From Millennium Pharmaceuticals, Cambridge, Massachusetts 02139
Of the many genetic obesity syndromes, none have
been as intensively studied as ob/ob and
db/db mice. These two mutant mice were originally
identified over 30 years ago (1, 2) and shown to be a result of two
distinct single gene mutations residing on mouse chromosomes 6 (ob) and 4 (db). The phenotypes and
pathophysiologies of these two mice have been studied for decades and
described in well over 1000 publications. However, the nature of the
lesions or primary defects was not revealed until very recently.
Perhaps the most informative early studies on the nature of the
ob and db primary defects were the parabiosis
experiments (partial connection of the circulatory systems of animals
through grafting) performed throughout the 1970s (3) (reviewed in Ref. 4). Parabiosis of an ob/ob mouse and a lean
control resulted in partial normalization of body weight in the
ob/ob mutant mouse. This led to the proposal that
ob/ob mice were deficient in a circulating factor
that could be restored through the blood of the lean animal. However,
db/db mice that underwent parabiosis with lean
controls did not exhibit body weight normalization. This suggested that db/db mice may be defective in their ability to
respond to the putative satiety factor, perhaps because they were
defective in the receptor for this molecule.
The Obese (ob) Gene and Its Product, Leptin Despite intensive interest in the nature of the putative satiety
factor missing in ob/ob mice, biochemical
strategies failed to identify it. It was not until a genetic/positional
cloning strategy was employed that the gene corresponding to the
ob locus and its gene product were ultimately identified
(5). The wild type ob gene encodes a protein of about 16 kDa
that is preceded by a secretory hydrophobic signal peptide. It is
expressed in adipose tissue in multiple mammalian species including
mice and humans. The development of antibody reagents confirmed that
this factor (leptin) is found at high levels in blood, consistent with the previous parabiosis studies (6).
Since the cloning of the ob gene numerous studies have
described the regulation of the leptin mRNA and protein. Although
the purpose of this review is not to comprehensively examine the
growing literature on the regulation of the leptin ligand, it is
important to briefly summarize a few aspects of leptin expression and
regulation that are key in interpreting the biology of the leptin
receptor. The leptin transcript appears to be expressed fairly
specifically in adipose tissue (5), although it is also detectable in
human placenta on poly(A)+ Northern
blots.1 Steady state levels of the leptin
mRNA and protein are elevated in a variety of rodent obesity models
(6-9). These observations have led to the proposal that leptin serves
as an "adipostat," informing the body of the status of energy
storage in the adipose tissue so that appropriate changes in appetite,
metabolism, and nutrient partitioning can be signaled via the leptin
receptor. Dramatic regulation of the leptin transcript and protein has
also been observed in response to short term alterations in food intake (7, 9-11); fasting results in dramatic down-regulation, and excessive
caloric intake results in up-regulation. It is therefore plausible that
an important role for leptin is mediating the response to starvation
(12). Acute effects on leptin mRNA and protein have also been
observed in response to a variety of stimuli including glucocorticoids, cytokines, and insulin (10, 13, 14).
There has now also been considerable analysis of leptin regulation in
humans. The leptin mRNA is regulated in humans by both changes in
the percentage of body fat as well as changes in food intake (6,
15-18). However, the degree of mRNA regulation in humans is less
impressive than that seen in rodents. Importantly, at the protein level
human leptin is dramatically regulated, with changes approaching those
seen in the rodent obesities and fasting (6, 15). The protein is much
higher in individuals with an increased percentage of body fat and is
down-regulated during body weight loss. This parallel regulation in
mice and humans may imply that leptin is functioning in humans as it is
in rodents, although further studies are required to directly address
this.
An obvious and important question has been whether a significant
portion of human obesity can be due to mutations in the ob gene. The ob coding region has been sequenced from hundreds
of human individuals, but mutations have not been found (16, 19, 20).
Although mutations affecting mRNA levels can reside outside of the
coding region, individuals with severely reduced leptin mRNA levels
have also not yet been described (15-18), suggesting that the number
of such individuals will not be high. On the other hand, genotyping of
microsatellite markers that span the ob gene region has
suggested linkage of this region with extreme human obesity (21,
22).
Considerable excitement has been generated by the observations that
administration of recombinant leptin to rodents results in food intake
reduction and weight loss (23-26). Although the potency of leptin is
highest in mice that are completely deficient in this protein
(ob/ob), significant effects can be seen at
higher doses in normal mice and mice with diet-induced obesity. Such studies have brought hope that leptin may be an effective treatment even in some obesities that are not due to leptin deficiencies. Of
particular interest are studies that have investigated the effects of
centrally administered leptin. These studies showed that leptin
injection into the lateral or third brain ventricle produced reduction
in food intake and weight loss, strongly implying that leptin could act
directly on receptors within the central nervous system (23, 26).
Weight loss in rodents following leptin administration appears to be
due to not only decreases in food intake but also increases in energy
expenditure (24, 25). Although the mechanisms of increased energy
utilization are likely to be complex, one important component may
involve the activation of brown adipose tissue (27).
The observation that leptin deficiencies are not common in human
obesity and, in fact, that leptin levels appear to be up-regulated as
the percentage of body fat increases has suggested that resistance to
normal or elevated levels of leptin may be more important than inadequate leptin production in human obesity (15). This line of
thought has been further strengthened by the parallel situation of type
II diabetics, many of whom exhibit severe insulin resistance while
producing elevated levels of insulin. These observations further
stimulated interest in the identification of the receptor for leptin
and the analysis of leptin signal reception. They also heightened
interest in the db/db mouse, a model of total
leptin resistance (23-26) and elevated leptin levels (8).
Cloning of the Leptin Receptor (OB-R) The identification of the receptor for the leptin protein was
realized through an expression cloning strategy (28). Tagged versions
of leptin were generated either through a traditional iodination
strategy or by generating recombinant fusion proteins between leptin
and secreted placental alkaline phosphatase. These tagged reagents were
then used to identify a tissue source expressing a cell surface leptin
binding activity (28, 29). Significant and specific leptin binding was
detected in the mouse choroid plexus. To clone this leptin binding
activity a murine choroid plexus cDNA library was constructed, and
cells transfected with this library were screened with a
leptin-alkaline phosphatase fusion protein. From this screen, cDNAs
were identified that encoded a cell surface leptin receptor (OB-R) with
an affinity for leptin of about 0.7 nM (28).
Sequencing of the original murine cDNA identified in the expression
cloning screen revealed a single membrane-spanning receptor of the
class I cytokine receptor family (28). This observation was consistent
with a previous structural prediction indicating that the leptin
protein would be expected to fold into a cytokine-like structure (30).
The closest relatives of OB-R were gp130 (31), the
G-CSF receptor (32), and the leukemia inhibitory factor receptor (33).
The predicted extracellular domain of OB-R was quite large, about 816 amino acids, while the predicted intracellular domain was fairly short,
34 amino acids, suggesting that this protein might not have
signal-transducing capability. However, further screening and analysis
of cDNA libraries using the original OB-R cDNA sequence as a
guide soon revealed that there were multiple forms of OB-R in both mice
and humans, including a long form with an intracellular domain of about
303 amino acids (28, 34-36) (Fig. 1). The intracellular
domain of the long form contained sequence motifs suggestive of
intracellular signal-transducing capabilities.
The extracellular domains of the short and long forms of OB-R are
identical throughout their entire length, since differences in the
receptor forms arise from alternative RNA splicing at the most
C-terminal coding exon, resulting in OB-R intracellular domains with
differing length and sequence composition (34, 35) (see below).
Additional short intracellular domain forms have now been identified,
all of which terminate shortly after the point of divergence
(34-36)3 (after amino acid 29 of the
intracellular domain). The most obvious grouping of OB-R forms is to
distinguish the long intracellular domain form (OB-RL), the
likely signaling form, from the growing list of short intracellular
domain forms (OB-RS1-n) (Fig. 1). A transcript potentially
encoding a soluble form of OB-R (lacking a transmembrane domain) has
also been described (35). The murine and human receptors are highly
similar in amino acid sequences of both the extracellular (78%
identity) and intracellular domains (71% identity) (28, 34).
The mRNA expression profile of OB-R must be considered with
care, due to the fact that there are multiple receptor forms encoded by
distinct transcripts. Northern, polymerase chain reaction, or in
situ analysis with probes generated from the extracellular domain
(common to all OB-R forms) shows expression in multiple tissues at
varying amounts (28, 35-37). In the mouse, the highest OB-R mRNA
levels are found in the choroid plexus, lung, and kidney, and somewhat
lower levels of expression are detected in nearly all tissues (28).
However, subsequent analysis has shown that the vast majority of
transcripts detected by these assays are transcripts encoding short
intracellular domain forms (OB-RS) (38).4 The mRNA species encoding the
long intracellular domain is much less abundant. Although this form can
be detected by RNase protection or polymerase chain reaction in both
mice and humans in nearly all tissues, in most tissues it is expressed
at very low levels (38).4 An exception to this is in the
hypothalamus (Fig. 1), where the OB-RL transcript is
expressed at much higher levels (38)4 and can be detected
by in situ hybridization (39). In fact, of the tissues so
far tested, only hypothalamus expresses more long form transcript than
the most predominant short form (OB-RS1).1
Within the hypothalamus, the long form transcript has been detected in
the arcuate, ventromedial, paraventricular, and dorsomedial nuclei
(39), regions previously thought to be important in body weight
regulation.
The Mouse Leptin Receptor Is Encoded by the Diabetes
(db) Gene After the cloning of the leptin receptor an important remaining
question was whether the gene encoding it corresponded to the
db locus, as had been predicted by the parabiosis studies of
decades ago. Genetic mapping of the gene encoding OB-R localized its
position to a narrow interval on mouse chromosome 4 (28). This position
was within the small genetic interval to which the db locus
had been mapped by genetic strategies. Sequencing of the gene encoding
the leptin receptor from normal and db mice revealed that a
mutation was, in fact, present in this gene in db/db mice (original allele) (34, 35). The
mutation is a single nucleotide substitution (G
It has also been shown that an OB-R defect is responsible for the
obesity phenotype of the fa rat (40-42). Genetic mapping studies in the rat had previously shown that the fa mutation
mapped to rat chromosome 5, a region syntenic with the db
region on mouse chromosome 4 (43). Sequencing of OB-R in the
fa rat revealed a single amino acid substitution in the
extracellular domain (41, 42). Although this mutation does not appear
to affect binding affinity of the receptor, it strongly affects cell
surface expression (42). Whether this mutation also affects signaling
potency is currently under study.
The obese phenotype of the Koletsky rat also appears to be due to a
defect in OB-R (44). In this case, a codon for an extracellular domain
amino acid (common to all OB-R isoforms) has been mutated to a stop
codon. This suggests that Koletsky rats are deficient in all cell
surface leptin receptors, and therefore these rats are likely to
represent the OB-R null phenotype.
Leptin Receptor Short Forms (OB-RS) The roles of the short intracellular domain forms of OB-R remain
to be defined. It is tempting to speculate that the high levels of the
short intracellular domain form in the choroid plexus play a role in
transporting leptin from the blood into the CSF, where it can then move
by diffusion to the brain centers that regulate body weight. It has
already been shown that leptin enters the brain by a specific and
saturable transport mechanism (45), although a critical role of
OB-RS in this process has not been demonstrated.
This saturable transport process may in fact be a critical feature of
leptin resistance. Differences between obese and lean individuals in
leptin levels are greater in blood than in CSF (46, 47). This suggests
that adipose-derived leptin levels in blood may not be properly
translated into rising CSF levels and therefore result in an apparent
leptin resistance. This form of leptin resistance would be quite
different from the intrinsic cellular resistance that is characteristic
of insulin resistance (48). Therefore, it is possible that the central
nervous system tissues of even obese individuals may be
leptin-responsive if only the elevated leptin levels were capable of
reaching these sites. This suggests that a leptin receptor agonist that
could more freely cross the blood-brain barrier may be important
therapeutically.
Short OB-R forms may play a role not only in transport but also in
clearance or as a source of soluble receptor (assuming proteolytic
mechanisms exist for releasing the extracellular domain from the cell
surface). The reason that several distinct short intracellular domain
forms of OB-R are produced is not clear. The different short forms
clearly have distinct tissue distributions (35). However, it seems
unlikely that there will be significant functional differences
demonstrated between the different short forms.
Leptin Receptor Long Form (OB-RL) The mechanism of leptin receptor triggering and signal
transduction is of obvious interest both for pharmaceutical
considerations as well as insight into the molecular mechanisms of body
weight homeostasis. The homology of OB-R to class I cytokine receptors immediately provided important clues as to possible intracellular mediators of leptin receptor activation. The class I cytokine receptors
are known to act through JAK and STAT proteins (49, 50). Typically, JAK
proteins are associated constitutively with membrane-proximal sequences
of the receptor intracellular domain (ICD) and phosphorylate the
receptor ICD upon ligand binding. The phosphorylated ICD then provides
a binding site for a STAT protein, which is activated upon binding the
phosphorylated receptor ICD. The activated STAT proteins then
translocate to the nucleus and stimulate transcription.
Transient co-transfection studies have revealed that the
OB-RL protein is capable of activating STAT proteins in
response to ligand binding (OB-RS is inactive in these same
assays) (38, 51). Two published reports each show that STAT3 and STAT5
are stimulated in COS cells by OB-R triggering but disagree on the activation of STAT1 and STAT6 (38, 51). Which STAT protein is most
important for OB-R regulation of body weight in vivo remains to be resolved. It is unlikely that either STAT1 or STAT6 is critically involved, since the corresponding gene knockouts have not been reported
to exhibit dramatic obesity (52-56). It is important to note that
although OB-R is capable of activating several known STAT proteins in
COS cells, the actual STAT proteins activated in vivo (and
their relative ratios) may differ considerably from that observed in
cell lines and may possibly include previously unidentified STAT
proteins. So far only STAT3 activation has been detected in mice when
recombinant leptin is exogenously administered (57). It is also
important to note that some class I cytokine receptors are able to
stimulate signal transduction cascades that are distinct from the
JAK/STAT pathway, such as the mitogen-activated protein and
phosphatidylinositol 3-kinase pathways (49, 50), and it is possible
that OB-R's ability to control body weight may depend upon these
signals as well.
Reconstitution experiments in hepatoma cells have shown that OB-R is
also able to stimulate transcription via response elements known to be
stimulated by members of the class I cytokine receptor family (51). As
observed for STAT protein activation, the major naturally occurring
mouse OB-R short form (also equivalent to the mutant long form in
db/db mice) has so far appeared incapable of
stimulating transcription (51).
In vitro, OB-R seems to have a signaling repertoire
similar to the IL-6 type cytokine receptors (51). Additionally, OB-R shares significant homology with both gp130 and the leukemia inhibitory factor receptor (which uses gp130 as a signal-transducing component). However, anti-gp130 antibodies (which are capable of blocking signaling
via IL-6 type cytokine receptors) do not block signaling by OB-R in
HepG2 cells (51). Therefore, OB-R signaling is unlikely to require
interaction with the gp130 signal-transducing component of the IL-6
receptor family.
The question of whether OB-R utilizes a second transmembrane component
for signaling is an important one. Several cytokine receptors have been
shown to interact with signaling components other than gp130. For
example, IL-3, granulocyte-macrophage colony-stimulating factor, and
IL-5 all interact with a common chain critical for signal transduction
(IL-3R OB-R could conceivably interact with a novel signaling chain.
Alternatively, it may transduce all relevant signals through homodimerization or homo-oligomerization as has been shown for a number
of class I cytokine receptors such as growth hormone receptor,
erythropoietin receptor, and G-CSFR (49, 50). To begin to distinguish
among these possibilities, chimeric receptors were constructed between
the extracellular domain of the G-CSFR and the intracellular domain of
OB-R (G-CSFR/OB-R), as well as the reciprocal chimeric receptor
encoding the extracellular domain of OB-R and the intracellular domain
of GCSF-R (OB-R/G-CSFR) (58). Both of these receptor fusions were able
to signal, and in fact, the signaling repertoire and potency were not
significantly different from that of the native receptors. The
observation that a G-CSFR/OB-R chimera could transduce signals in
response to G-CSF demonstrates that simple homodimerization of OB-R
intracellular domains is sufficient for transducing at least some
signals. This suggests that OB-R does not require an accessory binding
chain and may be activated by simple homodimerization or
homo-oligomerization. Furthermore, the ability of the leptin ligand to
generate a signal from the OB-R/G-CSFR chimera indicates that the
leptin ligand is capable of homo-oligomerizing OB-R extracellular
domains, since this is the requirement for G-CSFR intracellular domain
triggering.
Intracellular domain mutagenesis of OB-R has shown that there are at
least two distinct regions capable of generating intracellular signals
(58). A mutant receptor in which the most C-terminal 50 amino acids are
removed loses the ability to stimulate transcription via an IL-6
response element. However, this 50-amino acid C-terminal truncation
does not affect the ability of OB-R to stimulate transcription via a
hematopoietic receptor response element. Similar data were observed
when the most C-terminal tyrosine (Tyr-1141) was mutated to
phenylalanine (51). In addition, this single amino acid substitution differentially affected the ability of OB-R to activate different STAT
proteins (strongly affecting its ability to activate STAT1 and STAT3
and minimally affecting its ability to activate STAT5). The mutational
separation of distinct activities emanating from the OB-R intracellular
domain suggests that multiple distinct signals may be propagated by
OB-R triggering. Which of the identified or as yet unidentified
signaling regions is most important for body weight regulation will
require the transgenic introduction of mutant receptors into the
db mouse and testing for complementation of the mutant
phenotype.
There are numerous important questions remaining about the role of
OB-R in body weight regulation. Identification of the cell types in
which direct receptor signaling must occur is important to our
understanding of this pathway. It is possible that leptin need only
directly bind to a small number of neurons within the hypothalamus and
that receptor triggering in this small group of cells transmits the
leptin message to the large number of tissues affecting appetite,
metabolism, and nutrient partitioning. It is also possible that there
are peripheral tissues with sufficient quantities of OB-RL
and that direct receptor triggering at these sites is important for
leptin to function.
Additionally, the intracellular molecules that mediate signal
transduction via OB-RL are of tremendous interest. Although an understanding of some OB-R signal transduction capabilities is well
under way, it is critical to define which of these signals, or as yet
unidentified signals, is most important in body weight regulation. The
observation that OB-R is capable of activating STAT proteins and
promoter response elements suggests that leptin binding will result in
changes in gene expression within those cells expressing
OB-RL. The identification of genes induced by leptin in the
relevant cell types and the determination of the role they play in body
weight regulation will address many still unanswered questions
concerning body weight homeostasis.
The nature of leptin resistance is of paramount importance to the
understanding of human obesity. It is possible that this resistance
occurs as early in this pathway as the transport of leptin from the
blood into the brain. Another more remote possibility beginning to be
explored (59) is that significant numbers of humans will have a
polymorphism in the OB-R gene, which results in the production of
receptors with different signaling potency. More likely, perhaps, is
the possibility that leptin resistance is a result of flaws within the
signal-transducing pathway of the leptin receptor, more analogous to
the mechanism of insulin resistance. Discerning between these
possibilities and designing pharmacological strategies to correct them
are the exciting challenges ahead.
Fig. 1.
OB-R short and long forms. The
extracellular, transmembrane, and intracellular domains of
OB-RS and OB-RL are shown schematically.
wsxws, Trp-Ser-X-Trp-Ser motif; aa,
amino acids; mu vs hu, murine versus human.
[View Larger Version of this Image (74K GIF file)]
T transversion)
within an exon containing the extreme C terminus and 3
-untranslated
region of the predominant short intracellular domain form of OB-R
(OB-RS1). This mutation results in the generation of a new
splice donor site, creating an exon that becomes inappropriately
spliced into the transcript encoding the long intracellular domain form
of OB-R (Fig. 2). This new exon is composed of the last
6 codons and first 88 base pairs of the 3
-untranslated region of the
primary short form (OB-RS1) and is inserted exactly where
the long and short intracellular domains diverge. As a result of this
insertion, the long form transcript in db/db mice
would encode a protein in which the majority of the intracellular
domain has been truncated and is identical to the major short form
(OB-RS1). The demonstration that the defect in
db/db mice is in the OB-R gene provided
validation for the importance of this receptor in body weight
regulation. In addition, the near identity of the
ob/ob (leptin defect) and
db/db (OB-RL defect) phenotypes
suggests that without OB-RL leptin can exert no control
whatsoever over body weight regulation.
Fig. 2.
Model of OB-R splicing. Exons are
illustrated as blue or yellow boxes, and introns
are indicated by solid lines. Yellow corresponds
to exon sequences that are translated, and blue corresponds to 3-untranslated sequences. In a wild type mouse (wt),
splicing can result in transcripts that encode either short
(dashed line above exons) or long (dashed line
below exons) intracellular domains. In db/db
mice, these same splicing decisions can still occur; however,
transcripts encoding the long intracellular domain are interrupted by a
partial exon created by the db mutation. This results in a
stop codon prematurely terminating the long intracellular domain. The
mutation is indicated by an asterisk. Exons and exon segments are not drawn to scale. pA, poly(A) adenylation
site.
[View Larger Version of this Image (17K GIF file)]
), as do IL-2, IL-4, IL-7, and IL-9 (IL-2R
) (49, 50).
However, OB-R was found to transduce signals in hepatoma cell lines
which do not express IL-3R
or IL-2R
(51).