Hepatic Insulin Clearance: Mechanism and Physiology

Abstract

Upon its secretion from pancreatic β-cells, insulin reaches the liver through the portal circulation to exert its action and eventually undergo clearance in the hepatocytes. In addition to insulin secretion, hepatic insulin clearance regulates the homeostatic level of insulin that is required to reach peripheral insulin target tissues to elicit proper insulin action. Receptor-mediated insulin uptake followed by its degradation constitutes the basic mechanism of insulin clearance. Upon its phosphorylation by the insulin receptor tyrosine kinase, carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) takes part in the insulin-insulin receptor complex to increase the rate of its endocytosis and targeting to the degradation pathways. This review summarizes how this process is regulated and how it is associated with insulin-degrading enzyme in the liver. It also discusses the physiological implications of impaired hepatic insulin clearance: Whereas reduced insulin clearance cooperates with increased insulin secretion to compensate for insulin resistance, it can also cause hepatic insulin resistance. Because chronic hyperinsulinemia stimulates hepatic de novo lipogenesis, impaired insulin clearance also causes hepatic steatosis. Thus impaired insulin clearance can underlie the link between hepatic insulin resistance and hepatic steatosis. Delineating these regulatory pathways should lead to building more effective therapeutic strategies against metabolic syndrome.

Introduction

Insulin is cleared mainly in liver and kidney, with the former contributing to a greater extent than the latter in removing endogenously released hormone. Skeletal muscle, adipocytes, and other extra-hepatic insulin target tissues/cells contribute to peripheral insulin clearance (169). Following its pulsatile secretion from pancreatic β-cells (171), insulin is transported rapidly via the portal vein to reach the hepatocytes by crossing the fenestrations in the liver sinusoidal endothelium (134).

Under normal physiological conditions, up to 80% of secreted insulin is cleared during its first passage through the liver (50), and the half-life of insulin in the portal circulation is ~3–5 min (52). Receptor-mediated insulin uptake followed by insulin degradation in hepatocytes constitutes the basic mechanism of hepatic insulin clearance (204). Unprocessed insulin is delivered into the systemic circulation to exert its action in target peripheral tissues and brain, and to undergo further receptor-mediated uptake and degradation in the periphery (204). Unprocessed insulin can also return to the liver via the hepatic artery for a second round of insulin degradation (204). Contrary to the liver, insulin transport across endothelial cells of peripheral tissues is the rate-limiting step for insulin uptake (12). In this manner, endothelial cells in extra-hepatic insulin target tissues contribute to the regulation of systemic insulin homeostasis and action (107, 116).

The association between Type 2 diabetes (T2D) and impaired insulin clearance was initially reported in 1949 (22). Since then, several studies have identified defective insulin clearance as a critical factor in the pathogenesis of hyperinsulinemia in metabolic syndrome (97, 166), particularly among Hispanics and African Americans (77, 113, 125). Reduced insulin extraction has also been detected in aging (134) and in other metabolic abnormalities irrespective of ethnicity, including obesity (100, 102, 127, 206), non-alcoholic steatohepatitis (21, 109, 128), and polycystic ovarian syndrome (36). Moreover, hepatic, but not extra-hepatic, insulin clearance is lower in African Americans than in European American women (165), pointing to the dominant role of hepatic insulin clearance in maintaining insulin sensitivity in the African-American population. In Japanese T2D patients, however, peripheral rather than hepatic insulin clearance appears to be more adversely affected (154). Collectively, these studies link impaired insulin clearance to insulin resistance and other manifestations of metabolic syndrome.

Despite the wealth of evidence supporting a strong association between reduced insulin clearance and metabolic disorders, the cause-effect relationship between insulin resistance and its hallmark, hyperinsulinemia, remains obscure (189). Chronic hyperinsulinemia emerges from increased insulin secretion (68) as well as reduced insulin clearance (7, 77, 100, 109) to compensate for peripheral insulin resistance, in particular in individuals with obesity (69, 96, 98). Several epidemiological studies have suggested that reduced insulin clearance in obese subjects is more relevant than the increase in insulin secretion to compensate for peripheral insulin resistance (58, 59). In obese patients, reduced insulin clearance seems to be secondary to insulin resistance rather than to robust visceral obesity (101, 206). Consistently, a weight loss by ~10% of initial body weight increases insulin clearance and restores insulin sensitivity (96).

On the other hand, impaired insulin clearance can initiate hyperinsulinemia-driven systemic insulin resistance (16, 39, 170). Supportive evidence of the causative role of chronic hyperinsulinemia in the pathogenesis of secondary insulin resistance in obesity and T2D is mounting (131, 175, 201). Mechanistically, this could be mediated, at least in part, by desensitization and lysosomal downregulation of the insulin receptor by chronically elevated levels of insulin (106). Such a paradigm is supported by the systemic insulin resistance state that develops in response to impairing hepatic insulin clearance in mice with liver-specific inactivation (173) and deletion (75) of carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1), a glycoprotein that plays a central role in promoting hepatic insulin clearance (157, 173). In this review, we will describe how CEACAM1-dependent insulin clearance pathways regulate homeostatic insulin levels and, subsequently, insulin sensitivity. We will also describe how CEACAM1 coordinates its effect with insulin-degrading enzyme (IDE) to regulate intracellular insulin trafficking and targeting to the degradation process to remove insulin from the circulation.

Role of CEACAM1 in Receptor-Mediated Insulin Endocytosis

CEACAM1: General Structure

CEACAM1 is a transmembrane glycoprotein that belongs to the immunoglobulin superfamily of the CEA family of cell adhesion molecules (91, 143, 158). Before its nomenclature was revised in 1999 (14), CEACAM1 had been referred to as BGP1, CD66a, C-CAM1, Ecto-ATPase, and pp120/HA4 (93, 142).

CEACAM1 is ubiquitously expressed in particular in the liver and to a lower extent in the kidney, main sites of insulin clearance (93). The structure as well as the cellular and tissue distribution of CEACAM1 protein are highly conserved among species (33, 42, 174, 179). In humans and rodents, CEACAM1 is expressed as two alternatively spliced isoforms with CEACAM1–4L containing a long (L), highly conserved cytoplasmic tail of 71 amino acids (based on the rat sequence) (152), and the short (CEACAM1–4S) containing only 10 amino acids of the intracellular tail and lacking critical phosphorylation sites: Ser503, Tyr488, and Tyr513 residues (FIGURE 1). Genomic and cDNA analyses revealed that these isoforms result from the alternative splicing of exon 7 of the Ceacam1 gene (comprised of nine exons) (140). Recent studies have shown that alternative splicing is regulated by interferon regulatory factor 1 (44). CEACAM1–4L and CEACAM1–4S contain four immunoglobulin (IgG)-like structures in their extracellular domains that foster the cell adhesion property of CEACAM1, which is chiefly regulated by the NH₂-terminal IgV-like loop (93). The other common pair of CEACAM1 in humans and rodents is CEACAM1–2L and CEACAM1–2S, which possess two IgG loops in their extracellular domains emerging from alternative splicing of exons 3 and 4 (150).

FIGURE 1.

Regulation of Insulin Signaling by CEACAM1

Insulin (Ins) binding to the α-subunit of its receptor (IRα) activates its tyrosine kinase (TK) to catalyze its autophosphorylation on tyrosine 960 (pY960) in the juxtamembrane domain and on tyrosine 1316 (pY1316) on the COOH-terminus tail of the β-subunit (IRβ) of the receptor (1). CEACAM1 phosphorylation on Y488 requires phosphorylation on Y1316 in IRβ (1). pY488 on CEACAM1 binds to the SH2 domain of Shc, which, in turn, allows the binding of the PTB domain of Shc to pY960 in IRβ. In this manner, Shc mediates the formation of a complex between IR and CEACAM1 (2). CEACAM1 binds to SHP2 to sequester it. This protects IRS against the phosphatase and prolongs its activation. This also releases Shc from pY960 in IRβ (3).

CEACAM1-L contains in its cytoplasmic tail a basally phosphorylated Ser503 residue and two tyrosine residues (Tyr488 and Tyr513) that could be phosphorylated by Src-kinase and dephosphorylated by the inhibitory SH2-containing tyrosine phosphatases 1 and 2 (SHP-1 and SHP-2) (24, 139, 146). One of these tyrosine residues (Tyr488) undergoes phosphorylation by ligand-stimulated insulin receptor (IR) and epidermal growth factor receptor (EGFR) tyrosine kinases (2, 147). Tyr488 phosphorylation site of CEACAM1-L mediates several of its functions, including regulation of insulin and fatty acid metabolism in liver (71, 142, 149, 173) and inhibition of cell proliferation in response to insulin and EGF (2, 172), among others (13, 80, 93). Whereas the short isoform of CEACAM1 (CEACAM1-S) is devoid of these phosphorylation sites, it can still undergo phosphorylation by protein kinases to regulate cytoskeletal dynamics (54, 55, 104). In addition to the relative distribution of the long vs. the short isoform, the differential distribution to the lateral vs. the apical domains of the plasma membrane in polarized epithelial cells determines the ultimate function of CEACAM1 (199, 200). Under normal physiological conditions, CEACAM1-S is exclusively expressed on the apical domain, whereas CEACAM1-L is expressed on the lateral and apical domains of polarized epithelial cells (199, 200). A main mediator of CEACAM1’s regulatory role in insulin and fatty acid metabolism is the phosphorylation state of its long isoform, which is, in turn, regulated by an interplay of several kinases and phosphatases (142, 153). Consistently, experiments in transfected cell systems ruled out a significant role for CEACAM1-S in receptor-mediated insulin uptake (34, 71). Thus we will focus this review on the role of the long isoform in insulin clearance and fatty acid synthesis in hepatocytes and how this is mediated by its phosphorylation by the IR tyrosine kinase.

CEACAM1 Phosphorylation and Regulation of Insulin Signaling

Endogenous insulin is secreted from pancreatic β-cells into the portal vein in acute pulses to rapidly reach the sinusoidal domain of the hepatocyte via fenestrae in liver capillaries. Insulin binding activates the tyrosine kinase of the receptor to initiate a chain of phosphorylation/dephosphorylation reactions, starting with the autophosphorylation of the receptor by its own tyrosine kinase in the cytosolic tail of its β-subunits (IRβ) (41, 82, 164, 184, 213). In addition to the receptor, several other substrates have been identified, including insulin receptor substrate (IRS) family of proteins and Src-homology-2-containing protein (Shc) (185), which undergo phosphorylation in a spatiotemporal manner to avoid competition for binding to Tyr960 in the juxtamembrane domain of IRβ. The perinuclear distribution of Shc facilitates the relative delay in its binding to Tyr960 in IRβ subunit.

Expressed at a very high level on the sinusoidal surface membrane of hepatocytes, CEACAM1-L (from here on referred to as CEACAM1 for simplicity, unless otherwise mentioned) undergoes rapid phosphorylation by IR in intact cells (147). Phosphoamino acid analysis in H35 rat hepatoma cells indicated that CEACAM1 is basally and constitutively phosphorylated on serine and that this phosphorylation decreases significantly upon insulin treatment to be replaced by tyrosine phosphorylation (162). Site-directed mutagenesis revealed that Ser503 is the main site of basal serine phosphorylation (147). Neither the serine kinase that catalyzes this reaction nor the potential serine phosphatase that reverses its activity in the presence of insulin has been clearly identified, but Ser503 is located in a potential cAMP-dependent protein kinase phosphorylation site (Lys-Arg-Pro-Thr-Ser503) (124, 192), and the phosphatase activity appears to require an intact Tyr513 in the cytosolic tail of CEACAM1 (141).

In contrast to most other substrates, CEACAM1 is not phosphorylated by the IR’s close relative, the insulin-like growth factor receptor 1 (IGF-1R) (144). This differential phosphorylation is mediated by the non-conserved Tyr1316 in the COOH-terminus tail of IRβ, in contrast to the other substrates, the phosphorylation of which requires the conserved Tyr960 in the juxtamembrane domain of this subunit. This differential dependence on key tyrosine residues in IRβ precludes a direct competition for phosphorylation by IR between CEACAM1 and IRS proteins, and allows their simultaneous phosphorylation by the activated receptor (147).

Site-directed mutagenesis revealed that Tyr488 (Asp-Asp-Val-Ser-Tyr488), located downstream of two acidic amino acids (159), is the main site of CEACAM1 phosphorylation by the IR tyrosine kinase (FIGURE 1). This requires an intact Ser503 in its cytoplasmic tail, as demonstrated by the inability of insulin to stimulate Tyr488 phosphorylation when Ser503 is mutated to Ala (S503A mutant) (147). Even though Tyr513 is also located downstream of glutamic acid (Glu-Thr-Val-Tyr513), it does not undergo phosphorylation by IR (147). In response to insulin, phosphorylated Tyr488 binds to the Src Homology 2 (SH2) domain of Shc (24) and brings it closer to the juxtamembrane region to bind to Tyr960 of the receptor via its NH₂-terminal phosphotyrosine-binding domain (PTB) (FIGURE 1) (172). CEACAM1 interaction bestows on Shc the ability to compete with IRS for binding to Tyr960 of IR and decrease PI3Kinase/Akt activation in response to insulin. In addition, CEACAM1 binding to Shc sequesters this main coupler of Grb2 adaptor to the receptor and reduces the Ras/MAPKinase activity and, subsequently, the mitogenic effect of insulin (172). Thus, by binding to Shc, CEACAM1 phosphorylation by the receptor can regulate both arms of the insulin signaling pathways. CEACAM1 phosphorylation on Tyr488 can also regulate insulin signaling through its binding to SH2-containing phosphatases, such as SHP1 (46, 139) and SHP2/Syp (146, 153). We have shown that this interaction sequesters SHP2 and reduces its binding to IRS proteins, prolonging the activation of the IRS/PI3Kinase pathway in response to insulin (FIGURE 1) (146). Because of the opposing effect of CEACAM1 binding to Shc and SHP2 on IRS/PI3Kinase activation, it is conceivable that its interactions with these downstream signaling molecules are tightly regulated by the metabolic demand on the cell, as discussed below.

CEACAM1 Increases Receptor-Mediated Insulin Endocytosis and Targeting to the Degradation Process

It has been suggested that the internalization of the insulin-IR complex is a two-step process. The first step involves redistribution of the receptor from the microvilli to the nonvillous membrane of the cell, whereas the second step consists of anchoring the receptor in clathrin-coated pits (29). Activation of IR by insulin is required for the rapid internalization of the insulin-IR complex into clathrin-coated vesicles to be ultimately targeted to the degradation process (27, 28, 78, 130, 218). Whereas insulin undergoes degradation, the insulin receptor may either recycle back to the plasma membrane or translocate into lysosomes for degradation. The fate of the insulin receptor depends on the duration and saturating doses of insulin (45). Under normal physiological conditions, the internalized receptor is recycled to the plasma membrane (221), as opposed to being targeted to lysosomal degradation under conditions of prolonged hyperinsulinemia (70). Furthermore, non-receptor-mediated processes, such as pinocytosis, are more relevant in chronic hyperinsulinemic states (86).

Although there is a general agreement that the juxtamembrane domain of the receptor plays an important role in insulin endocytosis (10, 83, 99), the underlying mechanisms remain controversial despite ample investigations on the role of the two Tyr-centered sequences around Tyr953 (Gly-Pro-Leu-Tyr953) and Tyr960 (Asn-Pro-Glu-Tyr960) located in tight β-turn structures that conform to the internalization signaling motif. Moreover, the COOH-terminus tail of IRβ appears to play a significant role in stabilizing the receptor to prevent its targeting to the lysosomal compartment (56, 119). Thus we posited that, in light of its regulation by Tyr1316 in the COOH-terminus tail of IRβ, CEACAM1 phosphorylation plays a significant role in the intracellular trafficking of the insulin-IR complex, including lysosomal exclusion of the receptor under normo-insulinemic conditions.

Whereas Tyr960 of the juxtamembrane domain of IRβ is required for phosphorylation of IRS proteins, studies in transfected Chinese Hamster ovary cells ruled out a role for IRS-1 in insulin-receptor endocytosis (29). In contrast, co-transfecting CEACAM1-L, but not CEACAM1-S or the phosphorylation-defective mutants of CEACAM1 (Y488F, S503A, and Y488F/Y513F), with human IR-A form (hIR-A) in NIH-3T3 fibroblasts increased the rate of uptake of ¹²⁵[I]Insulin via its receptor and its degradation by two- to threefold relative to cells expressing hIR-A alone (71). On the other hand, inhibiting CEACAM1 expression in H4-II-E rat hepatoma cells by antisense mRNA transfection reduced receptor-mediated insulin internalization and degradation (71). Thus phosphorylation of CEACAM1 regulates receptor-mediated insulin uptake followed by its degradation (34).

Mutating Tyr960 to Phe in hIR-A did not affect CEACAM1 phosphorylation but abolished its stimulating effect on insulin endocytosis and degradation in NIH-3T3 fibroblasts (145), pointing to a role for IRβ's Tyr960 in mediating CEACAM1-dependent insulin endocytosis pathways. This is consistent with a model locating phosphorylated CEACAM1 in the insulin-IR endocytosis complex, the formation of which is mediated by the binding of Shc to phosphorylated Tyr488 of CEACAM1 via its SH2 domain and to phosphorylated Tyr960 of IRβ through its PTB domain (FIGURES 1 AND 2). This complex is targeted to clathrin-coated pits/vesicles via the collagen homology domain of Shc (FIGURE 2), which is known to associate with adaptor protein-2 (AP2) in the adaptin structure of the clathrin coat of endocytosis vesicles to participate in ligand-stimulated internalization of growth factor receptors (155).

FIGURE 2.

Coordinated regulation of receptor-mediated insulin intracellular trafficking by CEACAM1 and IDE

Step 1: insulin binding to its receptor activates it to form a complex with CEACAM1 via Shc (as described in the legend to FIGURE 1). Together with Y513 in CEACAM1, the collagen homology domain of Shc targets the complex to the AP2 adaptin in the clathrin-coated pits. Step 2: the complex is internalized inside the early endosomes, and the extracellular IDE becomes sequestered into the vesicular lumen. Progressive SHP2 binding to pY488 on CEACAM1 destabilizes the complex and causes a progressive loss of Shc to allow IRS binding to pY960. This leads to increased IRS phosphorylation and activation of PI3Kinase that, in turn, mediates the recruitment of cytosolic IDE to the outer membrane and maintainance of CEACAM1 at the endosomes. Luminal IDE degrades insulin. Step 3: as the endosomes mature into late endosomes and acidify (H⁺), IDE is inactivated. FASN binding to pY488 pulls off CEACAM1 from the insulin-IR complex to destabilize it and allow insulin to dissociate from its receptor to be degraded by the acidic thiol protease (AcTP). Step 4: the receptor is recycled back to the membrane during exocytosis, resulting from microtubular polymerization, which is possibly regulated by IDE at the outer membrane of the vesicle. Whether this IDE mediates the formation of the CEACAM1/FASN vesicle remains to be tested.

Additionally, Tyr513 of CEACAM1, located within the Tyr-x-x-Φ sequence (Tyr513-Ser-Val-Val) that is shared by proteins sorted into clathrin-coated vesicles (178, 205) targets CEACAM1 to AP2-associated clathrin-coated vesicles at the plasma membrane of polarized Madin-Darby canine kidney (MDCK) epithelial cells (199) (FIGURE 2, step 1). In contrast, Tyr488 fails to target CEACAM1 directly to clathrin-coated vesicles despite its localization in an AP2-targeted tyrosine-centered motif (Tyr488-Ser-Val-Leu) (199).

In addition to playing a key role in the inward routing of CEACAM1 into endocytosis vesicles, Tyr513, but not Tyr488, plays a dominant role in targeting the newly synthesized glycoprotein from the trans-Golgi network (TGN) via AP1-associated vesicles to its final destination on the lateral domain of the plasma membrane in MDCK cells (199). Tyrosine phosphorylation of this residue increases PI3Kinase activity that leads to endosomal accumulation of CEACAM1 and its rapid disappearance from the lateral membrane of MDCK (199). Additionally, Ser503 plays an essential role in regulating the posttranslational maturation of the glycosylated moieties of CEACAM1 during its vectorial transport from TGN to the plasma membrane (35), as shown by the higher diversion of the S503A CEACAM1 mutant to the lysosomal compartments of transfected NIH-3T3 cells (35), and of wild-type CEACAM1 when MDCK cells were treated with Staurosporine, a Ser/Thr kinase inhibitor (199). Thus it is conceivable that being subjected continuously to clathrin-mediated transport to and from the plasma membrane in a manner depending on key amino acid residues in its cytosolic tail, CEACAM1 serves as a faithful chaperone to the insulin-IR complex during its intracellular trafficking.

That CEACAM1 phosphorylation is required to mediate its stimulatory effect on insulin endocytosis and degradation is further supported by the observations that 1) CEACAM1 fails to regulate receptor-mediated IGF-1 endocytosis (145) and 2) blunting the cell adhesion property of CEACAM1 by mutating Arg98 in the NH₂-terminal IgV-like ectoplasmic domain to Ala (193) without modulating its phosphorylation state does not adversely affect the ability of CEACAM1 to stimulate insulin endocytosis in NIH-3T3 fibroblasts (195).

To undergo degradation, insulin has to dissociate from its receptor in the acidic environment of late endosomes (11). This requires destabilization of the receptor-Shc-CEACAM1 complex. As we have shown (149), fatty acid synthase (FASN) binds to phosphorylated Tyr488, which would, in turn, pull CEACAM1 off the complex to destabilize it and allow the separation of insulin from its receptor. The pathways mediating the vesicular separation of the CEACAM1-FASN complex are not known but are consistent with CEACAM1 not being involved in receptor-mediated insulin retro-endocytosis (120), which appears to be regulated mainly by a protein kinase C (PKC)-mediated process (72).

In agreement with the observation that insulin endocytosis is not required for its signaling but mainly for its removal (130), our studies summarized above demonstrated that insulin-induced phosphorylated CEACAM1 increases the cellular uptake of insulin, likely by stabilizing the insulin-IR complex via its binding to Shc, which together with Tyr513 targets the complex to the endocytosis vesicles to increase the rate of insulin degradation (130, 218). Because Tyr1316 in the COOH-terminus tail of IRβ is required for CEACAM1 phosphorylation (196), it is likely that the association of phosphorylated CEACAM1 with the insulin-IR complex mediates the stabilizing effect of the COOH-terminus tail of IRβ to prevent receptor’s targeting to lysosomes under normo-insulinemic conditions.

CEACAM1 Links Insulin Removal with Low Fatty Acid Synthesis in Hepatocytes

Under non-fasting conditions, insulin receptor (210), CEACAM1 (93), and FASN (95, 188) are highly expressed in hepatocytes, a major site of insulin clearance from the portal circulation (93, 142). In human adult hepatocytes, the insulin receptor gene is expressed as two alternatively spliced isoforms that differ by the absence (IR-A) or presence (IR-B) of a 12-amino acid segment encoded by exon 11 at the COOH terminus of the α-subunit of the receptor (60, 137, 187, 212). IR-A is less abundant but has a higher affinity for insulin than IR-B. This differential binding affinity to insulin may represent a mechanism whereby hepatic insulin receptors are less sensitive to insulin-induced receptor downregulation to protect the liver against the physiologically higher levels of insulin in the portal vein (208). We have shown that CEACAM1 preferentially promotes insulin endocytosis via IR-A isoform without affecting its recycling to the surface membrane despite being equally phosphorylated by both receptor isoforms (120). This is consistent with CEACAM1 promoting insulin uptake by IR-A to target insulin to the degradation process and protect the expression of the low abundant high-affinity receptor (IR-A) against insulin-induced downregulation, in particular when plasma insulin levels are high. By regulating the relative distribution of IR-A and IR-B in hepatocytes, CEACAM1-dependent pathways contribute to maintaining insulin sensitivity in the liver.

Moreover, we have shown that, in response to acute insulin, internalized CEACAM1 binds to FASN on its phosphorylated Tyr488 residue to separate FASN from the insulin-IR complex and suppress its enzymatic activity. The physiological implication of this phospho-CEACAM1/FASN interaction is not only to destabilize the insulin-IR complex in late endosomes but also to restrict hepatic de novo lipogenesis in the face of the higher levels of hepatic than extra-hepatic lipogenic gene expression resulting from the physiologically higher insulin concentration in the portal than in the systemic circulation (208). Thus we presented the first line of evidence that, in contrast to its chronic stimulation of lipogenic gene transcription (156), acutely released pulses of insulin from pancreatic β-cells suppress FASN activity in the hepatocyte in a manner depending on insulin-mediated phosphorylation of CEACAM1 and internalization. Using Ob/Ob obese mice (149), we have demonstrated that, under conditions of hyperinsulinemia, when the pulsatility of insulin release diminishes (129), and insulin signaling and CEACAM1 phosphorylation are compromised, cellular insulin uptake and its acute effect on FASN activity are blunted, giving way to the long-term positive effect of chronically elevated levels of insulin on lipogenesis that are mediated by the sterol regulatory element binding protein (SREBP1c), a master transcriptional regulator of lipogenic genes (156), and the upstream stimulatory factor (USF-1) (216). This intertwined modulatory effect of CEACAM1 on insulin and lipid metabolism is consistent with a physiological model whereby CEACAM1-dependent pathways protect the liver against the physiologically high levels of insulin in the portal vein by increasing insulin uptake via its high-affinity IR-A receptor to suppress de novo lipogenesis and remove insulin while maintaining homeostatic levels of IR-A in hepatocytes. Disturbance in this system leads to impaired insulin clearance with exaggerated hyperinsulinemia that could lead to downregulation of the insulin receptor and hepatic steatosis.

Insulin Regulates Its Hepatic Clearance

Several studies have shown that insulin regulates its synthesis from pancreatic β-cells (76). Even though CEACAM1 is expressed in these endocrine cells, insulin secretion is not compromised in mice with global null deletion of Ceacam1 gene (Cc1^–/–) (43). Thus CEACAM1 plays a minor role in the regulation of insulin secretion and regulates homeostatic insulin levels mainly by promoting insulin clearance in hepatocytes.

As stated above, hepatic insulin clearance is intimately associated with insulin signaling by its dependence on receptor-mediated insulin delivery to the intracellular compartments where it is degraded. Whereas insulin action is mediated by insulin signaling pathways, the process is itself regulated by homeostatic circulating levels of insulin, which is in turn determined, not only by insulin secretion from pancreatic β-cells, but also by hepatic insulin clearance.

Insulin clearance in hepatocytes is initiated by the activation of the insulin receptor by acute release of insulin pulses from pancreatic β-cells, which in turn phosphorylates CEACAM1 to mediate insulin targeting to its degradation process. Additionally, acute insulin pulses during the first few hours of refeeding following an overnight fast induce the transcriptional activity of Ceacam1 promoter (140) to stimulate CEACAM1 expression (176). Thus, by inducing CEACAM1 transcription and phosphorylation, insulin stimulates its own clearance in hepatocytes, which in turn regulates its homeostatic levels to secure the propagation of its multiple functions in peripheral target cells.

The physiological implication of how insulin regulates its own clearance to coordinate with secretion its homeostatic levels is supported by several studies. To mention a few, fenofibrate maintains insulin sensitivity in mice fed a high-fat diet despite its negative effect on insulin secretion by exerting a parallel downregulatory effect on insulin clearance to maintain normo-insulinemia (177). Exenatide, a glucagon-like peptide-1 (GLP-1) receptor agonist, maintains physiological insulin level, despite its stimulatory effect on insulin secretion by causing a parallel increase in insulin clearance in mice fed a high-fat diet (74). The inability of fenofibrate (177) and exenatide (74) to reverse the effect of high-fat diet in Cc1^–/– mice demonstrates the dependence of their concomitant modulation of insulin secretion and clearance on hepatic CEACAM1-dependent pathways. Fenofibrate regulates Ceacam1 transcription by activating the peroxisome proliferator-activated receptor α (PPARα) (177), and exenatide binds to the PPRE response element in the Ceacam1 promoter to activate PPARγ to stimulate Ceacam1 transcription (74). Additionally, exenatide stimulates CEACAM1 expression in vivo through its ability to release insulin, which, in turn, stimulates Ceacam1 transcription (140).

Moreover, at fasting when insulin release is low, CEACAM1 expression is reduced by a PPARα-dependent mechanism to upregulate FASN activity and, in turn, reduce the level of malonyl-CoA and its inhibitory effect on fatty acids β-oxidation (176). Low level of CEACAM1 at fasting limits insulin extraction in the face of its low production to maintain homeostatic insulin levels. This is consistent with the observation that selective reduction in β-cell mass restricts pulsatility of insulin secretion followed by a compensatory reduction of postprandial hepatic insulin extraction in minipigs (105). Collectively, these findings emphasize the central role that hepatic insulin clearance plays in coordinating with insulin secretion the regulation of insulin homeostasis and, ultimately, its function (82).

Impaired Insulin Clearance, Hyperinsulinemia, and Insulin Resistance

The hallmark of insulin resistance is hyperinsulinemia. Mechanistically, the cause-effect relationship between these two parameters remains elusive.

As stated above, elevated plasma insulin levels can be caused by a compensatory increase in insulin secretion from β-cells and reduced insulin clearance. Several studies in humans (17) and animals (7, 103) have shown that reduced insulin clearance can cooperate with increased insulin secretion to properly regulate glucose homeostasis. In this regard, reduced insulin extraction alleviates the burden of pancreatic β-cells and limits the need for a compensatory increase in insulin synthesis and release. Because hepatic insulin clearance depends on receptor-mediated entry of insulin, blunting insulin signaling can restrict insulin uptake and its degradation, leading to reduced insulin clearance and providing a mechanistic underpinning for the negative effect of hepatic insulin resistance on insulin clearance.

On the other hand, impaired insulin clearance leads to hyperinsulinemia, which in turn, causes insulin resistance through several mechanisms. These include 1) desensitization and lysosomal downregulation of the insulin receptor leading to its reduced expression on the surface membrane (106), 2) limiting the pulsatility of insulin release (129, 151, 186), and 3) decreasing brown adipogenesis and, consequently, energy expenditure (175).

Hyperinsulinemia Causes Insulin Resistance: A Lesson from Animal Models of Impaired CEACAM1-Dependent Pathways

That impaired insulin clearance causes secondary insulin resistance is supported by our studies on genetically modified mouse models targeting the Ceacam1 gene globally or in a liver-specific manner. Using these mice, we have demonstrated that 1) CEACAM1 is a key player in promoting insulin clearance (43, 75, 173, 217), 2) by promoting insulin clearance, CEACAM1 regulates insulin sensitivity and links insulin to lipid metabolism (89, 149), and 3) reduced hepatic CEACAM1 level/function impairs insulin clearance to cause hepatic insulin resistance independently of lipolysis (75, 181).

Cc1^–/– mice with global null deletion (43) and AlbCre+Cc1^fl/fl mice with liver-specific deletion (75) of the Ceacam1 gene display chronic hyperinsulinemia caused by impaired insulin clearance at 2 mo of age without significant changes in insulin secretion, plasma glucagon levels, and mass of pancreatic β- or α-cells. Hepatic insulin resistance does not develop in both groups until ~6–7 mo of age, when both null deletions are propagated on a pure inbred C57BL/6J background (43, 75, 173, 217). Whereas Cc1^–/– mice develop hepatic insulin resistance with compromised ability of insulin to suppress hepatic gluconeogenesis, they do not develop fasting hyperglycemia, consistent with their intact pancreatic β-cell function (43).

Mutant mice with global (43, 217) or liver-specific deletion of Ceacam1 gene (75) and L-SACC1 mice harboring a liver-specific overexpression of the S503A dominant-negative phosphorylation-defective CEACAM1 mutant (173) also manifest hepatic steatosis concomitantly with hepatic insulin resistance, resulting from increased hepatic de novo lipogenesis and reduced fatty acid β-oxidation. Increased hyperinsulinemia-driven de novo lipogenesis [resulting from SREBP1c activation and the loss of the counter-regulatory effect of insulin on FASN activity (149)] tips the balance in favor of reesterification and redistribution of VLDL-triacylglycerol preferentially to white adipose tissue when mice are propagated on the C57BL/6J genetic background (75, 149, 157). This leads to visceral obesity and systemic insulin resistance in association with hyperinsulinemia-driven downregulation of the insulin receptor and compromised insulin signaling in extra-hepatic tissues such as hypothalamus, aorta, and heart, among others (181, 182). The progressive manifestation of metabolic abnormalities in AlbCre+Cc1^fl/fl mice with liver-specific deletion of Ceacam1 gene (75) demonstrates that hyperinsulinemia-driven blunted insulin signaling elevates hypothalamic FASN activity, which, together with hyperleptinemia, contributes to hyperphagia and, subsequently, to systemic insulin resistance and an increase in lipolysis (75). This provides evidence that hepatic insulin resistance in Ceacam1-null mutants develops independently of lipolysis, mainly as a consequence of hyperinsulinemia caused by impaired hepatic CEACAM1-dependent insulin clearance pathways.

The fact that loss of CEACAM1 in liver plays a critical role in the metabolic abnormalities of Cc1^–/– mice was further demonstrated by the restoration of normal metabolic phenotype, including hyperphagia, upon exclusive transgenic restoration of CEACAM1 in the liver and recovering normal insulin clearance. Given the role of CEACAM1 in suppressing lipogenesis (149), it is possible that loss of hepatic CEACAM1 induced hepatic de novo lipogenesis, which could in turn be associated with hepatic insulin resistance (163), independently of insulin clearance (110). Mechanistically, increased hepatic lipid production could also be followed by lipid redistribution to white adipose tissue and lipolysis. Lipolysis-derived free fatty acids, initially reaching the liver via the portal vein, could in turn cause systemic insulin resistance (161) that triggers compensatory increase in insulin secretion and hyperinsulinemia. This possible scenario would rule out a key role for impaired insulin clearance in the pathogenesis of insulin resistance in Cc1^–/– mice. However, blunting lipolysis by nicotinic acid treatment fails to restore insulin clearance and reverse hyperinsulinemia and systemic insulin resistance in Cc1^–/– mice (180). Nicotinic acid also fails to reverse hyperinsulinemia-driven downregulation of the insulin receptor level and activation in hepatic and extra-hepatic tissues (such as white adipose tissue and hypothalamus) in Cc1^–/– mice (181). This gain-of-function model (Cc1^{–/–xliver+} rescue mice) substantiates the findings of the liver-specific loss-of-function model (AlbCre+Cc1^fl/fl mice) that lipolysis is the consequence rather than the trigger of systemic insulin resistance in Cc1^–/– mice and that hyperinsulinemia stemming from impaired CEACAM1-dependent hepatic insulin clearance pathways plays a causative role in the pathogenesis of insulin resistance in Ceacam1 mutants.

Altered CEACAM1 Expression in Metabolic Diseases

In addition to insulin resistance, visceral obesity, and hepatic steatosis, L-SACC1 and Cc1^–/– mice develop features of non-alcoholic fatty liver disease (NAFLD) with hepatic inflammation (89, 148) and cardiovascular abnormalities such as hypertension and endothelial and cardiac dysfunction (94, 182). Rescuing CEACAM1 expression in the liver reverses these cardio-metabolic abnormalities (181, 182), strengthening the notion that reduction in hepatic CEACAM1 levels impairs insulin clearance to cause hyperinsulinemia, followed by cardio-metabolic anomalies. Consistently, hepatic CEACAM1 is downregulated and insulin clearance is impaired in several rat models of obesity, insulin resistance, and hepatic steatosis [such as obese Zucker hyperphagic rats without diabetes (fa/fa) or with diabetes (Zucker Diabetic Fatty rats; ZDF), and obese spontaneous hypertensive Koletsky rats (f/f)] (90). Similarly, rats selectively bred for low aerobic running capacity (LCR) exhibit all features of metabolic syndrome, including hyperinsulinemia, insulin resistance, visceral obesity, hypertension, endothelial dysfunction, and NAFLD, compared with age-matched high-capacity runners (HCR) (18, 202). Interestingly, this selection for low aerobic capacity lowers hepatic Ceacam1 mRNA (215) and protein levels (18), and, subsequently, hepatic insulin clearance in LCR relative to HCR. Whereas high-intensity interval endurance training for 8 wk restores the cardiovascular abnormalities in LCR, it fails to reduce hepatic de novo lipogenesis of LCR and restore its hepatic CEACAM1 levels to the level of HCR (87). In contrast, caloric restriction reverses all of the metabolic abnormalities in parallel to restoring hepatic CEACAM1 levels and insulin clearance (18).

We (90) and others (115) have shown that hepatic CEACAM1 is reduced in obese human subjects with insulin resistance and hepatic steatosis and that the severity of fat deposition in the liver is linear with the degree of the loss of hepatic CEACAM1 levels (115). Moreover, CEACAM1 protein content is remarkably reduced in primary hepatocytes derived from steatotic livers of obese humans relative to age-matched lean subjects, demonstrating an intrinsic loss of hepatic CEACAM1 in obesity (90). In light of the reduction of Ceacam1 transcription by fatty acid-activated PPARα (176, 177), lower CEACAM1 levels in hepatocytes from obese humans could result from the rise in plasma free fatty acids that are mobilized during lipolysis. Increased lipolysis-driven hepatic fatty acid β-oxidation in humans with uncomplicated obesity (81), and its role in regulating hepatic de novo lipogenesis (108, 203), propose an important role for the loss of hepatic CEACAM1 in the regulation of lipid homeostasis in hepatocytes. In fact, we have shown that high-fat feeding of normal C57BL/6 mice progressively represses Ceacam1 mRNA and protein levels in the hepatocyte by fatty acids’ activation of PPARα-dependent mechanisms (176) to provide a positive feedback mechanism on fatty acid β-oxidation and to reduce the burden of excessive energy supply by limiting its adverse effect on insulin sensitivity (180). However, when more than half of hepatic CEACAM1 is lost, insulin clearance followed by chronic hyperinsulinemia and hepatic steatosis develops (180). This is consistent with reduction of hepatic insulin clearance in dogs (214) and rats (161, 220) by the rise of free fatty acids in the portal circulation. In further support of the critical role of CEACAM1 along this adipocyte-hepatocyte axis in mediating dietary-induced insulin resistance, forced transgenic CEACAM1 overexpression (5) and adenoviral-mediated redelivery of CEACAM1 to the liver protect the liver against diet-induced impairment of insulin clearance and associated metabolic abnormalities (118, 180).

Molecular Mechanism of Insulin Degradation: The Role of Insulin-Degrading Enzyme (IDE) Revisited

Insulin-degrading enzyme (IDE; EC 3.4.99.45), is the most-abundant protease that degrades insulin in the cytosol, in particular in kidney (47, 48), although insulin-degrading enzyme activity also has been reported in lysosomes and membrane fractions. It has been proposed that cytosolic IDE, and other insulin proteases, may initiate the “specific” cleavage of intact insulin, yielding different intermediate products, which synergistically are degraded by non-specific membrane-associated proteases that fail to act on intact insulin (47, 48, 88, 92).

Insulin-Degrading Enzyme: General Information

IDE, also known as insulin protease, insulinase, insulysin, insulin-glucagon protease, neutral thiol protease, metalloendoprotease, or peroxisomal protease, is a ubiquitously expressed and well-conserved Zn²⁺ metallo-endopeptidase (8, 22, 49, 50). In rat liver, IDE is present in parenchymal cells, particularly near the portal tract, and in the epithelium of the bile duct (4).

IDE has a high affinity for insulin (K_m ~0.1 µM) and degrades insulin into several fragments (49, 50). However, despite its comparable affinity, proinsulin is a poor substrate of IDE (51). IDE can also degrade insulin growth factor (IGF) I and II in in vitro systems more slowly than insulin (3, 132). Thus proinsulin and IGF I could in principal serve as competitive IDE inhibitors (133). Albeit with lower affinity for insulin, IDE can also degrade glucagon (6, 47), amylin (15), β-amyloid (Aβ) peptide (112), Aβ precursor protein intracellular domain (APP-AICD) (53), amyloid Bri (ABri) and amyloid Dan (ADan) (135), cytochrome c (211), atrial natriuretic peptide (138), calcitonin and β-endorphin (67), transforming growth factor-α (TGF-α) (73), oxidized hemoglobin (61), growth hormone-releasing factor (206), and chemokine ligand (CCL)3 and CCL4 (123).

Molecular Mechanisms Underlying IDE Action

Receptor-bound insulin is internalized into hepatocytes resulting in its degradation or release of the intact or partially degraded hormone into circulation (85). IDE can degrade insulin in various cellular compartments, including the plasma membrane (219). However, most of the receptor-bound insulin is internalized into early/sorting endosomes where degradation is rapidly initiated in their neutral non-acidic environment mostly by IDE (84, 219). Subsequent maturation of late endosomes and their acidification allow dissociation of insulin from its receptor to undergo degradation (84) with no significant transfer to lysosomes (160). However, this proteolytic activity is not attributed to IDE but rather to acidic thiol protease (9), because cellular acidosis inactivates IDE by modulating its oligomerization state (79). In addition to intact insulin that escapes endosomal degradation, partially degraded insulin fragments are completely degraded in lysosomes (10, 11, 50). Degradation of internalized insulin may also occur in the cytosol, nucleus, and Golgi, proposing multiple intracellular pathways for insulin degradation.

The role of IDE in the endosomal degradation of internalized insulin remains controversial, suggesting a role for an alternative insulin-specific endosomal protease distinct from IDE that regulates hepatic insulin degradation (11). The rationale for this dispute is based on several observations. First, changes in endosomal pH produce different insulin degradation products, implying engagement of several enzymes with different pH-dependent enzymatic activities (37). Second, insulin dissociation from its receptor and its full degradation require an acidic pH that inactivates IDE (79). Third, IDE is mostly cytosolic, whereas endosomal degradation of insulin is membrane-associated, which implies that different proteases, including IDE, gain access inside the lumen of vesicles. It has been reported that a targeting sequence at the COOH terminus and an alternative splice isoform with extended NH₂ terminus serve to direct IDE to peroxisomes and mitochondria, respectively (117, 136). However, there are no identified specific sequences in IDE that could target it into the lumen of endosomes. It might be possible that plasma membrane-associated IDE is internalized together with the insulin receptor into the lumen of vesicles. Alternatively, proteolytic vesicles containing IDE might fuse with early endosomes. Interestingly, phosphatidylinositol (3-5)-triphosphate (PIP3), a second messenger produced by the intracellular insulin signaling pathway, stimulates the recruitment of IDE to early endosomes (194). However, this fails to explain how IDE gains access inside late endosomes. Some studies have proposed that this occurs during the micro-autophagy of cytosolic proteins due to IDE’s weak interaction with phosphatidylinositol phosphates (183, 194). On the other hand, insulin signaling-mediated localization of IDE to early endosomes could represent a way in which IDE regulates the sorting process of these vesicles, which, in turn, determines whether endocytosed insulin receptors are targeted to lysosomal degradation or plasma membrane recycling. However, this non-proteolytic function of IDE has not been confirmed experimentally. Together, this suggests that IDE may exert proteolytic as well as non-proteolytic activities in regulating hepatic insulin metabolism.

IDE and Insulin Clearance

IDE has been proposed to be a major enzyme involved in regulating insulin signaling and clearance (8, 49). Earlier studies demonstrated that ablation of Ide gene leads to hyperinsulinemia, supporting the notion of a physiological role for IDE in insulin clearance (1, 63). However, we recently demonstrated that liver-specific deletion of the Ide gene in mice (L-IDE-KO) does not result in hyperinsulinemia, as expected from deleting a protein that is postulated to play a major role in hepatic insulin clearance (207). Instead, it impairs insulin signaling by downregulating the surface membrane expression of the insulin receptor. These findings are consistent with a non-proteolytical function of IDE in the regulation of insulin action.

A novel insight on the role of renal IDE in regulating systemic insulin clearance has recently been reported (122). Insulin increases the association of IDE with sorting nexin 5 (SNX5), a cytoplasmic and a membrane-associated protein that regulates intracellular trafficking in the brush-border membrane of proximal tubules in human and rodent kidneys (122). Renal-selective depletion of SNX5 by the subcapsular infusion of Snx5-specific siRNA reduces IDE levels by ~40% and leads to a rise in circulating insulin levels with reduction in urinary insulin excretion. Moreover, silencing of Snx5 in mice decreased the expression level of the insulin receptor in renal proximal tubule cells, a major site of IDE action (121). Taken together, these data question the current notion that the main function of renal IDE in regulating systemic insulin clearance through its protease activity and prompt the formulation of a new paradigm whereby IDE regulates renal insulin clearance via a non-proteolytical function (i.e., regulating the level of insulin receptors on the brush border membrane of proximal tubule cells). Further research is warranted to fully elucidate the molecular mechanisms underlying the regulation of systemic insulin clearance by IDE.

Role of IDE in Insulin Clearance in Obesity

Controversy surrounds the impact of obesity on hepatic IDE levels and insulin clearance. Feeding female Swiss mice (10–12 wk old) a cafeteria diet for 8 wk causes an ~22% increase in body weight, glucose intolerance, and insulin resistance, and a reduction in insulin clearance in association with reduced hepatic protein levels of IDE (19). Interestingly, hepatic IDE mRNA levels from the canonical catalytic efficient 15a-IDE spliceoform and the catalytic inefficient 15b-IDE spliceoform were significantly reduced by the cafeteria diet (19, 62). In contrast, feeding female Wistar rats (10–12 wk old) a cafeteria diet for 17 wk causes hyperinsulinemia with insulin resistance in parallel to an increase in IDE mRNA levels and IDE activity, suggesting enhanced plasma insulin clearance (30). Feeding C57BL/6 mice (6–8 wk of age) a high-fat diet (58% kcal in fat) for 6 mo caused higher hepatic IDE protein levels and activity (209). Male Wistar rats fed a high-fructose diet (20% fructose in drinking water) for 6 wk developed obesity, hyperinsulinemia, hyperglucagonemia, hyperglycemia, and insulin resistance (57). However, hepatic IDE protein levels were significantly higher than in control rats (57). This was attributed to either a hyperinsulinemia-driven negative feedback mechanism or to hyperglucagonemia inducing hepatic IDE expression (57).

The discrepancies between these studies could arise from the use of different experimental models, as well as to the length and composition of diet. Nonetheless, no conclusive information could be drawn concerning the cause-effect relationship between IDE and impaired insulin clearance in the setting of obesity and diabetes.

Role of IDE in Insulin Clearance in Diabetes

Compelling evidence has demonstrated an association between reduced IDE levels and lower insulin clearance in Type 2 diabetics (19, 168, 206). Other studies have shown lower IDE activity and insulin degradation in both subcutaneous and visceral fat depots, with a greater extent in the visceral depot, of pre-diabetics and Type 2 diabetics compared with non-diabetic patients (65).

Conversely, IDE activity was increased in T2D patients treated with sulfonylureas (197). In human islets, IDE protein levels were higher in T2D patients treated with insulin than in those treated with oral hypoglycemic drugs (66). In contrast, well controlled Type 1 diabetics did not show changes in IDE activity (206). Moreover, the route of insulin administration can affect the response of IDE activity in diabetic patients, with subcutaneous but not intravenous injections being associated with changes in IDE activity (206).

The potential use of IDE inhibitors for the treatment of T2D has attracted a lot of attention in recent decades. However, despite the discovery of a variety of compounds with diverse chemical structures, none of them demonstrated valid therapeutic use due to toxicity, low potency or selectivity (see Ref. 167 for a comprehensive review). For instance, long-term inhibition of IDE may cause chronic hyperinsulinemia, resulting in secondary insulin resistance and impaired insulin secretion. Furthermore, hyperinsulinemia could be related to the upregulation of cell proliferation in tissues such as liver. In addition to insulin, IDE degrades glucagon, amylin, and Aβ. It is uncertain whether inhibiting IDE could increase hepatic glucose output and glucose intolerance due to altered insulin-to-glucagon ratio. Acute inhibition of IDE by 6bK seems to predict this unfavorable scenario (126). Likewise, deposition of amylin in pancreas and of Aβ in the brain could, respectively, cause cytotoxicity in pancreatic β-cells in parallel to reduced insulin secretion and to brain damage due to neuroinflammation. In summary, whether acute or long-term inhibition of IDE activity constitutes an effective strategy to regulate insulin clearance remains to be established.

Proposed Coordination of Insulin Trafficking and Metabolism by CEACAM1 and IDE-Dependent Pathways

Activation of the insulin receptor and intracellular signal transduction are intimately associated with a highly regulated receptor-mediated insulin internalization and removal process. Signal transduction by receptor-bound insulin is not confined to the cell surface but continues upon the entry of the endocytosis complex into early endosomes, with the α-subunit of the receptor being in the vesicular lumen and its tyrosine kinase on IRβ exposed to substrates in the cytoplasm. This feature allows the insulin receptor to continue to target its downstream substrates to transmit insulin signaling while it carries its cargo for removal from the circulation.

CEACAM1 has been assigned a major role in promoting the rate of uptake of insulin via its receptor and its translocation to the degradation process in hepatocytes as well as in isolated renal proximal tubule cells (5). Insulin degradation has been ascribed to proteolytic enzymes, among which IDE has been postulated to play a major role, in particular in kidneys (50). Intact insulin clearance in L-IDE KO mice has challenged this paradigm and revealed a non-proteolytic activity of IDE involving maintenance of insulin signaling (207). Although IDE may play a cooperative role with CEACAM1 in targeting the insulin-IR complex through its intracellular trajectory, it appears to contribute to a minor extent to insulin degradation in the early endosomes in hepatocytes. The novel conceptual model of the coordinated regulation of insulin metabolism by CEACAM1 and IDE is captured in FIGURE 2. According to this model, insulin binding to its receptor leads to a conformational change that induces autophosphorylation of the insulin receptor. Alhtough it has not been demonstrated, it is possible that cytosolic IDE participates in a conformational change that favors this step, similar to its interaction with the ~50 amino acid cytoplasmic tail of the Type A scavenger receptor A (26), and with androgen and glucocorticoid receptors (111).

Nonetheless, insulin-activated IR-A phosphorylates CEACAM1 that, in turn, takes part in the formation of a stable insulin-IR complex via Shc, which, together with Tyr513 in the COOH-terminus tail of CEACAM1, targets the complex to AP2-associated clathrin-coated pits and early endosomes (FIGURE 2, step 1). It is possible that IDE located extracellularly and on the plasma membrane is sequestered inside the vesicle during this process (25).

In parallel, insulin induces the production of PIP3 from phosphatidylinositol by PI3Kinase to stimulate IDE recruitment to the outer layer of the early/sorting endosomes (194) and to retain CEACAM1 in endosomes (199). Activation of PI3Kinase likely results from the progressive binding of SHP2 to CEACAM1, an event that sequesters the phosphatase and prevents dephosphorylation of IRS (146). In this manner, CEACAM1 contributes to the recruitment of a cytosolic IDE pool to the outer side of early endosomes while prolonging insulin signaling via the IRS/PI3Kinase pathway as a compensatory mechanism to insulin degradation by vesicular IDE (FIGURE 2, step 2). Binding of CEACAM1 to SHP2 requires the progressive loss of Shc-CEACAM1 interaction and, subsequently, progressive detachment of CEACAM1 from the insulin-IR complex by binding to FASN on Tyr488 residue. It is possible that IDE in the outer layer of the vesicle contributes to microtubule reorganization and polymerization to form the CEACAM1-FASN vesicle. This remains to be investigated. Nonetheless, as the insulin-IR complex is destabilized in part by the detachment of CEACAM1, and as the endosome matures and undergoes acidification (FIGURE 2, step 3; H⁺), insulin dissociates from its receptor to be degraded by the acidic thiol protease (FIGURE 2, step 3; AcTP) and/or by the aspartic acid protease cathepsin D (9), but not by IDE that is inactivated by the acidic milieu of late endosomes (FIGURE 2, step 3).

Insulin receptor, free of its hormone and of CEACAM1, is recycled back to the plasma membrane (FIGURE 2, step 4). The pool of IDE-bound to the outer side of late endosomes might interact with α-synuclein and/or other proteins involved in regulating vesicular trafficking of IR-A back to the plasma membrane. That a non-proteolytical activity of IDE is involved in vesicular trafficking is supported by the observation that IDE binds to the COOH terminus of α-synuclein, an amyloidogenic protein, to reduce the formation of α-synuclein oligomers (190, 191), thereby increasing tubulin polymerization and microtubule-dependent trafficking for proper autophagy (32, 64, 114). In fact, depletion of IDE from pancreatic β-cells is associated with increased levels of α-synuclein, in parallel to reduced tubulin polymerization, autophagy, and microtubule-dependent recruitment of secretory granules containing insulin (198). Moreover, insulin increases the association of SNX5 with IDE in the plasma membrane of renal proximal tubule cells (122), and depletion of renal SNX5 results in reduced expression of the insulin receptor and lower signal transduction (122). Because IDE is the primary enzyme responsible for insulin degradation in kidney, the dynamic interaction between IDE and SNX5 regulates systemic circulating insulin levels (122). Thus IDE might similarly regulate the levels of α-synuclein oligomers in hepatocytes to facilitate microtubule polymerization and trafficking of late endosomes (with their cargo) to the plasma membrane.

The proposed model compartmentalizes the coordinated functions of CEACAM1 and IDE in regulating insulin disposal. CEACAM1 regulates insulin action by promoting receptor-mediated insulin endocytosis and targeting to the degradation process. Although IDE initiates the degradation of insulin in early endosomes, its main effect on insulin action appears to be mediated by its non-proteolytic activity on the intracellular trafficking of the insulin-IR complex and, in particular, on insulin receptor recycling to the plasma membrane, an important step in insulin retro-endocytosis. Reduced expression level and phosphorylation of the insulin receptor (and CEACAM1) on the membrane of hepatocytes derived from L-IDE-KO mice and their hepatic insulin resistance are consistent with the significant non-proteolytic regulation of insulin action by IDE (207). Low IR level in hepatocytes isolated from these mice could also stem from increased trafficking of IR-A to lysosomes. More studies are needed to delineate the underlying mechanisms.

Concluding Remarks

Insulin secretion and clearance coordinate their regulation of insulin homeostasis and action. In this respect, the liver is equipped with multiple mechanisms that protect it against the high levels of insulin in the portal vein. This protective repertoire is mediated by the pulsatility of insulin release and by the high level of hepatocytic CEACAM1 that mediates excess insulin removal upon its phosphorylation by ligand-activated insulin receptor to maintain normo-insulinemia. Through targeting the insulin-IR complex to the intracellular degradation process, CEACAM1 also mediates a negative effect of the acute pulses of insulin on FASN activity. This restricts hepatic de novo lipogenesis and limits hepatic steatosis in the face of the higher level of insulin in the portal than systemic circulation. Thus, when hyperinsulinemia develops and pulsatility of insulin release diminishes, CEACAM1 phosphorylation is compromised to impair insulin clearance, causing hepatic insulin resistance (downregulation of the insulin receptor by lysosomal degradation) and giving way to the lipogenic effect of chronically elevated levels of insulin. Through substrate redistribution to white adipose tissue, elevated hepatic de novo lipogenesis triggers visceral obesity followed by systemic insulin resistance. Thus we posit that insulin resistance is not selective with respect to insulin’s ability to acutely reduce hepatic gluconeogenesis, as has been widely accepted (23, 38), but it also includes its inability to acutely suppress hepatic de novo lipogenesis. Nonetheless, our studies show that impaired CEACAM1-dependent insulin clearance pathways link hyperinsulinemia-driven systemic insulin resistance to hepatic steatosis (as in NAFLD).

Mechanistically, hepatic insulin clearance is mediated by receptor-mediated insulin uptake and degradation in hepatocytes, a process that is upregulated by CEACAM1. Devoid of enzymatic activity, CEACAM1 does not engage in the proteolytic degradation pathway, per se. Instead, it acts as a chaperone that delivers the complex to late endosomes where insulin ultimately dissociates from its receptor to be fully degraded. Whereas IDE catalyzes insulin degradation in early endosomes, its proteolytic activity is inhibited in the acidic environment of late endosomes, giving way to resident acidic proteases to degrade insulin. Normo-insulinemia in mice with liver-specific deletion of the Ide gene rules out a major role for IDE in insulin degradation. Instead, it proposes a non-proteolytic regulation of insulin action by IDE, namely via promoting the surface content of insulin receptor in hepatocytes. This is consistent with disagreement over the therapeutic effect of IDE inhibition as a strategy to ameliorate insulin resistance. In light of the undisputed critical role of CEACAM1 in regulating insulin action by promoting insulin clearance, we propose that therapy based on inducing CEACAM1 is likely to be more beneficial in the prevention as well as in the treatment of insulin resistance and hepatic steatosis. In support of this notion, hyperinsulinemia in patients with NAFLD appears to be more strongly associated with impaired insulin clearance than with increased insulin secretion (21). Upregulation of CEACAM1 transcription by agonists of PPARγ (pioglitazone) and GLP-1R (exenatide) (74) strengthened the rationale for the use of these drugs in the treatment of T2D and NAFLD (20, 31, 40). Studies are underway to document changes in hepatic CEACAM1 levels in patients with metabolic syndrome, but reduced insulin clearance is emerging as a major risk factor for this disease, including NAFLD (21, 77, 113, 166). This provides impetus to carry out further studies to validate reduced hepatic CEACAM1 level as a biomarker for metabolic syndrome, with an overarching goal to target insulin clearance as a therapeutic strategy against this complex disease.