2025 Volume 72 Issue 12 Pages 1269-1286
Comparative endocrinology is a research subfield in endocrinology that delves into deeper understanding of the endocrine system from an evolutionary or phylogenetic perspective. To date, this approach has contributed significantly to the development of endocrinology by elucidating the evolutionary history of hormone molecules and their functions from invertebrates to vertebrates. In this review, the author initially introduces how the comparative approach has expanded and enlightened the view in endocrinology using the concept of hormones as an example. The expansion of the hormone concept blurs boundaries between signaling molecules of the three homeostatic systems, namely, the endocrine, nervous, and immune systems. Subsequently, the evolutionary history of the endocrine system is introduced in terms of both molecules and functions using the insulin superfamily as a model. This hormone family is one of the most ancient hormonal systems in animal (metazoan) phylogeny and the homologous hormones are identified in the most ancient metazoans such as sponges and hydra. In addition, this hormonal system was chosen as a topic of this review, because insulin is one of the most focused research topics in modern medicine in relation to insulin resistance and metabolic syndrome. Finally, the ancestral molecule of the insulin superfamily and its original or essential function will be discussed with some speculations to illustrate the value and joy of comparative studies that can create an original concept of the endocrine system from the evolutionary viewpoint. The comparative approach certainly helps deeper understanding of the insulin superfamily of humans.

The hormone in the classic concept of endocrinology is described as a biologically active substance for information transfer within the body, which is secreted from specialized cells in the endocrine tissue and acts on receptive cells in distant target tissues via blood circulation [1]. Indeed, the pituitary, thyroid, adrenal and pancreas represent typical and historically well-known endocrine tissues, as disturbance of their hormonal function produces serious clinical syndromes. Over time, however, the definition of hormones has expanded in terms of molecular species and mode of action. In addition, many tissues that were previously known to perform specialized functions, such as bones, cardiac muscles, skeletal muscles, adipose tissues and immune tissues, have been newly found to produce and secrete bioactive peptides that are now classified as hormones [2].
Hormone molecules are typically categorized into peptide (protein) hormones, amine hormones and steroid hormones. In addition, several chemical substances that were not previously designated as hormones, such as eicosanoids (prostaglandins) and gasotransmitters (e.g., nitric oxide), are newly accepted as hormones and are described in endocrinology textbooks [1, 3]. In support of this idea, prostaglandins are not only intracellular messengers that act locally within the cells but also secreted into the circulation and act via membrane receptors of distant cells [4]. Nitric oxide (NO) is secreted from vascular endothelial cells and acts in a paracrine fashion on the adjacent smooth muscle cells to promote relaxation [5]. As discussed below, many newly recognized hormones secreted from non-endocrine tissues act directly on the neighboring cells, and the paracrine route seems to be a major mode of their actions. The receptor for NO is a soluble guanylyl cyclase in the cytoplasm, which produces cGMP as a second messenger after NO binding. The cytoplasmic receptor for NO is reminiscent of other receptors for lipophilic hormones.
Recent progress in bioinformatics has identified many new hormone genes in the genome database of various animals [6, 7], which more than doubled the number of known hormones within a few decades and allowed us to expand the concept of hormones [1, 3]. Many of the newly identified hormones are paralogs of known hormones, and it is now generally accepted that almost all of these hormones are members of families that consist of peptides originating from a single ancestral gene. Likewise, the families of peptides also act on families of respective receptors with different affinities. A typical example is the insulin-like peptide family, which is the topic of this review. Importantly, most of the newly identified hormones are not secreted into the circulation but into the intercellular space to act on neighboring cells or on the secreting cells themselves. Thus, the paracrine route seems to be the most common mode of action for hormonal molecules.
If the endocrine route is not a major mode of hormone action, it seems difficult to distinguish the three homeostatic systems in terms of information mediators, namely, hormones for the endocrine system, neurotransmitters and neuromodulators for the nervous system, and cytokines for the immune system, all of which act in a paracrine fashion (Fig. 1). Neurotransmitters are used for rapid information transfer from neuron to neuron using long axons, synapses and ionotropic receptors in the target cell. Meanwhile, neuromodulation uses so-called neuromodulators that act on metabotropic receptors in the target neurons. The neuromodulators include paracrine hormones such as appetite-regulating neuropeptides including insulin. Cytokines are secreted from immune cells and tissues and act on the neighboring cells, but some cytokines are secreted into the circulation and act on distant cells. In fact, growth hormone/prolactin, erythropoietin and leptin are traditionally classified as hormones but act on the cytokine receptors as cytokines [8]. Many cytokines secreted from non-endocrine tissues such as the liver (hepatokine), adipose tissues (lipokine) and skeletal muscle (myokine) are indistinguishable from hormones [2, 9, 10]. Based on the apparent expansion of the hormone concept, it seems appropriate to suggest that hormone is the first messenger secreted from various tissues and is used for cell-to-cell communication via metabotropic receptors in the target cells. In addition to neuromodulators and cytokines, several growth factors that are basically involved in growth, differentiation and proliferation in early developmental stages are also classified into the category of hormones such as insulin-like growth factors (Fig. 1).

Comparative endocrinology is a discipline that aims to understand the endocrine system from a phylogenetic or evolutionary approach [11-13]. For deeper understanding of this section for non-biologists, let’s first show the phylogenetic tree of Metazoa, which is also called Animalia that includes all extant animals (Fig. 2). A few hormone-like genes have been identified in the genome of all metazoans ranging from sponges (Porifera) to vertebrates (Chordata). These include transforming growth factor (TGF)-β, insulin-like peptides, and glycoprotein hormones such as gonadotropins of vertebrates and bursicons of insects [14]. However, hormone molecules and their actions are subject to profound changes during evolution. The comparative approach attempts to analyze the diversity and to extract the original and essential role(s) of a hormone family by tracing the history of molecular and functional evolution [7, 14, 15]. During phylogenetic evolution, the vascular system first appeared in the advanced groups of bilaterians such as Mollusca (octopus and squids), Arthropoda (insects and crabs) and Chordata including vertebrates (Fig. 2), while genes for hormones and their receptors are present in more primitive ‘pre-bilaterian’ groups such as Porifera (sponges), Placozoa (tiny, blob-like animals), Ctenophora (comb jellies) and Cnidaria (hydras and jellyfish) [16]. Thus, the endocrine (hormonal) system already existed before the emergence of the circulatory system. This fact also supports the concept that the paracrine route is an original mode of hormone action from the phylogenetic viewpoint.

The nervous system first appeared in Ctenophora and Cnidaria as a diffused nervous system or a nerve net [17], where nerve cells communicate with each other by neuropeptides such as RF amides, rather than by acetylcholine/catecholamine used in vertebrates [18]. In addition, the nervous system has not yet developed in Porifera and Placozoa, but hormone genes are already present in these animal groups. Secretory neuroid cells reminiscent of endocrine cells are in close contact with digestive cells that express receptor proteins in sponges [19], and neuropeptide-producing endocrine cells are found in placozoans [16]. These results suggest that paracrine hormones already existed before the development of the nervous system. In this way, comparative endocrinology can provide a unique perspective from the phylogenetic viewpoint on the relationship between the endocrine system and the nervous system (Fig. 1).
Endocrine system vs. immune systemCytokines are the mediator of the acquired immune system, which first emerged in vertebrates [20]. The innate immune system is present in all metazoans [21], but the presence of cytokines has not yet been demonstrated in invertebrates except for the TGF-β gene [22]. TGF-β is not cytokine but is recently regarded as an immune regulator of tumorigenesis [23]. Phylogenetically, therefore, hormone genes were present before the emergence of the acquired immune system, where cytokines are used as cell-to-cell communicators.
Cytokines are characterized by their binding to their specific receptors, called cytokine receptors, which initiate intracellular signaling cascade through the JAK/STAT pathway [24]. Most growth factors bind not to the cytokine receptors but to another type of receptors, called receptor tyrosine kinases (RTKs), which have a cytoplasmic kinase domain. Despite such differences, several groups of growth factors are now categorized into cytokines [24]. As growth hormone and leptin bind to the cytokine receptors as mentioned above, the boundary between hormones, cytokines and growth factors becomes more ambiguous when viewed from the phylogenetic approach (Fig. 1).
Finally, the evolutionary history of hormones is variable among hormone families; some existed in the most ancient animal groups (pre-bilaterians) before divergence of deuterostomes and protostomes in the phylogenetic tree (Fig. 2), while others have emerged more recently in vertebrates. The oldest hormone families thus far known are the insulin superfamily [25] and the TGF-β superfamily, because orthologous genes are detectable in the most ancestral animal group, Porifera [26]. The classification rank of “superfamily” is given to these hormone families because they consist of (sub)families of different types of hormones. Of interest is that both the insulin superfamily and the TGF-β superfamily have their origins in growth factors, which bind to RTK or receptor serine/threonine kinase [27]. Although the TGF-β superfamily is as important and intriguing as the insulin superfamily from the phylogenetic viewpoint, this review will focus on the insulin superfamily and discuss its molecular and functional evolution.
After the discovery of insulin more than 100 years ago [28] and subsequent amino-acid sequence determination of human insulin [29], intensive and extensive studies have been performed on this intriguing peptide [30]. Insulin is now recognized as a sole and powerful hypoglycemic hormone in mammals, especially in humans, and its disorder gives rise to hyperglycemia (diabetes mellitus) and subsequent various forms of metabolic syndrome through insulin resistance [31]. Thus, insulin has now become a major research focus in both clinical and basic endocrinology.
Insulin superfamily in vertebratesThe molecular structures of the insulin superfamily and their receptors in vertebrates are described briefly here to provide a basis for comparison with those of invertebrates. Please refer to the comprehensive reviews for detailed structures of hormones and their receptors [32-35]. The intracellular signaling pathway after insulin binding to its receptor, called insulin signaling, is one of the hottest research topics in endocrinology [36, 37], but this topic is beyond the scope of this review.
Hormone moleculesAmino acid sequences of insulin superfamily peptides in various vertebrate species are shown in the review by Conlon [38]. Insulin is a member of the insulin superfamily, the members of which are roughly classified into two groups in vertebrates based on the receptor to which they bind [39]; the insulin/insulin-like growth factor (IGF) family consisting of insulin, IGF-1 and IGF-2, and the relaxin/insulin-like peptide (INSL) family consisting of relaxin-1, 2 and 3 and INSL-3, 4, 5 and 6. Insulin-like peptides identified in invertebrates are abbreviated as ILPs (see below). The two families are grouped together as a superfamily, as all members appear to have originated from a single ancestral gene.
The basic structure of insulin superfamily peptides is a heterodimer composed of an A-chain and a B-chain, which are connected by two interchain disulfide bridges [32]. A C-peptide between the two chains in proinsulin is proteolytically cleaved off during peptide maturation. The mature insulin hormone contains a third, intrachain disulfide bond in the A-chain. Meanwhile, two IGFs are single-chain peptides with A-chain and B-chain connected by the C-peptide, and an additional peptide, called D-peptide, is attached at the C-terminus of the mature peptide. All members of the insulin superfamily have a characteristic sequence in common in the A-chain, called insulin signature (Cys-Cys-3X-Cys-8X-Cys), where X is a variable amino acid. In the relaxin/INSL family, relaxin-1–3 have Arg-3X-Arg-2X-Ile/Val sequence in the B-chain named relaxin-binding cassette, but INSL peptides lack this cassette. Three relaxin genes are present in humans, but most other mammals do not have the relaxin-2 gene. The relaxin-1 gene was duplicated to generate the relaxin-2 gene only in a primate linage including humans 30–40 million years ago [40].
ReceptorsThe distinction between the insulin/IGF family and the relaxin/INSL family is in their receptors (Fig. 3). Insulin/IGF uses RTK for their specific receptor, while relaxin/INSL binds to the G protein-coupled receptor (GPCR). Three types of insulin/IGF receptors are found and named INSR-A, INSR-B and IGF-1R. INSR-A and INSR-B are produced by alternative splicing of the same gene. Among the three, two are coupled in various combinations, and each dimeric receptor serves as a receptor for insulin, IGF-1, or IGF-2 with different affinities [41]. For instance, homodimers of INSR-B and of IGF-1R are specific receptors for insulin and IGFs, respectively, while not only insulin but also IGF-2 binds homodimer of INSR-A and heterodimer of INSR-A and INSR-B with high affinities. Heterodimers of INSR-A/B and IGF-1R serve as receptors for insulin and IGFs with varying affinities. In addition, cation-independent mannose-6-phospate glycoprotein receptor is designated as an IGF-2 receptor. Relaxin receptors belong to the rhodopsin-like GPCR family with a leucine-rich repeat at the extracellular N-terminal region (LGR), and they are abbreviated as RXFPs (Fig. 3). Four types (RXFP1–4) have been identified thus far, and relaxin-1/2, INSL-3, relaxin-3 and INSL-5 bind to RXFP1, RXFP2, RXFP3 and RXFP4, respectively, with high affinities [33]. Specific receptors for INSL-4 and INSL-6 have not been identified yet.

All vertebrate species possess insulin superfamily peptides, but the number of each family member differs among different groups [38]. The most primitive extant vertebrates are agnathans (jawless fishes), and extant species are called cyclostomes that include hagfish and lamprey (Fig. 4). Cyclostomes have single insulin and IGF in their genome [42], and the IGF is equally similar to IGF-1 and IGF-2 of mammals but has higher affinity to mammalian IGF-2 receptor [43]. Insulin, IGF-1 and IGF-2 are all present throughout jawed vertebrates. Insulin receptors (RTKs) are also found in all vertebrate species [44].

The relaxin/INSL family peptides and their receptors are also present in cyclostomes, and they have been diversified during vertebrate evolution [45]. Relaxin-3 appears to be an ancestral molecule of the family as suggested by the molecular phylogenetic analysis. Thus, two relaxin-like peptides identified in cyclostomes (probably relaxin-3 and INSL-3) may be generated from relaxin-3 by the first-round whole genome duplication (WGD) that occurred during the transition from tunicates to agnathans (Fig. 4). The second-round WGD produced four peptides (relaxin-1/2, relaxin-3, INSL-3 and INSL-5) in jawed vertebrates [45]. In teleosts, an additional third-round WGD occurred only in their lineage to produce additional members of ligands and receptors, but some of them seem to have disappeared as pseudogenes in the extant teleosts. However, four IGF genes, named igf1, igf2a, igf2b and igf3, are retained in the zebrafish genome [46], suggesting important functions of IGF. Collectively, all classes of non-mammalian vertebrates have at least four types of relaxin/INSL peptides, although the number in each type is variable due to the specific gene duplication and loss in different linages [47].
Insulin superfamily in invertebratesAfter the discovery of insulin, the presence of insulin homologs has been sought in various invertebrate species, initially by immunological methods. For instance, insulin immunoreactivity was detected in the extract of the fruit fly, Drosophila melanogaster and the earthworm, Annelida oligochaeta using radioimmunoassay for human insulin [48]. Recent establishments of genome databases in various invertebrate species profoundly accelerated progress in identifying insulin-related peptide and receptor genes in both protostomes and deuterostomes (Fig. 2). The signaling pathway after binding to the receptor is well conserved across phylogeny [49].
ProtostomesUnlike vertebrates where INSL stands for insulin-like peptide, ILP is commonly used for the abbreviation in invertebrates. The first ILP identified in invertebrates is bombyxin, which was isolated from the brain of the silkmoth, Bombyx mori, as a neuropeptide that stimulates the release of ecdysone, a steroid hormone that induces ecdysis and metamorphosis in insects, from prothoracic glands [50]. Encouraged by the discovery of bombyxin, ILPs were sought in various species of invertebrates from pre-bilaterians to bilaterians that consist of protostomes and deuterostomes (Fig. 2). Thanks to the public release of genome databases and subsequent development of bioinformatic techniques, ILPs were identified in almost all phyla of metazoans by either cDNA cloning or data mining from the database, and they are assigned by various names (Table 1). Several ILPs have been identified in diverse protostome species, of which duplication of the ILP gene is particularly prominent in some species and the gene copy number increased to ca. 40 in the silkmoth [51] and in the nematode, Caenorhabditis elegans [52]. In these animals, many paralogous genes are localized together within a short distance on a chromosome [53, 54]. Such frequent gene duplication may be caused by ectopic recombination or retrotransposition, while retention of the duplicate genes may be due to the neo- or sub-functionalization as the function of ILPs in these invertebrates are diverse including specific functions for larval diapause and ecdysis. ILPs are also identified in several species of mollusks [55] and annelids [56] (Table 1).
| Phylum | Animal | Peptide name | Reference | |
|---|---|---|---|---|
| Pre-bilateria | ||||
| Placozoa | trichoplax (Trichoplax adhaerens) |
Insulin-like1 1–4 | [69] | |
| Cnidaria | hydra (Hydra magnipapillata) |
ILP2 1–3 | [70] | |
| Bilateria (Protostomia) | ||||
| Arthropoda | ||||
| Decapods | spiny lobster (Sagmariasus verreauxi) |
insulin-like1, relaxin-like1, gonadulin6, androgenin | [58] | |
| Insects | silkmoth (Bombyx mori) |
Bombyxins, IGF-like peptide | [50] [119] |
|
| fruit fly (Drosophila melanogaster) |
ILP 1–8 | [144] | ||
| Isopods | common pill woodlouse (Armadillidium vulgare) |
Androgenic gland hormone | [145] | |
| Nematoda | C. elegans (Caenorhabditis elegans) |
Insulin-like 1–37 | [52] | |
| Mollusca | great pond snail (Lymnaea stagnalis) |
Molluscan insulin-related peptide (MIP) | [146] | |
| Pacific oyster (Magallana gigas) |
MIP 1–4, ILP, MIP-like peptide | [55] | ||
| Annelida | leech (Helobdella robusta) |
polychaete worm (Capitella teleta) |
Insulin-like1 1–4 | [56] |
| Bilateria (Deuterostomia) | ||||
| Chordata | ||||
| Tunicata | vase tunicate (Ciona intestinalis) |
ILP, Relaxin-like, IGF-like | [63] | |
| Cephalochordata | amphioxus (Branchiostoma californiensis) |
ILP3 | [64] | |
| Hemichordata | acorn worm (Schizocardium californicum) |
IGF-like1, DILP7-like, octinsulin | [58] | |
| Echinodermata | blue bat star (Asterina pectinifera) |
Relaxin-like gonad-stimulating peptide (RGP)4 | [133] | |
| purple sea urchin (Strongylocentrus purpuratus) |
IGF-like1, GSS5, DILP7-like, octinsulin, multinsulin | [58] | ||
1These peptides are described as either insulin, IGF or relaxin in the paper.
2Inslin-like peptide is generally abbreviated as ILP in invertebrates and INSL in vertebrates.
3Six ILP are identified in another amphioxus species and classified into either insulin/IGF or relaxin/INSL family [65].
4RGP2 and other ILP are also identified after its discovery [138].
5This peptide corresponds to RGP.
6This peptide corresponds to ILP8 of Drosophila.
In several species of arthropods including insects (Drosophila) [57] and decapods (lobster) [58], identified ILPs are classified into three types (insulin-like, IGF-like and relaxin-like) by molecular phylogenetic analyses, functional analyses, or receptor preferences (Table 1). In this classification, the insulin-like and the IGF-like are distinguished by the presence of C-peptide in the mature peptide, and the relaxin-like is distinguished from others by binding to GPCR, as demonstrated in Drosophila [59]. Therefore, divergence of the insulin superfamily peptides into insulin/IGF and relaxin/INSL subfamilies occurred before divergence of protostomes and deuterostomes (Fig. 2).
Since Drosophila is a model species with extensive accumulation of research data, its ILPs (DILPs) have been investigated intensively [60]. Among the eight DILPs identified in this fly, DILP1-6 have high affinities to RTK-type insulin receptor, of which DILP6 is categorized as IGF-like because its A-chain and B-chain are connected by a short C-peptide. Meanwhile, DILP7 and DILP8 are classified into relaxin-like because they have high affinities to LGR4 and LGR3, respectively [61, 62] (Fig. 3). The DILP8-like genes are also found in the genomes of various arthropods, where they are primarily expressed in the gonads and thus named ‘gonadulin’ in a few species (Table 1). The presence of DILP8/gonadulin has not yet been determined in the silkmoth in which the greatest number of ILPs have been identified [53].
In another model species, C. elegans, forty ILP genes identified in the genome are named INS-1–39 and DAF-28 [52]. DAF stands for ‘abnormal dauer formation protein’ and it is so named because natural mutation of the gene caused abnormal dauer in the nematode. The 40 ILPs are classified into three types, α, β and γ, based on the location and number of disulfide bonds, and γ-type peptides have two interchain bridges and one intrachain bond in the A-chain as in other insulin superfamily peptides of vertebrates and invertebrates [54]. INS-1 (β-type) and INS-18 (γ-type) possess a C-peptide between A-chain and B-chain, which is thought to be cleaved off by prohormone convertases during maturation. Thus, INS-18 is structurally most akin to the vertebrate insulin or relaxin. Several other INSs have no C-peptide in the prohormone sequence and thus they are supposed to be single-chain peptides like IGFs after maturation. In contrast to diversified ligands, only one RTK receptor, named DAF-2, has been identified in C. elegans [26]. As several LGR genes are present in the genome, it is possible that relaxin-type INS also exists in C. elegans (Fig. 3). Some nematode INSs were shown to exhibit antagonistic effect on the intracellular signaling after binding to DAF-2, which contrasts with the exclusively agonistic effects of insulin superfamily peptides in vertebrates.
DeuterostomesIn deuterostomes, multiple ILPs have been identified in the major three phyla, chordates, hemichordates and echinoderms (Table 1), which can also be classified into three groups, insulin-like, IGF-like and relaxin-like, as in some species of protostomes [58, 63]. In chordates, multiple ILPs were found in tunicates, but only one ILP was identifiable in the amphioxus, Branchiostoma californiensis, in which this ILP is equally similar to insulin and IGF [64]. In a different amphioxus species, B. floridae, however, six ILP genes were identified and classified into insulin/IGF-like and relaxin-like by molecular phylogenetic analyses [65, 66]. However, the binding of the relaxin-like peptide to LGR has not been confirmed.
Pre-bilateriansIn addition to bilaterians, cloning of possible insulin cDNA was reported in a species of sponge, Geodia cydonium [67]. It was named insulin because its cDNA sequence is highly similar (~80%) to human insulin. The presence of its gene should be confirmed in the published genome database of this species. On the other hand, cDNAs of insulin receptor-like RTKs were cloned in three species of sponges by the same research group [68]. In other pre-bilaterians, four ILPs named insulin 1–4 were identified in the genome database of Placozoa, Trichoplax adhaerens [69] and three ILPs in Cnidaria, Hydra magnipapillata [70]. However, placozoan insulin has an untypical insulin signature, Cys-Cys-4X-Cys-8X-Cys, compared with that of bilaterian insulins (3X instead of 4X). It seems that the presence of ILP has not been reported in another pre-bilaterian, Ctenophora (Fig. 2).
It is of interest to note that immunoreactive and bioactive insulin were identified in the unicellular eukaryote, Tetrahymena pyriformis [71]. In addition, another research group suggested that the ciliate possesses insulin receptor, and application of mammalian insulin to the external medium increased cAMP in the ciliate cytoplasm [72]. They hypothesized that the ciliate communicates with each other using this insulin-like pheromone secreted into the external media. Ciliates are previously considered as animals and classified as Protozoa (meaning first or primitive animals). However, they are now excluded from Metazoa and grouped as Protista. If insulin is in fact present in Protista, insulin superfamily of peptides could have existed prior to the emergence of Metazoa. To my knowledge, however, the presence of the ILP gene has not been reported thus far in the Tetrahymena genome.
Invertebrate ILP receptorsIt seems that ILP receptors, either RTKs for insulin/IGF or LGRs for relaxin, already existed before their ligand hormones emerged and exerted any biological functions (Fig. 3). The origin of RTKs predates the dawn of Metazoa, as the presence of RTK-like genes was reported in pre-metazoan filastereans, Capsaspora owczarzaki and Ministeria vibrans [73]. Consistently, the RTK genes are found in pre-bilaterian genomes such as placozoan [69], and some of them are categorized into the insulin/IGF-like receptors by comparative genomic analyses [74].
LGRs are known to be the receptor of glycoprotein hormones (LH, FSH, TSH, bursicon, etc.) and relaxin family peptides in bilaterians [16]. LGRs belong to the class A GPCR, which is also called the rhodopsin-like receptor family. As its name indicates, the original function of the class A GPCRs may be photoreception in the unicellular eukaryotes. It is well established that GPCRs and membrane guanylyl cyclases (receptors for natriuretic peptides and guanylins), major hormone receptors that utilize cyclic nucleotides (cAMP and cGMP) as second messengers, were originally receptors for physical (e.g., photons) and/or chemical (e.g., odorants) signals, and they have evolved as hormone receptors when newly emerged hormones bound to them and initiate intracellular signaling cascade [75, 76].
Evolution of GPCRs is phylogenetically old, and genes that encode proteins associated with GPCR signaling systems can be found in all eukaryotic species including metazoans [77, 78]. The human genome contains ~800 annotated GPCR genes, of which about half have sensory functions for olfaction, taste, light perception, etc. It is interesting to note that GPCRs have undergone profound evolution in their function during environmental adaptation on the balance between constraints and plasticity [79]. Adaptive changes in GPCR function are observed not only in sensory systems (vision, taste, and smell) but also in hormonal systems regulating reproduction and energy homeostasis. Among GPCRs, relaxin receptor-like LGRs are detected in the genome of pre-bilaterians, Cnidaria and Placozoa [16, 69]. Genes of GPCRs and associated signaling molecules are identified even in the genomes of unicellular protists [80]. As several mammalian hormones act as chemoattractant or chemorepellent when applied to the external media of unicellular eukaryotes, it is likely that these prototype GPCRs used for cell-to-cell communication have evolved later as hormone receptors in metazoans.
The function of hormones is flexible even within the same family depending on the secretory stimulus, target tissues, affinity to different receptors and others, which generates functional variety among peptides in the same family and exemplifies the concept of ‘biased signaling’ [81]. Furthermore, the function of a hormone also changes during evolution from fishes to mammals and, more obviously, from invertebrates to vertebrates. In this section, therefore, the author will initially summarize what is known about the function of insulin superfamily in mammals to provide the framework for comparison with those of other animals, which will help elucidate an original or essential function of the insulin superfamily from the comparative viewpoint.
Biological actions in mammals InsulinAs mentioned above, insulin is a sole hypoglycemic hormone in mammals, and thus its deficiency causes diabetes and other metabolic diseases. The presence of a single hypoglycemic hormone, in contrast to multiple hyperglycemic hormones, may reflect a long history of starvation in early mammalian life. The powerful hypoglycemic effect of insulin that leads to nutrient storage could be explained in part by ‘thrifty gene hypothesis,’ in which the gene that increases fat deposit is advantageous in the food-deficient life, but it leads to obesity in the modern life where food is excessively available [82].
The mechanism of insulin action has been intensively investigated for a century, and the fruits of the research have been summarized in several excellent reviews [35, 37]. In terms of glucose metabolism, the principal target organs of insulin are the skeletal muscle, liver and white adipose tissue, where insulin promotes glucose uptake by activating glucose transporters (GLUTs). In the skeletal muscle, insulin recruits GLUT4 from the cytoplasmic vesicles to the plasma membrane to increase glucose influx, and a resultant increase in cellular glucose activates glycogen synthesis [83]. In the liver, insulin stimulates glycogen synthesis and protein synthesis by utilizing glucose that enters the hepatocytes through GLUT2 [84]. In the white adipose tissue, insulin stimulates lipogenesis using glucose taken up through the recruited GLUT4, while it potently inhibits lipolysis via suppression of hormone-sensitive lipase [85]. In this way, insulin is deeply involved in carbohydrate metabolism, but it also plays important roles in lipid metabolism, thereby functioning as one of the major anabolic hormones.
In relation to nutrient metabolism, restricted nutrient intake leads to delayed aging and longevity in mammals and other animals [86], and insulin seems to be involved in this effect on aging and lifespan. Mice lacking insulin receptors in adipose tissue show extended longevity [87], and mice with impaired insulin signaling systems also exhibit longer lifespan and delayed aging [88]. It should be emphasized that the involvement of ILP in the regulation of aging and longevity was first reported in C. elegans [89], followed by the report in Drosophila [90].
Another important insulin action concerns mitogenic signaling through activation of the mitogen-activated protein kinase (MAPK) pathway common to the actions via RTKs [91]. Insulin also acts on the brain across the blood-brain barrier to regulate brain functions such as feeding behavior (appetite) and cognitive behavior [92]. Insulin also regulates nutrient metabolism by central actions through the autonomic nervous system.
IGFsIGF-1 is synthesized in the liver in response to growth hormone, with hepatic synthesis accounting for almost 75% of the plasma IGF-1 [93]. Because of its action as a mediator of somatotropin (growth hormone), IGF-1 was originally named somatomedin C. The major function of IGF-1 is to promote skeletal growth, and the IGF-1 gene (igf1) knockout results in small body size and high post-natal mortality [94]. In addition to the endocrine action, IGF-1 is synthesized in various tissues and acts in a paracrine and autocrine fashion to promote cell proliferation and protein synthesis like other growth factors. In contrast to the postnatal actions of IGF-1, IGF-2 is expressed and functions most actively during the intrauterine life and plays important roles in fetal growth and development. This is partially because igf2 is maternally imprinted, and its expression in the fetus arises from the paternally inherited allele. Thus, igf2-null mice have intrauterine growth retarded and smaller size at birth compared with the wild type [95].
Evidence for the role of IGF-1 in the control of aging was provided through studies on female mice with mutation of the IGF-1 receptor gene (igfr1), which live longer than wild-type controls [88]. These mutants exhibited minimal reduction in growth and no alteration in sexual maturation, while the life extension was shown to be associated with the increased tolerance to oxidative stress. On the other hand, the mice with homozygous knockout of igf1r were not viable, but heterozygous knockout mice lived longer than their wild-type littermates, and the effect on lifespan was significantly greater in females than in males [96]. Thus, both insulin and IGF seem to have effects on longevity in rodents.
Another important characteristic of IGFs is the presence of IGF-binding proteins (IGFBP1-6), which bind IGFs in plasma and regulate their action and clearance [97]. IGFBPs have similar affinities to IGF-1 and IGF-2 but do not bind to other insulin superfamily members including insulin. The affinity of IGFs to IGFBPs is greater than to IGF receptors and thus limit the availability of IGFs for receptor activation. On the other hand, IGFBPs extend the half-life of IGFs in plasma by protecting them from plasma peptidases and from renal clearance. IGFBPs also have binding sites to proteoglycans on the cell surface, and their binding to the proteoglycan provides potential mechanisms for sequestration of IGFs to their receptors in specific tissue compartments. IGFBPs are divided into two main groups, IGFBP3/5/6 and IGFBP1/2/4, and the former IGFBPs are frequently localized in the nucleus and may have functions as transcription factors [98].
RelaxinsRelaxin is synthesized in various reproductive organs of mammals and is initially identified as a hormone that induces relaxation of pubic ligament and thus smooth parturition [33, 99]. As relaxin-2 occurred only in catarrhine primates and has a similar amino acid sequence and receptor selectivity of relaxin-1 (to RLF1 and 2), the term ‘relaxin’ used hereafter means relaxin-1/2 if not specified. After its discovery, relaxin was shown to have growth factor-like functions in female reproductive organs by inducing the growth of gonadal follicles, mammary gland, uterus and other tissues, resulting in the promotion of ovulation, implantation, parturition and lactation. Relaxin is produced also in male reproductive organs (prostate and testis), and the peptide released into the seminal fluid is thought to promote implantation of the ovulated follicle. In non-reproductive organs, relaxin shows effects on cardiovascular system (vasodilation), body fluid regulation (drinking elicitation), wound healing (connective tissue growth) and immune regulation [99]. It seems that relaxin still holds growth-promoting effects as do insulin and IGF.
In contrast to relaxin-1/2, relaxin-3 is synthesized most abundantly in the brain, with lesser expression in male reproductive organs [33]. Relaxin-3 preferentially binds to RXFP3, but it also interacts with RXFP1 and RXFP4. Relaxin-3 sequence is most conserved in the relaxin/INSL family throughout vertebrate species, but its physiological function is still obscure. As deduced from the high expression of relaxin-3 and RXFP3 in the brain, its central action has attracted attention. The possible central actions thus far reported include modification of stress response, feeding behavior, locomotor activity, etc. [33].
INSLsMolecular phylogenetic analyses of INSLs and their receptor preference classify them in the same clade with relaxin [100]. Among INSLs, biological function has been most rigorously investigated for INSL-3 [101], partly because it is a unique ligand for RXFP2. The cDNA coding for INSL-3 was first isolated from the testes of several mammalian species, while its gene is expressed also in the female reproductive organs such as the ovary, uterus and placenta. The insl3 or rxfp2 knockout mouse has no obvious phenotypic changes but is infertile in males with testes located high in the abdominal cavity, called cryptorchidism [33]. In females, the knockout of insl3 or rxfp2 gene results in impaired fertility with longer estrous cycles.
INSL-5 and its receptor, RXFP4, are pseudogenes in rats and dogs [102]. INSL-5 mRNAs are detected in various tissues of humans, but its physiological function is still under investigation. The INSL-4 and INSL-6 genes are located with the insulin-1 and insulin-2 genes in tandem on the same chromosome 9 of humans, and specific receptors for INSL-4 and INSL-6 are still unknown [33].
Non-mammalian vertebratesThe biological functions of insulin and IGF have not been reported in cyclostomes (Fig. 4). Elasmobranchs (sharks and rays) are tolerant to hypoglycemia, and mammalian insulin injections decreased plasma glucose slowly and only slightly [103]. In teleost fishes, hyperglycemia has been shown to increase plasma insulin concentrations slightly, while mammalian insulin administration slowly increased glucose uptake by the liver and skeletal muscle followed by an increase in tissue glycogen deposition [104]. Furthermore, insulin depressed the rate of gluconeogenesis and upregulated the expression of genes involved in glycolysis in the rainbow trout [105]. On the other hand, amino acids injections increased plasma insulin concentrations more potently than glucose injections in the flounder [106], whereas acute insulin injection decreased plasma free fatty acid concentrations linked to enhanced hepatic lipogenesis [107]. Thus, insulin is involved in various aspects of nutrient metabolism in teleost fishes, probably depending on their food habits [108]. IGF-1 and IGF-2 are involved in growth and development in fishes as in mammals [109], while IGF-3 identified in teleosts is expressed most abundantly in gonads and involved in ovulation in the zebrafish [46].
Studies on insulin physiology are still scarce in amphibians [103], but there are a few reports showing that toad pancreatic tissues incubated with glucose increased insulin release at 8 mM glucose in the medium [110]. The sensitivity to glucose is much lower than that of mammalian pancreatic tissue in vitro. In reptiles, isolated pancreatic islet tissue exhibits very low sensitivity to glucose for insulin secretion, and the half-maximal glucose concentration for the secretion was 19.4 mM in the anole lizard [111].
Birds are interesting species for research on glucose metabolism, as they have blood glucose concentration twice higher than that of humans without showing metabolic syndrome [112]. Thus, birds are excellent models for the study of tolerance to hyperglycemia. In fact, insulin has only a weak effect on glucose uptake into the skeletal muscle in birds [113], which is accounted for by the lack of the GLUT4 gene in birds [114]. More recent study showed that the GLUT4 gene is present in the chicken genome, but the gene is masked for detection by the high frequency of GC repeats around the gene [115]. In fact, GLUT4 is expressed in the chicken skeletal muscle but only slightly. It seems that GLUT1 is responsible for glucose uptake in the skeletal muscle and adipose tissue of birds [116]. Glucose taken up into adipocytes is used for lipogenesis, while insulin inhibits lipolysis, resulting in fat deposition. Fat deposits are used for effective energy production during long flights of migration.
Collectively, insulin is involved in nutrient metabolism in all vertebrates. Unlike mammals, however, not carbohydrate metabolism but lipid metabolism is a major target of regulation by insulin in non-mammalian species.
Biological actions in invertebrates ProtostomesThe biological function of ILPs has been investigated vigorously in insects (silkmoth [117], Drosophila [62]) and nematodes (C. elegans [54]), in which ILPs are shown to be involved in the various aspects of physiological regulations such as growth, development, reproduction, metabolism, and behavior as in vertebrates. The first ILP identified in invertebrates is bombyxin, which was initially thought to be ecdysone-releasing hormone [50], but later studies show that it is a pleiotropic hormone like insulin [53, 117]. The interesting story about the history of bombyxin discovery is described by Ishizaki [118]. Most of the bombyxin-like ILPs are secreted from the brain to regulate nutrient-dependent growth and metabolism. A silkmoth IGF-like peptide, named BIGFLP, is also identified, which is secreted mostly from the fat body to regulate growth during development from pupae to adults [119]. The fat body of insects is a multifunctional organ equivalent to the liver [120].
Among the ILPs of Drosophila, DILP1-7 are involved in diverse actions; 1) development and growth as a growth factor, 2) nutrient metabolism, lifespan and reproduction as a hormone, and 3) cognitive behavior and sleep-awake cycle as a neuromodulator [60, 121, 122]. It has been shown that the knockdown of each of the DILPs resulted in either increased lifespan, reduced fecundity, decreased carbohydrate and lipid metabolism, increased stress resistance [57, 62]. DILP6 is an IGF-like peptide of Drosophila and highly expressed in the fat body during the pupal stage [119]. Concerning lifespan, mutation of the receptor gene or chico, a gene orthologous to mammalian insulin receptor substrate, significantly increased lifespan in some Drosophila strains [88, 90].
On the other hand, DILP8 is a relaxin-like peptide and binds to Drosophila LGR3, which is orthologous to mammalian relaxin receptor RXFP1/2 [123]. DILP8 has been implicated in ovarian maturation and ovulation in Drosophila [124], suggesting homology to starfish relaxin-like gonad-stimulating peptide (see below). Similar reproductive function has been reported for a DILP8 homolog in malaria mosquito, Aedes aegypti [125] and other arthropods [126]. Gonadulins identified in gonads of a few species of arthropods [127] are homologs of DILP8. In the larva, DILP8 acts on the prothoracic gland to reduce ecdysone production and release and thus controls tissue growth and development for metamorphosis [123, 128]. Recently, Drosophila orphan LGR, LGR4, is suggested to be a receptor for another relaxin-like peptide, DILP7, and DLIP7-LGR4 system seems to modulate escape behavior of Drosophila larvae [129].
Unlike the presence of pseudogenes in silkmoth ILPs, all of 40 ILP (INS) genes appear to be expressed in C. elegans, and their functions are categorized as agonistic, antagonistic, or pleotropic based on experiments overexpressing each gene [26]. Because of the functional redundancy of highly similar ILPs, functional analysis of each gene by loss-of-function experiments is difficult. Pleiotropic functions of C. elegans ILPs include regulation of aging/lifespan, development, reproduction, behavior, memory/learning, dauer formation, and fat accumulation [54]. Like Drosophila, C. elegans is advantageous for research on aging and lifespan as it has a short life of a few weeks, conserved embryonic and post-embryonic developmental stages, genetic tractability, etc. Thus, this worm has been used frequently as a model animal for the study of lifespan [130, 131] and aging [132]. These studies contribute significantly to our understanding of the aging mechanisms of humans.
DeuterostomesSeveral ILPs have been identified in all deuterostome classes including chordates, hemichordates and echinoderms (Table 1). Among them, the function has been investigated for relaxin-like gonad-stimulating peptide (RGP) of the starfish, Asterina (or Patiria) pectinifera [133]. RGP was previously called gonad-stimulating substance (GSS), which is produced in the radial nerve cords of starfish and acts on gonads just like gonadotropins (FSH and LH) of vertebrates [134]. GSS induces final maturation of oocytes or meiotic resumption by acting on the ovarian follicle cells or testicular interstitial cells to produce maturation-inducing hormone, 1-methyladenine, resulting in egg or sperm shedding. RGP has been identified in all classes of echinoderms including sea cucumbers, brittle stars and sea lilies [58]. RGP is phylogenetically closest to relaxin-3, which may be an ancestral molecule of the relaxin family. The RGP receptor has also been identified in the starfish, which is a LGR as in mammals [135]. It was later shown that the receptor is a homolog of LGR3 identified in Drosophila as a receptor for DILP8 [136]. Transcriptomic analyses of the starfish radial nerve cords identified another RGP named RGP2 [137], and RGP2 also triggered spawning in starfish [138]. Both RGP and RGP2 act as ligands for LGR3 in the starfish [137].
In cephalochordates, the amphioxus, Branchiostoma californiensis, has a single ILP that is equally similar to insulin and IGF [64]. The ILP gene was expressed in the hepatic caecum, the homologous organ of the liver, and rat GH upregulated its expression in this amphioxus tissue just like IGF-1 in the liver of vertebrates [139]. Recombinant amphioxus ILP stimulated cell proliferation in flounder gills. The IGFBP-like gene was identified in B. floridae that has multiple ILPs [140]. The amphioxus IGFBP seems to have IGF-independent actions such as transcription factors as it is localized in the nucleus as reported in vertebrate IGFBP3/5/6. The tunicate, Ciona intestinalis, also has similar sets of ILPs (Table 1), but physiological functions of tunicate ILPs have not yet been reported.
It is important to trace the history of diversification of the insulin superfamily peptides during metazoan evolution and to identify the ancestral molecule, both of which provide a basis for consideration of the original function of the superfamily. In deuterostomes, tunicates and cephalochordates already possess three types of ILP genes, insulin-like, IGF-like and relaxin-like genes [63, 66], suggesting that divergence of the two types in vertebrates (insulin/IGF and relaxin/INSL) occurred in primitive chordates (Fig. 4). Compared with tunicates and cephalopods, vertebrates underwent two more WGD, resulting in quadruplicates of each gene in mammals [141]. Further linage-specific local gene duplications and losses caused three insulin/IGF genes and seven relaxin/INSL genes in the human genome (Fig. 5).

The ancestral gene common to insulin/IGF and relaxin/INSL may have existed in pre-bilaterians before divergence of deuterostomes and protostomes (Fig. 5), because both types of ILPs are found in both bilaterian groups (Table 1). The most reliable criterion to distinguish between the two families is their receptors, RTK for insulin/IGF and LGR for relaxin/INSL. It is suggested that Drosophila DILP7 binds to both RTK and LGR [58, 129]. Thus, the ancestral ILP molecule of pre-bilaterians may have possessed a weak receptor selectivity. In fact, both GPCR and RTK have a long history of evolution from the pre-metazoan era, and their genes are detected even in unicellular organisms [77, 142]. Thus, either insulin/IGF or relaxin/INSL could find a binding partner when metazoans emerged. It seems difficult, therefore, to determine whether the ancestral molecule belongs to the insulin/IGF family or the relaxin/INSL family only from its receptor.
Accordingly, let us consider the ancestral molecule from its function. It is obvious that the functions of insulin and IGFs are mutually related, but they are clearly distinguishable; insulin is principally a metabolic regulator, while IGFs are growth factors regulating cellular growth, differentiation, and apoptosis in all metazoan species [93]. Unlike consistent IGF actions throughout metazoans, the target nutrient of insulin for metabolism has changed even within vertebrates; the major target is carbohydrates in mammals, while lipids are the common target from fishes to mammals [85]. This is also true in invertebrates; manipulations of the insulin signaling system consistently affect lipid metabolism but did not change carbohydrate metabolism in Drosophila [143], although such manipulations change plasma trehalose level in the silkmoth [117]. It seems appropriate to speculate that ancestral peptide of insulin/IGF is involved in growth and development as a growth factor in early metazoans and the metabolic effects have evolved later to supply energy for such functions.
The function of relaxin family peptides is more directed to reproduction, and relaxin-like peptides play important roles in the gonadal maturation and gamete shedding as observed in starfish [138] and in growth and maturation of reproductive organs in mammals [99]. It is not known when the relaxin-type peptide emerged in metazoan evolution, but the first peptide may have serendipitously activated a GPCR in the reproductive organs and consequently developed a new function, although extant relaxin-like peptides still retain the effects on growth and development [123, 136]. It is of interest to note that both starfish RGP and vertebrate gonadotropins (FSH and LH) activate LGR and stimulate reproductive function. Reproduction is as important as growth and development for species conservation.
Based on the above discussion, it is hypothesized that the ancestral molecule of the insulin superfamily may be a mitogenic growth factor akin to IGF and act in a paracrine fashion in the earliest multicellular metazoans [34] (Graphical Abstract). Growth factors share RTK as cognate receptors, and they are not only phylogenetically old but also ontogenetically active during early stages of animal development. It is of interest to note that the TGF-β superfamily, another hormone family phylogenetically traceable to pre-bilaterians, also comprises growth factors [26]. It is quite intriguing to examine how insulin and relaxin peptides have acquired new functions to regulate metabolism and reproduction in addition to growth and development in early bilaterians.

The author thanks Dr. Christopher A. Loretz of the University at Buffalo for comments on the manuscript and editing English. He also expresses sincere gratitude to Dr. Akira Mizoguchi of Nagoya University, Dr. Masatoshi Mita of Showa University, and Dr. Takashi Kadowaki of Toranomon Hospital for critical reading of this manuscript as experts of the insulin superfamily of invertebrates and vertebrates.
The author has no potential conflicts of interest associated with this review.
This review is not supported by any grant from funding agencies.