Insulin Receptor Isoforms in Physiology and Metabolic Disease

Abstract

Insulin receptors (IRs) are ubiquitously expressed and essential for all cell types. Their signaling cascades are connected to key pathways involved in cell metabolism, proliferation, and differentiation, amongst others. Thus, dysregulation of IR-mediated signaling can lead to diseases such as metabolic disorders. In mammals, the IR pre-mRNA is alternatively spliced to generate two receptor isoforms, IR-A and IR-B, which differ in 12 amino acids in the α-chain involved in ligand binding. Given the isoforms have different affinities for their ligands insulin, proinsulin, and insulin-like growth factors (IGFs), it is speculated that IR amount and splicing regulation might contribute to a change in IR-mediated effects and/or insulin resistance. The aim of this chapter is to increase awareness of this subject in the research fields of diseases characterized by disturbances in insulin signaling. Here, we will describe the IR isoform distribution and discuss the current knowledge of their expression and ligand binding affinities as well as their signaling in physiology and during obesity and type 2 diabetes in humans and animal models. Moreover, we will discuss the necessary steps to gain a better understanding on the function and regulation of the IR isoforms, which could result in future therapeutic approaches against IR-related dysfunction.

1. Introduction

In invertebrates, one ancestral gene, DAF-2, encodes one receptor that binds insulin-like peptides [1]. With the emergence of vertebrates, three distinct receptors appeared, namely: the insulin receptor (IR), the type 1 insulin-like growth factor receptor (IGF1R), and an orphan receptor called the insulin receptor-related receptor (IRR). The genes encoding these receptors share similar genomic organization, with conserved α and β protein chains that are synthesized from one single pre-mRNA (reviewed in [2]). While originally all three receptors were formed by 21 exons, both IRR and IR acquired independently one extra exon, namely exon 11. The IRR exon 11 can be traced back to amphibians, whereas IR exon 11 is found exclusively in mammals [3, 4]. Thus, exclusively in mammals, the exon 11 of the IR gene is alternatively spliced to produce two protein isoforms called IR-A and IR-B. To trace the origin of IR isoforms, Hernández-Sánchez et al. analyzed their transcripts in different species by reverse transcriptase-polymerase chain reaction (RT-PCR) analysis [4]. In mouse tissues, they found tissue-specific expression of both IR-A and IR-B, while in chicken and frog tissues, only the IR-A isoform was detected [4]. The physiological reason for the evolutionary acquisition of the IR-B isoform in mammals is unclear. It seems that the IR-B provided the receptor with higher specificity for insulin and poor binding of other possible ligands of IR-A, which will be discussed later.

2. Characterization of the IR isoforms

2.1 Structure and assembly

In humans, the insulin receptor gene (INSR) maps to human chromosome 19 (in mice, it maps to chromosome 8) and spans more than 120 kb [5]. The insulin receptor complementary DNA (cDNA) was cloned in 1985 by two independent groups [6, 7], giving two different lengths and indicating two isoforms, dependent on the inclusion (IR-B) or exclusion (IR-A) of exon 11. The 36 base pairs of exon 11 (that account for the 12 amino acid difference) encode a portion of the C-terminus of the α-subunit in the vicinity of the ligand-binding domain (reviewed in [8]), resulting in isoform-specific properties of the receptors. A linear α-β amino acid sequence (IR pro-receptor precursor) is translated from the IR mRNA and includes a signal sequence at the N-terminus to enter the endoplasmic reticulum [9]. After cleavage of the signal peptide, the inter-α-chain disulphide dimerization occurs, forming the β-α-α-β structure of the IR [10]. The insulin pro-receptor is further processed in the Golgi apparatus by the protease furin, and the mature IR is then trafficked and inserted in the plasma membrane [11].

The IR and IGF1R belong to the same subgroup of receptor tyrosine kinases and can form either homo-receptors (two IR α-β subunits) or hybrid receptors, consisting of one IR α-β subunit linked to one IGF1R α-β subunit. Furthermore, the two IR splice variants enable the formation of both homo-dimers (IR-A/IR-A) or hetero-dimers (IR-A/IR-B), and similarly, two modalities of hybrid receptor (IR-A/IGF1R and IR-B/IGF1R). Hybrid receptors have been detected in all tissues and cell lines that express both receptor types [12] and it is presumed that both IR-A and IR-B isoforms are equally capable of forming hybrids with IGF1R [13]. The factors regulating their assembly are unknown; however, there is evidence to suggest that the formation of homo-receptors and hybrid receptors is proportional to the relative concentrations of each receptor type [12, 14, 15].

Crystal structures of the IR were determined in 1994 [16] and 2006 [17] and refined in 2016 [18]. Single-particle cryo-electron microscopy has since been used to explore receptor conformations and ligand-receptor complexes [19, 20, 21]. It is worth pointing out that these studies used the IR-A isoform to reconstruct and represent the IR. The functional IR consists of two covalently linked IR monomers, that is, two extracellular α-subunits linked by disulphide bonds and two transmembrane-spanning β-subunits. The α-subunit contains either 719 (IR-A) or 731 (IR-B) amino acids and has a molecular mass of approximately 130 kDa. This subunit is entirely extracellular and contains the ligand-binding sites. The transmembrane-spanning β-subunit contains 620 amino acids, has an approximate molecular mass of 95 kDa and is composed by extracellular, transmembrane, and cytosolic domains. The latter domain contains the receptor’s tyrosine kinase, which is activated by ligand-binding and conformational change of the IR. Two insulin-binding sites are located in the extracellular α-subunit of the IRs. The primary insulin-binding site (site 1) is formed from elements of the L1 domain and a C-terminal peptide of the α-subunit [22, 23, 24, 25, 26, 27]. The second insulin binding site (site 2) has lower ligand binding affinity and is formed from residues in the first and second type III fibronectin repeats [28]. A model for insulin binding to the IR has been showed, in which one single insulin molecule simultaneously engages site 1 of one α-chain and site 2 of the other, thus bridging the two IR monomers; while a second insulin molecule binds to the equivalent, symmetry-related site 1′, creating a second bridging with site 2′. The two insulin molecules effectively crosslink the two IR monomers and thereby activate the IR [8, 20, 21]. As to the possible implications of the differences in the α-chain C-terminal domains of the IR isoforms, comprehending the significance of the 12 extra amino acids is hampered by the lack of structural data using the IR-B isoform. Thus, only secondary structure predictions can be made. It can be inferred that the 12-amino acid fragment of IR-B is most certainly the reason for the lower binding affinity of insulin-like growth factor 1 (IGF1) and insulin-like growth factor 2 (IGF2) toward this receptor isoform in comparison to IR-A (as discussed in Section 2.2). Based on this, Menting et al. speculate that the additional residues in the IR-B α-chain C-terminus are devoid of secondary structure, thus making this structure longer and hindering steric accommodation of these ligands, with a similar situation to be expected regarding insulin binding [29]. Whittaker et al. used alanine-scanning mutagenesis of insulin binding site 1 of IR isoforms A and B transiently expressed in cells to study their insulin binding properties [30]. They found several mutations that compromised insulin binding, some of which produced differential effects between the two receptors, either reducing affinity or inactivating one specific isoform [30].

2.2 Ligand binding affinities of the IR isoforms

The different structures of the IR isoforms are responsible for their functional differences, for example, disparity in ligand affinities, internalization and recycling kinetics, signal transduction, and the activation of specific cellular pathways.

Insulin is the main ligand for the IR, although the receptor also binds IGF1 and IGF2, as well as proinsulin (hormone precursor to insulin) (Figure 1). Some groups set out to study the different ligand binding affinities of the IR isoforms and came to different conclusions. To measure isoform-specific ligand binding, the main technique used throughout the different studies was competition for radiolabeled insulin. All studies were conducted using human IR cDNA in mouse or rat cell lines (referred to as hIR-A and hIR-B), probably aiming to translating the findings in humans. Mosthaf et al. expressed hIR-A and hIR-B in Rat-1 cells and found that IR-A had ~2-fold higher affinity for insulin than IR-B, both in intact cells and using detergent solubilized, partially purified receptors [31]. In agreement, Kellerer et al. used partially purified receptors from Rat-1 cells and found that IR-A displayed a higher affinity for insulin compared to IR-B [32]. Accordingly, Yamaguchi et al. also reported a ~2-fold higher affinity for insulin in intact Chinese hamster ovary (CHO) cells expressing hIR-A, in comparison to those expressing hIR-B [33]. A second study by this group found a faster insulin association rate to hIR-A receptors in intact CHO cells, as well as an accelerated insulin dissociation from hIR-B receptors, proposing a biochemical basis for the differential ligand biding affinities [13]. A similar faster dissociation of insulin from hIR-B that could be responsible for the lower affinity of this receptor was shown for Rat-1 cells [34]. To note, contrarily to all studies that used stimulated native receptors on intact cell membranes, a single work using solubilized recombinant receptors found no significant difference between the affinities of the two IR isoforms for insulin [30].

Figure 1.

IR-A, IR-B, and IGF1R monomers, and their combinatorial possibilities of dimerization, with their possible ligands. The size of the different ligands represents the relative affinity for a given receptor. IGF1, insulin-like growth factor 1; IGF1R, type 1 insulin-like growth factor receptor; IGF2, insulin-like growth factor 2; IR, insulin receptor; IR-A, insulin receptor isoform A; IR-B, insulin receptor isoform B.

IR isoform affinities for IGF1 and IGF2 have also been investigated. Frasca et al. used R-cells, a mouse fibroblastic cell line that lack IGF1R, expressing either hIR-A or hIR-B [35]. They reported that IR-A, but not IR-B, binds IGF2 with high affinity (comparable to that of insulin). Further, IGF2 bound to IR-A with similar affinity to that of IGF2 to IGF1R [35]. Using the same cellular system, Sacco et al. reported that IR-A bound IGF2 with high affinity (4-fold lower than that for insulin), whereas IR-A’s affinity for IGF1 was 30-fold lower than that for insulin [36]. Proinsulin binding has been less studied compared to the other ligands and its ability to bind differentially the two IR isoforms as well as its possible signal transduction remain an enigma. One study in intact R-cells showed that proinsulin binds and activates both IR isoforms, but had a higher affinity for hIR-A than for hIR-B. Authors report that, similar to IGF2, proinsulin effectively stimulates cell proliferation and migration and curiously had no activity toward IGF1R or IR/IGF1R hybrid receptors [37]. Conversely, McClain’s work (mentioned previously) conducted in intact Rat-1 fibroblast cells found that hIR-A and hIR-B bound proinsulin with the same relative affinity [34].

Few studies have addressed the ligand affinities for the different hybrid receptors (IR/IGF1R). Using competition for tracer-labeled insulin and an enzyme-linked immunosorbent assay (ELISA)-based method, Slaaby et al. found that IR-A/IGF1R and IR-B/IGF1R hybrid receptors respond 20 to 50 times more effectively to IGF1 than to insulin [38, 39]. The increase in IGF1R expression and thereby its incorporation into hybrid formation with IR has prompted a potential role of hybrid receptors in reducing cell insulin responsiveness. Studies in CHO cells suggest that hybrids between IGF1R and both IR isoforms have low binding affinity for insulin and high affinity for IGF1 and IGF2 [40]. Another study in R-cells showed that IR-B/IGF1R receptors had high affinity only for IGF1, whereas hybrid IR-A/IGF1R receptors also bound IGF2 and insulin [41].

In summary, the accumulated knowledge suggests that IR-A has higher affinity for insulin, IGF1 and IGF2 compared to IR-B and that hybrid receptors show a preferential affinity for IGF1 (Figure 1). This could be due to the availability/inaccessibility of the different binding sites in hybrid receptors. The majority of these affinity studies were conducted using mainly tracer-labeled ligand technique and the works on binding affinity for insulin to IR-A/IR-B are dated more than two decades ago. Molecular chemistry methods have advanced since then and it could be of interest to verify these affinity studies with improved tools comparing primary cells (with double knockout of IR and IGF1R) where only one IR isoform (of the same species of the cells) is expressed at a time. Moreover, ligand binding could potentially be altered by different post-translational modifications of the α-chain such as specific glycosylation patterns and different lipid raft composition, which could both vary between cell types and under different culture conditions.

2.3 Functional differences between the IR isoforms

The IR isoforms seem to display equal receptor activation and kinase activity triggered by the binding of insulin. McClain et al. investigated insulin-stimulated tyrosine kinase activity in solubilized hIR-A and hIR-B receptors by exposure to insulin and radioactive ATP and found similar accumulated radioactivity in the Tyr-phosphorylated receptors [34]. On the contrary, Kellerer et al., preparing equal amounts of solubilized hIR-A and hIR-B, found higher radioactivity for hIR-B (2.5-fold) after stimulation with insulin and phosphorus-32 [32]. However, when performing the same experiment on native receptors in human embryonic kidney (HEK) intact cell membrane transiently expressing the isoforms, they detected no difference in tyrosine kinase activity. Neither did they report differences when they used short-time trypsinization to cleave the α-subunit and activate the tyrosine kinase [32]. The latter data suggests that solubilized receptors were able to undergo different activation compared to receptors integrated in the plasma membrane of intact cells, and that differences in the isoform α-subunit structures were responsible for their different kinase activities.

Few studies on the kinetics of IR isoform-specific internalization have been published. Yamaguchi et al. showed that in CHO cells expressing the isoforms, hIR-A displayed a 25% higher rate of ligand-stimulated internalization in comparison to hIR-B [33]. Further, work in Rat-1 fibroblasts showed that in cells expressing hIR-A, the maximum internalization reached ~65% after 10 minutes, followed by a high recycling rate of ~80% of internalized receptors after 20 minutes. In hIR-B expressing cells, the maximum internalization was ~60% and was reached within 15 minutes; however, no recycling was detectable within 30 minutes [42]. Clearly, these few data in specific cell lines warrants future research to dissect the different kinetics of internalization between the two IR isoforms.

Upon ligand-binding the IRs transduce diverse signaling pathways, which culminate in cellular functions ranging from glucose, lipid and protein metabolism to cell differentiation, proliferation and apoptosis. Insulin binding causes autophosphorylation and activation of the IR, which in turn allows the binding and activation of diverse downstream effectors. The availability and/or recruitment of specific binding partners could lead to distinct signal transductions and to the consequent activation of different pathways resulting in different biological endpoints/responses in different cell types. Up until now, only few cell types and IR isoform-specific signal transduction pathways have been investigated, especially in primary cells that express both isoforms. In the insulin-producing pancreatic β-cell, Leibiger et al. showed that the binding of insulin to IR-A or IR-B results in selective transcriptional activation of different target genes [43]. Insulin gene transcription was promoted through IR-A and the activation of PI3K class Ia/p70s6k-mediated signaling, while transcription of the glucokinase gene by signaling through IR-B PI3K class II-like activity and PKB [43]. In a subsequent study from the same group, Uhles et al. showed that isoform-specific insulin receptor signaling involves different plasma membrane domains [44]. By using tagged IR isoforms in a hamster β-cell line, they found that mutation of certain amino acids encoded by exon 11 resulted in both loss of signaling and shift in IR isoforms localization in the plasma membrane, suggesting an isoform-specific sorting to different microdomains of the plasma membrane [44]. Later, they demonstrated that spatial segregation allows simultaneous and selective signaling via IR-B in the same cell. They showed that in the pancreatic β-cell, insulin activated the glucokinase gene from plasma membrane-standing IR-B, while c-Fos gene activation was dependent on IR-B signaling from early endosomes [45]. In a following work, Leibiger et al. further confirmed this hypothesis by showing the co-distribution of IR-B, but not IR-A, with two proteins involved in signal transduction, PI3K-C2α and PKBα/Akt1, in the same plasma membrane microdomains [46]. In the only other system that studied the possible signaling differences between IR isoforms in primary cells, a recent work in human podocytes found that the lipid raft enzyme sphingomyelin phosphodiesterase acid-like 3b (SMPDL3b) interfered with the ability of IR-B to bind caveolin-1, thus interrupting signal transduction through this isoform. This demonstrated a mechanism via which sphingolipids may affect IR localization at the plasma membrane and IR signaling in an isoform-specific manner [47].

In summary, the small 12 amino acid difference between the IR isoforms is responsible for the differences in their function in studied cells where both receptors are expressed simultaneously. Thus, it will be pivotal in future research to address and consider the existence and distinction of two IR isoforms when studying insulin signaling in specific cell types.

3. Tissue and cell type IR isoform expression

The IR isoform expression is regulated in a developmental and tissue-specific manner (Figure 2). In human, adult tissues associated with the known metabolic effects of insulin, such as the liver, adipose tissue but also kidney, IR-B is the predominant isoform [48, 49, 50, 51]. The IR-A isoform is highly expressed in fetal tissues—where it enhances the effects of IGF2 during embryogenesis and fetal development [52]—and in several adult tissues, such as brain [53], spleen [31], ovary [54] and testis [55]. The up-regulation of IR-A during adult life has been associated with mitogenic effects and has been described in a wide variety of cancers (reviewed in [56]). Other tissues express both isoforms in closer proportions, such as in pancreatic islets [57] and skeletal muscle [49]. Of note, IR isoform tissue distribution is generally conserved amongst mammals, with some differences, as shown in Table 1 [31, 50, 51, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73].

Figure 2.

IR isoform expression in tissues and cell types. IR, insulin receptor; IR-A, insulin receptor isoform A; IR-B, insulin receptor isoform B.

			Brain						Adipose tissue						Liver		Kidney		Spleen		Heart		Duodenum
		Species			Perigonadal		Mesenteric		Retroperitoneal		Subcutaneous		Bat
Author	Year		A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B
Besic et al.	2015	H													18	82
Kaminska et al.	2014	H									40	60
Mosthaf et al.	1990	H													40	60	50	50	100	0
Seino and Bell	1989	H	100	0			70	30							15	85	40	60	100	0
Sesti et al.	1994	H					45	55							15	85
Vienberg et al.	2011	H									30	70
Norgren et al.	1994	H									25	75			10	90
Escribano et al.	2009	M													40	60
Vienberg et al.	2011	M	95	5	22	78	25	75	38	62					5	95	5	95	95	5	60	40
Muller et al.	2007	M
Moruzzi et al.a	2021	M	95	5	15	85					20	80	6	94	6	94	8	92	88	12	73	27	18	82
Huang et al.	1994	Mo	95	5			70	30							35	65	70	30			60	40
Vienberg et al.	2011	P	95	5			85	15	90	10	80	20			18	92	50	50	80	20	30	70
Amessou et al.	2010	R	95	5											8	92	30	70
Serrano et al.	2005	R			40	60			50	50			40	60	0	100	30	70			50	50
Vienberg et al.	2011	R	95	5	22	78	60	40	30	70	65	35			10	90	18	82
Vidal et al.	1995	R			59	41									2	98					34	66
McGrattan et al.	1998	S			29	70					29	70			20	80
									Muscle										Pancreas		Skin		Stomach
		Species	Hind limb		Rectus abdominis		Trapezius		EDL muscle		Soleus		GM		Vastus lateralis		White quadriceps		Islets
Author	Year		A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B	A	B
Hribal et al.	2003	H																	50	50
Malakar et al.	2016	H																	42	58
Mosthaf et al.	1990	H																	80	20	90	10
Norgren et al.	1993	H													30	70
Sell et al.	1994	H													18	82
Norgren et al.	1994	H			25	75
Malakar et al.	2016	M																	35	65
Vienberg et al.	2011	M							55	45	80	20
Moruzzi et al.a	2021	M									61	39	75	25					65	35
Huang et al.	1994	Mo													55	45							65	35
Vienberg et al.	2011	P					20	80	20	80	20	80
Serrano et al.	2005	R									100	0					100	0	80	20
Vienberg et al.	2011	R							90	10	90	10
Vidal et al.	1995	R	99	1
McGrattan et al.	1998	S			27	73

Table 1.

IR isoform mRNA expression in humans and animal models expressed as percentage of IR-A and IR-B.

^a

Control for the high-fat high-sucrose diet in that specific study.

BAT, brown adipose tissue; EDL muscle, extensor digitorum longus muscle; GM, gastrocnemius muscle; H, human; IR, insulin receptor; IR-A, insulin receptor isoform A; IR-B, insulin receptor isoform B; M, mouse; Mo, monkey; P, pig; R, rat; S, sheep.

Because every tissue is composed of a mix of different cell types, findings regarding the expression pattern of IR isoforms cannot be extrapolated without considering the specific cell types forming the tissue. For example, analysis of liver tissue shows ~90% of IR-B expression, suggesting that hepatocytes may express exclusively IR-B. However, this tissue also contains other cell types, such endothelial cells which are known to express predominantly IR-A [70, 74], and Kupffer cells (the resident macrophages in the liver) in which IR isoform expression has not been studied. Another example is the brain, where tissue analysis shows mainly IR-A expression; however, predominant IR-B expression has been described in human astrocytes [75]. Regarding other primary cells, Muller et al. applied single-cell RT-PCR to elucidate IR isoform distribution in human pancreatic islet cells and, notably, found no expression of IR-B in isolated α-cells [69]. Mouse adipose tissue resident macrophages have been found to express both IR-A and IR-B, while mouse lymphocytes and monocytes express only a low amount of IR-B. Of note, these cells were analyzed as a bulk sample after magnetic-column cell sorting, with the possibility of low contamination of other cell types that might express IR-B [70].

IR-A is predominant in progenitor and precursor cells, whereas IR-B is more abundant in differentiated cells (Figure 2). Different studies reported high levels of IR-A in brown and white pre-adipocytes, osteoblast precursors, monocytes, neural progenitors and intestinal epithelial stem cells, compared with the high IR-B levels that characterize their differentiated cell counterparts [70, 72, 76, 77, 78, 79]. Some studies have reproduced this IR-A to IR-B switch favoring cell differentiation in vitro by overexpressing IR-B, for example in murine hemopoietic [80] and human colorectal cancer cell lines [79], or by treating HepG2 hepatoma cells with dexamethasone, which is known to maintain an adult hepatocyte phenotype [81]. These studies suggest that alternative splicing of the IR gene is a highly regulated process during differentiation and could play an important role in cell specialization.

In summary, the IR isoforms have specific tissue and cell-type distribution, and deviations from the wild-type IR-A/IR-B ratios may affect the fine-tuning of insulin signaling, disturbing metabolic and mitogenic pathways and compromising cell function.

4. Evidence of IR isoform roles using tissue-specific IR-knockout models

Insulin receptor knockout (IR-KO) mouse models have been developed to study the function of the IR, and the reconstitution of IR signaling with only one type of receptor (IR-A or IR-B) has been exploited to understanding the distinct roles of the IR isoforms. Mice with a global deletion of IR are born with normal features and with only slight growth retardation. However, shortly after birth, metabolic control rapidly deteriorates, glucose levels increase upon feeding, and insulin levels rise up to 1000-fold above normal and the animals die of diabetic ketoacidosis within 48-72 h [52, 82]. This phenotype clearly indicates that the IR is necessary for postnatal glucose homeostasis but is not essential for prenatal growth.

Using IR-KO mice, Okamoto et al. demonstrated that lethality and diabetes in IR-KO mice could be rescued by reconstitution of IR-B in three organs: liver, brain, and pancreatic β-cells [83]. However, reconstitution of IR-B in only glucose transporter type 4 (GLUT4)-expressing tissues (e.g., muscle, fat) did not rescue the phenotype [84]. Their data suggests that insulin signaling in liver, brain, and pancreatic β-cells as insulin-target tissues is sufficient to prevent diabetes. The stable expression of either IR-A or IR-B in IR-KO tissue has been used as a platform by Benito’s group to study the differential role of the IR isoforms in the liver [85]. They developed adeno-associated viral vectors encoding IR-A or IR-B targeted to the liver and showed that hepatic expression of IR-A in inducible liver insulin receptor knockout (iLIRKO) mice could increase hepatic glucose utilization, thereby decreasing hyperglycemia and ameliorating the diabetic phenotype [85]. In another recent in vivo study, the same group explored whether overexpression of IR-A via adeno-associated viral vectors in the liver without ablation of the endogenous IR, could improve glucose homeostasis in a mouse model of high-fat diet-induced obesity [86]. Again, they observed that IR-A expression, but not IR-B expression, induced increased glucose uptake in the liver, improving insulin tolerance [86]. Of note, in both these studies a comparison of the expression levels of IR-A/B isoforms in the liver, achieved by the viral vectors, was not reported. In line with the in vivo findings, Nevado et al. generated immortalized neonatal hepatocyte cell lines from iLIRKO mice and reconstituted IR signaling with retroviruses encoding for human IR-A or IR-B [87]. The expression of IR-A, but not IR-B, in the iLIRKO immortalized neonatal hepatocytes restored basal glucose uptake to wild-type levels, indicating that IR-A works as a GLUT1/2-associated cotransporter to facilitate glucose uptake [87]. Consistently, Diaz-Castroverde et al. demonstrated that in iLIRKO immortalized neonatal hepatocytes, IR-A expression, but not IR-B expression, enhanced insulin signaling associated with elevated glycogen synthesis and storage [88]. It is worth noting that the primary neonatal hepatocytes per se might have a high IR-A/IR-B expression, which was not described. Moreover, primary hepatocytes isolated from neonatal mice were immortalized prior to the induction of IR gene knockout and reconstitution with hIR-A or hIR-B. The immortalization process may have caused an increase in IR-A expression in “wild-type cells.” Thus, the restoration of IR function to “wild-type levels” mediated by IR-A could be also due to IR-A being the main isoform expressed in the immortalized neonatal hepatocytes prior to IR deletion.

To date, most of the research investigating the specific function of IR isoforms has been conducted in cell lines in vitro. While these experiments served for the advancement in the field, there is a clear need for more in vivo data. For this, a much desired mouse model would be one in which a tissue-specific IR-KO is used to reintroduce only IR-A or IR-B, which could be studied in parallel to extrapolate the gain or loss of function in IR isoform-mediated signaling in specific cell types.

5. IR isoform changes during metabolic disease

A number of in vitro studies have been performed to investigate the changes in IR isoform mRNA expression in response to glucose or insulin treatment [64, 66, 68]. Although it is necessary to identify possible mechanisms leading to IR splicing, the utilization of tumor-derived cells and supra-pathological amount of insulin or glucose, make it difficult to interpret these findings and reproduce them in primary cells. Moreover, the cancer cell lines used in different laboratories might have specific phenotypes regulating gene transcription machinery and the ability to modulate splicing factors in response to specific stimuli. For this reason, we have limited our discussion below to primary cells or tissues derived from humans and animal models.

In primary mammal tissues, with the tools at hand, changes in IR splicing have been studied at the transcriptional level during metabolic disease such as obesity and type 2 diabetes mellitus (T2DM), extrapolating changes to the protein level and thus signaling. During diabetes and obesity, several interesting studies addressed the possibility of splicing alteration of IR in human tissues such as muscle, liver and adipose tissue, which are accessible for sampling. Moreover, these tissues are primarily involved in insulin resistance and T2DM development, being the site of glucose handling and insulin clearance (liver), energy storage (adipose tissue) and the main site of postprandial glucose uptake (muscle).

5.1 Human evidence

In muscle, Norgren et al. found a poor correlation between IR-A mRNA and insulin-stimulated glucose utilization as well as an increase of IR-B in muscle tissue obtained from vastus lateralis muscle of non-insulin-dependent diabetic subjects [67]. This study was in agreement with a contemporary study conducted by Mosthaf et al., in which authors found higher IR-B expression in vastus lateralis muscle of subjects with insulin resistance, as compared to subjects with normal insulin sensitivity [89]. Accordingly, in a previous publication, the same authors detected predominant expression of IR-A in gastrocnemius of healthy subjects and both isoforms in T2DM [48]. Using a polyclonal antibody, which can differently displace radiolabeled insulin from the two IR isoforms but could not be of use for Western blot analysis (and is not commercially available), the same group corroborated their findings at the protein level and in absence of muscle fiber changes. In this work, the authors calculated an IR-A/IR-B ratio of 7:3 for control subjects and the opposite for diabetic subjects [90]. However, in another study, Benecke et al. did not find any difference in rectus abdominis muscle between healthy subjects compared to non-insulin-dependent T2DM patients [91]. The consensus in these studies seems to be an increase of IR-B in muscle during T2DM and T2DM development. The differences between the results can be due to the type of muscle biopsied and to the stage of T2DM. In fact, it is still unclear if different muscle fibers express different amounts of IR isoforms, which might mirror their different metabolism and glucose utilization. Thus, further work is needed to dissect these changes in muscle and the underlying mechanism using alternative methods to visualize the IR ratio in situ.

Contrary to muscle tissue, one study in liver suggested that hyperinsulinemia can regulate the tissue ratio of IR mRNA favoring the IR-A isoform [63]. In this work, the authors measured IR isoform mRNA levels in liver samples from individuals with or without T2DM before gastric bypass surgery and after 1.5 years follow-up. They found that IR-A expression was higher in T2DM patients prior to surgery and that the abnormal liver IR-A/IR-B ratio normalized post-surgery in patients with remission of diabetes, following a decrease in IR-A expression [63]. Despite a limited sample size, a similar trend was shown in a previous study by Moller et al. [49].

In subcutaneous adipose tissue (SAT), Kaminska et al. found an increased IR-A expression in insulin-resistant, obese and T2DM subjects (compared to controls), which was reversed by weight loss [65]. Moreover, they found a correlation between high fasting insulin and IR-A, linking these alterations to possible changes in splicing factors, which would in turn regulate IR isoform expression. However, using a polyclonal antibody which can differently displace radiolabeled insulin from the two IR isoforms (used for studies in muscle as mentioned above [90]), a previous study in human isolated adipocytes showed an increase in IR-B expression in adipocytes of non-insulin-dependent T2DM patients [92]. Notably, this small cohort of patients had similar body mass index (BMI) values and insulin levels and were on different antidiabetic medications or under dietary intervention [92].

5.2 Evidence in animal models of obesity and T2DM

In mouse as well as in other mammals, several studies have addressed the amount of IR isoform transcripts in different organs in healthy animals (see Section 3). Surprisingly, studies of the changes in IR isoform mRNA during metabolic disease, aimed to find common patterns and possible mechanisms, are scarce. The IR mRNA splicing variants were analyzed by RNA template-specific PCR (RS-PCR) in several tissues of a small group of diabetic rhesus monkeys [71]. Here, the authors showed that hyperinsulinemic monkeys had significantly higher expression of IR-A in vastus lateralis and rectus abdominis muscles compared to normoinsulinemic monkeys [71]. However, the authors did not dig into the possible mechanisms. We recently conducted a study to understand the possible splicing changes in a broad range of tissues in genetic and diet-induced mouse models of obesity and T2DM [70]. Comparing the findings to the abovementioned human studies, we found similar changes in adipose tissue, an increase in IR-A versus IR-B in all obese and diabetic mouse models, while in soleus and gastrocnemius muscles we found an increase in IR-B in the genetic models of obesity and T2DM only. Using a new method to visualize the IR mRNA isoforms in situ and combining it with tissue fractionation, we linked the increase of IR-A in perigonadal adipose tissue to changes in tissue architecture, rather than to a change of splice isoform in a particular cell type (i.e., adipocytes). In fact, as previously addressed, the IR isoform pattern seems to be cell-type dependent and an infiltration of immune cells (expressing mainly IR-A) in adipose tissue was the reason for these changes. The modification of tissue architecture, rather than a change of splicing at the single cell level, can also be the reason for the changes in IR isoform ratio seen in other tissues during disease. This hypothesis is further supported by a study by Vidal et al., where inducing diabetes in rats with streptozotocin showed no changes in IR mRNA ratio in liver, heart or muscle [73]. However, we have to take in consideration that rat muscle had only 1% of IR-B, the animal sample size was limited and adipose tissue was not investigated. Also, in line with this hypothesis, 48 hours of fasting in rats—which causes a significant decrease of plasma insulin—did not alter the IR isoform mRNA ratios in liver, heart, muscle and adipose tissue [73].

5.3 Myotonic dystrophy as a model of IR isoform shift

In relation to metabolic diseases, it is worth mentioning that complications of myotonic dystrophy—an autosomal genetic disease characterized by muscle loss and weakness caused by the expansion of nucleotides repeat in 3′ untranslated region of different mRNAs—has been found to alter the IR pre-mRNA splicing [93, 94]. One of the mechanisms involves the function of the CUG-BP splicing factor (acting on CUG repeats), which together with MBNL1 and other splicing factors has been shown to be pivotal for IR gene regulation [95]. In both myotonic dystrophy types 1 and 2 (DM1 and DM2), insulin resistance and decreased muscle insulin sensitivity are common. This correlates with an isoform switch from IR-B to IR-A in muscle, without changes in the total IR protein levels, and can be considered the closest model of IR splicing changes in a specific cell type in vivo. In this context, the diabetes medication metformin has been shown to affect the alternative splicing in DM1 patient-derived myoblast as well as in peripheral blood lymphocytes in T2DM patients, leading to the increase of IR-B expression. The increase of IR-B expression was not shared by another glucose-lowering agent, the dipeptidyl peptidase-4 (DPP-4) inhibitor sitagliptin [96]. A follow-up of this effect of metformin during T2DM treatment in other primary cells and tissue, which express both IR isoform receptors, would be of importance to understand if this could be a further mechanism explaining the effect of this drug in lowering blood glucose levels.

5.4 Summing up the evidence and the possible mechanisms

In summary, in metabolic diseases such as T2DM or obesity, limited work has been conducted in humans and animal models to uncover IR isoform changes and underlying mechanisms during disease. It seems that long-term metabolic alterations such as the ones occurring during T2DM and obesity alter the IR isoform mRNA ratio in some of the studied tissues. It seems that IR-A/IR-B ratio decreases in muscle and increases in liver and adipose tissue during hyperinsulinemic and T2DM states in humans [48, 63, 67, 89, 90, 92]. It is current consensus in the field that a higher expression of IR-A (considered to drive more mitogenic signals rather than metabolic ones) would enable this isoform to compete with IR-B for insulin, thus reducing the action of IR-B in maintaining glucose homeostasis, leading to insulin resistance. This seems to be the case during diabetes mellitus, although there could be other possible metabolic alterations due to changes in alternative splicing of other genes. Alternatively, a change in IR isoforms ratio could be interpreted as a result of changes in tissue architecture and the increase/decrease of certain cell types expressing one or the other isoforms. In fact, besides one exception, the mentioned investigations were conducted measuring IR isoform mRNA in the whole muscle, liver and adipose tissue. Architecturally, all tissues are composed of different cell types, as well as different subpopulations (heterogeneity) within a specific cell type. For example, in liver and fat, zonation has been described, in which differences in transcriptomics shows that even cells of the same kind display different phenotypes and potentially even specific IR isoform ratios [97, 98]. Thus, without experiments focusing on dissecting the mechanism behind a change in IR isoforms at single-cell resolution, it will be difficult to draw conclusion on the mechanism behind this phenomenon during metabolic disease. Future efforts are therefore required to tackle this issue more in depth to provide a common denominator for the IR changes at cell resolution, possibly taking advantage of a cell type that expresses both IR isoforms, in order to detect if changes in splicing occur.

6. Challenges and perspectives

Since the discovery of the alternative splicing of the IR in 1989 by Seino et al. [50], many studies have focused on understanding the tissue expression patterns, binding affinity, crystal structures, differential signaling and alternative routes of internalization and recycling of the IR isoforms. However, what we currently know about the IR isoforms is only a fraction of what we have not discovered yet. The reason for the two isoforms conferring an evolutionary advantage in mammals, and why other vertebrates, such as birds and fish, exist with just IR-A as well as the reasons behind the complex interactions and redundancy of insulin and IGFs systems are extremely interesting and important questions. Uncovering these aspects together with understanding why IR-A is expressed during development, in stem and cancer cells, and why progenitor cells express mainly IR-A switching their expression to IR-B upon differentiation and specialization, would help decipher the complex regulation of IR-mediated signaling upon their ligand binding. Here below, we present key points, which should be addressed in the near future, along with the tools needed in order to achieve these goals.

6.1 Shift in focus to “single cell” research

Amongst the pioneers of the IR isoform research area, Seino and Mosthaf in the ‘90s pinpointed the importance of determining IR isoforms splicing and their signaling at single cell level [31, 50, 99]. Until now, almost all works investigating the change in IR splice isoforms were performed using whole tissues. Thus, a key question is where/if the IR isoforms are present at the single-cell level in vivo in heath and disease and the possible mechanisms behind changes in their ratio. For example, in the liver, little is known about the specific expression of IR isoforms in non-parenchymal liver cells, which constitute around 20% of total cells [100]. Similar considerations apply to the brain, on which contradictory studies have been published. Garwood et al. found predominantly IR-B mRNA in cultured astrocytes isolated from human and in primary astrocytes commercially available [75]. However, this finding is difficult to reconcile with the fact that the brain tissue as a whole expresses almost only IR-A. Considering that astrocytes outnumber neurons and that the latter probably express exclusively IR-A, the expected outcome in the whole tissue following this work would be a similar expression of both isoforms. Another possible explanation would be that an IR isoform switch occurs after isolation and culture of astrocytes, or that a much lower IR mRNA is expressed in these cells compared to neurons and others cells expressing IR-A, such as endothelial cells. In contradiction, Heni et al. using another available source of human astrocytes in vitro found that the majority of IR mRNA in astrocytes was IR-A [101]. One single work in human brain was performed using RT-PCR/FISH to detect IR isoforms in situ, which in our view, lacked validation for the two IR isoforms. Nevertheless, the authors found the majority of IR mRNA as IR-B in microglia cells, and mainly IR-A in neurons, with the occasional finding of IR-B [102].

6.2 IR isoform-mediated signaling

It is common, especially in the research field of metabolic disease to talk about “classical or canonical” and “non-classical or non-canonical” insulin target tissues. The first have most commonly been liver, muscle and fat, while recently also pancreatic islets and brain have been considered. However, all cells in the body express IRs and therefore this distinction might be obsolete. The type and amount of the ligands that can bind the IR and the downstream signaling proteins involved in IR-mediated signaling (i.e., IRSs, PI3Ks, MAPKs, AKTs, etc.) are expressed in different amounts in different cell types within a tissue and this heterogeneity increases the combinatorial possibility of signaling downstream of the receptor (reviewed in [103]).

The current consensus is that an increase of IR-A, with its higher affinity for insulin and IGF2, might induce a strong proliferative signal and decrease the metabolic effect of insulin (reviewed in [56, 104]). There is strong evidence that IR-A is increased in cancer, where it exerts a proliferative and survival advantage. However, the fact that IR-A transduces a mitogenic signal in non-cancer and non-stem cells, where both isoforms are present, seems not always to be the case. In pancreatic β-cell, where both IR isoforms are present, the IR-A induces a downstream activation of insulin gene transcription and does not confer proliferative effects [43]. Of note, the pancreatic β-cell is the only primary non-cancer cell, together with kidney podocytes, from which there is evidence of IR isoform-specific signal transduction, and for which downstream pathways have been partially uncovered [43, 46, 47, 105, 106]. Moreover, cells with high prevalence of IR-A such as neurons are clearly not proliferative, showing that alteration of ligands and downstream signaling might be the key factors for the proliferative effect through IR-A.

In cells that express both IR homo-receptors simultaneously, one hypothesis would be that the two isoforms regulate different signaling pathways in the same cell as shown for pancreatic β-cells, due to spatial segregation at the membrane and different downstream binding partners. This could explain the selective insulin resistance (as shown for podocytes [47]) seen in liver or adipose tissue, where only some downstream signals of the insulin signaling pathway are blunted during disease [107, 108]. However, the selective insulin resistance in liver or fat linked to the IR isoforms still depends on the possibility that hepatocytes and adipocytes express IR-A. Moreover, if we hypothesize that of the total IR in hepatocytes or adipocytes only 10% is IR-A, then IR-A homo-receptors should be absent and IR-A would be found forming hetero-receptors with IR-B, unless some mechanism of segregation is present to preserve homo-dimer assembly.

An understudied variable that could potentially alter insulin signaling during metabolic disease is the binding of proinsulin to IR-A and its possible downstream effect. In fact, it is still not clear if this binding results in a signal transduction, or if the IR acts as a “sponge” for proinsulin resulting in a decrease in insulin binding. This aspect could play an important role during T2DM development, where the proinsulin/insulin ratio is increased in plasma [37].

In summary, much work needs to be done to understand basic questions regarding the IR isoform expression and signaling in specific cell types and during disease. However, essential tools to discriminate between the two IR isoforms are lacking and critical to address these important issues and provide breakthroughs in the field.

6.3 The need for new tools

Up until now, measuring IR isoform mRNA has been useful to investigate the isoforms at tissue level, and more recently at cellular resolution in situ. However, IR mRNA expression might not be linked to actual protein levels in cells and tissue [70]. One intriguing future possibility to identify the existence of IR isoforms in single cells would be to utilize single-cell RNA sequencing (RNA-Seq). However, given the low levels of IR mRNA expression and that only a few base-pairs distinguish the two isoforms, this method is still technically challenging but probably possible in the near future [109].

Pivotal to understanding the dynamics and the regulation of the equilibrium of the IR isoforms (and therefore signaling) in health and disease is to develop tools with the possibility to visualize the amount and localization of the mature isoform receptors. Up to now, attempts to develop an isoform-specific antibody for Western blotting or immunostaining have failed, probably due to the small difference between the two denatured proteins and to the fact that the 12 differential amino acids reside in a poorly accessible area of the receptor. Means to visualize the IR isoform in live cells would allow studying the binding of ligands to the IR isoforms at the cell membrane. Additionally, other small molecules such as aptamers, nucleotides oligomers that could differentially bind to the IR isoforms could conjugate with fluorescent dyes and allow the visualization of endogenous IR receptor monomers or dimers in living cells. Such visualization would allow to track the receptors in situ and understand more about their intracellular dynamics during health and disease. That aptamers can block IR-mediated signaling in an isoform-specific manner in vitro has been demonstrated [106] and shows the potential use of these molecules for IR isoform detection.

Alternatively, the results of overexpression studies can be useful in understanding the possible segregation of the IRs due to different lipid membrane requirements, and possibly signaling [110]. However, a spill over of signaling and changes in IR distribution could take place due to overexpression and the results would need to be confirmed based on the endogenous receptor.

6.4 Therapeutic potential

Let us imagine having the available methods to visualize and measure the IR at the protein level as well as the cellular distribution of IR-A/IR-B monomers, dimers and hybrids receptors. With these tools it would be possible to study which signals are transduced by the IR isoforms in a determined cell type and in response to insulin, proinsulin, IGF1 and IGF2. In this scenario, new discoveries could be used to develop targeted treatments for metabolic diseases. For example, we could screen for selective activators and inhibitors of the different IR isoforms by measuring downstream activation of selective pathways in specific healthy or diseased cells.

We have discussed the necessity of being able to understand the specific signaling cascades initiated by the different receptors upon ligand binding. The two IR isoforms could be considered as targets to selectively increase or decrease signaling pathways driven by one of the two receptors in specific cell types. Focusing on diabetes, and provided that signaling pathways downstream of IR-B are the ones modulating glucose metabolism, the generation of selective IR-B activators could be beneficial in comparison to the current insulin sensitizers or insulin analogues, which do not target either isoform specifically. On the contrary, selectively targeting IR-A, using specific antibodies or other therapeutic molecules, could be directed to treat tumor proliferation by blunting its mitogenic signaling pathways and thus hamper cell growth and survival. Newly developed insulin analogues that preferentially bind one or the other IR isoform [111] could be pivotal for improving insulin therapy, but to define this feature we would need to have tools to measure selective downstream signaling in specific cells and tissues to understand the potency and effect of such molecules. Not only peptides could be used to modulate the IR isoform activity in a selective manner, but other small molecules such as aptamers could also prove useful. One promising publication showed that these molecules can be IR isoform-specific and can facilitate or block the signaling selectively, working as allosteric regulators [112].

Another intriguing therapeutic possibility to modulate IR splice isoforms would be to use a selective splice switcher, which has recently been tested (reviewed in [113]). This would also be a better strategy to study the IR isoforms in vitro, since the overexpression of specific isoforms might produce a spill over of signaling cascades, as explained above.

7. Conclusions

In this chapter, we aimed to summarize the state of the art research involving the IR isoforms, especially in the area of metabolic disease. We also wanted to draw attention to how important it is to understand the full implications of having two IR isoforms. The majority of the research focused on insulin signaling refers to the IR as one receptor, without considering that there are two structurally and functionally distinct isoforms in play. In our view, future research in this field would benefit from a focus on cell type-specific IR isoform signal transduction pathways, what differences there may be between cells of the same type but different localization in a tissue, and finally, what changes occur during disease. The development of the tools needed to address these questions would pave the way for important breakthroughs in comprehending the ubiquitous, but diverse, IR signaling. Finally, these tools will be essential for the development and testing of new therapeutic strategies to counteract diseases affected by IR signaling dysfunction.

Acknowledgments

We would like to thank Ingo and Barbara Leibiger for the critical reading of this chapter and their contribution as part of the Per-Olof Berggren lab in the field of IR isoforms in pancreatic islets.

Conflicts of interest

The authors declare no conflict of interest.