Download University of Groningen Metabolic adaptations in models of

University of Groningen
Metabolic adaptations in models of fatty liver disease
Hijmans, Brenda
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to
cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2017
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Hijmans, B. (2017). Metabolic adaptations in models of fatty liver disease: Of mice and math [Groningen]:
Rijksuniversiteit Groningen
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 15-06-2017
Chapter 6
Zonation of glucose and fatty acid metabolism: mechanism and
metabolic consequence
Brenda S. Hijmans, Aldo Grefhorst, Maaike H. Oosterveer and Albert K. Groen
Adapted from: Biochimie (2014): 96, 121-129.
103
104
Chapter 6
ABSTRACT
The liver is generally considered a relatively homogeneous organ containing
four different cell types. It is however well-known that the liver is not homogeneous and consists of clearly demarcated metabolic zones. Hepatocytes from
different zones show phenotypical heterogeneity in metabolic features, leading
to zonation of metabolic processes across the liver acinus. Zonation of processes
involved in glucose and fatty acid metabolism is rather flexible and therefore
prone to change under (patho)physiological conditions.
Hepatic zonation appears to play an important role in the segregation of
different metabolic pathways in the liver. As a consequence, perturbations in
metabolic zonation may be a part of metabolic liver disease. The metabolic syndrome is characterized by the inability of insulin to adequately suppress hepatic
gluconeogenesis, leading to hyperglycemia, hyperinsulinemia and eventually to
type II diabetes. As insulin promotes lipogenesis through the transcription
factor sterol regulatory element binding protein (SREBP)-1c, one would expect that lipogenesis should also be impaired in insulin-resistant states. In the
metabolic syndrome however, hepatic de novo lipogenesis is increased, leading
to hyperlipidemia and hepatosteatosis, primarily in the pericentral zone. These
observations suggest the co-existence of insulin resistant glucose metabolism
and insulin sensitive lipid metabolism in the metabolic syndrome. Here we provide a theoretical framework to explain this so-called insulin signaling paradox
in the context of metabolic zonation of the liver.
Hepatic zonation and insulin resistance
105
INTRODUCTION
Due to its rather uniform histological appearance the liver is often unjustly regarded as a homogeneous organ. Already in 1856, when describing the anatomy
of the liver in great detail, Lionel Beale noticed a heterogeneity of hepatocytes
with regard to bile secretion and deposition of oil [237]. After this, it took more
than a century until Jungermann and Sasse proposed a functional significance
of heterogeneous enzyme distribution in the liver, thereby introducing the concept of metabolic zonation [238].
It was recognized that the liver may be divided into functional units which
were designated acini. Inside the acinus blood flow is directional on basis of
which the acinus may be subdivided into different zones [239, 240]. The periportal zone receives nutrient rich blood from the portal vein and blood rich in
oxygen from the hepatic artery. In the pericentral zone blood is drained from
the liver by the central vein. Hepatocytes lining the sinusoids can be classified
according to their location on the portocentral axis of the acinus (see Figure
6.1). Three different zones are distinguished: 1 = periportal, 2 = intermediate, 3 = pericentral). The flow of blood through the liver generates gradients
of oxygen tension, hormones and nutrients which causes hepatocytes to be exposed to different metabolic conditions depending on their location along the
portocentral axis. Gradients in signals arise through interaction of blood borne
components with hepatocytes, such that periportal hepatocytes are exposed to
higher concentrations of blood borne components than pericentral hepatocytes
[241].
Hepatocytes from different zones of the liver show phenotypical heterogeneity in metabolic features, leading to zonation of metabolic processes across the
acinus [242]. Regarding glucose and fatty acid metabolism, periportal hepatocytes are more involved in gluconeogenesis and β-oxidation, while pericentral
hepatocytes are more engaged in glycolysis and lipogenesis [243, 244]. Other
metabolic properties of the liver, such as ammonia metabolism, xenobiotic reactions, cytoprotective functions and protein metabolism also show zonation
along the portocentral axis [241].
Zonation of glucose [244] and fatty acid metabolism [245] shows flexibi-lity
in different conditions in such a manner that dynamic adaptation of gene and
protein expression can be observed in different nutritional states [246]. Interestingly, metabolic pathways performing opposing functions seem to follow
inverse gradients and are distributed in a complementary manner. Interdependent metabolic pathways (e.g. lipogenesis and glycolysis) are co-localized
106
Chapter 6
Figure 6.1: Schematic depiction of the hepatic acinus.
to allow for synergistic action, whereas opposing pathways are segregated in
different zones, likely to avoid interference and thereby waste of energy. Altogether, the heterogeneity of the liver enables the simultaneous performance
of different and even opposing metabolic pathways while allowing for flexible
adaptation to differing circumstances [241]. This chapter will first briefly discuss the concept and mechanisms of hepatic metabolic zonation, with a special
focus on zonation of glucose and fatty acid metabolism. Second, the physiological consequences of segregation of metabolic pathways in the liver will be
discussed via examples of disturbed zonation in metabolic diseases of the liver.
Finally we will provide a theoretical framework that may explain the relevance
of hepatic zonation to aberrant metabolic states, such as insulin resistance and
non-alcoholic fatty liver disease (NAFLD).
Hepatic zonation and insulin resistance
107
METABOLIC ZONATION
Signals underlying hepatic metabolic zonation
Many different signals are involved in the establishment of zonal heterogenic
properties of hepatocytes. First of all, liver innervation seems to play a substantial role (e.g. [247]). Sympathetic and parasympathetic nerve fibers enter
the liver near the hepatic artery and the portal vein. The extent of further
innervation of the acinus differs between species. In rats and mice innervation
is limited to the periportal zone, while in humans also areas surrounding the
central vein are innervated [246]. Furthermore, the directional blood flow inside the acini establishes oxygen, nutrient, nutrient intermediate and hormonal
gradients along the portocentral axis [241, 246]. In the periportal zone oxygen
tension and concentrations of hormones and nutrients are higher. Hepatocytes
in the pericentral zone are exposed to blood enriched in CO2 and other products of metabolism.
While as of now it is not entirely clear how gradients of enzymes along
the portocentral axis arise, blood borne humoral factors (e.g. oxygen and hormones) are thought to be mainly involved in the dynamic adaptation of their
heterogeneous expression [248] as will be discussed below. This dynamic adaptation is mostly seen in enzymes that show a gradient-like distribution, such
as those involved in glucose [244] and fatty acid [245] metabolism. Other enzymes, such as carbamoyl phosphate synthetase I (CPS I) [249] and glutamine
synthetase (GS) [250] the key enzymes in ammonia metabolism, show a more
compartmentalized distribution. Recent studies have implicated different signaling pathways and molecules in the establishment of hepatic zonation. The
following sections will provide a brief description of several of these pathways.
Wnt/β-catenin signaling
The involvement of β-catenin was suggested by the finding that liver tumours
that contain mutations in activation of β-catenin express a pericentral-like transcription pattern [251, 252]. Immunohistochemical studies in mice have shown
that adenomatous polyposis coli (APC), which is part of the β-catenin degradation complex, shows a heterogeneous distribution along the portocentral axis
with a higher expression in periportal hepatocytes. The active unphosphorylated form of β-catenin was increased in hepatocytes surrounding the central
vein [253, 254]. Thus, there seems to be a mutually exclusive localization of
active β-catenin and one of its negative regulators across the acinus.
By modulating the Wnt signaling pathway it was shown that Wnt/β-catenin
108
Chapter 6
promotes the expression of pericentral genes while downregulating periportal
gene expression. When APC expression is downregulated, β-catenin is activated
and an induction of normally exclusive pericentral localized genes can be observed in periportal hepatocytes. On the other hand, specific blockade of Wnt
signaling induces periportal gene expression in pericentral hepatocytes [253].
Finally, pharmacological inhibition of glycogen synthase kinase 3β, a component of the β-catenin degradation complex, leads to suppression of periportal
gene expression and activation of pericentral gene expression in a resident liver
stem cell line [255].
Ha-RAS/MAPK signaling
While β-catenin signaling imposes a pericentral pattern of gene expression, the
periportal pattern is likely regulated by the Ha-RAS pathway. The first implications for involvement of Ha-RAS signaling in metabolic zonation were derived
from the observation that mutated Ha-RAS liver tumours display a periportal
pattern of gene expression [256]. Also, a portal to central gradient was found
for the phosphorylated form of extracellular signal-regulated kinase (ERK), a
downstream target of RAS. A discrepancy in this finding is however described:
in female rats phosphorylated ERK (p-ERK) levels are higher in pericentral
areas, especially during the pro-oestrous phase [257]. It may therefore be possible that the difference in zonation of pERK between females and males is
dependent on female sex-hormone levels.
Activation of the RAS/MAPK pathway induces ERK expression in cells
that do not express GS or other pericentral genes [252, 256]. Furthermore,
induction of the RAS/MAPK pathway by serum components also suppressed
GS expression [258]. Combined, these studies indicate that Ha-RAS favours a
periportal pattern of gene expression, while it abolishes a pericentral pattern.
HNF4α
Hepatic nuclear factor 4α (HNF4α) is a transcription factor that is highly expressed in liver, kidney, intestine and pancreas. HNF4α can bind to the promoter region of 12% of the genes expressed in the liver [259]. A rather homogeneous expression of this transcription factor was found throughout the
hepatic lobule, with slightly more transcripts in pericentral hepatocytes [260].
Liver-specific ablation of HNF4α induces GS expression in periportal hepatocytes while weakening the expression of periportal PEPCK [261]. This indicates
that HNF4α stimulates the expression of periportal enzymes while inhibiting
expression of pericentral enzymes. Interestingly, recent findings support a con-
Hepatic zonation and insulin resistance
109
vergence of the Wnt/β-catenin signaling and HNF4α in control of metabolic
zonation. Using co-immunoprecipitation HNF4α was shown to interact with
lymphoid enhancer binding factor-1 (LEF1), a Wnt target [255].
MicroRNAs
MicroRNAs (miRNAs) are a group of small noncoding RNAs of approximately
18-25 nucleotides long. In animals, miRNAs repress translation of their target messenger RNAs (mRNAs) through partial complementary binding to the
coding region or the 3'untranslated region (3'UTR) [262]. In addition to this,
miRNAs can direct cleavage of mRNAs, thereby reducing their amount [263].
Thus, miRNAs regulate gene expression at the posttranscriptional level, inhibiting mRNA translation into proteins. It is becoming increasingly clear that these
molecules play a prominent role in regulation of gene expression. One study
has examined a general role of miRNAs in metabolic zonation by knocking out
Dicer1. This gene encodes for the protein Dicer, which cleaves pre-miRNAs into
mature, functioning miRNAs. Loss of Dicer and thus a loss of miRNAs resulted
in impaired localization of the examined periportal proteins. Zonal expression
of pericentral proteins was less affected, with only expression of some proteins
extending beyond the normal boundaries while remaining mainly pericentrally
located [264].
Interestingly, ablation of either β-catenin or Dicer induced a similar phenotype. Both models show a loss of heterogeneous localization of periportally
expressed proteins, and a rather diffuse expression pattern throughout the liver
acinus [253, 255, 264]. This suggests that both β-catenin and Dicer are required
for normal localization of periportal enzyme expression. Conserved expression
of pericentral gene expression indicates that Wnt/β-catenin signaling is intact
in Dicer-deficient livers. Furthermore, it was found that Dicer expression is
not affected by β-catenin deficiency [265]. Microarray analysis did not reveal
individual miRNAs that are directly activated by β-catenin signaling but identified four miRNAs that are modestly upregulated [264]. Therefore it seems
that β-catenin in concert with yet to be specified microRNAs impact another
factor which in turn inhibits periportal gene expression.
Although loss of Dicer expression indicates a role for miRNAs in zonation,
it is far from clear which specific miRNAs are involved. Presently, over 1000
miRNA have been found in humans, which are predicted to regulate the activity
of more than 60% of all protein-coding genes. This suggests that miRNAs play
a role in most, if not all, (patho)physiological processes. However, bioinformatic predictions and experimental approaches indicate that a single miRNA
110
Chapter 6
may target more than a hundred mRNAs [265]. Given the current popularity
of this area of research our insight into the mechanisms of miRNA action is
expected to significantly increase within the coming years.
In summary, many signals have been implicated in zonation of protein expression alongside the portocentral axis. In this respect, the Wnt/β-catenin signaling and its antagonist, the Ha-RAS signaling pathway appear to have a major
contribution. It is however not clear which factors induce the heterogeneity
in these signaling pathways and how the different signals interact. MiRNAs
also play a role, but the mechanistic basis of their actions remains to be established. Although our insights into the mechanisms of metabolic zonation
are still limited, it is important to evaluate the zonation of metabolic processes
under different conditions. Before discussing this issue, further details about
zonation of glucose and fatty acid metabolism will be outlined.
Zonation of glucose metabolism
The liver plays a central role in the maintenance of glucose homeostasis. During
the fed state it takes up excess amounts of glucose and converts it to glycogen
and triglycerides (TG) while oxidizing it to CO2 . To maintain normoglycemia
during fasting states, the liver releases glucose derived from glycogenolysis and
gluconeogenesis into the bloodstream [266].
The bulk of research on acinar zonation of enzymes involved in glucose and
fatty acid metabolism dates back to the eighties and nineties of the last century.
Partitioning of different processes involved in glucose metabolism along the liver
acinus constituted the basis for the concept of metabolic zonation [241, 246].
According to the concept of metabolic zonation, gluconeogenesis mainly
takes place in the periportal zone, while glycolysis is performed in pericentral
hepatocytes. Many studies have reported increased gluconeogenic gene and
protein expression in periportal hepatocytes, whereas glycolytic enzymes show
higher expression in pericentral hepatocytes (for reviews, see: [267, 268, 269]).
More importantly the rate of glucose formation from gluconeogenic precursors
was found to be higher in cultured periportal hepatocytes isolated from fasted
animals after collagenase-digitonin perfusion [270, 271]. Glycogen synthesis
via the indirect pathway was also found to be higher in periportal cells [270],
showing that gluconeogenic flux is increased in periportal zones. While glucokinase activity is higher in pericentral hepatocytes, no zonal differences in
glycolytic flux were found [270, 271]. These in vitro studies were however not
performed under physiological oxygen tensions and did not account for the dif-
Hepatic zonation and insulin resistance
111
ferent oxygen tensions and hormonal gradients along the portocentral axis. It
has been shown that glucose metabolism is significantly modified upon oxygen
supply, and that glucose conversion to lactate is higher under pericentral-like
oxygen tension and hormonal concentrations [272]. Furthermore, it was shown
that perfusion of liver with glucose promotes glycogen deposition in pericentral zones while perfusion with lactate induces glycogen formation in periportal
hepatocytes [273]. This suggests that glucokinase flux is higher in pericentral
hepatocytes. It has recently been shown that glucose flux is required for nuclear
localization of β-catenin in enteroendocrine cells [274]. A higher glucose flux
through glucokinase in pericentral hepatocytes may therefore provoke a portocentral gradient of active β-catenin in liver.
Glycogen synthesis and utilization are also distributed in a zonated manner
across the acinus. Glycogen degradation is initiated in the periportal zone and
ends in the pericentral zone while glycogen stores are replenished in the opposite direction [275].
Zonation of fatty acid metabolism
Evidence is abundant that fatty acid oxidation takes place at higher rates in
periportal hepatocytes, while pericentral hepatocytes are more engaged in lipogenesis. It was shown that the rate of fatty acid synthesis is higher in pericentral
hepatocytes, while the rate of β-oxidation is higher in periportal cells [276, 277].
In microdissected rat liver tissue, Katz and colleagues reported higher activities
and higher presence of the lipogenic enzymes acetyl-CoA carboxylase (ACC)
and ATP citrate lyase (ACL) in pericentral areas. Furthermore, the activity
of carnitine palmitoyltransferase-1 (CPT1), a key enzyme in β-oxidation, was
found to be higher in periportal hepatocytes [278]. CPT-1 also shows a lower
sensitivity to inhibition by the lipogenic intermediate malonyl-coA in periportal
as compared to pericentral areas [243, 279]. Esterification of fatty acids and synthesis of very low density lipoprotein (VLDL) is slightly higher in hepatocytes
derived from pericentral areas. This uneven distribution has been suggested to
avoid TG accumulation due to increased lipogenesis in this part of the liver.
However, the same study found no differences in TG excretion rate for hepatocyte fractions derived from either periportal or pericentral zones of the liver
[243].
There are indications for higher fatty acid uptake in periportal hepatocytes.
Immunohistochemical staining shows that liver fatty acid binding protein (LFABP) is present in a portocentral gradient, declining in its expression from
periportal to pericentral in rodents [280] and in humans [281]. L-FABP is a
112
Chapter 6
cytoplasmic lipid binding protein involved in intracellular lipid transport. Its
heterogeneous distribution indicates a higher uptake and utilization of fatty
acids in periportal liver cells. However, digitonin/collagenase isolated periportal
and pericentral cells show no differences in radiolabeled oleic acid incorporation
into TG and phospholipids and no changes in lipogenic rates and water-soluble
oxidation products [282]. This does however not exclude the possibility that
in an intact organism, where a complex interplay of gradients of hormones and
substrates exists, a differential uptake of fatty acids in periportal versus pericentral hepatocytes may occur. Indeed, upon perfusion with a fluorescent-labeled
stearic acid a portal to central decline in fluorescence could be observed in isolated liver, indicating higher uptake of fatty acids in periportal as compared
to pericentral hepatocytes. Interestingly, this pattern was reversed upon retrograde perfusion, indicating the presence of a 'first pass'-effect [282, 283].
Flexibility of metabolic zonation
It is intriguing to realize that the localization of enzymes involved in fatty acid
and glucose metabolism shows flexibility. Some enzymatic distribution gradients partly overlap and may switch according to physiological needs [279]. The
zonal distribution of enzymes involved in glucose and fatty acid metabolism
may vary under different (patho)physiological circumstances. For instance, in
alloxan-induced diabetes in rats, zonal heterogeneity of gluconeogenic and glycolytic enzymes is perturbed. This diabetic state is characterized by a zonation
pattern resembling that of fasted rats [284]. Similarly, streptozotocin-induced
diabetic ketoacidosis was shown to increase hepatic glucose output and ketone
production in perfused rat liver. The same study also applied different degrees of digitonin-induced pericentral hepatocyte damage to demonstrate that
diabetic ketoacidosis led to an increase in gluconeogenesis in pericentral hepatocytes specifically [285].
Disturbances in zonation of hepatic fatty acid metabolism were also observed in a rat model for alcoholic fatty liver disease (AFLD) [286]. Ethanol
feeding caused a selective deposition of TG in pericentral hepatocytes, a pattern
also seen in humans. In control animals, lipogenic rate and ACC activity were
found to be higher in pericentral areas, whereas this heterogeneity was blunted
after ethanol feeding. The ethanol group exhibited a higher rate of β-oxidation
in pericentral hepatocytes, while this process was more abundant in periportal
areas of control animals. Furthermore, in this model VLDL secretion and TG
incorporation into VLDL were inhibited in pericentral hepatocytes. Combined,
these data show that AFLD is characterized by disturbed zonal heterogene-
Hepatic zonation and insulin resistance
113
ity of hepatic lipid metabolism. The outcome of this study also suggests that
a reduction in VLDL-TG excretion may be the primary cause of pericentral
steatosis in AFLD.
The collected evidence for altered heterogeneity in metabolic enzyme expression under pathophysiological conditions leads to the question whether
dysregulation of metabolic zonation is involved in or may even induce liver
pathology. AFLD and NAFLD are both characterized by disturbances in fatty
acid metabolism and show similar histological liver pathology. It is therefore
tempting to speculate that, as is the case for AFLD, also in NAFLD metabolic
zonation will be disturbed.
NON-ALCOHOLIC FATTY LIVER DISEASE
In NAFLD TG accumulate in hepatocytes due to imbalances in hepatic fatty
acid metabolism [11]. The precise mechanisms that lead to steatosis in NAFLD
have not been elucidated. The deposition of TG in the liver most likely has
different underlying origins that may vary between individuals. Hepatic TG
accumulation may for instance either result from an increased fatty acid influx,
a decreased fatty acid efflux, or a combination of the two. There are different
ways in which TG may reach the liver. Dietary TG that are transported via
chylomicrons from the intestine may be delivered directly to the liver, or may
first reach adipose tissue and reach the liver in the form of FFA after lipolysis.
Fatty acids for esterification inside the liver are provided through either the
plasma FFA pool or via de novo lipogenesis (DNL). Decreased efflux of hepatic
TG may be due to a decrease in β-oxidation or a decrease in transport of lipids
out of the liver in VLDL particles [11].
A dynamic balance of the above mentioned pathways is needed to maintain
homeostasis of hepatic TG content. To adequately evaluate the basis of steatosis in NAFLD, all pathways involved in influx and efflux of hepatic TG should
be studied in parallel. Donnelly et al. quantified fatty acid metabolism and the
origin of hepatic TG in NAFLD patients using 13 C-labeled isotope enrichment
of free fatty acids (FFA), TG and a fatty acid precursor in the hepatic TG pool.
They found that in NAFLD patients, postprandially, hepatic and VLDL TG
are for the largest part (59%) derived from circulating FFA. The contributions
of DNL and dietary fatty acids were found to be smaller, 26% and 15%, respectively. During fasting about 80% of hepatic and VLDL TG were derived from
the plasma compartment, suggesting that they are mainly derived from adipose
114
Chapter 6
tissue lipolysis [119]. While healthy subjects show an increase in fractional DNL
from 5% in the fasted to 23% in the postprandial state [287], NAFLD patients
[119] and hyperinsulinemic obese subjects [288] did not shown an elevation of
DNL in response to feeding. Instead, in NAFLD and hyperinsulinemic obese
subjects, DNL was already elevated in the fasted state and contributed for up
to 25% to hepatic and VLDL TG in both fasted and postprandial conditions
[119, 288]. Regarding the efflux of hepatic TG, in NAFLD patients generally an
induction of genes involved in β-oxidation is reported [289, 290]. In support of
this, higher levels of circulating ketone bodies in these subjects are indicative of
increased hepatic β-oxidation [291]. The excretion of TG-rich VLDL particles
was furthermore elevated [292]. Although VLDL-TG secretion increased with
higher intrahepatic lipid content, it reached a plateau at higher TG concentrations, suggesting that TG secretion in VLDL somehow fails to increase further
above a certain threshold [292]. Therefore it seems that plasma derived FFA
are the main contributor to hepatic TG in NAFLD, suggesting that the main
contributor of hepatic steatosis in the metabolic syndrome is increased adipose
tissue lipolysis.
Zonation of steatosis
Deposition of TG in NAFLD is often found to be distributed in a zonated manner. In humans, steatosis associated with AFLD, and many forms of NAFLD is
described to initiate in pericentral areas, with advancement to intermediate and
periportal areas upon disease progression [293, 294]. As stated earlier, under
normal physiological circumstances periportal zones show higher rates of fatty
acid oxidation while lipogenesis is more pronounced in pericentral zones. In
addition, changes in this segregation of fatty acid metabolism have been observed under aberrant metabolic conditions. Unfortunately, not many studies
have evaluated metabolic zonation in NAFLD. Some studies have found increased lipid peroxidation and oxidative damage in pericentral areas in liver
biopsies from human non-alcoholic steatohepatitis (NASH) patients [295, 296].
This indicates a zonation of mitochondrial dysfunction to pericentral areas in
steatohepatitis. Enhanced lipid peroxidation and oxidative stress are known
to occur at lower oxygen tensions [297], which may account for predominant
pericentral hepatocyte damage due to obesity or ethanol use.
While in adults steatosis is most often localized in pericentral zones, this
is not the case for paediatric steatosis. Some studies found paediatric steatosis
to be typically azonal or to have a homogeneous distribution [298, 299], while
others report a predominant periportal pattern [298]. In malnutrition disorders,
Hepatic zonation and insulin resistance
115
infections and toxic insults, steatosis seems to originate in the periportal zone
[300]. These differences indicate that under different pathophysiological circumstances zonation of steatosis may vary. More insight is needed into specific
zonal alterations that occur in the liver acinus under these different conditions.
Whether zone-dependent initiation and progression of steatosis and development into NASH exists is also not known. Further insight into disease progression is needed to reveal by what mechanisms TG accumulation progresses, and
how steatosis may develop into NASH.
Zonation of steatosis in animal models of NAFLD
Genetic or diet-induced animal models are commonly used to examine pathophysiology of hepatosteatosis. Unfortunately not many studies report on zonal
distribution of TG deposition. Those models in which a zonal distribution of
steatosis has been described will be discussed here.
Genetic models
A distinction can be made between spontaneous and induced genetic models
of NAFLD. The two most often used models are spontaneous genetic models:
ob/ob and db/db mice. Ob/ob mice exhibit a naturally occurring mutation
that results in the absence of the satiety hormone leptin that inhibits feeding
behaviour and stimulates energy expenditure [301]. These animals display hyperphagia and obesity. Db/db mice possess a genetic defect that leads to a
similar phenotype. These animals do not have functioning receptors for leptin,
resulting in a loss of leptin signaling (e.g. [302]). Both genetic models develop
severe obesity, accompanied by a reduction in insulin sensitivity, abnormalities
in (hepatic) fatty acid metabolism and hepatic steatosis [303].
Hepatosteatosis is located in pericentral hepatocytes in ob/ob mice [304]. To
our knowledge, it has not been examined whether lipogenesis or other aspects
of fatty acid metabolism are deregulated in a zonal manner in these animals.
To date, the localization of TG accumulation in db/db mice has not been reported. However, histological pictures of the liver reveal a clearly zonated fat
accumulation [305].
Diet-induced models
C57BL/6 mice fed a high fat diet (HFD) are often used as a model to study the
metabolic syndrome. On a normal rodent diet, this mouse strain spontaneously
develops obesity, hyperinsulinemia and glucose intolerance with age, which can
be induced earlier in life by feeding a HFD [306]. In addition, C57BL/6 mice
116
Chapter 6
develop hepatosteatosis on a HFD.
Not many studies report on zonation of diet-induced steatosis. Some recent articles report predominant fat accumulation in pericentral hepatocytes in
response to a HFD in mice [307, 308, 309]. However, in another study high
fat feeding was shown to predominantly lead to deposition of TG in hepatocytes in areas surrounding the portal vein [310]. Overfeeding mice a high fat
liquefied diet was shown to cause steatosis throughout the whole liver acinus.
Interestingly, this led to upregulation of lipogenic genes in pericentral hepatocytes [310]. It should be noted that the diets used in these studies differ in fat
and carbohydrate content. Because glucose and fructose are metabolized in a
different manner, the source of carbohydrates may also be of importance [311].
This content difference may account for the difference in histological findings,
particularly because the diet used in the study of Gaemers and colleagues had
a relatively high carbohydrate content (47%). It has been shown that different
dietary fat types may have different metabolic consequences, leading to differences in for instance liver TG content [312]. Diets high in carbohydrates seem
to induce periportal steatosis. In Sprague-Dawley rats high dietary concentrations of sucrose (40-50%) lead to periportal fat accumulation [313]. Mice fed
a high fat/high fructose diet initially also show TG accumulation in periportal
hepatocytes. In this model, small fat droplet accumulation in pericentral zones
was noticed after eight weeks of feeding [314]. There is no published research
on the basis of the difference in zonation of steatosis that occurs when feeding
different diets. A sucrose or fructose-rich diet would lead to an increased flux
of fructose to the liver via the portal vein. The loss of fructose in the urine in
persons that have an inherited deficiency of fructokinase [315] shows that fructose is not well metabolized in extrahepatic tissues. Furthermore, the isoform
fructokinase-C, which is considered the most active in metabolizing fructose
[316] is only expressed in liver, intestine and kidney [317]. A high availability of fructose will lead to rapidly decreasing hepatic intracellular ATP levels
and a decline in free phosphate due to the low Km of fructokinase for fructose
and the absence of negative feedback by energy status on its metabolism [318].
Therefore, triose-phosphate derivatives of fructose will be metabolized rapidly
by glycolysis or will enter the gluconeogenic pathway. It has been shown in humans that prolonged fructose, but not glucose consumption leads to increased
postprandial hepatic DNL and elevated postprandial plasma TG [319]. It could
be argued that a fructose rich diet will primarily result in periportal zonation
of steatosis, since in rats fructose is taken up mainly by periportal hepatocytes
[320]. On the other hand, as discussed in more detail in a subsequent section,
Hepatic zonation and insulin resistance
117
a high fat diet may lead to a portocentral gradient in insulin sensitivity, which
may account for pericentral accumulation of TG.
INSULIN RESISTANCE AND STEATOSIS
NAFLD is regarded as the hepatic component of the metabolic syndrome [2],
showing clear correlations with diabetes [321] and obesity [5]. Although a strong
correlation exists between insulin resistance and NAFLD, the causality behind
this association is uncertain. For instance, when hepatic steatosis is induced by
pharmacological inhibition of β-oxidation this does not lead to hepatic insulin
resistance [322]. In genetic diseases that promote hepatic FFA influx, such as
glycogen storage disease type 1, steatosis is also found without the presence of
insulin resistance [146]. Moreover, in liver insulin receptor knockout (LIRKO)
mice, which display complete hepatic insulin resistance, no hepatic steatosis
is found [323]. In liver X receptor (LXR) agonist treated mice severe hepatic
steatosis is induced without hepatic insulin resistance [324]. Finally, in mice
with a liver-specific deletion of histone deacetylase 3 (HDAC3), a fatty liver
is even accompanied by an increase in insulin sensitivity [309]. These findings
show that hepatic fat deposition per se does not lead to insulin resistance and
insulin resistance in itself does not produce steatosis. For further reading on this
subject the reader is referred to a recent review [325]. Since insulin resistance
and fatty liver are both aspects of the metabolic syndrome, it is important to
elucidate the mechanisms by which these symptoms are associated and may
influence each other. According to the concept of metabolic zonation, one can
even wonder whether it is relevant to correlate TG accumulation with the failure of insulin to inhibit gluconeogenesis because these processes are normally
localized in different hepatic zones. In NAFLD, pericentral hepatocytes more
often show TG accumulation [293, 294], while gluconeogenic activity is mainly
located in periportal hepatocytes [326].
The insulin signaling paradox
It has been suggested previously that in the metabolic syndrome the liver shows
a mixed pattern of insulin resistance and sensitivity [327, 328]. In persons suffering from the metabolic syndrome gluconeogenesis is not (sufficiently) suppressed
by insulin, leading to hyperinsulinemia, and eventually to hyperglycaemia and
type II diabetes. As insulin stimulates lipogenesis through SREBP-1c (reviewed
in [329]), one would expect that in the presence of hepatic insulin resistance,
118
Chapter 6
lipogenesis should also be impaired. However, a paradox exists with respect to
lipid metabolism: in type II diabetes DNL is increased, leading to hyperlipidemia and hepatosteatosis [288]. These findings suggest the presence of selective
insulin signaling in the metabolic syndrome, with insulin resistance of gluconeogenesis, while lipid metabolism appears to remain insulin sensitive. It has
been suggested that mixed insulin sensitivity may arise from different signaling
pathways for insulin to lipid and glucose metabolism. Insulin receptor substrate
(IRS) 1 was proposed to signal to lipid metabolism, while IRS 2 would signal
to glucose metabolism [330]. Recent studies have however pointed out that
IRS1 and IRS2 together signal to both hepatic glucose and lipid metabolism
under different dietary circumstances [331, 332]. Other manners of branching
of the insulin signaling pathway downstream of the insulin receptor have also
been suggested [333, 334, 335, 336], but do not (fully) explain mixed insulin
sensitivity and zonation of steatosis.
More recently, Sun and colleagues [309, 325] have proposed an interesting
theory which may explain the insulin signaling paradox from a biochemical
viewpoint. It is hypothesized that under pathological conditions of overnutrition the oversupply of metabolic intermediates is the driving force behind the
co-existence of gluconeogenesis and lipogenesis. In these conditions the flux of
FFA to the liver is thought to be too high for lipogenesis to keep up, thereby
leading to accumulation of lipid intermediates which may cause hepatic insulin
resistance [309]. As discussed before, gluconeogenesis mainly takes place in
periportal hepatocytes, while pericentral hepatocytes are more engaged in lipogenesis. Here, we propose an alternative theory which explains how hepatic
metabolic zonation may lead to selective insulin signaling.
The insulin signaling pathway that inhibits gluconeogenesis is well studied. Insulin binds to the insulin receptor present on cell membranes, leading
to a cascade of phosphorylation of downstream enzymes: IRS, phosphoinositide 3-kinase (PI3K) and Akt. Eventually phosphorylation of Forkhead Box
1 (Fox01) results in its nuclear exclusion, thereby reducing gene expression of
gluconeogenic enzymes (reviewed in [336]). The signaling pathway that leads
to increased SREBP-1c expression is however less well studied. Liver-specific
deletion of the insulin receptor in mice (LIRKO mice) leads to hyperinsulinemia
and hyperglycaemia with normal plasma and hepatic TG levels. These mice
show an increase in Srebp-1c expression [323], accompanied by an increase in
Pepck expression [337].
Thus, total hepatic insulin resistance leads to a phenotype that does not
match the metabolic syndrome. When downstream targets in the IRS/PI3K/Akt
Hepatic zonation and insulin resistance
119
pathway are inhibited, a similar phenotype is found: hyperglycaemia, with normal to low plasma lipid levels. For instance, mice deficient for the two major
insulin receptor substrates (IRS1 and IRS2) in the liver show hyperglycaemia,
but have normal plasma lipid values [331]. Moving even further downstream,
it has been shown that inhibition of PI3K or Akt leads to blocking of insulininduced increases in Srebp-1c and decreases in Pepck expression [333]. It is also
known that Akt drives lipogenesis in multiple ways. Activation of mTORC1
increases lipogenesis through induction of SREBP-1c [338]. Also, Akt mediated
inhibition of Insig2, which activity retains SREBP-1c in the endoplasmic reticulum, increases the active form of SREBP-1c [338]. Finally, transgenic mice that
express a constitutively active form of Fox01 in liver show hyperglycaemia, but
reduced plasma lipid levels [339]. Therefore it seems that insulin-induced upregulation of hepatic lipogenesis and downregulation of hepatic gluconeogenesis
may be regulated via a common signaling pathway. From these studies it may
be proposed that excessive insulin signaling rather than insulin resistance is
responsible for hepatosteatosis and dyslipidemia in type II diabetes.
Selective insulin sensitivity across the portocentral axis
How can glucose metabolism be insulin resistant and lipid metabolism insulin
sensitive if these metabolic pathways are regulated through a common arm of
the insulin signaling pathway? Here, we propose a model of zonation of insulin
sensitivity along the portocentral axis of the liver. Under normal circumstances
high circulating insulin levels suppress adipose tissue lipolysis in the postprandial state. In obese individuals, insulin resistance causes insufficient suppression
of hormone sensitive lipase (HSL), leading to excessive lipolysis in adipose tissue. It has indeed been reported that insulin does not suppress adipose tissue
lipolysis in NAFLD patients to the same extent as it does in healthy individuals
[291].
Metabolites of FFA, such as ceramides, may inhibit hepatocyte insulin signaling in two ways. First, ceramides interfere with the insulin signaling pathway by activating protein kinase C (PKC) ζ [340] and stabilizing interaction
between PKCζ and Akt, which blocks Akt translocation to the plasma membrane [341, 342]. Second, ceramides stimulate activity of protein phosphatase
2A (PP2A), which dephosphorylates and thereby deactivates Akt [343, 344].
Circulating FFA may be taken up predominantly by periportal hepatocytes,
perhaps due to a first-pass effect [282, 283]. The periportal accumulation of FFA
metabolites may cause insulin resistance in these hepatocytes, while pericentral hepatocytes may take up less FFA and may therefore remain relatively
120
Chapter 6
Figure 6.2: Proposed model for zonation of insulin signaling in liver.
PP = periportal, PC = pericentral. See text for details.
insulin sensitive. This will lead to a zonated maintenance of SREBP-1c mediated lipogenesis and may explain selective accumulation of TG in pericentral
hepatocytes.
We thus propose a model in which increased circulating FFA concentrations in obese subjects will predominantly impair insulin sensitivity in periportal hepatocytes. This will result in impaired insulin-mediated inhibition
of gluconeogenesis and glycogen synthesis and hence enhanced hepatic glucose
production by periportal hepatocytes. The increased blood glucose concentrations will promote pancreatic insulin release, hence inducing hyperinsulinemia.
The increased insulin levels will enhance pericentral insulin signaling resulting
in increased lipid synthesis in the pericentral area. This will lead to accumulation of TG in pericentral zones of the liver, accompanied by an increase in
the secretion of large, TG-rich VLDL particles, resulting in hyperlipidemia (see
Figure 6.2).
Hepatic zonation and insulin resistance
121
FINAL REMARKS
In human adults steatosis is in most cases localized in pericentral hepatocytes,
while in paediatric cases steatosis is more present in periportal zones of the
liver. Research in animal models of NAFLD has shown that under differing
circumstances steatosis may occur in different zones of the liver. Although it is
crucial for an animal model to closely reflect the human situation, one should
also acknowledge that there is a variety in the mechanisms that lead to steatosis in humans. By using these different NAFLD models, future work should
provide more detailed insights into associations between the location of hepatic
TG accumulation and possible changes in metabolic zonation.
Many zonated enzymes in the liver show flexibility of expression patterns under changing conditions. This shows the ability of hepatocytes to change their
repertoire to differing physiological conditions and metabolic demands. The
variety in zonation of fat deposition between steatosis models illustrates the
complexity and flexibility of metabolic zonation. Hepatocytes across the portocentral axis are not homogeneous, nor is TG accumulation. Therefore, zonespecific investigation of changes in fatty acid metabolism and insulin signaling
should generate more precise answers to the mechanisms underlying NAFLD.
New insights into regulation of metabolic zonation in the metabolic syndrome
may also lead to development of therapies that target zone-specific hepatocytes. In this respect, periportal-specific upregulation of insulin signaling may
be an interesting approach. It has been reported that adenovirus associated
vector-mediated modulation of gene expression is more effective in periportal
hepatocytes in non-human primates. This will however be a challenging approach, because an inverse pattern of induction is seen in mice [345].
In conclusion, elucidation of the mechanisms responsible for induction and
maintenance of hepatocyte heterogeneity remains a challenge in experimental
hepatology. It is however important to consider this heterogeneity of the liver,
because of possible influences of (aberrant) metabolic zonation on pathological
conditions.