonparenchymal cells using gene 67. Hirsch C, Gauss R, Horn SC, et al. The ubiquitylation
expression profiling. Hepatology. 2004;40(3):545–54. machinery of the endoplasmic reticulum. Nature. 2009;458
47. Zhang Z, Li XJ, Liu Y, et al. Recombinant human decorin inhibits (7237):453–60.
cell proliferation and downregulates TGF-beta1 production in 68. Wei Y, Wang D, Pagliassotti MJ. Saturated fatty acid-mediated
hypertrophic scar fibroblasts. Burns. 2007;33(5):634–41. endoplasmic reticulum stress and apoptosis are augmented by
48. Kalamajski S, Aspberg A, Lindblom K, et al. Asporin competes trans-10, cis-12-conjugated linoleic acid in liver cells. Mol Cell
with decorin for collagen binding, binds calcium and promotes Biochem. 2007;303(1–2):105–13.
osteoblast collagen mineralization. Biochem J. 2009;423(1):53–9. 69. Puri P, Mirshahi F, Cheung O, et al. Activation and dysregulation
49. Chen AA, Khetani SR, Lee S, et al. Modulation of hepatocyte of the unfolded protein response in nonalcoholic fatty liver
phenotype in vitro via chemomechanical tuning of polyelectrolyte disease. Gastroenterology. 2008;134(2):568–76.
multilayers. Biomaterials. 2009;30(6):1113–20. 70. Ni M, Lee AS. ER chaperones in mammalian development and
50. Wang Y, Lu C, Wei H, et al. Hepatopoietin interacts directly with human diseases. FEBS Lett. 2007;581(19):3641–51.
COP9 signalosome and regulates AP-1 activity. FEBS Lett. 71. Zhang L, Lai E, Teodoro T, et al. GRP78, but Not Protein-
2004;572(1–3):85–91. disulfide isomerase, partially reverses hyperglycemia-induced
51. Pearce C, Hayden RE, Bunce CM, et al. Analysis of the role of inhibition of insulin synthesis and secretion in pancreatic {beta}-
COP9 signalosome (CSN) subunits in K562; the first link between cells. J Biol Chem. 2009;284(8):5289–98.
CSN and autophagy. BMC Cell Biol. 2009;10:31. 72. Kammoun HL, Chabanon H, Hainault I, et al. GRP78 expression
52. Kato JY, Yoneda-Kato N. Mammalian COP9 signalosome. Genes inhibits insulin and ER stress-induced SREBP-1c activation and
Cells. 2009;14(11):1209–25. reduces hepatic steatosis in mice. J Clin Invest. 2009;119
53. Laplante JM, O'Rourke F, Lu X, et al. Cloning of human Ca2+ (5):1201–15.
homoeostasis endoplasmic reticulum protein (CHERP): regulated 73. Valero R, Marfany G, Gonzalez-Angulo O, et al. USP25, a novel
expression of antisense cDNA depletes CHERP, inhibits intracel- gene encoding a deubiquitinating enzyme, is located in the gene-
lular Ca2+ mobilization and decreases cell proliferation. Biochem poor region 21q11.2. Genomics. 1999;62(3):395–405.
J. 2000;348(Pt 1):189–99. 74. Valero R, Bayes M, Francisca Sanchez-Font M, et al. Character-
54. Vallee RB, Williams JC, Varma D, et al. Dynein: an ancient motor ization of alternatively spliced products and tissue-specific iso-
protein involved in multiple modes of transport. J Neurobiol. forms of USP28 and USP25. Genome Biol. 2001;2(10):
2004;58(2):189–200. RESEARCH0043.
OBES SURG (2010) 20:1698–1709 1709

75. Carlson MR, Zhang B, Fang Z, et al. Gene connectivity, function, molecular target. Proc Natl Acad Sci U S A. 2006;103
and sequence conservation: predictions from modular yeast co- (46):17402–7.
expression networks. BMC Genomics. 2006;7:40. 78. Saris CG, Horvath S, van Vught PW, et al. Weighted gene co-
76. MacLennan NK, Dong J, Aten JE, et al. Weighted gene co- expression network analysis of the peripheral blood from
expression network analysis identifies biomarkers in glycerol amyotrophic lateral sclerosis patients. BMC Genomics.
kinase deficient mice. Mol Genet Metab. 2009;98(1–2):203– 2009;10:405.
14. 79. Sharma MR, Polavarapu R, Roseman D, et al. Transcriptional
77. Horvath S, Zhang B, Carlson M, et al. Analysis of oncogenic networks in a rat model for nonalcoholic fatty liver disease: a
signaling networks in glioblastoma identifies ASPM as a microarray analysis. Exp Mol Pathol. 2006;81(3):202–10.