Introduction
Obesity can be described as a disproportionate accumulation or organization of body
fat and has become an increasingly important health problem worldwide, referred to
as the obesity pandemic [1]. Obesity is
usually assessed by measuring height and weight to calculate the body mass index
(BMI, kg/m2). Recently, however, it has emerged that this
parameter is not an efficient predictor of mortality risk and should be replaced by
the estimation of percent body fat despite the complexity and high financial cost
of
assessment [2].
Obesity contributes to chronic diseases, including type 2 diabetes (T2D), hepatic
steatosis and steatohepatitis, cardiovascular disease, stroke, dyslipidemia,
hypertension, and different types of cancers [3]
[4]. Moreover, progressive
caloric pressure on white adipose tissue (WAT) results in low-grade but persistent
inflammation, called “metaflammation”, associated with increased
macrophage infiltration responsible for the clearance of dysfunctional and dying
adipocytes [5]
[6]. Adipocytes and infiltrating immune
cells can produce and release a plethora of chemokines and cytokines that mediate
systemic inflammation in obese patients [7]. This continuous inflammatory status in WAT also leads to a metabolic
switch from a storage to an inflammatory phenotype, causing ectopic lipid deposition
in secondary tissues such as the liver or muscle, which in turn results in
deregulated systemic insulin signaling [8].
Adipose tissue macrophages (ATMs) play a central role in the development and
progression of inflammation in WAT. In fact, inhibition or silencing of various
mediators produced by these immune cells was shown to be sufficient to ameliorate
the multiple pathological consequences of obesity [9]
[10]. By investigating the molecular mechanism of this polarization switch
in macrophages, inositol-requiring enzyme 1a and peroxisome proliferator-activated
receptor gamma were identified as important regulators in this process, as they
create an inflammatory environment [11]
[12]. To reduce WAT
inflammation and adipocyte hypertrophy, caloric restriction is widely used as a
treatment strategy. However, this therapy paradoxically increases the number of ATMs
and the expression of related cytokines in humans and mice [13]
[14]
[15]
[16]. In addition, various pharmacological
approaches have been used in an attempt to reduce obesity. For instance, melatonin,
a powerful antioxidant, has been demonstrated to reduce obesity-related problems by
lowering inflammatory adipokines such as interleukin-6 (IL-6), monocyte
chemoattractant protein-1, leptin, and tumor necrosis factor-α [17]. The numerous beneficial effects of
glucagon-like peptide-1 (GLP-1) render this hormone an interesting candidate for the
development of pharmacotherapies to treat obesity, diabetes, and neurodegenerative
disorders [18]. In fact, a weekly dose of
the GLP-1R agonist semaglutide was associated with a sustained reduction in body
weight [19]. In mice, semaglutide reduced
adipocyte hypertrophy and macrophage infiltration and activated adipocyte browning
and mitochondrial biogenesis to promote weight loss [20].
Other studies focused on discovering the non-inflammatory role of immune cells
induced in obesity. Under conditions of increased adiposity, secreted factors from
WAT trigger a program of lysosome biogenesis in ATMs to buffer the huge amount of
lipids from adipocytes [21]. Unexpectedly,
ATMs from obese mice do not polarize toward one of the classical M1 or M2
phenotypes. The authors, therefore, hypothesized that these are not qualitative
changes in the expression profile of ATMs, but that these cells increase in number
and, thus, no clear polarization was observed [21]. Obesity reprograms ATMs into a pro-inflammatory metabolically
activated state that is transcriptionally, mechanistically, and functionally
distinct from M1- or M2-like phenotypes [22]. A unique pleiotropic phenotype in WAT health has been attributed to
these macrophages, which varies between beneficial (removal of dead adipocytes) and
deleterious (release of pro-inflammatory cytokines) determined by the duration of
high-fat diet feeding, at least in mice [23]. Single-cell RNAseq studies have recently identified a new macrophage
subset induced during obesity, termed lipid-associated macrophages, characterized
by
the expression of triggering receptors expressed on myeloid cells 2 [24]. These cells appear to activate an
expression profile that involves phagocytosis, lipid catabolism, and lysosomal
pathways, which is a common phenotype of macrophages in different inflamed tissues
such as the liver, brain, and atherosclerotic plaques [24]
[25]
[26]
[27].
In the last 10 years, the search for new genes that are differentially regulated in
the WAT of obese mice has substantially increased. The discovery that glycoprotein
non-metastatic melanoma protein B (GPNMB) is highly expressed in the WAT of obese
animals [28], opened the way for many
studies on this protein. This review aims to emphasize the importance of GPNMB in
the context of ATMs, lysosomal function, and obesity.
Structure, function, and regulation of glycoprotein non-metastatic melanoma
protein B
GPNMB was first identified in 1995 in a screen using high and low metastatic
human melanoma cell lines [29]. It is
a type I transmembrane glycoprotein, also known as osteoactivin or dendritic
cell heparan sulfate proteoglycan integrin-dependent ligand [30]
[31]. GPNMB was detected in a range of cell types, including
osteoblasts and osteoclasts in bone, melanocytes, keratinocytes, microglia in
the central nervous system, as well as macrophages and dendritic cells [31]. It was found to be increased in a
variety of inflammatory diseases such as colitis, renal diseases, different
types of cancers, and neurodegenerative disorders [32]
[33]. In addition, GPNMB was associated with other disorders, such as
senescence [34], vitiligo [35], glaucoma [35], myocardial infarction [37], and atherosclerosis [38]. Mutations in Gpnmb cause
hypopigmented lesions and pigmentary glaucoma in mouse models [39]
[40]
[41] and recessive and
semi-dominant amyloidosis cutis dyschromica in humans [42]
[43].
The human Gpnmb gene, located at chromosome 7p15, encodes for 2
alternative splicing isoforms of 572 and 260 amino acids [44]. Mouse Gpnmb codes for a
protein of 574 amino acids and shares 70.16% sequence identity with the
human protein [45]
[46]. In its extracellular domain, GPNMB
contains an N-terminal signal peptide (SP), an integrin-binding RGD motif and a
polycystic kidney disease domain, a single-pass transmembrane domain, as well as
an immunoreceptor tyrosine-based activation-like motif and a lysosomal targeting
di-leucine motif in the cytoplasmic tail ([Fig. 1a, b]) [47]
[48]. The protein has 12 potential
N-glycosylation sites, described in numerous cell types [49]
[50]
[51]. It is
predominantly located in endosomal/lysosomal compartments, where it
promotes the recruitment of light chain 3 (LC3/Atg8) to the phagosome
for lysosomal fusion ([Fig. 1c])
[52]
[53]
[54]
[55]. In addition to
phagocytosis, GPNMB was also associated with efferocytosis, the clearance of
apoptotic and necrotic cells primarily by macrophages [56]. IL-6, under the control of the
phosphorylated-signal transducer and activator of transcription 3 (pSTAT3), was
shown to be a positive regulator of this progress [57]. Although Gpnmb-deficient
bone marrow-derived macrophages can initiate phagocytosis, they are unable to
digest the cargo content, as pSTAT3 activation is not sustained over time.
Moreover, this impairment does not allow macrophages to correctly switch from an
inflammatory to a restorative phenotype, underscoring the link between GPNMB,
phagocytosis, and tissue repair [57].
Fig. 1 GPNMB structure and localization. (a) Model of human
GPNMB. The structure of the protein was predicted using the BIOZENTRUM
SWISS-MODEL tool [62].
(b) Schematic model of the GPNMB structure, including
N-glycosylation sites and cleavage site for AD-AM10. SP, signal peptide;
RGD, RGD motif; PKD, polycystic kidney disease domain; TMD,
transmembrane domain; ITAM, immunoreceptor tyrosine-based
activation-like motif; DL, di-leucin motif. (c) Gpnmb expression
is regulated by melanogenesis associated transcription factor (MITF) and
transcription factor EB (TFEB). It localizes to the
endo/lysosomal com-partment, where it recruits LC3/Atg8
for phagosome fusion. This image was created with Bio-render.com
(accessed on September 22nd 2023). [rerif]
Although the precise mechanisms driving this process are unknown, a disintegrin
and metalloproteinase 10, a proteolytic enzyme belonging to the matrix
metalloproteinase (MMP) family, contributes to GPNMB extracellular domain
shedding [58]. This soluble form
(sGPNMB) can bind to a variety of receptors, including
Na+/K+-ATPase, CD44, epidermal
growth factor receptor, vascular endothelial growth factor receptor, and other
molecules such as integrins, heparin, and syndecan-4 [31]
[59]. In addition, GPNMB signaling increases extracellular
signal-regulated kinase and protein kinase B phosphorylation in many disease
models [60]
[61]
[62]
[63]. GPNMB is tightly
transcriptionally regulated, with melanogenesis-associated transcription factor
(MITF) as one of the major players. MITF overexpression increased GPNMB
expression by binding and activating its promoter in both human and animal cells
[55]
[64]
[65]. In addition, transcription factor EB was identified as a
regulator of GPNMB expression [34].
Endo/lysosomal localization of glycoprotein non-metastatic melanoma
protein B and its role in lysosomal storage diseases
Since GPNMB is localized to endo/lysosomes, studies focused on
understanding its role in the biology of these organelles and its link to
macrophages, one of the cell types that primarily utilize lysosomal degradation
to generate energy in response to the nutritional status of the cell. The link
between GPNMB and lysosomal function is supported by many in vitro
studies. Numerous inducers of lysosomal stress, such as HEPES, sucrose,
chloroquine, bafilomycin, concanamycin A, or palmitate, increase GPNMB
expression in macrophage cell lines [28]
[66]. Moreover, the
lysosomal/endocytic marker lysosomal associated membrane protein 2 was
reported to co-localize with GPNMB in osteoclasts [55]. GPNMB is essential for the
recruitment of the autophagy protein LC3/Atg8 to the surface of
autophagosomes and subsequent acidification and fusion with lysosomes [54], highlighting its close association
with this organelle.
The accumulation of lysosomal macromolecules and the resulting stress condition
may be due to a genetic deficiency of lysosomal enzymes, leading to lysosomal
storage disorders (LSDs). Tissue macrophages are among the primary storage cells
involved in LSDs because they contribute to the cleavage of various substrates.
Biomarkers such as chitotriosidase (CHIT1) and chemokine (C-C motif) ligand 18
(CCL18) have been identified in patients with LSD but cannot be used in mouse
models because CHIT1 is not expressed in phagocytes [67] and a CCL18 homolog is absent in
rodents [68]. The urgent need to find
a new marker for this group of diseases led to the discovery of GPNMB. Van Ejik
and colleagues were among the first to demonstrate an increase in GPNMB in
Gaucher disease spleen and, in particular, in Gaucher cells, the lipid-laden
macrophages characteristic of this pathology, accompanied by several
hundred-fold increase in circulating sGPNMB concentrations [69]. These discoveries paved the way
for many other studies that underscored the importance of GPNMB in Gaucher
disease [70]
[71]
[72]
[73] and other LSDs
[74]
[75]
[76] and, like CHIT1 and CCL18, confirmed its strong association with
LSDs and lipid-laden macrophages.
Further findings provided important insights into the possible molecular
mechanisms underlying the increase in GPNMB in LSDs. Another important player
during lysosomal stress is the mammalian target of rapamycin complex 1 (mTORC1),
a protein localized to the surface of lysosomes and implicated in the control of
autophagy [77]
[78]. In several models of impaired
lysosomal function, mTORC1 was downregulated, and Mitf, the main
transcription factor regulating Gpnmb expression, was upregulated
[28]
[79]. Moreover, lysosomal
Ca2+ release, as a consequence of organelle stress, was
shown to induce nuclear translocation and activation of transcription factor EB,
another important transcription factor for Gpnmb
[80].
Glycoprotein non-metastatic melanoma protein B and obesity
Across several models of obesity, expansion of WAT induces a program of lysosome
biogenesis in ATMs associated with lipid catabolism but not a classic
inflammatory phenotype [21], arguing
that the increase in the inflammatory profile of WAT associated with obesity
derives primarily from quantitative increases in immune cell populations. Thus,
in addition to genetic defects, lysosomal lipid accumulation is also triggered
when the amount of fat exceeds the storage capacity of the adipocytes, which
eventually undergo apoptosis and recruit macrophages. When the WAT is no longer
able to process lipids properly, they may accumulate in ectopic tissues, such as
the liver or skeletal muscle.
Since the reports that GPNMB is drastically induced in WAT of several obese
animal models [28]
[81]
[82], many studies have focused on describing the role of this protein
in obesity and associated metabolic disorders ([Fig. 2]). GPNMB was identified as a
negative regulator of macrophage inflammatory responses and only reparative,
anti-inflammatory M2-like macrophages activated by TGFβ retain
full-length GPNMB on their surface [83]. Pro-inflammatory macrophages activated by interferon γ
and lipopolysaccharide secrete sGPNMB [83]. sGPNMB, which is abundantly produced by hypertrophied
adipocytes, was also suggested to reduce the inflammatory capacity of
macrophages by inhibiting nuclear factor-κB signaling mainly through
binding to CD44. Thus, chronic WAT inflammation was severely exacerbated in
high-fat diet-fed Gpnmb-deficient mice, accompanied by a pronounced
increase in crown-like structures [84]. These data emphasize the critical function of GPNMB in macrophage
activation and the subsequent inflammatory response in obese WAT.
Fig. 2 Overview of the role of GPNMB in macrophage function and
obesity. (a) Membrane bound GPNMB is retained on the surface of
anti-inflammatory macrophages, whereas soluble (s)GPNMB is released by
pro-inflammatory cells. (b) Obesity induces the production of
sGPNMB by adipocytes, which lowers the inflammatory capacity of
macrophages by interacting with CD44 on the cell surface and inhibiting
the function of NF-kB. (c) To reduce oxidative stress, lipid
accumulation, and fibrosis in the liver, obese adipocytes release
sGPNMB, which interacts with calnexin on Kupffer and stellate cells.
(d) In obese WAT, the hepatokine sGPNMB activates SREBP1c to
promote lipogenesis by binding to CD44 on adipocytes. This image was
created with Biorender.com (accessed on September 24th 2023).
[rerif]
The phenomenon that GPNMB plays an essential role in decreasing WAT inflammation
during obesity by reducing the number of ATMs was absent when GPNMB was
over-expressed in adipocytes and macrophages of mutant mice [81]. Whether the discrepancy in the
observed phenotype is due to the different high-fat diet (coconut oil [81] versus lard [83]) remains elusive.
In fact, only palmitic acid present in lard is able to trigger insulin resistance
[85] and GPNMB expression [28], leading to a stronger effect on
WAT of obese mice. However, both diets were very effective in inducing liver
steatosis. Furthermore, both studies showed that sGPNMB secreted by adipocytes
from obese mice was responsible for decreased oxidative stress, fat deposition,
and fibrosis in the liver by interacting with calnexin on Kupffer and stellate
cells. However, sGPNMB was also described as a hepatokine that activates SREBP1c
and thus lipogenesis in obese WAT by binding CD44 on adipocytes, resulting in a
positive correlation between sGPNMB and BMI [86]. These findings indicate that GPNMB is a strong risk factor for
obesity.
GPNMB was also linked to T2D, one of the diseases potentially associated with
obesity. Numerous sequelae, such as acute renal injury, cardiovascular disease,
muscle failure, ocular pathologies, and cognitive dysfunction, frequently
accompany the development of T2D. Indeed, GPNMB was found to be increased in
many of these T2D-associated disorders [87]
[88]
[89], once more emphasizing the
important role of this protein as a biomarker in obesity and its related
conditions.
