Lipodystrophy syndromes are conditions in which the adipose tissue mass of an individual is altered inappropriately. The change in adipose mass can range from a relatively modest and subtle redistribution in some individuals with partial lipodystrophy to a near-complete absence of adipose tissue in the most severe forms of generalized lipodystrophy. The common feature is a disconnection between the need of the individual for a safe, healthy lipid storage capacity and the available adipose mass to perform this critical role. The inability to partition lipids for storage in appropriately functioning adipocytes leads to lipid accumulation in other tissues, which typically results in conditions such as diabetes, dyslipidemia, fatty liver, and cardiovascular disease. Several genes have been identified whose disruption leads to inherited forms of lipodystrophy. There is a link between some of these genes and adipose dysfunction, so the molecular basis of disease pathophysiology appears clear. However, for other lipodystrophy genes, it is not evident why their disruption should affect adipose development or function or, in the case of partial lipodystrophy, why only some adipose depots should be affected. Elucidating the molecular functions of these genes and their cellular and physiological effects has the capacity to uncover fundamental new insights regarding the development and functions of adipose tissue. This information is also likely to inform better management of lipodystrophy and improved treatments for patients. In addition, the findings will often be relevant to other conditions featuring adipose tissue dysfunction, including the more common metabolic disease associated with obesity.
Introduction
Lipodystrophy syndromes are disorders featuring altered adipose mass, distribution, or function (1). Lipodystrophy may be classified as partial or generalized, depending on the severity of the adipose tissue loss. In congenital generalized lipodystrophy (CGL), severe lack of adipose tissue may be the principal feature driving lipid accumulation in other tissues, insulin resistance, and metabolic disease. In familial partial lipodystrophy (FPLD), adipose mass may be only modestly reduced, but dysregulated insulin sensitivity and lipid handling may result in exaggerated metabolic consequences (1). The often subtle reduction of adipose mass and/or redistribution means that partial forms of lipodystrophy syndromes are likely to be significantly underdiagnosed, and recent screening suggests a much higher prevalence than previously thought (2). Related to this, studies of FPLD may reveal more insight as to precisely why lower body subcutaneous adipose tissue is typically beneficial to metabolic health while upper body visceral adipose tissue in particular may be detrimental. However, it should be noted that in some cases of acquired partial lipodystrophy, accumulation of lower body adiposity can be associated with metabolic disease (3). Understanding adipose tissue dysfunction in lipodystrophy has implications beyond the treatment of this rare group of patients, as it starkly illustrates the importance of appropriately functioning adipose tissue in the maintenance of metabolic health. Indeed, the inability to further expand metabolically active adipose tissue is understood to underlie many of the metabolic consequences of obesity, where the healthy expansion limit of adiposity may have been exceeded (4). Hence, defining the mechanisms underlying lipodystrophy may lead to better understanding of metabolic control in more common conditions where adipose dysfunction plays a role, particularly in obesity-related disease.
Detailed clinical descriptions of different forms of genetic and acquired lipodystrophy have been published recently elsewhere (1,5). These provide a comprehensive overview of the condition, including the rarer or more complex forms of lipodystrophy, and highlight key clinical management and treatment strategies (1,5). This review aims to highlight mechanistic links between different forms of the disease and explore some key outstanding questions.
Mechanisms Underlying FPLD: A Complex Picture
Although the biological functions of genes whose disruption can cause FPLD have been extensively studied, in several cases the precise molecular mechanisms via which they cause the condition remain unclear (1). Although it is important to consider effects in other tissues, most have direct functions within developing and mature adipocytes, as depicted in Fig. 1. The genetic basis of FPLD type 1 (FPLD1) remains uncertain but appears to be polygenic. Indeed, it has been noted recently that FPLD1 might be better considered the extreme of a distribution of common, metabolically unhealthy “apple-shaped” adipose distribution in women (1). FPLD type 2 (FPLD2) is caused by mutations in the LMNA gene, encoding the nuclear lamina proteins lamins A and C, which differ in their C-terminal protein sequences (1,6). This represents the most common cause of FPLD and probably the most intriguing. LMNA mutations can cause not only lipodystrophy but also a wide range of conditions, including muscular dystrophies and progeroid syndromes as well as several neuropathic conditions, cardiac conduction system disorders, or cardiomyopathy syndromes. However, this typically involves mutations distinct from those causing FPLD2 (6). The diversity of phenotypes in these “laminopathies” reflects the diversity of roles that lamin proteins play within the cell (6). Lamin A/C is involved in multiple processes, including communication between the nuclear envelope and gene transcription, nuclear pore protein activity, nuclear stiffness, and mechanotransduction. It remains unclear which of these functions underpins the importance of LMNA in developing or mature adipocytes, although several recent advances have revealed new potential mechanisms.
Functions of key lipodystrophy genes within developing and mature adipocytes. The schematic shows the principle functions of genes whose mutations cause inherited forms of lipodystrophy (in red), which may contribute to disease causation. Other proteins relevant to their functions are shown in green. Note that FPLD1 appears polygenic in nature (see main text). FPLD2 is caused by mutations in LMNA, encoding lamin A/C, a nuclear lamina matrix protein, which may influence adipocyte gene expression and function by multiple mechanisms. Mutations in the key adipocyte transcription factor PPARγ can cause FPLD3. PPARγ functions as a heterodimer with RXRα to control the expression of multiple genes essential for adipocyte development and function. Mutations in PLIN1, encoding perilipin 1 (FPLD4), LIPE, encoding HSL (FPLD5), and CIDEC (FPLD6) are all likely to cause lipodystrophy by altering storage due to effects on adipocyte lipolysis and lipid droplet biogenesis. Under basal conditions, perilipin 1 sequesters ABHD5, an activator of ATGL, but FPLD4 mutation may prevent this, leading to increased and poorly controlled lipolysis. The subsequent actions of HSL and monoacylglycerol lipase (MGL) at the lipid droplet complete the breakdown of stored triacylglycerol to fatty acids and glycerol. ADRA2A normally reduces lipolysis, and its mutation can also cause FPLD, likely due to increased lipolysis. Mutations in AKT2, a key enzyme of insulin signaling in adipocytes, can also impair adipogenesis and cause FPLD. PCYT1A encodes phosphate cytidylyltransferase 1α, a critical enzyme in phosphatidylcholine synthesis whose disruption impairs adipocyte differentiation, potentially via altering nuclear membrane phosphatidylcholine content and stress pathways and/or synthesis of signaling lipids. CGL1 is caused by mutations in AGPAT2. AGPAT2 is a key enzyme in triacylglycerol synthesis but also impairs adipogenic gene expression by an unknown mechanism. Disruption of BSCL2, encoding the protein seipin, causes CGL2. Seipin is an ER membrane protein that can bind AGPAT2 as well as GPAT3 and lipin 1, all of which can influence adipogenic gene expression as well as generate diacylglycerol, which diacylglycerol acyltransferase (DGAT) enzymes can use to generate triacylglycerol. Seipin can also bind the key regulator of lipid droplet biogenesis LDAF1 and thereby alter lipid droplet dynamics. Mutations in CAV1, encoding caveolin 1, or PTRF, encoding polymerase I and transcript release factor, can cause CGL3 and CGL4, respectively. Both are key proteins of membrane caveolae whose integrity is essential for normal hormonal signaling and lipid uptake and transport in adipocytes.
Functions of key lipodystrophy genes within developing and mature adipocytes. The schematic shows the principle functions of genes whose mutations cause inherited forms of lipodystrophy (in red), which may contribute to disease causation. Other proteins relevant to their functions are shown in green. Note that FPLD1 appears polygenic in nature (see main text). FPLD2 is caused by mutations in LMNA, encoding lamin A/C, a nuclear lamina matrix protein, which may influence adipocyte gene expression and function by multiple mechanisms. Mutations in the key adipocyte transcription factor PPARγ can cause FPLD3. PPARγ functions as a heterodimer with RXRα to control the expression of multiple genes essential for adipocyte development and function. Mutations in PLIN1, encoding perilipin 1 (FPLD4), LIPE, encoding HSL (FPLD5), and CIDEC (FPLD6) are all likely to cause lipodystrophy by altering storage due to effects on adipocyte lipolysis and lipid droplet biogenesis. Under basal conditions, perilipin 1 sequesters ABHD5, an activator of ATGL, but FPLD4 mutation may prevent this, leading to increased and poorly controlled lipolysis. The subsequent actions of HSL and monoacylglycerol lipase (MGL) at the lipid droplet complete the breakdown of stored triacylglycerol to fatty acids and glycerol. ADRA2A normally reduces lipolysis, and its mutation can also cause FPLD, likely due to increased lipolysis. Mutations in AKT2, a key enzyme of insulin signaling in adipocytes, can also impair adipogenesis and cause FPLD. PCYT1A encodes phosphate cytidylyltransferase 1α, a critical enzyme in phosphatidylcholine synthesis whose disruption impairs adipocyte differentiation, potentially via altering nuclear membrane phosphatidylcholine content and stress pathways and/or synthesis of signaling lipids. CGL1 is caused by mutations in AGPAT2. AGPAT2 is a key enzyme in triacylglycerol synthesis but also impairs adipogenic gene expression by an unknown mechanism. Disruption of BSCL2, encoding the protein seipin, causes CGL2. Seipin is an ER membrane protein that can bind AGPAT2 as well as GPAT3 and lipin 1, all of which can influence adipogenic gene expression as well as generate diacylglycerol, which diacylglycerol acyltransferase (DGAT) enzymes can use to generate triacylglycerol. Seipin can also bind the key regulator of lipid droplet biogenesis LDAF1 and thereby alter lipid droplet dynamics. Mutations in CAV1, encoding caveolin 1, or PTRF, encoding polymerase I and transcript release factor, can cause CGL3 and CGL4, respectively. Both are key proteins of membrane caveolae whose integrity is essential for normal hormonal signaling and lipid uptake and transport in adipocytes.
Most molecular and cellular studies have used knockout or knockdown approaches or have overexpressed FPLD-causing mutant forms of lamin A/C. These have revealed some potential means whereby lamins may control adipogenesis. Early studies implicated abnormal regulation of the key adipogenic transcription factor SREBP1c (7–10) and variously invoke SREBP1c sequestration and increased or decreased DNA interactions as potential mechanisms. Other work has proposed that lamin A/C makes direct repressive contacts with chromatin at multiple loci in mesenchymal stem cells (MSCs), some of which contain proadipogenic genes, genes favoring alternative cell fates or genes involved in maintaining stem cell identity. Once adipogenesis is induced, contacts with proadipogenic genes are then selectively relinquished, facilitating adipocyte development (11). Wild-type lamin A and lamin A bearing the FPLD2 mutation p.R482W (LMNA-R482W) exert different effects on lamin-associated domains (LADs) in chromatin, and LMNA-R482W may specifically alter LADs that include genes with known effects in adipogenesis (12). In addition, expression of LMNA-R482W, but not wild-type lamin A, can increase the expression of the microRNA miR-335 in MSCs (13). miR-335 can inhibit adipogenesis, while inhibition of miR-335 in LMNA-R482W-expressing cells can restore the induction of adipogenic genes (13).
The FPLD2-causing R482L mutation in LMNA can impair activation of the key developmental Notch signaling pathway in MSCs and so inhibit adipogenesis in these cells (14). Recently, a study using cells isolated from subcutaneous white adipose tissue (WAT) and brown adipose tissue (BAT) from patients with FPLD2 has revealed that these pathogenic LMNA mutations can alter autophagy pathways during adipogenesis, leading to altered gene expression and a more WAT-like phenotype of BAT cells (15). This was consistent with the observed phenotype of BAT in patients and indicates that dysregulated autophagy also underpins at least some of the effects of LMNA mutation on adipose dysfunction in FPLD2.
Thus, a range of different molecular pathways have emerged that are regulated by lamins and may be altered in FPLD2. However, none so far convincingly explains the selective loss of subcutaneous, particularly lower body, adipose tissue that is typically metabolically beneficial while permitting preservation of metabolically harmful visceral fat. This remains the critical unanswered question not only in FPLD2 but also in other forms of partial lipodystrophy (Fig. 2). In the author’s view, not only better animal models (described below) but also more precise cellular models are required to probe the molecular mechanisms underlying FPLD2 and other partial lipodystrophy. Reliable immortalized human cell models of the different adipose stem cells that occur in different adipose depots are likely key to defining the molecular mechanisms via which lamin A and lamin C alter adipose tissue function. Combining these with the knock-in of different LMNA mutations and complementary in vivo experiments may finally reveal why different depots are selectively affected and why this leads to metabolic disease, even where some adipose depots appear preserved.
Multiple adipose tissue depots are variably affected by different forms of FPLD and CGL. Adipose depots with clearly defined beneficial effects on metabolic health are highlighted in green. In particular, the lower body subcutaneous depots, such as gluteal and thigh fat, are typically disproportionately reduced in FPLD, and this is likely to contribute significantly to metabolic disease. Adipose depots whose expansion is typically less healthy are shown in amber. Visceral/mesenteric adipose tissue expansion is typically preserved in FPLD and in individuals with obesity. Expansion of these adipocytes, shown in purple, is particularly associated with poor metabolic health. Several adipose depots, such as plantar, orbital, and joint fat, show variable levels of preservation in different forms of lipodystrophy. Their functional properties are relatively poorly understood but are increasingly studied and clearly have important local functions. These may contribute to some of the less well-characterized aspects of the disease. Overall, the variable effects of different genetic disruptions on different adipose depots highlight the unique properties and/or stem cell origins of each and the importance of the affected genes in their development and maintenance.
Multiple adipose tissue depots are variably affected by different forms of FPLD and CGL. Adipose depots with clearly defined beneficial effects on metabolic health are highlighted in green. In particular, the lower body subcutaneous depots, such as gluteal and thigh fat, are typically disproportionately reduced in FPLD, and this is likely to contribute significantly to metabolic disease. Adipose depots whose expansion is typically less healthy are shown in amber. Visceral/mesenteric adipose tissue expansion is typically preserved in FPLD and in individuals with obesity. Expansion of these adipocytes, shown in purple, is particularly associated with poor metabolic health. Several adipose depots, such as plantar, orbital, and joint fat, show variable levels of preservation in different forms of lipodystrophy. Their functional properties are relatively poorly understood but are increasingly studied and clearly have important local functions. These may contribute to some of the less well-characterized aspects of the disease. Overall, the variable effects of different genetic disruptions on different adipose depots highlight the unique properties and/or stem cell origins of each and the importance of the affected genes in their development and maintenance.
Until recently it was considered that patients with FPLD2 exhibited changes in adipose tissue distribution and attendant metabolic disease only during or after puberty. However, it has now been demonstrated that both male and female children with FPLD-linked LMNA variants have reduced lower limb adiposity that precedes puberty and that serum triglycerides are elevated in females (16). Nonetheless, it remains the case that metabolic disease and lipodystrophy typically are diagnosed only after puberty and are more severe in females. Thus, there do appear to be influences of pubertal changes in adipose depots and/or the regulation of these processes by endocrine signals during puberty. Concerning potential mediators of this, sex steroids can influence multiple aspects of adipose tissue biology that are directly affected by genes mutated in lipodystrophy, including adipocyte differentiation, lipolysis, lipogenesis, and adipokine secretion (17). Adipose tissue is both a target and a source of estrogens, which can influence adipose distribution. Hence, there is clear scope both for lipodystrophy genes to influence adipose distribution via altering sex steroid production or signaling and for sex steroids to contribute to differences in lipodystrophy phenotypes between sexes and over the course of life. This is another area that clearly deserves further targeted research, as it may offer opportunities to intervene early in life to dramatically improve the course of disease in patients with lipodystrophy.
Much more is known regarding how pathogenic mutations in the PPARG gene can cause FPLD type 3 (FPLD3) (18). Peroxisome proliferator–activated receptor γ (PPARγ) is a key transcription factor regulating adipocyte development and the target of the thiazolidinedione class of antidiabetic drugs. However, PPARγ is also essential for placentation, making PPARγ deficiency lethal, so lipodystrophy-causing mutations in PPARG do not simply result in complete loss of function. Instead they may affect cofactor or DNA binding to various degrees and thereby impair the normal actions of the PPARγ protein in developing and/or mature adipocytes (18). Given that PPARγ regulates genes controlling adipocyte development and function (Fig. 1), any given pathogenic mutation may affect adipogenesis and the insulin sensitivity, lipid storage capabilities, and metabolic function of mature adipocytes. Indeed, the presence of significant adipose depots in patients with FPLD3 implies that adipocyte function, rather than development per se, is the main mechanism via which PPARγ mutations cause lipodystrophy and metabolic disease. The functions of PPARγ are reviewed elsewhere (18). PPARγ provides a clear example of where detailed knowledge of the molecular function and structure of a protein can aid the understanding of potential pathogenic mutants. High-throughput systematic mutation of all residues of PPARγ and examination of their effects has been possible because well-recognized downstream readouts are available to quantitatively assess PPARγ activity (19). The resulting data offer a valuable reference compendium for new mutations identified with high-throughput genomic screens and illustrates the potential power of similar analyses of other lipodystrophy genes. A key limiting factor of applying this approach to other lipodystrophy genes is the requirement to have a functional readout amenable to high-content screening methods in a readily manipulated cell model. Recent work using morphological analyses in cultured human adipocytes offers one example of how this type of approach can be applied (20).
FPLD Genes Affecting Adipocyte Lipolysis
FPLD type 4 (FPLD4) is caused by homozygous frameshift mutations in the gene PLIN1, encoding the key adipocyte lipid droplet protein perilipin 1 (21,22). Perlipin 1 plays critical roles in the regulation of lipolysis by controlling the association between adipose triglyceride lipase (ATGL) and its activator, CGI-58. Perilipin sequesters CGI-58 until hormonal sympathetic stimulation of adipocytes leads to perilipin phosphorylation, CGI-58 release, and subsequent activation of ATGL. FPLD4-causing frameshift mutations in perilipin 1 lead to increased basal lipolysis. In at least some cases this is due to loss of CGI-58 sequestration, which causes persistent ATGL activation (21,22). However, other frameshift mutations in PLIN1 have been found in individuals with no evidence of FPLD (23). It remains possible that some of these cause very subtle alterations in adiposity and metabolic health or, alternatively, that FPLD4-causing PLIN1 mutations have specific functional consequences. A combination of further detailed phenotyping of individuals with a range of PLIN1 mutations and corresponding detailed cellular and molecular analyses will be valuable in resolving this question. Such studies are also likely to bring new insights regarding perilipin 1 function relevant to adipocyte lipid storage and metabolic health. Lipodystrophy has also been reported in individuals homozygous for loss-of-function mutations in hormone-sensitive lipase (HSL), the product of the LIPE gene (FPLD6) (24). Although initially it was proposed that this led to altered adipose distribution via reduced lipolysis, a recent study has also revealed altered adipocyte development in adipose stem cells isolated from affected individuals (25). A single family has been reported with an atypical form of FPLD due to mutations in adrenoceptor-α 2A (ADRA2A), an adrenergic receptor normally involved in suppressing lipolysis in adipocytes (26). Mutation of the lipid droplet regulator CIDEC can also lead to FPLD5 (27). CIDEC regulates lipid droplet fusion and growth, controlling the transfer of lipids between droplets but also influencing lipolysis.
As with LMNA and PPARG mutations, it is not clear why only certain adipose depots appear affected by mutations in genes that encode key players in adipocyte lipid handling and lipolysis. This may reflect differences in how lipid storage is governed in these depots and/or how hormonal responsiveness of lipolysis varies within their constituent adipocyte populations. More detailed analysis of affected versus unaffected depots and comparison between these different forms of FPLD should reveal exciting new lessons regarding how adipocyte lipid droplet regulation is linked to metabolic health.
Genes Disrupted in CGL
The majority of cases of CGL for which the genetic cause is known result from mutations in the gene encoding 1-acylglycerol-3-phosphate-O-acyltransferase 2 (AGPAT2) or in BSCL2, which encodes an endoplasmic reticulum (ER) transmembrane protein named seipin (5) (causing CGL1 or CGL2, respectively). AGPAT2 is an enzyme that catalyzes the acylation of lysophosphatidic acid to form phosphatidic acid, which is important for glycerophospholipid and neutral lipid synthesis, including the generation of triglyceride. However, cellular studies suggest that preadipocytes lacking AGPAT2 exhibit a defect in their capacity to induce adipogenic genes rather than just an inability to generate triglyceride within the maturing adipocyte (28,29). Given the ability of AGPAT2 products to feed into the synthesis of multiple lipid species, a credible explanation is that the lack of this enzyme alters lipid signaling. Increased cyclic phosphatidic acid generation, altered intracellular signaling downstream of phosphatidylinositol 3 kinase, and a failure to induce several key proadipogenic transcription factors have been implicated in impaired adipogenesis in AGPAT2-deficient preadipocytes (29,30). However, the molecular mechanisms underlying CGL1 have not yet been fully defined.
The seipin protein appears to function as a molecular scaffold protein in the ER membrane, and several investigators have shown that seipin and its nonmammalian orthologs form homo-oligomers of between 9 and 12 subunits. Most recently, this has included detailed cryoelectron microscopy analyses showing human seipin may form 11-subunit oligomers (31,32). Multiple studies have reported seipin interacting proteins, which include AGPAT2, glycerol-3-phosphate acyltransferase 3 (GPAT3), lipin1/3, ADRP, SERCA2A, 14-3-3β, perilipin 1, promethin/LDAF1, and Reep1 in mammalian cells and Pet10p, Ldo45, and Ldo16 in yeast (33). Some pathogenic mutations in seipin impair protein stability or normal oligomerization of seipin and may thereby inhibit complexing of key binding partners (34).
Seipin plays a significant role in the regulation of lipid droplets that is highly evolutionarily conserved from mammals to yeast (33). Moreover, the human seipin protein can functionally rescue normal lipid droplet morphology in yeast lacking the seipin ortholog. Seipin oligomers may play a physical role in the development of nascent lipid droplets within the ER membrane by regulating surface tension, modifying membrane curvature, or directing the sequestration of lipids into the droplet (32). Alternatively, as several lipid droplet-modifying proteins and enzymes have been shown to associate with seipin, the key role of seipin may be to scaffold these together, thereby recruiting them to participate in lipid droplet development. Indeed, the association of seipin with proteins such as promethin/LDAF1 has been shown to play a key role in this process. Most likely seipin plays a dual role as both a physical regulator of droplet biogenesis and a scaffold protein to recruit other key regulators of the process (33). However, it remains unclear if the loss of this lipid droplet-modifying activity underlies the lack of adipose tissue in patients with CGL2.
Multiple studies have shown that loss of seipin expression in precursor cells impairs adipocyte development in culture. Seipin has been shown to bind directly to AGPAT2, offering the enticing possibility that disruption of either of these two genes causes CGL via a common mechanism (30,35). Seipin can also bind to GPAT enzymes, which generate the substrate for AGPAT2 and are key regulators of phosphoglycerolipid and triglyceride synthesis (35,36). Seipin loss in cultured preadipocytes was shown to increase GPAT activity, while inhibiting GPAT3 expression could improve adipogenesis in seipin-deficient preadipocytes (35). This provided a potential unifying hypothesis whereby seipin loss, by increasing GPAT activity, could cause the formation of disorganized and/or unusually large lipid droplets in multiple cell types but inhibit adipogenesis in adipose stem cells. However, congenital loss of GPAT3 in seipin null mice did not substantially rescue adipose development, although it did improve metabolic health (37). To add further complexity, while seipin appears critical for adipogenesis in cellular models and adipose development in vivo (38–40), there is also evidence that seipin is needed for adipocyte maintenance (41,42). Hence, CGL due to seipin disruption could at least partly result from postnatal loss of adipocytes. Finally, seipin is highly expressed in the brain and clearly detectable in the liver as well as other tissues, where it could contribute to metabolic disease in CGL2. However, despite its conserved role in lipid droplet formation, it does not appear that seipin loss specifically in the liver significantly contributes to hepatic steatosis in CGL2 (43).
Overall, while multiple seipin binding partners have been identified, including other proteins implicated in lipodystrophy, such as AGPAT2 and perilipin 1 (20,30), the precise molecular mechanism via which seipin loss leads to severe CGL remains unclear. However, these links hint at the possibility of some shared underlying mechanisms that might ultimately offer therapeutic targets relevant to multiple forms of lipodystrophy.
The mechanisms via which CAV1 and PTRF mutations cause CGL3 and CGL4, respectively, are also incompletely understood. CAV1 encodes caveolin 1, a protein involved in the formation and function of caveolae (44). Homozygous loss-of-function mutations in CAV1 have been shown to cause CGL, although heterozygous CAV1 mutations have also been linked to cases of both CGL and FPLD (1,45). Caveolae are plasma membrane invaginations important for multiple cellular functions, including the activation of intracellular signaling (including by insulin in adipocytes), lipid uptake, and lipid transport (44). PTRF encodes Cavin 1, a critical protein for the stabilization of caveolae (44). This suggests that a common mechanism links the lipodystrophy observed in patients with CGL3 and CGL4. Consistent with this, Cav1 null mice are insulin resistant with reduced expression of the insulin receptor, and similar features can also be observed in Ptrf null mice (44). Intriguingly, a recent study of Agpat2 null mice has revealed significantly decreased caveola formation in adipocytes lacking AGPAT2 that are briefly present in the early postnatal period (46). This provides a potential mechanistic link between CGL1, CGL3, and CGL4 that clearly warrants further investigation. Overall, the involvement of caveolae in insulin signaling and lipid handling in adipocytes implies that adipose dysfunction and lipodystrophy result from the disruption of these functions. However, the diverse functions and dynamic nature of caveolae, caveolins, and other proteins associated with these organelles means that significant further work is required to define the mechanism by which caveolin 1 and Cavin 1 loss causes CGL.
Understanding Depot-Selective Differences in Adipose Tissues
As highlighted above, a key question remains how mutations in genes causing lipodystrophy can have depot-selective effects. This is most evident in considering FPLD, where there is maintenance or even expansion of certain adipose depots (1). However, there are also significant depot-selective effects in CGL (Fig. 2). Notable is the retention of adipose depots such as those within the joints, the plantar fat pads, and retro-orbital fat in patients with CGL1 (AGPAT2 deficiency) and CGL3 (CAV1 deficiency), which are typically absent from patients with CGL2 (BSCL deficiency) and CGL4 (PTRF deficiency) (Fig. 2). This implies that these depots are inherently different from those that are absent from all forms of CGL and that the genes affected have depot-selective functions. Similarly, bone marrow adipose tissue appears to be absent from CGL1 and CGL2 but preserved in CGL3 and CGL4. The origins and properties of bone marrow adipose tissue (of which at least three subtypes may exist) are known to differ from other adipose depots (47), but the molecular basis of this altered susceptibility to the loss of different CGL-linked genes is not clear. The differences between the stem cells and properties of classical subcutaneous and visceral depots are becoming more clearly defined, with multiple distinct stem cell populations making various contributions to the adipocytes within each adipose depot (4). The local microenvironment also contributes depot-specific signals that affect adipose stem cell proliferation in addition to inherent differences in the properties of resident stem cells (48). Given lineage-tracing methods have been required to define these individual stem cell populations, most studies have involved murine rather than human adipose tissues. However, there is an increasing body of data characterizing differences between the stem cells giving rise to subcutaneous and visceral adipose tissue and upper versus lower body subcutaneous adipose depots in humans (4). Depot-selective effects of different lipodystrophy genes may arise because these genes are only required for adipogenesis by a subset of adipose stem cells or for the survival of the resulting adipocytes. In this scenario, even depots that may appear to be preserved in FPLD might in fact be comprised of different constituent adipocytes than one would see in healthy individuals. Adding to the complexity, it is evident that the characteristics of adipose depots and their adipocytes change as metabolic disease progresses in obesity and probably also in lipodystrophy. Hence, it will prove challenging to differentiate primary from secondary changes that occur in adipose depots of patients with lipodystrophy, and detailed longitudinal studies may be required. As noted above, another important step will be comparing the effects of disrupting lipodystrophy genes in specific different adipose stem cell populations. Together, such studies may ultimately reveal why adipose depots are differentially affected in the various forms of FPLD and CGL and point to novel strategies for treating these patients. This may include pharmacological or gene therapy approaches but would also inform specific targeted stem cell therapies. Importantly, this information will be more broadly valuable in understanding healthy versus unhealthy adipocyte profiles, which will be relevant to more common metabolic disease in obesity.
Murine Models of Lipodystrophy: Limitations and Opportunities
Murine disease models have a critical role as preclinical models to develop novel therapies for multiple conditions, including more common forms of diabetes and dyslipidemia. Moreover, mouse models with severe generalized lipodystrophy, such as the A-ZIP transgenic mice, demonstrated the critical importance of adipose tissue in maintaining metabolic health (49). However, the use of murine models specifically harboring mutations in genes causing lipodystrophy in humans, particularly those causing FPLD, has proved challenging. In part this may reflect difficulties in accurately mapping the anatomical equivalents of different human adipose depots to those in mice, along with differences in lipid and glucose handling and energy expenditure. The autosomal-dominant inheritance of most FPLD also requires knock-in models rather than knockouts. The knock-in models of the FPLD3-causing P467L mutation in PPARγ (P465L in mice) exemplifies this, as restricted islet function or leptin deficiency was required to unmask an overt metabolic phenotype (50,51). Further studies have revealed other subtle but metabolically relevant effects of this mutation, such as those on lipidemia (52).
Studies of accurate knock-in models of FPLD2-causing mutations in LMNA are conspicuous by their absence, given that these account for so many cases of FPLD. Nonetheless, new insights have come very recently from adipose-selective disruption of Lmna in mice (53), which leads to a progressive lipodystrophy very similar to that of FPLD2. Adipogenesis from tissue-resident predipocyte stem cells appeared relatively unaffected by Lmna loss, and early postnatal adipose development was relatively well preserved. However, as mice aged to adulthood, adipose depots reduced in size and impaired glucose metabolism became evident. Metabolic disease is milder in adipose-specific Lmna null mice than in patients with FPLD2 (LMNA) (1,53). However, when adipose-specific Lmna null mice were challenged with a high-fat diet, they became more overtly glucose and insulin intolerant than their control littermates. This is reminiscent of the observations with knock-in mice modeling the FPLD3-causing P467L mutation in PPARγ noted above, where additional metabolic stress or drivers for adipose expansion were needed to uncover clear metabolic disease (50,51,53). While adipose-specific Lmna null mice do not genetically model FPLD2, there are clearly very significant similarities to the phenotype seen in the patients. Hence, they should offer a valuable new model to examine the role of LMNA in adipose function with which to identify potential therapeutic approaches for the treatment of lipodystrophy.
Overall, murine in vivo studies of CGL indicate that mouse models of generalized lipodystrophy more accurately recapitulate the human disease and offer a better opportunity for preclinical studies. Nonetheless, there remain some limitations with these models compared with the corresponding human disorders. For example, while AGPAT2 disruptions generate a more severe CGL than BSCL2 deficiency in mice, BSCL2 mutations cause the more severe CGL in humans (1,38,54). Although the reason for this is unclear, it may be related to the finding that in rodents Bscl2 disruption does not prevent BAT formation per se and that seipin-deficient BAT still accumulates lipids (38,39,55). BAT is absent from Agpat2 null mice, indicating that Agpat2 is needed for both BAT and WAT development or maintenance. In contrast, BAT in adult humans appears more comparable to murine brite adipose tissue and may be reliant on BSCL2/seipin for development or maintenance (54,56). Thus, residual adipose tissue in Bscl2 null mice could arise partly from stem cells capable of developing via BAT-specific adipogenic pathways. Along with residual BAT, this could result in a less severe lipodystrophy than that seen in Agpat2 null mice and patients with mutations in BSCL2. Also of note, while Bscl2 null mice have a severe generalized reduction in adipose tissues, they do not exhibit the hypertriglyceridemia observed in BSCL2-deficient individuals unless placed on a dyslipidemic Ldl receptor null background (57). Regardless of these caveats, several studies have used mice modeling CGL2 to examine novel potential therapies for the condition and have shown beneficial effects of leptin, PPARγ agonists, FGF21, sodium–glucose cotransporter inhibitors, and adipose transplantation (42,57–60).
Thus, despite the limitations of murine models in studying the mechanisms of lipodystrophy-associated disease, it is important not to underestimate their utility. Rather, it will be important to identify the most appropriate background strain, diet, and housing conditions and define the parameters that best predict therapeutic benefit in humans. Replacement of the adipocyte-derived hormone leptin (or metreleptin injection) is effectively the only dedicated treatment for lipodystrophy. The low availability due to high cost highlights the difficulties of accessing therapeutics selectively beneficial in rare diseases. In addition, while metreleptin is typically highly effective as a treatment to improve glycemic control, dyslipidemia, hepatic steatosis, and appetite regulation in CGL, it is not effective in all cases of lipodystrophy, especially in some patients with FPLD. The significant benefits and the limitations of leptin therapy have been reviewed in detail elsewhere (61), but it is clear other treatment options are needed. Repurposing existing drugs as treatments for lipodystrophy or developing therapies that improve metabolic health in both obesity and lipodystrophy should circumvent this problem. Well-defined preclinical models of lipodystrophy are likely to play a critical role in achieving this.
Poorly Understood Features of Lipodystrophy and the Need for More Targeted Treatments
Most research so far has focused on the metabolic and endocrine aspects of lipodystrophy affecting patients. These include the loss of subcutaneous fat, hyperglycemia, and lipoatrophic diabetes with associated complications such as hyperlipidemia, chronic pancreatitis, severe insulin resistance, retinopathy, neuropathy, and chronic, debilitating hyperphagia derived from hypoleptinemia (1). However, patients also suffer a range of lesser-known or acknowledged consequences of the condition. These include both musculoskeletal and neuropathic chronic pain, debilitating chronic fatigue, and mental health difficulties such as depression, anxiety, and low self-esteem (1,62). Recent studies involving patients with lipodystrophy and complementary work in mouse models have provided some insight regarding the skeletal muscle hypertrophy that is observed in individuals with lipodystrophy, which may be linked to fatigue experienced by patients. A reduction in type 1 (oxidative) and increase in type 2 (glycolytic) muscle characteristics has previously been observed in patients with CGL (63). A recent, detailed study of patients with FPLD2 reports a lower oxidative muscle phenotype consistent with this as well as evidence of impaired mitochondrial function, possibly underlying the observed earlier onset of muscle fatigue (64). Reduced protein degradation may partly explain increased muscle mass, which is found in various forms of FPLD and CGL as well as in mouse models of CGL. Incomplete lipid oxidation has been reported in patients with FPLD2 as well as in mice lacking Bscl2, modeling CGL2. In the latter case, this was not due to Bscl2 disruption in the skeletal muscle itself. Together the evidence points to increased skeletal muscle mass and muscle fatigue being linked to systemic consequences of adipose insufficiency in lipodystrophy, given that it can be seen in patients with FPLD and CGL as well as in mouse models of the latter condition. Nonetheless, how this occurs remains unresolved but might result from metabolic changes in skeletal muscle caused by altered substrate availability, changes in adipokines or other adipose-derived secreted factors, or alterations in neuronal signals to the muscle. This exemplifies the type of “nonmetabolic” features of lipodystrophy where a combination of molecular, cellular, and in vivo animal and human studies has significant potential to reveal novel targets to develop therapies that would improve the experience of patients. Moreover, the findings may well prove relevant to other conditions that feature chronic fatigue or muscle weakness.
Conclusions
The molecular mechanisms underlying why and how fat mass or fat distribution is altered in lipodystrophy, and how this leads to particular pathologies, remains incompletely understood. Key outstanding questions surround how mutations in genes causing partial lipodystrophy can selectively affect only certain adipose depots and why the effects of this can so significantly impact metabolic health. However, there has been a significant recent acceleration in our understanding of how the genes disrupted in inherited forms of lipodystrophy exert their effects. By applying novel cutting-edge techniques, there is a huge opportunity to transform further our understanding in this area. Such research also offers a unique means to understand the fundamental biology of adipose tissue development and function as well as how adipose tissue influences health.
Article Information
Acknowledgments. The author would particularly like to thank Dr. Rebecca Sanders (Lipodystrophy UK) and Andra Stratton (Lipodystrophy United) for their valuable input and insights. The author would also like to thank Pat Bain, Rowett Institute, University of Aberdeen, for assistance in generating the figures. Unfortunately, due to the limitations of the article format, it was not always possible to cite all relevant original articles, especially reports identifying individual mutations in genes causing lipodystrophy. The author apologizes to any investigators whose contributions have not been individually recognized as a result of this.
Funding. This work was supported by Diabetes UK (18/0005884) and the Biotechnology and Biological Sciences Research Council (BB/V015869/1).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.